ELECTRICAL SYNAPSES

Although their existence in the mammalian central nervous system (CNS) has been known for a long time, electrical synapses, or gap junctions, between neurons were thought to be of relatively little importance for the functioning of the adult mammalian CNS. Only recently has it become apparent that these synapses are quite common and that they may underlie important neuronal functions.

An electrical synapse is effectively a low-resistance pathway between cells that allows current to flow directly from one cell to another and, more generally, allows the exchange of small molecules between cells. Electrical synapses are present in the CNS of animals from invertebrates to mammals. They are present between glial cells, as well as between neurons. Electrical coupling of neurons has been demonstrated for most brain regions, including the inferior olive, cerebellum, spinal cord, neocortex, thalamus, hippocampus, olfactory bulb, retina, and striatum.

A gap junction is the morphological correlate of an electrical synapse (see also Chapter 1). These junctions are plaque-like structures in which the plasma membranes of coupled cells become closely apposed (the intercellular space narrows to approximately 3 nm) and filled with electron-dense material (Fig. 6-1). Freeze-fracture electron micrographs of gap junctions display regular arrays of intramembrane particles that correspond to proteins that form the intercellular channels connecting the cells. The typical channel diameter is large (1 to 2 nm), thus making it permeable not only to ions but also other small molecules up to approximately 1 kDa in size.

Figure 6-1 Gap junction structure. A, Schematic view of the gap junction showing narrowing of the intercellular space to 3.5 nm at the junction. The gap junction has multiple channels, with each channel formed by two connexon hemichannels. Each connexon in turn comprises six connexin subunits. B, Electron micrograph of part of a complex synaptic arrangement called a glomerulus that is found in the inferior olive and some other CNS regions. Two dendritic spines are coupled by a gap junction (small black arrows). An axon terminal packed with synaptic vesicles fills the upper right part of the panel. Large arrowheads point to the electron-dense material that marks the active zone. Black dots are immunogold labeling for GABA, thus identifying this terminal as GABAergic.

Electrical synapses are fast (essentially no synaptic delay) and bidirectional (i.e., current generated in either cell can flow across the gap junction to influence the other cell). In addition, they act as low-pass filters. That is, slow electrical events are much more readily transmitted than are fast signals such as action potentials. One important role for neuronal gap junctions appears to be synchronization of network activity. For example, the activity of inferior olivary neurons is normally synchronized but becomes uncorrelated when pharmacological blockers of gap junctions are injected into the inferior olive. It also appears that the patterns of electrical coupling by gap junctions may be highly specific. For example, neocortical interneurons almost exclusively couple to interneurons of the same type. This specific gap junction—coupling pattern suggests that multiple, independent, electrically coupled networks of interneurons may coexist across the neocortex.

AT THE CELLULAR LEVEL

Each gap junction channel is formed by two hemichannels (called connexons), one contributed by each cell. Each connexon, in turn, is a hexamer of connexin protein subunits, which are encoded for by a gene family of at least 21 different members in mammals. (Recently, a second family of proteins that form gap junctions, the pannexins, has also been identified.) Gap junctions formed by different connexins have distinct biophysical properties (gating and conductance) and cellular distributions. Although at least 10 connexin types are expressed in the CNS, connexin 36 (connexins are named according to their molecular weight; thus, the number refers to the approximate molecular weight of the connexin in kilodaltons) is the major neuronal connexin in the adult CNS. Other connexin types found in the CNS form gap junctions between glial cells or are primarily expressed transiently during development.

Finally, although electrical synapses are generally regarded as relatively simple and static in comparison to chemical synapses, they may actually be fairly dynamic entities. For example, the properties of electrical synapses can be modulated by several factors, including voltage, intracellular pH, and [Ca⁺⁺]. Moreover, they are subject to regulation by G protein—coupled receptors, and the connexins contain sites for phosphorylation. These factors can change the coupling between cells by causing changes in single-channel conductance, the formation of new gap junctions, or removal of existing ones.

CHEMICAL SYNAPSES

Chemical synaptic transmission was first demonstrated between the vagus nerve and the heart by a simple experiment by Otto Loewi. The vagus nerve of a frog was stimulated to slow the heart rate down while the solution perfusing the heart was collected. This solution was then used to perfuse a second heart, whose beating slowed on being perfused. The chemical responsible was found to be acetylcholine, which we now know is also a neurotransmitter at the neuromuscular junction and at synapses in the peripheral and central nervous systems.

Unlike the situation at electrical synapses, at chemical synapses there is no direct communication between the cytoplasm of the two cells. Instead, the cell membranes are separated by a synaptic cleft of some 20 μm, and interaction between the cells occurs via chemical intermediaries known as neurotransmitters. Chemical synapses are generally unidirectional, and thus one can refer to the presynaptic and postsynaptic elements that are diagramed in Figure 6-2. The presynaptic element is often the terminal portion of an axon and is packed with small vesicles whose exact shape and size vary with the neurotransmitter that they contain. In addition, the presynaptic membrane apposed to the postsynaptic element has regions, known as active zones, of electron-dense material that corresponds to the proteins involved in transmitter release (Fig. 6-1, B). Moreover, mitochondria and rough endoplasmic reticulum are typically found in the presynaptic terminal. The postsynaptic membrane is also characterized by electron-dense material, which in this case corresponds to the receptors for the neurotransmitter.

Figure 6-2 Schematic of a chemical synaptic terminal releasing all three main classes of neurotransmitter. For each, the mechanisms of release, sites of action, and mechanisms for termination of activity are shown. Real synapses release transmitter from one or more classes.

Chemical synapses occur between different parts of neurons. Traditionally, focus has been placed on synapses formed by an axon onto the dendrites or soma of a second cell (axodendritic or axosomatic synapses), and our description will be based primarily on such synapses. However, there are many additional types of chemical synapses, such as axoaxonic (axon to axon), dendrodendritic (dendrite to dendrite), and dendrosomatic (dendrite to soma). Furthermore, complex synaptic arrangements are possible, such as mixed synapses, in which cells form both electrical and chemical synapses with each other; serial synapses, in which an axoaxonic synapse is made onto the axon terminal and influences the efficacy of that terminal’s synapse with yet a third element; and reciprocal synapses, in which both cells can release transmitter to influence the other. Figure 6-1, B, shows a complex synaptic arrangement, called a glomerulus, that involves both chemical and electrical synapses among the participating elements.

Much of what we know about chemical synapses comes from the study of two classic preparations, the frog neuromuscular junction (the synapse from a motor neuron onto the muscle) and the squid giant synapse (the synapse from a second-order neuron onto third-order neurons that innervate the muscle of the squid’s mantle; i.e., the motor neurons, which are the cells whose axons were used to characterize the conductances underlying the action potential [see Chapter 5]). The principles governing transmission at these synapses mostly apply to synapses within the mammalian CNS as well, at least with regard to synapses using what are called the “classic” neurotransmitters (see the section Neurotransmitters). Thus, much of the following discussion will be based on results from these two preparations; however, some differences in CNS synapses will also be pointed out.

Synaptic transmission at a chemical synapse may be summarized as follows. Synaptic transmission is initiated by arrival of the action potential at the presynaptic terminal. The action potential depolarizes the terminal, which causes Ca⁺⁺ channels to open. The subsequent rise in [Ca⁺⁺] within the terminal triggers the fusion of vesicles containing neurotransmitter with the plasma membrane. The transmitter is then expelled into the synaptic cleft, diffuses across it, and binds to specific receptors on the postsynaptic membrane. Binding of transmitter to receptors then causes the opening (or less often, the closing) of ion channels in the postsynaptic membrane, which in turn results in changes in the potential and resistance of the postsynaptic membrane that alter the excitability of the cell. The changes in membrane potential of the postsynaptic cell are termed excitatory and inhibitory postsynaptic potentials (EPSPs and IPSPs) (Fig. 6-3), depending on whether they increase or decrease, respectively, the cell’s excitability, which can be defined as its probability of firing action potentials. The transmitter acts for only a very short time (milliseconds) because reuptake and degradation mechanisms rapidly clear the transmitter from the synaptic cleft.

Figure 6-3 IPSPs and EPSPs recorded with a microelectrode in a cat spinal motor neuron in response to stimulation of appropriate peripheral afferent fibers. Forty traces are superimposed. Note that these IPSPs are hyperpolarizing, but in some cases IPSPs can be depolarizing’see text for an explanation.

(Redrawn from Curtis DR, Eccles JC: J Physiol 145:529, 1959.)

The succeeding sections will amplify specific points of this summary. However, it is worth mentioning at this point that some of the nonclassic types of neurotransmitters (e.g., neuropeptides and gaseous neurotransmitters such as nitric oxide) and the discovery of metabotropic receptors have required modifications of several aspects of this basic conception (a metabotropic receptor does not contain an ion channel but, instead, is coupled to a G protein that initiates second messenger cascades that can ultimately affect ion channels, whereas an ionotropic receptor contains the ion channel as an integral part of itself). Some of the differences between classic and peptide transmitters are listed in Table 6-1. More details on the properties of peptide and gaseous transmitters are provided in the relevant parts of the Neurotransmitters section of this chapter, and metabotropic receptors are covered in the Receptors section.

Table 6-1 Distinctions between Classic Nonpeptide Neurotransmitters and Peptide Neurotransmitters

Nonpeptide Transmitters	Peptide Transmitters
Synthesized and packaged in the nerve terminal	Synthesized and packaged in the cell body; transported to the nerve terminal by fast axonal transport
Synthesized in active form	Active peptide formed when it is cleaved from a much larger polypeptide that contains several neuropeptides
Usually present in small, clear vesicles	Usually present in large, electron-dense vesicles
Released into a synaptic cleft	May be released some distance from the postsynaptic cell There may be no well-defined synaptic structure
Action of many terminated because of uptake by presynaptic terminals via Na⁺-powered active transport	Action terminated by proteolysis or by the peptide diffusing away
Typically, action has short latency and short duration (msec)	Action may have long latency and may persist for many seconds

Calcium Entry Is the Signal for Transmitter Release

Depolarization of the presynaptic membrane by the action potential causes voltage-gated Ca⁺⁺ channels to open, which makes it possible for Ca⁺⁺ to flow into the terminal and trigger the release of transmitter. However, Ca⁺⁺ will enter the terminal only if there is a favorable electrochemical gradient to do so. Recall that it is the combination of the concentration and voltage gradients that determines the direction of ion flow through open channels. Extracellular [Ca⁺⁺] is high relative to intracellular [Ca⁺⁺], which favors entry into the terminal; however, during the peak of the action potential, the membrane potential is positive, and the voltage gradient opposes the entry of Ca⁺⁺ because of its positive charge. Thus, at the peak of the action potential, relatively little Ca⁺⁺ enters the terminal because although the membrane is highly permeable to Ca⁺⁺, the overall driving force is small. In fact, by using a voltage clamp, one can experimentally make the membrane potential positive and equal to the Nernst equilibrium potential for Ca⁺⁺. If this is done, no Ca⁺⁺ will enter the terminal despite Ca⁺⁺ channels being open, and as a result no transmitter is released and no postsynaptic response is observed. This voltage is known as the suppression potential. If the membrane potential is rapidly made negative again (because of either the end of the action potential or by adjusting the voltage clamp), Ca⁺⁺ rushes into the terminal as a result of the large driving force (which arises instantaneously on repolarization) and the high membrane permeability to Ca⁺⁺ (which remains high because it takes the Ca⁺⁺ channels several milliseconds to close in response to the new membrane potential), thereby resulting in release of transmitter and a postsynaptic response (Fig. 6-4).

Figure 6-4 Presynaptic Ca⁺⁺ current and its relationship to the postsynaptic response. A, Schematic of a squid giant synapse preparation. Electrodes 1 and 2 are used to voltage-clamp the presynaptic terminal and record its voltage and current. Note that TTX and TEA were present to block Na⁺ and K⁺ conductance, in order to isolate the Ca⁺⁺ conductance. Electrode 3 records the membrane potential of the postsynaptic axon. The presynaptic terminal was voltage-clamped to increasingly more depolarized levels (blue traces). With a small depolarization (B), a small Ca⁺⁺ current starts shortly after the voltage step, continues to grow for the duration of the step (on current), and then decays exponentially after its termination (off or tail current). A larger voltage step (C) increases both the on and the off components of the Ca⁺⁺ current, and now distinct on and off responses are observed in the postsynaptic response. D, The voltage step is to the Nernst potential for Ca⁺⁺, so there is no Ca⁺⁺ current during the step, but a large tail current and off response are observed. TTX-tetrodotoxin; TEA, tetraethyl ammonium.

(Based on data of Llinas R, et al: Biophys J 33:323-351, 1981.)

Synaptic Vesicles and the Quantal Nature of Transmitter Release

How neurotransmitter is stored and how it is released are questions fundamental to synaptic transmission. Answering these questions began with two observations. The first was the discovery of small round or irregularly shaped organelles known as synaptic vesicles in presynaptic terminals by electron microscopy (Fig. 6-2). The second observation came from recordings of postsynaptic responses at the neuromuscular junction. Normally, an action potential in a motor neuron causes a large depolarization in the postsynaptic muscle, termed an end plate potential (EPP), which is equivalent to an EPSP in a neuron. However, under conditions of low extracellular [Ca⁺⁺], the EPP amplitude is reduced (because the presynaptic Ca⁺⁺ current is reduced, leading to a smaller rise in intracellular [Ca⁺⁺], and transmitter release is proportional to [Ca⁺⁺]). In this condition, the EPP is seen to fluctuate among discrete values (Fig. 6-5). Moreover, small, spontaneous depolarizations of the postsynaptic membrane, termed miniature end plate potentials (mEPPs), are observable. The amplitude of the mEPP (≤1 mV) corresponds to that of the smallest EPP evoked under low [Ca⁺⁺], and the amplitudes of other EPPs were shown to be integral multiples of the mEPP amplitude; thus, it was natural to propose that each mEPP corresponded to the release of transmitter from a single vesicle and that EPPs represented the combined simultaneous release of transmitter from many vesicles.

Figure 6-5 A, Spontaneous mEPPs recorded at a neuromuscular junction in a fiber of frog extensor digitorum longus. B, EPPs evoked by nerve stimulation under low-[Ca⁺⁺] conditions, which reduce the probability of transmitter release. The small-amplitude EPPs evoked under these conditions vary in amplitude in a step-like manner, where the size of the step is equal to the smallest EPP, which in turn equals the size of the mEPPs (note that in these conditions the stimulus often fails to evoke any response, as indicated by a flat response).

(A, Data from Fatt P, Katz B: Nature 166:597, 1950; B, data from Fatt P, Katz B: J Physiol 117:109, 1952.)

This linking of mEPPs and vesicles implies that each mEPP is caused by the action of many molecules of neurotransmitter binding to postsynaptic receptors. The alternative that each mEPP could be caused by a single transmitter molecule binding to and opening a single postsynaptic receptor was rejected, in part because responses smaller in amplitude than mEPPs could be generated experimentally by directly applying dilute solutions of acetylcholine to the muscle. In fact, mEPPs were calculated to be caused by the action of approximately 10,000 molecules, which corresponds well to estimates of the number of neurotransmitter molecules contained within a single vesicle.

Many additional studies have confirmed the vesicle hypothesis of neurotransmitter release. For example, biochemical studies have shown that neurotransmitter is concentrated in vesicles, and fusion of vesicles to the plasma membrane and their depletion in the terminal cytoplasm after action potentials have been shown with electron microscopic techniques.

Molecular Apparatus Underlying Vesicular Release

The small vesicles that contain nonpeptide neurotransmitters can fuse with the presynaptic membrane only at specific sites, called active zones. To become competent to fuse with the presynaptic membrane at an active zone, a small vesicle must first dock at the active zone. It must then undergo a priming process before the vesicle can fuse and release its transmitter into the synaptic cleft in response to an increase in local cytoplasmic [Ca⁺⁺]. On the order of 25 proteins may play roles in docking, priming, and fusion. Some of these proteins are cytosolic, whereas others are proteins of the vesicle membrane or the presynaptic plasma membrane. The functions of most of these proteins are incompletely understood; however, knowledge of the molecular details of transmitter release has increased dramatically in recent years.

As with other exocytotic processes, neurotransmitter release involves SNARE (soluble N-ethyl maleimide-sensitive factor attachment protein receptor) proteins: v-SNARES in the vesicle membrane and t-SNARES on the (target) presynaptic plasma membrane. Zipper-like interactions between synaptobrevin (a v-SNARE) and syntaxin and SNAP-25 (two t-SNARES) bring the vesicle membrane and the presynaptic plasma membrane close together before fusion. The SNARE proteins are targets for various botulinum toxins, which disrupt synaptic transmission, thus demonstrating their critical role in this process. Nevertheless, they do not bind Ca⁺⁺, so another protein must be the Ca⁺⁺ sensor that triggers the actual fusion event. Although several proteins in the terminal do bind Ca⁺⁺, synaptotagmin is almost certainly the Ca⁺⁺ sensor.

Calcium channels are located in the active zone membrane at sites adjacent to the docked vesicles. When they open, a small area of high [Ca⁺⁺], which lasts for less than a millisecond and is termed a microdomain, is created at the active zone. This local high concentration allows the rapid binding of Ca⁺⁺ to a protein called synaptotagmin, and it is thought that this binding causes a conformational change in synaptotagmin that triggers the fusion event of a docked vesicle. Indeed, the time from Ca⁺⁺ influx to vesicle fusion is about 0.2 msec.

Synaptic Vesicles Are Recycled

During synaptic transmission, vesicles must fuse with the plasma membrane to release their contents into the synaptic cleft. However, there must be a reverse process; otherwise, not only would it be hard to sustain the vesicle population, but the presynaptic membrane’s surface area would also grow with each bout of synaptic transmission, and its molecular content and functionality would likewise change (because, as just discussed, the protein content of the vesicle membrane is distinct from that of the terminal membrane).

There appear to be two distinct mechanisms by which vesicles are retrieved after release of their neurotransmitter content (Fig. 6-6). One mechanism is the endocytotic pathway commonly found in most cell types. Coated pits are formed in the plasma membrane, which then pinch off to form coated vesicles within the cytoplasm of the presynaptic terminal. These vesicles then lose their coat and undergo further transformations (i.e., acquire the correct complement of membrane proteins and be refilled with neurotransmitter) to become once again synaptic vesicles ready for release.

Figure 6-6 Vesicle recycling pathways. Synaptic vesicles have been thought to fuse with the membrane while emptying their contents and then be recycled by forming clathrin-coated pits that are endocytosed to form coated vesicles (1 → [2 or 2′] → 3′ → 1). An alternative pathway that may allow more rapid recycling of vesicles has been proposed. This pathway, called “kiss and run,” involves only transient fusion of the vesicle to the presynaptic membrane to form a pore through which the vesicle contents may be emptied, followed by detachment of the vesicle from the membrane (1 → 2 → 3 → 4 → 5 → 1).

(Redrawn from Valtorta F, Meldolesi J, Fesce R: Trends Cell Biol 11:324, 2001.)

Recently, evidence for a second, more rapid recycling mechanism has been obtained (Fig. 6-6). It involves transient fusion of the vesicle to the synaptic membrane and has been called “kiss and run.” In this case, fusion of the vesicle with the synaptic membrane leads to the formation of a pore through which the transmitter is expelled, but there is no wholesale collapse of the vesicle into the membrane. Instead, the duration of the fusion is very brief, after which the vesicle detaches from the plasma membrane and reseals itself. Thus, the vesicle membrane retains its molecular identity. Its contents can then simply be replenished, thereby making the vesicle ready for use again.

The relative importance of these two mechanisms is still being debated. However, at central synapses, which tend to be small and contain relatively few vesicles in comparison to the neuromuscular junction, the rapid time course of the kiss-and-run mechanism may help avoid the problem of vesicle depletion and the consequent failure of synaptic transmission during periods of high activity (many neurons in the CNS can show sustained firing rates of several hundred hertz, and a few types of neuron can fire at rates of approximately 1000 Hz).

Postsynaptic Potentials

When an action potential triggers release of neurotransmitter from a motor neuron, an EPP is generated in the muscle. More generally, at excitatory synapses throughout the nervous system, action potentials trigger EPSPs in the postsynaptic cell. In both cases there is a depolarization of the membrane that increases the excitability of the cell (i.e., makes it more likely to fire an action potential or, if it is already active, increases the firing rate). The EPP is so large that under normal circumstances it depolarizes the sarcolemma well above the action potential threshold and thus always triggers a spike leading to contraction of the muscle cell. This is an example of a synapse with a high (>1) safety factor (ratio of synaptic potential to the amplitude needed to reach threshold), which makes sense for the neuromuscular junction because each muscle cell is contacted by only a single motor neuron and if that motor neuron is firing, the nervous system has basically made the decision to contract that muscle. In contrast, most neurons receive thousands of excitatory synapses from many different cells. Here, each synapse generates a small EPSP, and thus it takes the summed EPSPs of multiple active synapses to trigger an action potential in the postsynaptic neuron.

In both situations the basic process leading to the EPSP is the same: the neurotransmitter binds to receptors in the postsynaptic cell that open channels to allow an inward current to flow, which in turn leads to depolarization of the membrane. These channels are termed ligand gated because their opening and closing are primarily controlled by the binding of neurotransmitter. This mechanism can be contrasted with the channels underlying the action potential, which open and close in response to changes in membrane potential. However, there are some channels, most notably the NMDA (N-methyl-D-aspartate) channel, that are both ligand and voltage gated.

It is also worth noting here that the preceding description and what follows in this section refer to what happens when neurotransmitter binds to receptors in which the ion channel is part of the receptor itself. These receptors are referred to as ionotropic receptors and underlie what is now called “fast” synaptic transmission. There is also “slow” synaptic transmission, mediated by what are called metabotropic receptors, in which the receptor and ion channel are not part of the same molecule and binding of neurotransmitter to the receptor initiates biochemical cascades that lead to much slower responses (see the section Receptors for details). Despite the differing time courses, many of the same basic principles apply to both types of synaptic potential.

Once the EPSP channels are open, the direction of current flow through them is determined by the electrochemical driving force for the permeant ions. It turns out that the pores of most channels that underlie EPSPs are relatively large and therefore allow passage of most cations with similar ease. As an example, consider the acetylcholine-gated channel that is opened at the neuromuscular junction. Na⁺ and K⁺ are the major cations present (Na⁺ extracellularly and K⁺ intracellularly); therefore, the net current through the channel is approximately the sum of the Na⁺ and K⁺ currents (I^net = I^Na + I^K). Recall that the current through a channel from a particular ion is dependent on two factors: the conductance of the channel to the ion and the driving force on the ion. This relationship is expressed by the equation

Equation 6-1

where g_x is the conductance of the channel to ion x, V_m is the membrane potential, and E_x is the Nernst equilibrium potential for ion x. In this case g_x is similar for Na⁺ and K⁺, so the main determinant of net current is the relative driving forces (V_m − E_x). If the membrane is at its resting potential (typically around −70 mV), there is a strong driving force (V_m − E_Na) for Na⁺ to enter the cell because this potential is far from the Na⁺ Nernst potential (about +55 mV), whereas there is only a small driving force for K⁺ to leave the cell because V_m is close to the K⁺ Nernst potential (about −90 mV). Thus, if acetylcholine-gated channels open when the membrane is at its resting potential, a large inward Na⁺ current and a small outward K⁺ current will flow through the acetylcholine channel, thereby resulting in a net inward current, which acts to depolarize the membrane.