The proteins that function as voltage-gated ion channels have three remarkable properties that enable nerve cells to conduct an electric impulse: (1) opening in response to changes in the membrane potential (voltage gating); (2) subsequent channel closing and inactivation; and (3) like all ion channels, exquisite specificity for those ions that will permeate and those that will not. In this section, we describe the molecular analysis of these voltage-gated ion-channel proteins, starting with an important technique for measuring their functional properties. One of the surprising results to emerge from molecular cloning is that all voltage-gated ion channels, be they Na⁺, K⁺, or Ca²⁺ channels, are related in structure and in function. We examine in some detail the voltage-gated K⁺ channel, the first to be cloned and the most widely distributed, and then briefly discuss other members of this family of ion channels, which also includes certain non-voltage-gated channels, such as the K⁺ channels that generate the resting membrane potential.

The technique of patch clamping enables workers to investigate the opening, closing, and ion conductance of a single ion channel. As illustrated in Figure 21-19, this technique measures the electric current caused by the movement of ions across a small patch of the plasma membrane. In general, the membrane is electrically depolarized or hyperpolarized and maintained (clamped) at that potential by an electronic feedback device. Thus the membrane potential cannot change, in contrast to the situation during an action potential. The patch-clamping technique can be used in various ways as shown in Figure 21-20.

The inward or outward movement of ions across a patch of membrane can be quantified from the amount of electric current needed to maintain the membrane potential at the designated "clamped" value. To preserve electroneutrality, the entry of each positive ion (e.g., a Na⁺ ion) into the cell across the plasma membrane is balanced by the entry of an electron into the cytosol from the electrode placed in it; the resulting current flow is measured by an electronic device. Conversely, the movement of each positive ion from the cell (e.g., a K⁺ ion) is balanced by the withdrawal of an electron from the cytosol.

The patch-clamp tracings in Figure 21-21 illustrate the use of this technique to study the properties of ion channels in the plasma membrane of muscle cells. In one study, two patches of muscle membrane, each containing one voltage-gated Na⁺ channel, were depolarized about 10 mV and clamped at that voltage. Under these circumstances, the transient pulses of current that cross the membrane result from the opening and closing of individual Na⁺ channels (see Figure 21-21a). Each channel is either open or closed; there are no graded permeability changes for individual channels. From such tracings, it is possible to determine the time that a channel is open (about 0.7 ms) and the ion flux through it (9900 Na⁺ ions/ms). Replacement of the NaCl within the patch pipette (corresponding to the outside of the cell) with KCl or choline chloride abolishes current through the channels, confirming that they conduct only Na⁺ ions.

Several properties of voltage-gated K⁺ channels can be deduced from the tracings in Figure 21-21b. At the depolarizing voltage of 10 mV, the channels in the membrane patch open infrequently and remain open for only a few milliseconds, as judged, respectively, by the number and width of the "upward blips" on the tracings. Further, the ion flux through them is rather small, as measured by the electric current passing through each open channel (the height of the blips). Depolarizing the membrane further to +20 mV causes these channels to open about twice as frequently. Also, more K⁺ ions move through each open channel (the height of the blips is greater) because the force driving cytosolic K⁺ ions outward is greater at a membrane potential of +20 mV than at 10 mV. Depolarizing the membrane further to +50 mV, such as at the peak of an action potential, causes the opening of more K⁺ channels and also increases the flux of K⁺ through them. Thus, by opening during the peak of the action potential, these K⁺ channels cause the outward movement of K⁺ ions and the repolarization of the membrane potential toward the resting value.

The initial breakthrough in identification and cloning of voltage-gated ion channels came from analysis of fruit flies (Drosophila melanogaster) carrying the shaker mutation. These flies shake vigorously under ether anesthesia, reflecting a loss of motor control and a defect in excitable cells. The axons of giant nerves in shaker mutants have an abnormally prolonged action potential (Figure 21-22), suggesting that the mutation causes a defect in voltage-gated K⁺ channels that prevents them from opening normally immediately upon depolarization. Thus the X-linked shaker mutation was thought to cause a defect in a K⁺ channel protein.

The wild-type shaker gene was cloned by chromosome walking (see Figure 8-24). To show that the encoded protein indeed was a K⁺ channel, the wild-type shaker cDNA was used as a template to produce shaker mRNA in a cell-free system. Expression of this mRNA in frog oocytes and patch-clamp measurements on the newly synthesized channel protein showed that it had properties identical to those of the K⁺ channel in the neuronal membrane, demonstrating conclusively that the shaker gene encodes that K⁺ channel protein (Figure 21-23).

A functional Shaker K⁺ channel is built of four identical subunits arranged in the membrane around a central pore. (These are equivalent to the four domains of the voltage-gated Na⁺ channel shown in Figure 21-13.) The 656-amino-acid polypeptide encoded by the shaker gene, like most K⁺-channel polypeptides, contains six membrane-spanning a helixes, designated S1S6. Three regions in each subunit are critical to the functioning of the K⁺ channel: a P segment between S5 and S6, which lines the pore of the channel; an amino-terminal segment, the "ball," which moves into the open pore, inactivating the channel; and the S4 a helix, which acts as a voltage sensor (Figure 21-24). As we will see later, homologous regions are present in voltage-gated Na⁺ and Ca²⁺ channels.

In Drosophila, at least five different shaker polypeptides (isoforms) are produced by alternative splicing of the primary transcript of the shaker gene. In the oocyte expression assay, these Shaker isoforms exhibit different voltage dependencies and K⁺ conductivities. Thus differential expression of Shaker isoforms can affect the timing of repolarization during an action potential, accounting for differences in the electrical activity of different types of neurons. Also, if a single K⁺ channel is constructed from two different isoforms, the properties of the resultant "hybrid" channel differ from those of both of the channels formed from four identical polypeptides.

Using the shaker gene as a hybridization probe, workers have isolated genes encoding more than 100 K⁺-channel proteins from vertebrates. Although all the encoded channel proteins have a similar overall structure, they exhibit different voltage dependencies, conductivities, channel kinetics, and other properties. However, many open only at strongly depolarizing voltages, a property required for the channel to repolarize the membrane during an action potential. The shaker gene also has been used to isolate genes encoding a large number of K⁺-channel proteins from humans, illustrating how research on invertebrates can advance understanding of the human nervous system.

The P segments of the Shaker protein were implicated as forming part of the pore in experiments with two inhibitors that physically plug the channel mouth: a neurotoxin from scorpion venom (charybdotoxin) and tetraethylammonium ion. Site-specific mutation of any of several amino acids in the P segment rendered the channel, when expressed in oocytes, resistant to inhibition by these agents. In addition, K⁺ channels from different sources vary in their conductance (number of ions transported per millisecond) and sensitivity to tetraethylammonium. Chimeric proteins in which the P segment from one protein was replaced with that from another exhibited conductances and sensitivities to tetraethylammonium that correlated with the origin of the P segment, not with the origin of the rest of the channel protein. In the case of the K⁺ channel, four P segments, one from each subunit, are thought to form the ion-selective pore that constricts the center of the channel.

The recent determination of the three-dimensional structure of a bacterial K⁺ channel provides insight into how the remarkable ion specificity of channel proteins is achieved. Each subunit in this bacterial channel contains a P segment highly homologous in sequence to that in all known K⁺ channels. Eight transmembrane helices, two from each subunit (analogous to the S5 and S6 helices in the Shaker channel), form an "inverted teepee," generating a water-filled cavity in the outer portion of the channel that leads to the actual selectivity filter (Figure 21-25a). The filter, which is lined by four extended loops that are part of the P segments, is wide enough to accommodate a K⁺ ion only if it is stripped of its water of hydration. The ability of this filter to select K⁺ over Na⁺ is due mainly to backbone carbonyl oxygens on glycine residues located in a Gly-Tyr-Gly sequence that is found in an analogous position in the P segment in every known K⁺ channel. Mutation of these residues disrupts the channel's ability to discriminate between K⁺ and Na⁺ ions.

As a K⁺ ion enters the narrow selectivity filter, it loses its water of hydration but becomes bound with a similar geometry to four backbone carbonyl oxygens, one from the loop in each P segment lining the channel (Figure 21-25 a, b); thus a relatively low activation energy is required for passage of K⁺ ions. Because a Na⁺ ion is smaller than a K⁺ ion, after it loses its water of hydration it cannot become perfectly coordinated to the carbonyl oxygens that line the selectivity filter, so the activation energy for passage of Na⁺ ions is relatively high. As a result, 1000 K⁺ ions pass through the channel for every 1 Na⁺ ion.

Since voltage-gated channels open when the membrane depolarizes, some segment of the protein must "sense" the change in potential. Sensitive electric measurements suggest that the opening of each voltage-gated K⁺ (and Na⁺) channel is accompanied by the movement of 10 to 12 protein-bound positive charges from the cytosolic to the exoplasmic surface of the membrane; alternatively, a larger number of charges may move a shorter distance across the membrane. The movement of these gating charges (or voltage sensors) under the force of the electric field triggers a conformational change in the protein that opens the channel. Several lines of evidence indicate that the voltage sensor is the S4 transmembrane a helix present in each Shaker polypeptide (see Figure 21-24); similar S4 helices are found in voltage-gated Na⁺ and Ca²⁺ channel proteins. These voltage-sensing helices, often called gating helices, generally have a positively charged lysine or arginine every third or fourth residue. In the closed resting state, half of each S4 helix is exposed to the cytosol; when the membrane is depolarized, these amino acids move outward and become exposed to the exoplasmic surface of the channel.

The role of the S4 helix in voltage sensing has been demonstrated in studies with mutant Shaker K⁺ channels. In one experiment, one or more arginine or lysine residues in the S4 helix were replaced with neutral or acidic residues. When these mutant proteins were expressed in frog oocytes, patch-clamping measurements showed that fewer positive charges than in wild-type channels moved across the membrane immediately in response to a membrane depolarization, indicating that the arginine and lysine residues in the S4 helix do indeed more across the membrane. In other studies, mutant proteins in which various S4 residues were converted to cysteine were tested for their reactivity with a membrane-impermeant cysteine-modifying chemical agent. The results indicated that in the resting state amino acids near the C-terminus of the S4 helix face the cytosol, but when the membrane is depolarized, some of these same amino acids become exposed to the exoplasmic surface of the channel. These experiments directly demonstrate movement of the S4 helix across the membrane, as depicted in Figure 21-13 for the analogous segment in Na⁺ channels.

An important characteristic of voltage-gated channels is inactivation: soon after opening they close spontaneously, forming an inactive channel that will not reopen until the membrane is repolarized. The N-termini of each of the four Shaker polypeptides forms a globular "ball," one of which swings into the open channel, inactivating it (Figure 21-26a). Two key experiments, explained in Figure 21-26b, have shown that inactivation depends on the ball domains, occurs after channel opening, and does not require the ball domains to be part of the protein. In addition, deletion of part of the 40-amino-acid segment (the "chain") connecting the ball to the first membrane-spanning helix increases the rate of inactivation. The shorter the chain, the more rapid the inactivation, as if a ball attached to a shorter chain can move into the open channel more readily. Conversely, addition of random amino acids to lengthen the normal chain slows channel inactivation.

The pioneering research we've just described, starting with isolation of a mutant fruit fly that shakes under anesthesia, has led to the current detailed picture of the structure and operation of voltage-gated K⁺ channels. Purification, molecular cloning, and analysis of other ion-channel proteins has revealed that many have a structure remarkably similar to that of voltage-gated K⁺ channels (Figure 21-27). For instance, as we discuss later, both the visual and olfactory systems contain ion channels whose opening and closing are controlled by binding of cAMP or cGMP. Like the Shaker K⁺ channel, these channels are tetramers, each subunit containing six transmembrane a helices and a pore-lining P segment. The resting K⁺ channel, which generates the resting potential, also contains four identical subunits; these have only two membrane-spanning a helices, one on either side of a P segment, analogous to the two in the bacterial K⁺ channel (see Figure 21-25a) and the S5 and S6 helices in the Shaker channel. However, none of the a helices in the resting K⁺ channel or nucleotide-gated ion channels acts as a voltage sensor, since these channels are not voltage gated.

Although voltage-gated Na⁺ and Ca²⁺ channels are monomeric proteins, both contain four homologous transmembrane domains, each similar in sequence and structure to the central part of the Shaker polypeptide (see Figure 21-27c,d). These domains are connected and flanked by nonhomologous cytosolic segments. Each of the four homologous domains contains a nonhelical P segment, between helices S5 and S6, and a voltage-sensing S4 a helix. Sitespecific mutation studies have shown that the ion specificity of Na⁺ and Ca²⁺ channels, like that of K⁺ channels, is determined by amino acid side chains in the P segments. Unlike K⁺ channels, however, Na⁺ and Ca²⁺ channels have only a single channel-inactivation segment, located in the cytosol-facing loop between domains III and IV. The Na⁺ and Ca²⁺ channels contain additional, essential regulatory subunits that affect the rate at which the channel opens and becomes inactivated, and the voltage dependence of channel inactivation. The subunit organization and number of transmembrane a helices varies in some other ion channels from that shown in Figure 21-27; nonetheless, they all contain four copies of the highly conserved P segment flanked by homologs of the S5 and S6 transmembrane a helices.

Although the density of voltage-gated K⁺ and Na⁺ channels is very low, both types of channel proteins have been isolated. As we saw earlier, isolation of a K⁺ channel from Drosophila was greatly aided by discovery of shaker mutants. Likewise, certain neurotoxins facilitated the purification and study of Na⁺ channels. For example, tetrodotoxin produced by the puffer fish and saxitoxin produced by certain red marine dinoflagellates specifically bind to and inhibit the voltage-gated Na⁺ channels in neurons, preventing action potentials from forming. One molecule of either toxin binds to one Na⁺ channel with exquisite affinity and selectivity. Measurements of the amount of radioactive tetrodotoxin or saxitoxin that binds to a typical unmyelinated invertebrate axon have shown that it contains 5500 Na⁺ channels per square micrometer of membrane, equivalent to about 1 part per million of total plasma-membrane protein molecules. This value agrees with the number of channels estimated from patch-clamp studies. The Na⁺ channels in these unmyelinated membranes are thus spaced, on average, about 200 nm apart. However, as noted earlier, they are packed much closer together at the nodes of Ranvier.

The structural and functional similarities among voltage-gated Na⁺, Ca²⁺, and K⁺ channels suggest that all three proteins evolved from a common ancestral gene. The distribution of these channels among different organisms provides clues about a likely evolutionary pathway.

Voltage-gated K⁺ channels have been found in all yeasts and protozoa studied, whereas only multicellular organisms have voltage-gated Na⁺ channels. Intermediate in distribution are voltage-gated Ca²⁺ channels, which function in synaptic transmission (discussed below): These are present in multicellular organisms but in only a few of the more complex protozoa, such as Paramecium. Thus voltage-gated K⁺ channel proteins probably arose first in evolution. The voltage-gated Ca²⁺ and Na⁺ channel proteins are believed to have evolved by repeated duplication of an ancestral one-domain K⁺-channel gene. Since all K⁺ channels share a similar pore-lining P segment, it is probable that voltage-gated and non-voltage-gated K⁺ channels also evolved from a common progenitor.

• Patch-clamping techniques, which permit measurement of ion movement through single channels, are used to determine the potential at which a channel opens, its ion conductivity, and the rate of channel inactivation and closing (see Figure 21-21).
• Recombinant DNA techniques allow the expression and study of cloned channel proteins in frog oocytes and other types of cells (see Figure 21-23).
• Voltage-gated K⁺ channels, such as the Shaker protein from Drosophila, are assembled from four similar subunits, each of which has six membrane-spanning a helices and a nonhelical P segment that lines the ion pore. The S4 a helix in each subunit acts as a voltage sensor.
• Voltage-gated Na⁺ and Ca²⁺ channels are monomeric proteins containing four homologous domains each similar to a K⁺-channel subunit (see Figure 21-27).
• The ion specificity of channel proteins is due mainly to coordination of the selected ion with specific residues in the P segments, thus lowering the activation energy for passage of the selected ion compared with other ions (see Figure 21-25).
• Voltage-sensing a helices have a positively charged lysine or arginine every third or fourth residue. Their movement outward across the membrane, in response to a membrane depolarization of sufficient magnitude, causes opening of the channel (see Figure 21-13).
• Voltage-gated K⁺, Na⁺, and Ca²⁺ channel proteins contain one or more cytosolic domains that move into the open channel thereby inactivating it (see Figure 21-26).
• Non-voltage-gated K⁺ channels and nucleotide-gated channels lack a voltage-sensing a helix, but otherwise their structures are similar in many respects to voltage-gated K⁺ channels.
• Most likely, voltage-gated channel proteins and possibly all K⁺ channels evolved from a common ancestral gene.