12: Transmitter Release

• Transmitter Release Is Regulated by Depolarization of the Presynaptic Terminal

• Release Is Triggered by Calcium Influx

  • The Relation Between Presynaptic Calcium Concentration and Release

  • Several Classes of Calcium Channels Mediate Transmitter Release

• Transmitter Is Released in Quantal Units

• Transmitter Is Stored and Released by Synaptic Vesicles

  • Synaptic Vesicles Discharge Transmitter by Exocytosis and Are Recycled by Endocytosis

  • Capacitance Measurements Provide Insight into the Kinetics of Exocytosis and Endocytosis

  • Exocytosis Involves the Formation of a Temporary Fusion Pore

  • The Synaptic Vesicle Cycle Involves Several Steps

• Exocytosis of Synaptic Vesicles Relies on a Highly Conserved Protein Machinery

  • The Synapsins Are Important for Vesicle Restraint and Mobilization

  • SNARE Proteins Catalyze Fusion of Vesicles with the Plasma Membrane

  • Calcium Binding to Synaptotagmin Triggers Transmitter Release

  • The Fusion Machinery Is Embedded in a Conserved Protein Scaffold at the Active Zone

• Modulation of Transmitter Release Underlies Synaptic Plasticity

  • Activity-Dependent Changes in Intracellular Free Calcium Can Produce Long-Lasting Changes in Release

  • Axo-axonic Synapses on Presynaptic Terminals Regulate Transmitter Release

• An Overall View

Some of the brain's most remarkable abilities, such as learning and memory, are thought to emerge from the elementary properties of chemical synapses, where presynaptic terminals release chemical transmitters that activate receptors in the membrane of the postsynaptic cell. In the last three chapters we saw how postsynaptic receptors control ion channels that generate the postsynaptic potential. Here we consider how electrical and biochemical events in the presynaptic terminal lead to the secretion of neurotransmitters. In the next chapter we examine the chemistry of the neurotransmitters themselves.

What are the signals at the presynaptic terminal that lead to the release of transmitter? Bernard Katz and Ricardo Miledi first demonstrated the importance of depolarization of the presynaptic membrane through the firing of a presynaptic action potential. For this purpose they used the giant synapse of the squid, a synapse large enough to permit insertion of electrodes into both pre- and postsynaptic structures. Two electrodes are inserted into the presynaptic terminal—one for stimulating and one for recording—and one electrode is inserted into the postsynaptic cell for recording the excitatory postsynaptic potential (EPSP), which provides an index of transmitter release (Figure 12–1A).

When the presynaptic neuron is stimulated it fires an action potential, and after a brief delay an EPSP large enough to trigger an action potential is recorded in the postsynaptic cell. Katz and Miledi then asked how the presynaptic action potential triggers transmitter release. They found that as voltage-gated Na⁺ channels are blocked by application of tetrodotoxin, successive action potentials become progressively smaller. As the action potential is reduced in size, the EPSP decreases accordingly (Figure 12–1B). When the Na⁺ channel blockade becomes so profound as to reduce the amplitude of the presynaptic spike below 40 mV (positive to the resting potential), the EPSP disappears altogether. Thus the amount of transmitter release (as measured by the size of the postsynaptic depolarization) is a steep function of the amount of presynaptic depolarization (Figure 12–1C).

Katz and Miledi next investigated how presynaptic depolarization triggers transmitter release. The action potential is produced by an influx of Na⁺ and an efflux of K⁺ through voltage-gated channels. To determine whether Na⁺ influx or K⁺ efflux is required to trigger transmitter release, Katz and Miledi first blocked the Na⁺ channels with tetrodotoxin. They then asked whether direct depolarization of the presynaptic membrane, by current injection, would still trigger transmitter release. Indeed, depolarization of the pre synaptic membrane beyond a threshold of about 40 mV positive to the resting potential elicits an EPSP in the postsynaptic cell. Beyond that threshold, progressively greater depolarization leads to progressively greater amounts of transmitter release. This result shows that during a normal action potential presynaptic Na⁺ influx is not necessary for release. Rather Na⁺ influx is important only insofar as it depolarizes the membrane enough for transmitter release to occur (Figure 12–2B).

To examine the contribution of K⁺ efflux to transmitter release, Katz and Miledi blocked the voltage-gated K⁺ channels with tetraethylammonium at the same time they blocked the voltage-sensitive Na⁺ channels with tetrodotoxin. They then injected a depolarizing current into the presynaptic terminals and found that the EPSPs were of normal size, indicating that normal transmitter release occurred (Figure 12–2C). Thus neither Na⁺ nor K⁺ flux is required for transmitter release.

In the presence of tetraethylammonium the current pulse elicits a maintained presynaptic depolarization because the K⁺ current that normally repolarizes the presynaptic membrane is blocked. As a result, transmitter release is sustained throughout the current pulse as reflected in the prolonged depolarization of the postsynaptic cell. The sustained depolarization increased the accuracy of the measurements and permitted Katz and Miledi to determine a complete input–output curve relating presynaptic depolarization to transmitter release (Figure 12–2D). They confirmed the steep dependence of transmitter release on presynaptic depolarization. In the range of depolarization over which transmitter release increases (40–70 mV positive to the resting level) a 10 mV increase in depolarization produces a 10-fold increase in transmitter release. Depolarization of the presynaptic membrane above an upper limit produces no further increase in the postsynaptic potential.

Katz and Miledi next turned their attention to Ca²⁺ ions. Earlier, Katz and José del Castillo had found that increasing the extracellular Ca²⁺ concentration enhanced transmitter release, whereas lowering the concentration reduced and ultimately blocked synaptic transmission. Because transmitter release is an intracellular process, these findings implied that Ca²⁺ must enter the cell to influence transmitter release.

Previous work on the squid giant axon membrane had identified a class of voltage-gated Ca²⁺ channels, the opening of which results in a large Ca²⁺ influx because of the large inward electrochemical driving force on Ca²⁺. The extracellular Ca²⁺ concentration, approximately 2 mM in vertebrates, is normally four orders of magnitude greater than the intracellular concentration, approximately 10 ^–7 M at rest. However, because these Ca²⁺ channels are sparsely distributed along the axon they cannot, by themselves, provide enough current to produce a regenerative action potential.

Katz and Miledi found that the Ca²⁺ channels were much more abundant at the presynaptic terminal. There, in the presence of tetraethylammonium and tetrodotoxin, a depolarizing current pulse was sometimes able to trigger a regenerative depolarization that required extracellular calcium, a calcium spike. Katz and Miledi therefore proposed that Ca²⁺ serves dual functions. It is a carrier of depolarizing charge during the action potential (like Na⁺) , and it is a special chemical signal—a second messenger—conveying information about changes in membrane potential to the intracellular machinery responsible for transmitter release. Calcium ions are able to serve as an efficient chemical signal because of their low resting concentration, approximately 10⁵ fold lower than the resting concentration of Na⁺. As a result, the small amounts of ions that enter or leave a cell during an action potential can lead to large percentage changes in intracellular Ca²⁺ that can trigger various biochemical reactions. Proof of the importance of Ca²⁺ channels in release has come from more recent experiments showing that specific toxins that block Ca²⁺ channels also block release.

The properties of the voltage-gated Ca²⁺ channels at the squid presynaptic terminal were measured by Rodolfo Llinás and his colleagues. Using a voltage clamp, Llinás depolarized the terminal while blocking the voltage-gated Na⁺ channels with tetrodotoxin and the K⁺ channels with tetraethylammonium. He found that graded depolarizations activated a graded inward Ca²⁺ current, which in turn resulted in graded release of transmitter (Figure 12–3). The Ca²⁺ current is graded because the Ca²⁺ channels are voltage-dependent like the voltage-gated Na⁺ and K⁺ channels. Calcium ion channels in squid terminals differ from Na⁺ channels, however, in that they do not inactivate quickly but stay open as long as the presynaptic depolarization lasts.

Calcium ion channels are concentrated in presynaptic terminals at active zones, the sites where neurotransmitter is released, exactly opposite the postsynaptic receptors (Figure 12–4). Calcium ions do not diffuse long distances from their site of entry because free Ca²⁺ ions are rapidly buffered by calcium-binding proteins. As a result, Ca²⁺ influx creates a sharp local rise in Ca²⁺ concentration at the active zones. This rise in Ca²⁺ in the presynaptic terminals can be visualized using Ca²⁺-sensitive fluorescent dyes (Figure 12–4B). One striking feature of transmitter release at all synapses is its steep and nonlinear dependence on Ca²⁺ influx; a twofold increase in Ca²⁺ influx can increase the amount of transmitter released by 16-fold. This relationship indicates that at some site, the calcium sensor, the cooperative binding of several Ca²⁺ ions is required to trigger release.

The Relation Between Presynaptic Calcium Concentration and Release

How much Ca²⁺ is necessary to induce release of neurotransmitters? To address this question Bert Sakmann and Erwin Neher and their colleagues measured synaptic transmission in the calyx of Held, a large synapse in the mammalian brain stem that is part of the auditory pathway. This synapse is specialized for very rapid and reliable transmission to allow for precise localization of sound in the environment.

The calyx forms a cup-like presynaptic terminal that engulfs a postsynaptic cell body (Figure 12–5A). To ensure that every action potential results in reliable synaptic transmission, the calyx synapse includes almost a thousand active zones that function as independent synapses. In contrast, synaptic terminals of a typical neuron in the brain contain only a single active zone. Because the calyx terminal is large, it is possible to insert electrodes into both the pre- and postsynaptic structures, much as with the squid giant synapse, and directly measure the synaptic coupling between the two compartments. This paired recording allows a precise determination of the time course of activity in the presynaptic and postsynaptic cells (Figure 12–5B).

These recordings revealed a brief lag of 1 to 2 ms between the onset of the presynaptic action potential and the postsynaptic excitatory synaptic potential, which accounts for what Sherrington termed the synaptic delay. Because Ca²⁺ channels open more slowly than Na⁺ channels, Ca²⁺ does not begin to enter the presynaptic terminal until the membrane has begun to repolarize. Surprisingly, once Ca²⁺ enters the terminal, transmitter is rapidly released with a delay of only a few hundred microseconds. Thus the synaptic delay is largely attributable to the time required to open Ca²⁺ channels. The astonishing speed of Ca²⁺ action indicates that, prior to Ca²⁺ influx, the biochemical machinery underlying the release process must already exist in a primed and ready state.

A presynaptic action potential normally produces only a brief rise in presynaptic Ca²⁺ concentration because the Ca²⁺ channels open only for a short time. In addition, Ca²⁺ influx is localized at the active zone. These two properties contribute to a concentrated local pulse of Ca²⁺ that induces a burst of transmitter release (Figure 12–5B). As we shall see later in this chapter, the duration of the action potential regulates the amount of Ca²⁺ that flows into the terminal and thus the amount of transmitter release.

To determine how much Ca²⁺ is needed to trigger release, the Neher and Sakmann groups introduced into the presynaptic terminal an inactive form of Ca²⁺ that was complexed within a light-sensitive chemical cage. They also loaded the terminals with a fluorescent dye that alters its fluorescence upon binding Ca²⁺ and so can be used to assay the intracellular Ca²⁺ concentration. By uncaging the Ca²⁺ ions with a flash of light they could trigger transmitter release through a uniform, known increase in Ca²⁺ concentration. These experiments revealed that a rise in Ca²⁺ concentration of less than 1 μM is sufficient to induce release of some transmitter, but approximately 10 to 30 μM Ca²⁺ is required to release the amount normally observed during an action potential. Here again the relationship between Ca²⁺ concentration and transmitter release is highly nonlinear, consistent with a model in which four or five Ca²⁺ ions must bind to the Ca²⁺ sensor to trigger release (Figure 12–5C,D).

Several Classes of Calcium Channels Mediate Transmitter Release

Calcium channels are found in all nerve cells and in many non-neuronal cells. In skeletal and cardiac muscle cells Ca²⁺ channels are important for excitation-contraction coupling; in endocrine cells they mediate release of hormones. Neurons contain five classes of voltage-gated Ca²⁺ channels: the L-type, P/Q-type, N-type, R-type, and T-type, each encoded by distinct genes or gene families. Each type has specific biophysical and pharmacological properties and physiological functions (Table 12–1).

Calcium channels are multimeric proteins whose distinct properties are determined by their pore- forming subunit, the α₁-subunit. The α₁-subunit is homologous to the α-subunit of the voltage-gated Na⁺ channel, comprised of four repeats of a domain with six membrane-spanning segments that includes the S4 voltage-sensor and pore-lining P region (see Figure 7–14). Calcium channels also have auxiliary subunits (termed α₂, β, γ, and δ) that modify the properties of the channel formed by the α₁-subunit.

Four of the voltage-gated Ca²⁺ channels—the L-type, P/Q-type, N-type, and R-type—require fairly strong depolarization to be activated (voltages positive to –40 to –20 mV are required) and thus are often referred to as high-voltage-activated Ca²⁺ channels. In contrast, T-type channels open in response to small depolarizations around the threshold for generating an action potential (–60 to –40 mV) and are therefore called low-voltage-activated Ca²⁺ channels. Because they are activated by small changes in membrane potential, the T-type channels help control excitability at the resting potential and are an important source of the excitatory current that drives the rhythmic pacemaker activity of certain cells in both brain and heart.

In neurons the rapid release of conventional transmitters associated with fast synaptic transmission is mediated mainly by the P/Q-type and N-type Ca²⁺ channels because these are the channel types concentrated at the active zone. The localization of N-type Ca²⁺ channels at the frog neuromuscular junction has been visualized using a fluorescence-labeled snail toxin that binds selectively to these channels (see Figure 12–4A). The L-type channels are not found in the active zone and thus do not normally contribute to the fast release of conventional transmitters such as ACh and glutamate. However, Ca²⁺ influx through L-type channels is important for slower forms of release that do not occur at specialized active zones, such as the release of neuropeptides from neurons and of hormones from endocrine cells. As we shall see below, regulation of Ca²⁺ influx into presynaptic terminals controls the amount of transmitter release and hence the strength of synaptic transmission.

Voltage-gated Ca²⁺ channels are responsible for certain acquired and genetic diseases. A point mutation in the α₁-subunit of P/Q-type channels underlies an inherited form of migraine. A mutation in the α₁-subunit of L-type channels inhibits the voltage-dependent inactivation of these channels, and underlies Timothy syndrome, a pervasive developmental disorder involving both impaired cognitive function and a severe form of autism. Patients with Lambert-Eaton syndrome, an autoimmune disease associated with muscle weakness, make antibodies to the L-type channel α₁-subunit that decrease total Ca²⁺ current.

How does the influx of Ca²⁺ trigger release? Katz and his colleagues provided the key insight into this question by showing that transmitter is released in discrete amounts they called quanta. Each quantum of transmitter produces a postsynaptic potential of fixed size, called the quantal synaptic potential. The total postsynaptic potential is made up of a large number of quantal potentials. EPSPs seem smoothly graded in amplitude only because each quantal (or unit) potential is small relative to the total potential.

Katz and Fatt obtained the first clue as to the quantal nature of synaptic transmission in 1951 when they observed spontaneous postsynaptic potentials of approximately 0.5 mV in the nerve-muscle synapse of the frog. Like end-plate potentials evoked by nerve stimulation, these small depolarizing responses were largest at the site of nerve-muscle contact and decayed electrotonically with distance (see Figure 9–5). Small spontaneous potentials have since been observed in mammalian muscle and in central neurons. Because the postsynaptic potentials at vertebrate nerve-muscle synapses are called end-plate potentials, Fatt and Katz called these spontaneous potentials miniature end-plate potentials.

Several results convinced Fatt and Katz that the miniature end-plate potentials represented responses to the release of small amounts of ACh, the neurotransmitter used at the nerve-muscle synapse. The time course of the miniature end-plate potentials and the effects of various drugs on them are indistinguishable from the properties of the end-plate potential. Like the end-plate potentials, the miniature end-plate potentials are enhanced and prolonged by prostigmine, a drug that blocks hydrolysis of ACh by acetylcholinesterase. Conversely, they are reduced and finally abolished by agents that block the ACh receptor. The miniature end-plate potentials represent responses to small packets of transmitter that are spontaneously released from the presynaptic nerve terminal in the absence of an action potential. Their frequency can be increased by a small depolarization of the presynaptic terminal. They disappear if the presynaptic motor nerve degenerates and reappear when a new motor synapse is formed.

What could account for the small, fixed size of the miniature end-plate potential (around 0.5–1 mV)? Del Castillo and Katz first tested the possibility that each event represents a response to the opening of a single ACh receptor-channel. Small amounts of ACh applied to the frog muscle end-plate elicited depolarizing postsynaptic responses that were much smaller than the 0.5 mV response of a miniature end-plate potential. This finding made it clear that the miniature end-plate potential represents the opening of more than one ACh receptor-channel. In fact, Katz and Miledi were later able to estimate the voltage response to the elementary current through a single ACh receptor-channel as being only approximately 0.3 μV (see Chapter 9). Based on this estimate a miniature end-plate potential of 0.5 mV would represent the summation of the elementary currents of approximately 2,000 channels. Later work showed that a miniature end-plate potential is the response to the synchronous release of approximately 5,000 molecules of ACh.

What is the relationship of the large end-plate potential evoked by nerve stimulation and the small, spontaneous miniature end-plate responses? This question was addressed by del Castillo and Katz in a study of synaptic signaling at the nerve-muscle synapse bathed in a solution low in Ca²⁺. Under this condition the end-plate potential is reduced markedly, from the normal 70 mV to about 0.5 to 2.5 mV. Moreover, the amplitude of each successive end-plate potential now varies randomly from one stimulus to the next; often no response can be detected at all (termed failures). However, the minimum response above zero—the unit EPSP in response to a presynaptic action potential—is identical in amplitude (approximately 0.5 mV) and shape to the spontaneous miniature end-plate potentials. Importantly, the amplitude of each end-plate potential is an integral multiple of the unit potential (Figure 12–6).

Now del Castillo and Katz could ask: How does the rise of intracellular Ca²⁺ that accompanies each action potential affect the release of transmitter? They found that increasing the external Ca²⁺ concentration does not change the amplitude of the unit synaptic potential. However, the proportion of failures decreases and the incidence of higher-amplitude responses (composed of multiple quantal units) increases. These observations show that an increase in external Ca²⁺ concentration does not enhance the size of a quantum of transmitter (that is, the number of ACh molecules in each quantum) but rather acts to increase the average number of quanta that are released in response to a presynaptic action potential (Box 12–1). The greater the Ca²⁺ influx into the terminal, the larger the number of transmitter quanta released.

Thus three findings led del Castillo and Katz to conclude that transmitter is released in packets with a fixed amount of transmitter, a quantum: The amplitude of the end-plate potential varies in a stepwise manner at low levels of ACh release, the amplitude of each step increase is an integral multiple of the unit potential, and the unit potential has the same mean amplitude as that of the spontaneous miniature end-plate potentials.

In the absence of an action potential the rate of quantal release is low—only one quantum per second is released spontaneously at the end-plate. In the presence of a normal concentration of extracellular Ca²⁺ the firing of an action potential in the presynaptic terminal at the vertebrate nerve-muscle synapse releases approximately 150 quanta, each approximately 0.5 mV in amplitude, resulting in a large end-plate potential. Thus when Ca²⁺ enters the presynaptic terminal during an action potential, it dramatically increases the rate of quantal release by a factor of 150,000, triggering the synchronous release of about 150 quanta in about one millisecond.

What anatomical features of the cell might account for the quantum of transmitter? The physiological observations indicating that transmitter is released in fixed quanta coincided with the discovery, through electron microscopy, of accumulations of small clear vesicles in the presynaptic terminal. Del Castillo and Katz speculated that the vesicles are organelles for the storage of transmitter, each vesicle stores one quantum of transmitter (amounting to several thousand molecules), and each vesicle releases its entire contents into the synaptic cleft in an all-or-none manner at sites specialized for release.

The sites of release, the active zones, contain a cloud of synaptic vesicles that cluster above a fuzzy electron-dense material attached to the internal face of the presynaptic membrane (see Figure 12–4A). At all rapid synapses the vesicles are typically clear, small, and ovoid, with a diameter of approximately 40 nm. Although most synaptic vesicles do not contact the active zone, some are physically bound. These are called the docked vesicles and are thought to be the ones immediately available for release. At the neuromuscular junction the active zones are linear structures (see Figure 12–4), whereas in central synapses they are disc-shaped structures approximately 0.1 μm² in area with dense projections pointing into the cytoplasm. Active zones are always precisely apposed to the postsynaptic membrane patches that contain the neurotransmitter receptors. Thus presynaptic and postsynaptic specializations are functionally and morphologically attuned to each other. As we shall learn later, several key active zone proteins involved in transmitter release have now been identified and characterized.

Quantal transmission has been demonstrated at all chemical synapses so far examined with only one exception: at the synapse between photoreceptors and bipolar neurons in the retina (Chapter 26). Nevertheless, the efficacy of transmitter release from a single presynaptic cell onto a single postsynaptic cell varies widely in the nervous system and depends on several factors: (1) the number of synapses between a pair of presynaptic and postsynaptic cells (ie, the number of presynaptic boutons that contact the postsynaptic cell); (2) the number of active zones in an individual synaptic terminal; and (3) the probability that a presynaptic action potential will trigger release of one or more quanta of transmitter at an active zone (Box 12–1).

In the central nervous system most presynaptic terminals have only a single active zone where an action potential usually releases at most a single quantum of transmitter in an all-or-none manner. However at some central synapses, such as the calyx of Held, the presynaptic terminal may contain many active zones and thus can release a large number of quanta in response to a single presynaptic action potential. Central neurons also differ in the number of synapses that a typical presynaptic cell makes with a typical postsynaptic cell. Whereas most central neurons form only a few synapses with any one postsynaptic cell, a single climbing fiber forms up to 10,000 terminals on a single Purkinje neuron in the cerebellum! Finally, the mean probability of transmitter release from a single active zone also varies widely among different presynaptic terminals, from less than 0.1 (that is, a 10% chance that a presynaptic action potential will trigger release of a vesicle) to greater than 0.9. This wide range of probabilities can even be seen among the individual boutons at different synapses between one specific type of presynaptic cell and one specific type of postsynaptic cell.

Thus central neurons vary widely in the efficacy and reliability of synaptic transmission. Synaptic reliability is defined as the probability that an action potential in a presynaptic cell leads to some measurable response in the postsynaptic cell—that is, the probability that a presynaptic action potential releases one or more quanta of transmitter. Efficacy refers to the mean amplitude of the synaptic response, which depends on both the reliability of synaptic transmission and on the mean size of the response when synaptic transmission does occur.

Most central neurons communicate at synapses that have a low probability of transmitter release. The high failure rate of release at most central synapses (ie, their low release probability) is not a design defect but serves a purpose. As we discuss below, this feature allows transmitter release to be regulated over a wide dynamic range, which is important for learning and memory. In synaptic connections where a low probability of release is deleterious for function, this limitation is overcome by simply having many active zones in one synapse, as is the case at the calyx of Held and the nerve-muscle synapse. Both contain hundreds of independent active zones, so an action potential reliably releases 150 to 250 quanta, ensuring that a presynaptic signal is always followed by a postsynaptic action potential. Reliable transmission at the neuromuscular junction is essential for survival. An animal would not survive if its ability to move away from a predator was hampered by a low-probability response.

Not all chemical signaling between neurons depends on synapses. Some substances, such as certain lipid metabolites and the gas nitric oxide (see Chapter 11), can diffuse across the lipid bilayer of the membrane. Others can be moved out of nerve endings by carrier proteins if their intracellular concentration is sufficiently high. Plasma membrane transporters for glutamate or γ-aminobutyric acid (GABA) normally take up transmitter into a cell from the synaptic cleft following a presynaptic action potential (see Chapter 13). However, in certain glial cells of the retina, the direction of glutamate transport can be reversed under certain conditions, causing glutamate to leave the cell through the transporter into the synaptic cleft. Still other substances simply leak out of nerve terminals at a low rate. Surprisingly, approximately 90% of the ACh that leaves the presynaptic terminals at the neuromuscular junction does so through continuous leakage. This leakage is ineffective, because it is diffuse and not targeted to receptors at the end-plate region, and because it is continuous and low level rather than synchronous and concentrated.

Synaptic Vesicles Discharge Transmitter by Exocytosis and Are Recycled by Endocytosis

The quantal hypothesis of del Castillo and Katz has been amply confirmed by direct experimental evidence that synaptic vesicles do indeed package neurotransmitter and that they release their contents by directly fusing with the presynaptic membrane, a process termed exocytosis.

Forty years ago Victor Whittaker discovered that the synaptic vesicles in the motor nerve terminals of the electric organ of the fish Torpedo contain a high concentration of ACh. Later, Thomas Reese and John Heuser and their colleagues obtained electron micrographs that caught vesicles in the act of exocytosis. To observe the brief exocytotic event, they rapidly froze the nerve-muscle synapse by immersing it in liquid helium at precisely defined intervals after the presynaptic nerve was stimulated. In addition, they increased the number of quanta of transmitter discharged with each nerve impulse by applying the drug 4-aminopyridine, a compound that blocks certain voltage-gated K⁺ channels, which increases the duration of the action potential, thereby enhancing Ca²⁺ influx.

These techniques provided clear images of synaptic vesicles at the active zone during exocytosis. Using a technique called freeze-fracture electron microscopy, Reese and Heuser noted deformations of the presynaptic membrane along the active zone immediately after synaptic activity, which they interpreted as invaginations of the cell membrane caused by fusion of synaptic vesicles. These deformations lay along one or two rows of unusually large intramembranous particles, visible along both margins of the presynaptic density and now thought to be the voltage-gated Ca²⁺ channels (Figure 12–7). The particle density (approximately 1,500 per μm²) is approximately that of the Ca²⁺ channels essential for transmitter release. Moreover, the proximity of the particles to the release site is consistent with the short time interval between the onset of the Ca²⁺ current and the release of transmitter.

Finally, Heuser and Reese found that these deformations are transient; they occur only when vesicles are discharged and do not persist after transmitter has been released. Thin-section electron micrographs revealed a number of omega-shaped (Ω) structures that have the appearance of synaptic vesicles that have just fused with the membrane, prior to the complete collapse of the vesicle membrane into the plasma membrane (Figure 12–7B). Heuser and Reese confirmed this idea by showing that the number of Ω-shaped structures is directly correlated with the size of the EPSP when they varied the concentration of 4-aminopyridine to alter the amount of transmitter release. These morphological studies provide striking evidence that transmitter is released from synaptic vesicles by means of exocytosis.

Following exocytosis, the excess membrane added to the presynaptic terminal is retrieved. When Heuser and Reese obtained images of presynaptic terminals 10 to 20 seconds after stimulation, they observed new structures at the plasma membrane, the coated pits, which represent membrane retrieval through the process of endocytosis (Figure 12–7C). Several seconds later the coated pits are seen to pinch off from the membrane and appear as coated vesicles in the cytoplasm. As we will see below, endocytosis through coated pit formation represents one of several means of vesicle membrane retrieval.

Capacitance Measurements Provide Insight into the Kinetics of Exocytosis and Endocytosis

In certain neurons with large presynaptic terminals the increase in surface area of the plasma membrane during exocytosis can be detected in electrical measurements as increases in membrane capacitance. As we saw in Chapter 6, the capacitance of the membrane is proportional to its surface area. Neher discovered that one could use measurements of capacitance to monitor exocytosis in secretory cells.

In adrenal chromaffin cells (which release epinephrine and norepinephrine) and in mast cells of the rat peritoneum (which release histamine and serotonin) individual dense-core vesicles are large enough to permit measurement of the increase in capacitance associated with fusion of a single vesicle. Release of transmitter in these cells is accompanied by stepwise increases in capacitance, followed somewhat later by stepwise decreases, which reflect the retrieval and recycling of the excess membrane (Figure 12–8).

In neurons the changes in capacitance caused by fusion of single, small synaptic vesicles are usually too low to resolve. In certain favorable synaptic preparations that release large numbers of vesicles (such as the giant presynaptic terminals of bipolar neurons in the retina), membrane depolarization triggers a transient smooth rise in the total capacitance of the terminal as a result of the exocytosis and retrieval of the membrane from hundreds of individual synaptic vesicles (Figure 12–8C). These results provide direct measurements of the rates of membrane fusion and retrieval.

Exocytosis Involves the Formation of a Temporary Fusion Pore

Morphological studies of mast cells using rapid freezing suggest that exocytosis depends on the formation of a temporary fusion pore that spans the membranes of the vesicle and plasma membranes. In electrophysiological studies of capacitance increases in mast cells, a channel-like fusion pore was detected in the electrophysiological recordings prior to complete fusion of vesicles and cell membranes. This fusion pore starts out with a single-channel conductance of approximately 200 pS, similar to that of gap-junction channels, which also bridge two membranes. During exocytosis the pore rapidly dilates, probably from around 5 to 50 nm in diameter, and the conductance increases dramatically (Figure 12–9A).

The fusion pore is not just an intermediate structure leading to exocytosis of transmitter, as transmitter can be released through the pore prior to full fusion. This was first shown by amperometry, a method that uses an extracellular carbon-fiber electrode to detect certain amine neurotransmitters, such as serotonin, based on an electrochemical reaction between the transmitter and the electrode that generates an electrical current proportional to the local transmitter concentration. Firing of an action potential in serotonergic cells leads to a large transient increase in electrode current, corresponding to the exocytosis of the contents of a single dense-core vesicle. In some instances these large transient increases are preceded by smaller, longer-lasting current signals that reflect leakage of transmitter through a fusion pore that flickers open and closed several times prior to complete fusion (Figure 12–9B).

Transmitter can also be released solely through transient fusion pores, that is, without full collapse of the vesicle membrane into the plasma membrane. Capacitance measurements for exocytosis of both large dense-core vesicles and small clear vesicles in neuroendocrine cells show that the fusion pore can open and close rapidly and reversibly. The reversible opening and closing of a fusion pore represents a very rapid method of membrane retrieval. The circumstances under which the small clear vesicles at fast synapses discharge transmitter through a fusion pore, as opposed to full membrane collapse, are uncertain.

The Synaptic Vesicle Cycle Involves Several Steps

When firing at high frequency, a typical neuron is able to maintain a high rate of transmitter release. This can result in the exocytosis of a large number of vesicles, more than the number originally present within the presynaptic terminal. To prevent the supply of vesicles from being rapidly depleted, used vesicles are rapidly retrieved and recycled. Because nerve terminals are usually some distance from the cell body, replenishing vesicles by synthesis in the cell body and transport to the terminals would be too slow to be practical.

Synaptic vesicles are released and reused in a simple cycle (Figure 12–10A). Vesicles fill with neurotransmitter and cluster in the nerve terminal. They then dock at the active zone where they undergo a complex priming process that makes vesicles competent to respond to the Ca²⁺ signal that triggers the fusion process.

Three mechanisms exist for retrieving the synaptic vesicle membrane following exocytosis and each has a distinct time course. The first, most rapid mechanism involves the reversible opening and closing of the fusion pore, without the full collapse of the vesicle membrane into the plasma membrane (Figure 12–10B1). In the kiss-and-stay pathway the vesicle remains at the active zone after the fusion pore closes, ready for a second release event. In the kiss-and-run pathway the vesicle leaves the active zone after the fusion pore closes, but is competent for rapid rerelease. Vesicles are thought to be preferentially recycled through these pathways during stimulation at low frequencies.

Stimulation at higher frequencies recruits a second, slower recycling pathway that uses clathrin to retrieve the vesicle membrane after fusion with the plasma membrane (see Figure 12–10B2). (The clathrin-coated vesicle membranes are the coated pits observed by Heuser and Reese.) In this pathway the retrieved vesicular membrane must be recycled through an endosomal compartment before the vesicles can be reused. Clathrin-mediated recycling requires up to a minute for completion, and appears to shift from the active zone to the membrane surrounding the active zone (see Figure 12–7).

A third mechanism operates after prolonged high-frequency stimulation. Under these conditions large membranous invaginations into the presynaptic terminal are visible, which are thought to reflect membrane recycling through a process called bulk retrieval (Figure 12–10B3).

Biochemists have isolated and purified many key proteins of synaptic vesicles, as well as their interacting partners in the plasma membrane (Figure 12–11). One key class of vesicle proteins are the neurotransmitter transporters (Chapter 13). These transmembrane proteins use energy stored in a proton gradient across the vesicle membrane to pump transmitter molecules into the vesicle from the cytoplasm, against their concentration gradient.

Other synaptic vesicle proteins target vesicles to their release sites, participate in the discharge of transmitter by exocytosis, and mediate recycling of the vesicle membrane. The protein machinery involved in these three steps is conserved from worms to humans, and forms the basis for the regulated release of neurotransmitter. We consider each of these steps in turn.

The Synapsins Are Important for Vesicle Restraint and Mobilization

The vesicles outside the active zone represent a reserve pool of transmitter. Paul Greengard discovered a family of proteins, synapsins, that are important regulators of the reserve pool of vesicles (Figure 12–11). Synapsins are peripheral membrane proteins that are bound to the cytoplasmic surface of synaptic vesicles. They also bind adenosine triphosphate (ATP) and actin.

The synapsins are substrates for both protein kinase A and Ca²⁺/calmodulin-dependent protein kinase I. When the nerve terminal is depolarized and Ca²⁺ enters, the synapsins become phosphorylated by the kinase and are thus released from the vesicles, a step that is thought to mobilize the reserve pool of vesicles for transmitter release. Indeed, genetic deletion of synapsins or application of a synapsin antibody leads to a decrease in the number of synaptic vesicles in the nerve terminal and a decrease in the ability of a terminal to maintain a high rate of transmitter release during repetitive stimulation.

SNARE Proteins Catalyze Fusion of Vesicles with the Plasma Membrane

Because a membrane bilayer is a stable structure, fusion of the synaptic vesicle and plasma membrane must overcome a large unfavorable activation energy. This is accomplished by a family of fusion proteins now referred to as SNAREs (soluble N-ethylmaleimide–sensitive factor attachment receptors) (Figure 12–12).

SNAREs are universally involved in membrane fusion, from yeast to humans. They mediate both constitutive membrane trafficking during the movement of proteins from the endoplasmic reticulum to the Golgi apparatus to the plasma membrane, as well as synaptic vesicle trafficking important for regulated exocytosis. SNAREs have a conserved protein sequence, the SNARE motif, that is 60 residues long. They come in two forms. Vesicle SNAREs or v-SNAREs (also referred to as R-SNAREs because they contain an important central arginine residue) reside in the vesicle membranes. Target-membrane SNAREs or t-SNAREs (also referred to as Q-SNAREs because they contain an important glutamine residue) are present in target membranes, such as the plasma membrane. In biochemical experiments using purified v-SNAREs and t-SNAREs in solution, four SNARE motifs bind tightly to each other to form an α-helical coiled-coil complex (Figure 12–12B).

Each synaptic vesicle contains a single type of v-SNARE called synaptobrevin (also called vesicle-associated membrane protein or VAMP). By contrast, the presynaptic active zone contains two types of t-SNARE proteins, syntaxin and SNAP-25. (Synaptobrevin and syntaxin have one SNARE motif; SNAP-25 has two.) The first clue that synaptobrevin, syntaxin, and SNAP-25 are all involved in fusion of the synaptic vesicle with the plasma membrane came from the finding that all three proteins are substrates for botulinum and tetanus toxins, bacterial proteases that are potent inhibitors of transmitter release. James Rothman then provided the crucial insight that these three proteins interact in a tight biochemical complex.

How does formation of the SNARE complex drive synaptic vesicle fusion? During exocytosis the SNARE motif of synaptobrevin, on the synaptic vesicle, forms a tight complex with the SNARE motifs of SNAP-25 and syntaxin, on the plasma membrane (Figure 12–12). The crystal structure of the SNARE complex suggests that this complex draws the membranes together. The ternary complex of synaptobrevin, syntaxin, and SNAP-25 is extraordinarily stable. The energy released in its assembly is thought to draw the negatively charged phospholipids of the vesicle and plasma membranes in close apposition, forcing them into a prefusion intermediate state (Figure 12–12). Such an unstable state may start the formation of the fusion pore and account for the rapid opening and closing (flickering) of the fusion pore observed in electrophysiological measurements. The SNARE proteins may form part of the fusion pore based upon studies that have shown that mutations in the transmembrane region of syntaxin alter the single-channel conductance of the fusion pore.

However, the SNAREs do not fully account for fusion of the synaptic vesicle and plasma membranes. Reconstitution experiments with purified proteins in lipid vesicles indicate that synaptobrevin, syntaxin, and SNAP-25 can catalyze fusion; but the in vitro reaction shows little regulation by Ca²⁺ and the reaction is much slower and less efficient than vesicle fusion in a real synapse. One important additional protein required for exocytosis of synaptic vesicles is Munc18 (mammalian unc18 homolog). Homologs of Munc18, referred to as SM proteins (sec1/Munc18-like proteins), are essential for intracellular fusion reactions in general. Munc18 binds to syntaxin before the SNARE complex assembles by an unknown mechanism. Deletion of Munc18 prevents all synaptic fusion in neurons. The core fusion machinery is thus composed of SNARE and SM proteins that are modulated by various accessory factors specific for particular fusion reactions.

After fusion, the SNARE complex must be disassembled for efficient vesicle recycling to occur. Rothman discovered that a cytoplasmic ATPase called NSF (N-ethylmaleimide-sensitive fusion protein) binds to SNARE complexes via an adaptor protein called SNAP (soluble NSF-attachment protein, not related to the SNARE protein SNAP-25). NSF and SNAP use the energy of ATP hydrolysis to dissociate SNARE complexes, thereby regenerating free SNARE (Figure 12–12A). NSF also participates in the cycling of AMPA-type glutamate receptors in dendritic spines.

Calcium Binding to Synaptotagmin Triggers Transmitter Release

Because fusion of synaptic vesicles with the plasma membrane must occur within a fraction of a millisecond, it is thought that most proteins responsible for fusion are assembled prior to Ca²⁺ influx. According to this view, once Ca²⁺ enters the presynaptic terminal it binds a Ca²⁺ sensor on the vesicle, triggering immediate fusion of the membranes.

A family of closely related synaptic vesicle proteins, the synaptotagmins, has been identified as the major Ca²⁺ sensors that trigger fusion. The synaptotagmins are membrane proteins with a single N-terminal transmembrane region that anchors them to the synaptic vesicle (Figure 12–13A). The cytoplasmic region of each synaptotagmin protein is largely composed of two domains, the C2 domains, which are a common protein motif homologous to the Ca²⁺ and phospholipid-binding C2 domain of protein kinase C. That the C2 domains bind not only Ca²⁺ but also phospholipids is consistent with their importance in calcium-dependent exocytosis. Moreover, the synaptotagmins bind Ca²⁺ over a concentration range similar to the Ca²⁺ concentration required to trigger transmitter release.

The two C2 domains bind a total of five Ca²⁺ ions, the same number of Ca²⁺ ions that electrophysiological experiments reveal are required to trigger release of a quantum of transmitter (Figure 12–13B). The binding of the Ca²⁺ ions to synaptotagmin is thought to act as a switch, promoting the interaction of the C2 domains with phospholipids. The C2 domains of synaptotagmin also interact with SNARE proteins, through both calcium-independent and calcium-dependent reactions.

Studies with mutant mice in which synaptotagmin 1 is deleted or its Ca²⁺ affinity is altered provide further evidence that synaptotagmin is the physiological Ca²⁺ sensor. When the affinity of synaptotagmin for Ca²⁺ is decreased twofold, the Ca²⁺ required for transmitter release is changed by the same amount. When synaptotagmin 1 is deleted in mice, flies, or worms, an action potential is no longer able to trigger fast synchronous release. However, Ca²⁺ is still capable of stimulating a slow form of release referred to as asynchronous release (Figure 12–13C). Thus, although synaptotagmin 1 is not required for all forms of calcium-mediated transmitter release, it is essential for fast synaptic transmission.

The Fusion Machinery Is Embedded in a Conserved Protein Scaffold at the Active Zone

A defining feature of fast synaptic transmission is that neurotransmitter is released by exocytosis at the active zone. Other types of exocytosis, such as that which occurs in the adrenal medulla, do not require a specialized domain of the plasma membrane. The active zone is thought to coordinate and regulate the docking and priming of synaptic vesicles to ensure the speed and tight regulation of release. This is accomplished through a conserved set of proteins that form one large macromolecular structure.

An exceedingly detailed view of the structure of the active zone at the frog neuromuscular junction has been obtained by Jack MacMahan using a powerful ultrastructural technique called scanning electron microscopic tomography. This technique has shown how synaptic vesicles are tethered to the membrane by a series of electron-dense projections, termed ribs and beams, that attach to defined sites on the vesicles and to particles (pegs) in the presynaptic membrane that may correspond to voltage-gated Ca²⁺ channels (Figure 12–14).

A key goal in understanding how the various synaptic vesicle and active zone proteins are coordinated during exocytosis is to fit the various proteins that have been identified into this high-resolution electron micrograph. Several cytoplasmic proteins have been identified that are thought to be components of a cytoskeletal matrix at the active zone. These include two large cytoplasmic multidomain proteins, Munc13 (not related to the Munc18 protein discussed above) and RIM, which form a tight complex with each other and may well comprise part of the ribs and beams. The binding of RIM and Munc13 is an important component of the priming of synaptic vesicles for exocytosis. Phosphorylation of RIM by cAMP-dependent protein kinase is likely to be a key regulatory mechanism in the long-term enhancement of transmitter release implicated in certain forms of learning and memory. As we will see below, regulation of Munc13 by second messengers is involved in short-term forms of synaptic plasticity. RIM binds several other synaptic vesicle proteins, including the Rab3 family of low molecular weight guanosine triphosphatases (GTPases). There are four Rab3 isoforms (Rab3A, B, C, and D) that transiently associate with synaptic vesicles as the GTP–Rab3 complex (Figure 12–11). The binding of RIM and Rab3 is thought to regulate the interaction of synaptic vesicles with the active zone during the vesicle cycle.

The effectiveness of chemical synapses can be modified for short and long periods, a property called synaptic plasticity. Such functional modification can be effected by intrinsic or extrinsic signals. An example of an intrinsic signal is rapid firing; extrinsic signals include direct synaptic input from other neurons and more diffuse actions of neuromodulators.

Synaptic strength can be modified presynaptically, by altering the release of neurotransmitter, or postsynaptically, by modulating the response to transmitter, or both. Long-term changes in presynaptic and postsynaptic mechanisms are crucial to development and learning (Chapters 66 and 67). Here we focus on how synaptic strength changes through modulation of the amount of transmitter released.

Transmitter release can be modulated dramatically and rapidly—by several-fold in a matter of seconds—and this change can be maintained for seconds, to hours, or even days. In principle, such changes can be mediated by two different mechanisms: changes in Ca²⁺ influx or changes in the amount of transmitter released in response to a given Ca²⁺ concentration.

Synaptic strength commonly is enhanced by intense activity. In many neurons a high-frequency train of action potentials is followed by a period during which a single action potential produces successively larger postsynaptic potentials (Figure 12–15A). High-frequency stimulation of the presynaptic neuron, which in some cells can generate 500 to 1,000 action potentials per second, is called tetanic stimulation. The increase in size of the EPSP during tetanic stimulation is called potentiation; the increase that persists after tetanic stimulation is called post-tetanic potentiation (Figure 12–15A). This enhancement usually lasts several minutes, but it can persist for an hour or more at some synapses. The opposite effect, a decrease in the size of postsynaptic potentials, occurs in response to more prolonged periods of high-frequency stimulation. This effect is referred to as synaptic depression (Figure 12–15B).

Activity-Dependent Changes in Intracellular Free Calcium Can Produce Long-Lasting Changes in Release

Because transmitter release depends strongly on the intracellular Ca²⁺ concentration, mechanisms that affect the concentration of free Ca²⁺ in the presynaptic terminal also affect the amount of transmitter released. Normally the rise in Ca²⁺ in the presynaptic terminal in response to an action potential is rapidly buffered by cytoplasmic calcium-binding proteins and mitochondria; Ca²⁺ is also actively transported out of the neuron by pumps and transporters.

However, during tetanic stimulation so much Ca²⁺ flows into the axon terminals that the Ca²⁺ buffering and clearance systems become saturated. This leads to a temporary excess of Ca²⁺, called residual Ca²⁺. The residual free Ca²⁺ enhances synaptic transmission for many minutes or longer by activating certain enzymes that are sensitive to enhanced levels of resting Ca²⁺, such as, the Ca²⁺/calmodulin-dependent protein kinase. Activation of such calcium-dependent enzymatic pathways is thought to increase the priming of synaptic vesicles in the terminals.

Here then is the simplest kind of cellular memory! The presynaptic cell stores information about the history of its activity in the form of residual Ca²⁺ in its terminals. This Ca²⁺ acts by multiple pathways that have different halftimes of decay. In Chapters 66 and 67 we shall see how posttetanic potentiation at certain synapses is followed by an even longer-lasting process (also initiated by Ca²⁺ influx), called long-term potentiation, which can last for many hours or even days.

During prolonged tetanic stimulation synaptic vesicles become depleted at the active zone, resulting in synaptic depression. To counteract this depression, the synapse utilizes multiple mechanisms that often involve a complex of two of the active zone proteins discussed above, Munc13 and RIM. For example, prolonged tetanic stimulation activates phospholipase C, which produces inositol 1,4,5-trisphosphate (IP₃) and diacylglycerol. Diacylglycerol directly interacts with a protein domain on Munc13 called the C1 domain (no relation to the C2 domain) homologous to the diacyl glycerol-binding domain in protein kinase C. Binding of diacylglycerol to Munc13 during prolonged repetitive stimulation helps maintain a high rate of transmitter release even in the face of depression by accelerating the rate of synaptic vesicle recycling. As a result, deletion of Munc13 or RIM results, among other actions, in an enhancement in depression (Figure 12–15B).

Axo-axonic Synapses on Presynaptic Terminals Regulate Transmitter Release

Synapses are formed on axon terminals as well as the cell body and dendrites of neurons (see Chapter 10). Although axosomatic synaptic actions affect all branches of the postsynaptic neuron's axon (because they affect the probability that the neuron will fire an action potential), axo-axonic actions selectively control individual terminals of the axon. One important action of axo-axonic synapses is to increase or decrease Ca²⁺ influx into the presynaptic terminals of the postsynaptic cell, thereby enhancing or depressing transmitter release, respectively.

As we saw in Chapter 10, when one neuron releases transmitter that hyperpolarizes the cell body (or dendrites) of another, it decreases the likelihood that the postsynaptic cell will fire; this action is called postsynaptic inhibition. In contrast, when a neuron makes synapses onto the axon terminal of another cell, it can reduce the amount of transmitter that will be released by the postsynaptic cell onto a third cell; this action is called presynaptic inhibition (Figure 12–16A). Other axo-axonic synaptic actions can increase the amount of transmitter released by the postsynaptic cell; this action is called presynaptic facilitation (Figure 12–16B). Both presynaptic inhibition and facilitation can occur in response to activation of ionotropic or metabotropic receptors in the membrane of the presynaptic terminals. For reasons that are not well understood, presynaptic modulation usually occurs early in sensory pathways.

The best-analyzed mechanisms of presynaptic inhibition and facilitation are in invertebrate neurons and vertebrate mechanoreceptor neurons (whose cell bodies lie in dorsal root ganglia). Three mechanisms for presynaptic inhibition have been identified in these cells. One mechanism depends on the activation of ionotropic GABA receptor-channels in the presynaptic terminal. The opening of these channels leads to an increased conductance to Cl^‒, which decreases (or short-circuits) the amplitude of the action potential in the presynaptic terminal. The smaller depolarization activates fewer Ca²⁺ channels, thereby decreasing transmitter release.

The other two mechanisms both result from the activation of presynaptic G protein-coupled metabotropic receptors. One type of action results from the modulation of ion channels. As we saw in Chapter 11, the βγ-subunit complex of G proteins can simultaneously close voltage-gated Ca²⁺ channels and open K⁺ channels. This decreases the influx of Ca²⁺ and enhances repolarization of the presynaptic terminal, thus diminishing transmitter release. The second type of G protein- dependent action depends on a direct action by the βγ-subunit complex on the release machinery itself, independent of any changes in ion channel activity or Ca²⁺ influx. This second action is thought to involve a decrease in the Ca²⁺ sensitivity of the release machinery.

In contrast, presynaptic facilitation can be caused by enhanced influx of Ca²⁺. In certain molluscan neurons serotonin acts through cAMP-dependent protein phosphorylation to close K⁺ channels in the presynaptic terminal (including the Aplysia S-type K⁺ channel discussed in Chapter 11). This action increases the duration of the presynaptic action potential, thereby increasing Ca²⁺ influx by enabling the voltage-dependent Ca²⁺ channels to remain open for a longer period. In other cells activation of presynaptic ionotropic receptors increases transmitter release. This facilitation can be caused directly in the case of Ca²⁺-permeable receptor-channels by directly enhancing Ca²⁺ influx, or indirectly in the case of voltage-gated Ca²⁺ channels through depolarization of the presynaptic terminal, which leads to increased Ca²⁺ influx.

Thus, presynaptic terminals are endowed with a variety of mechanisms that allow for the fine-tuning of the strength of synaptic transmission. Although we know a fair amount about short-term changes in synaptic strength—changes that last minutes and hours—we are only beginning to learn about mechanisms that support changes that persist for days, weeks, and longer. These long-term changes often require alterations in gene expression and growth of presynaptic and postsynaptic structures in addition to alterations in Ca²⁺ influx and enhancement of transmitter release from existing terminals.

In his book Ionic Channels of Excitable Membranes, Bertil Hille summarizes the importance of calcium in neuronal function as follows:

Neither Na⁺ influx nor K⁺ efflux is required to release neurotransmitter at a synapse. Only Ca²⁺, which enters the cell through voltage-gated channels in the presynaptic terminal, is essential. Synaptic delay—the time between the onset of the action potential and the release of transmitter—largely reflects the time it takes for voltage-gated Ca²⁺ channels to open and for Ca²⁺ to trigger the discharge of transmitter from synaptic vesicles.

Transmitter is packaged in vesicles that each contain approximately 5,000 transmitter molecules. The all-or-none release of transmitter from a single vesicle results in a quantal synaptic potential. Synaptic potentials evoked by nerve stimulation are comprised of integral multiples of the quantal potential caused by the release of transmitter from multiple synaptic vesicles. Increasing the extracellular Ca²⁺ does not change the size of the quantal synaptic potential. Rather, it increases the probability that a vesicle will discharge its transmitter. As a result, the number of vesicles that release transmitter is increased, leading to a larger postsynaptic potential. In addition, transmitter is spontaneously released at low rates from synaptic vesicles, producing what are called spontaneous miniature synaptic potentials.

Electron microscopic images of presynaptic terminals that have been rapidly frozen following electrical stimulation reveal that synaptic vesicles release transmitter by exocytosis. The vesicle membrane fuses with the plasma membrane in the vicinity of the active zone, allowing the transmitter to flow out of the cell toward receptors on a postsynaptic cell. A conserved protein machinery mediates the exocytosis and rapid retrieval of synaptic vesicles that allows nerve terminals to maintain a high rate of transmitter release, even during prolonged trains of action potentials.

Vesicle fusion is thought to be catalyzed by the assembly of a synaptic vesicle SNARE protein (v-SNARE) with a pair of SNARE proteins in the presynaptic target membrane (t-SNARES) into a tight complex that forces the synaptic vesicle and plasma membrane into close proximity. Two other proteins, Munc13 and Munc18, are essential for synaptic vesicle fusion, probably because they enable SNARE complex assembly. Calcium triggers the opening of the vesicles during a late step in the synaptic vesicle cycle by binding to synaptotagmin, a synaptic vesicle protein that serves as the major Ca²⁺ sensor in exocytosis. The machinery that executes fusion and Ca²⁺ triggering is embedded in the active zone by a conserved protein scaffold that also regulates transmitter release. A critical component of this scaffold is a large multidomain protein called RIM that binds to Rab3, a GTP-binding protein on synaptic vesicles, and also interacts with Munc13.

The amount of transmitter released from a neuron is not fixed but can be modified by both intrinsic and extrinsic modulatory processes. High-frequency stimulation produces an increase in transmitter release called posttetanic potentiation. This potentiation, which lasts a few minutes, is caused by an action of the residual Ca²⁺ remaining in the terminal after the large Ca²⁺ signal that occurs during the train of action potentials has been largely dissipated by diffusion, buffering, and extrusion. Neurotransmitters acting at axo-axonic synapses can facilitate or inhibit release of transmitter by altering the steady-state level of resting Ca²⁺, the Ca²⁺ influx during the action potential, or the functioning of the release machinery.