begin to increase after the birth (12). Coronal sections (400 Am thick) of brains were obtained with a

Vol. 83, pp. 3022-3026, May 1986 Neurobiology

Acetylcholine raises excitability by inhibiting the fast transient potassium current in cultured hippocampal neurons

(A-current/muscarinic

agonists/4-aminopyridine/whole-ceil

recording)

YASUKO NAKAJIMA, SHIGEHIRO NAKAJIMA, REID J. LEONARD, AND KAZUHIKO YAMAGUCHI

Department ofBiological Sciences, Purdue University, West Lafayette, IN 47907 Communicatedby Charles F. Stevens, December 23, 1985

ABSTRACT The effects of

acetylchoLine

oncultured hip- pocampal neurons were investigated by using the whole-cell version of thepatch-clamp technique. TheCAl region of the hippocampus was excised frombrain slices ofyoung rats (12-19 day old), incubated in a papain solution, and dissociated.

Neurons wereplated on a glial feeder layer. The experiments were conducted mostly on neurons cultured for2-6 days. Upon depolarization under voltage clamp, these cellsexhibited a fast transient outward current (A-current), which wasinhibited by 4-aminopyridine(2.5 mM).

Acetylcholine

(0.1p) also inhib- ited thisA-current, as did themuscarinicagonistsbethanechol and muscarine. As expected from their inhibition of the A-current, acetylcholine and 4-aminopyridine both increased theamplitude oftheaction potential and prolonged its dura- tion. Weconclude that the inhibition of the A-current consti- tutes amechanismby which acetylcholine exerts itsexcitatory influenceonhippocampal neurons.

Neuronal excitation produced by the muscarinic action of acetylcholine(AcCho)is animportant cellulareventforbrain function, but details of its ionic mechanism are not fully understood. In 1971 Krnjevid et al. (1) investigated the muscarinic action of AcCho on cortical neurons by using intracellular microelectrodes, and they concluded that Ac- Choprobably produces itsexcitatory action byreducingboth resting and delayed potassium currents. Recently, Brown and Adams (2) and Halliwell and Adams (3) analyzed the muscarinic action under voltage clamp and explained muscarnnicexcitationasbeingdueprimarilyto aninhibition ofapotassium current,whichtheynamed the M-current.

We haveinvestigatedtheactionofcholinergic agonistson culturedhippocampalneuronsby using the whole-cell patch- clamptechnique. Ourresults show thatcholinergicagonists inhibit the fast transient potassium current, which was designatedasthe A-currentbyConnor andStevens(4).This inhibition of the A-current would play a role in

cholinergic

excitation. Theeffect ofAcChoresemblesthatof4-aminopyri- dine (APd),which hasbeen showntobearelatively

specific

inhibitor ofthe A-current(5).Previously,it hasbeen shownthat thyrotropinreleasinghormone (6), noradrenaline (7), and the intracellular level ofcyclic AMP (8, 9) may

modify

neural activityvia theireffectsonA-current.

MATERIALS AND METHODS

Culture.The culturemethodwassimilartothatdeveloped for dissociated cell cultures of locus coeruleus (10) and nucleus basalis (11). Young rats (Long-Evans, postnatal, 12-19dayold; Charles RiverBreedingLaboratories)rather than newborn or embryonic rats were used in the main, because it is known that muscarinic receptors in the hip-

pocampus begin to increase after the birth (12). Coronal sections (400

Am

thick) of brains were obtained with a vibratome. From these brain slices, the CA1 region of hippocampuswasexcised under a dissecting microscope. We begantheworkbyusing trypsin dissociation (10, 11), but we soonrealized that the papain treatment (13) yielded a health- ier culture, particularly when carried out with cells from young rather than newborn animals. Thus, the main work was done using the papain treatment. The dissected frag- ments wereincubatedtwice(15 min each)in an oxygenated L-15culture medium (pH 7.3) containing papain (12

units/

ml), DL-cysteine(0.2mg/ml), and bovineserumalbumin(0.2 mg/ml)at370C. Tissuefragmentswere thendissociated by trituration in modified Eagle's minimal essential culture medium with Earle's salt, L-glutamine (0.292 mg/mi), D- glucose(6mg/ml), NaHCO3 (3.7 mg/ml),L-ascorbicacid(10

pg/ml), penicillin (50 units/ml), streptomycin (50 Ag/ml),

and 10% heat-inactivatedrat serum (14). In some cultures, 10% fetal bovine serum and 10%o heat-inactivated horse serum wereused insteadofrat serum. Weusedasmall well atthe centerofaPetridish as aculture chamber, andthe dissociatedneurons wereplated on afeederlayerofglial cells (11).Mostexperimentswereconductedonmaterials cultured for2-6days,butinsomeexperiments materials culturedfor up to 27dayswereused.

Electrophysiology. The techniques were similar to those described (11).Weused thewhole-cellversion ofthepatch- clamp method (15).During the experiments,theculturewas superfused continuously with oxygenated Krebs solution containing148mM (or145.5mM)NaC1,2.5mM(or5mM) KCl,2.41mMCaCl2, 1.3mMMgC2,5 mMHepesNaOHbuffer, 11mMglucose(pH 7.4). Thepatchpipettewasfilledwith either internalsolution I(0.5 mM EGTACabuffer)orinternalsolution II (5 mM EGTA,

Ca-free).

Internal solution I was 120 mM potassium aspartate/40 mM

NaCl/5

mM Hepes/KOH buffer/0.5mM

EGTA/KOH/0.25

mM

CaC12/3

MMMgCl2/2 mM Na2ATP, pH7.1-7.2. Internal solution IIwas 115 mM

potassium aspartate/40

mM

NaCl/5

mM

Hepes/KOH/5

mM

EGTA/KOH/3

mM

MgC12/2

mMNa2ATP,pH7.1-7.2.Drugs were applied by pressure ejection (16) from micropipettes withatipdiameter of 4-5

pum

placedwithin 15-20

A&m

of the neuronal somasurface. The values of membrane potentials werecorrectedfora9-mVliquid junctionpotentialbetween the internal solution and the bathing solution. The bath temperaturenearthe neuron waskeptat

30'C-320C.

RESULTS

The neurons that we used from the CA1 area of the

hip-

pocampus were -20 ,um in diameter and were

pyramidal,

fusiform, ormultipolar. Previously, it has been shown that hippocampalneuronsin brain slicesorin culture

exhibit

fast transient outward currents (A-currents; refs.

17-19).

Our

Abbreviations:AcCho, acetylcholine; APd, 4-aminopyridine.

3022 Thepublicationcostsof this article^weredefrayedin partbypagecharge payment.This article must thereforebeherebymarked"advertisement"

inaccordancewith 18 U.S.C.§1734 solelytoindicate thisfact.

Proc.Natl. Acad. Sci. USA 83(1986) 3023

AcCho APd

10sec

FIG. 1. AcCho (0.1,uM)and APd (2.5 mM)suppress the fasttransient outwardcurrentsin a culturedhippocampalneuron.Theculture was superfused with Krebs solutioncontainingtetrodotoxin(1AM),and the cell wasvoltage-clamped bythe whole-cellclamptechnique.Thepipette contained internal solution II (5mMEGTA).The upper records are currents and lower records arepotentials.Theoutwardcurrentswere activatedrepeatedly, by steps, to -35 mV from aholding potentialof -90 mV. The resting potentialofthe cell was -78 mV. The barsindicate the time while the drug was pressure-ejectedfrom the pipette. The drug pipettes wereinitiallylocatedfarawayfromthe soma.Thepipettewas brought near the soma(within20Am)immediatelybefore the drug ejection was started. After the endoftheejection,thedrugpipettewasagain removedfrom thevicinityof the cell. Note that the transient outward currents began to be suppressed justbeforethe startof the APdejection:

this is due to a leak of APd from theapproaching pipette.

cultured neurons from young rats also exhibited a fast transientoutward current under the whole-cell patch clamp.

InFig. 1 the cellwasdepolarized repeatedly from a holding potential of -90 mV to the level of -35 mV under voltage clamp. With the slow time base used in Fig. 1, the fast transient outward currents are seen asspike-like upstrokes, which rapidly decline, followed by more slowly decaying outwardcurrents.Theapplication of APd (2.5 mM) or AcCho (0.1

AuM)

produced asubstantial reduction of the transient outward currentandamuch smallerreduction of the delayed outwardcurrent(Fig. 1).

InFig. 2 theoutward currents evoked by three different depolarizationsaredisplayed. Inthe controlsolution, depo- larizations to -43 mV or to -25 mV evoked large outward currents, reaching apeak at 6-11 msec and then declining rapidlywithatimeconstantof 15-18 msec(Fig.2A2 and A3).

RecordsinFig.2B and C illustrate theeffects of AcCho (0.1

,4M)

or APd (2.5 mM) on these outward currents. At a

moderate depolarization (-43 mV), the fast transient out- ward currentswerealmost completely eliminated by either AcChoorAPd(Fig. 2 B2 and C2), but at ahigherdepolar- ization (-25 mV), the inhibition of the transient outward currentsbythesedrugswas notcomplete(Fig.2 B3 andC3).

To measureAcCho- or APd-sensitive currents, digital sub- traction of currents inthe presence ofAcCho or APd fromthe control currents was used (Fig. 2 D and E) (cf. ref. 19forthe APd-sensitive currents). The time courses of the AcCho- sensitive andthe APd-sensitive currents look very similar, except thatthe latter hasalarger residualcurrent at-25mV, indicatingthat APdexerts a greatereffect than AcChoonthe delayed partof theoutwardcurrents.Theseresultsindicate that the fast transient outward currents in our cultured neurons are mostprobably A-currents and that AcCho hasan inhibitory effectonthe A-current.

Thedosedependency of the inhibitory action of AcChoon the A-current has not been worked out completely. But Al Control (mean) B1 AcCho Cl APd DI(Control-AcCho)El (Control-APd)

+

JVNV

626MV

A2 52 C2

62

~~~~~~D2

^E

-1A ~~-43mV

B3 C3

Ij

FIG. 2. AcCho and APdeffectsonoutwardcurrentsina neuronbathed in the Krebs solution containing tetrodotoxin(1 ,uM).Thepatch pipette contained internal solution II (5 mM EGTA).The membranewasdepolarized, bysteps,fromaholding potentialof-108mVto-62 mV(Al-El),-43mV(A2-E2), andto-25mV (A3-E3). (AJ-A3) Averages oftwocontrolcurrentsbeforeandafter the drug applications.While AcCho(0.1,M)orAPd(2.5 mM)wasapplied(for25 sec), depolarizations of various amplitudeswereappliedtorecordthecurrentsin BJ-B3 orinCJ-C3. Arrowsindicate thezerocurrentlevels.In DJ-D3 (orEl-E3),theAcChorecords (Bl-B3) (or the APd records)weresubtracted from thecontrolrecords (Al-A3).Digital sampling^ratewas5 kHz.

A3

D3 E3

---1.

Iik

^Im

]-25^mV

100misc

Neurobiology: Nakajima

etal.

AcCho at high concentrations (1 ,uM or more) did not

necessarily producelargereffectsonthe A-current thanat0.1 ,uM.Verylowconcentrations (=0.01,uMAcCho)orcontrol solution (without AcCho) didnotproduceinhibitoryeffects

onthe A-current. Sometimes the control solutionproduceda

slight enhancement of the A-current, buttheeffect^wasvery small, and itmayhaveatrivialexplanation.

We have attempted to analyze the mechanisms of the AcCho and APd effects by studying their voltage dependen- cies. In Fig. 3, the open symbols show the steady-state inactivation of the A-current ^{as a} function of membrane potential. The membranepotentialwasheldat -40mV and then was hyperpolarized to various levels (for 700 msec) before it was depolarized to -22 mV. The peak current

produced by this depolarizationwasplotted against the level ofhyperpolarization. These datawerefitted byanequation similartotheonethat describes the steady-state Na inacti- vation (20); namely,I = TI{1 + exp[(V ^- V)/k]}-1 + C,in which k and ^Varetheparametersfor thesteepnessand the half-inactivated voltage, respectively, and C is the current that is notinactivated (including the leakcurrent). Table 1 Eves theaveragevalues of k and^V for six cells. The data for V indicate that the inactivation curves are shifted in the depolarizing direction by 13 mVor7 mV in thepresenceof AcChoorAPd. The value ofI representsthemagnitude of the maximum activatable A-currentat -22 mV. For the six cells inTable 1, AcCho (0.1

jLM)

reduced^Tby 55% 5%

(mean ^± SEM)and APd(2.5mM)reduceditby 41% 8%, indicating that the potencies of 0.1 ,uM AcCho and 2.5 mM APdarecomparable.

The solid symbols in Fig. 3 illustrate the relationship between potential and the activation of A-currents. The potential washeld at -108 mV and various steps ofdepo- larization were applied. The peak of the A-current was

plotted againstthedepolarization.These data showthat the A-currents are activated at about -50 to -60 mV in the control and also suggest that the curves are shifted in the depolarizing direction by AcCho^orAPd. We didnotfurther analyze the activation mechanisms in thispaper. Neverthe- less,ourAPdresultsagreewellwith therecentdatabyKasai

1500

C

a)

as)

(6

1000

500

-120 -100 -80 -60 -40 -20

Membrane potential, mV

FIG. 3. Activation(closedsymbols)and thesteady-stateinacti- vation(opensymbols) of the fast transientcurrents.*and o,control;

* ando,AcCho(0.1 ,uM); *and A, APd(2.5 mM).Thepatch pipette contained internalsolutionII(5mMEGTA).Theactivationplotwas

obtained by depolarizing the membrane to various levels froma

holdingpotential of-108 mV. Theinactivationplotwasobtainedby holdingthepotentialat-40mV,hyperpolarizingtovariouslevels for 700msec,and thendepolarizingto-22 mV. The inactivationcurves werefittedby the equationI=T{1 + exp[(V- V)/k]}-1+C. The initialresting potentialwas-79 mV.

Table 1. Steady-state inactivation ofA-current

AcCho APd

Control (0.1AM) (2.5mM)

V,mV -73± 1 -60±2 -66± 2

k,mV 5.7 ±0.3 6.2 ±0.4 5.6±0.2 Valuesare mean+SEM (n = 6). Theinactivation curve was fitted

bythe

equation

I=T

{1

+ exp[(V^-

7)/k]}-1

+ C.

(21), who found similar effects of APd on activation and inactivationby single channel analysis.

To eliminate the possibility of contamination from

Ca2'

currents, wealsoperformedexperiments on cells bathed in a Krebs solution containing 5 mM cobalt chloride. The A-currentpersistedinthe presence of cobalt ions, although its activation curve tended to be shifted in the depolarizing direction, in agreement with the effect of manganese on A-currents in othermammalian neurons (22). In the cobalt solution,weobservedthat theA-current wasstillinhibited by AcChoaswell asby APd.

ActionPotentials. Action potentials recorded by the whole- cellcurrentclampareshown inFig.4.Thecontrolrecord in Alshows that onlyoneaction potential wasproducedatthis current intensity near the threshold. In the presence of AcCho(0.1ALM) or APd (2.5 mM), however, trains ofspikes wereelicited by the same intensity of current (Fig. 4B) and Dl). This phenomenonhasbeen reported by Segal et al. (18) forAPd andby Cole and Nicoll (23) for AcCho.Therecords inFig.4A2-E2illustratetheactionpotentialat thebeginning ofeachspike trainat afaster time scale. Theactionpotentials undertheinfluence of AcChoorAPdarelargerand oflonger duration than the control. In six cells, AcCho (0.1

AM)

increasedthespike amplitude by11%± 1%(mean± SEM), and the duration (measured at the midpoint between the threshold and thepeak) wasincreasedby24% ± 9%; APd (2.5 mM) increased the amplitude by 11% ± 2% and the durationby81% ± 14%.

Receptors for the AcCho Effect. Hippocampalandcortical neurons are known to be rich in muscarinic receptors.

Excitatory effects of AcCho that Krnjevid et al. (1) and Brown and Adams (2) obtained have been ascribed to a muscariniceffect. In the presentstudy, atropineat0.5 ,uM partially antagonizedthe inhibition ofA-current byAcCho (0.1

AM).

However,athigher concentrations (1

AM

ormore), atropine sometimes produced an agonistic effect, itselfin- hibitingthe A-current. Bethanechol(250 ,uM),amuscarinic agonistwithoutnicotinicaction, hadaneffectsimilartothat ofAcCho, as did muscarine at 2.5-20 ,uM. These results suggestthat the AcCho inhibition ofthe A-current is medi- atedthroughmuscarinicreceptors.

DISCUSSION

TheA-current has been described in many neurons.Itisan outward K current with arapidly

inactivating

time course similar to the one described in this paper and also with activationandinactivationcurveslike those in

Fig.

3 (4, 7, 17-19, 21, 22, 24-28).Inaddition,the A-currentofmamma- lianneuronsisinhibitedbyAPd. In the present

experiments,

AcChoproducedalmost thesameinhibition of the transient outward current,asdid APd.Thus,weconclude that AcCho inhibits the A-current inour

hippocampal

neurons.

We propose that A-currentinhibition constitutesamech- anism by which AcCho exerts its

excitatory

influence through muscarinicreceptors. Asshown in

Fig.

4,under the influence of

AcCho,

neuronstendto

produce high-frequency

repetitive firing.Atfirst

sight,

this

repetitive firing

doesnot

Proc. Natl.Acad. Sci. USA 83 (1986) 3025

Al Control BI AcCho C1 Control Dl APd El Control

,1~~~~~_ 1114X lA.lU-

-X e

I

I=

^]C

As - B2 C - DP E2 1

5msec

FIG. 4. Actionpotentialsrecordedfromaculturedhippocampalneuron.Thehorizontallinesindicate thezeropotentiallevels.(Al)Control spike(upper trace) elicitedbyasquareconstant current(lowertrace).Duringtheapplicationof 0.1,uMAcCho(Bl)or2.5 mMAPd(Dl),the samecurrentsgenerated trainsofspikes.These effectswerereversible,asshown inCl and El.(A2-E2)Thefirst actionpotentialsseeninAl-El on afaster time scale. Theheightandduration of the actionpotentialsin the presence of thedrugs (B2andD2)werelargerthanthoseofthe controls (A2, C2, and E2). In alltherecords,the cellwasunderwhole-cellcurrentclamp.Thepatchpipettecontained solution I(0.5mMEGTA).

The overallfrequencyresponseof therecordingsystem, determinedbytheplaybackspeedratio of the tape recorder andbythefrequency responseof the penrecorder,was2kHz.

appear to becaused by the A-currentinhibition, sinceafter the endof the firstspike, the membrane potentialwas never sufficiently negative for inactivation of the A-currentto be removed. Nevertheless,accordingtothe recentsinglechan- nel databy Kasai (21), the A-currentis inactivated with two time constants, 100 msec and4sec. Thus, it is verylikelythat after the end of the control spike (Fig. 4AI), a part of A-current stillremains withoutbeingcompletelyinactivated.

Inhibition of this remaining A-current by AcCho would produce a substantial effect on the ease of firing of the subsequent spikes. The role of the A-current in modulating the spikerepetition rate has already been well explained by Connor and Stevens (29). It should benoted,however,that some effects of AcCho or APd other than the inhibitionof A-currentcould exist, and theseeffects mayplay a role in

producing

the

repetitive

tendency.In

fact,

Fig.2shows that lessinward current isrequiredtohold thepotential at -108 mVin the presenceof AcCho or APd, an effect that may not beattributed to aneffect on the A-current. Coleand Nicoll (23) suggested that inhibition of the Ca-induced K current underlies AcCho-inducedrepetitivefiring.

Inhibition of the A-current would allow asingle spike to be evoked moreeasily, provided the resting potential before the spike wassufficientlynegative. This situationwaspredicted byConnor et al. (30) and has been recentlyobserved by Segal etal. (18)inconnectionwithinhibitionof the A-current by APd. Thus, an excitatory postsynaptic potential, which would have been below threshold, may now, under the influence ofAcCho, be able to initiate aspike.

Theincrease in theamplitude and the duration of the spike byAcCho and APd, seen inFig. 4 B2 and D2, would certainly be related to theinhibition of A-currents. The A-current, with its fastkinetics,is an importantfactor in determining the size of the action potential and its time course ofrepolarization;

the role played by the A-current in determining the repolarization phase was emphasized by Belluzzi et al. (28).

Anenhancement of spike height and duration would exert a powerful excitatory action if these changes took place at presynaptic nerve terminals: there, small changes of action potentialheightandshape produce remarkable changes in the amountof transmitter released(31-34).Infact, a presynaptic excitatory effect mediated by muscarinic receptors has re- cently been suggested by Raiteri et al. (35). In summary, AcCho mimics the excitatory action of APd, a convulsant, whichwould exert itsexcitatory influence mainly through the inhibitionof the A-current.

All the excitatory effects wehavedescribed in this study took place in the probable absence of M-currents. Underour experimental conditions, we could rarely observe M-cur- rents,perhaps because of thepresenceof 3 mMmagnesium in thepatch pipette (36). Under more physiological condi- tions, the modification of both the M-current and the A- current could work together in producing the excitatory action of AcCho. The muscarinic effects are multifaceted phenomena (1-3, 23, 37-40). Foramuscarinic effecttotake place, the cellmusthavethefollowingthreecomponents: (i) thereceptor, (ii)themessenger,and(iii)the targetchannel.

Different responses could arise not only because of the existenceofavarietyof target channels but also becauseof theexistence ofmultiple classes ofmuscarinicreceptors(41) or because of the possible involvement of more than one messenger. Probably our experimentalconditions were fa- vorableforonefacet of these complexevents tobeobserved.

Wethank Dr. Peter R. Stanfield for criticallyreadingthe manu- script.Thanksarealso due to Ms. PamellaSchroeder and Ms.Anita Robinson for their technical and clerical help. The research was supported by National Institutes of Health Grants AG06093 and NS08601 and by an Alzheimer's Disease and Related Disorders AssociationGrant.

1. Krnjevid, K., Pumain, R. & Renaud, L. (1971) J. Physiol.

(London)215, 247-268.

2. Brown, D. A. &Adams, P. R. (1980) Nature(London) 283, 673-676.

3. Halliwell,J. V. &Adams,P. R.(1982) Brain Res. 250,71-92.

4. Connor, J. A. & Stevens, C. F. (1971) J. Physiol. (London) 213, 21-30.

5. Thompson, S. H. (1977) J. Physiol.(London)265,465-488.

6. Barker,J. L.,Dufy,B.,Owen,D.G. &Segal,M.(1983)Cold SpringHarborSymp. Quant. Biol. 48, 259-268.

7. Aghajanian,G. K. (1985)Nature(London) 315, 501-503.

8. Kaczmarek,L. K.&Strumwasser, F.(1984) J.Neurophysiol.

52, 340-349.

9. Strong,J. A. (1984)J. Neurosci. 4, 2772-2783.

10. Masuko, S., Nakajima, Y. & Nakajima, S. (1984) Soc.

Neurosci.Abstr. 10, 659.

11. Nakajima, Y., Nakajima, S., Obata, K., Carlson, C. G. &

Yamaguchi, K. (1985) Proc. Natl. Acad. Sci. USA 82, 6325-6329.

12. Dudai, Y., Ben-Barak, J., Silman, I. & Gazit, H. (1980) in Neurotransmitters and Their Receptors, eds. Littaue, U.Z., Dudai, Y., Silman, I.,Teichberg, V. I. & Vogel, Z. (Wiley, Chichester,UnitedKingdom),pp.217-239.

13. Leifer, D., Lipton,S.A.,Barnstable,C. J. &Masland,R. H.

Neurobiology: Nakajima

etal.

(1984) Science 224, 303-306.

14. Dichter, M. A. (1978) Brain Res. 149, 279-283.

15. Hamill, 0. P., Marty, A., Neher, E., Sakmann, B. &

Sigworth, F. J. (1981)PflugersArch. 391, 85-100.

16. Choi, D. W. & Fischbach, G.D. (1981) J. Neurophysiol. 45, 605-620.

17. Gustafsson,B., Galvan, M.,Grafe,P.&Wingstram,H. (1982) Nature(London)299,252-254.

18. Segal,M.,Rogawski,M. A. & Barker,J. L. (1984) J. Neuro- sci.4, 604-609.

19. Zbicz, K. L. & Weight, F. F. (1985) J. Neurophysiol. 53, 1038-1058.

20. Hodgkin, A. L. & Huxley, A. F. (1952) J. Physiol. (London) 117, 500-544.

21. Kasai, H.(1985) Soc. Neurosci. Abstr. 11, 955.

22. Galvan, M. & SedImeir, C. (1984) J. Physiol. (London)356, 115-133.

23. Cole, A. E. & Nicoll, R. A. (1984) J.Physiol. (London)352, 173-188.

24. Hagiwara, S., Kusano, K. & Saito, N. (1961) J. Physiol.

(London) 155, 470-489.

25. Nakajima, S. & Kusano, K. (1966) J. Gen. Physiol. 49, 613-628.

26. Nakajima,S.(1966)J. Gen. Physiol.49, 629-640.

27. Neher, E. (1971) J. Gen.Physiol. 58,36-53.

28. Belluzzi, O., Sacchi, 0. & Wanke, E. (1985) J. Physiol.

(London)358, 91-108.

29. Connor, J. A. & Stevens, C. F. (1971) J. Physiol. (London) 213, 31-53.

30. Connor, J. A., Walter, D. & McKown, R. (1977) Biophys. J. 18, 81-102.

31. Takeuchi, A. & Takeuchi, N. (1962) J. Gen. Physiol. 45, 1181-1193.

32. Katz, B.&Miledi,R.(1967) J. Physiol. (London) 192, 407-436.

33. Heuser, J. E., Reese, T.S., Dennis, M. J., Jan, Y., Jan, L. &

Evans, L. (1979) J. Cell Biol. 81, 275-300.

34. Shimahara,T.(1983) Brain Res. 263, 51-56.

35. Raiteri, M., Leardi, R. & Marchi, M. (1984) J. Pharmacol.

Exp.Ther. 228, 209-214.

36. Galvan, M., Satin, L. S. & Adams, P. R. (1984) Soc. Neuro- sci.Abstr. 10, 146.

37. Akasu, T. & Koketsu, K. (1982) Neurosci. Lett. 29, 41-45.

38. Nowak, L. M. & Macdonald, R. L. (1983) J. Neurophysiol.

49,792-803.

39. Madison,D.V.,Lancaster, B.,Nicoll,R. A. &Adams,P. R.

(1985)Soc. Neurosci.Abstr. 11, 467.

40. Spain, W., Schwindt, P. C., Chubb, M.C. & Crill, W.E.

(1985) Soc. Neurosci.Abstr.11, 786.

41. Birdsall,N.J. M.&Hulme, E. C. (1983) Trends Pharmacol.

Sci. 4, 459-463.