C) ADENOSINE RECEPTORS. The adenosine receptors could be considered as autoreceptors in many systems, since adenosine in the extracellular compartment is formed by ATP breakdown after it is coreleased with the neurotransmitter. In mouse motoneurons, A₁ receptors were found to act via G protein cascade to directly inhibit presynaptic N-type calcium channels (579). The inhibition via G protein was demonstrated also in rat motoneurons (317).

D) GABA RECEPTORS. -Aminobutyric acid is probably the most studied endogenous modulator of synaptic transmission causing inhibition. Its influence on other channels is discussed elsewhere; here we summarize its influence on calcium channels. -Aminobutyric acid acts on GABA_B receptors in the terminal surface membrane (207, 215, 354, 367, 438, 666, 871, 933, 979), which activates a G protein, inhibiting HVA calcium channels in the terminal (331, 354, 367, 666, 871). It is generally accepted that the action of GABA is on the N- and/or P/Q-type channels, but its influence on L-type channels has also been shown. An inhibitory action of GABA on presynaptic calcium channels has been demonstrated in several preparations. Such an action that modulates the ability of cells to transmit information could be one of the ways for maintaining brain plasticity and an important mechanism for therapeutic use. -Aminobutyric acid also inhibits R-type calcium channels (976).

In guinea pig hippocampal synaptosomes, protein kinase C (PKC) phosphorylation increases release of glutamate by increasing VDCC activity (875). In rat cerebral cortex synaptosomes, release of glutamate and calcium current are reduced by dephosphorylation effected by calcineurin (788), and in rat hippocampal synaptosomes, it was found by measuring [Ca²⁺]_i that there is a dynamic balance between phosphorylation by PKC (activation) and dephosphorylation (inactivation) by protein phosphatase, both being tonically active in terminals (55). Protein kinase C phosphorylation occurs on an intracellular loop of the ₁-subunit that binds the inhibiting G protein complex mentioned above (995).

Most of the calcium channel regulation in presynaptic nerve terminals, described in the previous sections, was directed toward the probability of an opening of a calcium channel that is already present in the surface membrane of the nerve terminal. Recently, a very interesting possibility was suggested for human neuroblastoma and for rat pheochromocytoma cell lines (638), which if applicable to nerve terminals opens new directions for regulation of calcium channels and for frequency modulation (see Ref. 686) of synaptic transmission. It was found that secretagogues such as KCl, ionomycin, and phorbol esters (see Ref. 770) cause a translocation of N-type calcium channel from the intracellular pool to the surface membrane (see Ref. 638). If this mechanism exists also in the nerve terminals, then an appropriate stimulation that causes the insertion of new calcium channels in the surface membrane will increase the synaptic strength for subsequent stimuli. This increase in synaptic strength will last as long as the additional calcium channels dwell in the surface membrane in their active state.

In rat corpora striata synaptosomes, extracellular, but not cytosolic, acidification inhibits the release of dopamine by blocking voltage-gated calcium channels (216).

In rat soleus muscle, voltage-dependent calcium influx of the nerve terminal could be increased by muscle stretching (985).

In the sympathetic nerve in kidney and spleen and in heart and blood vessels, PGE₂ and PGI₂ inhibit the release of norepinephrine, inhibitors of cyclooxygenase enhances norepinephrine release via -conotoxin-sensitive channels (506).

In rat cerebral cortex synaptosomes, glucocorticoids promote calcium channel activity by enhancing calmodulin-dependent activation of the channels (853).

In rat brain stem synaptosomes, stimulation of presynaptic receptors coupled to the G_s-adenylate cyclase system may lead to a facilitation of the release of neurotransmitters, through a cAMP-dependent enhancement of the opening of the calcium channels located on nerve terminals. Maitotoxin increases the ⁴⁵Ca entry and [Ca²⁺]_i. These are significantly enhanced by addition of dibutyryl cAMP and by the pretreatment with cholera toxin (903).

In the buccal ganglion of Aplysia, FMRFamide facilitates ACh release by increasing the presynaptic calcium influx. These neuromodulators control only the influx of calcium through N-type channels, since their action on transmitter release can be prevented by -conotoxin but not by funnel web spider toxin. FMRFamide increased calcium influx by shifting the voltage sensitivity to activation of the channels (267).

In lobster abdominal muscles, blockers of N-type channels enhanced excitatory junctional current suppression exerted by high pressure (310).

In Torpedo electric organ synaptosomes, lactate production by stimulated postsynaptic electroplaques may inhibit ACh release from presynaptic nerve terminals. Lactate reduced the release of ACh triggered by depolarization with potassium via inhibition of voltage-dependent calcium influx mediated by the natural calcium channel (286).

A large number of different calcium stores have been described in several cells (for a review, see Ref. 667). They include mitochondria and the intracellular reticulum (endoplasmic and sarcoplasmic reticulum) where at least two different types of channels have been described, namely, IP₃ and ryanodine; the number of different channel molecules is much larger since at least four different genes have been shown to encode the IP₃ channel. In addition, it was proposed that the following intracellular structures can serve as calcium stores: the Golgi complex, secretion vesicles and granules, endosomes and lysosomes, the nucleus, and cytosolic calcium-binding molecules (see Ref. 667). In the nerve terminal, there are no nuclei, but many of the other organelles have been suggested to take part in the regulation of transmitter release, by opening of calcium release channels and by uptake of the cytosolic calcium after nerve stimulation and at rest. We would like to dwell briefly on two intracellular organelles: the reticulum and the mitochondria.

The interest in the role of mitochondria in intracellular calcium homeostasis was very substantial in the 1970s (see Ref. 124). With regard to the presynaptic nerve terminal, it was shown then (15) that mitochondrial calcium uptake inhibitors augment transmitter release, suggesting that mitochondria may be part of the intracellular machinery regulating free [Ca²⁺]_i. In situ morphological studies done on liver mitochondria (92, 810) suggested that the calcium content of mitochondria is rather low while calcium was found in the endoplasmic reticulum (529). These and other findings led to reduced efforts to show that the mitochondrion is an important component of [Ca²⁺]_i regulation under physiological conditions. The situation changed dramatically in the last 5 years when the calcium reporter aequorin was targeted to intact intracellular mitochondria and showed that not only the intramitochondrial calcium content is very substantial, but also that it responds rapidly to channel and receptor activation of the surface membrane (709-711, 735). Recently, it was shown that mitochondria take up and release calcium in physiological calcium loads (190, 868). Hence, it is tempting to predict that the interest in role of mitochondria in the function of the nerve terminal in physiological and pathophysiological conditions (hypoxia, neurotrauma, and aging) will regain momentum (see Ref. 540).

There is no doubt that the ryanodine and the IP₃ channels are expressed in many types of cells (270, 847), including cells in the nervous system (279, 335, 353, 359, 396, 415, 429, 478, 479, 610, 728, 783-785, 909). However, most of the studies do not deal specifically with the nerve terminal.

Recently, however, evidence was obtained on the existence of ryanodine-sensitive calcium stores that take part in transmitter release (801) at the sympathetic nerve terminals of the guinea pig. When the presynaptic nerve was stimulated at low frequencies, all the transmitter release was inhibited by calcium channel blockers acting on calcium channels at the surface membrane. At high frequencies (5 stimuli every 2 s at 20 Hz), a residual release was observed that was time-locked to the stimulus. This residual release was blocked by ryanodine. Frequency of stimulation-dependent calcium release through ryanodine calcium channels was also observed in cerebellar Purkinje neurons (396). These experiments are of potential interest regarding two different aspects. First, they suggest the functional presence of ryanodine calcium channels in the sympathetic nerve terminals. Second, the time locking shows that the communication between the surface and the ryanodine channels is very fast.

The intracellular IP₃ receptor-channel complex is present in many tissues, including the nervous system (see, for example, Ref. 275). This provides possible links among surface membrane receptors (such as metabotropic glutamate receptors), changes in [Ca²⁺]_i, transmitter release, and synaptic plasticity (2, 25, 72, 185, 274, 426, 476, 780, 874). There are also pathophysiological implications since mice deficient in this receptor show ataxia and epilepsy (520).

1.  Lambert-Eaton myasthenic syndrome

A) CLINICAL CHARACTERISTICS OF LAMBERT-EATON MYASTHENIC SYNDROME. Lambert-Eaton myasthenic syndrome (LEMS) is an autoimmune disease that often occurs in association with malignancy. The syndrome is characterized by weakness, myalgias, and a tendency to develop fatigue. The disorder afflicts men more than women. The incidence of associated malignancy is 70% in men and 25% in women. In most cases of either sex, the tumor is a small cell carcinoma of the lung. Other tumors associated with LEMS affect the breasts, the prostate, and the stomach. The diagnosis of LEMS may be suspected clinically but must be confirmed with electrophysiological testing (526), which involves the recording of the electrical response of a muscle to supramaximal stimulation of its motor nerve by repetitive (2-3 Hz) shocks.

B) PATHOPHYSIOLOGY OF LEMS. Lambert-Eaton syndrome is associated with diminished quantal release of ACh at the neuromuscular junction (see Ref. 442). In paraneoplastic LEMS, physiological data incriminate antibodies directed against VDCC present both on the tumor and at distal motor nerve terminals (779). The physiological features of the syndrome can be passively transferred to mice by the patients' sera, consistent with an autoantibody interfering with the function of VDCC (713).

D) OTHER PRESYNAPTIC BINDING SITES FOR LEMS ANTIBODIES. The question of whether LEMS antibodies bind also to other sites on the presynaptic terminal is still debatable. Sera from LEMS patients have a heterogeneous spectrum of antibodies, and some studies have suggested that this heterogeneity reflects the immune response to various synaptic proteins, including not only multiple voltage-gated calcium channels but also synaptosecretory complex proteins (863). It has been shown that plasma from patients with LEMS contains antibodies that bind to the synaptic vesicle protein synaptotagmin (234). This led to the suggestion that since the interaction between synaptotagmin and -conotoxin-sensitive calcium channels may play a role in the docking synaptic vesicles at the plasma membrane before rapid neurotransmitter release, autoantibody binding to a synaptotagmin-calcium channel complex may be involved in the etiology of LEMS (464, 515).

E) MICROSCOPIC FINDINGS. Electron microscopy of presynaptic motor nerve terminals at the neuromuscular junction in LEMS revealed a decrease in the number of structures believed to correspond to voltage-sensitive calcium channels. The same picture was found by passive transfer of LEMS IgG from human to mouse (276). In addition, it was found that the active particles, normally arranged in double parallel rows, move closer together, form clusters, and are reduced in number (583).

2.  Amyotrophic lateral sclerosis

A) CLINICAL CHARACTERISTICS OF AMYOTROPHIC LATERAL SCLEROSIS. Amyotrophic lateral sclerosis (ALS) is a motoneuron disease, manifested by weakness and wasting of muscles. Although reflexes are relatively preserved, the wasting and fasciculation of the muscles is prominent. In the progressive muscular atrophy form of ALS, the anterior horn cells of the spinal cord are primarily affected. There are no sensory changes. Frequently in ALS, there is a combination of spasticity and hyperreflexia in the lower limbs and weakness, wasting, and fasciculation in the upper limbs. In the progressive form of the disease, there is difficulty in swallowing, which may lead to aspiration and pulmonary disease. Intellectual function is usually spared.

B) ETIOLOGY OF ALS. The cause of ALS is not fully understood. There are four main hypotheses: excitotoxicity linked to glutamate receptor overactivation, mutation of the superoxide dismutase gene, neurofilament accumulation, and production of autoantibodies to calcium channels (see Ref. 364). Only the latter alternative is reviewed here.

C) ANTIBODIES TO CALCIUM CHANNELS IN ALS. During the last decade, evidence has accumulated on the involvement of autoantibodies in ALS that lead to motoneuron degeneration and death (27). In animal models, inflammatory foci within the spinal cord and IgG at the neuromuscular junction as well as within upper and lower motoneurons support the role of autoimmune mechanisms in motoneuron destruction (804). Immunoglobulin G fraction from ALS patients can passively induce physiological changes at the neuromuscular junction in mice. The interaction of autoantibodies with calcium channels and, therefore, the alteration of their function have been shown both by biochemical and electrophysiological assays. Furthermore, it has been documented that motoneurons may be selectively vulnerable, since they have a low calcium-buffering capacity (14).

Hibernation is a very puzzling physiological condition. An animal survives a prolonged almost total ischemia of the brain without major deficit. Recently, it was shown that synaptosomes isolated from hibernating ground squirrels exhibit a reduced calcium entry upon depolarization (289). A pharmacological dissection of the phenomenon indicates that it is probably associated with a reduction in the activity of Q-type calcium channels (289). The interaction between hypoxia and the activity of calcium channels occurs not only in synaptosomes but also in smooth muscle (697). It will be of interest to see whether mechanisms operating in hibernation can be used as neuroprotection.

	III. POTASSIUM CHANNELS IN NERVE TERMINALS

Top Previous Next References

C.  Functional Properties According to Channel Type

1.  Delayed-rectifier potassium channels, K channels

The outward potassium current in giant axons of the squid, described by Hodgkin and Huxley (348, 349), was called the delayed rectifier, because it activated, with a delay, upon depolarization and rose more slowly than the sodium current. Ever since, voltage-dependent potassium currents with similar kinetics, lacking rapid inactivation, and exhibiting no activation by intracellular calcium have been referred to as delayed rectifiers. On the basis of their activation threshold, gating kinetics and sensitivity to pharmacological agents such as aminopyridines and tetraethylammonium (TEA), it appears that K channels are not homogeneous and include several subtypes, the determinants of which are not yet fully known.

Table 8 presents nerve endings (with the exception of the Aplysia sensory neuron) in which delayed rectifier channels and currents were directly detected and measured. In the motor nerve terminals of squid, crayfish, frog, lizard, and mouse, K channels contribute to keeping the action potentials short. The high potassium permeability originating from the rapidly activating K channels repolarizes the presynaptic terminal after an action potential. By regulating the action potential duration, the delayed rectifier channels play a critical role in transmitter release. Thus Augustine (42) showed that in squid nerve terminals, the channel blocker 3,4-diaminopyridine (DAP) causes broadening of the presynaptic action potential leading to an increase in calcium entry and an enhancement of transmitter release. Likewise, action potentials in the excitor axon terminals of the crayfish are kept short by a delayed rectifier. Sivaramakrishnan et al. (794) showed that 4-aminopyridine (4-AP), which blocked this current in crayfish terminals, led to action potential broadening and an increase in transmitter release. In frog motor nerve terminals, where similar actions were seen, Hevron et al. (341) and Shakiryanova et al. (769) found in addition that ACh suppressed the delayed rectifier, and it was suggested that this modulation may be significant in cases of acetylcholinesterase inhibition, accounting for nerve terminal membrane hypersensitivity. In mouse motor nerve endings, two delayed rectifiers were distinguished, one fast and one slow. The fast component is responsible for action potential repolarization as evident from its blockade by external TEA, while the slow component may play a role in regulating transmitter release since its selective blockade by the snake venom dendrotoxin (DTX) leads to facilitation (217, 854). Action potentials are kept short by K channels in CNS nerve terminals as well. Thus, in rat calyx of Held nerve endings, which utilize glutamate as their transmitter, a fast-activating delayed rectifier was recorded by Forsythe (264). This K channel permits short action potentials and high firing rates and participates in producing prominent afterhyperpolarizations, and its block by 4-AP leads to potentiated release.

Some neuronal cells, with much longer action potentials than those characteristic of motor axons, also employ delayed rectifiers to aid repolarization, but use a different K channel subtype whose gating kinetics are much slower. Thus, in sensory neurons of Aplysia, K channels activate within 100-200 ms and inactivate within 0.5 s (422). Byrne and Kandel (117) summarized the contribution of such delayed rectifiers to spike broadening. Another relatively slow K channel, with an activation time of 100 ms, was recorded by Bennett and Ho (66) in chick calyciform nerve terminals. Finally, the patch-clamp technique was used to study posterior pituitary nerve endings of axons originating in the hypothalamus, and a delayed rectifier was found with a single-channel conductance of 27 pS (76). This channel, too, has relatively slow activation kinetics and is selectively blocked by DTX. Because of its slow activation, it cannot contribute to action potential repolarization; rather, this delayed rectifier may be activated during long bursts of activity known to occur in the hypothalamus-pituitary system. Opening of the K channel would raise the threshold for generating action potentials and hence limit secretion at the end of a long burst.

A channels are voltage-gated potassium channels giving rise to currents that have been termed fast transient, transient outward, early outward, and rapidly inactivating potassium currents. They were first observed in molluscan neurons by Connor and Stevens (170), who called them the A current. A channels function in different neurons by regulating firing rates, determining first spike latencies and in some cases contributing to action potential repolarization. Compared with other potassium channels, these channels activate and inactivate more rapidly. Their steady-state inactivation is usually almost complete near the resting potential, and most important, their activation threshold is lower than that of other potassium channels. Therefore, A channels function in the subthreshold range for generating action potentials, opening transiently upon small depolarizations that start from hyperpolarized potentials. Because of these characteristics, A channels are able to regulate repetitive firing at low rates, and they are prevalent in sensory cells that fire at rates that reflect stimulus intensity, in axons with low-frequency firing rates (169), and in other cells (733). After an action potential, the ensuing hyperpolarization removes A channel inactivation, and the transient outward current flowing through these channels slows down the return of the membrane potential toward action potential threshold so that the interval between the past action potential and the next one is prolonged. The upshot is that A channels serve as dampeners of interspike intervals.

A channels are presumed to be present in the presynaptic terminals of a number of sensory cells as well as in the nerve endings of CNS neurons, but to date, in most cases they have only been recorded in the soma of these cells. Among others, they were found in sensory neurons of Aplysia (422, 781) and guinea pig (398). In the latter, a single-channel conductance of 20 pS was found, and the channels inactivated with two time constants of 100 ms and 4 s. The slow inactivation component was suggested to provide the basis for building central delays into neuronal networks (186). In the CNS, a notable example of A channels was recorded by Dolly et al. (210) in guinea pig hippocampal slices, where an A current was specifically attenuated by DTX, leading to facilitation and spontaneous epileptiform activity in intact cell populations. It is also noteworthy that in cultured hippocampal neurons, an A current was found to be modulated by GABA_B agonists, which induced a positive shift in the voltage dependence of channel inactivation (738). If similar A channels were present in the nerve terminals, GABA_B agonists could make action potentials briefer by potentiating the A current and hence depressing transmitter release; at the resting potential, no change in potassium conductance would be expected, leaving the excitability of the nerve terminal intact.

Table 9 presents the only two prominent instances in which A channels were recorded directly from "native" nerve endings. Ironically, in these two cases, the classical functional role of the A channel is hardly evident. In the axon terminals of retina horizontal cells (which receive synaptic inputs from photoreceptors), A channels were detected and characterized, although their role remains unclear (296). A channels and A currents have also been detected in rat posterior pituitary nerve endings (76, 380, 382, 882). These channels are thought to participate in repolarizing the terminal after a spike. In addition, during high-frequency stimulation, reduction of the A current by inactivation may lead to action potential broadening.

Another noteworthy A channel was detected in fused synaptosomes prepared from the electric organ of Torpedo (228, 229). These pinched off cholinergic nerve terminals were patch-clamped in the cell-attached and inside-out configurations, and an A channel was recorded and found to have a single-channel conductance of 24 pS, a bursting activity and voltage-dependent activation, and inactivation kinetics that conferred upon it a "statistical memory" (114, 681) and oscillatory behavior (682). The channel may be important in frequency modulation phenomena such as facilitation, tetanic potentiation, and posttetanic potentiation.

Recently, evidence was obtained for the presence of an inactivating potassium current in the motor nerve terminals of the frog (553). Thus an A-type potassium channel may be directly involved in synaptic facilitation.

Delayed rectifier potassium currents with extremely slow activation kinetics were originally described in Purkinje fibers of the heart (597). The K_s channels underlying these currents were further characterized in atrial and ventricular myocytes (521, 790). It appears that the first demonstration of such a current in a neuron was in a nerve terminal. A novel slowly activating K_s channel was found in isolated rat pituitary nerve endings that are a part of magnocellular neurons (417). The activation time constant of the current was 4 s at 50 mV, decreasing to 700 ms at 40 mV, and the half-activation occurred at 37 mV. Magnocellular neurons secrete oxytocin following bursts of activity that are sustained by plateau currents that last for seconds (22). Repetitive stimulation for more than 5 s was found to lead to inactivation of excitability (99). A long-lasting potassium current with kinetics of seconds would be required for this phenomenon; indeed, a calcium-gated potassium current of suitable properties was found and suggested to be involved in terminating action potential bursts or regulating the duration of interburst intervals (943). The novel slowly activating K_s channel may be another potassium channel that contributes to the modulation of excitability and release, through its activation kinetics (395).

4.  Other potassium channels

A) K(CA) CHANNELS. The functional properties of calcium-gated potassium channels are described in section IV.

B) K(ATP) CHANNELS. Potassium channels that are voltage insensitive and shut by direct binding to cytosolic ATP are called K(ATP) channels. They were found in pancreatic -cells, various muscle cells, and several neurons (727). At physiological glucose levels, K(ATP) channels are shut, but a decrease in intracellular ATP concentration increases their open probability so that these channels hyperpolarize cells when energy reserves dwindle. Deist et al. (198) found that mouse motor nerve terminals probably contain such channels, since the antidiabetic drug sulfonylurea, which is a specific blocker of K(ATP) channels, reduces part of the potassium current in these terminals. Furthermore, bathing the terminals in glucose-free solution increases the size of the sulfonylurea-sensitive current. It was suggested that under hypoglycemic conditions, when the cytosolic ATP level decreases and the ATP-dependent pumps become less effective, the opening of K(ATP) channels becomes essential for maintaining the resting potential by preventing membrane depolarization. With normal glucose levels, on the other hand, these channels do not contribute to the resting membrane potential.

C) S CHANNELS. These potassium channels, which are shut by 5-HT, are only weakly sensitive to voltage if at all and provide a voltage-independent background conductance to Aplysia sensory neurons in which they are found. Although not directly detected in nerve terminals, Klein et al. (422) showed that in Aplysia sensory neurons, the S channels contribute substantially to the total outward current and that S current reduction causes action potential broadening, leading to presynaptic facilitation.

As noted by Hille (345), each excitable membrane uses a mix of several potassium channels to fulfill its functional needs. Table 10 presents the particular mixes employed by various nerve endings. The most conspicuous facts that emerge from Table 10 are the presence of K(Ca) channels in nearly all nerve terminals, the presence of at least two potassium channel types in most nerve endings, and the cohabitation of delayed rectifier and calcium-activated potassium channels at motor nerve terminals (see also Ref. 380).

The potassium channels employed by each particular terminal have different gating kinetics and sensitivities. These usually ensure action potential repolarization, a link between calcium-triggered secretion and cessation of release, and in many cases, the possibility for frequency modulation of transmitter liberation.

The pharmacological agents commonly used in the study of potassium channels are not very useful tools for distinguishing unambiguously among the various channel types. Tetraethylammonium and cesium ions, for example, block most potassium channels, at least to some degree. Nevertheless, certain agents, particularly toxins, appear to be relatively selective in their actions, although few instances of absolute specificity for one channel type may be noted (for general reviews of potassium channel pharmacology, see Refs. 171, 321, 555, 733).

Delayed-rectifier channels may usually be blocked by externally applied TEA and 4-AP. Tetraethylammonium is impermeant, and its blocking effects differ when applied externally or internally, suggesting two distinct binding sites (33, 345, 815). Indeed, extensive in vitro mutagenesis studies have identified the two separate sites and the K channel protein residues that participate in TEA binding (330, 502). A more specific K channel blocker is the scorpion peptide noxiustoxin, which was found to act on the delayed rectifier current in the squid axon (127).

In guinea pig hippocampal cells, Dolly et al. (210) demonstrated that the snake venom DTX blocked K channels. In nerve terminals, DTX was shown to block the D channel of posterior pituitary nerve endings (76) and the slower delayed rectifier component of the mouse motor nerve terminal (217). Other blockers of the two delayed rectifiers in mouse motor nerve terminals are guanidine and uranyl ions (854), although a later study reported that while uranyl ions do partially block the faster K channel, the slower delayed rectifier current is actually enhanced by these ions (147).

Fast transient A channels in nerve terminals are blocked, at least partially, by external application of TEA and millimolar concentrations of 4-AP (76, 296). No evidence for their block by DTX has been reported in nerve terminals, unlike that found in hippocampal neurons (316). Capsaicin, a plant toxin, blocks the A current of horizontal cell axon terminals at 20 mM (296).

Calcium-gated potassium channels of the SK type are usually TEA resistant and are blocked by the peptide apamin, a component of bee venom (112), at nanomolar concentrations. Less than 1 mM of externally applied TEA usually blocks BK channels as do nanomolar levels of the scorpion venom charybdotoxin (CTX) (547). The action of CTX was shown to be both voltage and state dependent (21). At the frog motor nerve terminal, CTX was found to block the K(Ca) channel at a nanomolar concentration (20). Recently, Harvey et al. (322) reported K(Ca) channel block in mouse motor nerve terminals by a series of CTX homologs and unveiled some of the critical peptide residues responsible for the block. The site of CTX binding and the particular amino acids participating in the binding have been identified in cloned potassium channels (500, 501). Further pharmacological details of K(Ca) channels are presented in section IVE.

The K(ATP) channel does have a specific blocker, namely, the antidiabetic drugs sulfonylureas (171). Indeed, on the basis of its effect on the outward potassium current, Deist et al. (198) proceeded to detect the K(ATP) channel in mouse motor nerve terminals.

S channels have, as their "natural blocker," the transmitter serotonin. These channels are turned off by cAMP-dependent protein kinase phosphorylation, resulting from cAMP level increases produced by 5-HT.

Thus, for example, in the squid nerve terminal, which is large enough to study with intracellular voltage clamping, a K channel and a K(Ca) channel were found. Augustine and Eckert (41) showed that the calcium-activated potassium current could be suppressed by blocking calcium current using cadmium or by external application of TEA, to which the delayed rectifier in this terminal was found to be resistant. Furthermore, the slow activation of the K(Ca) current during depolarization pulses, in combination with external TEA, enabled them to characterize the voltage properties and kinetics of the delayed rectifier. Augustine (42) showed further that by employing TEA and the aminopyridine DAP, which blocks both current components, the functions of the two channels in the terminal could be demonstrated.

Another prominent example is the nerve terminal of posterior pituitary axons. Here, patch clamping was successfully carried out by Bielefeldt et al. (76), and three different potassium channels were recorded: a K channel, an A channel, and an unusual CTX-resistant BK-type K(Ca) channel with single-channel conductances of 37, 32, and 134 pS, respectively. The unusual K(Ca) channel in this terminal was found to be extremely sensitive to calcium, activating during depolarizing pulses even at the low resting calcium levels. Consequently, its separation from the other two potassium current components in whole cell current measurements was not readily achieved by cadmium. Fortunately, in this terminal, the delayed rectifier could selectively be blocked by DTX. An additional separation problem arose from the fact that the BK current inactivation, although slower, was similar to the inactivation of the A current. Luckily, again, the currents could be separated on the basis of their activation properties: the A current activated within 1 ms by strong depolarizations, while the activation of the K(Ca) channel was much slower. In cell-attached patches, the individual activities of the three channel types could more readily be distinguished on the basis of their amplitudes and thresholds for activation.

The lack of a clear-cut means of distinguishing among K channel types does not only cause methodological problems. As pointed out by Rudy (733), it has also furnished considerable confusion in the literature; for example, A currents, often described as such in certain cells, turn out to be very similar to currents identified as delayed rectifiers in other cells. Molecular biology has begun to shed some light on the structural features of potassium channels, and the ongoing research in this area may alter our classical view of channel classification. There is growing evidence that rather than envisaging K channels and A channels as two discrete channel types, they should be viewed as placed on a continuum.

F.  Molecular Biology of Potassium Channels

1.  Genetic classification of potassium channels

Voltage-gated potassium channels are composed of integral membrane -subunits (388, 746) that form the channel pore, and these are sometimes associated with accessory cytoplasmic -subunits. Each -subunit contains six hydrophobic segments, S1S6, that are assumed to span the membrane, with a special pore-forming domain, called H5 or P, between S5 and S6, and a positively charged S4 segment that acts as a voltage sensor. The -subunits cloned so far appear to belong to two superfamilies that are denoted Kv and Kir for voltage-activated and inwardly rectifying, respectively (721). Only the former is of interest here, since just one case of detecting a Kir channel in nerve endings has so far been reported. Usually, four -subunits are required to form a functioning Kv potassium channel. The -subunits, when present, can modify the channel gating properties (531, 699).

On the basis of mutants exhibiting abnormal motor behavior, different genes encoding for potassium channel subunits were originally characterized in Drosophila (reviewed by Papazian et al., Ref. 632): Shaker, encoding for an -subunit a Kv channel; hyperkinetic (Hk), encoding for a Kv channel -subunit; ether-a-go-go (eag), for a different Kv channel -subunit; and slow-poke (slo), for a calcium-activated potassium channel subunit. Changes in A current were found to be associated in Drosophila with Shaker, changes in delayed rectifier current with eag, and changes in IK(Ca) with slow-poke. Subsequently, four separate Shaker-like genes, mapped to different chromosomal sites, were classified in Drosophila and were called Shaker, Shal, Shab, and Shaw (116). In Drosophila, expressed channels of Shaker inactivate rapidly, those of Shal less rapidly, while expressed channels of Shab exhibit little inactivation and those of Shaw virtually none.

The Shaker probes from Drosophila led the way to cloning-related potassium channels from other invertebrate and vertebrate neurons, and Shaker, Shal, Shab, and Shaw-like genes were classified in several animals (951). Today, it is recognized that each of the four members of the Drosophila gene family has one or more mammalian homologs and thus defines a subfamily (525, 745, 951). The sequence identity within the four genes in Drosophila is ~40%, whereas significantly higher identity (70%) is found among Drosophila Shaker and its homologs across the phyla. The same holds true for each of the Drosophila Shal, Shab, and Shaw genes and its homologs (745). The channels coded for by homologous genes in different animals may or may not share the same biophysical properties. Thus the Shab-related delayed rectifiers in Drosophila and mouse are very similar (626), whereas most mammalian channels of Shaker homologs inactivate slowly in contrast to the Drosophila Shaker coded channel, which acts as a rapidly inactivating A channel (951).

In Drosophila, alternative splicing of primary Shaker-related RNA transcripts is extensive, leading to great potassium channel diversity, whereas in vertebrates, diversity is created by mRNA expression from separate genes (663). The number of -subunit coassembly combinations is enormous, since both homomultimers and heteromultimers can be expressed. This great diversity may be useful for a vast number of cells, since it allows for the blending of multiple potassium channel types in a fine-tuned manner according to particular needs. As pointed out by several workers (663, 745), however, an unrestrained mixing of subunits would be undesirable for cells that require functionally independent and discrete potassium currents, and indeed, a molecular barrier to heteromeric association of -subunits of different subfamilies seems to exist (179).

The Shaker-Shal-Shab-Shaw genes place the A channels and K channels within one group of voltage-activated potassium channels with intermediate features. Furthermore, originally shown by Zagotta et al. (994), inactivation could be restored to a noninactivating mutant Shaker-related potassium channel in vitro, by adding to the mutant channel a synthetic peptide "ball" that blocked the internal channel mouth in a "ball-and-chain" mechanism. Later, Covarrubias et al. (180) showed that rapidly inactivating A channels can be converted into noninactivating delayed rectifier K channels by phosphorylation of the inactivation domain. In parallel, Rettig et al. (699) demonstrated how association of a Kv -subunit with -subunits conferred rapid inactivation on delayed rectifiers.

A Babel-like confusion afflicts the terminology of potassium channel genes. For example, virtually identical rat, mouse, and human cDNA clones of the Shaker subfamily have been called RCK1 (or RBK1), MBK1 (or MK1), and HK1, respectively, and similarly, the Shab subfamily-related genes in mouse and rat have been termed Mshab and DRK1, respectively. As cloning proceeded, "conversion tables" were needed, and several have indeed been put forward to make sense of the various voltage-gated potassium channel genes (386, 663). Table 11 presents the simplified nomenclature for gene subfamilies that was proposed by Chandy (144), which seems to be the one currently used by most investigators. Here, the Shaker-related genes are classified as Kv x,y, where x is a number denoting a subfamily and y a number denoting the specific member of the subfamily.

Chandy and Gutman (145) reviewed the six subfamilies Kv1 to Kv6, and recently new subfamilies have emerged (363). It was found that -subunits of the same subfamily, but not of different subfamilies, can coassemble in forming channels as summarized by Shen and Pfaffinger (771). It was also found that cloned -subunits assemble specifically with Kv1 -subunits (768).

A detailed structural model of voltage-gated potassium channels was proposed by Durrell and Guy (224), and many molecular biological studies have provided insights into the factors that determine channel properties such as ionic selectivity, voltage gating, kinetics of activation and inactivation, and binding of blockers (for review, see Ref. 663). Unfortunately, however, the pharmacological investigation of clonal potassium channels has not yielded simple selective tools for identifying particular potassium channel types in native cells.

Several investigators have combined in situ hybridization techniques and immunohistochemistry to follow the localization of potassium channels in mammalian brain. Site-specific polyclonal antibodies were prepared for the different -subunits of many members of the Kv subfamilies encoding for a variety of both A-type and K-type channels. Various channel subunits that were found to be directed to presynaptic nerve terminals in a number of brain structures are summarized in Table 12.

Subunit localization in nerve terminals and other subcellular areas vary from neuron to neuron, and no stereotypic patterns emerge from the data collected so far (918). The differential expression of Kv proteins, however, as well as their targeting to contrasting subcellular compartments and their particular colocalizations in different neurons, obviously contribute to channel mixing diversity, which may be tailored to specific functional requirements. Thus, for example, Sheng et al. (775) found that two A channels, assembled from Kv1.4 and Kv4.2 subunits, were segregated in vivo to different subcellular locations, with the former concentrated in axon terminals and the latter in somatodendritic domains of rat CNS neurons. They suggested that a Kv1.4 channel is involved in the regulation of presynaptic transmission, whereas the Kv4.2 channel regulates postsynaptic signals. Veh et al. (918) pointed out that the presynaptic colocalization of the A-type channel subunits Kv1.4 and Kv3.4 in particular hippocampal terminals may be related to the long-term potentiation known to occur at these presynaptic points, which is probably related to memory mechanisms (87). The localization of subunits that encode for delayed rectifier channels in various CNS terminals (e.g., Kv3.2 and many of the Kv1 subfamily members) most probably contributes to action potential repolarization and regulation of transmitter release (566, 918, 953).

Finally, it should be noted that only one Kv -subunit has so far been detected in brain nerve endings (704). This -subunit was found to be concentrated in both hippocampal perforant path terminals and in cerebellar basket cell terminals. Similarly, just one instance has been reported on a slo-related subunit of the maxi calcium-activated potassium channel in rat brain. In this case, the slo protein was found to be concentrated in terminal areas of several major projection tracts (424).

As mentioned in section IIIF, most potassium channels are members of either the voltage-gated superfamily Kv or the inward-rectifying superfamily Kir. The Kir protein subunits appear to have two membrane-spanning sections, M1 and M2, and a P domain, but they lack an S4-like voltage sensor (721). The Kir superfamily, which includes G protein-gated inward rectifiers and K(ATP) channels, may have arisen from a deletion of four of the six membrane-spanning segments of the more ancient Kv channels (160). Inward rectifying potassium channels are prevalent in skeletal muscle, heart cells, and central neurons, where they are important mediators of transmitter actions (345). There is no electrophysiological evidence for the presence of Kir channels in nerve terminals, but in one recent immunohistochemical study, Ponce et al. (662) reported that inward rectifier potassium channel proteins have been detected in rat brain nerve endings. The authors, who were surprised to find Kir channels in thalamocortical and hippocampal nerve terminals, suggested that the channels may mediate the actions of µ-opiate receptors, which are known to be present in these terminals.

	IV. CALCIUM-GATED POTASSIUM CURRENTS AND CHANNELS

Top Previous Next References

Calcium-gated potassium currents are consistently found at the presynaptic nerve terminals in almost every preparation studied, including neuromuscular junction of crayfish (88, 794), frog (20, 409), lizard (23, 473, 569), and mouse (20, 217, 507, 513, 890, 916); rat brain synaptosomes (54, 245, 598); cochlear nerve terminals (948); rat posterior pituitary (neurohypophysis) (73, 76, 214, 937, 940); avian ciliary ganglion (66); barnacle photoreceptor (327); and goldfish retinal bipolar cells (741). Based on their biophysical and pharmacological properties, they can be subdivided into fast and slow current [I_K(Ca)] (569; reviewed in Ref, 535). In general, fast I_K(Ca) (507, 535, 569) activate within milliseconds and account in part for the repolarizatory phase of the action potential. This current can also be selectively blocked by CTX (569). Slow I_K(Ca) (23, 66, 88, 794, 940) activates with a delay of several tens of milliseconds and accounts mainly for the hyperpolarizing afterpotential. This current in lizard neuromuscular junction can be selectively blocked by apamin (569). In rat neurohypophysis, it is insensitive to apamin (940).

Calcium-gated potassium channels (reviewed in Ref. 85) can be divided into two groups based on their single-channel conductance: BK channels and SK channels. Large-conductance calcium-gated potassium channels typically have single-channel conductance of >200 pS and are blocked by CTX and TEA (see e.g., Refs. 214, 245, 598, 948). Presynaptic SK channels generally have unitary conductance <100 pS and are less sensitive to TEA (245).

Both types of calcium-gated potassium channels are highly selective for potassium ions and their activity is determined by two factors: membrane potential and intracellular calcium levels. Calcium concentrations required for the activation of the channels range from 10 nM to 10 µM for different channels (73, 245, 598). Membrane depolarization is also required for activation. In rat brain synaptosomes, calcium-gated potassium channels open at 40 to 30 mV when calcium concentration is 10-100 µM (245, 598). In rat neurohypophysis, the channels open at a potential of 30 mV at calcium concentrations of 0.1 µM (73). The channels are voltage dependent, and the open probability increases e-fold per 8-15 mV of depolarization (73, 245, 598). Increase in intracellular calcium shifts the voltage dependence of activation to the negative side, i.e., less depolarization is required to activate calcium-gated potassium channels at higher calcium concentrations. This action of calcium ions is cooperative with a Hill coefficient of 1.6-1.8 (73, 598).

Single-channel properties of calcium-gated potassium channels at presynaptic nerve terminals are summarized in Table 13.

To understand the functions of calcium-gated potassium channels, it is helpful to look first at the location of these channels relative to the sites of transmitter release. Calcium-gated potassium channels have been demonstrated to be clustered at the active zones in the vicinity of calcium channels in hair cells (376, 714, 715), in presynaptic terminals of barnacle photoreceptor (830), and in frog neuromuscular junction (718, 720). This strategic location of the channels allows them to activate rapidly in response to the rise in calcium concentration (720) and thus makes them likely candidates to be involved in the local regulation of the duration of depolarization at the neurotransmitter release sites (718).

Blockade of calcium-gated potassium channels causes two notable changes in the shape of the action potential invading the nerve terminal. The first is a slight prolongation of the depolarization (88, 473, 569, 794), and the second is disappearance of the hyperpolarizing afterpotential several milliseconds after the action potential (66, 88, 217, 569, 794). The first effect is attributed to the fast calcium-gated potassium current, which activates almost immediately after depolarization, and the second to the slow current (569).

The role of calcium-gated potassium channels in transmitter release is not firmly established. An intriguing hypothesis has been proposed by Stanley and Ehrenstein (822) whereby calcium-gated potassium channels in synaptic vesicles open in response to the rise in intraterminal calcium concentration and allow potassium into the synaptic vesicle. The entry of potassium and concomitantly of water causes swelling of the synaptic vesicles and fusion with the plasma membrane. However, this does not seem to be the case (44, 1009).

Although it is conceivable that calcium-gated potassium channels may regulate the duration of the action potential invading the nerve terminal and, thus, limit calcium entry, studies on the effects of blockade of calcium-gated potassium channels on transmitter release are inconclusive. Some investigators report an increase in transmitter release when a selective blocker of calcium-gated potassium channels, CTX (513, 718, 719), or its homolog CTX-2 (513), is applied to the nerve terminal, whereas several other reports show that selective inhibition of calcium-gated potassium channels produces little effect on transmitter release (20, 473, 559). Furthermore, it is argued that the effect of CTX and CTX-2 is due to their action on other potassium channels rather than on calcium-gated potassium channels (513).

Calcium-gated potassium channels are likely to come into play during the repetitive activity of the nerve terminal. Both a significant increase in intraterminal calcium concentration (105, 382) and a rise in calcium-activated potassium current (327, 432) during repetitive stimulation have been demonstrated. Activation of the calcium-gated potassium conductance may cause hyperpolarization of the membrane, limit calcium entry, and subsequently reduce transmitter release. At least two observations support this hypothesis. The first is calcium-dependent hyperpolarization of the terminal membrane and action potential failures at high-frequency stimulation (73). The second is a decrease in paired pulse facilitation at crayfish neuromuscular junction when calcium-gated potassium channels are not blocked (795). Consequently, calcium-gated potassium channels have been hypothesized to be involved in termination of bursts of activity and regulation of duration of interburst intervals (940) as well as in the fatigue of secretion observed at the rat neurohypophysis (73). In auditory hair cells, which are stimulated at different frequencies in vivo, calcium-gated potassium channels have been proposed to be involved in frequency tuning (34, 361, 376, 715) and in protection against overstimulation (948).

To the best of our knowledge, gene or genes encoding the calcium-activated potassium channels at the presynaptic nerve terminals have not been cloned, and the structure of the channel has not yet been unraveled. However, there is some information about the molecular properties of the small calcium-gated potassium channels blocked by apamin (446, 724, 984).

In different tissue preparations, including brain synaptosomes, apamin binds a 250-kDa protein (448, 724). Affinity labeling with ¹²⁵I-apamine results in a covalent labeling of a single-chain polypeptide of ~30 kDa (448, 724). Thus the apamin receptor is postulated to be either a part of the calcium-gated potassium channel macromolecule (448) or a protein associated with calcium-gated potassium channel (807), and the calcium-gated potassium channel is proposed to be a large oligomeric structure containing a 30-kDa subunit (724). Sokol et al. (807) used antibodies against apamin-binding protein from bovine brain synaptosomes to clone a cDNA from a porcine vascular smooth muscle expression library. The gene encoding the apamin receptor (Kcal 1.8) encodes the protein of 438 amino acids, which has four potential transmembrane domains, one putative calcium binding site, a PKC phosphorylation site, and a leucine zipper motif. It does not possess any significant sequence homologies with known ion channels or receptors (807).

Mutations of Drosophila slow-poke (slo) locus, located on the third chromosome (40), specifically abolish calcium-gated potassium currents in adult and larval muscles and in larval neurons (236, 425, 739, 792). The molecular analysis of this locus provided information about the structural components of the calcium-gated potassium channels. The slo locus contains an open reading frame of 3,552 nucleotides, which encodes a protein of 1,184 amino acids. Hydropathy analysis of this protein reveals seven hydrophobic domains near the NH₂ terminal. This arrangement resembles the structure of other potassium channels (40). The fourth hydrophobic region (S4) and the region between the fifth and the sixth hydrophobic domains (H5) show sequence similarity with voltage-gated potassium channels (40). Other regions of the protein also show some degree of sequence similarity with voltage- and nucleotide-gated channels (387). Thus this putative calcium-gated potassium channel is thought to be a member of the superfamily that includes voltage-gated sodium, calcium, and potassium channels and second messenger-gated ion channels (387). Several cDNA derived from the slo locus can be found in the fly head cDNA library. These cDNA encode closely related proteins of ~1,200 amino acids (4). The RNA transcribed from the slo cDNA expressed in oocytes give rise to several functionally diverse calcium-gated potassium channels (4, 789), thus suggesting that alternative splicing of slo mRNA gives rise to a large family of calcium-gated potassium channels (4). Similar genes encoding calcium-gated potassium channels in mice (115) and humans (892) have also been cloned. It is noteworthy that the expression pattern of slo protein in rat brain is consistent with the targeting of the protein to the presynaptic nerve terminals (424).

Recently, another gene and protein associated with calcium-gated potassium channels (the -subunit) have been characterized (893). The gene encodes a polypeptide of 191 amino acids, which may contain 2 transmembrane domains. No evidence for alternative splicing of the gene product has been found. Coexpression of this protein with the -subunit results in a channel with increased calcium and voltage sensitivity. The function of the -subunit, its relationship to the channel molecule, and whether it is expressed to a significant extent at presynaptic terminals are unclear.

Similar to many intracellular processes, the activity of calcium-gated potassium channels can be modulated by phosphorylation and dephosphorylation. In rat posterior pituitary nerve terminals, macroscopic calcium-gated potassium current disappears in ~5 min when the ATP is omitted from the pipette filling solution. The open probability of calcium-gated potassium channels in excised patches is dependent on the ATP concentration at the intracellular side of the channels. Adenosine 5'-triphosphate prevents the decline (rundown) in the channels' activity and restores it after rundown occurs (75). Furthermore, addition of the catalytic subunits of the ATP-dependent protein kinase increases the activity of calcium-gated potassium channels from rat brain synaptosomes incorporated into lipid bilayers (245). Studies with a nonhydrolyzable ATP analog, nonspecific kinase inhibitor IQMP, and magnesium-free solutions show that ATP acts as a phosphate donor rather than a ligand directly affecting the channels (75). In excised patches, okadaic acid blocks the rundown at 50 nM, but not at 1 nM, thus indicating that the rundown is caused by dephosphorylation by phosphatase 1. The addition of GTPS to the cytoplasmic face of the excised patches accelerates the rundown in the absence of okadaic acid, suggesting that G protein is present in excised patches and may activate phosphatase 1 (75). Kinetic and statistical analyses of phosphorylation and dephosphorylation, as well as the voltage dependence of phosphorylation (74, 75), indicate that the calcium-gated potassium channel and its kinase form a macromolecular complex, whereas phosphatase is probably a separate protein that diffuses in a membrane and must encounter the channel for the dephosphorylation to occur (74).

Calcium-gated potassium channels play a role in action potential failures at high-frequency stimulation of the pituitary nerve terminals (73). Agents that interfere with phosphorylation and dephosphorylation influence action potential failures in a manner consistent with the enzymatic modulation of calcium-gated potassium channels, suggesting that this modulation may play a role under physiological conditions (75).

Calcium-gated potassium channels can be modified by various toxins, neurotransmitters, drugs, and ions. The substances affecting the calcium-gated potassium channels may be subdivided into two broad categories: 1) agents that inhibit or block calcium-gated potassium channels and 2) agents that increase the activity of the channels. Toxins and several drugs belong to the first category, alcohols to the second. Actions of various substances on calcium-gated potassium channels and currents are summarized in Table 14.

Tetraethylammonium specifically blocks BK and SK calcium-gated potassium channels when applied extracellularly or intracellularly (23, 76, 88, 245, 473, 569, 794, 890, 940, 948). Large-conductance calcium-gated potassium channels seem to be more sensitive to internal application of TEA than SK channels (245).

Several toxins have been used in the study of calcium-gated potassium channels. These include the scorpion toxins CTX (20, 73, 76, 88, 245, 409, 473, 513, 569, 741, 794, 916, 940, 948) and iberiotoxin (88, 916), the bee venom toxin apamin (20, 73, 217, 446, 569, 724, 940, 984), and the red tide toxin brevetoxin B (890). Kaliotoxin (322, 444, 725) may block presynaptic calcium-gated potassium channels; its effect, however, has not been definitely established.

Clinically used drugs that are known to block calcium-gated potassium channels include aminoglycoside antibiotics (598), antipsychotics (63), and the antimalarial drug quinine sulfate (54). Aminoglycoside antibiotics block calcium-gated potassium channels, when applied at the cytosolic side, in a dose- and voltage-dependent manner. The potency of different antibiotics depends on the number of positively charged amino residues, indicating electrostatic interactions between the agent and the receptor site at the channel (598). Extracellularly applied tetrandrine, a drug used in China for the treatment of hypertension and cardiac arrhythmias, blocks presynaptic calcium-gated potassium channels with high affinity (936, 937).

Alcohols such as ethanol, propanol, and butanol increase the activity of calcium-gated potassium channels (214, 984). Studies on excised patches show that ethanol alters the gating properties of calcium-gated potassium channels, causing them to spend more time in an open state (214). The fact that ethanol exerts its effect in excised patches suggests that it probably affects the channels directly, rather than via second messengers (214). Calcium-stimulated potassium current probably plays an important role in alcohol intoxication, since intraventricular injection of apamin inhibits ethanol-induced narcosis in mice (984).

Zinc ions have been reported to affect presynaptic calcium-gated potassium channels (552). At frog neuromuscular junction, zinc ions block voltage-dependent and calcium-activated potassium channels, when applied extracellularly (552). The effect of zinc is not specific, and prolonged exposure to the ion leads to an irreversible disruption of all ionic conductances of the terminal (552). Barium ions also block presynaptic calcium-gated potassium channels when applied to the extracellular or the intracellular side of the channels (23, 409, 507, 940).

F.  Pathophysiology of Potassium Channels

1.  Acquired neuromyotonia (Isaacs' syndrome)

Myotonia is a disorder in which sustained contraction of muscle is triggered by either a voluntary contraction or percussion. Its electrophysiological equivalent is the occurrence of repetitive discharges after activation of muscle, with these discharges then diminishing in amplitude or frequency. Acquired neuromyotonia (Isaacs' syndrome) is a disorder characterized by hyperexcitable motor nerves and sometimes by central abnormalities (320). There are several lines of evidence suggesting that this syndrome is due to an immunologically mediated injury causing hyperexcitability, since it responds well to plasmapheresis (373), azathioprine (706), methylprednisolone (231), phenytoin (146), and valproic acid (608).

In Isaacs' syndrome, a number of specific antibodies were detected to human brain voltage-gated potassium channels expressed in Xenopus oocytes (320), as well as to potassium channels of a neuronal cell line (PC-12) (31, 811). Thus Isaac's syndrome nerve hyperexcitability is probably the result of the immunological attack on the voltage-dependent potassium channels located along the motor nerve or at the nerve terminal.

	V. SODIUM CHANNELS IN NERVE TERMINALS

Top Previous Next References

The main function of presynaptic nerve terminals is to release transmitter upon the arrival of an appropriate signal, which in most cases is the action potential. In most terminals, the action potential is generated by voltage-dependent sodium channels. The number of transmitter quanta liberated from the presynaptic nerve ending depends on the amount of calcium ions entering the terminal. This in turn depends on the amplitude and duration of the depolarizing pulse, as shown in the pioneering articles of Katz and Miledi (401-403, 405). Here we deal only briefly with the presynaptic excitability of nerve terminals; other aspects are well described in the review by Jackson (380).

In addition to the primary role of generating the action potential, sodium channels also have a role in determining the intracellular sodium concentration ([Na⁺]_i), which in turn affects the [Ca²⁺]_i. The initial observations pointed to a paradox. If the amplitude and the duration of the action potential are of importance in determining the number of quanta liberated by the nerve impulse (m), then reduction of extracellular sodium concentration should cause a reduction in m. However, the opposite was observed (167, 677). The discovery of the sodium/calcium exchanger resolved this paradox. The activity of this protein is controlled in most cases by five parameters: the intracellular and extracellular concentrations of sodium and calcium as well as the membrane potential. Hence, sodium channels can affect the function of the nerve terminal by several of these parameters. It was found repeatedly that elevation of the [Na⁺]_i causes a marked increase in transmitter release (98, 129, 219, 246, 280, 284, 465, 494, 542, 558, 577, 602, 641, 684, 701, 708, 833, 1012). The accumulation of sodium ions (and the resulting accumulation of calcium ions) has been implicated in a number of pathophysiological conditions, including aging and cell death (77, 86, 122, 129, 218, 450, 455, 494, 503, 838, 919).

The third functional aspect of the sodium channels in the presynaptic nerve terminals is their role as targets of the action of toxins and drugs. Some of the toxins are among the most potent biological weapons. By blocking the sodium channels they abolish transmitter release from the nerve terminals and thus paralyze the prey.

The reader is referred to several recent reviews regarding the cellular and molecular biology of sodium channels (268, 269, 298, 586, 835, 836, 908).

Sodium channels were found in the nerve terminal of almost all the preparations studied. They were first demonstrated in the pioneering work of Katz and Miledi at the neuromuscular synapse of the frog and thereafter in many other species (23, 24, 168, 187, 188, 191, 308, 401, 403-405, 427, 535, 643, 734). Sodium channels were found in many other nerve terminals (49, 138, 162, 172, 246, 454, 457, 468, 518, 748).

Tetrodotoxin is the best-known toxin in this group. Found in pufferfish and porcupinefish (572), it is an alkaloid containing a guanidino moiety. Its binding to the sodium channel prevents the passage of ions. The blocking effect of tetrodotoxin at nerve terminals has been known for more than three decades (402, 403, 405, 596). Other toxins, such as saxitoxin, neosaxitoxin, and gonyautoxins, also block sodium channels (136). Further details on tetrodotoxin and saxitoxin binding and use can be found in a number of reviews (51, 136, 137, 831).

Another group of sodium channel-blocking toxins, the µ-conotoxins, was discovered in the venom of marine snails from the Conus family (183, 306, 530). µ-Conotoxins do not block sodium channels in frog or mammalian motor nerves or in rat brain (183), but they bind to sodium channels in electric eel electroplax (986) and Aplysia neurons (530). The µ-conotoxins bind to the same site on the sodium channel as tetrodotoxin and saxitoxin. Conotoxin GS isolated from the venom of Conus geographus is a toxin with similar binding properties to those of µ-conotoxins (987).

The first compound of natural origin found to affect sodium open channel characteristics is the plant alkaloid veratridine. The affected channels open at normal levels of membrane potential, but their inactivation is inhibited (135). Activation of sodium channels in synaptoneurosomes by veratridine can be blocked by preincubation with tetrodotoxin (371).

Tityus toxin, the purified venom of the Brazilian scorpion Tityus serrulatus, produces pre- and postsynaptic actions at the neuromuscular junction. Presynaptically, Tityus toxin increases the spontaneous release of transmitter (949). A neurotoxin from Tityus serrulatus, TsTX-V, was shown to delay the inactivation of sodium channels (28). The toxin TsTX-VI from the same scorpion produces a dose-dependent release of neurotransmitters, this effect being blocked by tetrodotoxin (749). Seven similar toxic peptides were isolated and sequenced from the venom of Tityus bahiensis and Tityus stigmurus (59). Tityus bahiensis toxin IV-5b selectively affects sodium channel inactivation. In contrast, sodium channel activation is negligibly delayed, and the deactivation is unaffected (888).

Batrachotoxin from the dart-poison frogs (Dendrobatidae) is an alkaloid, which acts by preventing voltage-dependent sodium channels from closing or inactivating (941). At the mammalian neuromuscular junction, batrachotoxin causes a transient increase in the spontaneous release of ACh, followed by a block of both spontaneous and evoked release (389).

The brevetoxins are lipid-soluble toxins acting on sodium channel receptor site 5 (887) and can be isolated from the red-tide dinoflagellate Ptychodiscus paresis. Brevetoxin-B has been shown to depolarize giant axons of squid (38) and nerve terminals (975). The depolarization is sodium dependent and can be blocked by tetrodotoxin (142).

Ciguatoxin enhances quantal transmitter release at the neuromuscular junction (556), probably due to the prolonged sodium channel opening in nerve membranes (120). Ciguatoxin-1b induces a tetrodotoxin-sensitive swelling of the nodes of Ranvier by modifying sodium currents and increasing intracellular sodium concentration (67).

A series of polypeptide toxins was isolated from the sea anemone Anemonia sulcata (68), the most abundant of which is ATxII (143). Similar toxins have been isolated from a number of other organisms (412, 449). ATxII has been shown to prolong nerve action potentials and to cause spontaneous and repetitive activity (723). The toxin ATxII causes depolarization (which can be prevented by tetrodotoxin) and repetitive firing of motor nerves by prolonging the inactivation phase of sodium currents (557, 723).

Another class of protein toxins affecting sodium channels has been discovered in the venom of the spider Agelenopsis aperta. Termed µ-agatoxins, they cause gradual and irreversible spastic paralysis in flies. This is probably due to an effect on neuronal sodium channels, which leads to repetitive action potentials originating in presynaptic axons and nerve terminals (3).

Most sodium channels activate rapidly and thereafter inactivate (348, 349). The inactivation is usually a rapid process and occurs in a time scale of milliseconds, unless toxins are present as described in a previous section. Recently, however, persistent sodium channels have been described in a number of cell types, and they may have a substantial physiological and pathophysiological importance (48, 182, 260, 329, 394, 618, 675, 837). They usually activate at more negative potentials than the "normal" sodium channels, and their activity persists for long periods. There is presently no direct evidence that these channels exist at the nerve terminals. If they do exist, they may affect the membrane potential of the terminal and thus affect tonic release of transmitters (see Ref. 634).

	VI. CHLORIDE CHANNELS IN NERVE TERMINALS

Top Previous Next References

Chloride channels are located both in the plasma and intracellular organelle membranes of animal cells. They play major roles in the regulation of intracellular pH (110), cell volume (350, 594), nerve excitability, and determination of the resting potential (12). Chloride channels were found in fused synaptic vesicles (980), mitochondria (81, 510), microsomes (441), and other intracellular organelles (10). This review deals only with chloride channels known to be present in the presynaptic terminals. Excellent reviews dealing with the structure and function of other chloride channels can be found elsewhere (390, 594).

Chloride is the most abundant extracellular physiological anion. In animal cells, the cytoplasmic chloride concentration is lower than the plasma concentration, and in most nerve cells, the chloride equilibrium potential (E_Cl) is near the resting potential. The contribution of chloride channels to the resting potential of the nerve terminal depends on their relative number and conductance. Chloride channels would be expected, when activated, to reduce normal excitability and help repolarize the cell during an action potential (345). In nerve terminals, chloride channels were found to be involved in presynaptic inhibition (222). Opening of these channels causes an increase in the total membrane conductance; this leads to decreased efficiency of subsequent inward currents to produce a depolarization and, hence, transmitter release (731). A mathematical model was developed suggesting that depolarization associated with the increase in chloride conductance (when the E_Cl is above the resting potential) may inactivate some calcium and sodium channels in the nerve terminal, contributing to presynaptic inhibition (303).

Chloride channels may also be involved in the determination of the size of the fraction of the nerve terminal arborization activated at any given time (934). Usually nerves branch extensively, and branches innervate many different postsynaptic targets. At the branch points, there is a lower safety factor of impulse conduction (694). By increasing the chloride conductance, the safety factor of impulse conduction is decreased even further. Thus an increase in chloride conductance (by activation of GABA_A receptors, see Ref. 1004) may determine which nerve terminals will be functional (230). It should be noted that not only chloride channels may serve this function (606).

At the surface membrane of the presynaptic nerve terminal, three chloride channel types were found: 1) ligand-gated chloride channels composed of GABA-gated chloride channels (490, 1004; for reviews, see Refs. 392, 496, 963) and glycine-gated chloride channels (932) (for a general review, see Ref. 688); 2) calcium-dependent or calcium-activated chloride channels (619) (for a general review, see Ref. 763); and 3) voltage-dependent chloride channels in fused synaptosomes (227).

The ligand-gated chloride channels belong to a large family of channel molecules that include also the nicotinic ACh (397), 5-HT₃ (512), glycine, (436), and GABA (588) receptor channels (70). The functional channels are composed of five homologous membrane-spanning subunits that form ion-conducting channels across the membrane. They contain ligand-binding sites in the extracellular domains and modulatory sites in the cytoplasmic domains of the pentamer. The deduced amino acid sequences show significant homology within the members of the family (20-30%). Moreover, the general structural characteristics of the subunits are very similar.

-Aminobutyric acid is an important inhibitory neurotransmitter in the CNS (94). Its role was first described in the invertebrate nervous system (377, 625). In the CNS, it can modulate synaptic transmission by the activation of three classes of receptors: GABA_A, GABA_B, and GABA_C. Of those, only GABA_A and GABA_C are ion channels.

A) GABA_A RECEPTOR-CHANNEL COMPLEX. I) functional properties and localization. Binding of an agonist to the GABA_A receptor-channel complex leads to the opening of an anion channel permeable to chloride and to a lesser extent to bicarbonate (94). In presynaptic nerve terminals, two mechanisms were proposed for the inhibition of transmitter release by activation of chloride channels (732): retardation in the propagation of action potentials into the nerve terminal arborization (934) and a decrease in the amount of transmitter released by each action potential which succeeds in reaching the terminal. -Aminobutyric acid-mediated chloride currents may produce both forms of inhibition by shunting membrane currents and by reducing the probability of opening of other ion channels (380).

B) GABA_C RECEPTOR-CHANNEL COMPLEX. I) Functional properties. Although GABA_C receptor-channel complex was first described in 1975 (393), its classification as a GABA channel subtype was confirmed only in recent years. The channel and pharmacological properties differ from those of the GABA_A family of channels (93, 392, 952). The GABA_C channels were found in presynaptic nerve terminals in vertebrate retina (491) and seem to be involved in visual information processing by the modulation of transmitter release from bipolar cells (490, 672, 974, 1002; for review, see Ref. 490). The GABA_C channels were also found in rat anterior pituitary (97) and brain (391) and lobster thoracic ganglia (378).

It has been known for more than three decades that glycine is present in the CNS and can serve as an inhibitory transmitter (193, 954). Several lines of evidence support the existence of glycine-activated chloride channels at presynaptic nerve terminals. Glycine was shown to inhibit potassium-evoked transmitter release in cultured cerebellar granule cells (932). Glycine-gated chloride channels have been demonstrated in synaptoneurosomes (239). The application of glycine induced the influx of chloride, which was inhibited by the glycine postsynaptic inhibitor strychnine. The effects of strychnine and picrotoxin, a specific inhibitor of GABA_A channel, were additive, indicating the coexistence of the two channels.

Calcium-activated or -dependent chloride currents were shown to be involved in the regulation of cell volume in a wide variety of cells (for review, see Ref. 351). In neurons, the activation of these channels may account for the decrease in axon excitability, thereby decreasing transmitter release and inhibiting repetitive firing. In presynaptic terminal of goldfish retinal bipolar cells, a tail current following depolarization was identified as a calcium-dependent chloride current, since it was suppressed by the intracellular application of the calcium chelator 1,2-bis(2-aminophenoxy)ethane-N,N,N',N'-tetraacetic acid and its reversal potential was very close to the equilibrium potential of chloride (619). In a detailed characterization of calcium-activated chloride channels in embryonic cultured spinal neurons from Xenopus, two kinds of calcium-activated chloride channels were described in the soma: the maxi- and the mini-chloride channels (365). The maxi-channels show numerous subconductance states, voltage-dependent inactivation above and below ±20 mV, mean conductance of 310 pS, and a reversal potential close to E_Cl. The mini-channels show conductance ranging from 42 to 69 pS, and they are not voltage inactivated. It is suggested that the activation of these channels is mediated by a second messenger cascade activated initially by the elevation of intracellular calcium.

A voltage-dependent chloride channel was found in fused synaptosomes of the Torpedo electric organ. The mean conductance of this channel is 11.7 pS. The open probability of the channel increases with depolarization and acidity (227). Among the channels found in this preparation, the chloride channel is the second most abundant, with the first being the bursting potassium channel.

	VII. PRESYNAPTIC LIGAND-GATED CHANNELS

Top Previous Next References

Receptors for neurotransmitters are not found exclusively on the postsynaptic membrane, where they mediate the response of the postsynaptic cell. Presynaptic nerve endings also possess a variety of receptors that are activated by the transmitter released from the same or adjacent terminals. The main function of these receptors is modulation of transmitter release and, consequently, modification of synaptic efficacy (for review, see Ref. 528).

Several criteria must be met to demonstrate clearly that a presynaptic receptor is involved in modulation of transmitter release. First, the receptor must be present at the presynaptic site. Obviously, the endogenous ligand needs to be present in the vicinity of the receptor to activate it. Furthermore, activation of the receptor must alter the release of neurotransmitter, and finally, this activation must be necessary and sufficient to account for the changes in transmitter release (528).

Various preparations and methods are being used in the study of presynaptic receptors. Anatomic localization of the receptors in different parts of the nervous system can be determined with the use of radiolabeled ligand or receptor-specific antibodies (291, 644). These techniques, however, do not always provide a resolution high enough to distinguish between presynaptic and postsynaptic structures.

The activity of ionotropic receptors can be assayed by measuring changes in [Ca²⁺]_i (130, 132) and ionic fluxes (466, 563, 564, 849) through the plasma membrane during pharmacological modulation of the receptors. With this approach, one cannot definitely rule out the effect of a retrograde messenger, diffusing from the postsynaptic site, unless the experiments are performed on the isolated nerve endings. In suitable preparations, electrical currents passing through the plasma membrane, following activation of the receptors, can be directly measured. This is feasible in large presynaptic terminals such as rat superior olivary complex (calyxes of Held) (264), chick ciliary ganglion (827), secretory endings in neurohypophysis (380, 756, 1004-1007), cones in tiger salamander retina (652), or terminals of the goldfish bipolar neurons (47, 522).

Functions of the presynaptic receptors are usually assessed by the effect of a pharmacological activation or inhibition of the receptors on transmitter release. The release of neurotransmitter is usually estimated directly by measuring release of radiolabeled transmitter from isolated presynaptic terminals, referred to as synaptosomes (158, 261, 808), or in intact preparation (740, 869, 1008). Changes in neurotransmitter release can also be estimated indirectly from measurement of the postsynaptic response (52, 554, 645, 755, 772). In that case, presynaptic actions of applied pharmacological agents must be separated from their postsynaptic effects (e.g., alteration of the sensitivity of postsynaptic receptors). This is usually achieved by statistical analyses of the postsynaptic currents (295, 554, 645; see also Ref. 242).

Presynaptic receptors for virtually all neurotransmitters are found in the CNS (881). These can be classified as metabotropic and ionotropic receptors. Activation of metabotropic receptors triggers second messenger cascades, whereas ionotropic receptors are coupled to ion channels. We concentrate here on the latter category of receptors.

Ionotropic receptors to three neurotransmitters have been identified on the presynaptic nerve terminals: GABA receptors, glutamate receptors, and nAChR. An additional class of receptors, purinergic P₂ receptors, has also been demonstrated.

Neuronal nAChR can be subdivided into two classes, based on their affinity to nicotine and to muscle nicotinic receptor antagonist -bungarotoxin (-BTX). One class binds nicotine with a high affinity and the other one has a higher affinity to -BTX (528).

Neuronal receptors share many properties with their muscle siblings (originally described by Sakmann and co-workers, Refs. 590, 742). Both types are nonselective ion channels with similar electrophysiological characteristics (603, 828). These similarities suggest that two kinds of receptors may function in an analogous fashion.

Nicotinic receptors are widely distributed in the central and peripheral nervous systems. In the CNS they are not limited exclusively to cholinergic nerve endings but are also found on glutamatergic, dopaminergic, GABAergic, and serotoninergic terminals. There is a strong evidence of the existence of presynaptic nicotinic receptors in rat striatum (195, 413, 690, 730, 744, 759), hippocampus (163, 203, 333, 873), hypothalamus (620, 621, 759), and cortex (344, 841). Presynaptic nicotinic receptors have also been demonstrated in goldfish retina (1011), mouse thalamus (459), and human cortical slices (599). In addition, axons in chick optic tectum may contain presynaptic nicotinic receptors (851).

In the peripheral nervous system, presynaptic nicotinic receptors have been demonstrated at neuromuscular junction (phrenic nerve terminals) of rat (177, 955, 957), mouse (420, 896, 925), and guinea pig (955); buccal ganglion of Aplysia (265); and nerve terminals in rat sublingual acini (1000).

A) MOLECULAR BIOLOGY AND STRUCTURE. Like their postsynaptic counterparts, the neuronal nicotinic receptors are proposed to consist of five subunits. These are mainly - and -subunits, with a ratio of 2 to 3 (173). Receptors composed of three different types of subunits, as well as homo-oligomers composed of only -subunits, are also possible (301, 920). Eight -subunits (_2-9) and three -subunits (_2-5) have been characterized so far in different tissues and species (527, 751); human genes encoding _3,4,5,7- and _2-4-subunits have also been cloned (309). Thus the family of the neuronal nicotinic receptors seems to be quite diverse. With possible combinations of the subunits taken into account, the number of different receptors that may exist in the nervous system can reach thousands (722).

B) MODULATION. Various substances have been found to modulate the activity of the presynaptic nicotinic receptors. Both inhibitors and activators of the receptors have been described and are summarized in Table 16.

C) POSSIBLE PHYSIOLOGICAL ROLES. Unlike postsynaptic receptors, presynaptic nicotinic receptors most likely serve as modulators of transmitter release. Activation of several tens of receptors could significantly depolarize the terminal membrane and cause a significant change in intraterminal calcium concentration (828). Thus it seems likely that their main function both in central and peripheral nervous systems would be augmentation of transmitter release.

Glutamate activates three different classes of ionotropic receptors (reviewed in Ref. 285), each having a distinct pharmacology and physiology. The receptors are named according to the agonists that elicit specific physiological responses: N-methyl-D-aspartate (NMDA) receptors, -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors, and kainate receptors. In addition, glutamate can act on metabotropic receptors coupled to G proteins (307).

There are anatomic and physiological data supporting the existence of all three classes of presynaptic glutamate receptors in the CNS. Immunocytochemistry techniques have been used to demonstrate AMPA receptors on the nerve terminals in rat retina (644) and in hypothalamohypophysial tract in rat and monkey (291). Electrophysiological studies have shown that glutamate opens chloride channels in cones of tiger salamander and that this effect of glutamate is limited to presynaptic terminals (652, 750). Presynaptic kainate receptors in rat hippocampus have been shown to possess a dual action. Activation of the receptors causes a transient increase in transmitter release, followed by a depression of glutamatergic transmission (158). Activation of NMDA receptors in guinea pig cortex synaptosomes causes an increase in the release of glutamate and norepinephrine (562). Interestingly, this effect is decreased by inhibitors of nitric oxide synthase (562). This is not entirely surprising, however, since it is known that nitric oxide is necessary to sustain opening of the NMDA receptor ion channel (9). In rat hippocampus, AMPA and NMDA receptors probably coexist and increase norepinephrine release (687). The cooperation of these two receptors seems to be necessary for their proper function (658). In cultured chick retina cells, all three types are likely to coexist, since agonists to each type of ionotropic glutamate receptors increase GABA release (130).

A) MOLECULAR BIOLOGY OF A PRESYNAPTIC GLUTAMATE RECEPTOR. Similar to other ligand-gated ion channels, glutamate receptors are heteropentamers. The exact subunit composition is not completely known (for review, see Ref. 584). Recently, the transmembrane topology of the glutamate receptor subunit was elucidated. The subunit is a 439-amino acid polypeptide with a molecular mass of ~48 kDa (969). The NH₂ terminus is likely to be extracellular, and the COOH terminus is intracellular. The polypeptide contains three transmembrane domains M1, M3, and M4 (64, 969). The M2 domain forms a hairpin loop lining the intracellular pore of the channel. This configuration is similar to that of potassium channels (64, 971), suggesting the possibility of a common origin of these two superfamilies of ion channels (971).

B) PHARMACOLOGY OF PRESYNAPTIC GLUTAMATE RECEPTORS. The pharmacological properties of presynaptic glutamate receptors are studied with the agents known to affect postsynaptic receptors. We failed to find substantial differences between the pharmacology of post- and presynaptic receptors. The table in Reference 307 summarizes the pharmacological modulation of glutamate receptors.

C) POSSIBLE PHYSIOLOGICAL ROLES. Both facilitatory and inhibitory functions have been attributed to presynaptic glutamate receptors. The role played by a specific receptor is likely to depend on the species and on the location of that receptor in the CNS. The NMDA receptors have been demonstrated to increase release of ACh in mouse brain slices (1008), norepinephrine in rat hippocampus synaptosomes (687) and slices (128), and norepinephrine and glutamate in guinea pig cortex synaptosomes (562). Of special interest is the possible involvement of NMDA receptors in long-term potentiation. Increased amounts of mRNA encoding a putative presynaptic NMDA receptors are detected for at least 5 h after induction of long-term potentiation in rat hippocampus (800).

The AMPA receptors cause an increase in the release of norepinephrine in rat hippocampal synaptosomes (659) and GABA in chick retina cells in culture (130).

Kainate receptors in rat hippocampus seem to play a dual role in the regulation of transmitter release. The initial effect of their activation is a transient increase in the release of glutamate, which is followed by a depression of neurotransmission (158).

In cones of tiger salamander retina, glutamate receptors cause a hyperpolarizing current (652), thus reducing the excitability of cells. Opposite effects are elicited by glutamate receptors in chick retina (130).

There is a growing body of evidence that ATP and other nucleotides can function as extracellular signaling molecules. They act on a large and diverse family of P₂ purinoceptors, which can be divided into two distinct classes: ligand-gated ion channels (P_2x receptors) and G protein-coupled receptors (P_2y, P_2u, P_2t, and P_2d receptors) (155).

In rat hippocampal slices, the P_2x receptor antagonist pyridoxal-phosphate-6-azophenyl-2',4'-disulfonic acid (PPADS) reduces the amplitude of evoked EPSC. Inasmuch as PPADS did not reduce the amplitude of the postsynaptic currents elicited by direct application of glutamate, the presynaptic site of action for PPADS has been proposed (574). P_2x receptors have been demonstrated in chick ciliary ganglion by Sun and Stanley (845). The P_2x channels exhibited a conductance of 17 pS and were blocked by suramin. In dorsal root ganglion neurons, the presynaptic ATP P_2x receptors cause the release of a neurotransmitter (dopamine) (312). They not only modulate transmitter release but may also initiate sensory signals in CNS without peripheral input. This interesting observation makes presynaptic P_2x receptors a possible target for pain therapy.

Seven different types of 5-HT receptors have been described (see Ref. 314). Most of them are coupled to G proteins, and one of them, the 5-HT₃ receptor, is a cation channel. The electrophysiological characteristics of the channels located at cell bodies have been studied quite extensively. Several receptors have been described in different preparations, ranging from 0.31 to 16.5 pS (for a review, see Refs. 383, 650). Presynaptic localization of the 5-HT₃ receptors has been demonstrated in various locations in the nervous system in different preparations. They were found in rat cortex (181), rat striatum (592), and human cortex (523). They have also been demonstrated on nonsynaptic cholecystokinin-releasing terminals in rat cortex and nucleus accumbens (640).

The pharmacological agents used to investigate presynaptic 5-HT₃ receptors are those known to affect the receptors at other locations. We have failed to find substantial evidence of the agents specific for the presynaptic receptors. The classical agonists include 5-HT and its derivative 2-methyl-5-HT. Antagonists include medications used in the treatment of nausea and vomiting, such as tozacopride, ondasetron, tropisetron, dolasetron, and granisetron (for review, see Ref. 314). Newer agents include agonists 1-phenylbiguanide (523, 640) and antagonists (3-tropanyl)-1H-indole-3-carboxylic acid ester and MDL-72222 (640).

Similar to other ligand-gated ion channels, 5-HT₃ receptors are oligomers that are probably composed of five subunits. The exact composition of the receptors is not yet completely understood, but there is evidence supporting existence of both homo- and heteromers (366). The receptor subunit is probably a polypeptide of 487 amino acids with a molecular mass of ~56 kDa. Hydropathy analysis predicts four transmembrane domains named M1 through M4. The sequence of the subunit exhibits some homology with nicotinic, GABA_A, and glycine receptors (511).

The function of the presynaptic 5-HT₃ receptors is not clear. These ligand-gated ion channels are likely to act as heteroreceptors in the modulation of transmitter release. In some preparations, pharmacological activation of the receptors can cause an increase in intracellular calcium levels (592) and the release of cholecystokinin (640), whereas in others, 5-HT₃ receptors mediate inhibition of transmitter release (523). The extent and physiological significance of this modulation are yet to be established.

The ATP-gated potassium channels are inhibited by millimolar levels of intracellular ATP (35) and therefore open when the ATP concentration decreases below a certain limit. Thus they provide the means of coupling the electrical activity of the cell to its energy content. These channels were demonstrated originally in pancreatic -cells (726), where they play a key role in the regulation of insulin secretion (for review, see Refs. 103, 889). They have been demonstrated in other cells as well, including cardiac myocytes, skeletal and smooth muscle cells, and neurons (for review, see Refs. 447, 631). Presynaptic ATP-gated potassium channels have been demonstrated in noradrenergic nerve terminals of rat cortex (452, 453, 865), in GABAergic terminals of rat substantia nigra (816, 950, 991), and in cholinergic mouse motor nerve endings (205). In fused presynaptic terminals from rat motor cortex, one ATP-gated potassium channel has been characterized with the use of patch-clamp technique (452). This channel has a conductance of 52.5 pS and demonstrates an inward rectification at voltages positive to 20 mV. Intracellular ATP at the concentration of 1 mM is sufficient to inhibit the activity of this channel. The channel's open probability is independent of intracellular calcium, and it is blocked by 100 µM tolbutamide.

Several agents have been found to modulate the behavior of ATP-gated potassium channels. The activity of the channels is inhibited by TEA in high concentrations (205) and by sulfonylureas such as glibenclamide and related agents (205, 288, 452, 866, 950, 991). The activators of the channels include cromakalim and its L-enantiomer lemakalim, a benzothiadiazine compound diazoxide, and its pyridinic analogs (288, 614, 657, 816, 865, 991). Experimental evidence has shown that the CNS possesses an endogenous inhibitor of ATP-gated potassium channels. This molecule, called endosulfine, has been isolated from porcine (924) and ovine (923) brain.

Knowledge about the molecular biology of presynaptic ATP-gated potassium channels is based on studies of the gene encoding a similar channel in rat heart. This gene may encode presynaptic channels as well, since in situ PCR demonstrates its mRNA in multiple areas of peripheral nervous system and CNS. The putative ATP-gated potassium channel (or its subunit) is a 417-amino acid peptide that contains two hydrophobic, most likely transmembrane segments. These segments flank a putative pore region, homologous to other potassium channel pore sequences. The protein is proposed to be oriented so that the NH₂ and COOH termini are on the cytosolic side of the membrane. Electrophysiological characterization of channels expressed in human or hamster epithelial cells demonstrates that several properties of cloned channels are similar to those of native channels. The channel exhibits bursts of activity, interspersed among long silent periods, and has a conductance of 70.2 ± 20 pS. Single-channel current shows inward rectification even in the absence of internal magnesium, which becomes more prominent with the increase in magnesium concentration on the intracellular side of the channel. The open probability of the channel shows no significant dependence on membrane potential. The activity of the channel is inhibited by intracellular ATP, with a half-maximal inhibitory concentration of <100 µM. On the other hand, the pharmacology of the cloned ATP-gated potassium channel is different from that of the native channels. The channel is activated by pinacidil (30 µM); however, the application of sulfonyureas has no effect on the open probability of the channel. One possible explanation for this fact is that the sulfonylurea receptor is not a part of the channel molecule, but rather represents a separate molecular entity (36).

The ATP-gated potassium channels are likely to be involved in maintaining the membrane potential and the reduction of transmitter release when the energy content of the nerve terminal is low, namely, during ischemia, hypoxia, and hypoglycemia. Several lines of evidence support this hypothesis. Hypoxia has been demonstrated to cause a decrease of evoked norepinephrine release from rat cortical (866) and substantia nigra (950) slices. This decrease is abolished by the application of sulfonylureas. Moreover, hypoxia also increases the ⁸⁶Rb efflux. Linear correlation is observed between the evoked norepinephrine release, intracellular ATP concentration, and ⁸⁶Rb efflux, suggesting the role of ATP-gated potassium channels in reduction of transmitter release (866). The sulfonylureas tolbutamide and glibenclamide block fast potassium component of presynaptic current and enhance the release of norepinephrine in rat cortical slices. These effects are more prominent under hypoxic and glucose-free conditions (205, 865). Because the activation of ATP-gated potassium channels during metabolic stress causes an outward current that would tend to hyperpolarize the terminal plasma membrane, these channels may play a neuroprotective role during anoxia and ischemia. Indeed, pharmacological activation of the channels has been demonstrated to reduce anoxic depolarization in hippocampal neurons, probably by means of reduction of glutamate release (62).

The role of the ATP-gated potassium channels under normal metabolic conditions is less clear. In mouse motor nerve endings, ATP-gated potassium current contributes a little (~8%) to the total potassium current in the nerve terminal (205). However, in rat substantia nigra slices, the application of glucose or sulfonylureas induces an increase in release under normal conditions (18). This finding suggests that some of the ATP-gated potassium channels are active and therefore may play a role in the regulation of transmitter release, under normal conditions as well.

	VIII. OTHER CHANNELS

Top Previous Next References

A large, nonselective channel has been described in fused synaptosomes from the Torpedo electric organ (536). This channel is highly voltage dependent, has a unitary conductance of 850 ± 18 pS, and does not distinguish between anions and cations. It is permeable to calcium ions and thus may serve as a route of calcium entry into the nerve terminal. This channel is discussed in section IIH.

Two nonselective and voltage-independent ion-gated channels have been identified in the peptidergic nerve terminals of the crustacean neurohemal organ, the sinus gland (457, 834). Both channels do not distinguish between sodium and potassium. One has a unitary conductance of 69 pS and is activated by intracellular sodium (834); the other channel has a conductance of 213 pS and is activated by [Ca²⁺]_i above 1 µM (457). A different, voltage-dependent, and calcium-activated nonselective channel was found in insect ganglia synaptosomes fused with liposomes (870). The properties of the nonselective channels are summarized in Table 17.

The physiological role of these nonselective channels is largely unclear. It has been proposed that the channels participate in sustaining prolonged activation of the nerve terminal during the burst activity. Activation of the channels evokes changes in intracellular ion concentrations, which reinforce further openings of the channels. This may help the nerve terminal to stay depolarized during the period of burst of activity (834). Although this hypothesis seems appealing, experimental evidence to support it is insufficient.

It is well known from the early work of Fatt and Katz in 1952 (247) that a stretch of the nerve muscle synapse produces an increase in spontaneous quantal transmitter release. In many cells, including secretory cells, it was found that stretch-activated ion channels of low selectivity exist (100, 370, 737). This prompted the suggestion that stretch-activated ion channels couple the length of the terminal, with its ability to release transmitter (154). However, the situation may be more complicated and may involve the participation of the cytoskeleton (153).

	IX. CHANNELS IN SYNAPTIC VESICLES

Top Previous Next References

In recent years, ion channels were discovered in the membranes of many intracellular organelles, where they play a crucial role in numerous cellular functions. Thus intracellular ion channels were found in skeletal and heart muscles (176, 279, 965), where they take part in the excitation-contraction coupling, and in the kidney (5), bone (82, 754), taste buds (472), and many other cell types (10, 11, 347, 565). In this section, we briefly describe the ion channels in one specific type of intracellular organelle, namely, synaptic vesicles in the presynaptic nerve terminals.

The main problem in the study of ion channels in the membrane of the synaptic vesicle is their small size. The typical size of a patch pipette is a little less than 1 µm (see Ref. 743). Because most synaptic vesicles have a diameter of 50-90 nm (see Ref. 1010), it is impossible with the current technology to create a seal between a single synaptic vesicle and the patch pipette. Two main methods were used to overcome this difficulty: fusion of many synaptic vesicles into a large structure amenable to patch-clamp recording (202, 680) and incorporation of synaptic vesicles into planar lipid membranes (972). Each of these methods permits the study of ion channels but has obvious disadvantages.

The disadvantages of the fusion method are the usage of fusogenic substances (such as polyethylene glycol 1500 and DMSO) and the marked change in curvature. In addition, the intravesicular content is largely diluted during the fusion procedure.

The disadvantages of the incorporation method are the necessity to incorporate the vesicle into a large membrane of usually foreign lipids and the large increase in the curvature and the surface area. Because membrane capacitance is proportional to the surface area, the time constant of the membrane is large. Hence, fast transients are frequently concealed during the recording procedure.

Ion channels were found in synaptic vesicles isolated from nerve endings of the Torpedo electric organ (202, 410, 602, 678, 679, 973, 980) and in vesicles isolated from neurosecretory endings of the hypophysis (451). They were found also in the membranes of other secretory cells (32, 823).

Six different types of ion channels have been identified to date (see Ref. 973). For one type of channel, the molecular structure is known at present, and it corresponds to synaptophysin (878).

Before speculating about the physiological roles of the ion channels in synaptic vesicles, it is important to note the permeability properties of these ion channels. Most of the ion channels described in synaptic vesicles are nonselective in nature and allow for the transport of different ions (602, 678, 878, 973, 980). In addition, chloride channels were found in synaptic vesicles (10, 980). Some of the channels are voltage dependent (980), and the channel at the neurohypophysial endings is calcium dependent (451).

Synaptic vesicles have a complex life cycle, which includes synthesis, filling with transmitter (or transmitters), transport to the release part of the nerve terminal, docking at the active zone, activation, fusion with the surface membrane, release of the stored transmitter, retrieval of the vesicle into the cytoplasm, and refilling with transmitters. One can envisage a role of ion channels in almost every stage in the life cycle of the synaptic vesicle (248, 255, 339, 340, 636, 678, 683, 717, 760, 814, 839, 913, 926, 931, 1009, 1010).

Transmitter is released from the nerve terminals as preformed, multimolecular quanta (200, 247; for summary, see Refs. 60, 400, 428, 915). After it was found that the presynaptic nerve terminals possess vesicles (716), it was proposed that the vesicles form the structural basis for quantal release (201). Many observations over the last four decades confirm this hypothesis (see Refs. 139, 140, 337-339). However, the vesicle hypothesis for secretion is not devoid of problems, as summarized recently (683). The three main problems are the hyperosmolarity of the contents of the synaptic vesicle, the slow rate of release as measured by amperometry in mast cells, and the lack of complete fusion of the vesicle membrane with the surface membrane during quantal transmitter release (16, 159, 560, 589, 607, 812, 929).

In the past, it was proposed (911, 912) that charged transmitter molecules are stored in the synaptic vesicles in a bound form, bound to an ion-exchange matrix. For positively charged transmitters, such as ACh and 5-HT, the ion-exchange matrix has a negative charge. Proteoglycans and other charged polymers that were found in the synaptic vesicle (416, 434, 585, 813, 814) can subserve this role. Hence, there is a "hunger for cations" in the releasing synaptic vesicle. The positively charged transmitter molecules have to be replaced with cations to be released through the fusion pore. If this is the case, then the nonselective ion channels in the vesicle membrane may have a crucial role in quantal transmitter release. During the fusion process, the membrane of the vesicle changes its membrane potential according to the potential of the surface membrane. If the voltage change is in the range of the activation of the nonselective ion channels, then they open and cations enter the vesicle, release the transmitter from the ion-exchange matrix, and make it available for release through the fusion pore. Such a mechanism will have a limited value if a complete fusion occurs between the vesicle membrane and the surface membrane but will be of great importance if a temporary fusion, nicknamed "kiss and run" (255), happens. If such a sequence of events does take place, then a postfusion control of transmitter release is possible. Many questions have to be clarified, however, before the generality of the postfusion hypothesis can be accepted, such as the similarities and the differences in the fusion machinery at different secreting sites, the existence of ion exchange molecules at synapses releasing negatively charged transmitters, the relative contribution of soluble counter ions, and the regulation and modulation of vesicle channels

	X. EPILOGUE

Top Previous Next References

In his inaugural lecture delivered at the University of London on 31 January 1952, Sir Bernard Katz stated that "The story of the nerve endings is in itself a most fascinating chapter of physiology and is full of mysteries." Now, almost half a century later, one of the mysteries, namely, the control of transmitter release by ion channels, is rapidly unfolding before our eyes. However, with the unfolding, the plot thickens. Who would have imagined just a few years ago that presynaptic nerve terminals have at least 12 different major categories of ion channels, representing at present several dozens of different ion channel types. It should come as no surprise if in the very near future this number increases to hundreds. Each channel type in turn is composed of many different ion channel molecules. This enormous number of ion channels molecules will no doubt have a significant place in understanding normal synaptic communication and plasticity. We venture to speculate that they will be of major significance in understanding pathophysiological processes as well. If this speculation turns out to be correct, the possibilities of pharmacological interventions are immense. But then we do not envy the future reviewers of this topic. The present review is based on over 7,000 articles, most of them published in the last several years. What will the number be in the near future?