5: Ion Channels

• Rapid Signaling in the Nervous System Depends on Ion Channels?

• Ion Channels Are Proteins That Span the Cell Membrane

• Currents Through Single Ion Channels Can Be Recorded

• Ion Channels in All Cells Share Several Characteristics

  • The Flux of Ions Through a Channel Is Passive

  • The Opening and Closing of a Channel Involve Conformational Changes

• The Structure of Ion Channels Is Inferred from Biophysical, Biochemical, and Molecular Biological Studies

  • Ion Channels Can Be Grouped into Gene Families

  • The Closed and Open Structures of Potassium Channels Have Been Resolved by X-Ray Crystallography

  • The Structural Basis of Chloride Selectivity Reveals a Close Relation Between Ion Channels and Ion Transporters

• An Overall View

Signaling in the brain depends on the ability of nerve cells to respond to very small stimuli with rapid and large changes in the electrical potential difference across the cell membrane. In sensory cells, the membrane potential changes in response to minute physical stimuli: Receptors in the eye respond to a single photon of light; olfactory neurons detect a single molecule of odorant; and hair cells in the inner ear respond to tiny movements of atomic dimensions. These sensory responses ultimately lead to the firing of an action potential during which the membrane potential changes up to 500 volts per second.

The rapid changes in membrane potential that underlie signaling throughout the nervous system are mediated by ion channels, a class of integral membrane proteins found in all cells of the body. The ion channels of nerve cells are optimally tuned to respond to specific physical and chemical signals. They are also heterogeneous—in different parts of the nervous system different types of channels carry out specific signaling tasks.

Because of their key roles in electrical signaling, malfunctioning of ion channels can cause a wide variety of neurological diseases (see Chapter 14). Diseases caused by ion channel malfunction are not limited to the brain; cystic fibrosis, skeletal muscle disease, and certain types of cardiac arrhythmia, for example, are also caused by ion channel malfunction. Moreover, ion channels are often the site of action of drugs, poisons, or toxins. Thus ion channels have crucial roles in both the normal physiology and pathophysiology of the nervous system.

In addition to ion channels, nerve cells contain a second important class of proteins specialized for moving ions across cell membranes, the ion transporters or pumps. These proteins do not participate in rapid neuronal signaling but rather are important for establishing and maintaining the concentration gradients of physiologically important ions between the inside and outside of the cell. As we will see in this chapter and Chapter 6, ion transporters differ in important aspects from ion channels, but also share certain common features.

Ion channels have three important properties: (1) They recognize and select specific ions, (2) they open and close in response to specific electrical, mechanical, or chemical signals, and (3) they conduct ions across the membrane. The channels in nerve and muscle conduct ions across the cell membrane at extremely rapid rates, thereby providing a large flow of electric charge. Up to 100 million ions can pass through a single channel each second. This current causes the rapid changes in membrane potential required for signaling, as will be discussed in Chapter 7. The fast flow of ions through channels is comparable to the turnover rate of the fastest enzymes, catalase and carbonic anhydrase, which are limited by diffusion of substrate. (The turnover rates of most other enzymes are considerably slower, ranging from 10 to 1,000 per second.)

In light of such an extraordinary rate of ion flow, channels are surprisingly selective for the ions they allow to permeate. Each type of channel allows only one or a few types of ions to pass. For example, the negative resting potential of nerve cells is largely determined by a class of K⁺ channels that are 100-fold more permeable to K⁺ than to Na⁺. In contrast, during the action potential a class of Na⁺ channels is activated that are 10- to 20-fold more permeable to Na⁺ than to K⁺. Thus a key to the great versatility of neuronal signaling is the regulated activation of different classes of ion channels, each of which is selective for specific ions.

Many channels open and close in response to a specific event: Voltage-gated channels are regulated by changes in membrane potential, ligand-gated channels by chemical transmitters, and mechanically gated channels by pressure or stretch. However, some channels are normally open in the cell at rest. The ion flux through these "resting" channels contributes significantly to the resting potential.

Ion channels are limited to catalyzing the passive movement of ions down their thermodynamic concentration and electrical gradients. For example, Na⁺ ions enter a cell through voltage-gated Na⁺ channels during an action potential because the external Na⁺ concentration is much greater than the internal concentration; the open channels allow Na⁺ to diffuse into the cell down its concentration gradient. With such passive ion movement the Na⁺ concentration gradient would eventually dissipate were it not for ion pumps. Different types of ion pumps maintain the concentration gradients for Na⁺, K⁺, Ca²⁺, Cl⁻, and other ions.

These pumps differ from ion channels in two important details. First, whereas open ion channels have a continuous water-filled pathway through which ions flow unimpeded from one side of the membrane to the other, each time a pump moves an ion, or a group of a few ions, across the membrane, it must undergo a series of conformational changes. As a result, the rate of ion flow through pumps is 100 to 100,000 times slower than through channels. Second, pumps that maintain ion gradients use energy, often in the form of adenosine triphosphate (ATP), to transport ions against their electrical and chemical gradients. Such ion movements are termed active transport. The function and structure of ion pumps is considered in detail at the end of this chapter and in Chapter 6.

In this chapter we examine four questions: Why do nerve cells have channels? How can channels conduct ions at such high rates and still be selective? How are channels gated? How are the properties of these channels modified by various intrinsic and extrinsic conditions? In addition, we compare the molecular structure of various channels and consider how channel structure explains function. Finally, we compare ion movements through channels and transporters. In succeeding chapters we consider how resting channels and pumps generate the resting potential (Chapter 6), how voltage-gated channels generate the action potential (Chapter 7), and how ligand-gated channels produce synaptic potentials (Chapters 8, 9, and 10).

To appreciate why nerve cells use channels, we need to understand the nature of the plasma membrane and the physical chemistry of ions in solution. The plasma membrane of all cells, including nerve cells, is approximately 6 to 8 nm thick and consists of a mosaic of lipids and proteins. The core of the membrane is formed by a double layer of phospholipids. Embedded within this continuous lipid sheet are integral membrane proteins, including ion channels.

The lipids of the membrane do not mix with water—they are hydrophobic. In contrast, the ions within the cell and those outside strongly attract water molecules—they are hydrophilic (Figure 5–1). Ions attract water because water molecules are dipolar: Although the net charge on a water molecule is zero, charge is separated within the molecule. The oxygen atom in a water molecule tends to attract electrons and so bears a small net negative charge, whereas the hydrogen atoms tend to lose electrons and therefore carry a small net positive charge. As a result of this unequal distribution of charge, positively charged ions (cations) are strongly attracted electrostatically to the oxygen atom of water, and negatively charged ions (anions) are attracted to the hydrogen atoms. Similarly, ions attract water; in fact they become surrounded by electrostatically bound waters of hydration (see Figure 5–1).

An ion cannot move away from water into the noncharged hydrocarbon tails of the lipid bilayer in the membrane unless a large amount of energy is expended to overcome the attraction between the ion and the surrounding water molecules. For this reason it is extremely unlikely that an ion will move from solution into the lipid bilayer, and therefore the bilayer itself is almost completely impermeable to ions. Rather, ions cross the membrane through ion channels, specialized pores or openings in the membrane where, as we shall see, the energetics favor ion movement.

Ion channels are not simply holes in the lipid membrane but are distinct protein structures that span the lipid bilayer. Although their molecular nature has been known with certainty for only approximately 25 years, the idea of ion channels dates to the work of Ernst Brücke at the end of the 19th century. Physiologists had long known that, despite the fact that the cell membrane acts as a barrier, cell membranes are nevertheless permeable to water and many small solutes, including some ions. To explain osmosis, the flow of water across biological membranes, Brücke proposed that membranes contained channels or pores that allow water but not larger solutes to flow. Over 100 years later Peter Agre found that a family of proteins termed aquaporins form channels with a high selective permeability to water. At the beginning of the 20th century William Bayliss suggested that water-filled channels would permit ions to cross the cell membrane easily, as the ions would not need to be stripped of their waters of hydration.

The idea that ions move through channels leads to a question: How can a water-filled channel conduct ions at high rates and yet be selective? How, for instance, does a channel allow K⁺ ions to pass while excluding Na⁺ ions? Selectivity cannot be based solely on the diameter of the ion because K⁺, with a crystal radius of 0.133 nm, is larger than Na⁺ (crystal radius of 0.095 nm). Because ions in solution are surrounded by waters of hydration, the ease with which an ion moves in solution (its mobility or diffusion constant) depends on the size of the ion together with the shell of water surrounding it.

The smaller an ion, the more highly localized is its charge and the stronger its electric field. As a result, smaller ions attract water more strongly. Thus, as Na⁺ moves through solution its stronger electrostatic attraction for water causes it to have a larger water shell, which tends to slow it down relative to K⁺. Because of its larger water shell, Na⁺ behaves as if it is larger than K⁺. In fact, there is a precise relationship between the size of an ion and its mobility in solution: the smaller the ion, the lower its mobility. We therefore can construct a model of a channel that selects K⁺ rather than Na⁺ simply on the basis of the interaction of the two ions with water in a water-filled channel (Figure 5–1).

Although this model explains how a channel can select K⁺ and exclude Na⁺, it does not explain how a channel could select Na⁺ and exclude K⁺. Moreover, the model cannot account quantitatively for the very high ionic selectivity exhibited by biological K⁺ channels. This problem led many physiologists in the 1930s and 1940s to abandon the channel theory in favor of the idea that ions cross cell membranes by first binding to a specific carrier protein, which then shuttles the ion through the membrane. In this carrier model selectivity is based on the chemical binding between the ion and the carrier protein, not on the mobility of the ion in solution.

Even though we now know that ions can cross membranes by means of a variety of transport macromolecules, the Na⁺ -K⁺ pump being a well-characterized example (Chapter 6), many properties of membrane ion conductances do not fit the carrier model. Most important is the rapid rate of ion transfer across membranes. This transfer rate was first examined in the early 1970s by measuring the transmembrane current initiated when the neurotransmitter acetylcholine (ACh) binds its receptor in the cell membrane of skeletal muscle fibers at the synapse between nerve and muscle. Using measurements of membrane current noise, small statistical fluctuations in the mean ionic current induced by ACh, Bernard Katz and Ricardo Miledi concluded that the current conducted by a single ACh receptor is 10 million ions per second. In contrast, the Na⁺ -K⁺ pump transports at most 100 ions per second.

If the ACh receptor acted as a carrier, it would have to shuttle an ion across the membrane in 0.1 μs (one ten-millionth of a second), an implausibly fast rate. The 100,000-fold difference in rates strongly suggests that the ACh receptor (and other ligand-gated receptors) must conduct ions through a channel. Later measurements in many voltage-gated pathways selective for K⁺, Na⁺, and Ca²⁺ also demonstrated large currents carried by single macromolecules, indicating that they too are channels.

But we are still left with the problem of what makes a channel selective. To explain selectivity, Bertil Hille extended the pore theory by proposing that channels have narrow regions that act as molecular sieves. At this selectivity filter an ion must shed most of its waters of hydration to traverse the channel; in their place weak chemical bonds (electrostatic interactions) form with polar (charged) amino acid residues that line the walls of the channel (Figure 5–1). Because shedding its waters of hydration is energetically unfavorable, the ion will traverse a channel only if its energy of interaction with the selectivity filter compensates for the loss of the energy of interaction with its waters of hydration. Ions traversing the channel are normally bound to the selectivity filter for only a short time (less than 1 μs), after which electrostatic and diffusional forces propel the ion through the channel. In channels where the pore diameter is large enough to accommodate several water molecules, an ion need not be stripped completely of its water shell.

How is this chemical recognition and specificity established? One theory was developed in the early 1960s by George Eisenman to explain the properties of ion-selective glass electrodes (which are similar to pH electrodes but select among the alkali metal cations). According to this theory, a binding site with a high negative field strength—for example, one formed by negatively charged carboxylic acid groups of glutamate or aspartate—will bind Na⁺ more tightly than K⁺. This selectivity results because the electrostatic interaction between two charged groups, as governed by Coulomb's law, depends inversely on the distance between the two groups.

Because Na⁺ has a smaller crystal radius than K⁺, it will approach a negative site more closely than K⁺ and thus will derive a more favorable free-energy change on binding. This compensates for the requirement that Na⁺ lose some of its waters of hydration to traverse the narrow selectivity filter. In contrast, a binding site with a low negative strength—one that is composed, for example, of polar carbonyl or hydroxyl oxygen atoms—would select K⁺ over Na⁺. At such a site the binding of Na⁺ would not provide a sufficient free-energy change to compensate for the loss of the ion's waters of hydration, which Na⁺ holds strongly. However, such a site would be able to compensate for the loss of a K⁺ ion's associated water molecules as the larger K⁺ ions interact more weakly with water. It is currently thought that ion channels are selective both because of specific chemical interactions and because of molecular sieving based on pore diameter.

A complete understanding of how channels work requires three-dimensional structural information. X-ray crystallographic and other structural analyses have been informative in the study of enzymes and other soluble proteins but have only recently been applied to integral membrane proteins, such as ion channels, because their transmembrane hydrophobic regions make them difficult to crystallize. Nevertheless, for the past 35 years single-channel recording has provided important functional information that has yielded significant insights into channel structure.

Before it became possible to resolve the small amount of current that flows through a single ion channel in biological membranes, channel function was studied in artificial lipid bilayers. In the early 1960s Paul Mueller and Donald Rudin developed a technique for forming functional lipid bilayers by painting a thin drop of phospholipid over a hole in a nonconducting barrier that separates two salt solutions. Although lipid membranes are highly impermeable to ions, ionic permeability of the membrane increases dramatically when certain peptide antibiotics are added to the salt solution. Early studies with gramicidin A, a 15-amino acid cyclic peptide, were especially informative. Application of low concentrations of gramicidin A brings about small step-like changes in current across the membrane. These brief pulses of current reflect the all-or-none opening and closing of the single ion channel formed by the peptide.

The current through a single gramicidin channel varies with membrane potential in a linear manner; that is, the channel behaves as a simple resistor (Figure 5–2). The amplitude of the single-channel current can thus be obtained from Ohm's law, i = V/R, where i is the current through the single channel, V is the voltage across the membrane, and R is the resistance of the open channel. The resistance of a single open channel is approximately 8 × 10¹⁰ ohms (Figure 5–2C). In dealing with ion channels it is more useful to speak of the reciprocal of resistance or conductance (γ = 1/R), as this provides an electrical measure related to ion permeability. Thus, Ohm's law can be expressed as i = γ × V. The conductance of the gramicidin A channel is approximately 12 × 10⁻¹² siemens (S), or 12 picosiemens (pS), where 1 S equals 1/ohm.

The insights into basic channel properties obtained from artificial membranes were later confirmed in biological membranes by the patch-clamp technique, developed by Erwin Neher and Bert Sakmann in 1976 (Box 5–1). A glass micropipette containing ACh—the neurotransmitter that activates the ACh receptors in the membrane of skeletal muscle—was pressed tightly against a frog muscle membrane. Small unitary current pulses representing the opening and closing of a single ACh receptor channel were recorded from the area of the membrane under the pipette tip. As with gramicidin A channels, the relation between current and voltage in the ACh receptor channel is linear (Figure 5–2), with a single-channel conductance of approximately 25 pS. This generates a unitary current of 2 pA (picoamperes) at a membrane potential of −80 mV (millivolts), which corresponds to a flux of around 12.5 million ions per second.

Most cells are capable of local signaling, but only nerve and muscle cells are specialized for rapid signaling over long distances. Although nerve and muscle cells have a particularly rich variety and high density of membrane ion channels, their channels do not differ fundamentally from those of other cells in the body. In this section we describe the general properties of ion channels found in a wide variety of cells.

The Flux of Ions Through a Channel Is Passive

The flux of ions through ion channels is passive, requiring no expenditure of metabolic energy by the channels. The direction and eventual equilibrium for this flux are determined not by the channel itself, but rather by the electrostatic and diffusional driving forces across the membrane.

Ion channels allow particular types of ions to cross the membrane, selecting either cations or anions to permeate. Some types of cation-selective channels allow the cations that are usually present in extracellular fluid—Na⁺, K⁺, Ca²⁺, and Mg²⁺—to pass almost indiscriminately. However, many other cation- selective channels are permeable primarily to a single type of ion, whether it is Na⁺, K⁺, or Ca²⁺. Most types of anion-selective channels are also highly discriminating; they conduct only one physiological ion, chloride (Cl⁻) .

The kinetic properties of ion permeation are best described by the channel's conductance, which is determined by measuring the current (ion flux) through the open channel in response to an electrochemical driving force. The net electrochemical driving force is determined by two factors: the electrical potential difference across the membrane and the concentration gradients of the permeant ions across the membrane. Changing either one can change the net driving force (see Chapter 6).

As we have seen, in some open channels the current varies linearly with driving force—that is, the channels behave as simple resistors. In others the current is a nonlinear function of driving force. This type of channel behaves as a rectifier—it conducts ions more readily in one direction than in the other because of asymmetry in the channel's structure or environment. Whereas the conductance of a resistor-like channel is constant—it is the same at all voltages—the conductance of a rectifying channel is variable and must be determined by plotting current versus voltage over the entire physiological range of membrane potential (Figure 5–4).

The rate of net ion flux (current) through a channel depends on the concentration of the permeant ions in the surrounding solution. At low concentrations the current increases almost linearly with concentration. At higher concentrations the current tends to reach a point at which it no longer increases. At this point the current is said to saturate.

This saturation effect is consistent with the idea that ion flux across the cell membrane is not strictly a function of laws of electrochemical diffusion in free solution but also involves the binding of ions to specific polar sites within the pore of the channel. A simple electrodiffusion model would predict that the ionic current should continue to increase as long as the ionic concentration also increases; that is, the more charge carriers in solution, the greater the current.

The relation between current and ionic concentration for a wide range of ion channels is well described by a simple chemical binding equation, identical to the Michaelis-Menten equation for enzymes, suggesting that a single ion binds within the channel during permeation. The ionic concentration at which current reaches half its maximum defines the dissociation constant, the concentration at which half of the channels will be occupied by a bound ion. The dissociation constant in plots of current versus concentration is typically quite high, approximately 100 mM, indicating weak binding. (In typical interactions between enzymes and substrates the dissociation constant is below 1 μM.) This weak interaction indicates that the bonds between the ion and the channel are rapidly formed and broken. In fact, an ion typically stays bound in the channel for less than 1 μs. The rapid rate at which an ion unbinds is necessary for the channel to achieve the very high conduction rates responsible for the rapid changes in membrane potential during signaling.

Some ion channels are susceptible to occlusion by certain free ions or molecules in the cytoplasm or extracellular fluid. Passage through the channel can be blocked by particles that bind either to the mouth of the aqueous pore or somewhere within the pore. If the blocker is an ionized molecule that binds to a site within the pore, it will be influenced by the membrane electric field as it enters the channel. For example, if a positively charged blocker enters the channel from outside the membrane, then making the cytoplasmic side of the membrane more negative—which, according to convention, corresponds to a more negative membrane potential (see Chapter 6)—will drive the blocker into the channel through an electrostatic attraction, increasing the block. Although blockers are often toxins or drugs that originate outside the body, some are common ions that are normally present in the cell or its environment. Physiological blockers of certain classes of channels include Mg²⁺, Ca²⁺, Na⁺, and polyamines such as spermine.

The Opening and Closing of a Channel Involve Conformational Changes

In all ion channels so far studied the channel protein has two or more conformational states that are relatively stable. Each of these stable conformations represents a different functional state. For example, each ion channel has at least one open state and one or two closed states. The transition of a channel between these different states is called gating.

Relatively little is known about the molecular mechanisms of gating. Although the picture of a gate swinging open and shut is a convenient image, it probably is accurate only for certain cases (for example the inactivation of Na⁺ and K⁺ channels, which we describe in Chapter 7). More commonly, channel gating involves widespread changes in the channel's conformation. For example, evidence from high-resolution electron microscopy and image analysis suggests that the opening and closing of gap junction channels (which we consider in Chapter 8) involve a concerted twisting and tilting of the six subunits that make up the channel. Similar evidence indicates that ACh receptor channels open through a coordinated twisting and bending of the α-helixes of each of the five subunits that form the channel pore. In K⁺ channels movement of a ring of α-helixes that forms the internal mouth of the pore is thought to act as a gate. The molecular rearrangements that occur during the transition from closed to open states appear to enhance ion conduction through the channel not only by creating a wider lumen, but also by positioning relatively more polar amino acid constituents at the surface that lines the aqueous pore.

The primary function of ion channels in neurons is to generate transient electrical signals, and three major gating mechanisms have evolved to control channel opening (Figure 5–5). Certain channels are opened by the binding of chemical ligands, known as agonists. Some ligands bind directly to the channel either at an extracellular or intracellular site; transmitters bind at extracellular sites whereas certain cytoplasmic constituents, such as Ca²⁺, cyclic nucleotides and GTP-binding proteins, bind at intracellular sites. Other ligands activate cellular signaling cascades, which can covalently modify a channel through protein phosphorylation. Other ion channels are regulated by changes in membrane potential. Some voltage-gated channels act as temperature sensors; changes in temperature shift their voltage gating to higher or lower membrane potentials, giving rise to heat- or cold-sensitive channels. Finally, some channels are regulated by mechanical stretch of the membrane (Figure 5–6).

The rapid gating actions necessary for moment-to-moment signaling may also be influenced by certain long-term changes in the metabolic state of the cell. For example, the gating of some K⁺ channels is sensitive to intracellular levels of ATP. Some channel proteins contain a subunit with an integral oxidoreductase catalytic domain that is thought to alter channel gating in response to the redox state of the cell.

These regulators control the entry of a channel into one of three functional states: closed and activatable (resting), open (active), or closed and nonactivatable (refractory). A change in the functional state of a channel requires energy. In voltage-gated channels the energy is provided by the movement of a charged region of the channel protein through the membrane's electric field. This region, the voltage sensor, contains a net electric charge because of the presence of basic (positively charged) or acidic (negatively charged) amino acids. The movement of the charged voltage sensor through the electric field imparts a change in free energy to the channel that alters the equilibrium between the closed and open states of the channel. For most voltage-gated channels, channel opening is favored by making the inside of the membrane more positive (depolarization).

In transmitter-gated channels gating is driven by the change in chemical-free energy that results when the transmitter binds to a receptor site on the channel protein. For mechanically activated channels the energy associated with membrane stretch is thought to be transferred to the channel either through the cytoskeleton or more directly by changes in tension of the lipid bilayer.

The stimuli that gate the channel also control the rates of transition between the open and closed states of a channel. For voltage-gated channels the rates are steeply dependent on membrane potential. Although the time scale can vary from several microseconds to a minute, the transition tends to require a few milliseconds on average. Thus, once a channel opens it stays open for a few milliseconds before closing, and after closing it stays closed for a few milliseconds before reopening. Once the transition between open and closed states begins, it proceeds virtually instantaneously (in less than 10 μs, the present limit of experimental measurements), thus giving rise to abrupt, all-or-none, step-like changes in current through the channel.

Ligand-gated and voltage-gated channels enter refractory states through different processes. Ligand-gated channels can enter the refractory state when their exposure to the agonist is prolonged. This process, called desensitization, is discussed in Chapter 11. The mechanisms underlying desensitization of ion channels are not yet completely understood. In some channels desensitization appears to be an intrinsic property of the interaction between ligand and channel, although in others it is a result of phosphorylation of the channel molecule by a protein kinase.

Many, but not all, voltage-gated channels enter a refractory state after opening, a process termed inactivation. In the inactive state the channel is closed and can no longer be opened by positive voltages. Rather, the membrane potential must be returned to its initial negative resting level before the channel can recover from inactivation so that it can again open in response to depolarization. In voltage-gated Na⁺ and K⁺ channels inactivation is thought to result from a conformational change, controlled by a subunit or region of the channel separate from that which controls activation. In contrast, the inactivation of certain voltage-gated Ca²⁺ channels is thought to require Ca²⁺ influx. An increase in internal Ca²⁺ concentration inactivates the Ca²⁺ channel by binding to the regulatory molecule calmodulin, which is permanently associated with the Ca²⁺ channel protein (Figure 5–7).

Exogenous factors, such as drugs and toxins, can also affect the gating control sites of an ion channel. Most of these agents tend to close the channel but a few open it. Some bind to the same site at which the endogenous agonist normally binds and thereby interfere with normal gating. The binding can be weak and reversible, as in the blockade of the nicotinic ACh-gated channel in skeletal muscle by curare, a South American arrow poison (see Chapter 9). Or it can be strong and irreversible, as in the blockade of the same channel by the snake venom α-bungarotoxin.

Some exogenous substances act in a noncompetitive manner and affect the normal gating mechanism without directly interacting with the transmitter- binding site. For example, binding of the drug diazepam (Valium) to a regulatory site on Cl⁻ channels that are gated by γ-aminobutyric acid (GABA), an inhibitory neurotransmitter, prolongs the opening of the channels in response to GABA (Figure 5–8). This type of indirect, allosteric effect is found in voltage- or stretch-gated channels as well.

What do ion channels look like? How does the channel protein span the membrane? What happens to the structure of the channel when it opens and closes? Where along the length of the channel protein do drugs and transmitters bind?

Biochemical and molecular biological studies have provided a basic understanding of channel structure and function. All ion channels are large integral-membrane proteins with a core transmembrane domain that contains a central aqueous pore spanning the entire width of the membrane. The channel protein often contains carbohydrate groups attached to its external surface. The pore-forming region of many channels is made up of two or more subunits, which may be identical or different. In addition, some channels have auxiliary subunits that modify their functional properties. These subunits may be cytoplasmic or embedded in the membrane (Figure 5–9).

The genes for all the major classes of ion channels have been cloned and sequenced. The amino acid sequence of a channel, inferred from its DNA sequence, can be used to create a structural model of the channel protein. Regions of secondary structure—the arrangement of the amino acid residues into α-helixes and β-sheets—as well as regions that are likely to correspond to membrane-spanning domains of the channel are predicted based on the structures of related proteins that have been experimentally determined using electron and X-ray diffraction analysis. This type of analysis identified the presence of four hydrophobic regions, each around 20 amino acids in length, in the amino acid sequence of a subunit of the ACh receptor channel. Each of these regions is thought to form an α-helix that spans the membrane (Figure 5–10).

Additional insights into channel structure and function have been obtained by comparing the amino acid sequences of the same type of channel from different species. Regions that show a high degree of similarity (that is, have been highly conserved through evolution) are likely to be important in maintaining the structure and function of the channel. Likewise, conserved regions in different but related channels are likely to serve a common biophysical function. For example, voltage-gated channels selective for different cations have a specific membrane-spanning segment that contains positively charged amino acids (lysine or arginine) spaced at every third position along an α-helix. This motif is observed in all voltage-gated cation channels, but not in transmitter-gated channels, suggesting that this charged region is important for voltage gating (see Chapter 7).

Once a structure for a channel has been proposed, it can be tested in several ways. For example, antibodies can be raised against synthetic peptides that correspond to different hydrophilic regions in the protein sequence. Immunocytochemistry can then be used to determine whether the antibody binds to the extracellular or cytoplasmic surface of the membrane, thus defining whether a particular region of the channel is extracellular or intracellular.

The functional consequences of changes in a channel's primary amino acid sequence can be explored through a variety of techniques. One particularly versatile approach is to use genetic engineering to construct channels with parts that are derived from the genes of different species—so-called chimeric channels. This technique takes advantage of the fact that the same type of channel can have somewhat different properties in different species. For example, the bovine ACh receptor channel has a slightly greater single-channel conductance than the ACh receptor channel in electric fish. By comparing the properties of a chimeric channel to those of the two original channels, one can assess the functions of specific regions of the channel. This technique has been used to identify the membrane-spanning segment that forms the lining of the pore of the ACh receptor channel (see Chapter 9).

The roles of different amino acid residues or stretches of residues can be tested using site-directed mutagenesis, a type of genetic engineering in which specific amino acid residues are substituted or deleted. Finally, one can exploit the naturally occurring mutations in channel genes. A number of inherited and spontaneous mutations in the genes that encode ion channels in nerve or muscle produce changes in channel function that can underlie certain neurological diseases. Many of these mutations are caused by localized changes in single amino acids within channel proteins, demonstrating the importance of that region for channel function.

Ion Channels Can Be Grouped into Gene Families

The great diversity of ion channels in a multicellular organism is underscored by the recent sequencing of the human genome. Our genome contains nine genes encoding variants of voltage-gated Na⁺ channels, 10 genes for different Ca⁺ channels, over 75 genes for K⁺ channels, 70 genes for ligand-gated channels, and more than a dozen genes for Cl⁻ channels. Fortunately, the evolutionary relationships between the genes that encode ion channels provide a relatively simple framework with which to categorize them.

Most of the ion channels that have been described in nerve and muscle cells fall into a few gene superfamilies. Members of each gene superfamily have similar amino acid sequences and transmembrane topology and, importantly, related functions. Each superfamily is thought to have evolved from a common ancestral gene by gene duplication and divergence. Several superfamilies can be further classified into families of genes with more closely related structure and function.

One superfamily encodes ligand-gated ion channels that are receptors for the neurotransmitters ACh, GABA, glycine, or serotonin. All of these receptors are composed of five subunits, each of which has four transmembrane α-helixes (Figure 5–11A). In addition, the extracellular domain that forms the receptor for the ligand contains a conserved loop of 13 amino acids flanked by pair of cysteine residues that form a disulfide bond. As a result, this receptor superfamily is referred to as the cys-loop receptors. Ligand-gated channels can be classified by their ion selectivity in addition to their agonist. The genes that encode glutamate receptor channels belong to a separate gene family.

Gap-junction channels, which bridge the cytoplasm of two cells at electrical synapses (see Chapter 8), are encoded by a separate gene superfamily. A gap-junction channel is composed of 12 identical subunits, each of which has four membrane-spanning segments (Figure 5–11B).

The genes that encode the voltage-gated ion channels responsible for generating the action potential belong to another family. These channels are usually activated by membrane depolarization and are selective for Ca²⁺, Na⁺, or K⁺. All voltage-gated channels have a similar architecture, with a core motif composed of six transmembrane segments termed S1–S6 (Figure 5–11C). The S5 and S6 segments are connected by a highly conserved P-region, which loops into and out of the extracellular face of the membrane and forms the selectivity filter of the channel. Voltage-gated Na⁺ and Ca²⁺ channels are composed of a large subunit that contains four repeats of this basic motif. Voltage-gated K⁺ channels are composed of four separate subunits, each containing one motif. This structure is shared by other, more distantly related channel families (see Chapter 7).

The major gene family encoding the voltage-gated K⁺ channels is distantly related to two families of K⁺ channels, each with distinctive properties and structure (Figure 5–12). One family consists of the genes encoding inward-rectifying K⁺ channels, which are open at the resting potential and close rapidly during depolarization. Each channel subunit has only two transmembrane segments connected by a pore-forming P-region. A second family is composed of subunits with two repeated pore-forming segments. These channels may also contribute to the resting K⁺ conductance.

The sequencing of the genomes of a variety of species has led to the identification of additional ion channel gene families, found in organisms from bacteria to humans. Channels with related P-regions have been identified that are only very distantly related to the family of voltage-gated channels. These channels include the glutamate-gated channels, in which the P-region is inverted: It enters and leaves the internal surface of the membrane (see Figure 5–12D). Finally, the transient receptor potential (TRP) family of nonselective cation channels (named after a mutant Drosophila strain in which light evokes an abnormal transient receptor potential in photoreceptors) comprises a very large group of channels that contain a P-region. Like the voltage-gated K⁺ channels, TRP channels also contain six transmembrane segments but are usually gated by intracellular ligands. TRP channels are important for Ca²⁺ metabolism in all cells, visual signaling in insects, and pain, heat, and cold sensation in the nervous system of higher animals. Recent evidence implicates TRP channels in mechanosensation and hearing in insects and fish and in certain taste sensations in mammals.

A number of other families of channels have been identified, distinct from those considered above. These include Cl⁻ channels that help set the resting potential of certain nerve and skeletal muscle cells, and a class of ligand-gated cation channels activated by ATP, which functions as a neurotransmitter at certain synapses. With the completion of the human genome project, it is likely that nearly all of the major classes of ion channel genes have now been identified.

However, the diversity of ion channels is even greater than the large number of ion channel genes. Because most channels in a given subfamily are composed of multiple subunits, each of which may be encoded by a family of closely related genes, combinatorial permutations of these subunits can generate a diverse array of heteromultimeric channels with different functional properties. Further diversity can be produced by alternative splicing of precursor mRNA transcribed from a single gene. Finally, the sequence of a transcript can be altered by a process termed RNA editing. As with enzyme isoforms, variants of a channel with slightly different properties may be expressed at distinct stages of development (Figure 5–13), in different cell types throughout the brain (Figure 5–14), and even in different regions within a cell. These subtle variations in structure and function presumably allow channels to perform highly specific functions.

The rich variety of ion channels in different types of neurons may make it possible to develop drugs that can activate or block channels in selected regions of the nervous system. Such drugs would, in principle, have maximum therapeutic effectiveness with minimum side effects.

The Closed and Open Structures of Potassium Channels Have Been Resolved by X-Ray Crystallography

Rod MacKinnon and his colleagues provided the first high-resolution X-ray crystallographic analysis of the molecular architecture of an ion-selective channel. To overcome the difficulties inherent in obtaining crystals of integral membrane proteins, they initially focused on a bacterial K⁺ channel, termed KcsA. These channels were useful for crystallography as they can be expressed at high levels for purification, are relatively small, and have a simple transmembrane topology that is similar to inward-rectifier K⁺ channels present in higher organisms, including mammals (see Figure 5–12B). The structure of the channel was further simplified using molecular engineering to truncate cytoplasmic regions that are not essential for forming the ion-selective pore.

The crystal structure determined from the modified KcsA protein provides several important insights into the mechanisms by which the channel facilitates the movement of K⁺ ions across the hydrophobic lipid bilayer. The channel is made up of four identical subunits arranged symmetrically around a central pore (Figure 5–15A). Each subunit has two membrane-spanning α-helixes, an inner and outer helix, that are connected by the P-loop, which forms the selectivity filter of the channel. At the extracellular end of the channel the two α-helixes tilt away from the central axis of the pore so that the structure resembles an inverted teepee.

The four inner α-helixes from each of the subunits line the cytoplasmic end of the pore. At the intracellular mouth of the channel these four helixes cross, forming a very narrow opening—the "smoke hole" of the teepee. The inner helixes are homologous to the S6 membrane-spanning segment of voltage-gated K⁺ channels. At the extracellular end of the channel the pair of transmembrane helixes from each subunit are connected by a region consisting of three elements: (1) a chain of amino acids that surrounds the mouth of the channel (the turret region); (2) an abbreviated α-helix (the pore helix) approximately 10 amino acids in length that projects toward the central axis of the pore; and (3) a stretch of 5 amino acids near the C-terminal end of the P-region that forms the selectivity filter (Figure 5–15B).

The shape and structure of the pore determine its ion-conducting properties. Both the inner and outer mouths of the pore are lined by acidic amino acids with negative charges that help attract cations from the bulk solution. Going from inside to outside, the pore consists of a medium wide tunnel 18 Å in length that leads into a wider (10 Å diameter) spherical inner chamber (Figure 5–15D). This chamber is lined predominantly by the side chains of hydrophobic amino acids. These relatively wide regions are followed by the very narrow selectivity filter, only 12 Å in length, which is rate-limiting for the passage of ions. A high ion throughput rate is ensured by the fact that the inner 28 Å of the pore, from the cytoplasmic entrance to the selectivity filter, lacks polar groups that could delay ion passage by binding and unbinding the ion.

An ion passing from the polar solution through the nonpolar lipid bilayer encounters the least energetically favorable region in the middle of the bilayer. The high energetic cost for a K⁺ ion to enter this region is minimized by two details of channel structure. The inner chamber is filled with water, which provides a highly polar environment, and the pore helixes provide a dipole whose electronegative carboxyl ends point toward this inner chamber (Figures 5–15C and 5–18A).

The high energetic cost incurred as a K⁺ ion sheds its waters of hydration is partially compensated by the presence of 20 oxygen atoms that line the walls of the selectivity filter and form favorable electrostatic interactions with the permeant ion. Each of the four subunits contributes four main-chain carbonyl oxygen atoms from the protein backbone and one side-chain hydroxyl oxygen atom to form a total of four binding sites for K⁺ ions. Each bound K⁺ ion is thus stabilized by interactions with a total of eight oxygen atoms, which lie in two planes above and below the bound cation. In this way the channel is able to compensate for the loss of the K⁺ ion's waters of hydration. The amino acid side chains of the selectivity filter, which are directed away from the central axis of the channel, help to stabilize the filter at a critical width, such that it provides optimal electrostatic interactions with K⁺ ions as they pass but is too wide for smaller Na⁺ ions to interact effectively with all eight carbonyl oxygens atoms at any point along the length of the filter (Figure 5–15C).

In light of the extensive interactions between a K⁺ ion and the channel, how does the KcsA channel manage its high rate of conduction? Although the channel contains a total of five potential binding sites for K⁺ ions—four in the selectivity filter and one in the inner chamber—X-ray analysis shows that the channel can be occupied by at most three K⁺ ions at any instant. One ion is normally present in the wide inner chamber, and two ions occupy two of the four binding sites within the selectivity filter (Figure 5–15D). Because of electrostatic repulsion, two K⁺ ions never simultaneously occupy adjacent binding sites within the selectivity filter; rather, a water molecule is always interspersed between K⁺ ions.

During conduction a pair of K⁺ ions within the selectivity filter hop in tandem between pairs of binding sites. If only one ion were in the selectivity filter it would be bound rather tightly, and the throughput rate for ion permeation would be compromised. But the mutual electrostatic repulsion between two K⁺ ions occupying nearby sites ensures that the ions will linger only briefly, thus resulting in a high overall rate of K⁺ conduction.

X-ray crystallographic analysis has also begun to provide insight into the conformational changes that underlie the opening and closing of K⁺ channels. Studies by Clay Armstrong in the 1960s suggested that a gate exists at the intracellular mouth of voltage-gated K⁺ channels of higher organisms. Small organic compounds such as tetraethylammonium can enter and block the channel only when this internal gate is opened by depolarization. We now know that this internal gate is the narrow opening formed by the crossing of the four α-helixes at the intracellular mouth of the channel. The small opening at the helix bundle crossing in KcsA revealed by X-ray crystallography turns out to be too narrow to allow ions to pass. Thus the X-ray crystal structure is that of a closed channel.

What do K⁺ channels look like when they are open? Although we do not know the answer for KcsA, MacKinnon and his colleagues determined the open structure for a related bacterial K⁺ channel, MthK. Each subunit of this channel has two transmembrane segments, similar to KcsA. Unlike KcsA, MthK has a cytoplasmic binding domain for Ca²⁺ and can be opened by high concentrations of internal Ca²⁺. MacKinnon and colleagues determined the structure of the open state by growing crystals of MthK in the presence of Ca²⁺. Remarkably, the inner helixes that form the tight bundle crossing in KcsA are bent in MthK at a flexible glycine residue that causes them to splay outward, forming an internal orifice that is dilated to 20 Å in diameter, wide enough to pass K⁺ as well as larger compounds such as tetraethylammonium (Figure 5–16). This mechanism is likely to be a general one because the inner helixes of many K⁺ channels of bacteria and higher organisms have a conserved glycine residue at this position. The presence of a bend at this glycine gating hinge was recently confirmed in the X-ray crystal structure of a mammalian voltage-gated K⁺ channel.

The Structural Basis of Chloride Selectivity Reveals a Close Relation Between Ion Channels and Ion Transporters

Ions move across cell membranes by active transport by ion transporters (or pumps) as well as diffusion through ion channels. Ion transporters are distinguished from ion channels because (1) they require a source of energy to actively transport ions against their electrochemical gradients and (2) they transport ions at rates much lower than those of ion channels, too low to support fast neuronal signaling. Nevertheless, some types of transporters and certain ion channels may have a similar structure according to studies of a bacterial membrane protein that transports protons and Cl⁻. These studies also yield insights into the structural basis for Cl⁻ channel selectivity.

A large family of Cl⁻ channels, the ClC channels, is widely expressed in neurons and other cells of vertebrates. In vertebrate skeletal muscle the ClC-1 channels are important in maintaining the resting potential. Mutations in the genes for these channels underlie certain inherited forms of myotonia; the loss of Cl⁻ channel activity leads to a depolarizing after-potential following an action potential, resulting in repetitive firing of action potentials that produces abnormally prolonged muscle contraction. MacKinnon and his colleagues obtained a high- resolution X-ray crystal structure for a ClC protein from the bacterium Escherichia coli. Based on the close similarity of their amino acid sequences, it seems likely that the three-dimensional structures of the E. coli protein and vertebrate ClC channels would be very similar. Indeed, the E. coli ClC structure is consistent with the findings of a large number of previous studies of the effects of mutagenesis of vertebrate ClC channels.

It therefore came as a surprise when Chris Miller and colleagues found that the E. coli ClC is a transporter, not a channel. The E. coli ClC transporter uses the energy stored in a H⁺ gradient across the cell membrane to move Cl⁻ against its electrochemical gradient from the inside of the cell to the outside in exchange for the transport of H⁺ from the outside of the cell to the inside, down its electrochemical gradient. This type of transporter is termed an ion exchanger. In the E. coli ClC two Cl⁻ ions are transported in exchange for one proton. The rate of Cl⁻ transport through the E. coli ClC by this exchange mechanism is 100- to 1,000-fold slower than that of vertebrate ClC channels.

The E. coli ClC protein, similar to vertebrate ClC channels, is a homodimer composed of two identical subunits. Each vertebrate ClC channel subunit forms a separate Cl⁻ -conducting pore that gates independently from the other half (Figure 5–17A, B). The structure of the E. coli ClC transporter is much more complex than that of a K⁺ channel. Each subunit contains 18 α-helixes divided into related N-terminal and C-terminal halves, each containing nine helixes (Figure 5–17C, D). Surprisingly, the two halves are found in a head-to-head arrangement so that the helixes in each half have an opposite orientation in the membrane (helix 1 is related to helix 18, not helix 10). Unlike the pore of a K⁺ channel, which is widest in the central region, each pore of the E. coli ClC has an hourglass profile (Figure 5–18B). The neck of the hourglass, a tunnel 12 Å in length that forms the selectivity filter, is just wide enough to contain fully dehydrated Cl⁻ ions.

Although the structures of the E. coli ClC transporter and K⁺ channel differ in significant respects, the two structures have evolved in four similar ways to achieve a high degree of ion selectivity. First, both can be occupied by multiple ions. The E. coli ClC contains three binding sites for Cl⁻ ions within each selectivity filter. It appears that they can all be occupied simultaneously, thus creating a metastable state that ensures that the ions pass through the pore quickly. Second, in both cases the ion binding sites are formed by polar, partially charged atoms, not by fully ionized atoms. Thus the Cl⁻ binding sites are formed by main chain amide nitrogen atoms, which bear a partial positive charge, and by side-chain hydroxyl groups. The binding energy provided by these polar groups is relatively weak, ensuring that the Cl⁻ ions do not become too tightly bound. Third, in both structures permeant ions are stabilized in the center of the membrane by the partial charges of α helixes (Figure 5–18). In E. coli ClC the positively polarized (N-terminal) ends of two α-helixes dip partway into the membrane to lower the energetic barrier for Cl⁻ ions within the nonpolar environment of the membrane. Fourth, in both cases wide water-filled vestibules at either end of the selectivity filter allow ions to approach the filter in a partially hydrated state (see Figure 5–18).

Thus we see that, although the K⁺ channel and E. coli ClC pores differ fundamentally in amino acid sequence and in secondary, tertiary, and quaternary structures, strikingly similar functional features have evolved in these two classes of membrane proteins to ensure both a high degree of ion selectivity and efficient throughput. These features have been conserved with surprising fidelity from prokaryotes through humans.

Why does the E. coli ClC function as a H⁺ /Cl⁻ exchanger, whereas the vertebrate ClC proteins function as conventional channels? A likely explanation is that, unlike ion channels, the E. coli ClC does not have a continuous open pathway for ion movement from the outside of the membrane to the inside. Rather, like other pumps it is thought to have two gates, one external and one internal. Importantly, the two gates would never open simultaneously. Rather, ion movements and gate movements are presumed to be highly coupled in a tight cycle of reactions (Figure 5–19).

The relatively slow opening and closing of two gates in each cycle of ion transport, which takes place on a scale of milliseconds, explains why the rate of ion transfer through a transporter is several orders of magnitude slower than that of an ion channel. In the crystal structure of the E. coli ClC transporter the Cl⁻ ion in the selectivity filter is trapped by a negatively charged side chain of a nearby glutamate residue on one side and an electronegative hydroxyl group from a serine residue on the other side. Protonation of this glutamate is likely to cause its side chain to rotate out of the way, permitting Cl⁻ movement.

One or both of the E. coli ClC gates is thought to be missing in vertebrate ClC channels, providing an unobstructed pathway for the rapid diffusion of Cl ⁻ down its electrochemical gradient. Thus, despite the conservation of certain fundamental mechanisms in ClC proteins, relatively small structural changes give rise to two very distinct mechanisms of ion transport.

In all cells ion channels permit the flow of ions across an otherwise impermeable membrane. In nerve and muscle cells ion channels are important for controlling the rapid changes in membrane potential associated with the action potential, receptor potentials and postsynaptic potentials of target cells. In addition, the influx of Ca²⁺ ions controlled by these channels can alter many metabolic processes within cells, leading to the activation of various enzymes and other proteins, release of neurotransmitter, and even changes in gene expression.

Channels differ in their ion selectivity and in the factors that control their opening and closing, or gating. Ion selectivity is achieved through physical–chemical interactions between the ion and various amino acid residues that line the walls of the channel pore. Gating involves a change of the channel's conformation in response to an extrinsic stimulus, such as voltage, a ligand, or stretch or pressure.

Three methodological advances have greatly increased our understanding of channel function. First, the patch-clamp technique has made it possible to record the current through single channels. Second, genome-wide sequencing has determined the primary amino acid sequences of nearly all genes that encode ion channels. From these results many of the channels described so far can be grouped into three major gene families: voltage-gated channels and related subfamilies, a large family of ligand-gated channels, and gap-junction channels. Finally, X-ray crystallography has provided a detailed view of the three-dimensional structure of two types of channels.

The activity of channels can be modified by cellular metabolic reactions, including protein phosphorylation; by various ions that act as blockers; and by toxins, poisons, and drugs. Channels are also important targets in various diseases. Certain autoimmune neurological disorders, such as myasthenia gravis, result from the actions of specific antibodies that interfere with channel function. Other diseases, such as hyperkalemic periodic paralysis, result from ion channel defects because of genetic mutations. Detailed knowledge of the genetic basis of channel structure and function may one day make it possible to devise more specifically targeted pharmacological therapies for various neurological and psychiatric disorders.