The Electrochemical Basis of Nerve Function

Forces Due to Concentration Gradients

Pure water is 55.5 M (mol/kg) in concentration. The logic behind this startling notion is that the molecular weight of water is 18 and the weight of a liter of water is a kilogram: 1000 ÷ 18 = 55.5. Any solute added to water takes up space, displacing water molecules and so reducing their concentration. Table 3.2 provides one example of how common ions might be found in a peripheral nerve cell and its surrounding fluid. Also shown are the protein concentrations, with the cell protein content being 10-fold greater than the protein in the body’s interstitium and 100-fold greater than that in the CSF, which is the brain’s interstitial fluid. Because of this higher concentration inside the cell, the cell protein displaces, and thus dilutes, the neuron’s water. Since all substances spontaneously tend to move from regions of a higher concentration to regions of a lower concentration, water will tend to move into the cell from the interstitium, causing the cell to swell.

Maintenance of proper cell volume is so important that a variety of systems have evolved to counter the presence of cell proteins and to adjust to changing interstitial conditions. Indeed, pathologists generally see abnormal swelling in metabolically compromised cells, when the processes that counter this tendency are no longer adequately functioning. The following sections explain the physical principles used by these systems.

Osmotic forces measure the tendency of water to move down its concentration gradient, but our analytic instruments measure the solutes (sodium, potassium, chloride, sucrose—the dissolved substance), not the water. Hence it is the solute that gets all the attention and not the water. As a result, we dissemble when we say that osmotic forces tend to move water from a more dilute solution ( of solute, that is ) to a more concentrated one. In truth, the higher concentration of water is in the solution that has the lower concentration of solute, and water does in fact move as required by the laws of entropy, namely, from the solution of higher (water) concentration to the solution of lower (water) concentration.

Aquaporins ( AQPs, Fig. 3.1 ) are the proteins in cell membranes that allow cells to reach osmotic equilibrium rapidly. AQPs contain transmembrane pores so specialized for the transport of water that they allow it to move almost as fast as in bulk solution—3 × 10 ⁹ molecules per second for each pore—while excluding all other molecules. (Such a membrane is termed semipermeable. ) At least 10 distinct aquaporins are present in various cells in the human body, distributed in a tissue-specific manner. Congenital abnormalities result from the absence of specific AQPs: lack of AQP0 leads to cataracts; lack of AQP4 leads to deafness because it is needed for proper cochlear function; and AQP1 is required for adequate intraocular pressure.

Cells counter the osmotic force exerted by the high intracellular concentrations of protein by making a predominantly extracellular particle (sodium) impermeant as well. Consequently, sodium is generally more concentrated outside the cell (extracellular) than inside the cell (intracellular) ( Table 3.2 ). The osmotic force (Π for pressure) exerted by this one ionic gradient is huge, being proportional to the difference between the two concentrations ( Table 3.3 , the van’t Hoff equation ):

Thus it is supremely important for the integrity of the cell that the contributions of all osmotically active particles sum to as close to zero as possible. The adjustments available to the cell that correct for small imbalances are the topic of the next section.

Electrical Forces

Cell proteins generally carry negative charges, and this large quantity of impermeant charge has important electrical consequences. In any solution, the number of positive and negative charges must be equal—the principle of microscopic electroneutrality. Because many of the cell’s negative charges are on proteins, the remaining intracellular anions (like chloride) must be reduced relative to their extracellular concentration. In osmotic terms, this concentration gradient acts together with sodium to minimize water flux across the cell membrane. In addition, chloride is also charged, and so electrical forces (measured as voltages) must also come into play, as can be understood by examining the numerical consequences of these interactions.

The best way to visualize the electrical forces involved in this chloride gradient is an approach that is analogous to the van’t Hoff equation ( Table 3.3 ), namely, to calculate the force (in this case a voltage) that is generated by the ionic gradient using the Nernst equation (also in Table 3.3 ). For instance,

This mental image of chemical forces giving rise to electrical forces can be expressed in three equivalent ways: first, this voltage (V _Cl ) is the potential at which Cl _o (chloride outside) is in equilibrium with Cl _i (chloride inside); second, the concentration gradient $\left({\text{Cl}}_{\text{o}}/{\text{Cl}}_{\text{i}}\right)$ is just offset by electrical forces at V _Cl ; and third, at V _Cl , chloride movements into the cell are just equal to chloride movements out of the cell. Hence this Nernst potential is also called the equilibrium potential for chloride, or simply the chloride potential.

Cations have the opposite valence of anions (the z term in the Nernst equation), and so (by the properties of the logarithm) the concentration gradient is inverted. Thus, for potassium, whose equilibrium potential is similar to that of chloride, the intracellular concentration exceeds the extracellular concentration ( Table 3.2 ).

Returning to the case of sodium, its equilibrium potential is very different from that of potassium or chloride:

This difference is easily tolerated because sodium is the effectively impermeant ion in most cells. Indeed, the steep sodium concentration gradient is used to great advantage in the effective transport of fluid and generation of APs, as we shall soon see.

Membrane Potential

In the long run, individual neurons exist at a steady state, with osmotic forces across the cell membrane balanced and with the concentration gradients of the permeant ions offset by a characteristic voltage. The relationship among these electrochemical parameters was first visualized by Goldman and further developed by Hodgkin and Katz as being governed by the permeabilities of ions across the cell membrane ( Table 3.3 ). This relation is derived from the Nernst-Planck equation, with the algebraic contributions of individual ions being in proportion to P _S , their steady-state membrane permeability. Although the P _S is itself difficult to measure and awkward to express, relative permeabilities are more straightforward concepts. For instance, it is easily demonstrated experimentally that the potassium permeability of a nonmyelinated nerve axon is 100 times that of sodium (and then easy to say that $P_{Na}/P_K=1/100$ ), largely due to K _2P membrane channels. By use of these relative permeabilities, and postponing consideration of chloride until later, a more tractable form of the Goldman-Hodgkin-Katz voltage equation would be

Potassium is now obviously the dominant ion in this calculation, with only a modest 8-mV contribution from extracellular sodium.

This calculation is exactly the reason that we have introduced the concepts of electrical and concentration forces, namely, to demonstrate that the membrane potential is primarily due to the diffusion of potassium from the cell, withdrawing positive charges until the electrical potential across the membrane becomes approximately equal—but opposite—to the force generated by the potassium concentration gradient.

Cell chloride differs from the cations we have discussed not only because it is negatively charged but also because it is often in electrochemical equilibrium with its surroundings, which is to say that the chloride reversal potential is equal to the membrane potential or that V _Cl = V _m . This is true in some nerve and muscle cells where chloride serves to stabilize the membrane potential. In these cases where the chloride equilibrium potential equals the membrane potential, V _Cl contributes nothing to the Goldman-Hodgkin-Katz calculation and can be safely omitted, as shown in the previous example. In other cells, chloride ions are pumped out, making V _Cl more negative than V _m . This is true in some postsynaptic nerve terminals, and when inhibitory neurotransmitters open chloride permeant channels, chloride ions diffuse passively into the neuron, transiently making the membrane potential more negative and thus more difficult to trigger an AP. Finally, chloride is actively accumulated in epithelial cells that secrete fluid, which is the subject of the next section.

Fluid Transport: Pumps and Channels

The nervous system uses a multitude of different pump and channel proteins to harness the electrochemical forces that are the subject of this chapter. We have seen two examples of membrane channels: aquaporin and the K _2P potassium channel that is responsible for the resting membrane potential. Channels are water-filled pores permitting the passive diffusion of select molecules down their electrochemical gradient. The number, location, and open-state characteristics of a particular channel are regulated as needed.

Pumps move ions and other molecules through dynamically structured, discontinuous, water-filled pathways. Such an arrangement allows molecules to be transported against their electrochemical gradient ( active transport) by coupling the movements to adenosine triphosphate (ATP) hydrolysis ( primary active transport) or to the gradient of another ion (commonly sodium, and termed secondary active transport).

Central to the electrical and transport functions of the nervous system is the sodium pump (a sodium-potassium ATPase, or P-type ATPase, Fig. 3.2 ), a molecule present in all cells that transport sodium out of the cell and potassium into the cell. This exchange generates a steep concentration gradient for sodium—less so for potassium—and is fueled by the chemical energy of ATP. The pump functions as it does because it has two distinct conformations: E1, with a pocket near the N-terminal region that is open to the cytoplasm but not the extracellular space, and E2, with a pocket that is open to the extracellular space but not the cytoplasm. With three sodium ions from the cytoplasm populating the E1 space, ATP can bind, causing a conformational shift to E2. When a critical aspartate near the C-terminal end of the protein is protonated, the pump’s relative affinity for sodium ions is decreased, and they are released to the outside surface, to be quickly replaced with two potassium ions. Next, the pump shifts back to the E1 conformation. Now being open to the cytoplasm, the proton and the two potassium ions exit the pump into the cytoplasm. Mutations near the C-terminal protonation site underlie the congenital diseases rapid-onset dystonia-parkinsonism and familial hemiplegic migraine type 2, illustrating the importance of the protonation step in the proper function of the sodium pump.

The distinction between pump and channel is not absolute, as the sodium pump is briefly open to both the cytoplasm and the extracellular space. Palytoxin can arrest the pump cycle at this point, allowing the free diffusion of sodium into the cell and potassium out of the cell, with fatal consequences. This toxin is made by palythoa, a polyp of the phylum Cnidaria, native to Hawaii and the Mediterranean Sea; death ensues following dermal abrasion, ingestion of fish or crustaceans that feed on the polyp, by inhalation of sea aerosols during infestations, and during warfare that uses spears poisoned with palytoxin.

The judicious placement of secondary active transport molecules and specific channels allows the nervous system to generate the specialized fluids in the extracellular spaces of the brain, cochlea, and eye, as required for the proper function of these organs. The brain is bathed in CSF, a solution low in protein that is generated by the choroid plexus and removed through the arachnoid villi. The cochlear endolymph is high in potassium, and the ciliary body of the eye is continuously producing a nutrient solution that flows past the lens and is taken up by specialized veins along the margin of the iris.

Epithelia are the tissues designed to move ions and fluids in one—and only one—direction. Epithelial cells have two functionally distinct surfaces: the base and the sides (or basolateral surface ), which are in contact with the interstitial fluid of the body, and the apical surface, which faces the lumen. Almost all epithelia restrict the sodium pump to the basolateral surface; the two exceptions are the choroid plexus and the retinal pigmented epithelium, in which the sodium pump is exclusively in the apical membrane. Individual epithelia then distinguish themselves by distributing characteristic channels and transporters on their apical and basolateral surfaces.

For instance, the epithelia that generates cochlear endolymph has the sodium pump and NKCC1 (a Na ⁺ /K ⁺ /2 Cl ^– cotransporter and gene product of SLC12A2) localized to their usual basolateral side. This arrangement concentrates cell chloride and potassium. These epithelial cells have their potassium channels on the apical side, causing the potassium and chloride pumped into the cell by the basolateral NKCC1 cotransporter to exit together into the scala media, along with water as required by osmotic forces, generating a cochlear endolymph that is high in potassium.

The sodium gradient is also the basis of the production of large quantities of CSF. The sodium pump, NBCe2 (a Na ⁺ /HCO ₃ ^– cotransporter and gene product of SLC4A5) and NKCC1 reside on the apical (ventricular) surface of the cells in the choroid plexus, moving large amounts of sodium bicarbonate–rich saline-like fluid into the cerebral ventricles. In addition, large quantities of carbonic anhydrase are present in the cells lining the choroid plexus, generating HCO ₃ ^– ions that accompany chloride into the CSF. These chloride and bicarbonate ions are accompanied by passive movements of water molecules as dictated osmotic forces. Other cotransport mechanisms move nutrients, antibiotics, and a wide variety of other organic molecules into the CSF, generating liters of CSF a week, all the time keeping an effective barrier against erythrocytes, leukocytes, and plasma proteins.

Ohm’s Law and the Equivalent Electrical Circuit of the Neuron

The resting membrane potential refers to the neuron at a steady state and is largely the result of the potassium, chloride, and sodium diffusion potentials; in addition, membrane currents —electrical charges carried by ions crossing the cell membrane—modify this voltage as described by Ohm’s law:

Thus current (I, in amperes) flowing through a conductance (G, in siemens) or across a resistance (R, in ohms) will generate a voltage drop ( Fig. 3.3 ). By a convention established by Benjamin Franklin, electrical current is the flow of positive charges: Positive charges leaving the cell are defined as a positive current. Equivalently, negative charges entering the cell are also a positive current. Conversely, positive charges entering the cell are a negative current, as is the exit of negative charges. A familiar example is the sodium pump, which is electrogenic because it cycles three sodium ions out of the cell for every two potassium ions in; this net positive current removes positive charges from the cell interior, causing the membrane potential to be more negative than predicted by the Goldman-Hodgkin-Katz voltage equation. In a cell as large as a skeletal muscle fiber, this amounts to ∼2 to 5 mV. In small nerve terminals, where the input resistance is much greater, this current can hyperpolarize the membrane by 15 mV or more.

Of even more interest is the flow of current through open membrane channels because the number of open channels varies when the nerve is stimulated in any of a wide variety of ways. From Ohm’s law, the magnitude and direction of the flow of an individual ion S through the cell membrane equals the driving force on that ion times its conductance:

The electrical driving force on S is the difference between the voltage across the membrane (V _m ) and that voltage where the ion is at electrochemical equilibrium (V _S , the Nernst potential for substance S). These relationships are summarized diagrammatically, for those familiar with electrical circuits, in Fig. 3.4 . Thus the magnitude of the ionic flow will increase as the driving force—or the conductance—of the ion increases and decrease as it decreases. In the face of an increasing conductance and a decreasing driving force, a situation that is described in the section on the AP, specific calculations are required to determine the final outcome.

We can use the circuit theory in Fig. 3.3 to calculate the number of charges (Q) on a cell membrane that has a given voltage (V) because Q = V × C _m , and the capacitance of a cell membrane (C _m ) has been measured to be 0.9 μF/cm ² . If a hypothetical neuron were spherical (for simplicity) and 20 μm in diameter, it would have a surface area of almost 1300 μm ² , a capacitance of 11 pF per cell, and thus a charge of 1 pC (picocoulomb) when the membrane voltage is 90 mV. The 1 pC of charge on the membrane represents 6 million ions. Although this may seem to be a lot, the cell volume of this neuron would be 4 pL (picoliters) and contain ∼250 billion potassium ions and ∼25 billion chloride ions. Together, these two ions alone are ∼40,000 times the number required to charge the membrane.

Pain and a Syndrome of Periodic Paralysis

Pathologic conditions may alter the concentration of ions ordinarily seen in nerve cells ( Table 3.2 ). For instance, tissue injury causes a local increase in the potassium concentration as these cells release their contents. This increased extracellular potassium moves the steady-state membrane potential toward 0 mV, which generates APs when it is done quickly enough. Thus one source of pain is simply the direct stimulation of nerve endings by elevated potassium concentration in the tissue interstitium.

In rare individuals who have certain genetic abnormalities, extracellular concentration of potassium can fall dramatically when epinephrine or insulin stimulates its uptake by muscle cells, leading to muscle weakness and even paralysis. This condition is called hypokalemic periodic paralysis. Surprisingly, the muscle membrane potentials are less negative than normal, just the opposite of what the Nernst equation predicts. For reasons not yet fully understood, the cell membrane loses its ability to select potassium over sodium, which is to say ${\text{P}}_{\text{Na}}/{\text{P}}_{\text{K}}$ declines markedly. The effect is so great that the cell is depolarized because the membrane potential moves away from V _K toward V _Na , as the Goldman-Hodgkin-Katz voltage equation predicts. This membrane potential change is slow, allowing the muscle fiber to undergo accommodation and so become inexcitable. (Accommodation is explained more fully in the section on APs.)

Attack by the Immune System: Membrane Attack Complexes and Perforins

The nervous system can be damaged by bacteria and even the body’s own immune system in ways that short-circuit membrane potentials and destroy the cell’s integrity by insertion of nonselective channels into cell membranes. Two distinct families of pore-forming proteins kill bacteria and those cells that harbor viruses. The first are the defensins, made by phagocytes and epithelial cells; the second are the MACPF/CDCs (membrane attack complex [MAC] perforin-like/cholesterol-dependent cytolysins), such as the perforins ( Fig. 3.5 ) stored whole in lytic granules of killer T lymphocytes, and the MAC , which is constructed from elements of the complement cascade C5b, C6, C7, C8, and a polymer of C9, which together form pores with diameters that are 10 nm and greater. (Although healthy cells use the protein CD59 to inhibit the formation of MACs, certain diseases destroy the myelin sheaths of motor neurons, causing a paralysis, as discussed later in this chapter.) At 16 nm, perforins are even larger.

The channels through MACs and perforins are wide enough to easily pass sodium, potassium, chloride, and sucrose, discriminating little among them. The channels’ conductances are correspondingly large: 2 nS (nanosiemens) for the MAC and 6 nS for perforin. As a consequence, the formation of just a single C9 aggregate or insertion of a single perforin molecule results in a large flow of ions. As in all nonselective channels, the currents flowing through MAC and perforin channels are carried primarily by sodium ions because the magnitude of the driving force for sodium is the greatest: V _m − V _Na >> V _m − V _Cl > V _m − V _K . If V _m = −90 mV (to take a simple example), the driving force on the sodium will be

and the current flowing through the MAC is easy to calculate:

Because one ampere is the flow of 6.3 × 10 ¹⁸ charges per second, the number of sodium ions flowing through a single MAC is 2 × 10 ⁹ per second, a rate so fast that the membrane potential of our previous example (the 20-μm neuron) would be neutralized in [(6 × 10 ⁶ ) ÷ (2 × 10 ⁹ )] = 3 × 10 ⁻³ seconds.

By following the movements of the various ions due to the electrical forces on them, it is possible to see that the cell gains osmotic particles and so swells. For instance, because the net driving force on sodium is negative, it will enter the cell, with its positive charge tending to cancel the negativity of the membrane potential. In fact, the cell membrane potential will rapidly go to zero and remain there as long as the MACs are in the membrane. As a consequence, the driving force on chloride will increase; (V _m − V _Cl ) being positive, the current will be positive, meaning that chloride is entering the cell along with the sodium. Thus, when sodium and chloride enter the cell, water follows, and the cell swells. Even more important, the MAC complexes are so large that molecules the size of ATP can diffuse from the cell. Thus attack by complement or killer T cells leads inexorably to cell swelling and lysis both because important metabolic contents are lost and because the osmotic pressures exerted by the remaining cell proteins cause cell swelling and death.

Microbial Attacks: Antibiotics

Insertion of ion channels into cell membranes is also a weapon deployed by many microorganisms. Antibiotics such as amphotericin and gramicidin, and α-staphylotoxins from Staphylococcus aureus, lyse cells by broaching their membranes with large pores. When it is used clinically, amphotericin preferentially attacks fungal cells in fungal meningitis, but there is a narrow therapeutic range because amphotericin also attacks cell membranes in the nervous system. In overwhelming sepsis, α-staphylotoxins attack all cells of the body, leading to multiple organ failure and cardiovascular collapse, primarily due to the loss of the integrity of cell membranes.

Generator Potentials

All body sensations are graded, with most transduction mechanisms generating bigger electrical signals—and consequently more APs—for bigger stimuli ( Table 3.4 ). The most striking exception concerns vision, where a photon acts to isomerize the G protein–coupled receptor (GPCR) rhodopsin, leading to the closure of cyclic guanosine monophosphate (cGMP)-gated cation channels, a reduction in the cell’s membrane current, and hyperpolarization of the photoreceptor. These graded responses are generator potentials that can be the direct result of the stimulus opening or closing membrane channels or increasing the current through existing membrane channels. More often, intermediary chemical signals connect the initial sensation to the opening of membrane channels, processes that are discussed further in Chapter 4 .

Mechanotransduction—the sensing of touch, of hearing, of cell volume change—is the direct result of stretch-activated channels opening as the cell membrane is deformed. The channels involved are drawn from highly disparate families, with a two being unique (Piezo1 and Piezo2, Fig. 3.6 ), but others belonging to ion channel families with many more members, such as TRPV2 (a transient receptor potential cation channel, Fig. 3.7 ) and KCNK4 (also called TRAAK; a two-pore potassium channel ). Structurally, mechanosensitivity can derive from extracellular (the piezo proteins) or intracellular sequences (the TRPV2 protein), but in either case, it appears that stretch-activation is caused by tension exerted by the cell membrane on the channel. Functionally, these channels, along with assorted accessory proteins, are used for distinct purposes. For instance, one use of Piezo2 is to sense prolonged or transient vibrations in Merkel cells or their Aβ sensory nerve fibers. By contrast, TRPV channels are more complex proteins that respond to a wide array of noxious stimuli, including the mechanical signal that results—via their ankyrin linkage to the cell’s cytoskeleton—from changes in cell volume. Pathologically, mutations of mechanoreceptors underlie certain congenital syndromes of extreme contractures that are present at birth; two Piezo2 variants that have an increased probability for the channel to be in the open state underlie distal arthrogyrposis type 5 .

Synaptic Potentials

Vertebrates communicate neuron-to-neuron by chemical transmission, as already introduced in Chapter 1, Chapter 2 . These signals bind to specific sites on receptors, to trigger electrical events (the subject of this chapter) or alter enzymatic activity and the second messengers calcium, cyclic adenosine monophosphate (cAMP), cGMP, the inositol phosphates, diacylglycerol, and the βγ subunit of the G protein (explained in Chapter 4 ). Ligands that bind to receptors may either activate them ( agonists ) or keep them from functioning ( antagonists ). Our body uses many different transmitters acting on many, many distinctly different receptors to confer specificity of action throughout the nervous system.

As with the generator potentials, neurotransmitters act to open membrane channels either directly or via intermediary signals. Direct activation of synaptic channels occurs in the cys-loop and glutamatergic receptor families as well as in one purinergic receptor type ( Table 3.1 ). In the best understood example of a cys-loop receptor, acetylcholine (ACh) binds directly to the nicotinic receptor (nAChR) at the neuromuscular junction (NMJ) of skeletal muscles and opens its nonspecific cation channel. The other cys-loop receptors bind either glycine or γ-aminobutyric acid (GABA) in the case of the GABA _A receptor or serotonin in the case of the 5-HT ₃ receptors. Each of these receptors are closely related structurally and functionally to the nAChR.

Receptor Binding and Channel Gating

All synaptic receptor molecules have a distinct region that specifically binds the transmitter. In the case of the NMJ, one or two ACh molecules bind reversibly in a highly specific manner to the large extracellular portion of the pentameric nicotinic ACh receptor molecule (whose subunits are (α ₁ ) ₂ β ₂ εδ, Fig. 3.8 ). The binding sites each span two subunits: the α ₁ ε and the α ₁ δ interfaces. In the presence of the transmitter, three loops of the α subunit come together with a loop of the ε or δ subunit to form a box of nonpolar and aromatic amino acids, primarily tryptophans and tyrosines. The open, conducting conformation is stabilized when ACh is in this box. Because there are two α ₁ subunits in the receptor complex, two ACh molecules must be bound before ion flow can begin. Once the binding of one of the AChs reverses, its α1 subunit is free to close, at which time ion flow ceases.

The (α ₁ ) ₂ β ₂ εδ nAChR is found only at the neuromuscular junction in the body, so a drug such as curare will specifically relax skeletal muscle fibers (for instance, during surgery) but have no unwanted side effects on ACh’s actions at other receptors on the heart or vascular smooth muscle contractility. A plant product, curare was the first nondepolarizing muscle relaxant used clinically. The specificity, duration of action, and potency of effect are largely due to the extremely high affinity of the drug to the binding site, which in turn reflects how well the drug fits into the box-like geometry of the amino acids that comprise the α ₁ ε and the α ₁ δ interfaces.

Nerves also have nicotinic receptors—pentamers constructed from one or more of 16 possible genes in a way that is different from the nAChR found at the neuromuscular junction. For instance, autonomic ganglia have the (α ₃ ) ₂ (β ₄ ) ₃ nAChR, which is blocked by hexamethonium, a drug that has no effect at the NMJ. Because of this distinction, hexamethonium was used to block the sympathetic nervous system at the ganglionic level as an antihypertensive. Similarly, ganglionic and many central nervous system nAChRs are blocked by bupropion, an antidepressant often used to limit the nicotine effects of cigarettes during smoking cessation therapy.

The other cys-loop receptors also have characteristic activators and inhibitors that are analogous to those active at the nicotinic receptor. Strychnine binds to the glycine receptor, blocking its inhibitory activity and leading to a hyperexcitable state, a tool used long ago by medical students to remain alert for examinations; this was effective but only within a very narrow range of dosing because slightly higher doses cause convulsions. Benzodiazepines bind to GABA _A receptors, increasing the effectiveness of the endogenous GABA; barbiturates bind to GABA _A receptors, inhibiting their activity. Thus specificity of action within the nervous system is conferred by the different structures of the various receptor molecules in their transmitter-recognition region.