The Formation of Chemical Synapses Between Cell-Cultured Neuronal Somata

The Journal of Neuroscience, March 1988, 8(3): 1032-l 038

P. G. Haydon Department of Zoology, Iowa State University, Ames, Iowa 50011

The study of the development and plasticity of chemical releaseof neurotransmitter from the growth cone, and within synaptic connections is frequently restricted by the lack of secondsof contact, miniature postsynaptic currents can be de- access to the synaptic terminals. This can, in part, be over- tected in the postsynaptic muscle cell (Xie and Poo, 1986). come by plating neurons into cell culture where all regions However, suchtransmission does not representthe fully differ- of a neuron are made experimentally accessible. However, entiated synapticconnection. For example,after the presynaptic the small size of synaptic terminals still makes direct ex- neurite contactsthe musclemembrane, alterations in the spatial perimental manipulation difficult. In this study we have found, distribution of postsynaptic AChRs occur (for review, see in the absence of neurite extension, directly contacting cell Schuetzeand Role, 1987).Although suchstudies have provided somata (diameter 50-100 pm) will form chemical synapses. insight into the development of synaptic transmission,little is Identified neurons 85 and B19 of Helisomawere plated into known about the physiological development of the presynaptic culture under conditions that promote adhesion between cell terminal. The small size of synaptic terminals has madeit dif- pairs. Under these conditions, neurite outgrowth was absent, ficult to directly record and control presynaptic membranepo- but action potentials in B5 evoked inhibitory postsynaptic tential during development, as has proved possiblein a limited potentials in B19 that were reversed in sign by the injection number of mature preparations(Auerbach and Bennett, 1969; of chloride ions and were blocked by tubocurare (10m5 M), Martin and Ringham, 1975; Llinas et al., 1981a, b; Wojtowicz reduced extracellular Ca2+, and Cd*+ ions. Such synapses and Atwood, 1984; Lemos et al., 1986). With the advent of exhibited classical properties of chemical synapses, includ- patch-pipettes, however, it has been possibleto study growth ing the spontaneous release of neurotransmitter. conesin cell culture. For example,using cell-attached and whole- Since somatic synapses represent an appropriate model cell recording techniques,it has been determined that growth of synaptic transmission, this system was utilized to study conescontain a variety of voltage-gatedion conductances(Co- the role of mutual neuronai contact in the development of han et al., 1985; Belardetti et al., 1986a, b) which sharecertain transmitter release capabilities. Future pre- and postsyn- properties with the membrane of neuronal somata(Belardetti aptic somata were cultured separately for 3 d, the period et al., 1986b).However, it hasnot yet beenpossible to determine required for the development of synaptic transmission under the developmental events within the presynaptic terminal fol- conditions of maintained contact. Then, neurons were made lowing contact with the postsynaptic cell. to contact and intracellular recordings taken within 0 to 4 Identified neurons of leech (Fuchs et al., 1981, 1982; Hen- hours. Postsynaptic potentials were detected as early as 10 derson, 1983; Henderson et al., 1983; Arechiga et al., 1986), set following contact. Thus, qualitatively the development Aplysia (Camardo et al., 1983; Bodmer et al., 1984; Schacher of transmitter release capabilities does not require main- et al., 1985), and Helisoma (Hadley et al., 1983, 1985) have tained contact. The ability to form synapses between directly been usedin cell-culture studiesof electrical and chemical syn- contacting spherical somata now permits the study of both aptogenesis.Isolated neuronsplated in cell culture extend neu- mature excitation-secretion coupling and the development rites and form synaptic connections.An attractive feature of the of synaptic transmission at a new level of resolution since leech cell-culture systemis that neuronsextend short neurites both the pre- and postsynaptic elements of a neuronal syn- and, as a consequence,microelectrode recordings made from apse are directly accessible to experimental manipulation the cell body are electrically close to the synaptic terminals and control. (Fuchset al., 1982; Pellegrinoand Simonneau,1984; Dietzel et al., 1986). While neuronal processescharacteristically form the Recent work on the development of synaptic connectionshas presynaptic elementsof synaptic profiles, recent evidence sug- increasedour understandingof the events leading to the for- geststhat nerve cell bodiesmay be accessiblemodels for study- mation of functional synaptic transmission.Motile growth cones ing presynapticfunction. For example, denervatedavian ciliary contain and can releaseneurotransmitter (Hume et al., 1983; ganglia releaseACh in responseto K+ depolarization, anti- Young and Poo, 1983). Contact with muscle induces further dromic stimulation, and cholinergic agonists;this led the au- thors to concludethat neurotransmitter is releasedfrom somata (Johnsonand Pilar, 1980). Additionally, Chow and Poo (1985) Received May 4, 1987; revised Aug. 24, 1987; accepted Sept. 10, 1987. demonstrated that spontaneousminiature potentials were de- This work was supported by PHS Grant NS24233. I wish to thank Drs. C. tected in myoblasts following contact with neuronal somata. Drewes, R. G. Wang, and M. Zoran for critical comments on the manuscript. Correspondence should be addressed to P. G. Haydon at the above address. Taken together, theseobservations raisethe possibility that a Copyright 0 1988 Society for Neuroscience 0270-6474/88/031032-07$02.00/O pair of neuronal cell bodiescould form a chemical synapse. The Journal of Neuroscience, March 1988, 8(3) 1033

bocurarine chloride (Sigma) was dissolved in standard Helisoma saline for application to cultured cells. ACh (Sigma) was focally applied to neuron B 19 using positive-pressure ejection from a micropipette. Neuronal morphology. The morphology of neuron pairs was determined by phase-contrast optics using a Nikon Diaphot inverted micro- scope. To further test for the presence of neurites extending from a neuronal cell body the fluorescent dye Lucifer yellow was iontopho- retically injected into one neuron of a cell pair as described previously (Cohan et al., 1987). Neuronal morphology was subsequently ascer- tained using fluorescence optics of the Nikon Diaphot on living tissue. In control experiments where neurons were plated into culture under Figure 1. Schematic illustrating the technique used to cause contact conditions that promote neurite extension, intrasomatically injected and, subsequently, adhesion between pairs of neurons. Lucifer yellow reliably filled neurites over distances in excess of 1 mm, and, when neurites passed over adjacent cell bodies, these processes were readily discernible with this dye. In this report a test is made of the hypothesis that in the absenceof neurite extensionneuronal cell bodiesare competent Results to form chemical synaptic connections. Specifically, identified Morphology neuronsB5 and B19 of Helisoma were utilized; theseneurons Neurons within the nervous systemof Helisomaare identifiable have previously been shownto form a unidirectional chemical on the basesof visual criteria, which include cell size, coloration, synapse in cell culture when contact is made through newly and position within the ganglion. Isolated neuron pairs plated extended neurites (P. G. Haydon and S. B. Kater, unpublished onto an adhesive substratefor recording retained their identi- observations). This study demonstratesthat neuronal somata fiability on the basesof 2 of thesecriteria: (1) neuron B5 is larger readily form chemicalsynaptic connections,and their properties than neuron B 19(Fig. 2), and (2) B5 containsa prominent region parallel those of the more classicalsynapses that are located on of pigmentation. Subsequentpenetration with intracellular mi- neuronal processes.Additionally, by manipulating the timing croelectrodesconfirmed cell identity, sincethe duration of neu- of initial contact betweenfuture pre- and postsynaptic partner ron B 19’s action potential is shorter than that of BS (Cohan et somata, the role of contact in regulating the development of al., 1985). transmitter releaseproperties is examined. Using phase-contrastoptics, cell pairs appearedas spherical structures devoid of net&es. However, to confirm that cell Materials and Methods interactions were mediated directly by somata, the fluorescent In all experiments, adult specimens of the albino strain (red) of the pond dye Lucifer yellow wasiontophoretically injected into the soma snail Helisoma were used. Animals were laboratory-reared, maintained of one neuron of a pair. Subsequentvisualization using flu- in aquaria, and fed lettuce and trout chow. As described in detail pre- orescent optics never revealed the presence of neurites overlying viously (e.g., Wong et al., 198 1; Haydon et al., 1985), snails were de- the membraneof the adjacentpaired cell. Each injected neuron shelled, and the paired buccal ganglia were acutely isolated from individual snails in antibiotic saline (ABS). Paired buccal ganglia were treated had the appearanceof a sphere(Fig. 2). From dye injections with a 0.2% solution of trypsin (Sigma, type III) in defined medium made on days 2-7 of culture, neuriteswere never detected(n = (DM; 50% Leibowitz-15 (CiBCO) with Helisoma salts) for 20-45 min, 7) indicating that neuronsB5 and B 19 make direct contact by followed bv 15 min in 0.2% trvnsin inhibitor (Sigma type II-S). Using their cell somata. Furthermore, examination of serial sections sterile procedures, enzymaticaliy treated ganglia were -pinned to the of cell pairs with transmissionelectron microscopy has not re- silicon rubber floor of a dissection dish for single-cell excision. Using an electrolytically sharpened tungsten microknife, a small slit was made vealed the presenceof neuritesextending from one cell over the in the connective sheath adjacent to the visually identified neuronal cell other (Emery and Haydon, unpublishedobservations). body. By applying gentle pressure to the ganglion, the cell of choice “popped” out through the slit, and by applying negative pressure through Synaptic connectivity a glass suction pipette, the neuronal soma and attached proximal axon were reliably removed from the ganglionic environment (Haydon et al., Since adhesionbetween directly apposedsomata is maintained 1985). in the absenceof neurite extension, the state of connectivity of Cell culture. Neurite extension critically relies on adhesion to an ap- such neuron pairs was determined. At this time, the presence propriate substratum. Thus, to prevent neurite extension, neurons were of both electrical and chemical synapseswas tested. plated into a nonadhesive culture environment. Nonfiltered snail hemolymph when applied to a culture dish reliably prevents cell adhesion Hyperpolarizing current was injected into either B5 or B19, to the substrate. Hemolymph was extracted from snails on the day of and the responsein the paired neuron wasdetermined to assay culture, as described by Hadley and Rater (1983). Neurons were plated for electrical coupling. In 63% of preparationsDC current did into a 40 ~1 droplet ofa 50% solution ofhemolymph in DM. To optimize not pass between cell pairs (n = 22 of 35; days 1-8; Fig. 3) contact and subsequent adhesion between cell pairs, a depression was made in the base of a Falcon #lo08 petri dish into which neuron pairs although in remaining preparations electrical connectionsdid were plated (Fig. 1). In this manner, greater than 95% of cell pairs form (e.g., Fig. 6). By contrast, intracellular recordingsrevealed adhered to one another and never to the substrate. After a period of 4- the reliable presenceof a unidirectional chemical synapse. 9 hr, cell pairs were removed from this droplet, using a suction pipette, The connectivity of paired neurons was tested on days 1-8 and transferred to 2 ml of a 1% solution of hemolymph in DM. For of culture. On days l-2 of culture chemicalsynaptic connections prolonged culture, hemolymph was changed every 34 d. Electrophysiology. In order to test for synaptic connectivity, neuronal were found in 1 of 7 pairs examined. After a latent period of 3 cell oairs were transferred to a Falcon #lo08 petri dish containing DM d, however, action potentials in neuron B5 evoked synaptic or, more characteristically, to a polylysine-coated Falcon #300 1 culture responsesin B19 in 100% of cell pairs tested (Fig. 3; n = 28, dish (Wong et al., 1981), where they rapidly adhere to the substrate. days 3-8 of culture), with latenciesin the rangeof 0.4-9.3 msec Such immobilized cells were penetrated with glass microelectrodes filled (n = 7; mean = 4.6 msec). Chemical synapsesformed irrespec- with either 2 M potassium acetate (&AC) or with 1.5 M potassium chloride (see Results) and had DC resistances in the range of 8-25 MQ. tive of whether a conjoint electrical connection was present. Standard electrophysiological techniques were utilized throughout. Tu- Initial recordingswere made from cell somatausing 2 M potas- 1034 Haydon l Chemical Synapses Between Neuronal Somata

Time (days) B19 I !jrnV 110 iz 85 IL Figure 3. Isolated neuronal somata form chemical synaptic connections in cell culture. Intracellular recordings made from neurons B5 and B19 demonstrate that somata form chemical synaptic connections. Bot- tom, Action potentials in B5 evoke inhibitory synaptic potentials in B19 (left), which are reversed in sign with hyperpolarization of B19’s membrane potential &%-filled microelectrodes). In this cell pair, DC current did not pass between neuronal somata, indicating that they were not electrically coupled (right). Top, Time course of formation of the B5-B 19 somatic chemical synapse. Neuron pairs were removed from culture on days 1-8 of culture and the state of connectivity was examined Figure 2. Neuron pairs maintain a spherical structure in cell culture. (n = 35). This chemical synapse is first reliably detected on day 3 of Cell pairs maintained contact for the total duration of culture and, when culture and is maintained throughout the experimental period of 8 d. viewed with phase-contrast optics (top: B5, left; B19, right), were seen to adopt a characteristic spherical morphology. The injection of Lucifer yellow into neuron B5 confirmed that somata did not extend neurites somataand to test the cholinergic nature of suchtransmission, (bottom). Scale bar, 50 Wm. we performed the following tests: tubocurarine chloride was addedto the bathing mediumof synaptically connectedcell pairs sium acetate-filled glass microelectrodes. Under such condi- and ACh was focally applied to the membrane of B19. Tubo- tions, action potentials in B5 evoke inhibitory synaptic re- curare (1O-5 M) reliably reducedthe amplitude of the B5 evoked sponsesin B 19, which reversedin sign by hyperpolarization of synaptic action in B 19 (n = 4; Fig. 4). Such data implicate ACh the postsynaptic membanepotential (Fig. 3). In previous ex- as being the neurotransmitter utilized at this synapse.This is periments performed on cells with net&es, we observed that further corroborated by the application of ACh to the membrane impaling B19 with a KCl-filled electrode reversedthe polarity of B 19 (days O-3). When usinga potassiumacetate electrode to of the B5 evoked ipsp recordedin B19 (P. G. Haydon and S. B. record from B19, ACh applied to B19 evoked a hyperpolar- Kater, unpublished observations). Therefore, we determined ization that was reversed in direction when B 19 was hyperpo- whether KC1 electrodeshad similar effects on the somaticipsp. larized by DC current injection. When using a KCl-filled elec- Penetration of B 19 with a KCl-filled microelectrode causeda trode to record from B19, ACh evoked a large depolarization rapid reversalof the evoked ipsp (Figs.4-7). Thesedata indicate of B19 (n = 4; Fig. 4). Furthermore, tubocurarine (1O-5 M) that the synapsewhich forms between contacting somata has reversibly reduced the amplitude of suchACh-evoked depolar- the sameproperties as the connection betweennewly extended izations (n = 3). Taken together, such evidence indicates that neurites, and that the synaptic potential is mediated, at leastin neuron B5 forms a cholinergic chemicalsynapse with B 19 when part, by chloride ions. For the remainder of this study, KCl- neuronal somataare in direct contact. filled microelectrodeswere utilized since,in addition to revers- Intracellular recordings made from B5 and B 19 have dem- ing the sign of the synaptic potential, its amplitude was aug- onstrated that neurotransmitter can be releasedspontaneously. mented. At resting potential, the isolated soma of B5 did not discharge The chemical connection that forms between neurites of B5 spontaneousaction potentials. However, simultaneousrecord- and B19 is believed to be cholinergic (P. G. Haydon and S. B. ings madefrom B 19 demonstratedthe presenceof spontaneous Kater, unpublished observations). To further substantiatethe postsynaptic potentials (Fig. 5). Depolarization of B5 increased chemical nature of synaptic transmission betweenpairs of cell the frequency of suchpsps. The injection of depolarizingcurrent The Journal of Neuroscience, March 1988, 8(3) 1035

Tubocurare

Bl9

B5 20mV 1; D )I 400msJ P.. Tubocurare O'gJ ------* * 20mV Gl 2mV Figure4. The synapticconnection that formsbetween somata of B5 I andB 19is sensitiveto tubocurare.Single action potentials in B5 evoke excitatorysynaptic potentials in B 19when the postsynapticneuron was impaledwith a KC1electrode (upper left). The additionof lOA M tubo- curareto the bathingsaline reduced the amplitudeof the B5 evoked synapticpotential (middle), which was reversed upon washout of this cholinergicantagonist (right). Pipette application of ACh to B19(lower left) evokeda similardepolarization of B19(KC1 electrode), which was Figure5. Synapticpotentials in B19detected in the absenceof action alsoreversibly reduced in amplitudeby applicationof 1O-5M tubocu- potentialsin BS.Prolonged depolarization of B5 evokeda singleaction rare. potential,which wasfollowed by a maintaineddepolarization of the cell.The actionpotential in B5evoked a single,large-amplitude synaptic potentialin B19, which wasfollowed by smalleramplitude synaptic potentials(upper truces). Furthermore, spontaneous synaptic potentials into B5 evoked a singleaction potential and associatedevoked couldbe detectedin B19 in the absenceof actionpotentials and de- synaptic potential in B 19. Frequently, maintained depolariza- polarizationof B5 (lowertruces). The maintaineddepolarization of B5 increasedthe frequencyof suchsynaptic potentials that arecomposed tion of B5 failed to evoke additional action potentials. Under of multipleamplitude events (note different time scalebetween upper theseconditions of sustaineddepolarization, however, a barrage andlower traces). of desynchronized pspswas detected in B 19 (Fig. 5).

Ca2+-dependenttransmitter release Neurotransmitter releaseat synaptic terminals is dependenton became immobilized and were impaled with microelectrodes an influx of Ca*+. The calciumdependence of transmitter release within minutes, or up to 4 hr following contact, in order to in this systemof neuronal somatawas demonstratedby the use determine connectivity. of a low-Ca2+saline and salinecontaining the Ca channelblock- Intracellular recordingsmade from neuronsthat first contact er Cd*+ (Kostyuk, 1980).The concentration of extracellular Ca2+ on day 3 demonstrate that maintained contact for the total wasreduced from 4.1 to 0 mM, while the Mg2+ concentration culture period is not required for the development of functional was increasedby 4.1-5.6 mM, or Ccl, was added to normal synaptic transmission.Chemical psps were detectedin B 19(n = saline(final concentration, 1 mM) to block CaZ+currents. Figure 8 of 10 cells examined; Fig. 7) as early as 12 min after contact 6 demonstratesthat nominally calcium-free salineblocks activ- (earliest time of recording). To make a more appropriate de- ity-dependent transmission,and the addition of Cd2+ to the termination of the time of contact required until detection of bathing saline blocks desynchronizedtransmission in response functional transmission,additional neurons B5 and B19 were to presynaptic DC depolarization. separately impaled with a microelectrode and micromanipu- Taken together, the experiments described above demon- lated to make contact (n = 3) while intracellular records were strate that, in the absenceof neurites, neuronalsomata are ca- beingtaken. In this manner,the earliestevoked psp wasdetected pable of forming functional chemical synaptic connections. 10 set following contact. By contrast, electrical connectionsdid Therefore, in order to addressquestions concerning the for- not form either when cells contacted while intracellular records mation of synaptic connections,neuronal somataact as an ex- were being taken (n = 3) or when they had contacted for up to perimentally accessiblemodel for developing synaptic termi- 4 hr before recordingswere taken (n = 10; Fig. 7). nals. The functional chemical connection had similar propertiesto the sameconnection that forms under conditions of maintained Do the fundamental features of the synapserequire maintained contact: spontaneous,evoked (Fig. 7) and desynchronizedre- contact betweenB5 and B19 for their development?To address leaseof neurotransmitter were detected shortly after contact. this question, neuronsB5 and B 19 were isolated from the ner- Additionally, the evoked pspsfollowed presynaptic action po- vous systemand immediately placedinto separateculture dishes tentials with a short latency, as is characteristic of cell pairs (2 ml 1% hemolymph). Then, after a period of 3 d, which cor- cultured under conditions of prolonged contact. This is not to respondsto the latent period for synaptogenesis(Fig. 3), either say, however, that prolonged contact has no effect on the de- B5 or B 19 wastransferred to the partner neuron’s culture dish velopment of the presynaptic transmitter releasecapabilities. so that contact wasmade between cell pairs. Immediately upon Neurons that contact immediately before recordingsare made contact, cells strongly and irreversibly adheredto one another, werecapable only of the intermittent releaseof transmitter. That permitting the subsequenttransfer of the cell pair into a standard is, pspsdetected in B 19 did not follow presynaptic action po- polylysine-coated recording chamber. Following transfer, cells tentials one for one. 1036 Haydon - Chemical Synapses Between Neuronal Somata

a OCa

BlQ _/J r-

Figure 6. Transmitter release from the soma of neuron B5 is Caz+ dependent. a, Burst of action potentials in B5 evoked multiple psps in B19 when bathed in standard snail saline (left). However, exchanging the saline to a 0 Ca”+, high-Mg2+ saline rapidly abol- b ished synaptic transmission (right). b, 1mM Cd DC depolarization of B5 evoked a maintained release of transmitter in the absence of presynaptic action potentials when bathed in standard snail saline (left). Note that 4 of these psps evoked action potentials in B 19 that are clipped by the pen-recorder. Addition of 1 rnr.4 Cd2+ to this standard saline rapidly abolished all transmitter release from the soma of B5. The residual DC coupling between B5 and B19 was due to the presence ofan electrical connection.

Discussion There is a striking similarity between the properties of the so- 0 3 days matic synapse that forms in cell culture and connections that * O-4h are normally present in vivo at synaptic terminals. For example, synaptic transmission is CaZ+ -dependent. Transmitter can be 0 released spontaneously. The DC depolarization of the presynaptic soma increases desynchronized transmitter release. Action potentials evoke transmitter release. The action of the neurotransmitter is blocked by a postsynaptic antagonist. Given our knowledge of synaptic function, these properties are to be ex- pected. However, the ability to control the formation of such chemical synapses at this isolated, experimentally accessible site in a predictable way readily permits the study ofthe fundamental mechanisms of synaptic transmission that have, to date, not been possible in more intact systems. In addition to questions concerning the mechanisms of function of a mature synapse are questions concerning the control of synaptic development.Toward this goal, much experimental attention has focused on elucidating the control of the spatial arrangement of postsynaptic AChRs (for review, see Schuetze and Role, 1987). The transformation of the growth cone into the presynaptic terminal has been less intensively studied. How- Figure 7. Maintained contact between pre- and postsynaptic neurons is not required for the development of transmitter release capabilities. ever, it is known that growth cones of developing neurons can Neurons B5 and B 19 were plated separately into culture. After a period release neurotransmitter prior to contact by postsynaptic targets of 3 d, contact was induced between cell pairs and the ability to release (Hume et al., 1983; Young and Poo, 1983) and subsequently, neurotransmitter from the adult soma of B5 tested. Left, Neurotrans- contact with the muscle target induces further release of neu- mitter release was reliably detected using B 19 as a bioassay, and action rotransmitter within the early minutes of contact (Xie and Poo, potentials in B5 could evoke postsynaptic responses. Right,By contrast, the injection of DC current into either B5 of B19 failed to evoke a 1986). Using somata it has now been possible to determine synchronous potential change in the partner neuron, indicating that whether similar properties exist for regenerating neurons. These detectable electrical connections do not form within 4 hr of contact. results demonstrate that when maintained for 3 d following (Different neuron pairs used in each set of traces.) The Journal of Neuroscience, March 1988, 8(3) 1037 axotomy, the normal period for synaptogenesis in cell culture, nature of the signals and the mechanisms involved in controlling adult neurons have similar capabilities to developing neurons. the appropriate spatial location of synaptic terminals. A priori Within the first minutes of contact between pre- and postsyn- it might be reasonable to assume that the specific location of a aptic neurons, transmitter release is readily detected. Thus, just synapse is not determined solely by presynaptic intrinsic signals, as in development, maintained contact between future pre- and but rather by a balance of signals arising from pre- and post- postsynaptic targets is not required for the development of the synaptic cells upon contact. In this way, the site of contact initial transmitter release capability. determines the synaptic site, rather than the postsynaptic cell A surprising observation in our experiments was the vari- having to seek a predetermined position of presynaptic mem- ability of synaptic delay. Since action potentials were evoked at brane. This is certainly the strategy utilized for the development the site of transmitter release, we would expect a short synaptic of the postsynaptic membrane of the neuromuscular junction, delay. In the squid giant synapse, transmitter release occurs after where the presynaptic growth cone induces the localized aggre- a delay of approximately 200 psec following the influx of cal- gation of postsynaptic AChRs at the site of contact. Consonant cium. In our experiments, however, latencies ranged from 400 with this theme, Chow and Poo (1985) demonstrated that the wsec to 9.3 msec. While several interpretations are possible, the ability to release neurotransmitter is not spatially restricted. variability and long delay of some preparations may result from Using a myoblast to detect neurotransmitter, they found that the fact that these cells were at different stages in development. growth cone, neurite, and soma can release neurotransmitter. In vivo neurons B5 and B19 are not synaptically intercon- Furthermore, the demonstration that neuronal somata of Heli- nected, but, in response to axotomy, this neuron pair is able to soma reliably form chemical synapses lends support to the no- form novel chemical and electrical connections. As demonstrat- tion that all neuronal regions are competent to form synaptic ed previously, neurons B5 and B 19 will form an electrical con- connections. However, if all regions are competent, why do nection in situ (Hadley and Kater, 1983) and in cell culture somata rarely become presynaptic elements of a synapse? Sev- (Haydon et al., 1984, 1987). These previous studies did not eral mechanisms could exert spatial regulation over synapto- report the formation of a novel chemical connection. However, genesis. Signals that control the direction of neurite extension these in situ experiments utilized high-Mg2+ saline to block the and glial elements that encapsulate neuronal somata of Heli- release of neurotransmitter, so that a higher-resolution analysis soma (Berdan and Bulloch, 1986) would reduce the incidence of electrical connections could be performed. When using Ca2+- of neurite-soma contact. Alternatively, while all regions are able containing saline we have now determined that a novel chemical to form a synapse, separate regions of a single neuron may connection can form between B5 and B19 in situ (P. Matthews compete for the ability to form a functional release site. Such a and P. G. Haydon, unpublished observations). Additionally, scheme might be suggested by the demonstration that trans- these same neurons can form a novel chemical connection in mitter release from somata is prevented if the neurites of the cell culture under conditions where contact is made through same neuron have previously contacted a myoblast (Chow and newly extended neurites (P. G. Haydon and S. B. Kater, un- Poo, 1985). published communications). The chemical connection that forms When adult Helisoma neurons are excised and plated together between BS and B 19, when contact is made through neurites, as pairs, synaptic connections are first detected after a latent has the same properties as the somatically mediated connection period of 3 d. Because of the accessibility of this synapse, it reported here: The connection is unidirectional, B5 being pre- should be possible to determine the developmental sequence of synaptic to B 19; psps are blocked by tubocurare; and B5 evokes events underlying this latent period. Application of ACh to ipsps in B19 that are reversed in sign by penetration of B19 neuron B 19 (days O-3) demonstrated that postsynaptic receptors with a KCl-filled electrode. Thus, while certain neurons re-form are present throughout the early stages of contact prior to the only appropriate chemical synapses in cell culture (Fuchs et al., detection of functional synaptic transmission. Such lines of evi- 198 1, 1982; Camardo et al., 1983), the formation of a novel dence point toward a presynaptic locus for the rate-limiting B5-B 19 chemical synapse is an additional example of a neuron step(s) in the development of functional transmission. Mature that forms connections not normally found in the nervous sys- synaptic transmission consists of a cascade of events triggered tem (Bodmer et al., 1984; Schacher et al., 1985). by membrane depolarization. Since the synaptic terminal was The focus of much of the previous work on neurons B5 and experimentally depolarized in the first days following contact B19 of Helisoma has been on the conditions required for the and transmitter release was not detected, these data point toward formation ofelectrical connections. Those studies demonstrated at least 1 of 3 loci for the rate-limiting step in synaptic devel- that electrical connections form under conditions of synchro- opment: the appearance of a CaZ+ conductance, the synthesis nous neurite extension (Hadley et al., 1983, 1985). If one cell of neurotransmitter, or the development of coupling between a of a pair has ceased extension, electrical connections do not calcium influx and transmitter release. Since we are able to form. Since neuronal somata do not extend neurites, it might voltage-clamp the presynaptic soma directly and to inject mol- be surprising that electrical connections did form in 37% of our ecules presynaptically, this system should permit the elucidation cell pairs. However, since previous studies examined when an of the development of each of these critical components to func- isolated neuron loses its competence to form an electrical con- tional transmission. nection, it is not known when a neuron gains the ability to form this type of connection-before neurite extension commences References or thereafter. Our data suggest that somata do have a limited Arechiga, H., M. Chiquet, D. P. Kuffler, and J. G. Nicholls (1986) capacity to form electrical connections; however, electrical con- Formation of specific connections in culture by identified leech neu- nections form reliably only under conditions of active neurite rones containing serotonin, acetylcholine and peptide transmitters. J. Exp. Biol. 126: 15-3 1. extension. Auerbach, A. A., and M. V. L. Bennett (1969) Chemically mediated Synaptic terminals in vivo are located in specific spatial lo- transmission at a giant fiber synapse in the central nervous system of cations of a given neuron. A fundamental question concerns the a vertebrate. J. Gen. Physiol. 53: 183-210. 1038 Haydon - Chemical Synapses Between Neuronal Somata

Belardetti, F., S. Schacher, E. R. Kandel, and S. Siegelbaum (1986a) Haydon, P. G., D. P. McCobb, and S. B. Kater (1987) The regulation The growth cones of Aplysiu neurons: Modulation by serotonin of of neurite outgrowth, growth cone motility, and electrical synapto- action potential duration and single potassium channel currents. Proc. genesis by serotonin. J. Neurobiol. 18: 197-2 15. Natl. Acad. Sci. USA 83: 7094-7098. Henderson, L. P. (1983) The role of 5-hydroxytryptamine as a trans- Belardetti, F., S. Schacher, and S. Siegelbaum (1986b) Action poten- mitter between identified leech neurones in culture. J. Physiol. (Lond.) tials, macroscopic and single channel currents recorded from growth 339: 309-324. cones of Aplysia neurones in culture. J. Physiol. (Lond.) 374: 289- Henderson, L. P., D. P. Kuffler, J. G. Nicholls, and R.-J. Zhang (1983) 313. Structural and functional analysis of synaptic transmission between Berdan, R. C., and A. G. M. Bulloch (1986) Gap junction specificity identified leech neurones in culture. J. Phvsiol. (Land.) 340: 347-358. in Helisoma ganglia. Neurosci. Abstr. 12: 539. Hume, R. I., L. W. Role, and G. D. Fischbach (1983) Acetylcholine Bodmer, R., D. Dagan, and I. B. Levitan (1984) Chemical and elec- release from growth cones detected with patches of acetylcholine re- trotonic connections between Aplysiu neurons in primary culture. J. ceptor-rich membranes. Nature 305: 632-634. Neurosci. 4: 228-233. Johnson, D. A., and G. Pilar (1980) The release of acetylcholine from Camardo, J., E. Proshansky, and S. Schacher (1983) Identified Aplysia postganglionic cell bodies in response to depolarization. J. Physiol. neurons form specific chemical synapses in culture. J. Neurosci. 3: (Lond.) 299: 605-619. 2614-2620. Kostyuk, P. G. (1980) Calcium ionic channels in electrically excitable Chow, I., and M.-M. Poo (1985) Release of acetylcholine from em- membrane. Neuroscience 5: 945-959. bryonic neurons upon contact with muscle cell. J. Neurosci. 5: 1076- Lemos. J. R.. .I. J. Nordmann. I. M. Cooke. and E. L. Stuenkel (1986) 1082. Single channels and ionic currents in pkptidergic nerve terminals. Cohan, C. S., P. G. Haydon, and S. B. Kater (1985) Single channel Nature 319: 410-413. activity differs in growing and non-growing growth cones of isolated Llinas, R., I. Z. Steinberg, and K. Walton (198 1a) Presynaptic calcium identified neurons of Helisoma. J. Neurosci. Res. 13: 285-300. currents in squid giant synapse. Biophys. J. 33: 289-322. Cohan, C. S., P. G. Haydon, A. J. Mercier, and S. B. Kater (1987) Llinas, R., I. Z. Steinberg, and K. Walton (1981 b) Relationship be- Formation, maintenance, and functional uncoupling of connections tween presynaptic calcium current and postsynaptic potential in squid between identified Helisoma neurons in situ. J. Neurobiol. 18: 329- giant synapse. Biophys. J. 33: 323-352. 341. Martin, A. R., and G. L. Ringham (1975) Synaptic transfer at a ver- Dietzel, I. D., P. Drapeau, and J. G. Nicholls (1986) Voltage depen- tebrate central nervous system synapse. J. Physiol. (Lond.) 251: 409- dence of 5-hydroxytryptamine release at a synapse between identified 426. leech neurones in culture. J. Physiol. (Lond.) 372: 19 l-205. Pellegrino, M., and M. Simonneau (1984) Distribution of receptors Fuchs, P. A., J. G. Nicholls, and D. F. Ready (1981) Membrane for acetylcholine and 5-hydroxytryptamine on identified leech neu- properties and selective connexions of identified leech neurones in rones growing in culture. J. Physiol. (Lond.) 352: 669-684. culture. J. Physiol. (Lond.) 316: 203-223. Schacher, S., S. G. Rayport, and R. T. Ambron (1985) Giant Aplysiu Fuchs, P. A., L. P. Henderson, and J. G. Nicholls (1982) Chemical neuron R2 reliably forms strong chemical connections in vitro. J. transmission between individual retzius and sensory neurones of the Neurosci. 5: 285 l-2856. leech in culture. J. Physiol. (Lond.) 323: 195-210. Schuetze, S. M., and L. W. Role (1987) Developmental regulation of Hadley, R. D., and S. B. Kater (1983) Competence to form electrical nicotinic acetylcholine receptors. Annu. Rev. Neurosci. 10: 403-457. connections is restricted to growing neurites in the snail, Helisoma. Wojtowicz, J. M., and H. L. Atwood (1984) Presynaptic membrane J. Neurosci. 3: 924-932. potential and transmitter release at the crayfish neuromuscular junc- Hadley, R. D., S. B. Kater, and C. S. Cohan (1983) Electrical synapse tion. J. Neurophysiol. 52: 99-l 13. formation depends on interaction of mutually growing neurites. Sci- Wong, R. G., R. D. Hadley, S. B. Kater, and G. C. Hauser (198 1) ence 221: 466-468. Neurite outgrowth in molluscan organ and cell cultures: The role of Hadley, R. D., D. A. Bodnar, and S. B. Kater (1985) Formation of conditioning factor(s). J. Neurosci. 1: 1008-1021. electrical synapses between isolated, cell cultured Helisoma neurons Xie, Z. P., and M.-M. Poo (1986) Initial events in the formation of requires mutual neurite elongation. J. Neurosci. 5: 3 145-3 153. neuromuscular synapse: Rapid induction ofacetylcholine release from Haydon, P. G., D. P. McCobb, and S. B. Kater (1984) Serotonin embryonic neuron. Proc. Natl. Acad. Sci. USA 83: 7069-7073. selectively inhibits arowth cone motilitv and svnantoaenesis of snecific Youna. S. H.. and M.-M. Poo (1983) Spontaneous release of neuro- identified-neurons.-Science 226: 561-564. - ^ - transmitter’ from growth cones of embryonic neurones. Nature 305: Haydon, P. G., C. S. Cohan, D. P. McCobb, H. R. Miller, and S. B. 634-637. Kater (1985) Neuron-specific growth cone properties as seen in identified neurons of Helisoma. J. Neurosci. Res. 13: 285-300.