Genesis of Fast Sodium Action Potentials and Properties of Sodium Channels

Neuronal cells use a single type of signaling based on all-or-nothing action potentials. Sodium action potentials such as those recorded in axons or cell bodies are relatively invariant in normal tissue, and thus the shape and duration of these electrical signals do not change significantly within neuronal subtypes in the nervous system. Calcium action potentials are similarly predictable, but the underlying ionic mechanism can be rather complex, depending on the cell type and its topographic location within the cell (see later). The terms sodium action potential and calcium action potential refer to the initial (depolarizing) phase of these rapid changes in membrane polarity. Although genetic or molecular alteration of I_Na and I_Ca can significantly affect neuronal firing and, ultimately, central and peripheral nervous system neurophysiology, it must be remembered that gross changes in neuronal excitability may also result by altering the repolarization phase of individual action potentials.

Action potentials have a characteristic shape once a certain threshold is reached. In normal tissue, stereotyped electrical events follow the initial depolarization (Fig. 49-4). The sequence can be described as follows: (1) RMP moves from an initial negative value (−65 to −80 mV for most neurons) toward the so-called threshold for activation of sodium channels (around −40 mV). This change can be slow and may occur spontaneously as a result of fluctuations in RMP; the threshold is reached rapidly when the initial depolarization is triggered by a synaptic potential (or a summation of synaptic potentials). (2) After reaching the threshold value, an extremely rapid (1 to 2 msec) depolarization occurs because of opening of sodium channels and massive influx of sodium ions into the cell. (3) Termination of the “upstroke” phase depends on a combination of factors: voltage– and time-dependent inactivation of I_Na and a concomitant decrease in the driving force for sodium occurring in parallel with voltage-dependent activation of I_K. (4) Further increases in the permeability to potassium ions restore the pre–action potential RMP values and force the membrane toward E_K. (5) Under most circumstances, an undershoot of cell resting voltage occurs (a few millivolts from RMP) as a result of residual activation of I_K and the contribution of the electrogenic sodium-potassium adenosine triphosphatase (Na⁺,K⁺-ATPase) extruding 3 Na⁺ ions in exchange for 2 K⁺. The return to pre–action potential voltage favors the so-called removal of inactivation, a necessary step that allows a subsequent cycle of depolarization-induced action potential firing. This stereotyped and relatively simple sequence of events is typically recorded in axons; other more complex interactions of various I_Na and I_K values may lead to slightly different voltage profiles.

FIGURE 49-4 Changes in conductance during generation of an action potential.

From a functional standpoint, it is important to remember that the genesis of fast sodium action potentials is a hallmark of neuronal function, to the degree that during neurophysiologic recordings, the presence or absence of Na⁺ spikes is frequently used to determine the neuronal or glial cell type. Recently, this notion has been challenged, and glial “action potentials” have been reported with increasing frequency. These responses, however, appear to usually be associated with pathologic conditions (brain tumors, epilepsy), and the old perception that neuronal cells are the exclusive tenants of sufficient I_Na density to promote active responses is still generally accepted. Within the same neuronal cell, Na⁺ channels involved in the generation of action potentials can be located heterogeneously, and it is common for clusters of channels to be located at specific and crucial membrane segments. The most commonly encountered clustering of Na⁺ channels occurs at the node of Ranvier of myelinated axons, but clustering also occurs at synaptic contacts, dendrites, and cell bodies, in proximity to the initial segment of axons.

Early pharmacologic studies attempting to elucidate the ionic correlates of the action potential greatly benefited from the availability of naturally occurring toxins that specifically and powerfully block I_Na. The magic bullet for sodium channels was tetrodotoxin (TTX),⁵ a lethal poison produced by a selected species of puffer fish (Fugu). TTX is present in large quantities in the liver and female genital organs, but the real culinary treat is the testes of this hermaphroditic animal. Separation of the toxic portions is paramount to survival of the consumer. The sea anemone toxin (ATX) and α-scorpion toxin are also powerful blockers of I_Na. TTX and other natural toxins block the external portion of the ion channel. Interestingly, although most sodium channels are blocked by micromolar concentrations of TTX, genetic ablation of one amino acid (Tyr or Phe) in the sequence of the ion channel protein confers relative resistance to TTX because binding of the toxin is restricted to this region.⁶ The amazing specificity of TTX binding suggested that normally occurring variation in the amino acid sequence of ion channels may prove to be an extremely important determinant of not only the pharmacologic properties of the channels but also their inactivation/activation/voltage dependency. In fact, although the poisonous actions of TTX are primarily due to direct blockade of Na⁺ fluxes through the channel’s pore, ATX and α-scorpion toxin bind to a portion responsible for channel inactivation. Mutations in these regions cause faulty inactivation, a condition linked to neuropathogenesis.

The biophysical correlates of Na⁺ channel function are well understood today. The general scheme of

explains the properties of whole-cell I_Na recorded from neurons. The voltage dependency of each process justifies the initial depolarization required to promote the opening of channels; the consequent depolarization induced by sodium current promotes further opening of channels, the process being terminated by time- and voltage-dependent closure of the channels. The passage from closed to open (and visa versa) is referred to as activation (deactivation), whereas the passage from open to inactivated is called inactivation. Removal of inactivation occurs when the channel returns to its closed state.

From a structural point of view, Na⁺ channels are constituted by αβ1β2 heterotrimers, often with four repeated domains each with six-membrane–spanning subunits. The voltage sensor is located on the fourth transmembrane domain. Different subunits are represented differently in the central and peripheral nervous systems. According to the recent literature, the following tissue-specific localization and pharmacology can be derived: for α subunits, SCN1A is primarily expressed in brain tissue and is blocked by TTX and saxitoxin (SXT). The most abundant brain α subunit is encoded by SCN2A1; this subunit is also found in peripheral nerves, in the initial axonal segment, and in the nodes of Ranvier (TTX and SXT sensitive). SCN2A2 and SC2NA3 are predominantly expressed in the brain, whereas SCN4A encodes channels found in muscle. In addition to TTX, these channels are sensitive to ω-conotoxins (GIIIA, GIIIB, GIIIC). Mutations of these channels are responsible for hyperkalemic periodic paralysis, paramyotonia, and myotonia. The SCN5A subunit is expressed in the heart and in denervated skeletal muscle; in the heart, these channels are responsible for upstroke of the action potential. These subunits are resistant to blockade by TTX and SXT; mutations of these genes are involved in subtypes of long QT syndrome, a cardiac condition that is sometimes associated with epileptic seizures. SCN6A (uterus), SCN10A, SCN11A, and SCN9A (dorsal root ganglion) are all sensitive to TTX. The molecular nature of glial sodium channels and the contribution of a specific α subunit are not fully understood at present. β Subunits are bound covalently to α subunits and provide inactivation kinetics to Na⁺ channels. Mutations of these SCN1B and SCN1A subunits have been linked to generalized epilepsy with febrile seizures (see Catterall and colleagues⁷ and Gambardella and Marini⁸ for review of voltage-gated ion channel–related hereditary diseases).

The fact that even slight mutations cause profound deficits in sodium channel function and that these mutations result in neurological diseases leads to the hypothesis that replacement of defective channels by gene therapy may repristinate the loss of function caused by the initial genetic deficit. If a single gene mutation is responsible (as in cystic fibrosis and in a number of neurological disorders) and if the tissue to be transfected with the repair mechanism is accessible (such as skeletal muscle), it is possible that viral delivery of genetic products may alleviate the consequences of faulty ion channels. A positive outcome with this approach is somehow dependent on the pathogenesis of the disease itself. If the observed deficit is the consequence solely of the inherited mutation, replacement by a normal genotype is likely to be successful. If, however, the initial deficit has led to extensive rearrangement of neuronal connections or if the deficit causes extensive neuronal cell death (such as in epilepsy), it is unclear whether restoring normal function by simple gene replacement may be beneficial. It is also worth remembering that although a small fraction of neurological disorders are clearly imputable to a single gene mutation affecting a particular ion channel, the most common forms of disease result from a complex interaction of initial genotypic changes followed by adaptive responses, including apoptosis or necrosis. Finally, given that mutations of crucial ionic mechanisms such as I_Na affect cardiac, neuronal, and muscular function, it is possible that a large number of mutations are nonvital and that the vital mutations are possible because of redundancy of gene expression or compensatory overexpression of similar ion channel proteins.

Phenotypic changes caused by relatively minor alterations in ion channel gating sometimes become clinically relevant when concomitant deficits not necessarily associated with action potentials are present. This is the case, for example, in mutations of the SCN4A subunit leading to hyperkalemic periodic paralysis. For the paralytic symptoms to occur, patients must concomitantly experience variations in plasma potassium (by either K⁺ intake or exercise followed by rest). This leads to opening of Na⁺ channels that switch to a non-inactivating mode, thereby leading to the development of a persistent inward Na⁺ current. The ensuing depolarization of muscle membrane will further increase [K⁺]_out via loss through voltage-dependent K⁺ channels and thus aggravate the initial trigger. Furthermore, the persistent depolarization causes inactivation of normal Na⁺ channels, which leads to rapid loss of tissue excitability and paralysis. This example demonstrates the complex interactions between normal and abnormal ion channels expressed in a certain cell type, the importance of the extracellular milieu in biophysical signaling via ion channels, and the difficulties associated with the diagnosis of altered ion channel phenotypes.

Calcium Action Potentials and Calcium Channels

The mechanism of calcium action potentials is somewhat different, but it follows the general principles of threshold for activation and rapid gating mechanisms. As is the case with sodium channels, Ca²⁺ channels are distributed heterogeneously in the CNS, and even within the same cells different subspecies of Ca²⁺ channels may be found. This inhomogeneous expression is functionally significant in that it allows Ca²⁺ influx to perform several different cellular tasks, including depolarization of dendrites and propagation of signals to the cell body, synaptic release of neurotransmitter, contraction, and second messenger function. As for sodium channels, membrane depolarization is the most common trigger for opening of channels; the kinetic properties of Ca²⁺ channels, however, are characterized by longer time constants. This kinetic behavior underlies the different durations of neuronal/axonal action potentials (I_Na, 1 to 3 msec) versus cardiac action potentials (several hundreds of milliseconds; large inward currents carried by I_Ca).* An additional difference between Na⁺ and Ca²⁺ action potentials (or the underlying ion channel mechanisms) is the fact that the threshold for I_Ca activation varies greatly between different Ca²⁺ channel families. Low-threshold Ca²⁺ channels (or low voltage activated) are also characterized by relatively rapid opening and closing and are also referred to as “T-type” (transient) currents. High-threshold Ca²⁺ channels (or high voltage activated) can be further subdivided into neuronal (N), L, and P types. The pharmacologic properties of the calcium channel families are equally complex (Table 49-3).^9,10 Although sodium action potentials are typically triggered by depolarization and terminated by depolarization (inactivation of I_Na and voltage-dependent activation of I_K), an additional mechanism is involved in repolarization: activation by intracellular Ca²⁺ channels that act as powerful terminators of cell depolarization. Ca²⁺ channels are modulated by important intracellular signals such as cyclic adenosine monophosphate (cAMP; cAMP-dependent protein kinases) and G proteins. These modulatory signals arise from receptor stimulation, thus coupling the activity of postsynaptic (or presynaptic in the case of presynaptic receptors) Ca²⁺ channels to the activity of neighboring cells.

Ca²⁺ channels contain four or five distinct subunits: α subunits display different tissue and peptide specificity. They are constituted by transmembrane-spanning proteins and act in both voltage sensor and selectivity filter capacities. The dihydropyridines verapamil and nifedipine bind to α₁ subunits. In Lambert-Eaton myasthenic syndrome, the autoimmune component of the disease is due to IgG binding to α_1A subunits. In the majority of cases, P/Q-type channels are involved; in a small percentage of cases, the α_1B subunit constituting N channels mediates the autoimmune response. Other subunits increase the amplitude of Ca²⁺ currents and bind the antiepileptic drug gabapentin (α_2δ). The β subunit is localized in the cytoplasm, associates with α subunits that are membrane bound, has regulatory functions, and mediates the effects of cAMP; phosphorylation of cAMP-dependent protein kinase modifies current, voltage dependence, and activation-inactivation. γ Subunits are exclusively localized within the membrane and lack a cytoplasmic component. Similar to subunits in other channels, γ subunits modulate channel voltage dependency.

Not all calcium channels are involved in transmembrane fluxes of calcium. Ca²⁺ release channels are located ubiquitously in intracellular organelles and regulate the cytoplasmic Ca²⁺ content of virtually every mammalian cell type. These channels belong to two separate families, those sensitive to the alkaloid ryanodine and those activated by inositol 1,4,5-triphosphate (IP₃). Ryanodine-sensitive Ca²⁺ release is triggered by the activity of dihydropyridine-sensitive Ca²⁺ channels and therefore acts as a signal amplifier. These channels are located mainly in skeletal and cardiac muscle. Disorders resulting from changes in these channels include malignant hyperthermia and central core disease. The IP₃ receptors are located primarily in glial and neuronal endoplasmic reticulum membranes. In response to stimulation of cell surface receptors, intracellularly generated IP₃ binds to these receptors and causes intracellular Ca²⁺ stores to be released. Thus, although IP₃ receptors link changes in intracellular calcium to chemical transmission and act as second messengers, the ryanodine-sensitive mechanism is optimized to link the electrical activity promoted by Ca²⁺ action potentials to calcium mobilization.

Given the large number of Ca²⁺ channel families and their widespread distribution in the nervous system, it is not surprising that numerous neurological disorders have been linked to mutation, autoimmune inactivation, or altered expression of I_Ca. Disorders affecting the presynaptic terminal of motor axons cause the aforementioned Lambert-Eaton myasthenic syndrome, and mutations of the α_1A subunit are responsible for a form of episodic ataxia type 2. Lambert-Eaton myasthenic syndrome is an autoimmune disorder associated with an immunologically abnormal response to neoplasms, whereas type 2 ataxia is caused by defective production of ion channels. Familial hemiplegic migraine is associated with missense mutations in transmembrane segments, and progressive ataxia is caused by either trinucleotide repeat expansion in an intracellular region near the C-terminal or by missense mutation.¹¹

Repolarization of Action Potentials and Maintenance of Resting Membrane Potential: Potassium Channels

There are many subtypes and functions of potassium channels in the mammalian CNS, with whole textbooks devoted to them. Different families of K⁺ channels are found in the mammalian CNS, including (1) Kv channels, which are activated by a change in transmembrane voltage, and (2) inward rectifier channels (K_IR channels), which have high conductance for K⁺ ions moving into a hyperpolarized cell (in contrast to K⁺ ions moving out of a depolarized cell) and are often regulated by intracellular messengers. Kv channels are typically closed at RMP and open rapidly after membrane depolarization. They are variably spliced tetramers composed of four homologous α subunits that each contains a voltage sensor and a sequence that provides cation selectivity for potassium. Kv channels are variably composed of alternative RNA splicing and subunit duplication of gene products from four subfamilies, Kv1 to Kv4, which correspond to the murine Shaker, Shab, Shaw, and Shal. Rates of channel inactivation vary dramatically between the different subtypes.

Potassium channels are the major contributor to both neuronal and glial RMP. As described earlier, neuronal membranes are predominantly permeable to potassium at rest, which results in cell RMP being near the potassium reversal potential E_K. Even more so than neurons, the majority of glial resting channels are K⁺ permeable, which results in a relatively hyperpolarized RMP. Because of their lack of electrical excitability, a hyperpolarized RMP was previously a defining criterion during physiologic recordings from glia in situ. However, it has been demonstrated over the past several years that glial cells can have a wide range of RMPs, depending on the glial cell subtype, ion channel endowment, location within the nervous system, and developmental stage.^12,13

Hodgkin and Huxley initially characterized the important role of (what are now termed) delayed rectifier Kv potassium channels in action potential repolarization.¹⁴ These channels, which are blocked by tetraethylammonium (TEA), are similar to sodium channels in that the probability of channel opening is increased by membrane depolarization, with increased opening in proportion to the amount of membrane depolarization. However, delayed rectifier potassium channels differ from sodium channels in that they open more slowly and do not inactivate in the presence of persistently maintained depolarization. As discussed earlier, it is this sodium channel inactivation together with persistent activation of outward potassium channels (and inward chloride channels that produce an outward current by allowing negatively charged chloride ions to enter the cell) that results in action potential repolarization. After repolarization of the action potential, there is a refractory period during which the threshold for initiation of another action potential is elevated. The absolute refractory period, during which an action potential cannot be generated, is followed by a relative refractory period, during which an increased stimulus is necessary to generate an action potential. The refractory period results from residual sodium channel inactivation and potassium channel activation; it limits the maximum firing frequency of different classes of neurons.

The M channel has distinctly different properties from the Kv potassium channels that are responsible for repolarization of the action potential. Although activated by membrane depolarizations, these channels are inhibited by muscarinic acetylcholine receptor binding, as well as by a variety of other neurotransmitters and neuroactive compounds. Rates of channel opening and closing are approximately 100 times slower than delayed rectifier channels. Based on these different characteristics, M channels are thought to have divergent functions in the CNS. On the one hand, by means of their slow kinetics, they prevent repetitive neuronal discharges and hyperexcitability, whereas on the other hand, their inhibition by modulatory neurotransmitters results in local increases in excitation. Inhibition of these channels is thus a double-edged sword that promotes local increases in excitation important to such processes as learning and memory while also potentially rendering areas of the brain proepileptic. The M channel KCNQ1 is mutated in a subtype of long QT syndrome, whereas KCNQ2 and KCNQ3 are mutated in two different forms of benign familial neonatal convulsions.¹⁵

Several other potassium currents present in CNS neurons are beyond the scope of this chapter.¹⁶ These include (1) channels that are opened by intracellular calcium (I_K(Ca)) and function in action potential repolarization and determination of the interspike interval between action potentials; (2) slowly hyperpolarizing currents (I_AHP), blockade of which promotes neuronal excitation; (3) mixed cation currents (I_ha) that are permeant to K⁺ and Na⁺ and are involved in rhythmic burst firing (see the later section “Expression of Ion Currents in Different Neuronal Populations”); (4) the transiently inactivating current I_A, which lengthens the interspike interval between action potentials; and (5) K_ATP channels, an inwardly rectifying subtype that opens in response to lowered intracellular adenosine triphosphate (ATP) and has been implicated in protection from ischemia, insulin-mediated hypothalamic neuronal glucose homeostasis, and glial potassium homeostasis.

By considering the clinical manifestation of diseases caused by altered expression of I_Na, I_K, and I_Ca, it is surprising that permanent changes in the normal endowment of ion channels can cause episodic diseases. This is true for a variety of inheritable cardiac conditions (arrhythmias), as well as neurological disorders such as episodic ataxia and epilepsy. How it is possible that most of the time patients affected by these disorders lack symptoms and what precipitates the clinical manifestations largely remains unknown. The modern techniques used to map and pinpoint the molecular mechanisms of diseases have thus far failed to determine the cofactors that transform a small ion channel deficit into a full-blown neurological disease. Understanding these coexisting conditions will perhaps provide information sufficient to chart an effective therapy. Channelopathies are summarized in Table 49-4.¹⁷

TABLE 49-4 Overview of Ion Channel Disorders Relevant to Neurological Surgery

AChR, acetylcholine receptor; BBB, blood-brain barrier.

Glial Ion Channels and Glutamate Release

Glial potassium channels are the most common electrophysiologic feature of both cultured and in situ astrocytes and can be categorized as follows: channels that allow inward but not outward current flow (inward rectifiers, K_IR), channels that allow outward but not inward current flow (delayed rectifier, I_DR; transient outward current, I_A), and channels that are opened by intracellular calcium (I_K[Ca)). Glial potassium channels differ in their sensitivity to blockers: inward rectifiers are blocked by submillimolar concentrations of external cesium and barium, and outward I_DR and I_A are both sensitive to TEA and 4-aminopyridine (4-AP), but I_A blockade by TEA requires high concentrations.¹⁸ Recently, a mixed cation channel (I_ha) permeant to K⁺ and Na⁺ and ether-a-go-go currents (HERG 1)¹⁹ have been described in both cultured and in situ astrocytes and may assist in potassium homeostasis in the CNS.

Voltage-dependent, TTX-sensitive and TTX-insensitive sodium channels are also expressed in both cultured and in situ glial cells.²⁰ Although astrocytes are incapable of producing action potential–like responses, possibly because of the relatively low Na⁺ current densities in these cells, a role of Na⁺ channels in extracellular buffering of potassium has been proposed. According to this hypothesis, Na⁺ influx sustains the Na⁺,K⁺-ATPase pump, which results in net uptake of K⁺. Finally, calcium channels are represented sparingly in glial cells and require either neuronal or otherwise differentiating factors for expression; whether I_Ca can be recorded from in situ hippocampal astrocytes is still unknown, but release of calcium from intracellular stores in response to neurotransmitters acting on astrocytes has clearly been demonstrated. Relevant to potassium homeostasis, micromolar [Ca⁺²]_in can cause opening of I_K(Ca) and may thus participate in the generation of outward potassium fluxes.

Astrocytes can release the excitatory transmitter glutamate, which acts on at least three families of receptors.^21–25 Astrocytic release of glutamate can occur through several mechanisms, some of which are specific to glia (e.g., reversal of glutamate uptake) or occur via Ca²⁺-dependent exocytosis, which shares similarities to neurotransmission. In addition to glutamate, astrocytes can release a variety of neurotransmitters such as taurine and adenosine. Unlike synaptic transmission, which is specific for a postsynaptic site, release of glutamate by a single astrocyte affects several adjacent neurons, thereby simultaneously controlling the excitability of several neighboring pyramidal cells. This may constitute one of the mechanisms of neuronal synchronization in epilepsy.

If astrocytes release glutamate and have neurotransmitter receptors, what differentiates neurons from glia? Are these phenomena operating in vivo, or are these findings limited to slice preparations? For example, glial cells display intrinsic activity in the absence of neuronal stimulation, but this finding was observed only in vitro. Astrocytes greatly outnumber neurons, and the ratio of astrocytes to neurons is larger in more evolved brain. There is no question about the importance of non-neuronal cells to CNS function and human disease; however, the laboratory findings need to be confirmed in preparations in which artifacts are controlled and known. An important physiologic aspect of the astrocyte in situ is its proximity to capillaries and the perivascular space of arterioles.²⁶ Astrocytes influence maintenance of the blood-brain barrier, which in turn regulates brain ion homeostasis.²⁷ Whether astrocytes also release significant levels of vasodilators or constrictors remains to be proved in vivo in preparations in which the surrounding tissue is intact. However, a clear effect of astrocytes on small-diameter, capillary-like structures has been demonstrated.²⁸ This effect may be important when capillary perfusion pressure is altered by brain edema or other disruptors of brain homeostasis.

Expression of Ion Currents in Different Neuronal Populations

Morphologically distinct neuronal populations in different areas of the CNS are endowed with distinct mixtures of ion channels, which results in widely differing electrophysiologic properties such as RMP, intrinsic ionic conductance, and rhythmicity. Several different patterns of neuronal activity can be classified. Brainstem and spinal cord motor neurons generate single spikes of action potentials that form trains of activity in direct correlation with the degree of depolarization. In contrast to this nearly linear firing pattern is that exhibited by many hippocampal and cortical pyramidal cells that display spike frequency adaptation, in which trains of action potentials decrease in frequency over time. Other neuronal populations, such as thalamic relay neurons, inferior olivary neurons, and some pyramidal cells, have intrinsic rhythmicity that allows the generation of bursts of activity without afferent stimulation. Other cells such as many GABAergic inhibitory interneurons are specialized to generate a high firing frequency (>300 Hz) of short-duration (<1 msec) action potentials. A final type of pattern is that exhibited by cholinergic, serotoninergic, noradrenergic, and histaminergic cells, which innervate large areas of the brain. These discrete cell populations carry out their modulatory function by spontaneously generating low firing frequencies (1 to 10 Hz).

This electrophysiologic heterogeneity affects the function of particular cell populations in the brain. Individual cell types possess particular combinations of the previously described Na⁺, K⁺, and Ca²⁺ currents that merge to result in the patterns of neuronal activity that are found in different areas of the CNS. These patterns can be investigated by in vitro isolated brain slice recordings, as well as by computer modeling simulations, to dissect the individual channel components. For example, relay neurons in the lateral geniculate nucleus that project to the visual cortex are characterized by an RMP of −70 mV during sleep versus −55 mV during the awake state, thus decreasing the transmission of information to the visual cortex during sleep. This sleep-wake variability is modulated by the low-threshold calcium current I_T, which together with the hyperpolarization-activated mixed cationic current I_ha generates rhythmic bursts of action potentials in thalamic relay neurons. During slow-wave sleep, thalamic relay neurons have a relatively hyperpolarized RMP that de-inactivates low-threshold calcium currents and thereby results in I_T-generated slow spikes of activity. Between these spikes, a slowly depolarizing potential is generated by activation of I_ha. Together, these two currents result in spontaneous synchronized bursts of low-frequency action potentials. In contrast, the transition from slow-wave sleep to either wakefulness or rapid eye movement sleep is characterized by relative depolarization of thalamic relay neurons. This depolarization inactivates I_T and I_ha, which results in conversion to a tonic action potential mode in which single spikes are produced one at a time in response to stimulation.¹⁶

Electrical Synaptic Transmission

Because electrical synapses transmit current between cells directly coupled by gap junctions, there is minimal synaptic delay, and intercellular transmission of current is nearly instantaneous. Gap junctions form a low-resistance pathway that allows electric current to flow from one cell to another, thereby resulting in depolarization of the postsynaptic cell. This depolarization can potentially trigger an action potential, thus linking electrical and chemical neurotransmission.

Intercellular electrical communication occurs through specialized channels called gap junctions. Each gap junction channel is made up of a pair of hemichannels contributed by the presynaptic and postsynaptic cell. The hemichannels together form an approximately 1.5-nm channel pore that directly communicates with the cytoplasm of the two cells and allows the passage of small molecules up to 1000 daltons, ions, and intercellular dyes to facilitate the study of gap junction function. Each hemichannel is composed of a connexon, which in turn consists of six identical connexin proteins. The cytoplasmic side of gap junction channels is sensitive to various modulators, including pH and intracellular calcium. Intracellular acidification and elevated intracellular calcium both result in closure of the gap junction channels by electrically uncoupling cells from one another. In contrast, alkalinization increases the degree of intercellular coupling. Electrical neurotransmission can be either bidirectional (nonrectifying) or directionally selective (rectifying), depending on whether the gap junction channels joining the two cells are voltage sensitive. The strength of electrotonic coupling between two cells can be changed by altering the shape or duration of the presynaptic impulse, the junctional conductance, or the conductance of nearby nonjunctional membrane.^30,31

Electrical synaptic transmission is extremely rapid, which facilitates the synchronous firing of large groups of neurons. Although the degree of neuronal electrical synaptic activity is probably underappreciated at the present time, the functional importance of this form of intercellular communication is increasingly being recognized. For example, the normal development of neuronal columnar domains is dependent on gap junction–mediated intercellular signaling. Spontaneous excitation of one or a few trigger neurons subsequently activates other columnar cells via gap junctions. Gap junctions may also coordinate hippocampal CA1 oscillatory behavior. Parvalbumin-containing GABAergic interneurons in the rat hippocampal CA1 region, known to form a network by mutual axosomatic chemical synaptic contacts, also form another network connected by dendrodendritic gap junctions. Gap junctions linking this dendritic network may facilitate the synchronization of oscillatory activities generated in the interneuron network.^32,33

Gap junction communication is found to a much larger degree between glial cells than between neurons in the adult mammalian brain. Glial cell gap junction signaling can be either homotypic (i.e., between two astrocytes) or heterotypic (between an astrocyte and an oligodendroglial cell). Astrocytes are extensively coupled by gap junctions, thus potentially forming a functional syncytium for the regulation of extracellular homeostasis of potassium ion concentration and pH. Neuronal stimulation causes an activity-dependent release of potassium, which results in a local increase in extracellular potassium ([K⁺]_out). After a single action potential, [K⁺]_out increases by 0.01 to 0.02 mM. Despite the firing of large populations of neurons, [K⁺]_out is extremely tightly regulated and even during seizure activity in the hippocampus does not rise above a ceiling level of 12 mM. Additionally, [K⁺]_out returns to normal in a relatively short period. Experiments have suggested that glial uptake and distribution via gap junctions play a pivotal role in conditions in which there is massive [K⁺]_out accumulation, such as epilepsy. Gap junctional communication between glial cells also provides a pathway for long-range metabolite or second messenger signaling. In response to direct astrocytic or neuronal stimulation, intercellular waves of Ca²⁺ can be generated in astrocytes. These waves spread from cell to cell by one of two mechanisms: either directly through gap junctions or via an extracellular route whereby ATP or glutamate released from astrocytes subsequently propagates via binding to P₂ purinergic or glutamatergic receptors.^30,31

Chemical Synaptic Transmission

In contrast to the direct intercellular communication between cells coupled by electrical synapses, chemical synapses are anatomically separated by a 20- to 40-nm-wide synaptic cleft across which neurotransmitter travels from the presynaptic to the postsynaptic cell. The presynaptic cell contains a region of membrane specialized for neurotransmitter release at which synaptic vesicles are clustered, termed the active zone. When an action potential arrives at the presynaptic terminal, voltage-gated calcium channels open. The resultant rise in intracellular calcium results in fusion of the synaptic vesicles with the presynaptic membrane, and neurotransmitters are released into the synaptic cleft by exocytosis. The neurotransmitter molecules are then able to diffuse to receptors on the postsynaptic cell terminal, which results in ion channel opening and alterations in the membrane properties of the postsynaptic cell. The specific postsynaptic neurotransmitter receptors determine whether the result of neurotransmitter binding is excitation or inhibition of the postsynaptic cell. Neurotransmitter receptors can be either ionotropic receptors that directly gate ion channels or metabotropic receptors that indirectly gate ion channels through modulatory activity by second messenger cascades. The direct synaptic activity produced by ionotropic receptors lasts on the order of milliseconds, whereas metabotropic receptor–gated activity lasts seconds to minutes.

Perhaps the best studied example of ionotropic synaptic transmission is at the neuromuscular junction. In a specialized region of muscle membrane termed the end plate, the motor neuron axon innervates the muscle. The axon ends in varicosities termed synaptic boutons, each of which contains a specialized area of membrane termed the active zone. The active zone contains the synaptic vesicles filled with acetylcholine, as well as voltage-gated calcium channels. The synaptic boutons are each positioned over an area of postsynaptic muscle fiber containing a clustering of acetylcholine receptors termed the junctional fold. When the acetylcholine released by presynaptic cells binds to its receptors in the junctional zone, an end-plate potential is generated by flow of both sodium and potassium ions through the nicotinic acetylcholine receptor–gated ionotropic channel. The end-plate potential thus generated results in a depolarization that is large enough to activate voltage-gated sodium channels at the base of the junctional folds; the end-plate potential is then converted into an action potential that propagates along the muscle fiber. In contrast to the acetylcholine-gated channel, which is permeable to both sodium and potassium and blocked by α-bungarotoxin, the voltage-gated sodium channel is permeable only to sodium and is blocked by TTX.

Similar to the neuromuscular junction, most CNS neurons use ionotropic receptors for rapid intercellular signaling. However, in contrast to the monosynaptic excitatory input at the neuromuscular junction, CNS neurons are usually innervated by both excitatory and inhibitory input from multiple cells using both excitatory and inhibitory neurotransmitters. Input from a single excitatory neuron generates a depolarizing excitatory postsynaptic potential (EPSP), whereas inhibitory innervation generates a hyperpolarizing inhibitory postsynaptic potential (IPSP). Excitatory and inhibitory synapses are morphologically different, as summarized in Table 49-5.

TABLE 49-5 Morphologic Differences between Central Nervous System Synapses

GABAergic, transmitting γ-aminobutyric acid.

Contribution of Astrocytes to Regulation of Neuronal Excitability: Spatial Buffering

We will focus on CNS potassium homeostasis as an example of the importance of astrocyte function in the maintenance of extracellular homeostasis. The action potential generation of large-voltage signals is accompanied by an equally impressive accumulation/depletion of ions in the extracellular space (ECS) or in the cytosol. The region of accumulation is easily derived from the Nernst equation: depolarizing potentials generated by the influx of Na⁺ or Ca²⁺ will cause an increase in [Na]_in and [Ca²⁺]_in, whereas repolarization mediated by K⁺ movements will cause an extracellular increase in [K⁺]_out. The concentration of potassium in the ECS (K_ECS) in the mammalian CNS increases measurably (from 3 to about 4 mM) during physiologic stimulation, to a larger extent (up to 12 mM) during seizures or direct, synchronous stimulation of afferent pathways, and to exceedingly high values (>30 mM) during anoxia or spreading depression. Despite these rapid and large changes in K_ECS, K⁺ values return to normal levels in a relatively short period.

Neuronal excitability is regulated by a complex interaction of excitatory and inhibitory potentials. In pyramidal neurons, the depolarizing ion conductances involved in action potential generation are regulated primarily by the voltage-dependent activation/inactivation properties of Na⁺ and Ca²⁺ channels; in addition, inward Na⁺ and Ca²⁺ currents underlie the generation of EPSPs. Termination of these depolarizing potentials occurs through the voltage- and calcium-dependent activation of intrinsic potassium conductances and by activation of interneurons that release inhibitory neurotransmitters to produce IPSPs; the latter are mediated by postsynaptic activation of chloride and potassium currents. Although I_Na, I_Ca, and I_EPSP are, under physiologic conditions, relatively independent of modest changes in the driving force for the permeant ions (because E_Na and E_Ca are remote with respect to cell resting potential), both repolarizing potassium and IPSP conductances are critically affected by even modest changes in cell RMP, [K⁺]_out, and [Cl]_in/[Cl]_out. Because neuronal RMP depends significantly, albeit not exclusively, on [K⁺]_out, maintenance of homeostatic control of extracellular potassium plays a crucial role in the regulation of neuronal firing.

Several mechanisms have been suggested to explain the rapid clearance of K⁺ from the ECS: passive diffusion through the ECS, active removal by blood flow, and neuronal reuptake. However, these mechanisms alone are not fast enough to account for the rapid removal of K⁺ from the ECS seen under experimental conditions. Several lines of evidence suggest that brain glial cells support homeostatic regulation of the neuronal microenvironment. CNS astrocytes are strategically located in proximity to excitable neurons and are sensitive to the changes in extracellular ion composition that follow neuronal activity.^34–39 Experiments have suggested that glial uptake plays a pivotal role in conditions in which there is massive K_out accumulation. In cortical regions, glial cells participate in genesis of the changes in extracellular field potential associated with neuronal depolarization and efflux of potassium in the ECS. The combination of potassium uptake into glial cells immediately followed by redistribution through electrotonically coupled glial gap junctions (“spatial buffering”) provides a valid working hypothesis to explain some of the features of K⁺ movement in the ECS (Fig. 49-7). Although a direct demonstration that K_IR or I_ha plays a role in CNS extracellular potassium homeostasis is still lacking, evidence from studies on cultured neocortical astroglia has demonstrated K⁺-induced inward currents sensitive to K_IR blockers. In addition, recent work has demonstrated evidence of glial spatial buffering during slow sleep oscillations in spike wave seizures in rodents.⁴⁰

FIGURE 49-7 Diagrammatic representation of the proposed “spatial buffering” mechanism underlying transcellular potassium movement across cortical astrocytes. Under resting conditions, both cells are bathed in similar [K⁺]_out, and cell 1 is characterized by a resting potential close to E_K by virtue of its high potassium permeability. Cell 2 is similarly sensitive to changes in E_K but has a more depolarized resting membrane potential. Under these “resting” conditions, gap junctions are kept closed because of a relatively acidic pH_i (1). Increases in [K⁺]_out in proximity to cell 1 cause influx of potassium into cell 1; during this accumulation process, cell depolarization will occur as a result of the sudden shift in E_K (2). Escape of intracellular potassium in this cell cannot occur because of the exclusive presence of potassium currents characterized by inward-going rectification; furthermore, electrochemical communication with cell 2 is precluded by the low conductance of gap junctions. As a result, a net increase in intracellular potassium sufficient to depolarize the cell occurs (3). This results in depolarization-induced alkalization (DIA), subsequent opening of gap junctions, and transport of potassium to cell 2 (4). Finally, outflow of K⁺ occurs from cell 2 (5). R_j represents the junctional resistance (100% at steady state, 50% decrease after DIA); R_e represents the resistance of the extracellular fluid. Only two cells of the syncytium are shown together with the equivalent electric circuit showing the resistive pathways involved in the transfer of extracellular potassium from the extracellular space close to cell 1 to cell 2; the final step implies restitution of potassium to the extracellular space, but other mechanisms, such as backward diffusion or passage across the blood-brain barrier (or both), may act in conjunction with spatial buffering.

It has long been known that in conditions involving high levels of neuronal activity (e.g., seizures), [K]_out accumulation is accompanied by cell swelling. The swelling that accompanies epileptiform neuronal discharge is due to excessive activity of the ionic mechanisms normally involved in the control of ECS homeostasis. One of the several proposed mechanisms associated with cell swelling, the Na⁺/K⁺/2Cl⁻ cotransporter, is also believed to participate in uptake of K⁺ into glia. This transporter is blocked by the loop diuretic furosemide. Treatment of “epileptic” hippocampal slices (treated with bicuculline, 0 Ca²⁺, or 4-AP) with furosemide has been shown to inhibit spontaneous burst discharge. It has been hypothesized that this mechanism is related to furosemide’s blockade of the swelling induced by the large ionic (and water) shifts that accompany Na⁺/K⁺/2Cl⁻ cotransporter activity.^41,42 More recent experimentation has suggested that furosemide-induced opening of glial inwardly rectifying potassium channels may contribute to the antiepileptic mechanism.⁴³

Neuronal K⁺ reuptake and part of the hyperpolarizing undershoot that follows the action potential are mediated in part by an energy-dependent process that requires Na⁺,K⁺-ATPase activity. Similarly, extracellular potassium accumulation in glia may depend on energy-dependent processes. Na⁺,K⁺-ATPase activity is regulated by both [Na⁺]_in and [K]_out. Thus, increases in extracellular potassium or influx of Na⁺ will cause activation of this electrogenic uptake mechanism. As a result, glial cells will accumulate potassium and extrude Na⁺, the net result being hyperpolarization. Whether Na⁺,K⁺-ATPase–dependent potassium uptake plays a significant role in K⁺ buffering is still controversial, partially because of the fact that selective pharmacologic blockade of the glial pump has been difficult.

Loss of Brain Homeostasis and Resultant Effects

It has been recognized for many years that one of the most peculiar histologic features of the mammalian brain is the infinitesimal size of the ECS separating various parenchymal cells. Because permeation of ions across neuronal membranes results in depletion/accumulation of ions undergoing transmembrane movement, it is important to understand how the brain maintains a constant extracellular and intracellular milieu. The essential nature of the latter function becomes apparent if one considers the volumetric rigidity of the skeletal structure encompassing the brain and the limited amount of osmotic changes that such a structure can withstand. This, together with the lack of lymphatic drainage from the CNS, implies that even a small accumulation of ion and water in the CNS requires immediate and complete clearance by mechanisms other than circulation of lymph. For simplicity, the following discussion focuses on changes in extracellular potassium and on the control of potassium homeostasis; similar considerations, however, apply to other physiologically relevant ions (H⁺, Ca²⁺, Na⁺, Cl⁻).

Potassium ions play a fundamental role in the control of neuronal (and cardiac) excitability. Action potential repolarization is curtailed by increased [K⁺]_out, and elevated potassium causes a dramatic decrease in some inhibitory potentials. The most direct effect of elevated extracellular potassium, however, consists of its direct depolarizing effects on neurons and neuronal terminals. Experimentally, the synaptic changes normally achieved by direct stimulation of presynaptic fibers (e.g., long-term potentiation) can be mimicked by small increases in potassium, whereas synchronous firing of thousands of cortical or hippocampal pyramidal cells is observed when [K⁺]_out is slightly elevated. The latter is commonly referred to as “potassium-induced epileptogenesis.”

The delicate homeostatic electrophysiologic balance of the CNS is readily perturbed in association with or as a result of various pathologic conditions such as seizures, brain edema, and brain tumors. In various forms of epilepsy, alterations in neuronal and glial ion channel expression, ECS size, and intercellular astrocytic gap junction coupling may all contribute to the development or maintenance of seizure activity. In temporal lobe epilepsy, the most common surgically treated form of epilepsy, a number of pathologic features are characteristically found, including selective neuronal loss in the CA1, CA3, and hilar regions of the hippocampus, reactive gliosis in these same subfields, and sprouting of granule cell excitatory axon collaterals (mossy fibers). Epileptic granule cells are characterized by alterations in GABA receptors that increase their zinc sensitivity at a time when “mossy fiber sprouting” of zinc-positive granule cell axonal projections has occurred.^44,45 In addition, alterations in either glial or neuronal electrotonic gap junction coupling may promote seizure activity. Neuronal synchrony is enhanced in conditions of increased intercellular coupling, whereas uncoupling agents that block gap junctions may prevent neuronal synchronization.

Glial regulation of ECS size and [K⁺]_out may similarly contribute to epileptogenesis. Astrocytes from animal model or human limbic epilepsy are characterized by decreased expression of the inwardly rectifying potassium channel K_IR.⁴⁶ This channel alteration would be expected to result in raised [K⁺]_out, as well as the need to use other glial mechanisms for potassium uptake such as the Na⁺,K⁺-ATPase pump and the Na⁺/K⁺/2Cl⁻ cotransporter. Together, these changes would be expected to increase epileptogenicity both by increasing [K⁺]_out and by decreasing ECS size. These hypotheses have been confirmed in normal rodent tissue with pharmacologic blockade of the inwardly rectifying potassium channel K_IR,¹² as well as in an animal model of traumatic brain injury.⁴⁷ Investigation is ongoing in human epileptic tissue.

Brain tumors are also associated with various electrophysiologic abnormalities. Gliomas are characterized both by selective expression of a chloride channel not expressed by normal astrocytes and by alterations in astrocytic mechanisms for the uptake of glutamate. The chloride currents expressed by glioma cells are volume activated and are proposed to contribute to the changes in cell shape and volume that are necessary for extracellular tumor cell migration.⁴⁸ Synaptically released glutamate and GABA are both removed from the ECS through the action of electrogenic Na⁺-dependent transporters. In the case of glutamate, five transporter subtypes have been identified, including the astrocytic transporters GLAST and GLT-1, which are the most abundant glutamate transporters in brain. In glioma cells, there is very little expression of GLAST and mislocalization of GLT-1 to the nuclear rather than the cytoplasmic membrane. Rather than take scavenging neuronally released glutamate, glioma cells locally release internally generated glutamate, thereby resulting in increased neurotoxicity and invasiveness. Blockade of glioma glutamate release may hold promise as a novel therapy for malignant gliomas.⁴⁹

Origin of Field Potentials in the Brain and Spinal Cord

Since the discovery of electroencephalographic (EEG) waves, the origin of extracellular field potentials underlying EEG activity has been studied intensively. There is general agreement that the generation, distribution, size, and polarity of extracellular recordings (or EEG recordings) are somehow related to the structural geometry of the CNS. Analogous to the heart and electrocardiogram, it has been assumed that extracellular potentials recorded with microelectrode techniques, scalp electrodes, or corticography are related to neuronal firing and that the polarity of the signals reflects both the distance from the generator cells and the “dipolar” nature of neuronal cells. At the present state of knowledge it is assumed that this explanation is valid only under particular circumstances. Two seemingly opposing points of view have been expressed to explain the origin of field potentials and their relationship to intracellularly recorded signals. For clarity, let us consider the characteristics and generators of each of the most commonly used extracellular recording techniques individually.

Electrical EEG activity measured at the surface of the skull represents a summation of dendritic synaptic potentials in cortical neurons. The EEG rhythms seen are thus thought to represent waves of thalamocortical excitatory synaptic potentials. These field potential spikes of activity are the extracellular reflection of the synchronized activity of large numbers of neurons. Because of the scalp location of the EEG recording electrodes, the surface EEG recording predominantly reflects the activity of cortical neurons that are close to the electrodes rather than activity of deeper hippocampal, thalamic, or brainstem cells. The overall EEG signal is on the level of microvolts, as opposed to the millivolt recordings obtained from single neurons intracellularly, because of distortion and attenuation of the signal between the electrode and the underlying brain tissue. In addition, because the EEG recording is a summated signal from a large number of neurons, the individual contributions of the cells reflected at the surface electrode will depend on whether the contributing cells are more superficial or deeper in the cortex and whether they are receiving EPSPs or IPSPs.

Recording Methods in Neurosurgery

The different recording methods used in the neurosurgical treatment of epilepsy, movement disorders, and pain syndromes are summarized in Table 49-6. Although single-unit recordings carried out with microelectrodes are used for research purposes in epilepsy patients, the majority of epilepsy recordings are EEG recordings using scalp electrodes, subdural or epidural electrode arrays, or intracerebral targeted-depth electrodes. These techniques allow both the lateralization and localization that is required in medically refractory patients before focal cortical resection of their seizure focus. Intraoperative electrocorticography (ECoG) is often used in patients with lesional epilepsies, such as those caused by tumors, focal cortical dysplasia, or cavernous malformations with refractory seizures, to determine whether the epileptogenic zone extends beyond the borders of the pathologic lesion. The use of ECoG in mesial temporal lobe epilepsy to guide surgical resection is more controversial. Intraoperative hippocampal ECoG can potentially be used to determine how much hippocampus to remove to maximize seizure-free outcome. Neocortical ECoG has not been shown to be of clear benefit in the intraoperative management of mesial temporal lobe epilepsy.⁵⁰

With the resurgence over the past several years of stereotactic surgery for the treatment of Parkinson’s disease, essential tremor, dystonia, and other movement disorders, the use of intraoperative physiology has similarly increased. Intraoperative macrostimulation for deep brain mapping of the thalamus, as well as microelectrode recording of neuronal units in the thalamus, globus pallidus, and subthalamic nucleus, is commonly used to guide the placement of lesions or stimulating electrodes for the treatment of these movement disorders. For determining proper lesion or stimulator location in the thalamus of patients with essential tremor or tremor-predominant Parkinson’s disease, thalamic “tremor” cells can be recorded in the ventrolateral motor thalamus. The firing pattern of these cells is synchronized with the tremor in the affected upper extremity, as recorded by electromyography. In addition, thalamic neurons with joint movement-activated (kinesthetic) receptive fields in the limb of interest are somatotopically organized in the target area. During the placement of lesions or stimulating electrodes in the globus pallidus interna or in the subthalamic nucleus in patients with Parkinson’s disease, characteristic neuronal recordings are obtained with microelectrodes in conjunction with frame-based imaging to physiologically and anatomically identify the desired target. In addition, stimulation can be carried out through the microelectrode to confirm proper target location and to minimize potential side effects or complications. Further details of the specific neuronal patterns of recordings obtained and the anatomic pathways mapped during these movement disorder procedures are detailed in the section of this text on movement disorders.

Excitotoxicity

Excitotoxicity refers to the ability of glutamate or other amino acid agonists at the glutamate receptors to mediate the death of central neurons. The most common excitotoxic mechanism consists of sustained exposure to large quantities of these agonists. Excitotoxic neuronal death is almost always a pathologic feature of CNS disorders and may contribute to the pathogenesis of human brain or spinal cord disease. Excitotoxicity has substantial cellular specificity and, in most cases, is mediated by glutamate receptors. N-Methyl-D-aspartate (NMDA) receptor activation is the most common signaling event that causes excitotoxicity, and NMDA triggers injury more rapidly than do α-amino-3-hydroxy-5-methyl-4-isoxazoleproprionate (AMPA) or kainate receptors (Fig. 49-8). It is commonly accepted that this depends on NMDA’s greater ability to induce calcium influx and subsequent cellular calcium overload. A good example of excitotoxicity is hippocampal sclerosis.^51,52 Hippocampal sclerosis is the most frequent pathologic finding in patients with temporal lobe epilepsy undergoing resective neurosurgery, and it is a common neuroimaging finding.⁵³

FIGURE 49-8 Complexity of synaptic transmission. Note the numerous sites that allow modulation of release, action, and metabolism of the excitatory mediator glutamate. AMPA, α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate; EPSP, excitatory postsynaptic potential; LTD, long-term depression; LTP, long-term potentiation; NMDA, N-methyl-D-aspartate.

Recent reports have overturned a series of dogmas that have been well entrenched in the neuroscience literature concerning NMDA-type glutamate receptors (NMDARs). The new data show that NMDARs exist on the myelin sheath formed by oligodendrocytes, that an uncompetitive NMDAR antagonist has successfully passed human clinical trials, and that NMDARs trigger multiple deleterious cascades to inflict cellular damage on both neurons and glia during cerebral ischemia (stroke). These recent findings bode well for clinical intervention with NMDAR antagonists in more neurological disorders than previously thought, including multiple sclerosis, cerebral palsy (periventricular leukomalacia), and spinal cord injury.

Epilepsy and Channelopathies

Epileptic seizures are electrical events that produce a complex behavioral response ranging from absence seizures to convulsions. Several investigators have studied the link between ion channels and seizure disorders. Animal data show that several features of epilepsy can be reproduced in rodents. Electrophysiologic recordings performed both in vivo and in vitro allowed the discovery of several important features that may contribute to the development of seizures. For example, the response to GABA and NMDA varies during development, and an attempt has been made to link abnormal GABA responses to infantile seizures.^54,55

In addition to developmental changes, several insults that are cofactors in human epilepsy have been shown to cause similar effects in animal models. Virtually all animal models have shown either altered excitatory or altered inhibitory loops, but more recent research has revealed a significant role for non-neuronal cells. For example, a specific deficit in potassium channels involved in glial potassium buffering has been discovered in both human tissue and animal model brain.^47,56,57 More recently, involvement of other members of the neurovascular unit has expanded the electrophysiologic mechanisms to other nonexcitable cells (e.g., blood-brain barrier endothelial cells).⁵⁸ Finally, it has to be emphasized that most of the work of the neurovascular unit consists of keeping the neuronal cells well nourished and living in a tightly controlled environment. As also mentioned in other sections of this chapter, altered potassium or calcium can have profound effects on neuronal firing.

A genetic contribution to the epilepsies has long been suspected, but progress in elucidating the specific genetic influences was relatively slow. A recent flurry of new experimental tools has allowed the discovery of several genetic features associated with epileptic disease. Among those that are most likely to influence neuronal discharges are “channelopathies,” or conditions involving the altered electrical property of neuronal or glial cells. Thus far, almost all of the progress in discovering “epilepsy genes” has come from analysis of rare families with mendelian modes of inheritance, and all but two of these genes encode voltage-gated or ligand-gated ion channels. Among these are KCNQ2 and KCNQ3 for potassium channels; SCN1A, SCN1B, and SCN2A for sodium channels; and GABRG2 for inhibitory chloride channels, among others. Genetic analysis of these genes is confounded by the numerous overlapping properties of ion channels. In other words, a specific mutation in a potassium channel may cause a variety of diseases in which seizures are not most prominent. On the other hand, it is possible that environmental or pharmacologic factors can mimic the genetic ablation of function even in the presence of a normal genotype. An example is the impotence drug Viagra (sildenafil), which acts on the KCNQ2 potassium channel. This compound acts as a specific phosphodiesterase type V inhibitor with a selective inhibitory effect on cyclic guanosine monophosphate (cGMP). Its enhancement of the release of nitric oxide (NO) leads to an increase in cGMP concentration, which is responsible for its main clinical effect, relaxation of smooth muscle in the corpus cavernosum. Sildenafil also affects KCNQ2 channels, which are regulated by cGMP/NO pathways. Not surprisingly, four instances of patients experiencing epileptic seizures during the clinical trials of sildenafil were reported, and anecdotal evidence supports the association between sildenafil and potassium channels expressed in the brain.⁵⁹