Different Types of Calcium Channels

Fig. 1 (taken from Nowycky, Fox & Tsien, 1984) illustrates the scheme that has been a useful test for multiple Ca²⁺ channel types on a single cell. The cell is a chick dorsal root ganglion neurone (DRG) and the only current carrier is 10 mmol l⁻¹ Ca²⁺ in the extracellular medium. With the cell held at –60 mV by voltage clamp, a pulse to 0 mV elicits a current which does not decline during the several hundred millisecond pulse. When the cell is held at –100mV, a decaying current is superimposed on the previous current. The hyperpolarization has unmasked a current which was inactivated at a holding potential of –60 mV and which decays within 100ms during a pulse to 0 mV. The decay appears to be voltage-dependent inactivation because exchanging Ba²⁺ for Ca²⁺ did not alter the decay rate as expected of an inactivation due to accumulation of intracellular Ca²⁺. The usual screening procedure for multiple channels is, therefore, selective steady-state inactivation. Like many aspects of the Ca²⁺ channel literature, multiple Ca²⁺ channel types can be traced back to work done by Susumu Hagiwara. The pioneering work, done on starfish eggs and published in 1975, distinguishes two Ca²⁺ current components by steady-state inactivation, activation threshold and selectivity as shown in Fig. 2 (taken from Hagiwara, Ozawa & Sand, 1975). The data connected by solid curves in Fig. 2A plot on the ordinate the peak current in response to voltage pulses to the potentials plotted on the abscissa. Currents were recorded with Ca²⁺, Sr²⁺ or Ba²⁺ as the charge carrier. In each case, there are two local minima in the current-voltage curve suggesting the presence of two channels which activate at different voltages. Hagiwara called the current peak centred at –30mV ‘Type I’ and the current activated at higher potentials ‘Type II. The crossing of curves indicates that the two currents have different selectivity. The type II channel passes greater Ba²⁺ currents than Ca²⁺ currents, as is typical of ‘classic’ Ca²⁺ channels; the order is reversed for type I channels.

Fig. 2B demonstrates the different inactivation properties of the two currents. The relative current amplitude during a test pulse is plotted against the potential during a conditioning prepulse intended to cause varying levels of inactivation. The triangles and circles are generated with test pulses designed primarily to activate the type I channels; the resulting inactivation curve is monotonic and clearly fitted by the standard Boltzmann distribution (solid curve). The squares are generated with test pulses to +22mV, which should activate both type I and type II currents. The resulting inactivation curve is unusually broad and appears biphasic in the vicinity of –40 mV. This has been fitted by summation of a scaled curve for the type I channel (solid curve) and a separate Boltzmann curve for the type II channel (dashed curve). This experiment remains the most clearly quantified separation of the inactivation kinetics of two Ca²⁺ channel types.

The properties of different activation thresholds and different inactivation ranges have been applied to a wide range of cells and multiple Ca²⁺ channel types have become the rule rather than the exception (for reviews see Reuter, 1985; Miller, 1985; Tsien et al. 1986). The following is an attempt to catalogue the cells in which multiple Ca²⁺ channel types have been observed: starfish eggs (Hagiwara et al. 1975), tunicate eggs (Okamota, Takahashi & Yoshii, 1976), Neanthes worm eggs (Fox, 1981; Fox & Krasne, 1984), inferior olivary neurones (Llinas & Yarom, 1981a,b), neuroblastoma cells (Fishman & Spector, 1981; Tsunoo, Yoshii & Narahashi, 1985; Yoshii, Tsunoo & Narahashi, 1985), rat hippocampal neurones (Halliwell, 1983; Gray & Johnston, 1986), dorsal root ganglion neurones (Nowycky et al. 1984, 1985; Carbone & Lux, 1984a,b;,Fedulova, Kostyuk & Veselovsky, 1985), GH₃ and GH₄ cells (Armstrong & Matteson, 1985; Cohen & McCarthy, 1985), sensory cranial neurones (Bossu, Feltz & Thomann, 1985), rat olfactory neurones (Brown et al. 1984), protozoan cilia (Deitmer, 1984), dog and frog atrial cells (Bean, 1985), guinea-pig ventricular cells (Nilius, Hess, Lansman & Tsien, 1985; Mitra & Morad, 1985), Aplysia bag cells (Chesnoy-Marchais, 1985), rat mesenteric artery smooth muscle (Bean, Sturek, Puga & Hermsmeyer, 1985, 1986), rat venous smooth muscle (Sturek & Hermsmeyer, 1985), rabbit ear artery smooth muscle (Aaronsonet al. 1986), skeletal muscle myotubes (Beam, Knudson & Powell, 1986; Cognard, Lazdunski & Romey, 1986) and pituitary lactotrophs (DeRiemer & Sakmann, 1986). Cells which appear to have only one type of Ca²⁺ channel are: adrenal chromaffin cells (Fenwick, Marty & Neher, 1982; Hoshi, 1985), hybridoma cells (Fukushima & Hagiwara, 1983) and type II astrocytes (Barres, Chun & Corey, 1985).

Fig. 3 shows records from the paper of Nowycky et al. (1985) showing whole-cell Ca²⁺ current and associated current-voltage curves from a DRG neurone. As noted before, pulses from a holding potential of –40mV elicit a non-inactivating current (Fig. 3A, smaller currents) represented by squares on the current-voltage curve (Fig. 3B). This prolonged current has been called ‘L’ (for Long-lasting, Large Ba²⁺ conductance; Nowycky et al. 1985) and ‘HVA’ (for High Voltage-Activated; Carbone & Lux, 1984a,b). From a holding potential of –100 mV, pulses to negative potentials elicit a current which inactivates completely and appears to reach its maximum amplitude near –10 mV. This component has been called ‘T’ (for Transient, Tiny Ba²⁺ conductance; Nowycky et al. 1985) and ‘LVA’ (for Low Voltage-Activated; Carbone & Lux, 1984a,b). Pulses to positive potentials from the – 100 mV holding potential elicit, in addition to the non-inactivating L current, an inactivating current of larger amplitude than expected of the T-type current. The component is like T in that it inactivates and requires a negative holding potential but is like L in its voltage range of activation. Fig. 3C plots the fraction of current which relaxes during the pulse. There are two clear components of the current-voltage curve with peaks at –10 mV and +20 mV. The authors proposed that this second peak of inactivating current was due to yet a third Ca²⁺ channel type. The challenge of this hypothesis is to prove the existence of a separate molecular entity despite the fact that it can only be activated under conditions that also activate the other two channels. This component is therefore called ‘N’ because proof that the current results from a separate channel requires evidence that it is caused by Neither T nor L.

The on-cell patch method (Hamill et al. 1981) allows the recording of individual ion channels by isolating a sub-microscopic patch of membrane. With this method, Nowycky et al. (1985) were able to demonstrate clearly the existence of three distinct Ca²⁺ channel types. Fig. 4 shows results from three different patches that show three different kinds of single channel activity. Fig. 4A shows a small-conductance channel activated with a protocol appropriate for T. Activity is bunched towards the beginning of the pulse as expected of an inactivating current. Fig. 4C shows a large-conductance channel activated from a depolarized holding potential that allows only L-type activity. Openings are scattered relatively evenly throughout the pulse as predicted for a non-inactivating channel. Fig. 4B shows a channel which is activated with pulses to positive potentials from a hyperpolarized holding potential as expected for the putative N channel. It has a distinctly larger single-channel amplitude than T yet its activity is towards the beginning of the trace, unlike L. Its single-channel conductance (13 pS) in 110 mmol l⁻¹ Ba²⁺ is roughly half that of the L channel (25 pS) and greater than the T channel (8pS) (making it iNtermediate in Ba²⁺ conductance). Thus, a third channel was demonstrated at the single-channel level with properties that could generate the N-type activity at the whole-cell level. Table 1 summarizes the single-channel conductances and kinetic features of these three channels.

Ligands that distinguish different Ca²⁺ channel types would be used both to dissect the different physiological functions of the channels and to serve as biochemical probes for isolation and purification. Two compounds seem useful in this regard.

Nifedipine and related dihydropyridines are the most potent organic compounds that block Ca²⁺ channels and are used clinically due to the dihydropyridine sensitivity of cardiac and smooth muscle Ca²⁺ channels. The family of dihydropyridines contains both blockers and agonists that enhance Ca²⁺ current. Fig. 5, from Hess, Lansman & Tsien (1984), shows the effect of an agonist, Bay K 8644, on the L-type channel in cardiac cells. The presence of Bay K 8644 causes a dramatic increase in the open time of single Ca²⁺ channels leading to an increase in the overall Ca²⁺ current. This effect occurs only with L-type channels in both DRG neurones (Nowycky et al. 1985) and cardiac cells (Nilius et al. 1985). Dihydropyridine agonists, therefore, provide a clear assay for the presence of L-type channels. While enhancement of single-channel currents by dihydropyridine agonists is clear and dramatic, block by antagonists is not as pronounced. Sensitivity of L-type channels to dihydropyridines varies among cell types and varies with the cell membrane potential (Bean, 1984; Sanguinetti & Kass, 1984). Neurones show relatively less sensitivity to dihydropyridines than does muscle tissue. At the negative holding potentials from which all channels can be activated, L channels in neurones cannot be completely blocked even at the highest dihydropyridine concentrations possible in aqueous media (10⁻⁵–10⁻⁴mol l⁻¹). The remaining L current is not readily distinguished from N and T currents because dihydropyridines induce a more rapid decay of L current (Lee & Tsien, 1983).

The only selective blocker known which is both potent and does not alter the kinetics of residual current comes from the venom of a marine snail, Conus geographus. This snail feeds on fish by stinging with a venom which causes virtually immediate paralysis and death. B. M. Olivera and his colleagues have characterized the peptides in Conus venoms (Olivera et al. 1985) and described the ω-VIA Conus geographus toxin (ω-CgTX), which has 14 hydroxyl groups among 27 amino acids and a net positive charge of 4 (Olivera et al. 1984). Kerr & Yoshikami (1984) showed that ω-CgTX blocks neuromuscular transmission in frogs by decreasing the quantal content of evoked acetylcholine release. They proposed that this was due to decreased Ca²⁺ flux in the presynaptic terminal because Ca²⁺ action potentials in DRG neurones were diminished by the toxin. The inhibition of the action potentials could have been due either to blockade of Ca²⁺ channels or to an enhancement of a hyperpolarizing conductance. Feldman & Yoshikami (1985), using voltage-clamp techniques, proved that the toxin blocked Ca²⁺ channels without an effect on the delayed rectifier potassium current.

Does the toxin distinguish among different channel types? Fig. 6, taken from McCleskey et al. (1986), shows recordings of T, N and L currents from DRG neurones recorded before and after a transient application of ω-CgTX. The T current, recorded at higher gain in Fig. 6A, is not blocked whereas the L current in Fig. 6C and the L and N currents in Fig. 6B are blocked. This selectivity has allowed, for the first time, recording of whole-cell T current on DRG neurones in the absence of N and L currents and provides a clear pharmacological distinction between the two inactivating currents, T and N.

Besides selecting among different channel types on a particular cell, the toxin is able to select among otherwise similar channels in different tissue. Table 2 summarizes results from eight preparations. L currents are blocked in the three neuronal cells but not blocked in the five muscle cell preparations. This specificity for neural Ca²⁺ channels may serve the purpose of selectively destroying neuromuscular transmission without lowering the toxin concentration through binding to the relatively high concentration of muscle Ca²⁺ channels. The specificity indicates that neuronal and muscle L-type Ca²⁺ channels cannot be entirely identical at the molecular level, though their kinetics and selectivity are similar.

The effect of ω-CgTX on N and L channels in DRG neurones shows little or no reversibility. During recordings lasting up to about 30 min, there has never been any recovery of current after a transient application of toxin. More than 80% of labelled ω-CgTX remains bound to chick brain synaptosome membranes after 2 h and the apparent dissociation constant with 30-min incubations is about 1 nmol l⁻¹ (Cruz & Olivera, 1986). The binding site is distinct from the dihydropyridine and verapamil binding sites (Cruz & Olivera, 1986) and, unlike dihydropyridines, blockage is not voltage-dependent and partial blockage does not alter the kinetics of the remaining current. Use of the ω-CgTX as a biochemical ligand for Ca²⁺ channels requires that it binds directly to the channel rather than acting through a second messenger. The toxin is effective at blocking L channels in cell-free, excised patches of membrane when applied to the external surface of the membrane. However, when the toxin is applied to the bulk of the cell surface it fails to block single channels protected from direct contact with the toxin by an on-cell recording pipette. Evidently, binding of ω-CgTX to the external surface of the Ca²⁺ channel is both necessary and sufficient for block.

ω-CgTX has several advantages over dihydropyridines as a biochemical ligand for Ca²⁺ channels. There is no ambiguity that Ca²⁺ channels are its primary target, it binds with little or no reversibility, block is independent of the gating state of the channel, and the toxin is not lipid-soluble. The dihydropyridine binding site of skeletal muscle has been purified (Glossman & Ferry, 1983; Curtis & Catterall, 1984; Borsotto, Barhanin, Norman & Lazdunski, 1984) and Ca²⁺ flux has been reconstituted into vesicles (Curtis & Catterall, 1986). Purification of ω-CgTX binding sites in neurones may provide the opportunity to compare the structures of L and N channels as well as L channels from nerve and muscle.

Ca²⁺ channels can make membrane electrical signals, intracellular chemical signals or a combination of both. L-type Ca²⁺ channels in cardiac cells clearly serve both electrical and chemical functions during the heartbeat. The fact that there is only very slow inactivation of L channels allows the maintained Ca²⁺ current that underlies the plateau phase of the cardiac action potential (Reuter, 1967). The Ca²⁺ entering the cell through L channels is necessary both for replenishing the Ca²⁺ stores of the sarcoplasmic reticulum (Chapman & Niedergerke, 1970) and possibly for triggering intracellular Ca²⁺ release.

L-type channels also underlie secretion in some cells. Dihydropyridine-sensitive channels are required for substance P release from DRG neurones (Perney, Hirning & Miller, 1986) and slowly inactivating L-like conductances are responsible for Ca²⁺ influx and transmitter release in adrenal chromaffin cells (Fenwick et al. 1982) and the squid giant synapse (Llinas, Steinberg & Walton, 1981; Augustine, Charlton & Smith, 1985).

A function of T-type Ca²⁺ channels in neurones was proposed before there had been a direct recording of T currents. Llinas & Yarom (1981a,b), in an elegant study of inferior olivary neurones, demonstrated a rapid Ca²⁺ action potential that could be induced only if the cell was hyperpolarized. This T-like activity would be inactive at all times except during the after-hyperpolarization that follows a burst of action potentials and it was proposed that the conductance helped control the duration of the interburst interval. A full cycle of bursting would utilize two different Ca²⁺ channels in different ways. The action potential is qualitatively like that in cardiac cells in that it consists of a sodium spike followed by a prolonged calcium plateau caused by a non-inactivating, L-like conductance. A burst of such spikes is terminated when the accumulation of intracellular Ca²⁺ sufficiently activates a Ca²⁺-dependent potassium conductance. The after-hyperpolarization caused by the increased potassium conductance would unmask the T-type channels which reprime excitability and help trigger the succeeding burst. Thus, T-type channels would play a crucial electrical role in neuronal coding.

Discussion of the function of N channels is speculative at this time but it is tempting to suggest that N-type channels serve a neurone-specific function since they have only been demonstrated in nerve. Could the N channel be responsible for neurotransmitter release from presynaptic terminals? The pharmacology of evoked release of acetylcholine from the frog neuromuscular junction is similar to N channels in that it is blocked by ω-CgTX (Kerr & Yoshikami, 1984) but not affected by dihydropyridines (Fairhurst, Thayer, Colker & Beatty, 1983). Richard Miller and colleagues have compared the pharmacology of transmitter release in various cultured neurone preparations and found a variety of responses. Potassium-evoked substance P release from DRG neurones is blocked by dihydropyridines but norepinephrine release from sympathetic neurones was not (Perney et al. 1986). However, norepinephrine release is blocked by ω-CgTX. We have shown that the L-type channel in sympathetic neurones is dihydropyridine-sensitive and have found a slowly inactivating N-like current that is toxin-sensitive. Thus, toxin-sensitive, N-like channels are a candidate for mediating release in sympathetic neurones whereas the dihydropyridine-sensitive, L-type channels seem to be ruled out. Overall, pharmacological data suggest that the identity of the channel responsible for neurotransmitter release may be L in some tissues and N in others. If so, ω-CgTX, which blocks both L and N channels in neurones but fails to block L channels in muscle, seems ideally designed to inhibit presynaptic release.

The past 2 years have been a watershed for the description of cells with multiple types of Ca²⁺ channels; a wide variety of neuronal, muscle and endocrine cells have been shown to have at least two kinds of Ca²⁺ channels. Dorsal root ganglion neurones have three types of Ca²⁺ channels, called T, N and L, which are distinguishable by steady-state inactivation, activation threshold, selectivity and pharmacology. A neurotoxin from the Conus geographus snail blocks only N- and L-type Ca²⁺ channels in DRG neurones, apparently by binding directly to the external surface of the channel; this promises to be a crucial probe for both the biochemistry and function of different channel types.

The toxin has created new ‘sub-classes’ of channels by distinguishing between L channels in nerve and muscle. In recording Ca²⁺ currents in different preparations, one notes that largely similar channels show enough subtle variations that a mere alphabet would be insufficient for cataloguing. On what should we base the nomenclature? Toxicology would be a poor choice. Sodium channels responsible for the rapid action potential in nerve transmission are homogeneous with respect to function, show subtle differences in kinetics and temperature dependencies, and show striking differences in toxicology. Should the sodium channels of tetrodotoxin-producing puffer fish be considered a separate kind of channel because they are tetrodotoxin-insensitive? In the case of sodium channels, function defines type. In the case of Ca²⁺ channels, elucidation of function is lagging behind the description of kinetics and pharmacology. In lieu of clearly defined functions, it seems safest to define Ca²⁺ channels by their most dramatic kinetic features, as these are most likely to reflect function.