14

Transmitter Release

RegulatedTransmitter Release IsPresynaptic Terminal

TransmiUer Release Is Triggered by Calcium Influx

TransmiUer Is Released in Quanta! Units

TransmiUer Is Stored and Released by Synaptic Vesicles

Synaptic Vesicles Discharge TransmiUer by Exocytosis

Exocytosis Involves the Formation of a Fusion Pore

Synaptic Vesicles Are Recycled

A Variety of Proteins Are Involved in the Vesicular Releaseof Transmitter

The Amount of TransmiUer Released Can Be Modulated byRegulating the Amount of Calcium Influx During theAction Potential

Intrinsic Cellular Mechanisms Regulate theConcentration of Free Calcium

Axo-axonic Synapses on Presynaptic Terminals RegulateIntracellular Free Calcium

An Overall View

S OME OF THE BRAIN'S MOST remarkable feats, such aslearning and memory, are thought to emerge fromthe elementary properties of chemical synapses.

The distinctive feature of these synapses is that actionpotentials in the presynaptic terminals lead to the re-lease of chemical transmitters. In the past three chapterswe saw how postsynaptic receptors for these transmit-ters control the ion channels that generate the postsyn-aptic potential. Now we return to the presynaptic celland consider how electrical events in the terminal arecoupled to the secretion of neurotransmitters. In the

chemistry of thenext chapter we shall examine theneurotransmitters themselves.

by Depolarization of the

Transmitter Release Is Regulatedby Depolarization of the Presynaptic Terminal

How does an action potential in the presynaptic celllead to the release of transmitter? The importance of de-polarization of the presynaptic membrane was demon-strated by Bernard Katz and Ricardo Miledi using thegiant synapse of the squid. This synapse is large enoughto permit the insertion of two electrodes into the pre-synaptic terminal (one for stimulating and one forrecording) and an electrode into the postsynaptic cell forrecording the synaptic potential, which provides an in-dex of transmitter release.

The presynaptic cell typically produces an action p0-tential with an amplitude of 110 mY, which leads totransmitter release and the generation of a large synapticpotential in the postsynaptic cell. The action potential isproduced by voltage-gated Na + influx and K+ efflux.Katz and Miledi found that when the voltage-gated Na +

channels are blocked upon application of tetrodotoxin,successive presynaptic action potentials become pro-gressively smaller, owing to the progressive blockade ofNa + channels during the onset of tetrodotoxin's effect.

The postsynaptic potential is reduced accordingly. Whenthe Na + channel blockade becomes so profound as to re-

duce the amplitude of the presynaptic spike below40 m V, the synaptic potential disappears altogether (Fig-ure 14-1B). Thus, transmitter release (as measured by thesize of the postsynaptic potential) shows a steep depen-dence on presynaptic depolarization.

Part ill / Elementary interactiON Between Neurons: Synaptic Transmission254

A Experimental setup

~~~

figure 14-1 The contribution of voltage-gated Na+ chan-nels to transmitter release is tested by blocking the chan-nels and measuring the amplitude of the presynaptic actionpotential and the resulting postsynaptic potential. (Adaptedfrom Katz and Miledi 1967a.)A. Recording electrodes are inserted in both the pre- and post-synaptic fibers of the giant synapse in the stellate ganglion of asquid.

~~

B. Tetrodotoxin (TTX) is added to the solution bathing the cellin order to block the voltag~ated Na+ channels. The ampli-tudes of both the presynaptic action potential and the postsyn-aptic potential gradually decrease. After 7 min the presynapticaction potential can still produce a suprathreshold synaptic p0-tential that triggers an action potential in the postsynaptic cell(1). After 14 and 15 min the presynaptic spike gradually be-comes smaller and produces smaller synaptic potentials (2 and

~

.How does membrane depolarization cause trans-

mitter release? One possibility, suggested by the aboveexperiment, is that Na + influx may be the important fac-

tor. However, Katz and Miledi were able to show thatsuch influx is not necessary. While the Na + channels

were still fully blocked by tetrodotoxin, Katz and Miledidirectly depolarized the presynaptic membrane bypassing depolarizing current through the second intra-cellular microelectrode. Beyond a threshold of about40 m V from the resting potential, progressively greateramounts of transmitter are released (as judged by theappearance and amplitude of the postsynaptic poten-tial). In the range of depolarization at which chemical

B Potential when Na+channels are blocked

C, Input-output curveof transmitter rele8se

TTX+7m1n

Pre

k:=-It:::=~4

TTX + 14 min

CaT1X + 16 min

TIX> 16min

3). When the presynaptic spike is reduced to 40 mV or less. itfails to produce a synaptic potential (4).C. An input-output curve for transmitter release can be inferredfrom the dependence of the amplitude of the synaptic potentialon the amplitude of the presynaptic action potential. This is 0b-tained by stimulating the presynaptic nerve as the Na + chan-nels for the presynaptic action potential are progressivelyblocked. 1. A 40 mV presynaptic depolarization is required toproduce a synaptic potential. Beyond this threshold there is asteep increase in amplitude of the synaptic potential in re-sponse to small changes in the amplitude of the presynapticpotential. 2. The semilogarithmic plot of the data in the input-output curve illustrates that the relationship between the pre-synaptic spike and the postsynaptic potential is logarithmic. A10 mV increase in the presynaptic spike produces a 1o-fold in-crease in the synaptic potential.

transmitter is released (40-70 mV above the restinglevel), a to mV increase in depolarization produces ato-fold increase in transmitter release. Thus, the presyn-aptic terminal is able to release transmitter without aninflux of Na +. The Na + influx is important only insofar

as it depo1arizes the membrane enough to generate theaction potential necessary for transmitter release.

Might the voltage-gated K+ efflux triggered by theaction potential be responsible for release of transmit-ter? To examine the contribution of K+ efflux to trans-mitter release, Katz and Meledi blocked the voltage-gated K+ channels with tetraethylammonium at thesame time they blocked the voltage-sensitive Na + chan-

A Experimental setup B Potentials whenNa+ channelsare blocked

11)(~-, Post ~Pre

-c2

3

~

:<'~Post

4

2OmV I

!O~2ma

Figure 14-2 Blocking the voltage-sensitive Na+ channelsand K+ channels in the presynaptic terminals affects theamplitude and duration of the presynaptic action potentialand the resulting postsynaptic potential, but does not blockthe release of transmitter. (Adapted from Katz and Miledi1967a.)A. The experimental arrangement is the same as in Figure 14-1, except that a current-passing electrode has been insertedinto the presynaptic cell. (TEA = tetraethylammonium.)B. The voltage-gated Na + channels are completely blocked by

adding tetrodotoxin (TTX) to the cell-bathing solution. Each setof three traces represents (from bottom to top) the depolarizingcurrent pulse injected into the presynaptic terminal (I). the re-sulting potential in the presynaptic terminal (Pre), and the post-synaptic potential generated as a result of transmitter releaseonto the postsynaptic cell (Post). Progressively stronger currentpulses are applied to produce correspondingly greater depolar-izations of the presynaptic terminal (2-4), These presynapticdepolarizations cause postsynaptic potentials even in the ab-sence of Na + flux. The greater the presynaptic depolarization,the larger the postsynaptic potential. indicating that membranepotential exerts a direct control over transmitter release. The

nels with tetrodotoxin. They then passed a depolarizingcurrent through the presynaptic terminals and foundthat the postsynaptic potentials nonetheless were ofnormal size, indicating that normal transmitter releaseoccurred (Figure 14-2). Indeed, under the conditions ofthis experiment, the presynaptic potential is maintainedthroughout the current pulse because the K+ currentthat normally repolarizes the presynaptic membrane isblocked. As a result, transmitter release is sustained(Figure 14-2C). Increases in the presynaptic potential

Chapter 14/ Transmitter Release 255

D Input-output curve of transmitter releaseC Potentials whenK+ channelsare blocked

TTX+TEA

Post LJ. 1

.I'L-~

'2

.J\....-

..r-- 3

.r-L-~~ 4 130mv

. T 2OOmV

~...r-"I-

presynaptic depolarizations are not maintained throughout theduration of the depolarizing current pulse because of the de-layed activation of the voltage-gated K+ channels. whichcauses repolarization.c. After the voltage-gated Na + channels of the action potential

have been blocked, tetraethylammonium (TEA) is injected intothe presynaptic terminal to block the voltage-gated K+ chan-nels as well. Each set of three traces represents current pulse,presynaptic potential, and postsynaptic potential as in partB. Because the presynaptic K+ channels are blocked, thepresynaptic depolarization is maintained throughout the currentpulse. The large sustained presynaptic depolarizations producelarge sustained postsynaptic potentials (2-4). This indicatesthat neither Na+ nor K+ channels are required for effectivetransmitter release.D. Blocking both the Na+ and K+ channels permits the mea-surement of a more complete input-output curve than that inFigure 14-1. In addition to the steep part of the curve, there isnow a plateau. Thus, beyond a certain level of presynaptic de-polarization, further depolarization does not cause any addi-tional release of transmitter. The initial level of the presynapticmembrane potential was about - 70 mY.

above an upper limit produce no further increase inpostsynaptic potential (Figure 14-20). Thus, neitherNa + nor K+ flux is required for transmitter release.

Transmitter Release Is Triggeredby Calcium Influx

Katz and Miledi then turned their attention to ea2+ions. Earlier, Jose del Castillo and Katz had found that

Part m / Elementary Interactions Between Neurons: Synaptic 1i'ansuUssion256

Pos1synepticpotential

~

~P\'8ynIpCIc t \ ~ 20 mV COIYWI'WId L

potentiII -" I

2ma

Figure 14-3 A simple experiment demonstrates that trans-mitter release is a function of Ca2+ influx into the presynap-tic terminal. The voltage-sensitive Na+ and K+ channels in asquid giant synapse are blocked by tetrodotoxin and tetraethyl-ammonium. The presynaptic terminal is voltage-clamped andthe membrane potential is stepped to six different commandlevels of depolarization (bottom traC88). The amount of pre-synaptic inward Ca2+ current (middle traces) that accompaniesthe depolarization correlates with.the amplitude of the resultingpost~ptic potential (top traces). This is because the amountof Ca + current through voItage-gated channels determines the

amount of transmitter released. which in tum determines thesize of the postsynaptic potential. The notch in the postsynap-tic potential trace is an artifact that results from turning off thepresynaptic command potential. (Adapted from Llin6s andHeuser 1977.)

increasing the extracellular Ca2+ concentration en-hanced transmitter release; lowering the extracellularCa2+ concentration reduced and ultimately blockedsynaptic transmission. However, since transmitter re-lease is an intracellular process, th~ findings impliedthat Ca2+ must enter the cell to influence transmitter re-lease.

Previous work on the squid giant axon had identi-fied a class of voltage-gated Ca2+ channels. As there is av~ large inward electrochemical driving force onCa + -the extracellular Ca2+ concentration is normally

four orders of magnitude greater than the intracellularconcentration-opening of voltage-gated Ca2+ channelswould result in a large Ca2+ influx. These Ca2+ channelsare, however, sparsely distributed along the main axon.Katz and Miledi proposed that the Ca2+ channels mightbe much more abundant at the presynaptic terminal andthat Ca2+ might serve dual functions: as a carrier of de-

tic Transmission

polarizing charge during the action potential (like Na +)and as a special signal conveying information aboutchanges in membrane potential to the intracellular ma-chinery responsible for transmitter release.

Direct evidence for the presence of a voltage-gatedCa2+ current at the squid presynaptic terminal was pro-vided by Rodolfo LliMs and his colleagues. Using amicroelectrode voltage clamp, Uinas depolarized theterminal while blocking the voltage-gated Na + and K+

channels with tetrodotoxin and tetraethylammonium,respectively. He found that graded depolarizations ac-tivated a graded inward ea2+ current, which in turn re-sulted in graded release of transmitter (Figure 14-3). Theea2+ current is graded because the ea2+ channelspossess voltage-dependent activation gates, like thevoltage-gated Na + and K+ channels. The ea2+ channelsin the squid terminals differ from Na + channels, how-ever, in that they do not inactivate quickly but stay openas long as the presynaptic depolarization lasts. Onestriking feature of transmitter release at all synapses isits steep and nonlinear dependence on Ca2+ influx-atwo-fold increase in ea2+ influx can increase transmitterrelease up to 16-fold. This relationship indicates that atsome site-ca1led the allcium sensor-the binding of upto four ea2+ ions is required to ~ release.

Even in the axon terminal Ca + currents are smalland are normally masked by Na + and K+ currents,

which are 10-20 times larger. However, in the region ofthe active zone (the site of transmitter release) ea2+ in-flux is 10 times greater than elsewhere in the terminal.This localization is consistent with the distribution ofintramembranous particles seen in freeze-fracture elec-tron micrographs and thought to be the ea2+ channels(see Figure 14-7 in Box 14-2).

The localization of ea2+ channels at active zonesprovides a high, local rise in ea2+ concentration at thesite of transmitter release during the action potential. In-deed, during an action potential the Ca2+ concentrationat the active zone can rise more than a thousandfold (to-100 JAM) within a few hundred microseconds. Thislarge and rapid increase is required for the rapid syn-chronous release of transmitter. The calcium sensor re-sponsible for fast transmitter release is thought to havea low affinity forea2+. On the order of ~loo JAM intra-cellular ea2+ is required to trigger release, whereas only-1 JAM of ea2+ is required for many enzymatic reac-tions. Because of the low-affinity calcium sensor, releaseonly takes place in a narrow region surrounding the in-tracellular mouth of a ea2+ channel, the only locationwhere the ea2+ concentration is sufficient to trigger re-lease. The requirement for a high concentration of Ca2+also ensures that release will be rapidly terminatedupon repolarization. Once the Ca2+ channels close, the

high local Ca2+ concentration dissipates rapidly (within1 ms) because of diffusion.

Calcium channels open somewhat more slowlythan the Na + channels and therefore Ca2+ influx doesnot occur until the action potential in the presynapticcell has begun to repolarize (Figure 14-4). The delay thatis characteristic of chemical synaptic transmission-thetime from the onset of the action potential in the presyn-aptic terminals to the onset of the postsynaptic poten-tial-is due in large part to the time required for Ca2+channels to open in response to depolarization. How-ever, because the voltage-dependent Ca2+ channels arelocated very close to the transmitter release sites, Ca2+needs to diffuse only a short distance, permitting trans-mitter release to occur within 0.2 ms of Ca2+ entry!

As we shall see later in this chapter, the duration ofthe action potential is an important determinant of theamount of Ca2+ that flows into the terminal. H the ac-tion potential is prolonged, more Ca2+ flows into thecell and therefore more transmitter is released, causing a

greater postsynaptic potential.Calcium channels are found in all nerve cells as well

as in cells outside the nervous system, such as skeletaland cardiac muscle cells, where the channels are impor-tant for excitation-contraction coupling, and endocrinecells, where they mediate release of hormones. Thereare many types of Ca2+ channels-called L, P /Q N, R,and T -with specific biophysical and pharmacologicalproperties and different physiological functions. Thedistinct properties of these channel types are deter-mined by the identity of their pore-forming subunit(termed the aI-subunit), which is encoded by a familyof related genes (Table 14-1). Calcium channels alsohave associated subunits (termed a2, 13, 'V, and 8) thatmodify the properties of the channel formed by theaI-subunits. All aI-subunits are homologous to thevoltage-gated Na + channel a-subunits, consisting offour repeats of a basic domain containing six transmem-brane segments (including an 54 voltage-sensor) and apore-lining P region (see Figure 9-14).-

Most nerve cells contain more than one type ofCa2+ channel. Channels formed from the differentaI-subunits can be distinguished by their differentvoltage-dependent gating properties, their distinctivesensitivity to pharmacological blockers, and their spe-cific physiological function. The L-type channels are se-lectively blocked by the dihydropyridines, a class ofclinically important drugs used to treat hypertension.The P/Q-type channels are selectively blocked by (J)-aga-toxin IV A, a component of the venom of the funnel webspider. The N-type channels are blocked selectively by atoxin obtained from the venom of the marine cone snail,the urconotoxin GVIA. The L-type, P /Q-type, N-type,

Chapter 14/ Transmitter Release 257

+10

01

-10

-20

~jIiw

...

-M)

...

-«J

-70

-60

I UrN tTime -Figure 14-4 The time course of ea2+ influx in the presynap-tic cell determines the onset of synaptic transmission. Anaction potential in the presynaptic cell (1) causes voltag~tedCa2+ channels in the terminal to open and a ea2+ current (2) toflow into the terminal. (Note that the Ca2+ current is turned onduring the descending phase of the presynaptic action potentialowing to delayed opening of the Ca2+ channels.) The Ca2+ in-flux triggers release of neurotransmitter. The postsynaptic re-sponse to the transmitter begins soon afterward (3) and. if suf-ficiently large. will trigger an action potential in the postsynapticcell (4). (EPSP = excitatory postsynaptic potentiaL) (Adaptedfrom Uinas 1982.)

and R-type channels all require fairly strong depolariza-tions for their activation (voltages positive to -40 to-20 mV are required), and are thus often referred to ashigh-voltage-activated Ca2+ channels. In contrast, T-typeea2+ channels are low-voltage-activated ea2+ channelsthat open in reponse to small depolarizations aroundthe threshold for generating an action potential (-60 to-40 mY). Because they are activated by small changesin membrane potential, the T-type channels help control

Part m / Elementary Interactions Between Neurons: Synaptic Transmission258

Table 14-1 Molecular Bases for Calcium Channel Diversity

Genel Ca2+ channel type

P/QN

A

B

C/D/S L

R

T

E

G/HIThe gene for the main pore-forming type of al-subunit.

excitability at the resting potential and are an importantsource of the excitatory current that drives the rhythmicpacemaker activity of certain cells, both in the brain andthe heart.

In neurons the rapid release of conventional trans-mitters associated with fast synaptic transmission ismediated by three main classes of Ca2+ channels: theP /Q-type, the N-type, and R-type channels. The L-typechannels do not contribute to fast transmitter release butare important for the slower release of neuropeptidesfrom neurons and of hormones from endocrine cells.The fact that Ca2+ influx through only certain types ofCa2+ channels, can control transmitter release is pre-sumably due to the fact that these channels are concen-trated at active zones. Localization of the N-type ea2+channels at the active zones has been visualized withfluorescently labeled crconotoxin at the frog neuromus-cular junction (Figure 14-5). By contrast, L-type chan-nels may be excluded from active zones, limiting theirparticipation to slow synaptic transmission.

.Transmitter Is Released in Quantal Units

How and where does Ca2+ influx trigger release? To an-swer that question we must first consider how transmit-ter substances are released. Even though the release ofsynaptic transmitter appears smoothly graded, it is ac-tually released in discrete packages called quanta. Eachquantum of transmitter produces a postsynaptic poten-tial of fixed size, called the quantal synaptic potential. Thetotal postsynaptic potential is made up from an integralnumber of quanta! responses (Figure 14-6). Synaptic p0-tentials seem smoothly graded in recordings only be-cause each quantal (or unit) potential is small relative tothe total potential.

lIssue Selective blockers Function

Cl)-agatoxin (spider venom)

IIHXmOtoxin (snail venom)

Dihydropyridines

Neurons

Neurons

Neurons, endocrine

Heart, skeletal muscle

Neurons

Neurons, heart

Fast release

Fast release

Slow release

(Peptides>

Fast release

Excitability

Paul Fatt and Bernard Katz obtained the first due asto the quanta! nature of synaptic transmission when theymade recordings from the nerv~muscle synapse of thefrog without presynaptic stimulation and observedsmall spontaneous postsynaptic potentia1s of about 0.5m V. Like the nerv~voked end-plate potentials, thesesmall depolarizing responses were largest at the site ofnerv~musde contact and decayed electronically withdistance (see FIgure 11-5). Similar results have since beenobtained in mammalian muscle and in central neurons.Because the synaptic potentials at vertebrate nerv~muscle synapses are called end-plate potentials, Fatt andKatz called these spontaneous potentials minillture end-plate potentillis.

The time course of the miniature end-plate poten-tials and the effects of various drugs on them are indis-tinguishable from the properties of the end-plate poten-tial evoked by nerve stimulation. Because acetylcholine(ACh) is the transmitter at the nerv~muscle synapse, theminiature end-plate potentials, like the end-plate poten-tia1s, are enhanced and prolonged by pl'O6tigmine, adrug that inhibits the hydrolysis of ACh by acetyl-cholinesterase. Likewise, the miniature end-plate poten-tials are reduced and finally abolished by agents thatblock the ACh receptor. In the absence of stimulation theminiature end-plate potentials occur at random inter-vals; their frequency can be increased by depolarizingthe presynaptic terminal. They disappear if the presyn-aptic motor nerve degenerates but reappear when a newmotor synapse is formed, indicating that these eventsrepresent small amounts of transmitter that are continu-ously released from the presynaptic nerve terminal.

What could account for the fixed size (around0.5 m V) of the miniature end-plate potential? Del Castilloand Katz first tested the possibility that each quantumrepresented a fixed response due to the opening of a sin-gle ACh receptor-channel. By applying small amounts of

Figure 14-5 Calcium channels are con-centrated at the neuromuscular junc-tion in regions of the presynaptic nerveterminal opposite clusters of acetyl-choline (ACh) receptors on the post-synaptic membrane. The fluorescent im-age shows the presynaptic Ca2+ channelsin red, after labeling with a Texasre~oupled marine snail toxin that bindsto Ca2+ channels. Postsynaptic ACh re-ceptors are labeled in green with boron-dipyromethane difluoride-labeleda-bungarotoxin, which binds selectively toACh receptors. The two images are nor-mally superimposed but have been sepa-rated for clarity. The patterns of labelingwith both probes are in almost preciseregister, indicating that the active zone ofthe presynaptic neuron is in almost per-fect alignment with the postsynapticmembrane containing the high concentra-tion of ACh receptors. (From Robitaille eta!. 1990.)

ACh to the frog muscle end-plate they were able to elicitdepolarizing responses much smaller than 0.5 m V. Fromthis it became clear that the miniature end-plate potentialmust reflect the opening of more than one ACh receptor-channel. In fact, Katz and Miledi were later able to estimatethe elementary current through a single ACh receptor-channel as being only about 0.3 tJ. V (see Chapter 6). Thisis about 1 /2000 of the amplitude of a spontaneous minia-ture end-plate potential. Thus a miniature end-plate p0-tential of 0.5 m V requires summation of the elementarycurrents of about 2000 channels. This estimate was laterconfirmed when the currents through single ACh-acti-vated channels were measured directly using patch-clamp techniques (see Box 6-2).

~Chapter 14/ Transmitter Release 259

Myelinatedaxon

Nervetem'lil18ls

Musclefibers

~

Pr~Dr. channels

PostsynllpticACh receptors

JunctioneI fold

Since the opening of a single channel requires thebinding of two ACh molecules to the receptor (one mol-ecule to each of the two a-subunits), and some of the re-leased ACh never reaches the receptor molecules (eitherbecause it diffuses out of the synaptic cleft or is lostthrough hydrolysis), about 5000 molecules are neededto produce one miniature end-plate potential. Thisnumber has been confirmed by direct chemical mea-surement of the amount of ACh released with each

quantal synaptic potential.We can now ask some important questions. Is the

normal postsynaptic potential evoked by nerve stimula-tion also composed of quantal responses that corre-spond to the quanta of spontaneously released transmit-

260 Part ill / Elementary Interactions Between Neurons: Synaptic Transmission

A

1

2

Responses f\. s~ "'-- 0u8drupIe

S3 J'o-- r r--- Unit

4 .1' Double

~ Double

---,./'-- Double

StJUIUS ~2 mV10 me

1

8

FIgure 14-8 Neurotransmitter is released in fixed incre-ments, or quanta. Each quantum of transmitter produces aunit postsynaptic potential of fixed amplitude. The amplitude ofthe postsynaptic potential evoked by nerve stimulation is equalto the unit amplitude multiplied by the number of quanta oftransmitter released.A. Intracellular recordings from a muscle fiber at the endplateshow the postsynaptic change in potential when eight consec-utive stimuli of the same size are applied to the motor nerve.To reduce transmitter output and to keep the end-plate poten-tials small, the tissue was bathed in a ea2+ -deficient (andMg2+ -rich) solution. The postsynaptic responses to the stimu-lus vary. Two presynaptic impulses elicit no postsynaptic re-sponse (failures); two produce unit potentials; and the othersproduce responses that are approximately two to four timesthe amplitude of the unit potential. Note that the spontaneousminiature end-plate potentials (5) are the same size as the unitpotential. (Adapted from Liley 1956.)

B. After many end-plate potentials were recorded, the numberof responses at each amplitude was counted and then plotted

ter? H so, what determines the number of quanta oftransmitter released by a presynaptic action potential?Does Ca2+ alter the number of ACh molecules thatmake up each quantum or does it affect the number ofquanta released by each action potential?

These questions were addressed by del Castillo andKatz in a study of synaptic signaling at the nerve-muscle synapse when the external concentration ofCa2+ is decreased. When the neuromuscular junction isbathed in a solution low in Ca2+, the evoked end-platepotential (normally 70 m V in amplitude) is reducedmarkedly, to about 0.5-2.5 m V. Moreover, the amplitude

B

J15

I::J

Z..c::

-8

i'0

Iz

in the histogram shown here. The distribution of responsesfalls into a number of peaks. The first peak, at 0 mV, representsfailures. The first peak of responses, at 0.4 mV, represents theunit potential, the smallest elicited response. This unit re-sponse is the same amplitude as the spontaneous miniatureend-plate potentials (inset). The other peaks in the histogramoccur at amplitudes that are integral multiples of the amplitudeof the unit potential. The red line shows a theoretical distribu-tion composed of the sum of several Gaussian functions fittedto the data of the histogram. In this distribution each peak isslightly spread out, reflecting the fact that the amount of trans-mitter in each quantum, and hence the amplitude of the post-synaptic response, varies randomly about the peak. The num-ber of events under each peak divided by the total number ofevents in the histogram is the probability that the presynapticterminal releases the corresponding number of quanta. Thisprobability follows a Poisson distribution (see Box 1 ~ 1 I. Thedistribution of amplitudes of the spontaneous miniature poten-tials, shown in the inset, is also fit by a Gaussian curve.(Adapted from Boyd and Martin 1956.1

of successively evoked end-plate potentials varies ran-domly from one stimulus to the next, and often no re-sponses can be detected at all (termed failures). However,the minimum response above zero-the unit synapticpotential in response to a presynaptic potential-isidentical in size (about 0.5 m V) and shape to the sponta-neous miniature end-plate potentials. All end-platepotentials larger than the quanta! synaptic potential areintegral multiples of the unit potential (Figure 1 ~).

Del Castillo and Katz could now ask: How does therise of intracellular Ca2+ that accompanies each actionpotential affect the release of transmitter? They found

Chapter 14 / Transmitter Release m

number of quanta that are released in response to apresynaptic action potential (Box 14-1). The greater theCa2+ influx into the terminal, the larger the number ofquanta released.

The findings that the amplitude of the end-plate p0-tential varies in a stepwise manner at low levels of AChrelease, that the amplitude of each step increase is an in-tegral multiple of the unit potential, and that the unit

Part m / Elementary Interactions Between Neurons: Synaptic Transmission262

Chapter 14/ TraIUlmitter Releaae 263

264 Part m / Elementary Interactions Between Neurons: Synaptic Transmission

is extremely small. A thin section through a convention- of the plasmaally fixed terminal at the neuromuscular junction of the this increase ifrog shows only 1 /4000 of the total presynaptic mem- surements asbrane. Moreover, the exocytotic opening of each small viding furthe:vesicle is of the same dimension as the thickness of the Chapter 8, thEultrathin (~100 nm) sections required for transmission tional to its 5electron microscopy. To overcome such problems, (which releaSEfreeze-fracture techniques began to be applied to the peritoneum hsynapse in the 1970s (Box 14-2). dense-oore Vel

Using these techniques, Thomas Reese and John ment of the irHeuser made three important observations. First, they sion of a singfound one or two rows of unusually large intramembra- cells is accomnous particles along the presynaptic density, on both lance, whichmargins (Figure 14-8A). Although the function of these stepwise deerparticles is not let known, they are thought to be reflect the retvoltage-gated ea + channels. Their density (about 1500 brane (Figure

per JLD12) is approximately that of the voltage-gated ea2+ tected at fast schannels essential for transmitter release. Moreover, the sion of a largeproximity of the particles to the release site is consistent 14-9C). Howe1with the short time interval between the onset of the with the fusioea2+ current and the release of transmitter. Second, they small to resolvnoted the appearance of deformations alongside therows of intramembranous particles during synaptic ac- E . I 'tivity (Figure 14-88). They interpreted these deforma- XOcytoS1S n

tions as representing invaginations of the cell membrane Exactly how fduring exocytosis. Finally, Reese and Heuser found that with the plasmthese deformations do not persist after the transmitter plays in cataly:has been released; they seem to be transient distortions Morphologicalthat occur only when vesicles are discharged. ing suggested .

To catch vesicles in the act of exocytosis, Heuser, formation of aReese, and their colleagues had to quick-freeze the tissue the vesicle an(with liquid helium at precisely defined intervals after of capacitancethe presynaptic nerve had been stimulated. The neuro- to complete fumuscular junction can thus be frozen just as the action detected in th.potential invades the terminal and exocytosis occurs. In 14-10). This fwaddition, they applied the drug 4-aminopyridine- conductance 0a compound that blocks certain voltage-gated K + chan- junction chamnels-to broaden the action potential and increase the During exocytnumber of quanta of transmitter discharged with each from around 1nerve impu1se. These techniques provided clear images creases drama!of synaptic vesicles during exocytosis. the fusion poll

The electron micrographs revealed a number of prior to complEomega-shaped (0) structures that correspond to vesicles Since tramthat have just fused with the membrane. Varying the cur within a fconcentration of 4-aminopyridine altered the amount of proteins that ntransmitter release. Moreover, there was an increase in brane are mos'the number of fi-shaped structures that was directly that bridges thcorrelated with the size of the postsynaptic response. fusion occurs.These morphological studies therefore provide inde- learned about ipendent evidence that transmitter is released by exocy- of two hemiclutosis from synaptic vesicles. and the plasn

The fusion of the synaptic vesicles with the plasma course of vesicmembrane during exocytosis increases the surface area flux would thE

264

of the plasma membrane. In certain favorable cell typesthis increase in area can be detected in electrical mea-surements as increases in membrane capacitance, pr0-viding further support for exocytosis. As we saw inChapter 8, the capacitance of the membrane is propor-tional to its surface area. In adrenal chromaffin cells(which release epinephrine) and in mast cells of the ratperitoneum (which release histamine), individual largedense-core vesicles are large enough to permit measure-ment of the increase in capacitance associated with fu-sion of a single vesicle. Release of transmitter in thesecells is accompanied by stepwise increases in capaci-tance, which in turn are followed somewhat later bystepwise decreases in capacitance, which presumablyreflect the retrieval and recycling of the excess mem-brane (Figure 14-9B). Capacitance increases can be d~tected at fast synapses after a rise in ea2+ due to the fu-sion of a large number of small synaptic vesicles (Figure14-9C). However, the increase in capacitance associatedwith the fusion of a single small synaptic vesicle is toosmall to resolve.

Exocytosis Involves the Formation of a Fusion Pore

Exactly how fusion of the synaptic vesicle membranewith the plasma membrane occurs and the role that Ca2+plays in catalyzing this reaction is under intensive study.Morphological studies from mast cells using rapid freez-ing suggested that exocytosis depends on the temporaryformation of a fusion pore that spans the membranes ofthe vesicle and plasma membrane. Subsequent studiesof capacitance increases in mast cells showed that priorto complete fusion a channel-like fusion pore could bedetected in the electrophysiological recordings (Figure14-10). This fusion pore starts out with a single-channelconductance of around 200 pS, similar to that of gap-junction channels, which also bridge two membranes.During exocytosis the pore rapidly dilates, probablyfrom around 1 nm to 50 nm, and the conductance in-creases dramatically (Figure 14-10A). In some instancesthe fusion pore flickers open and closed several timesprior to complete fusion (Fi~ 14-108).

Since transmitter release is so fast, fusion must oc-cur within a fraction of a millisecond. Therefore, theproteins that fuse synaptic vesicles to the plasma mem-brane are most likely preassembled into a fusion porethat bridges the vesicle and plasma membranes beforefusion occurs. Much like the gap-junction channels welearned about in Chapter 10, the fusion pore may consistof two hemichannels, one each in the vesicle membraneand the plasma membrane, which then join in thecourse of vesicle docking (Figure 14-1OC). Calcium in-flux would then simply cause the preexisting pore to

Cytoplasmic half of presynaptic membrane (freeze fracture)

Allure 14-8 The events of exocytosis at the presynaptic ter-minal are revealed by electron microscopy. The images onthe left are freeze-fracture electron micrographs of the cyto-plasmic half (P face) of the presynaptic membrane (compareFigure 1~7). Thin-section electron micrographs of the pre-synaptic membrane are shown on the right. (Adapted from Al-berts et al. 1989.)A. Parallel rows of intramembranous particles arrayed on eitherside of an active zone may be the voltage-gated Ca2+ channelsessential for transmitter release.B. Synaptic vesicles begin fusing with the plasma membranewithin 5 ms after the stimulus. Fusion is complete within an-other 2 ms. Each opening in the plasma membrane represents

Chapter 14/ 'Ii'ansmitter Release 265

Presynaptic membrane (thin section)

Ca

the fusion of one synaptic vesicle. In thin-section electron mi-crographs. vesicle fusion events are observed in cross sectionas G-shaped structures.C. Membrane retrieval becomes apparent as coated pits formwithin about 10 s after fusion of the vesicles with the pre-synaptic membrane. After another 10 s the coated pits begin topinch off by endocytosis to form coated vesicles. These vesi-cles include the original membrane proteins of the synapticvesicle and also contain molecules captured from the externalmedium. The vesicles are recycled at the terminals or are trans-ported to the cell body, where the membrane constituents aredegraded or recycled (see Chapter 4).

Part ill / Elementary Interactions Between Neurons: Synaptic Transmission266

A Mast cell before and after exocytosis

~

I---

5pm

Figure 14-9 Capacitance measurements allow direct studyof exocytosis and endocytosis.A. Exocytosis from mast cells. Electron micrographs of a mastcell before (top) and after (bottom) inducing exocytosis. Mastcells are secretory cells of the immune system that containlarge dense-core vesicles filled with the transmitter histamine.Exocytosis of mast cell secretory vesicles is normally triggeredby the binding of antigen complexed to an immunoglobulin(lgE). Under experimental conditions massive exocytosis canbe triggered by the inclusion of a nonhydrolvzable analog ofGTP in an intracellular recording electrode. (From Lawson et aI.,1977.1

B. Stepwise increases in capacitance rellect the successive fu-sion of individual secretory vesicles with the cell membrane.The step increases are unequal because of a variability in thediameter (and thus membrane area) of the vesicles. After exe-cytosis the membrane added through fusion is retrievedthrough endocytosis. Endocytosis of individual vesicles givesrise to the stepwise decreases in membrane capacitance. In

open and then dilate, allowing the release of transmitter.Recent advances in chemical detection suggest that

transmitter may be released through the fusion pore it-self, prior to full dilation and vesicle fusion (Figure14-1OC). An electrochemical method termed voltamdrypermits the detection of certain amine-containing trans-mitters, such as serotonin, using an extracellular

B Membrane capecitance during end after exocytosis of mast cell vesiclesMembr8ne capecit8nce

During exocyIOeia DOOng retriev8I of membr8ne

~26fF301

C Calcium-dependent exocytOlis of syneptic vesiclesori

I~v.

~~G.3

~0.2

0.1

0.0

...............I~

I I108

this way the cell maintains a constant size. (The units are infemtofarads, fF, where 1 fF - 0.1 ""m2 of membrane area.)(Adapted from Femandez et II. 1984.)C. Exocytosis and membrane retrieval from a neuronal presy-naptic terminal. Recordings were obtained from isolated synap-tic terminals of bipolar neurons in the retina of the goldfish.Transmitter release was triggered by a depolarizing voltage-clamp step (applied at arrow). which elicited a large sustainedCa2+ current (inset). The Ca2+ influx causes a transient rise inthe cytoplasmic Ca2+ concentration (bottom trace). This re-sults in the exocytosis of several thousand small synaptic vesi-cles. leading to an increase in total capacitance (top trace). Theincrements in capacitance due to fusion of a single smallsynaptic vesicle are too small to resolve. As the internal Ca2+concentration falls back to its resting level upon repolarization.the extra membrane area is rapidly retrieved and capecitancereturns to its baseline value. (Adapted from yon Gersdorff andMatthews 1994.)

carbon-fiber electrode (Figure 14-11). A large voltage isapplied to the electrode, which leads to the oxidation ofthe released transmitter. This oxidation reaction releasesfree electrons, which can be detected as a transient elec-trical current that is proportional to the amount of trans-mitter released. In response to action potentials largetransient increases in transmitter release are observed,

A

C 1 Fusion pore cIoIed

--

Figure 14-10 Transmitter is released from synaptic vesiclesthrough the opening of a fusion pore that connects a secre-tory vesicle with the presynaptic membrane.A. Patch-c!amp recording setup for recording current throughthe fusion pore. As a vesicle fuses with the plasma membrane,the capacitance of the vesicle (Cg) is initially connected to thecapacitance of the rest of the cell (Cm) through the high resis-tance (rpJ of the fusion pore. (From Monck and Fernandez1992.)B. Electrical events associated with the opening of the fusionpore. Since the membrane potential of the vesicle (lumenalside negative) is normally much more negative than the mem-brane potential of the cell, there will be a transient flow ofcharge (current) from the vesicle to the cell membrane associ-ated with fusion. This generates a transient current (I) associ-ated with the increase in membrane capacitance (C",}.The mag-nitude of the conductance of the fusion pore (gp) can be

.

corresponding to the exocytosis of the contents of asingle large dense-core vesicle. Often, these large tran-sient increases are preceded by a smaller longer-lastingsignal, corresponding to a period of release at alow rate (Figure 14-11C). Such events are thought toreflect leakage of transmitter through the fusionpore, prior to complete exocytotic fusion. A gooddeal of fast transmitter release may involve releasethrough fusion pores without the requirement for com-plete fusion.

Chapter 14 / Transmitter Release 267

B

s IiAIian ~2 FI.IIion pen open

calculated from the time constant of the transient current ac-cording to T = Grfp ... Gr/Qp. The fusion pore diameter can becalculated from the fusion pore conductance, assuming thatthe pore spans two lipid bilayers and is filled with a solutionwhose resistivity is equal to that of the cytoplasm. The fusionpore shows an initial conductance of around 200 pS, similar tothe conductance of a gap-junction channel, corresponding to apore diameter of around 2 nm. The conductance rapidly in-creases within a few milliseconds as the pore dilates to around7-8 nm (dotted line). (From Spruce et al. 1990.)C. Steps in exocytosis through a fusion pore. 1. A docked vesi-cle contains a preassembled fusion pore ready to open.2. During the initial stages of exocytosis the fusion pore rapidlyopens, allowing transmitter to leak out of the vesicle. 3. Inmost cases the fusion pore rapidly dilates as the vesicle under-goes complete fusion with the plasma membrane.

Synaptic Vesicles Are Recycled

If there were no process to compensate for the fusion ofsuccessive vesicles to the plasma membrane during con-tinued nerve activity, the membrane of a synaptic termi-nal would enlarge and the number of synaptic vesicleswould decline. This does not occur, however, becausethe vesicle membrane added to the terminal membraneis retrieved rapidly and recycled, generating newsynaptic vesicles (Figure 14-12).

r268 Part m / Elementary Interactions Between Neurons: Synaptic Transmission

A....

~

Ceroonfiber

/

c

Figure 14-11 Transmitter release through the fusion porecan be measured using electrochemical detection methods.A. Setup for recording transmitter release by voltametry. A cellis voltage-clamped with an intracellular patch electrode whilean extracellular carbon fiber is pressed against the cell surface.A large voltage applied to the tip of the electrode oxidizes cer-tain amine-containing transmitters (such as serotonin or norepi-nephrine). This oxidation reaction generates one or more freeelectrons. which results in an electrical current that can berecorded through an amplifier (Az) connected to the carbonelectrode. The current is proportional to the amount of trans-mitter release. Membrane current and capacitance arerecorded through the intracellular patch electrode amplifier

(A,).B. Recordings of transmitter release and capacitance measure-ments from mast cell secretory vesicles indicate that the fusionpore may "flicker" (open and close several times) prior to com-plete membrane fusion. During these brief openings transmit-ter can diffuse out through the pore. producing a "foot" of low-level release that precedes a large spike of transmitter release

B

S1and alone fticIc8fFlickering fusion

~~~~ ~~M~

~ J " U'U 1--Reversible openingof a fusion pore

D

~20pA1 ms

J'~500 p8

upon a full fusion event. Sometimes the reversible fusion poreopening and closing is not followed by full fusion, resulting in"stand alone flicker" in which transmitter is released only bydiffusion through the fusion pore. (From Neher 1993.)C-D. Similar patterns of release of the transmitter serotonin areobserved from Retzius neurons of the leech. The electron mi-crograph shows that these neurons package serotonin in bothlarge, dense-core vesicles and small, clear synaptic vesicles(arrow). Amperometry measurements show that Ca2+ eleva-tion triggers both large spikes of serotonin release (top trace)and smaller release events (bottom trace) (note the differencein current scales). These correspond to fusion of the largedense-core vesicles and synaptic vesicles, respectively. Thesynaptic vesicles release their contents rapidly, in less than1 ms. This rapid time course is consistent with the expectedrate of diffusion of transmitter through a fusion pore of 300 pS.Each large vesicle contains around 15,Q()()-300.000 moleculesof serotonin. Each small vesicle contains approximately 5000molecules of serotonin. (From Bruns and Jahn 1995.)

A

B

~

~~~Figure 14-12 The cycling of synaptic vesicles at nerve ter-minals involves several distinct steps.

A. Free vesicles must be tsrgetedto the active zone (1) andthen dock at the active zone (2). The docked vesicles must be-come primed so that they can undergo exocytosis (3). In re-sponse to a rise in Ca2+ the vesicles undergo fusion and re-lease their contents (4). The fused vesicle membrane is takenup into the interior of the cell by endocytosis (5). The endocy-tosed vesicles then fuse with the endosome, an internal mem-brane compartment. After processing, new synaptic vesiclesbud off the endosome, completing the recycling process.

B. Retrieval of vesicles after exocytosis is thought to occur viathree distinct mechanisms. In the first, classical pathway ex-cess membrane is retrieved by means of clathrin-coated pits.These coated pits concentrate certain intramembranous parti-

Although the number of vesicles in a nerve terminaldoes decrease transiently during release, the totalamount of membrane in vesicles, cisternae, and plasmamembrane remains constant, indicating that membraneis retrieved from the surface membrane into the internalorganelles. How the synaptic vesicles are recycled hasnot yet been resolved, but the process is known to in-volve clathrin-coating of the vesicle and the protein dy-namin (Chapter 4 and below) and is thought to be simi-lar to known mechanisms in epithelial cells (Figure14-12). According to this view, the excess membranefrom synaptic vesicles that have undergone exocytosis

Chapter 14 / Transmitter Release 269

Kiss end run Bulk endocytosis

cles into small packages. The pits are found throughout the ter-minal except at the active zones. As the plasma membrane en-larges during exocytosis. more membrane invaginations arecoated on the cytoplasmic surface. (The path of the coated pitsis shown by arrows after step 5.) This pathway may be impor-tant at normal to high rates of release. In the kiss-and-runpathway the vesicle does not completely integrate itself intothe plasma membrane. This corresponds to release throughthe fusion pore. This pathway may predominate at lower tonormal release rates. In the bulk endocytosis pathway excessmembrane reenters the terminal by budding from uncoatedpits. These uncoated cisternae are formed primarily at the ac-tive zones. This pathway may be reserved for retrieval aftervery high rates of release and may not be used during the usualfunctioning of the synapse. (Adapted from Schweizer et al.1995.)

is recycled through endocytosis into an intracellular or-ganelle called the endosome. Endocytosis and recyclingtakes about 30 seconds to one minute to be completed.

More rapid components of membrane recoveryhave been detected with capacitance measurements. Im-portantly, the rate of membrane recovery appears to de-pend on the extent of stimulation and exocytosis. Withrelatively weak stimuli that release only a few vesicles,membrane retrieval is rapid and occurs within a fewseconds (for example, see Figure 14-98). Stronger stim-uli that release more vesicles lead to a slowing of mem-brane recovery. The fastest form of vesicle cycling in-

270 Part m / Elementary Interactions Between Neurons:

volves the release of transmitter through the transientopening and closing of the fusion pore without fullmembrane fusion. The advantage of such llkiss-and-run" release is that it rapidly recycles the vesicle for sub-sequent release because it requires only closure of thefusion pore. Thus, different types of retrieval processesmay operate under different conditions (Figure 14-12).

A Variety of Proteins Are Involvedin the Vesicular Release of Transmitter

What is the nature of the molecular machinery thatdrives vesicles to cluster near synapses, to dock at activezones, to fuse with the membrane in response to Ca2+influx, and then to recycle? Proteins have been identi-fied that are thought to (1) restrain the vesicles so as toprevent their accidental mobilization, (2) target thefreed vesicles to the active zone, (3) dock the targetedvesicles at the active zone and prime them for fusion, (4)allow fusion and exocytosis, and (5) retrieve the fusedmembrane by endocytosis (Figure 14-13).

We first consider proteins involved in restraint andmobilization. The vesicles outside the active zone repre-sent a reserve pool of transmitter. They do not moveabout freely in the terminal but rather are restrained oranchored to a network of cytoskeletal filaments by thesytUlpsins, a family of four proteins (la, Ib, IIa, and lIb).Of these four, synapsins la and Ib are the best studied.These two proteins are substrates for both the cAMP-dependent protein kinase and the Ca2+jcalmodulin-dependent kinase. When synapsin I is not phosphory-lated, it is thought to immobilize synaptic vesicles bylinking them to actin filaments and other components ofthe cytoskeleton. When the nerve terminal is depolar-ized and Ca2+ enters, synapsin I is thought to becomephosphorylated by the Ca2+j calmodulin-dependentprotein kinase. Phosphorylation frees the vesicles fromthe cytoskeleta1 constraint, allowing them to move intothe active zone (Figure 14-14). ,

The targeting of synaptic vesicles to docking sitesfor release may be carried out by Rab3A and Rab3C, twomembers of a class of small proteins, related to the rasproto-oncogene superfamily, that bind GTP and hy-drolyze it to GDP and inorganic phosphate (Figure14-14B). These Rab proteins bind to synaptic vesiclesthrough a hydrophobic hydrocarbon group that is cova-lently attached to the carboxy terminus of the Rab pro-tein. Hydrolysis of the GTP bound to Rab, converting itto GDp, may be important for the efficient targeting ofsynaptic vesicles to their appropriate sites of docking.During exocytosis the Rab proteins are released fromthe synaptic vesicles into the cytoplasm.

SynapticTransmission

Following the targeting of a vesicle to its release sitea complex set of interactions occurs between proteins inthe synaptic vesicle membrane and proteins in thepresynaptic membrane. Such interactions are thought tocomplete the docking of vesicles and to prime them sothey are ready to undergo fusion in response to Ca2+ in-flux. Similar interactions are important for exocytosis inall cells, not only in the synaptic terminals of neurons.

As we have seen in Chapter 4, all secretory proteinsare synthesized on ribosomes and injected into the lu-men of the endoplasmic reticulum (ER). When theseproteins leave the ER they are targeted to the Golgi ap-paratus in vesicles formed from the membrane of theER. The vesicles then dock and fuse with the Golgimembrane, discharging their protein into the lumen ofthe Golgi, where the protein is modified. Other vesiclesshuttle the secretory protein between the cis and thetrans compartments (the different cisternae) of the Golgiapparatus until the protein becomes fully modified andmature. The mature protein is packaged in vesicles thatbud off the Golgi and migrate to the cell surface, wherethe protein is released through exocytosis. This type ofrelease is constitutive (that is the release is continuousand occurs independently of Ca2+) in contrast to regu-lated release, which occurs at synapses in response toCa2+ entry into the presynaptic terminal.

One prominent hypothesis for how membrane vesi-cles are docked and readied for exocytosis has been pro-posed by James Rothman, Richard Scheller, and Rein-hard Jahn. According to this theory, specific integralproteins in the vesicle membrane (vesicle-SNARES, orv-SNARES) bind to specific receptor proteins in the tar-get membrane (target membrane or t-SNARE) (Figure14-15). In the brain two t-SNARES have been identified:syntaxin, a nerve terminal integral membrane protein,and SNAP-25, a peripheral membrane protein of 25 kDamass. In the synaptic vesicle the integral membrane pro-tein VAMP (or synaptobrevin) has been identified as thev-SNARE.

The importance of the SNARE proteins in synaptictransmission is emphasized by the finding that all threeproteins are targets of various clostridial neurotoxins.All of these toxins act by inhibiting synaptic transmis-sion. One such toxin, tetanus toxin, a zinc endoprotease,specifically cleaves VAMP. Three other zinc endopro-teases, botulinum toxins A, B, and C, specifically cleaveSNAP-25, VAMp, and syntaxin, respectively. VAMP hasthe additional feature that it resembles a viral fusion

peptide.Reconstitution studies of purified proteins in lipid

vesicles indicate that VAMP, syntaxin, and SNAP-25may form the minimal functional unit that mediatesmembrane fusion. Moreover a detailed structural model

, MobiIiDtIon

i:!!

2 Tfllfficking

t I II, ' I II c.J+ ctl8nne1 I Neurexin I II A I I .-! Syntaxin Synt8xin : : : "J

';' J 1 , ,'a-I..atrotoICin .'

I...Tet8nu~ toxinBotuHnum toxin

SNAP-25

Figure 14-13 This diagram depicts characterized synapticvesicle proteins and some of their postulated receptors andfunctions. Separate compartments are assumed for (1) storage(where vesicles are tethered to the cytoskeleton), (2) traffickingand targeting of vesicles to active zones, (3) the docking ofvesicles at active zones and their priming for release, and (4) re-lease. Some of these proteins represent the targets for neuro-toxins that act by modifying transmitter release. VAMP (synap-tobrevin), SNAP-25, and syntaxin are the targets for tetanusand botulinum toxins, two zinc-dependent metalloproteases,and are cleaved by these enzymes. a-Latrotoxin, a spider toxinthat generates massive vesicle depletion and transmitter re-lease, binds to the neurexins. 1. Synapsins ate vesicle-associatedproteins that are thought to mediate interactions between thesynaptic vesicle and the cytoskeletal elements of the nerve ter-

has been proposed for how these proteins interact topromote membrane fusion (Figure 14-15B).

The ternary complex of VAMP, syntaxin, andSNAP-25 is extraordinarily stable. For efficient vesiclerecycling to occur this complex must be disassembledby the binding of two soluble cytoplasmic proteins:the N-ethylmaleimide-sensitive fusion (NSF) protein

Chapter 14/ Transmitter Release 271

V81ic18membr8ne

VeIicI8

3 DocIdng-priming 4FUIionpore

PI88mamembr8ne

Fusionporechlnnel?

minal. 2. The Rab proteins (see Figure 14-14B) appear to be in-volved in vesicle trafficking within the cell and also in targetingof vesicles within the nerve terminal. 3. The docking, fusion,and release of vesicles appears to involve distinct interactionsbetween vesicle proteins and proteins of the nerve terminalplasma membrane: VAMP (synaptobrevin) and synaptotagmin(p66) on the vesicle membrane, and syntaxins and neurexinson the nerve terminal membrane. Arrows indicate potential in-teractions suggested on the basis of in vitro studies. 4. Theidentity of the vesicle and plasma membrane proteins thatcomprise the fusion pore remains unclear. Synaptophysin, anintegral membrane protein in synaptic vesicles, is phosphory-lated by tyrosine kineses and may regulate release. Vesicletransporters are involved in accumulation of neurotransmitterwithin the synaptic vesicle (see Chapter 16).

and the soluble NSF attachment protein (SNAP-thisprotein is unrelated to SNAP-25; the similar names arecoincidental). The v-SNARES and t-SNARES serve asreceptors for SNAP (hence their name SNAP receptors),which then binds NSF. The NSF is an ATPase, utilizingthe energy released upon hydrolysis of ATP to unravelthe SNARE assembly.

Part ill / Elementary Interactions Between Neurons: Synaptic Transmission272

Figure 14-14 The mobilization, docking,and function of synaptic vesicles arecontrolled by Ca2+ and low-molecular-weight GTP-binding proteins.A. Synaptic vesicles in nerve terminals aresequestered in a storage compartmentwhere they are tethered to the cytoskele-ton, as well as in a releasable compart-ment where they are docked to the pre-synaptic membrane. Entry of Ca2+ into thenerve terminal leads to the opening of thefusion pore complex and neurotransmitterrelease. Calcium entry also frees vesiclesfrom the storage compartment throughphosphorylation of synapsins, thus increas-ing the availability of vesicles for docking atthe presynaptic plasma membrane.B. The Rab3A cycle targets vesicles totheir release sites. Rab3A complexed toGTP binds to synaptic vesicles. During thetargeting of synaptic vesicles to the activezone, Rab3A hydrolyzes its bound GTP toGDP. GTP hydrolysis may serve to make areversible reaction irreversible, preventingvesicles from leaving the active zone oncethey arrive. During fusion and exocytosis,Rab3A-GDP dissociates from the vesicle.There is then an exchange of GTP for GDP.This is followed by the association ofRab3A-GTP with a new synaptic vesicle,thus completing the cycle.

One additional integral membrane protein of thesynaptic vesicle, thought to be important for exocytosis,is synaptotagmin (or p6S). Synaptotagmin contains twodomains (the C2 domains) homologous to the regula-tory region of protein kinase C. The C2 domains bind tophospholipids in a calcium-dependent manner. Thisproperty suggests that synaptotagntin might insert intothe presynaptic phospholipid bilayer in response toea2+ influx, thus serving as the calcium sensor for exo-cytosis (see Figure 14-12). Synaptotagmin may alsofunction as a v-SNARE since it binds syntaxin and aSNAP isoform.

Several mutant animals that lack synaptotagminhave been created to test this protein's role in synaptictransmission. Based on these experiments two modelshave been proposed for the role of synaptotagmin. Ac-cording to one view synaptotagmin acts as a fusionclamp or negative regulator of release (preventing exo-cytosis in the absence of Ca2+). In this view, the influx ofea2+ rapidly frees this clamp, allowing synchronous re-

A Calcium control of vesicle fusion and mobilization

Q~~

B Rab3A control of vesicle fusion

lease. This hypothesis is attractive since the same ma-chinery involved in synaptic vesicle fusion (the SNAP-SNARE complex) also functions in constitutive releasethat is independent of external Ca2+. This model isbased on results from experiments with Drosophila andnematode mutants lacking synaptotagmin. which showgreatly impaired synaptic transmission in response toan action potential in the presynaptic terminal. More-over, in Drosophila the rate of spontaneous miniatureend-plate potentials is increased, suggesting that synap-totagmin has an inhibitory role.

The second hypothesis is that synaptotagmin servesas a positive regulator of release, actively promotingvesicle fusion. This view is based on the observationthat in mutant mice that lack a major isoform of synap-totagmin, fast synaptic transmission is blocked withoutan increase in spontaneous release. Since there are sev-eral isoforms of synaptotagmin in mammals, but onlyone isoform in invertebrates, it is possible that the dif-ferent mammalian isoforms have different roles: One

A,

-yCY

.

~

&~~~P &..

~

r-ATPt- ADP

& .&

~>'""d';~?/:> .'

.

8

may mediate regulated fast release and another maycontrol constitutive release.

Synaptotagmin may also play an additional role inendocytosis. Following exocytosis the fused membraneis retrieved by endocytosis. Excess membrane anywherein the terminal except at the active zone leads to the for-mation of a pit that is coated with c1athrin. The bindingof c1athrin to the membrane is enhanced by certainadaptor proteins. Synaptotagmin serves as a receptor

Chapter 14 / Transmitter Release 273

B

~

Figure 14-15 The molecular machinery for fusion and ex-

ocytosis.A. The SNARE hypothesis. Vesicle and target membranecompartments have distinct receptors-the v-SNARES (blue)and the t-SNARES (red)-that mediate docking and fusion(steps 1-4). Following fusion, two cytoplasmic proteins. NSFand SNAp, bind to the SNARE complex and disassemble it(steps 5 and 6).B. Model of the minimal fusion apparatus. At presynaptic ter-minals the v-SNARE VAMP (blue) binds to the two t-SNAREs:syntaxin (red) and SNAP-25 (green). The ternary complexconsists of a coil of four «-helices. one each from VAMP andsyntaxin and two from one molecule of SNAP-25. This coiled-coil structure is oriented parallel to the plane of the mem-brane. bringing the vesicle and target membranes in close ap-position and thus promoting fusion. The sites of cleavage bybotulinum (BoNn and tetanus toxin (TeNn are indicated.

for the dathrin adaptor protein AP-2. The clathrin coatforms a regular lattice around the pit, which finallypinches off as a small coated vesicle. The pinching off ofthe vesicle depends on a cytoplasmic GTPase calleddynamin, which forms a constricting helical ringaround the neck of the vesicle during endocytosis. ADrosophila mutant defective in dynamin is impaired insynaptic transmission owing to an inhibition of vesicle

recycling.

214 Part ill / Elementary Interactions Between Neurons: Synaptic 'Iransmission

Rgure 14-16 Changes in membrane potential ofthe presy-naptic terminal affect the intracellular concentration of Ca2+and thus the amount of transmitter released. When thepresynaptic membrane is at its normal resting potential, an ac-tion potential (top trace) produces a postsynaptic potential of agiven size (bottom). Hyperpolarizing the presynaptic terminalby 10 mV prior to an action potential decreases the steadystate Ca2+ influx, so that the same-size action potential pro-duces a smaller postsynaptic potential. In contrast. depolarizingthe presynaptic neuron by 10 mV increases the steady stateCa2+ influx, so that the same-size action potential produces apostsynaptic potential large enough to trigger an action poten-tial in the postsynaptic cell.

The Amount of Transmitter Released CanBe Modulated by Regulating the Amountof Calcium Influx During the Action Potential

The effectiveness of chemical synapses can be modi-fied for both short and long periods. This modifiabil-ity, or synaptic plasticity, is controlled by two types ofprocesses: (1) processes within the neuron that resultfrom changes in the resting potential or the firing of ac-tion potentials and (2) extrinsic processes, such as thesynaptic input from other neurons.

Long-term changes in chemical synaptic action arecrucial to development and learning, and we considerthese changes in detail later in the book. Here we shallfirst discuss the short-term changes-changes in theamount of transmitter released due to either changeswithin the presynaptic terminal or extrinsic factors.

Intrinsic Cellular Mechanisms Regulatethe Concentration of Free Calcium

As we saw at the beginning of this chapter, transmitterrelease depends strongly on the intracellular ea2+ con-

centration. Thus, mechanisms within the presynapticneuron that affect the concentration of free Ca2+ in thepresynaptic terminal also affect the amount of transmit-ter released. In some cells there is a small steady influxof Ca2+ through the presynaptic terminal membrane,even at the resting membrane potential. This ea2+ flowsthrough the L-type voltage-gated Ca2+ channels, whichinactivate little, if at all.

The steady state Ca2+ influx is enhanced by depo-larization and decreased by hyperpolarization. A slightdepolarization of the membrane can increase the steadystate influx of Ca2+ and thus enhance the amount oftransmitter released by subsequent action potentials. Aslight hyperpolarization has the opposite effect (Figure14-16). By altering the amount of Ca2+ that flows intothe terminal, small changes in the resting membrane p0-tential can make an effective synapse inoperative or aweak synapse highly effective. Such changes in mem-brane potential can also be produced by other neuronsreleasing transmitter at axo-axonic synapses that regu-late presynaptic ion channels, as described later. Theycan also be produced experimentally by injecting cur-rent.

Synaptic effectiveness can also be altered in mostnerve cells by intense activity. In these cells a high-ire-quency train of action potentials is followed by a periodduring which action potentials produce successivelylarger postsynaptic potentials. High-frequency stimula-tion of the presynaptic neuron (which in some cells cangenerate 500-1000 action potentials per second) is calledtetanic stimulation. The increase in size of the postsynap-tic potentials during tetanic stimulation is called potenti-ation; the increase that persists after tetanic stimulationis called posttetanic potentiation. This enhancement usu-ally lasts several minutes, but it can persist for an houror more (Figure 14-17).

Posttetanic potentiation is thought to result from atransient saturation of the various Ca2+ buffering sys-tems in the presynaptic terminals, primarily the smoothendoplasmic reticulum and mitochondria. This leads toa temporary excess of Ca2+, called residual Ca2+, the re-sult of the relatively large influx that accompanies thetrain of action potentials. The increase in the resting con-centration of free Ca2+ enhances synaptic transmissionfor many minutes or longer by activating certain en-zymes that are sensitive to the enhanced levels of rest-ing Ca2+, for example, the Ca2+j calmodulin-dependentprotein kinase. Activation of such ca1cium-dependentenzymatic pathways is thought to increase the mobiliza-tion of synaptic vesicles in the terminals, for examplethrough phosphorylation of the synapsins. Phosphory-lation of synapsin allows synaptic vesicles to be freedfrom their cytoskeletal restraint and to be mobilized into

and docked at release sites. As a result, each action po-tential sweeping into the terminals of the presynapticneuron will release more transmitter than before.

Here then is a simple kind of cellular memory! Thepresynaptic cell stores information about the history ofits activity in the form of residual Ca2+ in its terminals.The storage of biochemical information in the nerve cell,after a brief period of activity, leads to a strengthening ofthe presynaptic connection that persists for many min-utes. In Chapter 62 we shall see how posttetanic potenti-ation at certain synapses is followed bl an even longer-lasting process (also initiated by Ca + influx), called

long-term potentiation, which can last for many hours oreven days.

Axo-axonic Synapses on Presynaptic TerminalsRegulate Intracellular Free Calcium

Synapses are formed on axon terminals as well as thecell body and dendrites of neurons (see Chapter 12).Whereas axosomatic synaptic actions affect all branchesof the postsynaptic neuron's axon (because they affectthe probability that the neuron will fire an action poten-tial), axo-axonic actions selectively control individualbranches of the axon. One important action of axo-axonic synapses is to control ea2+ influx into thepresynaptic terminals of the postsynaptic cell, either de-pressing or enhancing transmitter release.

As we saw in Chapter 12, when one neuron hyper-polarizes the cell body (or dendrites) of another, it de-creases the likelihood that the postsynaptic cell will fire;this action is called postsynaptic inhibition. In contrast,when a neuron contacts the axon terminal of anothercell, it can reduce the amount of transmitter that will bereleased by the second cell onto a third cell; this action iscalled presynaptic inhibition (Figure 14-18A). Likewise,axo-axonic synaptic actions can increase the amount oftransmitter released by the postsynaptic cell; this actionis called presynaptic facilitation (Figure 14-18B). For rea-sons that are not well understood, presynaptic modula-tion usually occurs early in sensory pathways.

The best-analyzed mechanisms of presynaptic inhi-bition and facilitation are in the neurons of invertebratesand in the mechanoreceptor neurons (whose cell bodieslie in dorsal root ganglia) of vertebrates. Three mecha-nisms for presynaptic inhibition have been identifiedin these cells. One is mediated by activation ofmetabotropic receptors that leads to the simultaneousclosure of Ca2+ channels and opening of voltage-gatedK+ channels, which both decreases the influx of Ca2+and enhances repolarization of the cell. The secondmechanism is mediated by activation of ionotropicGABA-gated a- channels, resulting in an increased

Chapter 14 / Thmsmitter Release 275

Tetanic stimulation

~+40

0

-II

Poten- Posttetanictiation potentiationI' t r I

Figure 14-17 A high rate of stimulation of the presynapticneuron produces a gradual increase in the amplitude of thepostsynaptic potentials. This enhancement in the strength ofthe synapse represents storage of information about previousactivity, an elementary form of memory. The time scale of theexperimental record here has been compressed (each pre-synaptic and postsynaptic potential appears as a simple line in-dicating its amplitude). To establish a baseline (control), thepresynaptic neuron is stimulated at a rate of 1 per second. pro-ducing a postsynaptic potential of about 1 mY. The presynapticneuron is then stimulated for several seconds at a higher rateof 5 per second. During this tetanic stimulation the postsynap-tic potential increases in size, a phenomenon known as potenti-ation. After several seconds of stimulation the presynaptic neu-ron is retumed to the control rate of firing (1 per second).However, the postsynaptic potentials remain enhanced for min-utes, and in some cells for several hours. This persistent in-crease is called posttetanic potentiation.

conductance to CI-, which decreases (or short-circuits)the amplitude of the action potential in the presynapticterminal. As a result, less depolarization is producedand fewer Ca2+ channels are activated by the action p0-tential. The third mechanism is also mediated by activa-tion of metabotropic receptors and involves direct inhi-bition of the transmitter release machinery, independentof Ca2+ influx. This is thought to work by decreasingthe Ca2+ sensitivity of one or more steps involved in therelease process.

Presynaptic facilitation, in contrast, can be causedby an enhanced influx of Ca2+. In certain molluscanneurons serotonin acts through cAMP-dependent pro-tein phosphorylation to close K+ channels, therebybroadening the action potential and allowing the Ca2+influx to persist for a longer period (see Chapter 13). In

276 Part ill / Elementary Interactions Between Neurons: Synaptic Transmission

Presynaptic

B Presynaptic facilitation

Presynaptic

Figure 14-18 Axo-axonic synapses can inhibit or facilitatetransmitter release by the postsynaptic cell.

A. An inhibitory neuron (c,) contacts the terminal of a secondpresynaptic neuron (a). Release of transmitter by cell c, de-presses the Ca2+ current in cell a. thereby reducing the amountof transmitter released by cell a. As a result. the postsynapticpotential in cell b is depressed.

addition, the cAMP-dependent protein kinase also actsdirectly on the machinery of exocytosis to enhance re-lease in a manner that is independent of the amount ofCa2+ influx. In other cells activatioJ;l. of presynaptic ligand-gated channels, such as nicotinic ACh receptors or thekainate type of glutamate receptors, increases transmit-ter release, possibly by der:larizing the presynaptic ter-minals and enhancing Ca + influx.

Thus, regulation of the free Ca2+ concentration inthe presynaptic terminal is an important factor in a vari-ety of mechanisms that endow chemical synapses withplastic capabilities. Although we know a fair amountabout short-term changes in synaptic effectiveness-changes that last minutes and hours-we are only be-ginning to learn about changes that persist days, weeks,and longer. These long-term changes often require alter-ation in gene expression and growth of synapses in ad-

Postsynaptic

Postsynlptic

B. A facilitating neuron (C2) contacts the terminal of a secondpresynaptic neuron (a). Release of transmitter by cell C2 de-presses the K+ current in cell a. thereb~ prolonging the actionpotential in cell a and increasing the Ca + influx throughvoltage-gated Ca2+ channels. As a result, the postsynaptic ~tential in cell b is increased.

dition to alteration in Ca2+ influx and enhancement ofrelease from preexisting synapses.

An Overall View

In his book Ionic Channels of Excitable Membranes, Berti!Hille summarizes the importance of calcium in neu-ronal function:

Electricity is used to gate channels and channels are used tomake electricity. However, the nervous system is not primarilyan electrical device. Most excitable cells ultimately translatetheir electrical excitation into another form of activity. As abroad generalization, excitable cells translate their electricityinto action by Ca2+ fluxes modulated by voltage-sensitiveCa2+ channels. Calcium ions are intracellular messengers ca-

Fable of activating many cell functions. Calcium channels. . .serve as the only link to transduce depolarization into all thenonelectrical activities controlled by excitation. Without Ca2+channels our nervous system would have no outputs.

Neither Na + influx nor K+ efflux is required to re-

lease neurotransmitters at a synapse. Only ea2+, whichenters the cell through voltage-gated channels in thepresynaptic terminal, is essential. Synaptic delay-thetime between the onset of the action potential and therelease of transmitter-largely reflects the time it takesfor voltage-gated Ca2+ channels to open and for Ca2+ totrigger the discharge of transmitter from synaptic vesi-cles.

Transmitter is packaged in vesicles and each vesiclecontains approximately 5000 transmitter molecules. Re-lease of transmitter from a single vesicle results in aquanta! synaptic potential. Spontaneous miniaturesynaptic potentials result from the spontaneous fusionof single synaptic vesicles. Synaptic potentials evokedby nerve stimulation are composed of integral multiplesof the quantal potential. Increasing the extracellularea2+ does not change the size of the quantal synapticpotential. Rather, it increases the probability that a vesi-cle will discharge its transmitter. As a result, there is anincrease in the number of vesicles released and a larger

postsynaptic potential.Rapid freezing experiments have shown that the

vesicles fuse with the presynaptic plasma membrane inthe vicinity of the active zone. Freeze-fracture studieshave also revealed rows of large intramembranous par-ticles along the active zone that are thought to be Ca2+channels. These highly localized channels may be re-sponsible for the rapid increase, as much as a thousand-fold, in the Ca2+ concentration of the axon terminal dur-ing an action potential. One hypothesis about how Ca2+triggers vesicle fusion is that this ion permits the forma-tion of a fusion pore that traverses both the vesicle andthe plasma membrane. This pore allows the contents ofthe vesicle to be released into the extracellular space andmay further dilate so that the entire vesicle fuses withthe presynaptic plasma membrane.

Calcium also regulates the mobilization of thesynaptic vesicles to the active zone. These vesicles ap-pear to be bound to the cytoskeleton by synapsin, andea2+ is thought to free the vesicles by activating theCa2+/calmodulin~ependent protein kinase, whichphosphorylates the synapsins.

Several molecular candidates have been identifiedthat could account for the two other components of re-lease: targeting and docking. Targeting is thought to bemediated by the small GTP-binding Rab3A and Rab3Cproteins. Docking and fusion is thought to involve the

277ReleaseChapter 14/ Transmitter

synaptic vesicle v-SNARE VAMP (or synaptobrevin)and the plasma membrane t-SNARES, syntaxin andSNAP-25. Calcium binding to synaptotagmin may ac-tively promote vesicle fusion or remove an inhibitoryclamp that normally blocks fusion.

Finally, the amount of transmitter released from aneuron is not fixed but can be modified by both intrinsic

and extrinsic modulatory processes. High-frequencystimulation produces an increase in transmitter releasecalled posttetanic potentiation. This (intrinsic) potentia-tion, which lasts a few minutes, is caused by ea2+ left inthe terminal after the large Ca2+ influx that occurs dur-ing the train of action potentials. Tonic depolarization orhyperpolarization of the presynaptic neuron can alsomodulate release by altering steady state Ca2+ influx.The extrinsic action of neurotransmitters on receptors inthe axon terminal of another neuron can facilitate or in-hibit transmitter release by altering the steady statelevel of resting Ca2+ or the Ca2+ influx during the action

potential.In the next chapter we shall carry our discussion of

synaptic transmission further by examining the natureof the transmitter molecules that are used for chemicaltransmission.

Eric R. KandelSteven A. Siegelbaum

Selected Readings

Dunlap K, Luebke JI, Turner 'IJ. 1995. Exocytotic eaH chan-nels in mammalian central neurons. Trends Neurosci18:89-98.

Hanson PI, Heuser JE, Jahn R. 1997. Neurotransmitterrelease-four years of SNARE complexes. Curr OpinNeurobiol 7(3):31~315.

JesselllM, Kandel ER. 1993. Synaptic transmission: a bidi-rectional and self-modifiable form of cell-ce1l communica-tion. Cell 72:1-30.

Katz B. 1969. The Release of Neural Transmitter Substances.Springfield, IL: Thomas.

Lindau M, Almers W. 1995. Structure and function of fusionpores in exocytosis and ectoplasmic membrane fusion.Curr Opin Cell Bioi 7:509-517.

Matthews G. 1996. Synaptic exocytosis and endocytosis:capacitance measurements. Curr Opin Neurobiol 6(3):358-364.

Schweizer FE, Betz H, Augustine GJ. 1995. From vesicle

278 Put ill / Elementary Interactions Between Neurons: Synaptic Transmission

docking to endocytosis: intermediate reactions of exo- Hille B. 1992. jcytosis. Neuron 14(4):689-696. Sunderland

Smith SJ,Augustine GJ. 1988. Calcium ions, active zones and HimingLD,F(synaptic transmitter release. 1rends Neurosd 11:458-464. Miller RJ, T

Siidhof TC. 1995. The synaptic vesicle cycle: a cascade of pro- channels inrein-protein interactions. Nature 375:645-653. pathetic I\e1

Jones SW. 199References nets. J BioJ\lAlbertsB, BrayD, LewisJ,RaffM, Roberts I<, WatsonJD.I994. KandelER.l9'.

Molecular Biology of the Cell, 3rd ed. New York: Garland. to Be1rtJvioraAlmers W, Tse FW. 1990. Thansmitter release from synapses: Kandel ER. 1

Does a preassembled fusion pore initiate exocytosis? strength byNeuron 4:813-818. Katz B, Miled:

Bahler M, Greengard P. 1987. Synapsin I bundles F-actin in a in the amphosphorylation~ependent manner. Nature 326:704-707. 192:407-431

Balcer PF, Hodgkin AL, Ridgway EB. 1971. Depolarization Katz B, Milediand calcium entry in squid giant axons. J Physiol (Lond) neuromWK218:709-755. 189:535-54

Boyd lA, Martin AR. 1956. The end-plate potential in mam- Kelly RB. 19<-malian muscle. J Physiol (Lond) 132:74-91. Cell 72:43-

Breckenridge LJ, Almers W. 1987. Currents through the Klein M, Sha}:fusion pore that forms during exocytosis of a secretory the modulivesicle. Nature 328:814-817. Kretz R, ShaF

Bruns D, Jahn R. 1995. Real-time measurement of transmitter potentiaticrelease from single synaptic vesicles. Nature 377:62~. of the free

Couteaux R, Pecot-Dechavassine M. 1970. V&icu1es synap- Brain Res!tiques et poches au niveau des "'zones actives'" de la ;One- Kuffler SW, rtion neuromusculaire. C R Hebd ~ances Acad Sd ~ D Brain: A QSd Nat 271:2346-2349. tem,2nd ec

Del Castillo J, Katz B. 1954. The effect of magnesium on the Lawson D, Iiactivity of motor nerve endings. J Physiol (Lond) 1977. Mol.124:553-559. of exocytt

Erulkar SO, Rahamimoff R. 1978. The role of calcium ions in 72:242-59.tetanic and post-tetanic increase of miniature end-plate Uley AW. 19!

potential frequency. J Physiol (Lond) 278:501-511. end-plateFaber OS, Kom H. 1988. Unitary conductance changes at Llin4s RR. 1~

teleost Mauthner cell glydnergic synapses: a voltage- 247(4):56-clamp and pharmacologic analysis. J Neurophysiol Llin4s RR, H60:1982-1999. systems if

Fatt P, Katz B. 1952. Spontaneous subthreshold activity at Llin4s R, Stei

motor nerve endings. J Physiol (Lond) 117:109-128. presynaptFawcett DW. 1981. The Cell, 2nd ed. Philadelphia: Saunders. squid gUnFernandez }M, Neher E, Gomperts BD. 1984. Capacitance Martin AR.

measurements reveal stepwise fusion events in degranu- mechaniSJlating mast cells. Nature 312:453-455. Critiall, C

Geppert M, Sudhof TC. 1998. RAB3 and synaptotagmin: the edge and Cyin and yang of synaptic membrane fusion. Annu Rev lar Biolo~Neuroed 21:75-95. American

Heuser JE, Reese 1'5.1977. Structure of the synapse. In: ER Monck JR, FKandel (ed). Handbook of Physiology: A CritiCJII, Comprehen- J Cell BioIsive Presentation of Physiologiad Knowledge and Conapts, Neher E. 199Sect. 1, The Nervous System. VoL 1, Cellular Biology of NeIl- Nature 3trom, Part 1, pp. 261-294. Bethesda, MD: American Physi- Nicoll RA. 1ological Society. "yes" or '

Heuser JE, Reese 1'5. 1981. Structural changes in transmitter Peters A, Palrelease at the frog neuromuscular junction. J Cell Bioi the Nervo

88:564-580. PhiladelF

278

Hille 8. 1992. Ionk Chlmnels of Excitllble Membranes, 2nd ed.Sunderland, MA: Sinauer.

Himing LO, Fox AP, McCleskey EW, Olivera 8M, Thayer SA,Miller RJ, Tsien RW. 1988. Dominant role of N-type ea2+. .-channels in evoked release of norepinephrine from sym-pathetic neurons. Science 239(4835):57-61.

Jones SW. 1998. Overview of voltage-dependent Ca chan-nels. J 8ionerg 8iomembr 30(4):299-312-

Kandel HR. 1976. The Cellulllr Basis of Behtzuior: An Introduction

to BeJuwioral Neurobiology. San Francisco: Freeman.Kandel ER. 1981. Calcium and the control of synaptic

strength by learning. Nature 293:697-700.Katz 8, MUedi R. 1967a. The study of synaptic transmission

in the absence of nerve impulses. J Physiol (Lond)192:407-436.

Katz 8, MiIedi R. 1967b. The timing of calcium action duringneuromuscular transmission. J Physiol (Lond)189:535-544.

Kelly RB. 1993. Storage and release of neurotransmitters.Cell 72:43-53.

Klein M, Shapiro E, Kandel HR. 1980. Synaptic plasticity andthe modulation of the eaH current. J Exp 8iol89:1l7-157.

Kretz R, Shapiro E, Connor J, Kandel ER. 1984. Post-tetanicpotentiation, presynaptic inhibition, and the modulationof the free eaH level in the presynaptic terminals. Exp8rain Res Suppl 9'.2~283.

Kuffler SW, Nicholls JG, Martin AR. 1984. From Neuron toBrain: A Cellular Approach to the Function of the Nervous Sys-tem,2nd ed. Sunderland, MA: Sinauer.

Lawson D, Raft MC, Gomperts 8, Fewtrell C, GiIuJa NB.1977. Molecular events during membrane fusion. A studyof exocytosis in rat peritoneal mast cells. Cell 8iol72:242-59.

Uley AW. 1956. The quanta! components of the mammalianend-plate potential. J Physiol (Land) 133:571-587.

Uin4s RR. 1982. Calcium in synaptic transmission. Sci Am247(4):56-65.

Uin4s RR, Heuser JE. 1977. Depolarization-release couplingsystems in neurons. Neurosci Res Progr 8ullI5:555-687.

Uin4s R, Steinberg IZ. Walton I<. 1981. Relationship betweenpresynaptic calcium current and postsynaptic potential insquid giant synapse. 8iophys J 33:323-351.

Martin AR. 1977. Junctional transmission. II. Presynapticmechanisms. In: HR Kandel (ed). Handbook of Physiology: ACritical, Comprehensive Presentation of Physiologiall Knowl-edge and Concepts, Sect. 1, The Nervous System. Vol. 1, Cellu-IIlr Biology of Neurons, Part 1, pp. 329-355. Bethesda, MD:American Physiological Society.

Monck JR, Fernandez JM. 1992. The exocytotic fusion pore.J Cell 8ioI1l9:1395-1404.

Neher E. 1993. Cell physiology. Secretion without full fusion.Nature 363:497-498.

Nicoll RA. 1982. Neurotransmitters can say more than just"yes" or "no." 1iends Neurosci 5:369-374.

Peters A, Palay SL, Webster H deF. 1991. The Fine Structure ofthe Nervous System: Neurons and Supporting Cells, 3rd ed.Philadelphia: Saunders.

Redman S. 1990. Quantal analysis of synaptic potentials inneurons of the central nervous system. Physiol Rev70:165-198.

Robitaille R, Adler EM, Charlton MP. 1990. Strategic locationof calcium channels at transmitter release sites of frogneuromuscular synapses. Neuron 5:773-779.

Scheller RH. 1995. Membrane trafficking in the presynapticnerve terminal. Neuron 14:893-897.

Smith SJ, Augustine GJ, Charlton MP. 1985. Transmission atvoltage-clamped giant synapse of the squid: evidence forcooperativity of presynaptic calcium action. Proc NatlAcad Sci USA 82:622~25.

SOllner T, Whiteheart SW, Brunner M, Erdjument-BromageH, Geromanos S, Tempest P, Rothman IE. 1993. SNAP re-ceptors implicated in vesicle targeting and fusion. Nature362:318-324.

Spruce AE, Breckenridge LJ, Lee AK, Almers W. 1990. Prop-erties of the fusion pore that forms during exocytosis of amast cell secretory vesicle. Neuron 4:643-654.

Siidhof TC, Czernik AJ, Kao H-T, Takei K, Johnston PA,Horiuchi A, Kanazir SO, Wagner MA, Perin MS, De

Chapter 14 / Transmitter Release 279

Camilli P, Greengard P. 1989. Synapsins: mosaics ofshared and individual domains in a family of synapticvesicle phosphoproteins. Science 245:1474-1480.

Sutton RB, Fasshauer 0, Jahn R, Brunger AT. 1998. Crystalstructure of a SNARE complex involved in synaptic exo-cytosis at 2.4 A resolution. Nature 395(6700):347-353.

von Gersdorff H, Matthews G. 1994. Dynamics of synapticvesicle fusion and membrane retrieval in synaptic termi-nals. Nature 367:735-739.

Weber T, Zemelman BY, McNew JA, Westermann B, GmachlM, Parlati F, Sollner TH, Rothman JE. 1998. SNAREpins:minimal machinery for membrane fusion. Cell 92(6):759-772.

Wernig A. 1972. Changes in statistical parameters duringfacilitation at the crayfish neuromuscular junction. J Phys-iol (Lond) 226:751-759.

Zucker RS. 1973. Changes in the statistics of transmitter re-lease during facilitation. J Physiol (Lond) 229:787-810.