I. INTRODUCTION

Top Previous Next References

Life's chemistry of aqueous solutions employs ions as carriers of cell signals. Such a signal is the action potential. That such a simple all-or-nothing signal should require highly complex proteins and ion channels, rather than just a particle within membrane bilayer, was unknown when Hodgkin and Huxley first described the processes of activation and inactivation of cation currents during the action potential in the early 1950s. In the meantime, whole families of channels with high diversity of structure and function have been described. More than conveying rapid excitation by action potentials alone, other internal cell processes are initiated by ion signals. Accordingly, the expression of these proteins is not restricted to excitable cells such as neurons or muscle but can be observed in external and internal membranes of almost all cells.

It was by electrophysiological characterization of functional channel disturbances in skeletal muscle that the underlying genetic cause for the first ion channel disease was detected. Since then, over a dozen such disorders have been described that have some striking clinical similarities leading to the coining of the terms ion channel diseases or ion channelopathies. Elucidating the pathogenesis of hereditary ion channel diseases with electrophysiological methods has given rise to new approaches for basic research and has greatly contributed to knowledge of structure-function relationships of voltage-gated ion channels. They will be decisively involved in developing strategies for specific therapy in the future. In this review, basic patterns of structure and function of voltage-gated ion channels as well as functional changes brought about by naturally occurring mutations are discussed.

Further literature relevant to this topic may be found in reviews and handbooks of the structure and function of voltage-gated ion channels in general (8, 74, 197, 372) as well as for sodium (166, 255, 389), potassium (78, 171, 242, 402, 571), calcium (201, 317, 352, 395, 538), and chloride channels (231, 232, 414). Reviews on skeletal muscle ion channel disorders (22, 65, 200, 204, 239, 286-288, 313, 408) and cardiac channelopathies (252) have also been published as well as on neurological ion channel diseases (48, 169, 180). Overviews of the whole spectrum of voltage-gated channelopathies are also available (59, 145, 287, 288).

	II. CHANNEL FUNCTION AND STRUCTURE

Top Previous Next References

Ion-conducting membrane channels are opened by ligands or voltage changes (usually depolarization) and closed by a delayed inactivation that is simultaneously initiated with the activation. Sustained exposure to the ligand or the depolarization may lead to reopenings of the channel if the circumstances (time, voltage) allow the channel to recovery from the inactivated state. The ion-conducting pore is highly selective for a specific ion as in most voltage-gated channels, or it conducts cations or anions without high selectivity as in most ligand-gated channels. The structures of the pore, its selectivity filter, and its activation and inactivation gates show high evolutionary conservation that allows one to make deductions on structure-function relationships from one channel type to the next.

A.  Voltage-Gated Cation Channels

1.  General characteristics

Voltage-sensitive cation channels usually have at least one open state and two closed states, the resting state from which the channels can be activated, i.e., opened. At the resting potential, their open probability is extremely low, meaning that only very few channels open randomly. Depolarization causes channel activation by markedly increasing open probability. During maintained depolarization, open probability is time-dependently and not voltage-dependently reduced by channel inactivation, leading to a closed state from which the channels cannot immediately be reactivated. Instead, inactivated channels require repolarization and a certain time for recovery from inactivation. On the other hand, repolarization of the membrane before the process of inactivation will deactivate the channel, i.e., reverse activation leading to the closed resting state from which the channels can be activated. In this most simple approximative model, transitions from one state to another are possible in both directions, permitting also the transition from the resting to the inactivated state at depolarization as well as the recovery from inactivation via the open state. Forward and backward rate constants for the transitions determine the probabilities of the various channel states (Fig. 1).

View larger version (28K): [in this window] [in a new window]   Fig. 1. Scheme of 3 states of sodium channel that opens rapidly upon depolarization and then closes to a fast inactivated state from which it reopens very rarely. Repolarization of membrane, initiated by inactivation of channel, leads to recovery from inactivation (= resting state) from which activation is again possible. Outward movement of voltage sensor upon depolarization results in both opening of pore and exposure of a docking site for inactivation gate. (Scheme developed in collaboration with Dr. W. Melzer.)

A) STRUCTURE OF -SUBUNITS. Basic motif of this essential subunit is a tetramic association of a series of six transmembrane -helical segments, numbered S1-S6, connected by both intracellular and extracellular loops, the interlinkers (Figs. 2 and 3). Voltage-gated sodium and calcium channels and at least the potassium channels of the Shaker family show varying subunit composition. Of these, the -subunit determines main characteristics of the complex conveying ion selectivity and containing the ion-conducting pore, voltage sensors, gates for the different opened and closed channel states, and binding sites for endogenous and exogenous ligands.

View larger version (36K): [in this window] [in a new window]   Fig. 2. Classification of elementary units of cation channel -subunits on basis of their relation of transmembrane to pore segments. All functional voltage-gated -subunits consist of 4 units of 6 transmembrane segments each including voltage sensor (segment 4) characterized by several positive amino acid residues. All 4 units are encoded by a single sodium (or calcium) channel gene, whereas potassium channel genes code for only 1 unit. Although cyclic nucleotide monophosphate (cNMP)-activated channels contain a positive segment 4, they are not voltage sensitive at all, maybe due to uncoupling of sensor and activation gate. x, 1 ... n.

View larger version (38K): [in this window] [in a new window]   Fig. 3. Subunits of voltage-gated sodium channel. -Subunit consists of 4 highly homologous domains (repeats I-IV) containing 2 transmembrane segments each (S1-S6). S5-S6 loops form ion-selective pore, and S4 segments contain positively charged residues conferring voltage dependence to protein. Repeats are connected by intracellular loops; one of them, III-IV linker, contains supposed inactivation particle of channel. 1 and 2 are auxilliary subunits. When inserted in membrane, 4 repeats of protein fold to generate a central pore as schematically indicated on bottom right. To date, mutations have been described for -subunits of various species and tissues: human and equine adult skeletal muscle (Skm-1), human heart (hH-1), and murine brain. So far, only 1 mutation has been reported for a sodium channel subunit, i.e., 1 of human brain. Conventional 1-letter abbreviations are used for replaced amino acids whose positions are given by respective numbers of human skeletal muscle channel. Different symbols used for point mutations indicate resulting diseases as explained at bottom left.

B) -SUBUNIT PORE REGION. The pore region of voltage-gated potassium channels, the S5-S6 interlinker (P-region), was first defined by studies of external and internal binding sites of a pore-blocking agent, tetraethylammonium (TEA) (247, 248, 258, 323, 587). A combination of site-directed mutagenesis and toxin binding studies showed that neutralization of specific negative residues of this so-called SS1-SS2 or P-region abolished the ability of positively charged toxins to physically block the potassium channels and altered their ion specificity (371, 530, 587). This suggested that the P-region contributed to the lining of the channel pore and the negative charges surrounding the external mouth of the pore forming the selectivity filter of the channel. Selectivity filters are common structures in all voltage-gated cation channels, and modifications in only a few decisive residues in the P-region may make the pore selective for a different cation (155). For the sodium channel, for example, substitution of lysine and alanine residues by glutamate made the protein selective for calcium (193, 392).

C) ACTIVATION. Activation results from a depolarization-induced conformational change of the protein leading to the opening of the ion-conducting pore. Even when the pore is pharmacologically blocked, charge movements in the electrical membrane field are measurable, the so-called gating current, associated with movements of the highly conserved -helical S4 segments carrying arginines or lysines at every third amino acid residue. Replacing the positive charges with neutral or negatively charged residues reduces the steepness of the voltage dependence of activation (519), making them candidates for the voltage sensor of the channel. Size and shape of the hydrophobic (neutral) residues between the positive charges are equally important for the ability of S4 to move (21). During their outward movement, the S4 segments seem to move in a spiral path ("sliding helix" or "helical screw" model) outward through "canaliculi" of the channel protein, the outer charges becoming exposed on the cell surface while the inner charges become buried in the membrane during activation (335, 583).

D) INACTIVATION AND RECOVERY FROM INACTIVATION. Voltage-gated channels usually display two modes of inactivation, fast and slow. These nonconducting inactivated states are probably mediated by different molecular mechanisms. Fast inactivation describes the rapid and complete decay of currents observed in response to short millisecond depolarizations. Slow inactivation occurs when cells are depolarized for seconds or minutes. Recovery from inactivation takes place at membrane repolarization on similar time scales as inactivation itself.

E) FAST INACTIVATION. From studies employing intracellular perfusion of the giant axon with the proteolytic enzyme pronase that abolishes sodium channel inactivation (19), it has become obvious that the inactivation gate is accessible to cytoplasmic agents. The absence of charge movements during inactivation suggested the localization of the inactivation gate outside the membrane voltage field. Based on these studies, Armstrong and Bezanilla (18, 37) proposed a ball-and-chain model in which the ball, tethered to the cytoplasmic side of the channel by a chain, swings into the inner mouth of the pore where it binds and blocks ion fluxes. Although originally proposed for the sodium channel, this model has quite convincingly been shown for fast-inactivating potassium channels where the pore-blocking ball is part of or attached to the NH₂ terminal (so-called N-type inactivation) (203). For sodium channels, one or more of the cytoplasmic loops that connect the various domains could be involved in fast inactivation. That the loop between domains III and IV is essential was demonstrated by antibodies specifically directed against this region which slowed fast inactivation (547). Mutagenesis experiments confirmed the sodium channel III/IV loop as the putative inactivation gate. With elimination of the III/IV interlinker by mRNA cleavage and coexpression of the two resulting partial mRNA in Xenopus oocytes, inactivation was markedly slowed (519). The inactivation particle itself, i.e., the ball, seems to consist of three consecutive amino acids near the middle of the loop, namely, a phenylalanine flanked by two other hydrophobic amino acids (573). Functional similarity allows this III/IV loop of the sodium channel even to confer fast inactivation to slowly inactivating potassium channels (391).

Because of the resemblance of this III-IV loop to the hinged lids of allosteric enzymes controlling substrate access, a slight modification of the ball-and-chain model was proposed (Fig. 4). According to this hinged-lid model, the inactivation particle acts as a latch of a putative catch to be identified, and one of the hinges consists of a pair of glycines situated in the vicinity of the phenylalanine (391, 573). In potassium channels, the S4-S5 interlinkers putatively adjacent to the intracellular orifice of the pore may act as the acceptor for the N-type inactivation particle (202, 217, 497). Similar but not identical parts of the supposed S4-S5 helices and adjacent amino acids of the transmembrane segments S5 and S6 may form the catch of the sodium channel (136, 295, 344, 345, 363).

View larger version (34K): [in this window] [in a new window]   Fig. 4. Hinged-lid model of fast inactivation of sodium channels and effects of mutations at various locations on current decay. A: bird's eye view of channel consisting of 4 similar repeats (IIV). Channel is cut and spread open between repeats II and III to allow view on intracellular loop between repeats III and IV. Loop acts as inactivation gate whose "hinge" GG (= a pair of glycines) allows it to "swing" between 2 positions, i.e., noninactivated channel state (pore open; left) and inactivated state (pore blocked by "plug" IMF = amino acid sequence isoleucine, phenylalanine, methionine; right). [Modified from West et al. (573).] B: substitution of E (Glu) for Gly-1306 slows channel inactivation (left panels, cf. fast current decay in wild-type channel on far left) and leads to a life-threatening form of potassium-aggravated myotonia. Designed substitution of QQQ (Gln-Gln-Gln) for IFM (Ile-Phe-Met) completely abolishes channel inactivation (right panels) proving that loop between repeats III and IV is indeed inactivation gate. [Modified from West et al. (573) and Mitrovic et al. (361).] Bottom: effects of disease mutations on current. , increased; ø, no change.

F) SLOW INACTIVATION. Slow inactivation, also called core-associated or C-type inactivation, is not only kinetically distinct from fast inactivation, it also involves different structural elements. Present in almost all voltage-gated cation channels, it is detectable even when fast inactivation has been destroyed. It is the only inactivation possibility in potassium channels devoid of the ball. In potassium channels, C-type inactivation is influenced by cations of external and internal solutions that interact with certain amino acids of the ion-conducting pore (foot-in-the-door model of gating; Refs. 172, 315). The removal of inactivation by a point mutation in the pore led to the depiction of another term, the P-type (pore-type) inactivation (105). Further inactivation mechanisms involving modification of the intracellular vestibule by an auxiliary subunit have been described (365, 375).

G) -SUBUNIT TERTIARY STRUCTURE. Out of the homology of the four domains in sodium, calcium, and many potassium channels, a fourfold symmetry surrounding a central pore lined by the S5-S6 interlinkers dipping down into the membrane has been proposed. The S5 and S6 segments are thought to be located near the pore while the adjacent S4 segments are proposed to move outward upon depolarization in a space independent of the ion-conducting pore, and the S1-S3 segments are suggested to be situated adjacent to the lipid bilayer because of their amphipathic helical structure which allows interaction with lipid and polypeptides (181). The selectivity filter is the most narrow portion of the pore. A model for the selectivity filter was developed for the potassium channel from Streptomyces lividans in which main chain carbonyl oxygen lines the selectivity filter, which is held open by structural constraints to coordinate potassium ions but not smaller sodium ions. The selectivity filter contains two potassium ions, promoting ion conduction by exploiting electrostatic repulsive forces to overcome the interattractive forces (114).

H) ADDITIONAL SUBUNITS AND CHANNEL-ASSOCIATED MEMBRANE PROTEINS. Quarternary structure and channel function are dependent on additional subunits that may modify voltage sensitivity, kinetics, expression levels, or membrane localization (178). The auxilliary subunits of the different cation channels do not share homologous structures, indicating a large variety in possible mechanisms of channel modification. Usually, one of at least two different -subunits may bind to a single -subunit, e.g., the complete potassium channel tetramer binds up to four ₂-subunits. Although all cation channels consist of - and optional - subunits, only calcium channels consist of two additional proteins, ₂ and  encoded by a single gene, the transmembrane -subunit covalently binding the extracellular ₂-subunit by disulfide bonds. Specific for skeletal muscle, an additional transmembrane -subunit facilitating inactivation had been described (229). Recently, a similar subunit has been found for the brain (300). Link of -subunits to the underlying cytoskeleton by associated proteins like ankyrin and spectrin that bind to the brain sodium channel may help to control the mobility of the channels within the fluid lipid bilayer, thus enabling cell membrane topology (508).

I) VOLTAGE-INSENSITIVE CATION CHANNELS. Cyclic nucleotide-gated channels share common structural features with the voltage-sensitive channels, e.g., the -subunit composition with the four domains of six segments each as well as a voltage-sensing positive charge motif in the S4 (Fig. 2) that in the physiological voltage range is stabilized in an activated conformation that is permissive for pore opening (527). Functional channels show less selectivity for a specific cation, and they are activated by direct binding of cAMP or cGMP to the intracellular COOH terminus. Calcium ions permeate the channels and thereby block the pore for sodium, a mechanism important for biofeedback regulation of cyclic nucleotide synthesis and degradation. This block from the extracellular side of the pore takes place in a voltage-dependent fashion and is dependent on glutamate residues in the P-region, a selectivity filter also found in voltage-dependent calcium channels (see sect. IIA3B). Replacement of glutamate by glycine therefore reduces calcium permeability. A human disease is linked to mutations in channels of this group, i.e., missense and nonsense changes in the -subunit of cGMP channels: autosomal recessive retinitis pigmentosa (118), not further described in this review.

Membrane depolarization of excitable cells causes sodium channel activation in a positive-feedback mechanism along both the concentration gradient and the electric field. The resulting increase in sodium conductance of the membrane is associated with further depolarization and activation of further sodium channels. This produces an action potential that rises to a peak close to the sodium equilibrium potential within ~1 ms. The channel's intrinsic inactivation occurs within a few milliseconds (Fig. 5) and leads under unclamped, i.e., natural, conditions to repolarization of the membrane even in the absence of any voltage-gated potassium channels. After an action potential, the cell membrane is inexcitable for a short period of time, the so-called refractory period. The duration of this period of time is regulated by the kinetics of recovery of the channels from inactivation and is the limiting factor for the firing rate of the cells (Fig. 1).

View larger version (20K): [in this window] [in a new window]   Fig. 5. Two examples of faulty inactivation of mutant sodium channels of skeletal muscle. Patch-clamp recordings from normal (wt), Gly-1306-Val, and Val-1589-Met channels expressed in HEK 293 cells. Top panels: families of sodium currents recorded at various test potentials in whole cell mode show slowed decay and failure to return completely to baseline. Slowed inactivation is more pronounced with Gly-1306-Val; persistent inward sodium current is larger for Val-1589-Met. Bottom panels: traces of 5 single-channel recordings each obtained by clamping membrane potential to 20 mV. Mutant channels show reopenings that are reason for "macroscopic" current alterations shown in top panels. [Modified from Mitrovic and co-workers (361, 363).]

A) SODIUM CHANNEL -SUBUNITS. Ten different genes have been identified in the human genome that are known to encode four-domain -subunits of voltage-gated sodium channels (SCN1A to SCN10A, Table 1). Because voltage-gated sodium channels are responsible for the fast component of action potentials, most of these genes are expressed in excitable tissues such as brain, peripheral nerve, and skeletal muscle. Most important in the context of this review are SCN4A and SCN5A, the genes associated with diseases in humans: hyperkalemic periodic paralysis (also in horses, see Fig. 6), paramyotonia congenita, potassium-aggravated myotonia (see sect. IIIA), and long Q-T syndrome 3 (see sect. IV). Although the product of SCN4A is the only sodium channel -subunit detectable in the fully differentiated and innervated skeletal muscle, SCN5A is expressed in both cardiac muscle and in fetal skeletal muscle. Additionally, SCN8A is a candidate gene for inherited neurodegenerative diseases due to a deletion in the mouse homolog causing motor end-plate disease and ataxia (MED jolting, Fig. 3). Even in the fruit fly Drosophila melanogaster, mutations in the para locus encoding a neuronal voltage-gated sodium channel that is very similar to those of vertebrate sodium channels causes paralysis (see sect. VA).

View larger version (52K): [in this window] [in a new window]   Fig. 6. Top: a paralytic attack in hyperkalemic periodic paralysis elicited by rest after exercise. A naturally occurring animal model, affected quarter horse, is shown. (Courtesy of Dr. E. P. Hoffman.) Bottom: sequence of cellular events in hyperkalemic periodic paralysis patients that may result in muscle paralysis. [K+]e, extracellular K+.

B) PORE REGIONS IDENTIFIED BY TOXIN BINDING. Most -subunits expressed in brain and skeletal muscle are sensitive to tetrodotoxin (TTX), which blocks ion flux through the sodium channels by binding to the proximal part of the S5-S6 interlinker of repeat I and thereby occludes the external mouth of the pore in micromolar concentrations. Binding of TTX in these channels is conferred by an aromatic amino acid (Tyr or Phe) not present in the product of SCN5A, which is ~200-fold less sensitive to the toxin due to a cysteine residue at an equivalent position (83, 465). Origins of this obviously extremely poisonous substance are liver and ovaries of the Puffer fish Fugu whose flesh and testes are an exquisite Japanese dish. Separation of the ovaries from the testes in the hermaphrodite fish is a prerequisite for consumption, a task requiring expertise if the physiologically interested gourmet is to survive.

Sea anemone toxin (ATX) and -scorpion toxin (-ScTX) also bind to the extracellular surface of the channel at overlapping sites of the S3-S4 loop of repeat IV, in particular to a glutamate residue only present in the neuronal subunit (439). Both ATX and -ScTX impair fast inactivation, although the inactivation particle, part of the III-IV interlinker, can block the channel from the intracellular side. Slowing of fast inactivation may be explained by inhibition of the depolarization-induced outward movement of IVS4 that initiates both activation and coupled fast inactivation. A similar depolarization-induced movement of IIS4 is needed for the binding of a -ScTX to its extracellular end and to a receptor site that includes a highly conserved glycine in the adjacent S3-S4 loop, thereby shifting the voltage dependence of activation to more negative potentials and enhancing the closed state inactivation (75a). Extracellular application of another excitotoxin, -conotoxin P_VIA, produced by the inedible fish-hunting purple cone snail, Conus purpurascens, also slows fast inactivation, resulting in increased sodium current and membrane depolarization (488, 531).

C) STRUCTURES MEDIATING INACTIVATION. Disturbance of fast inactivation is not only a popular toxin mechanism but also is decisive for sodium channel disease pathogenesis (66, 68, 283, 284, 289), indicating a crucial function in the ability of cells to maintain their homeostasis. Regions important for the inactivation process in the sodium channel were identified by deletion of 10 amino acids at the NH₂-terminal end of the inactivation loop between repeats III and IV, which completely blocked fast channel inactivation (390). The inactivation particle itself is thought to be formed by three hydrophobic amino acids (Ile-Phe-Met) downstream within this loop, whereby replacement of Phe-1489 (of the rat brain IIa sodium channel corresponding to Phe-1311 in the human skeletal muscle isoform) alone already led to abolishment of fast inactivation (573) (Fig. 4). A cysteine substitution for phenylalanine is only accessible to intracellular thiol reagent in the hyperpolarized state, i.e., when the channel is not inactivated (254). The interaction of the inactivation particle with its receptor is likely to be hydrophobic, since there is a close correlation between the hydrophobicity of the Ile-Phe-Met substitutions studied and the extent of inactivation. Current research focuses on the hydrophobic parts of the putative intracellular orifice of the pore or its surrounding protein parts which may act as receptor of the inactivation particle (136, 295, 345). Identification of this acceptor site may give a clue to which residues contribute to the intracellular mouth of the permeation pathway.

D) VOLTAGE DEPENDENCE OF INACTIVATION. Fast inactivation derives most of its voltage dependence from coupling to activation. The conformational changes resulting from depolarization-induced activation increase the rate of inactivation. Although the structural nature of this coupling is unknown, electrophysiological experiments on naturally occuring mutants revealed that mutations in segment S4 of domain IV selectively affect the voltage dependence of inactivation time constants (76, 377).

E) SODIUM CHANNEL -SUBUNITS (FIG. 3). Several different genes encoding -subunits seem to exist (Table 1); however, only one, SCN1B, has been localized, and two gene products have been characterized to date. Only ₁ is expressed in skeletal muscle, whereas brain and heart additionally express ₂, which is covalently bound to the -protein (187) by disulfide bonds (222). Sequence analysis of the noncovalently bound ₁ suggests a structure consisting of a single transmembrane segment with an extracellular NH₂ terminal containing glycosylation sites (221). The 1:1 stoichiometry of -binding indicates only one of the four domains to mediate the binding site (268) that is located in the extracellular parts of the S5-S6 loops of the second and fourth domain (329). In Xenopus oocytes, the heterologously expressed -subunit shows slower inactivation than when coexpressed with the -subunit, with the slowing corresponding to additional channel reopenings or bursts (221, 266, 267). This suggests at least two gating modes, the one with the faster inactivation being stabilized by ₁ (220, 223, 234, 477). Site-directed mutagenesis of ₁ showed that small deletions in the extracellular domain slow both inactivation and recovery from inactivation, whereas deletion of the intracellular COOH terminus had no effect (82).

Although in Xenopus oocytes the ₁-subunit increased functional expression and current amplitude, shifted the steady-state activation and inactivation curves toward more negative potentials, and accelerated recovery from inactivation (328) of the -skeletal subunit, it had no effect on the gating of the cardiac channel (328). In mammalian cells without endogenous ₁-production, coexpression of ₁ had similar effects except for the almost missing acceleration of the already fast inactivation of the channel (223). This suggests the change in kinetics may be due to posttranslational modifications (phoshorylation and glycosylation) or lack of compatible endogenous auxiliary subunits in amphibian cells compared with mammalian expression systems. It is the ₁-subunit that shows mutations causing human disease: febrile seizures and generalized epilepsy (see sect. VA, Fig. 3).

F) VOLTAGE-INSENSITIVE SODIUM CHANNELS. A novel gene superfamily (SCNN, Table 1) that encodes voltage-insensitive ion channels involved in neurotransmission and in the control of cellular and extracellular volume as well as of distinct functions, such as mechanotransduction, contains the amiloride-sensitive epithelial sodium channel (ENaC) (64). This heteromultimeric protein, made up of three homologous subunits, , , and , each containing only two transmembrane segments, is rate-limiting for electrogenic sodium reabsorption in the distal part of the renal tubule, the distal colon, and the airways (60). Expression of the -protein is obligatory for the channel to function. A gain or change of function in human ENaC has been described in Liddle's syndrome, an autosomal dominant form of salt-sensitive hypertension (pseudohyperaldosteronism) resulting from point mutations in the - or -subunits (470, 486), whereas loss-of-function mutations cause salt wasting with hyperkalaemic acidosis (pseudohypoaldosteronism type I; Refs. 79, 176, 515). Because voltage-insensitive channels are not the topic of this review, these diseases are not discussed in more detail.

A) CLASSIFICATION. Voltage-gated calcium channels are classified into transient (T-type) and long-lasting (L-type) currents according to their inactivation properties, and B (brain), N (neuronal), P (Purkinje cell), and R (toxin resistant) channels are distinguished depending on their tissue expression pattern and toxin sensitivity. Although T-type channels are low-voltage activated (LVA), the thresholds for L-type and P-types are high-voltage activated (HVA). Pharmacologically, N-type channels are blocked by -conotoxin M_VIIC (198) and P-type channels by the funnel web spider toxin -Aga_IVA and by -conotoxin G_IVA (359). L-type channels are very sensitive to dihydropyridines (DHP; e.g., nifedipine), phenylalkylamines (e.g., verapamil), and benzothiazepines (e.g., diltiazem), which has led to the term dihydropyridine receptor, a misnomer because it suggests ligand activation when, in fact, the channel is activated voltage dependently. R-type channels are resistant to these toxins and drugs; they are opened by a depolarization smaller than that needed for HVA channels such as L-type channels and larger than that necessary for LVA channels like the T-type channels (426).

Voltage-gated calcium channels and especially the cardiac L-type channel are modulated by cAMP-dependent protein kinase A via certain G_s proteins (586). The ₁-subunits of N, P, and R channels may be voltage-dependently inhibited by G proteins by a direct interaction between the G complex and the ₁-subunit (111, 194, 208), e.g., somatostatin, carbachol, ATP, and adenosine are able to reduce inward current amplitude and to slow inactivation, effects that can be prevented by pertussis toxin (348), a G protein inhibitor. These stimulatory or inhibitory receptor-coupled mechanisms may coexist in synapses of the autonomous nervous system and its effector cells.

B) CALCIUM CHANNEL -SUBUNIT ENCODING GENES. Eight different ₁-subunit genes (CACNA to GACNG, Table 2) with homologous structure of their products to the sodium -subunits have been published in vertebrates (Fig. 2). The high selectivity for calcium over sodium is conferred by a group of conserved glutamate residues forming a high-affinity calcium binding site in the pore exhibiting an apparent dissociation constant of ~700 nM (582). Nevertheless, the channel conducts a reasonably high calcium flux, probably by the vicinity of a second binding site. When only one of the sites is occupied, which is the case at low concentration, calcium is bound tightly. However, as soon as the probability of double occupancy increases at higher calcium concentration, electrostatic repulsion drastically reduces the time that the ions spend at the site and calcium flows through the channel along its electrochemical gradient. Therefore, monovalent cations (e.g., sodium) pass the channel in the absence of divalents, micromolar calcium blocks the monovalent current, and millimolar external calcium leads to an almost pure calcium inward current (10). This binding site is conserved through all ₁-subunits of the calcium channel family.

C) CALCIUM CHANNEL -SUBUNITS OF BRAIN (FIG. 7 AND PARTICULARLY FIG. 17 IN SECTION VC). The neuronal _1A channel expressed in brain, presynaptical membrane of neuromuscular junction, axon-associated Schwann cells, and distal kidney convolute tubule is a master of disguise (574). Its subcellular localization in the cerebellum, for example, is determined by at least 10 different splice variants known to date, i.e., BI-1 is mainly expressed in the dendrites while rbA is located primarily in the membrane of soma and axon terminal (460). Not only the multiple splice variants modify its gating properties but also the variability in coexpression of auxiliary subunits (596). The resulting electrophysiological properties exhibit such different characteristics as the rapidly inactivating Q-type current (granular cell-type calcium channel) and the slowly inactivating P-type current (Purkinje cell calcium channel; Ref. 502) originally thought to be mediated by two totally different voltage-gated calcium channels. Additional variations result from interaction with N-type channels in vesicle formation and neurotransmitter release. The II-III loop of the protein binds syntaxin, which in turn has a functional effect on the PQN channel complex. Mutations in _1A cause familial hemiplegic migraine, episodic ataxia type 2, and spinocerebellar atrophy type 6 in humans as well as ataxia and seizures in mice (tottering and leaner) (see sect. VC).

View larger version (32K): [in this window] [in a new window]   Fig. 7. Subunits of voltage-gated calcium channel. -Subunit resembles that of sodium channel; however, function of various parts, e.g., III-IV linker, may not be same. 2/, 1 to 4, and  are auxilliary subunits. Mutations shown here, 1S-subunit of skeletal muscle L-type calcium channel (= dihydropyridine receptor), have been described for humans (HypoPP, MHS 5) and mice (mdg). Mutations have been also reported for 1A-subunit of brain P/Q-type channel (see Fig. 17) and 4-subunit causing tottering mice. Conventional 1-letter abbreviations are used for replaced amino acids whose positions are given by respective numbers of 1S-subunit. Symbols used for point mutations indicate resulting diseases as explained at bottom left.

D) CALCIUM CHANNEL -SUBUNITS OF THE VARIOUS MUSCLE TYPES (FIG. 7). Different splice variants of the same gene, CACNA1C, encode the L-type calcium channel of heart (_1Ca) and smooth muscle (_1Cb), whereas the main subunit of the L-type calcium channel of skeletal muscle, _1S, is encoded by another gene, CACNA1S. Physiologically, two _1S-isoforms are expressed: the rare 212-kDa complete protein and a likewise functional truncated form of 190 kDa comprising 95% of total channel population resulting by posttranslational proteolysis at amino acid 1690 (24, 106). An additional variant has been suggested to exist, at least in postnatal skeletal muscle (332). Further functional alterations have also been demonstrated, i.e., by phosphorylation; however, the physiological importance of these alterations, which are acknowledged for the cardiac channel, seems questionable for skeletal muscle (140). Mutations in _1S cause hypokalemic periodic paralysis and malignant hyperthermia susceptibility type 5 in humans as well as muscular dysgenesis in mice (see sect. IIIB).

E) ADDITIONAL SUBUNITS OF CALCIUM CHANNELS (FIG. 7). Even though the ₁-subunits form functional channels by themselves, maximally four additional subunits, ₂, , , and  copurified. The extracellularly located ₂ protein is anchored by disulfide bonds to the membrane-spanning subunit (230, 178), and the two proteins are encoded by a single gene. It was originally thought to possess an ion-conducting pore since expression in cells devoid of functional calcium channels resulted in an appreciable calcium current. In the meantime, this phenomenon can be explained by the drastic increase in expression of endogenous ₁-subunits by coexpression of ₂/-subunit (179). This subunit, which can bind the anticonvulsant drug gabapentin, not only increases ₁-expression rates and current density, but also accelerates inactivation kinetics and slightly shifts both steady-state inactivation and activation curves in hyperpolarizing directions (494).

Coexpression of any of the four -subunits with _1A markedly increases the number of channel complexes inserted into the membrane and the current amplitude (51). For _1S, -coexpression increased the number of DHP binding sites and accelerated current activation kinetics, however, without increasing current density (276, 546). Similar effects have been noted for _1C when expressed in oocytes increasing both rate of activation and current density (393, 572). In addition to the intracellular I/II loop, the COOH terminal also seems to act as a binding site for  (556). The type of  can decisively determine current characteristics of the whole channel complex, i.e., _2A induces the P-type current and _1B and ₃ induce the Q-type current when coexpressed with _1A (509). ₁ is an intracellular acidic protein and binds to the loop connecting domains I and II of the ₁-subunit, distinct from the consensus site for the G protein -complex (111). On the other hand, ₃ is capable of differential modulation of G protein inhibition of _1A and _1B subunits (437). Mutations in ₄ are found in the lethargic mouse (see sect. VC).

The -subunits consist of four transmembrane segments and are expressed in skeletal muscle and brain. Coexpression with skeletal muscle -subunits with cardiac ₁-subunits in amphibian and mammalian cell systems moderately increased calcium current amplitude and inactivation rate. The main effect is a marked shift of the voltage dependence of inactivation in the hyperpolarizing direction (494, 293). The brain -subunit stargazin has a similar effect when coexpressed with the P/Q-type calcium channel, and mutations therein cause absence epilepsy in the stargazer mouse (300) (see sect. VC). Additional -subunits may exist (124).

F) OTHER CALCIUM CHANNELS. Two distinct classes of channels mediating release of calcium ions from intracellular stores have been identified: they are sensitive either to inositol 1,4,5-trisphosphate (InsP₃) or to a nonphysiological ligand, the plant alkaloid ryanodine. The latter, the ryanodine receptor, releases calcium from the sarcoplasmic reticulum (SR) or endoplasmasmic reticulum (ER), an essential step for contraction of skeletal, heart, and smooth muscle (Figs. 8 and 9). In the heart, calcium release is initiated by the calcium influx through the fast-activating T-type calcium channel and maintained by L-type channels during the plateau phase of the action potential. Calcium activates RYR2, the ryanodine receptor expressed in cardiac and smooth muscle. In skeletal muscle, calcium release is initiated by the surface membrane action potential, which spreads via the transverse tubular (t-tubular) system within the fiber. A depolarization of the tubular membrane activates the L-type calcium channel leading to a fast conformational change of one or more intracellular loops which open RYR1, the ryanodine receptor expressed in skeletal muscle. This signal transmission between the t-tubular and SR membrane is referred to as excitation-contraction (EC) coupling (for review, see Ref. 149).

View larger version (28K): [in this window] [in a new window]   Fig. 8. Triadic junction between a transverse tubule and sarcoplasmic reticulum: position of 2 calcium channels of skeletal muscle, L-type calcium channel, also called dihydropyridine (DHP) receptor, and calcium release channel, also called ryanodine receptor. Coupling between the 2 channels is not fully elucidated. Mutations in respective genes cause hypokalemic periodic paralysis, malignant hyperthermia, or central core disease.

View larger version (46K): [in this window] [in a new window]   Fig. 9. Cartoon of homotetrameric ryanodine receptor, calcium release channel situated in membrane of sarcoplasmic reticulum (SR). Cytosolic part of protein complex, the so-called foot, bridges gap between transverse tubular system and SR. Mutations have been described for skeletal muscle ryanodine receptor (RYR1) that cause susceptibility to malignant hyperthermia and central core disease. Conventional 1-letter abbreviations are used for replaced amino acids whose positions are given by respective numbers of human RYR1. [Modified from Melzer et al. (352).]

Ryanodine receptors are among the largest known proteins consisting of homotetramers each over 5,000 amino acids with a molecular mass of ~565 kDa. Morphological studies revealed a quatrefoil structure with the hydrophobic parts of the four subunits forming a membrane-spanning base plate and the hydrophilic segments forming a cytoplasmic domain, the foot, which bridges the gap between t-tubular and SR membrane. Despite of the huge size, RYR1 channels can be functionally expressed by transient transfection and reveal very similar characteristics to those of the native channel, e.g., a calcium conductance of 116 pS in 50 mM calcium and an open time constant of 0.22 ms (85). RYR1 mutations cause susceptibility to malignant hyperthermia, type 1, a potentially lethal event during general anesthesia, and some of them lead to central core disease, a congenital myopathy (325) (see sect. IIIB).

Potassium channels constitute the most diverse class of ion channels with respect to kinetic properties, regulation, pharmacology, and structure (Tables 3 and 4). In vertebrates, over 13 subfamilies have been described, 8 of which show the typical voltage-dependent channel structural features with voltage sensor, toxin binding sites, and a single ion-conducting pore. Because of the high variability in structure, the channels can be classified according to the number of pore regions (P) and the number of helical structures that were thought to correspond to the number of transmembrane segments (termed T or S in the voltage-gated and M in the 4 ligand-gated channels).

As to be expected by their diversity, even voltage-gated potassium channels are expressed in nonexcitable tissues. This knowledge, which seems taken for granted today, was gained by a shift of weather-altering water currents in which giant squids roamed about near the coast of Irvine, California. Because of the absence of their beloved experimental animal, the potassium channel scientists were in immediate need of other plentiful research material, a problem they solved by changing to human blood lymphocytes also expressing the potassium channels of interest.

A) CLASSICAL VOLTAGE-GATED -SUBUNITS: KV AS HUMAN HOMOLOGS OF SHAKER, SHAB, SHAW, AND SHAL (FIG. 2). These channels inactivate at different rates and to a varying extent (fast N-type and slow C-type inactivation). The rapidly inactivating A-type Kv channels operate in the subthreshold range of an action potential and play a key role in the generation of pre- and postsynaptic signals; the slowly inactivating, delayed rectifying channels repolarize the cell membrane during an action potential and reduce cell excitability. Both types are found in almost all eukaryotic cells of the animal and plant kingdom (450) and are not only present in nerve or muscle cells, but also in lymphocytes, pancreatic islet cells, and others (173). The -subunits, typically forming tetramers, contain six transmembrane segments (6T). Four potassium channel families show this pattern (name of the corresponding gene in parentheses), namely, Shaker/Kv1 (KCNA), Shab/Kv2 (KCNB), Shaw/Kv3 (KCNC), and Shal/Kv4 (KCND) (78). Compatibility to aggregate as homotetramers (or within a channel subfamily also as heterotetramer) is conferred by a part of the intracellular NH₂ terminal (303). Tetramers show fast N-type inactivation if the NH₂ terminal of one of the -subunits carries the inactivation "ball," e.g., when Kv1.4 is involved. In the absence of the "N-type ball," rapid inactivation can also be achieved by the ₁-subunit (322, 428, 485) as for Kv1.1, but, in the presence of an N-type inactivation prevention (NIP region) as in Kv1.6, fast inactivation cannot be achieved even in heteromultimers with Kv1.4. Until now, one human disease is known to be associated to mutations in Kv1.1: episodic ataxia type 1 (see sect. VC).

Voltage-gated potassium channels with unusual characterisitics are encoded by the KCNQ family. Heterologous expression of KCNQ1 cDNA encoding a heart potassium channel led to currents whose kinetics were unlike known cardiac potassium currents showing more rapid activation and greater degree of inactivation than the slow delayed rectifier, Iks, despite a common pharmacology such as potentiation by cAMP and blockage by clofilium (584). However, when coexpressed with KCNE1 cDNA encoding MinK (see below), the potassium current density was much greater and became indistinguishable from Iks. Heterologous expression of KCNQ2 yielded a slowly activated current with a threshold at 60 mV and full activation at 0 mV with similiarities to KCNQ1 in voltage dependence and kinetics, but coexpression with MinK (KCNE1) did not change current characteristics (39). Mutations in KCNQ1 cause the long Q-T syndrome type 1 (see sect. IV), and mutations in KCNQ2/3 cause benign neonatal convulsions (see sect. VB).

B) VOLTAGE-SENSITIVE -SUBUNITS: SLO OR B_k AND IKR OR HERG (FIG. 2). HERG channels have an intracellular NH₂ terminus, whereas SLO has an extracellular one. Consequently, SLO possesses an additional transmembrane segment, called S0 (559). SLO, encoded by KCNMA1, is known as the large-conductance (>150 pS with symmetrical potassium concentrations) calcium-activated channel or B_K (B for big) channel that requires massive depolarization for activation in the absence of intracellular calcium. It possesses four additional hydrophobic segments S7-S10 that are thought to be associated with the inner surface of the membrane and convey calcium sensitivity. A -subunit, hslo-beta, interacts with SLO and increases its sensitivity to charybdotoxin and intracellular calcium.

Nearly a decade ago, Ikr (r for rapid) was electrophysiologically described as a relatively fast delayed rectifier in cardiac myocytes and contrasted to the very slow delayed-rectifying potassium current, Iks (s for slow; Ref. 578). Surprisingly, it showed similarity to the Drosophila ether-a-go-go (eag) potassium channel, leading to the term human ether-a-go-go-related gene (HERG). Heterologous expression indeed showed a potassium channel with gating and pharmacological properties similar to the Ikr current. A unique feature of the voltage-dependent gating of this channel is the relatively fast C-type inactivation due to high sensitivity to extracellular cations as well as the slow activation and deactivation in comparison with the rapid inactivation. This combination of gating kinetics produces a novel inward-rectifying behavior of the channel: inactivation is faster than activation so that membrane depolarization is not associated with an outward current, and recovery from inactivation is faster than deactivation so that repolarization causes a long-lasting tail current (474, 501). The NH₂ terminus stabilizes the open state and, by a separate mechanism, promotes C-type inactivation (565). Two alternatively processed, cardiac-specific isoforms have been identified: a splice variant that has a much shorter NH₂ terminus (280, 314) and another with a truncated COOH terminus that prevents current generation (274). When coexpressed with the full-length cDNA or by itself, the truncated NH₂-terminal isoform displays faster deactivation gating kinetics (280, 314). Heteromers may form the channels responsible for Ikr in vivo. HERG mutations cause the long Q-T syndrome type 2 (see sect. IV).

C) VOLTAGE-INSENSITIVE -SUBUNITS: CYCLIC NUCLEOTIDE-GATED, SK, AND IK (FIG. 2). Although these channels contain similar structures as the voltage-gated -subunits, i.e., S4 segments with positive residues at almost every third amino acid, they are not activated by depolarization. Cyclic nucleotide monophosphate (cNMP)-gated potassium channels possess an intracellularly located helical segment conveying a cyclic nucleotide binding site situated at the COOH terminus. These channels are not specific for potassium because the specificity filter is altered with the GYGD motif typical for the voltage-gated channels missing (189, 190). Calcium-gated voltage-insensitive potassium channels with small (SK1-3; ~10 pS) and intermediate (IK; ~40 pS) conductance also show the typical 6T structure; the latter is sensitive to charybdotoxin and clomitrazole and probably represents the Gardos channel (the Gardos effect is an increase in potassium conductance at ATP depletion in the presence of extracellular calcium as originally reported for red blood cells). The various SK/IK types are resistant to TEA; however, they differ in their affinity to gallamine, D-tubocurarine, and apamin, a bee venom (218, 551). KCa4, SK4, and IK1 seem to be products of the same gene. Because SK and IK are activated by high intracellular calcium, which may result from cell activity, e.g., calcium influx during action potentials or calcium release from intracellular stores, their hyperpolarizing effect may terminate a burst of action potentials. All these channels are not gated by voltage and are therefore not discussed further in this review, even though myotonic dystrophy, the most frequent muscle disorder of human adults, may be related to an abnormal expression of an apamin-sensitive potassium channel (for review, see Ref. 186).

D) 2T/1P -SUBUNITS: THE INWARD GOING RECTIFIERS (FIG. 2). The simplest motif for voltage-insensitive potassium channel -subunits are protein tetramers each consisting of only two membrane-spanning segments (M1 and M2) and an interlinker forming the pore. Its architecture has been recently demonstrated by crystallization of the corresponding S. lividans channel (17, 114). The homologous human channels are the inward-going rectifiers, K_ir, that are encoded by the over 15 genes of the KCNJ family. Rectification results from internal blockade by magnesium ions and the ubiquitously cytoplasmic polyamines such as spermine. Grade of rectification is determined by an important amino acid residue in M2 near the cytoplasmic side of the cell, whereby a negatively charged aspartate confers strong rectification and a neutral asparagine confers weak rectification (571). Although the NH₂ terminal does not carry an inactivation particle, fast inactivation is conferred by spermine. The presence of a lysine residue in the NH₂ terminal conveys sensitivity to pH (134).

K_ir3.4 and K_ir3.2 are well known as potassium channel -subunits that are blocked by intracellular ATP and unblocked at energy depletion (KATP1 and KATP2, respectively), whereas ROMK1 shows a paradoxical effect with ATP activating the channel (199). K_ir6.1 and K_ir6.2 interact with the sulfhydrylurea receptors SUR1 and SUR2 (see below) forming hetero-octameric K_ATP channel complexes with a (SUR-K_ir6.x)₄ stoichiometry and a tetrameric pore of 76 pS in the fully open state (13, 93), even though neither subunit alone exhibits channel activity. K_ATP complexes modulate insulin secretion and are activated by MgADP, cromakalim, pinacidil, and diazoxide and are inhibited by ATP and sulfonylureas, e.g., glibenclamide (213). Some of the inward-going rectifiers of the 2T/1P group are also directly gated by G proteins (GIRK1,2,4), e.g., GIRK1 is opened by parasympathetic nerve stimulation via activation of muscarinic acetylcholine receptors resulting in slower heart rate (272).

Several diseases are caused by mutations in channels encoded by genes of the KCNJ family: GIRK2, K_ir6.2, and ROMK1. A missense mutation in the pore region of the GIRK2 channel causes ataxia in homozygous weaver mice that reveal a lack of differentiation of cerebellar granule cells (388). To date, this is the only one gain-of-function potassium channelopathy. Mutations in the pancreatic islet inward rectifier subunit, K_ir6.2, cause a recessive disease, persistent hyperinsulinemic hypoglycemia of infancy (534), and mutations in ROMK1 or other channels or transporters that are involved in transepithelial chloride transport in the thick ascending limb of Henle, such as the basolateral chloride channel and a basolateral K-Cl cotransporter, lead to an impairment of the furosemide-sensitive salt reabsorption and cause the hyperprostaglandin E syndrome, also termed antenatal variant of Bartter syndrome (215, 492, 109). Because G protein-coupled and ATP-dependent inward rectifiers belong to the group of voltage-insensitive potassium channels, these diseases are not discussed in more detail here.

E) 4T/2P -SUBUNITS. Next to the 2T/1P type, 4T/2P types of channels have recently been cloned in humans and are encoded by the KCNK gene family. Each subunit is equivalent to two core complexes, M1-P-M2-M3-P-M4. Pore function has not been clarified sufficiently, with presumably only one pore being functional at a given time. This basic structure is found in TWIK (twin P-domain weakly inward-rectifying potassium channel encoded by KCNK1) and TASK (TWIK-related acid-sensitive potassium channel encoded by KCNK2) channels; the latter senses external pH variations near physiological pH (123). Both are weak inward rectifiers, probably involved in background potassium membrane conductance. The TREK channel (TWIK-related potassium channel encoded by KCNK3) acts as an outward rectifier (137). Other potassium channels such as ORK, a 4T/2P outward rectifier, or 8T/2P channels (6T as in voltage-gated channels plus 2T as in inward rectifiers) have been yet been identified in humans.

F) DEFINITIVE AND PUTATIVE -SUBUNITS. Several sequences and structures turned out to conduct potassium ions only when associated with one of the above-mentioned -subunits. Of the -subunit family interacting with the voltage-gated -subunit from the cytoplasm, two are known: the ₁-subunit encoded by KCNA1B that confers N-type inactivation as well as the ₂-subunit encoded by KCNA2B that increases the expression rate of the whole channel complex (428). A structurally different -subunit, hslo-beta, which contains two transmembrane segments, interacts with SLO and increases its sensitivity to charybdotoxin (185) and to intracellular calcium concentrations >100 nM that occur during cellular excitation and that functionally couple the two subunits (347). MinK proteins (also called ISK or Isk; encoded by KCNE1) contain a single transmembranous domain and conduct slowly activating potassium currents when expressed in oocytes and other cell lines that express KCNQ channels. The resulting currents are similar to those recorded in cardiac and some epithelial cells and are modulated after activation of various second messenger systems. MinK is suggested to act as the -subunit with so far unclear stoichiometry. The COOH terminus of MinK is supposed to interact with the pore region of KCNQ channels, resulting in prolonged openings and smaller single-channel conductance. Mutations in MinK cause long Q-T syndrome type 5 (see sect. IV). Various other transmembrane -subunits with tissue specificity are identified or postulated to exist for inward going rectifiers. Mutations in SUR1, the -subunit of the K_ATP channel exclusively expressed in pancreatic islet lead to the same disease as K_ir6.2 mutations, i.e., hyperinsulinemic hypoglycemia of infancy (122, 533; not further discussed here).

The voltage-dependent anion channel (VDAC) is a small, abundant pore-forming protein found in the outer membranes of all eukaryotic mitochondria and in the membranes of other intracellular organelles as well as in the plasmamembrane. Four human VDAC isoforms are known (40). The VDAC protein is believed to form the major pathway for movement of adenine nucleotides and anions. In contrast to the channels discussed in the following section, the anion selectivity is low and, dependent on the membrane potential, monovalent and perhaps divalent cations are also conducted.

Chloride channels are present in the plasma membrane of most cells playing important roles in cell volume regulation, transepithelial transport, secretion of fluid from secretory glands, and stabilization of membrane potential. They can be activated by extracellular ligands, intracellular calcium, cAMP, G proteins, cell swelling, mechanical stretch, and transmembrane voltage. Because mechanisms of activation may overlap, and expression of a given channel may not be restricted to a certain cell type, the classification is intriguing. Of the three superfamilies, the first consists of ligand- or transmitter-gated chloride channels that are predominantly expressed in nervous tissue. Here, the chloride conductance is important for inhibitory synapses at which the membrane will be hyperpolarized by opening of transmitter-gated anion channels such as GABA_A and glycine receptor channels mediating chloride influx. These channels consist of four transmembrane domains and may associate as pentamers. The second superfamily includes the ATP-binding cassette channels and cystic fibrosis transmembrane conductance regulator (CFTR) consisting of two blocks of six transmembrane domains followed by two nucleotide binding domains (520). The third superfamily is composed of the voltage-gated chloride channels, present in excitable and epithelial cells.

Voltage-dependent chloride channels also fulfill a variety of functions depending on their tissue distribution, i.e., stabilization of the membrane potential, regulation of cell volume, and concentration of extracellular medium (232). The channels can be found both in plasmalemma and in the lining of internal organelles. In axons, voltage-dependent chloride conductance is so small that it is usually neglected, whereas that of skeletal muscle is even larger than the resting conductance for potassium (49). Nevertheless, its electrophysiological identification and characterization at the single-channel level turned out to be very difficult. The reason why is its very low single-channel conductance near 1 pS as estimated from noise analysis (414). The large macroscopic chloride conductance, therefore, must result from an extremely high channel density in the skeletal muscle membrane.

A) CHLORIDE CHANNEL ENCODING GENES. Nine different human genes have been identified (CLCN1 to CLCN7; CLCNA/B, Table 5) to encode voltage-gated chloride channels, termed CLC1 to CLC7, CLCKA, and CLCKB, that do not structurally resemble any other ion channel family. According to hydrophobicity blots performed on the first recognized family member, the CLC0 channel from the electric organ of Torpedo, 13 putative transmembrane helical segments were originally assumed (233). Later, by a combination of glycosylation and electrophysiological experiments on mutant proteins, it became clear that both the NH₂ and COOH terminals must be located intracellularly and that the S8-S9 interlinker is extracellular because of a glycosylation site. Two different possibilities of configuration arose from these results, namely, a model that places S4 extracellularly and the hydrophobic core of S9-S12 crosses the membrane several times (Fig. 10; Refs. 232, 415, 472) or, alternatively, a model that places S2 extracellularly (2, 356).