Quantal analysis and long-term potentiation

Quantal analysis and long-term potentiation

Q Acodemie des sciences / Elsevier, Paris Quanta1 analysis and long-term An&se pntz’que Henri Korna*, Donald etpotentiation potentiation syn...

592KB Sizes 22 Downloads 76 Views

Q Acodemie

des

sciences

/ Elsevier,

Paris

Quanta1 analysis and long-term An&se pntz’que Henri

Korna*,

Donald

etpotentiation

potentiation

synaptique h long terme

S. Faberb

J Bioiogie ceilulaire et moleculaire du neurone, lnserm U261, lnstitut Pasteur, 25, rue du D’-ROUX, 75724 Paris cedex 15, France ” Department of Anatomy and Neurobiology, Allegheny University, 3200 Henry Avenue, Philadelphia,

PA 19129, USA

Abstract - Quanta1 analysis is useful for assessing the pre- and/or post-synaptic locus of the expression of long-term tetanic potentiation with the condition, however, that the studied synaptic potentials have been evoked by single cell stimulations, as is the case with paired recordings of identified neurons. The application of this methodology, primarily with indirect criteria, has produced conclusions which dance back and forth across the synaptic cleft. (0 Academic des sciences / Elsevier, Paris.] synapses

i neurotransmitters

/ receptors

/ quanta

/ synaptic

efficacy

An attractive example of activity-dependent synaptic plasticity in the vertebrate brain is long-term potentiation in the CA1 region oi hippocampal slices. However, whether synapses are stronger during this process because more transmitter is released (as is the case in Aplysia ]I] during sensitization), or because postsynaptic responsivity is greater, or both (figure I), has been a matter of fierce controversy over the past 20 years, even though tools of quanta1 analysis have been available for solving the problem.

1. Generalities

on quanta1 analysis

These tools were developed along with the quanta1 theory of neurotransmission, as formulated by Katz 121. Briefly, at the neuromuscular junction: i) evoked endplate potentials correspond to integral multiples of a basic ‘quantal’ unit, y which, ii) also occurs spontaneously in the absence of nerve impulses. Underlying this relationship is the fact that a particular synaptic connection involves one or several ‘active’ sites of secretion, each of which releases at most one quantum at a time 131 with a probability p. The amplitude distribution of responses elicited by presynaptic action potentials can therefore match statistical predictions such as the simple or compound binomial distributions 13-51, which allow the release para-meters p and n (which equates to the number of active sites) to be calculated, as well as g, the quantum size. The latter is determined by the number and properties of postsynaptic channels opened by a quantum and by the amount of transmitter packaged in a vesicle. Various indirect approaches can be applied to studies of

* Correspondence C. R. Acad. 1998. 321,

and

reprints

Sci. Paris, 125-130

Sciences

de

la vie

/ Life

Sciences

/ learning,

memory

synaptic plasticity such as changes in: i) the frequency of miniature events, their coefficient of variation (CV), and the number of failures of synaptic transmission; or ii) the quanta1 size, with the goal of identifying pre- or postsynaptic loci. Such is the mathematical theory, but its applicability in some brain structures has been complicated by numerous factors [5-i’] leading to ambiguities in identifying the quanta and their true CV, and to difficulties in resolving multi-quanta1 peaks particularly when the CV is large or the amplitude distribution of quanta is not gaussian but is skewed owing to cable properties of the recorded cell, for example. Also, it has been suggested that receptors associated with a release site are saturated (i.e. all occupied) by the molecules contained in a single synaptic vesicle 181 so that they may not be sensitive to variations in this parameter. These difficulties converge at excitatory junctions on CA1 pyramidal cells where, in
2. Preparations

for a square dance

In early studies the interpretation of experimental results was already oscillating in favor of one side of the synapse or the other. Microdialysis data iindicated that

H. Korn,

D. S. Faber

Is the synaptic current, i, increased during LTP due to :. ;‘I

potentintecl

1 ‘, / PRE releasing side

? _r;a_ 1) presynautic factors : more transmitter released - from a single site - or an increase in p

transmitter

POST

channels

receptor’s side Figure

1. Schematic

When bound in this released

.I nerve impulse activates the terminal bouton, synaptic vesicles release the transmitter molecules to the receptors, postsynaptic ionic channels are opened, thus allowing current (i) to flow across review is whether the increase in size of the postsynaptic potential observed during LTP (right) (presynaptic locus), to an enhanced postynaptic sensitivity (more channels opened, larger single

postsynaptic

cable

diagram

properties)

there is an increased lo], but reports also efficacy only occurs however, challenged are not spared during tion that paired pulse presynaptic Obviouly

help

The first hippocampal involvement

of a synapse

(left)

with

a microelectrode

2) postsynaDtic factors : more channels opened - or a difference in postsynaptic cable properties, etc. in the postsynaptic

release of glutamate during LTP 19, suggested that the enhanced synaptic at AMPA receptors [I 1 I. This was, by evidence this process facilitation,

phenomenon, was needed

that [121, usually

NMDA receptors and the observaassumed to be a

is not altered by LTP from other techniques.

[13].

analysis presynaptic mentioned

to

above in using this method with central synapses, there has been a subsequent resurgence of interest following two major and ingenious studies II 5, 161. In both, the authors recorded fluctuations in neurotransmission from trial to trial, and observed that the percentage of iailures of presynaptic action potential to evoke a postsynaptic response (pyramidal) was reduced by LTP, which, in addition, strongly modified the CV of the resulting amplitude distributions. These results pointed towards a presynaptic site of expression. Unfortunately theoretical calculations

and most ‘minimal’ activation intracellular

simulations modifications

[I71 called of the CV.

for caution Furthermore,

when this

of the subsequent studies were carried out with afferent stimulations, which do not guarantee of a single input axon alone, as would paired recordings. In fact, until recently, only a few

series of experiments were performed quanta1 transmission and LTP between postsynaptic hippocampal neurons.

126

they contain, and once the latter are the membrane. The question addressed is due to a larger amount of transmitter channel conductance or a difference in

or to both.

elaborate attempt to apply quanta1 LTP suggested a primarily [I 41 and, despite the difficulties

and computer interpreting

neuron.

118, 191 to study individual pre- and In one of them 1181

LTP was accompanied by a reduction oi failures, giving some credence to the presynaptic hypothesis. In the other an increase in quanta1 size was reported [I 91. The quanta1 to blur

introduction parameters the issue

of algorithms designed from non-peaked histograms by promoting a ‘compromise

Surprisingly, this still remains tion even after years of research ically, both quanta1 amplitude

to

extract seemed solution’.

the most reasonable solu(see Conclusion). Specif(q) and content (i.e. the np

product) appeared to become larger albeit to varying degrees, possibly value of p and to the experimental

during LTP [20-221 related to the initial pamdigm 1221. The

conclusion that the expression of LTP is the result of both preand postsynaptic changes was also reached with more conventional statistical approaches when clear peaks, indicative of multivesicular release, appeared in histograms of evoked responses 1231, although only quantal size increased when before and after potentiation

it could 1241.

be

directly

assessed

In view of all these contradictory conciusions, quanta1 principles rather than quanta1 analysis became the guidelines for further investigations, and several papers in particular stand out.

3. ‘Change

partners’

A method for detecting heterogeneity in transmitter release was designed 125, 261. It is based on the fact that the NMDA-channel blocker MK-801 inhibits the postsyC. R. Acad.

Sci.

Paris,

Sciences

de

la vie t Life Sciences 1998.321,125-130

Synaptic naptic responses only when the channel is open, the rate of this block with repeated trials depending upon p. Hence, any increase in p should accelerate the time course of this block, which was, however, unaltered during LTP [26]. The authors concluded in favor of a postsynaptic mechanism, although caution is necessary when using this method in slices. Also, as noted elsewhere [27], this procedure would have been insensitive to the addition of new synapses with the same p profile as preexisting ones. At the

same

time,

another

team

1281 chose

order to study LTP in single fiber inputs to CA1 cells. LTP was induced by tetanus, and resulted in a reduction in the proportion of failures, with no increase in the amplitude of the remaining evoked responses. This is most easily explained by an increase in release probability. But as

existing

ones,

had

not yet

4. Was the caller sleeping?

been

in similar conin g after LTP receptors are block of pre-

considered.

(‘maitre

de ballet’)

(although for multi-fiber inputs) would suggest yes. After LTP induction by pairing postsynaptic depolarization with low frequency presynaptic stimulations the CV of the AMPA component fell, but the mean and CV of the NMDA component remained unchanged. One explanation is that LTP induction uncovers clusters of previously latent AMPA receptors. This interesting concept was soon reinforced by the results of two groups [30, 311 who found, with minimal stimulation, that the failure rate was greater when recordings were obtained at resting potential, which favors expression of the AMPA component in isolation, than at a depolarized level that unmasks the NMDA component. Furthermore, LTP was associated with a specific decrease in the failure rate of the AMPA response. The basic assumptions underlying the silent synapse scenario, which have been demonstrated directly with paired recordings, but in another context, that of inhibitory synapses [32], were apparently contradicted by evidence [33] that in the hippocampus both shortand longterm potentiation are characterized by parallel enhanceof the

C. R. Acad. 1998. 321.

components

Sci. Paris, 125-130

Sciences

mediated de la vie

by the / Life

conclusions, transmitter

towards release.

Indeed, this ‘dual’ potentiation is in line with that observed in a careful study of LTP induced under different conditions of neurotransmitter release, achieved using adenosine agonists/antagonists or by changing extracellular Ca2’/Mg2’ [I 21. However, the proposed unmasking of silent synapses is not inconsistent accept the notion that functional fiable.

5. A presynaptic

with these results, ones are also

Experiments

two

Sciences

receptor

winner

if we modi-

conducted

with

in sight?

cultured

hippocampal

neurons, admittedly different from slices, also marily for a presynaptic involvement. Glutamate tions produce in these cells a potentiation frequency of miniature currents but leaves their unchanged function

by

[34]. Similarly, measuring the

direct tests rate at which

of

plead priapplicaof the amplitude

presynaptic terminals are

labeled upon bath application of antibody to synaptotagmin, a vesicle membrane protein, have shown an increase in synaptic vesicle cycling during LTP triggered by glutamate

Indeed, is it possible that clusters of previously silent AMPA receptors become functional during LTP [23], with no change in transmitter release? Results of an investigation [29] of the difference in amplitude fluctuations of the AMPA and NMDA components of the EPSC separately

ments

reversed once more the implying an increased

of learning

a different

strategy and observed the opposite effects. The authors used ‘minimal stimulation’, with the stimulus strength adjusted to give roughly 50% failures of transmission in

noted again [26] this does not explain why ditions, other groups had found an increase [21, 291. Also, ‘a scenario in which AMPA added’ and compensate for a postsynaptic

types. This a mechanism

locus

plus

low

Mg2’

1351.

Talmudic arguments directed against the presynaptic camp have been recently weakened by several groups. The most compelling answers against them have been obtained synapses

with paired recordings of neurons connected by bearing a single active zone 1361, a fortunate

design released reached

suggesting that no more than one quantum is at a time [3, 41. The same conclusion was for inhibitory synapses, which control the excita-

bility bear

of the teleost Mauthner only one active zone

also undergo an LTP, which pendent methods of quanta1 Stimulating

a presynaptic

per

cell [37J. These bouton [3] and

terminals they can

was analyzed with two analysis (figure 2). neuron

evoked

inde-

responses

with a relatively high failure rate, ranging from 0.5 to 0.8 among different cells. As expected, the distribution of the non-failure EPSCs was composed of only a single quanta1 peak. In agreement with results obtained at rnore complex junctions [28], there was a decrease in the fraction offailures after induction of LTP, with no change in the size or shape of the quanta1 response peak. This also argues that no new c:lusters of AMPA receptors were inserted at previously active synapses up to 30 min after induction of LTP. Careful additional controls were made. First, during paired pulse facilitation, the fraction of failures in response to the second stimulus was lower than that following the first one. Second, the fraction of failures was decreased to a low level when the preparation was exposed to high external Ca2+ concentrations.

127

H. Korn. D. S. Faber

A

1’ before

12’ after

post

ve

I %4kVIII

z N

comm.

4ms

B

20

2’ before

t

i = 0.90mV n = 14

6B ia $ : 2 B .s

5c E

4

0 0

3-

1

2

Y

mV

2-

I

1

I

(

0123456 (0.6%

n (potentiation

factor)

of VcoIlJ

10

0 0

i

1

Amplitude Figure

2. Evidence

for

a presynaptic

locus

of potentiation

at M-cell

Experimental protocol used for paired pre- and postsynaptic before) and potentiatcd responses (12 min after a tetanus) 9-C. Evrdence for a presynaptic involvement obtained with distributions of fluctuating unitary responses with optimal

A.

number enhanced

of active zones by more than

(n) and the quanta1 50% thus accounting

6. Conclusion:

inhibitory

mV

IPSPs

synapses.

recordings of connected interneurons and their target M-cell (left). Control (1 min are shown in the middle and right hand panels (presynaptic spikes not shown). the method of the coefficient of variation for four cells (B) and b,y fitting amplitude binomial functions fC, experiment 2 of 6). Note that 5 min after the tetanus the

size iq) are unchanged, for the potentiation

Ifrom

let us dance together

Thus, although the latest burst of evidence in this ongoing saga has returned the focus to presynaptic mechanisms during the expression and maintenance of LTP, a number of questions, in addition to that of the correct site, remain unsolved. One of them is why researchers navigated for so long from one side synapse to the other, with at every ‘volte-face’

of unitary

many of the solid

whereas Xi].

the

enhancement

of the

probability

of transmitter

release

(p) is

synapses [38], may also contribute to enhance the size of quanta1 events in hippocampal cells during LTP, provided there is an increase in glutamate release, which raises its concentration

in the

Several may rized

mechanisms

coexist both elsewhere

synaptic that

cleft regulate

[39]. presynaptic

pre- and postsynapticaily. 1401 the brief periods

activity in the input mate, which causes receptors channels,

of

strength As summasynchronous

fibers that trigger LTP release glutathe opening of postsynaptic NMDA producing a transient rise in intracel-

reasons for changing their minds? It could very well be that the protocol used to induce LTP, the length of time necessary for its induction (i.e. pairing versus tetanic stimulations), the age of the preparation (neonate, juvenile, adult), and the nature of the synapses under study all result in different forms of synaptic plasticity. In fact, yet another factor has been evoked recently. It has been suggested that spill-over of transmitter from one synapse to

lular calcium concentrations and activation of postsynaptic protein kinases. This rather brief phase of induction is followed by the longer lasting period of maintenance, which has been the focus of the debate mentioned above. This debate has not been easily resolved since obviously postsynaptic elements play a crucial role in triggering LTP. Thus the logical need for a transynaptic lilnk to the puta-

the

tively

128

next,

first

discovered

at lower

vertebrate

inhibitory

more C

effective R Acad.

presynaptic Sci. Paris,

Sciences

fibers de

has

become

la vie / Life Sciences 1’398.

321.

125-130

Synaptic apparent, raising gers 141-431.

the

issue

of so called

retrograde

the concept sides involved

messen-

tributions according

Thus, communication between the two sides of the synapse is a reality during induction and may also have a role in the expression of LTP. For example analyses of strontium-induced asynchronous release of quanta [44] have suggested that LTP is associated with an increase in quantal size as well as an enhanced frequency of quanta1 events. Also, a statistically rigorous quanta1 led to the conclusion that this process is both by an increase in quanta1 content, i.e. release, and by postsynaptic modifications

Consequently, pinpoint the on this topic

analysis 1451, accompanied of transmitter that enhance

it may not be locus for the has reinforced

7. References 11 1 Castelluci depression Pm. Nat/

[+l Kern a contra physica

of the synaptic wflcx in A,&+

lhi Rednun in thr, central

S.J. 1990. nervous

at a

system.

A.V., Jack J.B. Hid 5, 324-334.

O/w. 1995.

anallvsis Rlol.

vi synaptic 4, ihO-lOi.

Svnaptic

$astlc

transmlssron ity:

Contrihutlons rxprrwon

/ 121 A\rtrly F, Wlgstrom tion of Nh4DA receptor cdmpal CA1 region. fur

H., Cust~tsson. R. 1992. channels in the wprewon / Nrurot-I. 4. hB~h90.

I1 11 MrN,cughton B.L. potentiation in rat fascia >iO/. 124, .!49-‘!hL.

1982. dentata

,Ytxwrx

L.L. lYR.3. IC’;TC-C’ 10. 1 Oil-l

Long term ac r through

Long-term 069.

I”

hippocampd

the LTP.

potentiatlon glutamate

Presynaptic potmtlatlon

C. R. Acad. 1998 321

de

la vie

and short term mechanisms. /. Phyin

the

hippocampus.

enhancement in hipporampal

/ Life

and 242,

The relatwe contribuof LTP in the hippo-

Pwsvnaptic mechanism N,~tuw 346, 724-729.

116 MJinow R., Tslen R. 1990. wholr~cell recording of long-term N‘?fUW i4i,, 177-l 80.

Sciences

of quisqualate 01 LTP. Sr-rcnn,

enhancrmwt different

potrntiation

I1 51 Bckkc,rs J.M., Stevens C.F. 1990. term potentiation in the hippocampus.

SCI. Paris, 125-130

by the Odyssey

Sciences

for

H..

Fassnatht

Nature

synaptic responses varying not yet been elucidated.

of hippocampal

C’., Faber 6.316.

350,

long-

shown by slices.

LTP.

D S. 1991.

Transmission oscillations,

Is maintenance

121 l Kullmann D.M., Nicoll ted with increases In quantal 240-244.

of LTP pre-

between pairs of hipporxmpal and LTP. Scienre 252, 722-724.

B.L. 19Yl. to increased

Long-term quanta1

enhancement

srze,

1201 Voronin L.L.. Kuhnt U.. Cuscv 4.C. lY91. Changes meters oi CA1 synaptrc transmrwon are dependent or long-term potentialion. R;euroscf Rr. Com 9, l-7. R.A. 1992. ILong-term r-ontwt ,rnd quantal

not

quanta1

slice oiCA1 con-

in quantal parathe magnrtcrde ot

potenti,ttion amplitude.

is associ+

N,~tur~

357,

1221 Larkman A., Hannay T., Stratiord K., Jack J. 1991. Prrsynaptic release probabrlrty influences the locus of long-term potentiation. Naluw 36U, 70-73. 1231 Liao I)., Jones changes underlyIng 9. 1089-l 097

R-M., Errington M.L.. Lynch M.A. 19Rh. Corrr~ potentlatlon and release of endogcnous armno ot anaesthetized rats. i. Pys~ol. 377, 391-4013.

I1 1 / Muller D.. Jolv ,M.. I ynch G. 1988. hhlDA rel.eptors to the induction .Ind 1 hO4&1697.

1141 Voronin

in

in neurons

I91 Dolphin A.C., Errrngton M.L., Bliss T.V. 1081. Long-term 01 the lpetiorant path in viva is awori,ated with lncrcased reieaw. Nlture 297, 496-498. 11 01 Bliss F.V.P.. Douglas Iatlon between long-term a< idi from dentate gyrus

efficacy

an.llys~s ot synaptic potentials Phi%/ I&L.. 70, 165-198.

Qudntal

171 Bckkers J.M. lYY4. Qudntaf central nervous
plified

[19i Foiter TC., McNaughton synaptic transmlwon is due tent. H~pprKlmpus 1, 79-91.

141 Korn H., Triller A.. Mallet A., Faber D.S. 19Rl. Flue-tuating responses at a crntr.~l synapse: II of binomial tit prcdrcts number of Etained presynaptic I,nutons Scicwx* 213, R9%9Ol. D.S. 1991, Qu,rntal analysrs and synaptic system. TwncIs Rjrurosci. 14, 4 39-445.

both con-

approaches which have been so powerful in the past for advancing our understanding of synaptic function and the quanta1 processes of transmitter release (51. Without these aids statistical methods may meander endlessly, as exem-

1181 Malinow R. 1991. neurons: quanta1 levels,

H., Mallet A., Triller A.. Faber D.S. 1982. Transmission inhibitor\! synapse. II. Quanta1 description of releaw with correlate for the binomial II. /. Neurophyrd 48, 679-70;.

[17l, with their relative

required to solve the discrepancies listed above, electrophysiological and imaging techniques [35, 481 need to be complemented by a structural analysis of the synapses and immunocytochemical investigations of their associated receptors and by their mathematical models,

I1 71 Kern synaptic?

V., K,rndel E.R. 1974. A quanta1 analysis underlying habituation ofthejilll~wlthdrawal /k‘3d 5 i. urn 71, 5D04~5008.

I.5 Korn H., Faber the central nervnw

to potentiated to rules that have

synapse and

learning

of

Finally, LTP has a late, protein-synthesis-dependent phase, starting 2 h after its induction [40], which may be associated with the formation of new synapses as suggested by morphological data 1471. Among the studies

the synaptic current, as well as by the formation of ‘new sites’. At this stage it is also likely that postsynaptic properties can feed back on presynaptic terminals and influence synaptic strength [46]. surprising that rather than expression of LTP, the research

of the unity of the in this phenomenon,

locus

A.. Mdlinow long-term

R. 1992. potrntiation

Direct meawrement in CA1 hippoc.3mpus.

[Z-II Manabr T.. Rennrr P., hicoll long-term potentiatron rwt~aled rents. Natuw 355, io-55

R.A. 1992. PostsynaptIc hv the analysis of mini,lture

I251 Rosenmund C , Clement< probability of glutamate releaw 754.-757.

I.D., Westhrook at hippocampal

I261 (Manabe T., Nlcoll R.A. lY94. LonE-term against an ,ncrease in transmrttrr release prohahility the hippocampus. _\oenc-ra 2h5, 1888~1892. 1271 Malinow 11 Y 5-1196.

R. 1994.

LTP: Desperatflly

1281 Stevens C.F., Wang Y. 1 Y94. tion ai a mechanism for plasticity.

Changes

Naiurr

1291 Kullmnnn D.M. lY94. Amplitude EPSCs In hippocampal pyramidal tells: tiation. ,Nwron 12. 111 l-1 120 1301 ISddL synapses:

I.T.R., Nrroll implications

131 I Liao D., Hrssler cally silent synapse5 c-ampal slice. Natcuc,

R.A.. Malenka for the expression

N.A., during

375,

seeking

contrrhution synaptic

J.L. lY93. synapses.

to cur-

Non uniform irienc-r 262.

potentration: evidence in the (‘A I regron of resolution.

in reliahrllty

Scienc of synaptic-

t’266, func-

3 71, 704~707. fluctuations impllcatlons

01 dual-component tor long-tt,rm poten-

R.C. 1995. Evrdcnce for silent of LTP k’rurcln 15, 427434.

Mallnow K. 1995. Activation pairinK-induccti LTP in CA1

of post-synaptiregion of hippo-

400-4114.

1321 Charpier S.. Behrrnds I.(:.. Troller ,A., Faher D.S., Latent Inhibitory connrc tlons become functional during dent plastic itv. Proc. r\‘ai/ Ac ad. 5~ 1. L!SA 92, 11 7-l 20. 1331 Clark K.. Col1ingrldgc~C.L. ponent excitatory postsynaptic Go/. 482, 39-52.

of quanta Neuron

Korn H. 19~5. actlvity-depen-

1905. Synaptic potentiation of dual-c-omcurrents in the rat hippocampus. /. Phy-

129

H. Korn,

D. S. Fabel

1341 Malgaroli A., Tsien tiation of the frequency rons. Nature 357, 134-l

1351 Malgaroli Tsien, R.W., potentiation 268, 1624-1628.

R.W. 1992. of miniature

Glutamate-induced synaptic currents

long-term in cultured

potenneu-

39.

A., Ting A.E., Wendland B., Bergumaschi A., Villa, A., Scheller, R.H. 199.5. Presynaptic component of long-term visualized at individual hippocampal synapses. Science

1361 Bolshakov V.Y., Siegelbaum S.A. transmitter release during development 5cience 269, 1730-l 7.34.

1995.

Regulation of hippocampal and long-term potentiation.

Korn H. 1988. of transmitter.

Synergism

Proc Nat/. AC.&

13.91 Kullmann D.M., Erdemli G., Asztbly of AMPA and NMDA receptor-mediated expression and extrasynaptic glutamate 1401 Bliss long-term

T.V.P., CoIlingridge G.L. 1993. potentiation in the hippocampus.

141 I Schuman E.M., Madison potentiarion in the hippocampus.

130

at central

synapses

due

Sci. USA 85, 870%

F. 1996. Long-term potentiation signals: Evidence for presynaptic spill-over. Neuron 17, 461-474. A synaptic Nature

model of memory: 361, 31-39.

D.V. 1991. Locally distributed Scienw 2 54, 1503-l iO6.

synaptic

R.D., Kandel E.R., Arancio 0. 1991. Tests on substances in LTP: Evidence for nitric oxide as messenger. Proc. Nat/. Acad. ,515. USA 88,

1431 Zhuo M., Small S.A., Kandel E.R., Hawkins R.D. 1993. Nitric oxide and carbon monoxide produce activity dependent long-term synaptic enhancement in hippocampus. Science 260, 1946-1950. [441 Oliet of quanta1 1294-1297.

1371 Oda Y., Charpier S., Murayama Y.. Suma C., Korn H. 1995. l.ongterm potentiation of glycinergic Inhibitory synaptic transmission. 1. Neurophysio/. 74:1056-1074. 1381 Faber D.S., to lateral diffusion 8712.

1421 O’Dell T.J., Hawkins the roles of two diffusible a possible early retrograde 11285-11289.

S.H.R., Malenka size by synaptic

R.C., Nicoll R.A. 1996. Bidirectional control activity in the hippocampus. .Srienre 271,

1451 Stric-ker C., Field A.C., Redman S.J. 1996. Changes meters of EPSCs in CA1 neurons in vitro after the induction

sd. 490.443-454. 1461 Davis C.W..

Murphey R.K. 1994. Long-term transmitter release properties: retrograde signaling ment. Tends Neurosci. 17, 9-l 3.

in quanta1 pamof LTP. 1. Phy-

regulation of short-term and synaptic develop-

1471 Geinisman Y., de Toledo-Morrell L., Morrell F., Heller R.E., Rossi M., Parshall R.F. 1993. Structural synaptic correlate of long-term potentiation: Formation of axospinous synapses with multiple, completely partitioned transmission zones. Hippocampus3, 435-445. 1481 Kullmann D.M., NMDA receptor-dependent 997-l 002.

C. R. Acad.

Siegelbaum S.A. 1995. The site of expression LTP: new fuel for an old fire. Neuron

Sci.

Paris.

Sciences

de

la vie / Life Sciences 1998. 321. 125-130

of 15,