Ultrastructure of electrophysiologically-characterized synapses formed by serotonergic raphe neurons in culture

Neuroscience Vol. 67, No. 3, pp. 609-623, 1995

~

Pergamon

0306-4522(95)00010-0

Elsevier Science Ltd Copyright © 1995 IBRO Printed in Great Britain. All rights reserved 0306-4522/95 $9.50 + 0.00

ULTRASTRUCTURE OF ELECTROPHYSIOLOGICALLY-CHARACTERIZED SYNAPSES FORMED BY SEROTONERGIC RAPHE N E U R O N S IN CULTURE M. D. J O H N S O N * and A. G. Y E E Department of Neurobiology, Harvard Medical School, Boston, MA 02115, U.S.A. Almtract--Recent electrophysiological investigations in this laboratory have shown that cultured mesopontine serotonergic neurons from neonatal rats evoke serotonergic and/or glutamatergic responses in themselves and in non-serotonergic neuronsJ 7 Serotonergic nerve terminals in vivo are heterogeneous with respect to vesicle type, synaptic structure, and the frequency with which they form conventional synaptic contacts, but the functional correlates of this heterogeneity are unclear. We have therefore examined the ultrastructure of electrophysiologically-characterized synapses formed by cultured serotonergic neurons, and have compared the findings with the ultrastructural characteristics of serotonergic synapses reported in vivo. Dissociated rat serotonergic neurons in microcultures were identified by serotonin immunocytochemistry or by uptake of the autofluorescent serotonin analogue 5,7-dihydroxytryptamine, and were subsequently processed for electron microscopy. Unlabeled axon terminals formed numerous synapses on serotonin-immunoreactive somata and dendrites. Serotonin-immunoreactive axon terminals formed synapses on the somata, dendrites and somatodendritic spine-like appendages of serotonergic and non-serotonergic neurons. In microcultures containing a solitary serotonergic neuron that evoked glutamatergic or serotonergic/glutamatergic autaptic responses, both symmetric and asymmetric synapses were present. In addition to large dense core vesicles, individual neurons contained either microcanaliculi and microvesicles, clear round vesicles, or clear pleiomorphic vesicles. For a given cell, however, the subtypes of vesicles present in each axon terminal were similar. Thus, dissociated serotonergic and non-serotonergic raphe neurons formed functional, morphological synapses in culture. A direct examination of both the synaptic physiology and ultrastructure of single cultured serotonergic neurons indicated that these cells released serotonin and glutamate at synapses that were morphologically similar to synapses formed by serotonergic neurons in vivo. The findings also suggested that individual serotonergic neurons differ with respect to synaptic vesicle morphology, and are capable of simultaneously forming symmetric and asymmetric synapses with target cells.

Ultrastructural studies employing 5-HT immunocytochemistry or uptake of 3H-serotonin (5-HT) have indicated that m a m m a l i a n 5-HT axon terminals in vivo exhibit a range of morphological characteristics. Large dense core vesicles have been localized in most types of 5-HT axon terminals, and may contain either 5-HT or neuropeptides. In addition, clear round vesicles, pleiomorphic vesicles, or tubular elements and microcanaliculi have been identified in separate populations of serotonergic varicosities. 4,21,28

*To whom correspondence should be addressed. AHP, afterhyperpolarization; AMPA, - amino - 3 - hydroxy- 5 - methyl - 4 - isoxazolopropionate; D(-)-2-amino-5-phosphonovaleric acid; BSA, bovine serum albumin; CNQX, 6-eyano-7-nitroquinoxaline-2,3-dione; 5,7-DHT, 5,7-dihydroxytryptamine; 5-HT, 5-hydroxytryptamine; 8-OH-DPAT, (+)-8hydroxy-2-(di-N-propylamino)tetralin HBr; HEPES, N-2-hydroxyethylpiperazine-N'-2-ethanesulphonic acid; NMDA, N-methyl-D-aspartate; PBS, phosphatebuffered saline; TH, tyrosine hydroxylase; TrpH, tryptophan hydroxylase.

Abbreviations:

APV,

Serotonergic axon terminals also form several types of contacts with target cells in vivo. Both symmetric and asymmetric synapses with postsynaptic neurons have been reported. Additionally, 5-HT varicosities frequently form "non-junctional" contacts that lack the membrane specializations normally associated with conventional synapses. 33 A number of investigators have suggested that asymmetric (Gray Type I) synapses are excitatory and symmetric (Gray Type II) synapses are inhibitory, although exceptions to this hypothesis have been reported. 27 It is therefore natural to wonder whether 5-HT axon terminals forming asymmetric synapses are functionally different from those forming symmetric synapses. The incidence of symmetric, asymmetric, or nonjunctional contacts varies from one brain region to another. It is not known whether this morphological heterogeneity derives from the intrinsic properties of different subpopulations of 5-HT neurons, or whether single 5-HT neurons produce axon terminals and synapses with different morphological characteristics. 609

610

M . D . Johnson and A. G. Yee

The heterogeneity o f 5-HT axon terminals extends to the range o f co-transmitters that are present within them. Cytochemical studies have indicated that glutamate, G A B A and neuropeptides are colocalized with 5-HT in rat raphe neurons. 3'5A4'23 Moreover, extracellular stimulation o f the raphe nuclei can evoke a range o f responses at postsynaptic sites, only some o f which are mediated by 5-HT. 11'26 Because n o n - 5 - H T neurons were present at the sites o f stimulation, and because o f the morphological and cytochemical heterogeneity o f 5-HT terminals and synapses in slices and in vivo, it has been difficult to determine the functional role o f the subtypes o f 5-HT terminals and synapses in these responses. In this regard, a correlation o f the ultrastructural characteristics and synaptic function o f single 5-HT neurons is desirable. Recent studies in this laboratory have d e m o n strated that dissociated m e s o p o n t i n e 5-HT neurons from neonatal rats evoked glutamatergic a n d / o r serotonergic responses in themselves and in other neurons when maintained in microcultures for several weeks.17 The microculture format allows the soma, dendrites, axon and terminals o f an electrophysiologicallycharacterized neuron to be sectioned simultaneously, thereby facilitating a comprehensive light microscopic and ultrastructural survey o f the entire functional neuronal unit. ~2 Using this approach, we have investigated the ultrastructure o f synapses formed a m o n g postnatal rat raphe neurons in microcultures. EXPERIMENTAL

PROCEDURES

Preparation of microcultures The preparation of microcultures was as described previously/ 2'~9'3~ Briefly, glass cover slips were coated with a solution of 0.15% agarose (Sigma type IIA) and allowed to air dry. The. coverslips were then attached with Sylgard (Dow Corning Corp.) to the bottom of a 35 mm culture dish containing a central welP2 and cured overnight at 45°C. Droplets (200-500 #m in diameter) of a solution consisting of 0.75 mg/ml of human placental collagen (Sigma) in 0.125% acetic acid were then applied in a regular array using a glass micropipette. The droplets were allowed to air dry, and the dishes were then sterilized using ultraviolet irradiation. Cell culture Neonatal Long-Evans rats (P0-P3, Charles River) were decapitated, the brains were removed, and a 1 mm thick coronal section just caudal ~to the mesencephalic-pontine junction was obtained using two juxtaposed single-edged razor blades. A scalpel was then used to dissect from the medial region immediately ventral to the Sylvian aqueduct a diamond-shaped area of tissue (approximately 1.5 x 1.5 mm) that included the dorsal and median raphe nuclei. The tissue was incubated for 40 min at 37°C in a medium consisting of 81.4mM Na2SO4, 30mM K2SO4, 0.25mM CaC12, 10mM MgCI2, 20.4mM glucose, 5mM HEPES (pH 7.4), 1 mM kynurenate, 0.1 mg/ml L-cysteine/ ml, and 200 units of papain. The tissue was then incubated for 3 min in Minimal Essential Medium (Sigma Chemical Co.) containing 3 mg/ml of bovine serum albumin (BSA) and 3 mg/ml of trypsin inhibitor. After washing in Minimal Essential Medium, the cells were dissociated by trituration

and plated onto collagen microcultures or on to microcultures containing a pre-plated mesopontine, hippocampal or cortical glial layer. The neurons were maintained at 37°C in a 4.9% CO: atmosphere for up to 10 weeks, and were fed every one to four weeks with a Liebovitz-15 bicarbonatebased medium (modified from Furshpan and Potter ~3) which lacked glutamate, aspartate and fl-alanine. The medium contained 25#g/ml insulin, 15nM selenium, 90 #M putrescine, 25 #g/ml transferrin, and 5% rat serum, and lacked progesterone and the glutamate receptor antagonists used by Furshpan and Potter. 13

Electrophysiology Electrophysiological recordings were performed as described, ~2,17except that cultures were perfused with Hanks Balanced Salt Solution to which were added 1.0 mM MgC12, 2.8 mM CaC12, 5 mM HEPES, 25 mM glucose, and 10% Minimal Essential Medium. The extracellular Na + , K ÷ and C1- concentrations were 139, 5.8 and 148 mM, respectively. The bath temperature was maintained at 33-35°C. Intracellular microelectrodes contained 2 M potassium acetate and had resistances of 100-300 MfL The duration of intracellular current pulses ranged from 3 to 15 ms, and the amplitude was the minimum required to bring impaled neurons to threshold. After recording, cultures were stained for 5,7-dihydroxytryptamine (5,7-DHT) uptake, 5-HT immunoreactivity, or tryptophan hydroxylase (TrpH) immunoreactivity to identify 5-HT neurons. Uptake of 5,7-dihydroxytryptamine Dissociated mesopontine 5-HT neurons were labeled with the autofluorescent 5-HT analogue, 5,7-DHT, according to the method of Silva et al. 32 In brief, cultures were incubated at 37°C for 15-30min in Minimal Essential Medium containing 50#M 5,7-DHT and 0.5 mM ascorbate. The cultures were then washed with Minimal Essential Medium, and labeled neurons were visualized under epifluorescence using a Zeiss filter set (G 365 nm, FT 395 nm, LP 420 nm). The cultures were then fixed for electron microscopy or immunocytochemistry. Irnmunocytochemistry For fluorescence immunocytochemistry, the cultures were fixed in 4% paraformaldehyde (5-HT, TrpH, and tyrosine hydroxylase staining) or 4% paraformaldehyde with 0.14).3% glutaraldehyde (GABA staining) in 0.1 M sodium phosphate buffer (pH 7.4) for 30 min. The cultures were then washed in phosphate-buffered saline (PBS) and incubated for 90 min in primary antisera diluted in a buffer composed of 0.5 M NaC1, 0.3% Triton X-100, and 10% blocking serum in 10mM sodium phosphate buffer (pH 7.4). Blocking sera consisted of normal goat serum for 5-HT, tyrosine hydroxylase (TH), and GABA immunocytochemistry, or normal donkey serum for TrpH staining. As primary antisera, a rabbit anti-5-HT/BSA antiserum (Incstar) was used at a 1:400 dilution, a mouse monoclonal anti-TH-antibody (Boerhinger Mannhiem) was used at a 1:100 dilution, a rabbit anti GABA/BSA antiserum (Incstar) was used at a l : 3000 dilution, and a sheep anti-TrpH antiserum (Chemicon) was used at a 1:250 dilution. In double-labeling experiments for 5-HT and GABA immunoreactivities, the mouse monoclonal anti5-HT antibody was used in combination with the rabbit anti-GABA/BSA antiserum. When cultures were stained for both TH and 5-HT immunoreactivities, the mouse monoclonal anti-TH antibody and the rabbit anti-5-HT/ BSA antiserum were used. After incubation in primary antisera, cultures were washed and incubated for 30 min in the appropriate fluorescent secondary antibody diluted 1 : 50. Secondary antisera were: a fluorescein-labeled donkey anti-sheep antiserum (Jackson Immunoresearch), a fluorescein labeled goat anti-mouse antiserum (Atlantic Antibodies), a fluorescein-labeled goat anti-rabbit antiserum

Synaptic structure and function of cultured 5-HT neurons (Atlantic Antibodies), or a rhodamine-labeled goat antimouse antiserum (Atlantic Antibodies). The stained cultures were examined under epifluorescence using a Zeiss inverted microscope equipped with either a fluorescein (exciter BP 450M90 nm, FT 510 rim, LP 515-565 rim) or a rhodamine (exciter BP 546 rim, FT 580 rim, LP 590 rim) filter set. Control cultures in which the primary antisera were omitted lacked specific staining. Control cultures of dissociated sympathetic neurons contained TH immunoreactivity, but lacked immunoreactivity for GABA and 5-HT. Hippocampal cultures contained GABA immunoreactivity, but lacked immunoreactivity for TrpH, TH and 5-HT. All neurons that contained TrpH immunoreactivity accumulated 5,7-DHT. Previous work indicated that preadsorption of the 5-HT/BSA antiserum eliminated specific staining. 29 Moreover, this antiserum specifically stained the known serotonergic nuclei in 40/~m thick frozen sections of rat brain. TM

Electron microscopy For electron microscopy, avidin-biotin immunoperoxidase staining for 5-HT was used to identify 5-HTimmunoreactive neurons. Microcultures were fixed for 30 min in 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M sodium phosphate buffer (pH 7.4). Cultures were washed in PBS (pH 7.4) and incubated for 2 h at room temperature in PBS containing 10% normal goat serum, 0.05% Triton X-100, and a 1:400 dilution of rabbit anti5-HT/BSA antiserum (Incstar). The cultures were then washed and incubated for 1 h at room temperature in biotinylated goat anti-rabbit secondary antibody (Vector, 5/~l/ml) in the buffer described above. The cultures were washed in 50 mM Tris-buffered saline (pH 7.4) and incubated for 30 min in ABC reagent (Vector). The staining was then visualized by incubating the cultures for 1-3 min in Tris-buffered saline containing 0.015% hydrogen peroxide and 0.5mg/ml of diaminobenzidine. The cultures were washed, and the positions of labeled and unlabeled neurons in the microcultures were mapped in detail for subsequent relocalization. The cells were then fixed for an additional 20min in 3% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4. As an alternative to the avidin-biotin immunoperoxidase method, some 5-HT neurons in microcultures were identified by 5,7-DHT uptake. The position of labeled neurons in each microculture was mapped, and the cultures were fixed in 2.5% glutaraldehyde in 0.12M Sorensen's phosphate buffer (pH 7.4) for 30 min. Both, 5,7-DHT-labeled and 5-HT-immunolabeled cultures were postfixed for 1 h at room temperature in 2% osmium tetroxide containing 1.5% potassium ferrocyanide. The microcultures were then stained en bloc in 1% uranyl acetate in sodium maleate buffer, pH 6.0. After dehydration in a graded series of ethanol solutions, the cultures were embedded in epon/araldite. Microcultures containing previously-identified 5,7-DHT-labeled or 5-HT-immunoreactive neurons were relocated, and silver-grey thin sections were cut and examined in a JEOL 100CX transmission electron microscope.

RESULTS Initial studies examined the function and ultrastructure of contacts formed between dissociated 5-HT and non-5-HT neurons in multineuron cultures. Further studies employing single neuron microcultures were used to correlate the synaptic function and ultrastructure of solitary 5-HT neurons. Approximately 60 cultures of dissociated raphe

611

neurons were stained for 5-HT, G A B A , TrpH, or T H immunoreactivity. Electrophysiological recordings were obtained from 132 5-HT neurons and 48 non-5-HT neurons in 115 microcultures. Seven microcultures containing identified 5-HT neurons were selected for ultrastructural analysis.

Appositions between serotonergic and non-serotonergic neurons When cultures were simultaneously labeled for both 5-HT and T H immunoreactivities, no doublylabeled neurons were observed. However, 5-HTimmunoreactive axons were frequently observed in close apposition to the somata and dendrites of TH-immunoreactive neurons (Fig. IA~2). These points of contact were not limited to proximal somatodendritic regions, but extended along distal segments of TH-immunoreactive dendrites. THimmunoreactive axons were only faintly stained and, consequently, it was not possible to determine whether they also contacted 5-HT-immunoreactive neurons. Double-labeling for G A B A and 5-HT indicated that 5-HT-immunoreactive axons were frequently apposed to GABA-immunoreactive somata and dendrites (data not shown). Similarly, G A B A immunoreactive axons were frequently observed in close apposit!on to 5-HT-immunoreactive or 5,7-DHT-labeled neurons (Fig. I D - F ) . The G A B A immunoreactive fibers contained numerous varicosities that were observed primarily near the somata and proximal dendrites of 5-HT neurons. In some cases, GABA-immunoreactive axons enveloped 5-HT neurons.

Ultrastructural characteristics of cultured serotonergic neurons The neuronal somata and proximal dendrites of cultured raphe neurons were generally located superficial to a flattened glial cell layer, but the distal dendrites and axons were often interwoven between neuronal and glial elements (Fig. 2). Synaptic specializations were not observed between neurons and glia, although adhering junctions between these • two cell types were sometimes present. The nuclei of 5-HT neurons had multiple indentations and were often large and eccentrically located. Large numbers of free ribosomes, rough endoplasmic reticulum, mitochondria, lysosomes and a well developed Golgi complex were observed in the perikaryal cytoplasm, suggesting that the neurons were highly metabolically active. Occasional large dense core vesicles were also present in the cytoplasm of 5-HT neuronal perikarya. Somatodendritic spine-like appendages were consistently observed on cultured 5-HT neurons, and numerous vesicle-filled axons were closely apposed to the somata and dendrites of these cells. Although most of these axons were located on the undersurface of neurons or between neuronal and glial elements, some were observed on

612

M . D . Johnson and A. G. Yee

I

t

+

i

Fig. 1. Light microscopic observation of close appositions between 5-HT and non-5-HT neuronal elements in culture. One of the neurons shown in the phase micrograph in A (arrow) contained immunoreactivity for TH after 20 days in culture, as shown in the fluorescence micrograph in B. In the fluorescence micrograph in C, 5-HT-immunoreactive axons were seen in close apposition to the soma and dendrites of the TH-immunoreactive neuron. Numerous intensely stained, 5-HT-immunoreactive varicosities (small arrows) were observed in the somatodendritic region of the TH-immunoreactive neuron. Two of the raphe neurons shown in the phase micrograph in D (short arrows) accumulated 5,7-DHT as shown in E. Other neurons (long arrows in D) were not labeled with 5,7-DHT, but did contain GABA immunoreactivity in F. GABA-immunoreactive axons from these neurons formed a basket-like array around one of the 5,7-DHT-labeled cells. Scale bar = 50 #m.

the upper surfaces of neuronal somata and proximal dendrites.

Function and ultrastructure of synapses formed by non-serotonergic neurons The electrophysiological properties of cultured raphe neurons have been reported in detail elsewhere. ~7 Because the present study correlates the morphology and function of synapses formed by these cells, recordings of synaptic function are pre-

sented here along with the morphological data. Neurons that contained 5-HT immunoreactivity, T r p H immunoreactivity, or 5,7-DHT uptake were considered to be serotonergic. Intracellular stimulation of non-5-HT raphe neurons frequently evoked fast depolarizing or hyperpolarizing potentials in 5-HT or non-5-HT neurons. The fast inhibitory responses were blocked reversibly by the GABAA receptor antagonist, bicuculline ( 5 0 g M ) , indicating that they were mediated by

Synaptic structure and function of cultured 5-HT neurons

613

Fig. 2. Electron micrograph of a 5,7-DHT-labeled raphe neuron maintained in microculture for 21 days. The microculture also contained four unlabeled, non-5-HT neurons. The soma of the 5-HT neuron rested on the upper surface of a flattened glial layer (G1). The neuronal cytoplasm contained a well developed Golgi complex (Go) and numerous mitochondria, secondary lysosomes, endoplasmic reticulum and free ribosomes. Numerous axon terminals containing clear round or large dense cored synaptic vesicles formed synapses on the upper and lower surfaces of the soma and dendrites of the cell (arrows). One of the synapses (double arrow) is shown at higher magnification in the inset. Scale bar = 1/~m.

activation of GABA A receptors (Fig. 3A). The depolarizing potentials evoked by non-5-HT neurons were antagonized reversibly by glutamate receptor antagonists such as o(-)-2-amino-5-phosphonovaleric acid (APV, 50 p M), 6-cyano-7-nitroquinoxaline2,3-dione (CNQX, 20 #M), and kynurenate (1 mM; Fig. 3B). The rapid time course and pharmacology of these responses indicated that they were mediated by synaptic activation of the N M D A and kainate/ ~-amino- 3-hydroxy- 5-methyl-4-isoxazolopropionate (AMPA) subtypes of glutamate receptors. At the ultrastructural level, non-5-HT raphe neurons formed functional GABAergic and glutamatergic contacts on other non-5-HT neurons. Nonimmunoreactive axon terminals containing clear round or pteiomorphic vesicles formed synapses on non-immunoreactive somatic and dendritic profiles (Fig. 4A). Numerous vesicles were accumulated near the active zones, and electron dense material was frequently present within synaptic clefts. Large dense core vesicles were only occasionally observed in unlabeled axon terminals, but other cytoplasmic structures such as mitochondria were often

present. Axo-axonic specializations between nonimmunoreactive elements were not observed. Non-immunoreactive axon terminals also formed symmetric and asymmetric synapses on 5-HTimmunoreactive somata and dendrites, and were frequently observed on or near somatodendritic spine-like appendages (Fig. 4B). However, we could not exclude the possibility that, in some cases, apparent postsynaptic densities resulted from peroxidase reaction product that had accumulated beneath the postsynaptic membrane. Apposed 5-HT-immunoreactive and non-immunoreactive axons did not form membrane specializations.

Function and ultrastructure of synapses formed by serotonergic neurons In a previous study, 17 we reported that 5-HT neurons occasionally evoked slow inhibitory potentials in themselves or in target neurons that were blocked by 5-HT~A receptor antagonists such as (-)propranolol ( l # M ) or spiperone ( l - 1 0 p M ; Fig. 5A). The time course and pharmacology of the slow inhibitory potentials resembled the 5-HT~A

614

M.D. Johnson and A. G. Yee

A N

N2

BIC

CONT I

~

N 1-'---'-~

~""~

B

KYN

CONT

N,-A

N2~

NI----

t 0.5s

Fig. 3. Non-5-HT neurons evoked GABA- and glutamatemediated potentials in 5-HT neurons. (A) Intracellular recordings from two raphe neurons that were maintained in microculture for 27 days. Neuron Nl, but not neuron N2, accumulated 5,7-DHT. The resting potentials in N1 and N2 were - 5 0 and -54mV, respectively. In control medium (CONT), an action potential in the non-5-HT neuron, N2, evoked a fast hyperpolarization in the 5-HT neuron, NI. The GABAA receptor antagonist, bicuculline (BIC, 50/~M), blocked this hyperpolarization. Note that bicuculline also blocked the postspike AHP in N2, revealing the presence of an autaptic GABAA receptor-mediated component in this cell. The effect of bicuculline was reversed in control medium (not shown). (B) Intracellular recordings from two raphe neurons that had been maintained in a microculture for 63 days. The resting potentials in N~ and N 2 w e r e --63 and -58 mV, respectively. Neuron NI, but not neuron N2, contained 5-HT immunoreactivity. Intracellular stimulation of N2 in control medium (CONT) evoked a short latency depolarization in N1. The evoked excitatory postsynaptic potential was blocked by the glutamate receptor antagonist, kynurenate (KYN, 1 mM). The effect of kynurenate was reversed in control medium (not shown).

receptor-mediated inhibitory responses observed in slices and in vivo fl4'38 Over half of cultured 5-HT neurons formed functional glutamatergic contacts on themselves or on target neurons. 17 These short latency excitatory responses were blocked by APV, CNQX and kynurenate, but were unaffected by 5-HTIA, 5-HTjB, 5-HT~c, or 5-HT2 receptor antagonists. The failure of 5-HT receptor antagonists to affect many of the observed synaptic responses may have been due to the scarcity of 5-HT receptor-mediated responses among cultured 5-HT neuronsJ 7 About 100% of cultured 5-HT neurons evoked biphasic fast excitatory/slow inhibitory responses in themselves or in other cellsJ 7 5-HT neurons formed

synapses on non-5-HT neurons that were morphologically similar to 5-HT autapses. Whereas glutamate receptor antagonists reversibly blocked the fast excitation, 5-HT~A receptor antagonists blocked the slow inhibition (Fig. 5B). These data indicated that some 5-HT neurons formed dual serotonergic/ glutamatergic contacts with themselves or with other neurons. Electron microscopic analysis indicated that 5-HTimmunoreacfive neurons formed morphological synapses on themselves. Labeled axon terminals containing clear round vesicles and large dense core vesicles formed either symmetric or asymmetric synapses on the somata and dendrites of 5-HTimmunoreactive neurons (Fig. 5C). Large numbers of synaptic vesicles were frequently accumulated within presynaptic active zones. As with synapses formed by unlabeled nerve terminals, these synapses were sometimes located on or near somatodendritic spine-like appendages. No evidence for 5-HTimmunoreactive dendrites containing synaptic vesicles or forming membrane specializations with other dendrites was observed. Similarly, no axoaxonic synapses between 5-HT-immunoreactive axon terminals were seen.

Function and morphology of synapses formed by single serotonergic neurons In a related series of experiments, we investigated the feasibility of directly assessing the synaptic function and ultrastructural properties of a single 5-HT neuron. Combined electrophysiological and ultrastructural analyses were obtained from three 5,7DHT-labeled neurons in single neuron microcultures; data from two such neurons are presented below. As shown in Fig. 6, a brief intracellular current pulse in one of these neurons evoked an action potential, which was followed by a burst of excitation. ( - ) P r o p r a n o l o l (1 pM) failed to significantly alter the response, indicating that 5-HT~A/5-HTIB/ fl-adrenergic receptors were not responsible for the observed effect. However, the N M D A receptor antagonist, APV (50 p M), significantly reduced the excitation, and the kainate/AMPA receptor antagonist, CNQX (20#M), completely and reversibly blocked the response. The electrophysiological findings thus indicated that this 5,7-DHT-labeled raphe neuron exerted a powerful glutamatergic effect on itself. It is possible that this neuron also released 5-HT, but lacked functional 5-HTtA receptors. Ultrastructural analysis revealed that the soma of this neuron rested on several flattened glial cells and contained numerous mitochondria, free ribosomes, rough endoplasmic reticulum, lysosomes, a well developed Golgi complex, and occasional large dense core vesicles (Fig. 7). Somatodendritic spinelike appendages were also observed. Axons with numerous varicosities were held in close contact with the undersurface of the neuronal soma by numerous adhering junctions (puncta adhaerentia, data not

Synaptic structure and function of cultured 5-HT neurons

615

Fig. 4. Non-5-HT axon terminals formed symmetric synapses on non-5-HT and 5-HT neurons. This microculture contained one 5-HT-immunoreactive neuron and one non-immunoreactive neuron which had been maintained in culture for 10 days. In A, two non-immunoreactive axon terminals (A I and A3) formed symmetric synapses with a non-immunoreactive dendrite (D). Another terminal (A2) formed a synapse with a non-immunoreactive spine-like process (S). Note the accumulation of clear round and pleiomorphic vesicles within the active zones. Another vesicle-filled profile (A4)contained dark peroxidase reaction product, indicating the presence of 5-HT immunoreactivity. Another thin section from the microculture in A is shown in B. A non-immunoreactive axon terminal (AI) formed a symmetric synapse on a 5-HT-immunoreactive dendrite (D). A 5-HT-immunoreactive axon terminal (A2) was observed in close apposition to the non-immunoreactive axon (A0; however, no membrane specialization was observed. Scale bar = 0.5 l~m.

shown). Vesicle-containing axon terminals formed primarily symmetric synapses with postsynaptic elements; however, asymmetric synapses were also observed (Fig. 8). These terminals c o n t a i n e d primarily microcanaliculi a n d microvesicles, a l t h o u g h large dense core vesicles were sometimes present. O n average, the dimensions o f the microcanaliculi were 25 x 70 nm, a n d the microvesicles were a b o u t 35 n m in diameter. The large dense core vesicles h a d a n average outer d i a m e t e r o f 9 5 n m , while

the central dense cores were approximately 7 0 n m in diameter. In some cases, contacts between vesicle-filled axon terminals a n d adjacent n e u r o n a l s o m a t a or dendrities did n o t exhibit m e m b r a n e specializations. Because serial sections were n o t made, however, it was not possible to determine w h e t h e r these terminals formed c o n v e n t i o n a l or n o n - j u n c t i o n a l synapses. Also, no evidence for dendro-dendritic or axo-axonic synapses were observed.

616

M, D. Johnson and A. G. Yee

® 0.3s SPIP

(~)

CONT

0.Ss

CNQX N1

-

.-~

N2..... L PROP+CNQX

N1 N2-~

.

.

.

.

Fig. 5. Physiology and ultrastructure of synapses formed by cultured 5-HT neurons. (A) Intracellular recordings from a 5-HT-immunoreactive neuron maintained for 25 days in a microculture in which this was the only 5-HT neuron present. The resting potential was - 6 0 mV. In control solution (CONT), a brief intracellular current pulse elicited an action potential that was followed by a prolonged AHP with a duration of approximately 0.9 s, Bath perfusion with the 5-HT~A receptor antagonist, spiperone (SPIP, 1 pM) decreased the duration of the AHP to 0.5 s, indicating that this neuron released 5-HT onto itself. (B) Intracellular recordings from two 5-HT neurons (N1 and N2) maintained in microculture for 29 days. The resting potentials in N I and N 2 were - 5 7 and - 7 0 mV, respectively. When N2 was stimulated with a brief intracellular current pulse in control solution (CONT), an action potential followed by autaptic excitation and an AHP with a duration of 2.6 s was observed in N2, and a biphasic excitatory/inhibitory response was evoked in N1. The kainate/AMPA glutamate receptor antagonist, CNQX (20#M), significantly reduced the evoked excitation in both neurons. However, the slow inhibitory response remained in both cells. When the 5-HT~A receptor antagonist (-)propranolol (PROP, 1/~M) was added to the solution containing CNQX, the inhibitory response was blocked in both neurons. The effects of CNQX and (-)propranolol were reversed (not shown). (C, D) Electron micrographs obtained from a microculture containing two 5-HT-immunoreactive neurons and two non-immunoreactive neurons that had been maintained in culture for 10 days. In (C), a 5-HT-immunoreactive axon terminal made contact with a 5-HT-immunoreactive postsynaptic element in the region of a spine-like process (S). The immuno-labeled axon terminal was filled with clear round synaptic vesicles (arrows), some of which were clustered near synaptic zones (asterisks). In (D), a vesicle-filled 5-HT-immunoreactive axon terminal was closely apposed to a non-immunoreactive postsynaptic element. No synaptic membrane specializations were detected along the extensive region of contact shown. Scale bars = 0.5 pm C,D. Intracellular recordings from a multipolar 5-HT neuron in a n o t h e r single neuron microculture are s h o w n in Fig. 9. In this cell, a brief intracellular pulse

evoked an action potential, which was followed by autaptic excitation and a prolonged A H P with a d u r a t i o n o f approximately 2.5 s. The fact that

Synaptic structure and function of cultured 5-HT neurons

C

617

!

CONT

I PROP

.J

50ms

H.i,

APV

CNQX

I1 Fig. 6. Electrophysiology and morphology of a solitary 5-HT neuron maintained for 21 days in microculture. The microculture in (A) contained a solitary neuron (arrow) that accumulated 5,7-DHT in (B). No other neurons were present in the microculture. The substrate in the center of the microculture had retracted, and was traversed by neuritic processes. Scale bar = 100 #m. (C) Intracellular recordings from the neuron shown in A. In control solution (CONT), a brief (< 10 ms) intracellular current pulse evoked an action potential that was followed by robust excitation. (-)Propranolol (PROP, 1 pM) did not affect the excitatory response. However, the excitation was reduced by the NMDA glutamate receptor antagonist, APV (50 #M), and was blocked completely by the kainate/quisqualate glutamate receptor antagonist, CNQX (20 pM). The effect of CNQX and APV were reversed (not shown). spiperone (1/~M) blocked the prolonged A H P suggested that the neuron formed serotonergic contacts with itself. The k a i n a t e / A M P A receptor antagonist, C N Q X (20/~M), reversibly blocked the excitatory response. Thus, the neuron appeared to be functionally serotonergic/glutamatergic. An ultrastructural examination of the microculture indicated that the axon and terminals of this neuron contained both round and elliptical large dense core vesicles, as well as clear round vesicles (Fig. 10). The round dense core vesicles had an outer diameter of 95 nm, with a central dense core of 70 nm. The elliptical dense core vesicles averaged 140 x 70 nm, with a central dense core of 110 x 50 nm. The large dense core vesicles were more numerous in this cell than in the neuron of Fig. 8, and microcanaliculi and

microvesicles were not observed. Clear round vesicles with a diameter of about 50 nm often filled the entire length of axon segments. These axons frequently formed multiple symmetric synapses with apposed dendrites and dendritic spines. However, frequent asymmetric synapses with prominent postsynaptic densities were also present. Large dense core vesicles appeared to be randomly distributed within the axon terminals; however, the clear round vesicles were often accumulated within the presynaptic active zones. The types of vesicles present were generally similar in each terminal. Axo-axonic and dendrodendritic synapses were not observed. As in previous cases, some axo-somatic or axodendritic contacts failed to exhibit membrane specializations. Because serial sections were not done,

618

M. D. Johnson and A. G. Yee

Fig. 7. Electron micrograph of the same glutamatergic 5-HT neuron described in Fig. 6. The soma of the neuron (N) rested on the upper surface of a flattened glial cell (G1). Beneath the glial cell, a thin layer of collagen (C) was observed on top of the agarose (A) that coated the glass coverslip. The nucleus of the neuron (nu) was highly indented and eccentrically located. The cytoplasm contained a well developed Golgi complex, endoplasmic reticulum (ER), and a large number of free ribosomes, lysosomes and mitochondria. Numerous vesicle-filled axon terminals were observed in close apposition to the soma and dendrites of the neuron, as illustrated in Fig. 8. Scale bar = 5/lm.

however, the possibility remained that these terminals formed conventional synapses. This point was underscored by data obtained from semi-serial sections through the same axo-dendritic zone of apposition (Fig. 10). In several cases, terminals that initially appeared to be non-junctional were found in subsequent sections to form conventional symmetric or asymmetric synapses. DISCUSSION

Functional synapses between serotonergic and nonserotonergic neurons in multineuron microcultures Electrophysiological studies in slices and in vivo have indicated that mesopontine 5-HT neurons receive functional noradrenergic, GABAergic, and glutamatergic synaptic input, 22'25'36 and morphological studies on m a m m a l i a n brain sections have confirmed the presence of noradrenergic and G A B A ergic synapses on 5-HT neurons in the dorsal raphe nucleus. ~'39 We have recently reported the presence of functional transmitter-mediated interactions beween dissociated mesopontine 5-HT and non-5-HT

neurons in long term culture. ~7 The present study demonstrates that the glutamatergic and GABAergic contacts that mediate these interactions have the ultrastructural properties characteristic of conventional synapses. Both symmetric and asymmetric contacts were present. As in vivo, the synapses in culture were axo-somatic, axo-dendritic and axospinous, but not axo-axonic. 33'36 Dendro-dendritic synapses were not observed.

Morphology of serotonergic autapses in culture In vivo, mesopontine 5-HT neurons are thought to be influenced by 5-HT1A receptor-mediated autoinhibition. 35'37 However, it is not known whether this inhibition is mediated by axo-dendritic or dendrodendritic contacts. Although few in number, 5-HT immunoreactive axon terminals have been observed in the dorsal raphe nucleus of the rat, cat and monkey. 6'8'15'2° Similarly, 5-HT-immunoreactive dendrites have been reported to contain vesicles in the intact dorsal raphe nucleus and in embryonic raphe cells grafted to the hippocampus. 6'7 The present results indicate that cultured 5-HT neurons inhibited

Synaptic structure and function of cultured 5-HT neurons

619

Fig. 8. Both symmetric and asymmetric synapses were formed by the glutamatergic 5-HT neuron described in Figs 6 and 7. (A) Numerous axon terminals formed synapses on dendritic profiles (D). Each axon terminal of this neuron contained a similar population of elements. In addition to large dense core vesicles, microcanaliculi (mc) and microvesicles (mv) were present. Scale bar = 0.5/~m A. (B) Although most of the synapses formed by this 5-HT neuron were symmetric, a few asymmetric synapses were also present. One of the axon terminals (A~) in this micrograph formed a symmetric synapse (arrowhead) on a postsynaptic dendrite, while another (A2) formed an asymmetric synapse (open arrow). The types of vesicles in the two axon terminals were similar. Close appositions that lacked membrane specializations in this thin section were also present. Scale bar = 0.5/~m B.

themselves a n d n e i g h b o r i n g n e u r o n s via recurrent 5-HT a x o n collaterals t h a t formed b o t h symmetric a n d asymmetric s o m a t o d e n d r i t i c synapses. These findings are consistent with 5 - H T - m e d i a t e d inhibition via recurrent collaterals within the r a p h e nuclei in vivo. The inhibition was a p p a r e n t l y mediated by 5-HT~A receptors, a n d thus provides strong func-

tional a n d ultrastructural evidence t h a t 5-HT released at c o n v e n t i o n a l synaptic sites can activate such receptors. A l t h o u g h no functional or ultrastructural evidence for synaptic vesicle release by dendrites was o b t a i n e d in the present study, we c a n n o t rule out the possibility t h a t some fraction of the observed effects resulted from dendritic t r a n s m i t t e r release.

620

M.D. Johnson and A. G. Yee

li

SPIP

cNQX

60ms

C Fig. 9. Electrophysiology and morphology of a single 5,7-DHT-labeled raphe neuron maintained in microculture for 13 days. This multipolar neuron (arrow in A) was the only neuron in the microculture, and displayed in B the pale blue fluorescence characteristic of neurons that accumulate 5,7-DHT. Scale bar= 50#m. (C) A 10ms intracellular current pulse evoked an initial action potential that was followed by excitation and an AHP with a duration of approximately 2.5s. The 5-HT~A receptor antagonist, spiperone (SPIP, 1 #M), blocked the AHP. The effect of spiperone was not reversed. The glutamate receptor antagonist, CNQX (20/~M), completely and reversibly blocked the evoked excitatory response.

Ultrastructure o f serotonergic axon terminals that release glutamate Previous work in this laboratory revealed that cultured 5-HT neurons frequently evoke glutamatergic potentials in themselves and in target neurons. ~7 The latency and time course of these responses suggested that they were synaptic in origin, and ultrastructural data obtained in the present study confirmed this hypothesis. Interestingly, the structure of the glutamatergic axon terminals and associated synapses formed by 5-HT neurons in culture closely

matched that previously reported for 5-HT immunoreactive axon terminals in vivo. 2'4'21'28'33 Serotonergic axon terminals containing microcanaliculi and microvesicles have been observed in several brain regions, including the paratrigeminal nucleus, locus coeruleus, and the nucleus of the solitary tract. 4'2~'28 Although frequently non-junctional, they have been reported to form conventional symmetric and asymmetric synapses with target neurons. All 5-HT axon terminals in the locus coeruleus of the rat reportedly contain microcanaliculi and microvesicles.2~ The fact that this morphological subtype of 5-HT axon terminal released glutamate in culture raises the possibility that a similar phenomenon occurs in vivo. Extracellular stimulation of the dorsal raphe nucleus with 100-200/~A currents not only produced 5-HTmediated inhibition in the locus coeruleus, but short latency (3-5 ms) excitation as well. 3° Several investigators have observed short latency ( < 1 0 m s ) depolarizing synaptic potentials in other areas of the nervous system after stimulation of the raphe nuclei.H'26 In some cases, these depolarizing potentials have been blocked by glutamate receptor antagonists, and glutamate immunoreactivity has been localized in a majority of medullary rat 5-HT neurons. H'23 Serotonergic axon terminals with other morphological characteristics also released glutamate at functional synapses in culture. Figure 10 illustrates a neuron which evoked both glutamatergic and serotonergic effects in itself, and which contained clear round vesicles and numerous large dense core vesicles. The ultrastructural characteristics of the axon terminals of this neuron resembled those observed in vivo by many investigators, 33again raising the possibility that such terminals release both 5-HT and glutamate at functional synapses. Large dense core vesicles often contain monoamines and neuropeptides, are frequently not accumulated at synaptic zones, and may require trains of impulses for release. 34 Consistent with these findings, large dense core vesicles in cultured 5-HT neurons were not accumulated at synaptic zones. However, stimulation of the neurons with trains of impulses did not alter the frequency with which 5-HT-mediated synaptic effects were observed (unpublished observations). Heterogeneity o f synapses formed by a single serotonergic neuron The frequency with which 5-HT axon terminals form symmetric, asymmetric, or non-junctional contacts with target neurons varies from one region of the brain to another, 33 Axon terminals of embryonic 5-HT neurons grafted to the 5-HT-denervated hippocampus exhibit the same incidence of junctional versus non-junctional contacts as the normal 5-HT innervation. 7 Similarly, when 5-HT axon terminals were allowed to hyperinnervate the striatum after destruction of the endogenous dopaminergic inner-

Synaptic structure and function of cultured 5-HT neurons

621

Fig. 10. Synaptic morphology of the serotonergic/glutamatergic 5-HT neuron illustrated in Fig. 9. (A) The axon terminal (AI) contained numerous large dense core vesicles and clear round vesicles, and formed synapses on two small caliber dendritic branches (labelled D). (B) Numerous synaptic vesicles are present along the length of the axon segment (A0; the latter formed a synapse (arrowhead) on an apposed dendrite (labelled D). The axon also contacted a spine (S) of the dendrite. Each axon terminal of this neuron (e.g. A~ and A2) contained morphologically similar types of synaptic vesicles. Scale bar = 0.5/am A,B. (CO Another axon terminal (A) of this 5-HT neuron contained numerous large dense core vesicles (arrows) and clear round vesicles, and made close contact with a dendrite (labelled D). The arrowhead points to a zone of apposition that lacked membrane specialization. As shown in C 2, however, subsequent sections through the same profile revealed an asymmetric synapse with a prominent postsynaptic density. Note the presence of prominent presynaptic densities as well. Scale bar = 0.25/am C~,C2. vation, the intrinsic and relational features of the sprouted 5-HT terminals were the same as those in normal tissue. 9 One explanation for these results is that the target region plays a significant role in determining the relative n u m b e r of symmetric, asymmetric, or non-junctional 5-HT contacts. In support of this hypothesis, Beaudet and Sotelo 2 found that destruction of granule cells in the cerebellum converted a mostly non-junctional 5-HT innervation into a junctional one. A prediction of this "region specific" hypothesis is that the axon of a single 5-HT neuron has the intrinsic capacity to form various types of contacts with target cells. The data in the present study support this idea; the axon terminals from a single 5-HT neuron formed both symmetric and asymmetric synapses. Recent reports have raised the possibility that single dopaminergic neurons can also form both symmetric and asymmetric synapses, and that the nerve terminals forming these two types of synapses are cytochemically distinct. 16

Axo-dendritic or axo-somatic appositions with the characteristics of non-junctional complexes were also observed in the present study. However, serial sections required to demonstrate that these contacts lacked membrane specializations have yet to be done. Although the types of synaptic membrane specializations formed by the axon terminals of a single cultured 5-HT neuron varied, each terminal or varicosity of that neuron contained a similar complement of morphological subtypes of vesicles. No evidence for segregation of morphological subtypes of vesicles into separate axon terminals was observed. These morphological observations are thus consistent with Dale's Principle, i.e. that a neuron secretes the same transmitter(s) from all of its axon terminals. 1° The findings also support the hypothesis that the distinct morphological subtypes of 5-HT axon terminals observed in vivo arise from separate populations of 5-HT neurons, each producing certain types of synaptic vesicles. Because of the relative uniformity

622

M . D . Johnson and A, G. Yee

of the microculture environment compared with that observed in vivo, however, additional studies are needed to rule out the possibility that the morphology of synaptic vesicles within individual axon terminals of a single 5-HT neuron is locally determined by the microenvironment of each terminal. CONCLUSIONS Dissociated 5-HT and non-5-HT raphe neurons of the rat re-established synapses in culture that were functionally and ultrastructurally similar to those observed in vivo. Direct assays of both the synaptic function and morphology of solitary raphe neurons revealed that single serotonergic and/or glutamatergic neurons can simultaneously form both symmetric and asymmetric synapses. Individual 5-HT neurons differed with respect to synaptic vesicle morphology, but there was no evidence for segre-

gation of different types of synaptic vesicles into separate axon terminals of the same cell. These results raise the possibility that the morphological heterogeneity of synaptic vesicles observed within 5-HT axon terminals in vivo arises from separate subpopulations of raphe neurons. However, the incidence of symmetric, asymmetric, or non-junctional synapses formed by 5-HT axon terminals may be determined by the target tissue. Acknowledgements--The authors thank Dr D. D. Potter for

the use of equipment, helpful advice, and review of the manuscript. We also thank Dr Min-min Lu for technical assistance. This work was supported by the Freudenberger Fund and by NS02253-32 to D. D. Potter. M. D. Johnson is a recipient of a Harvard Ryan Fellowship, a Commonwealth Fund/Bristol-Myers Squibb Fellowship in Academic Medicine, an American Psychological Association Minority Fellowship, and a Medical Scientist Training Program Fellowship.

REFERENCES

I. Baraban J. M. and Aghajanian G. K. (1981) Noradrenergic innervation of serotonergic neurons in the dorsal raphe: demonstration by electron microscopic autoradiography. Brain Res. 204, 1-11. 2. Beaudet A. and Sotelo C. (1981) Synaptic remodeling of serotonin axon terminals in rat agranular cerebellum. Brain Res. 2116, 305-329. 3. Belin M. F., Nanopoulos D., Didier M., Aguera M., Steinbusch H., Verhofstad A., Maitre M. and Pujol J. F. (1983) Immunohistochemical evidence for the presence of gamma-aminobutyric acid and serotonin in one nerve cell. A study on the raphe nuclei of the rat using antibodies to glutamate decarboxylase and serotonin. Brain Res. 275, 329 339. 4. Chan-Palay V. (1978) The paratrigeminal nucleus--II. Identification and inter-relations of catecholamine axons, indoleamine axons, and Substance P immunoreactive cells in the neuropil. J. Neurocytol. 7, 419442. 5. Chan-Palay V., Jonsson G. and Palay S. L. (1978) Serotonin and substance P coexist in neurons of the rat's central nervous system. Proc. natn. Acad. Sci. U.S.A. 75, 1582 1586. 6. Chazal G. and Ralston H. J. (1987) Serotonin-containing structures in the nucleus raphe dorsalis of the cat: an ultrastructural analysis of dendrites, presynaptic dendrites, and axon terminals. J. comp. Neurol. 259, 317 329. 7. Daszuta A., Chazal G., Garcia S., Oleskevich S. and Descarries L. (1991) Ultrastructural features of serotonin neurons grafted to adult rat hippocampus: an immunocytochemical analysis of their cell bodies and axon terminals. Neuroscience 42, 793 811. 8. Descarries L., Watkins K. C., Garcia S. and Beaudet A. (1982) The serotonin neurons in nucleus raphe dorsalis of adult rat: a light and electron microscope radioautographic study. J. comp. Neurol. 207, 239 254. 9. Descarries L., Soghomonian J.-J., Garcia S., Doucet G. and Bruno J. P. (1992) Ultrastructural analysis of the serotonin hyperinnervation in adult rat neostriatum following neonatal dopamine denervation with 6-hydroxydopamine. Brain Res. 569, 1-13. 10. Eccles J. C. (1976) From electrical to chemical transmission in the central nervous system. R. Soc. Notes Records 30, 219-230. 11. Fung S. J., Liu R. H., Reddy V. K. and Barnes C. D. (1992) Glutamatergic raphespinal excitation of spinal motor output in cats. Soc. Neurosci. Abstr. 18, 1150. 12. Furshpan E. J., Landis S. C., Matsumoto S. G. and Potter D. D. (1986) Synaptic functions in rat sympathetic neurons in microcultures--I. Secretion of norepinephrine and acetylcholine. J. Neurosci. 6, 1061-1079. 13. Furshpan E. J. and Potter D. D. (1989) Seizure-like activity and cellular damage in rat hippocampal neurons in cell culture. Neuron 3, 199-207. 14. Glazer E. J., Steinbusch H., Verhofstad A. and Basbaum A. I. (1981) Serotonin neurons in nucleus raphe dorsalis and paragigantocellularis of the cat contain enkephalin. J. Physiol. Paris 77, 241 245. 15. Harandi M., Aguera M., Gamrani H., Didier M., Maitre M., Calas A. and Belin M. F. (1987) Gamma-aminobutyric acid and 5-hydroxytryptamine interrelationship in the rat nucleus raphe dorsalis: combination of radioautographic and immunocytochemical techniques at light and electron microscopy levels. Neuroscience 21, 237 251. 16. Hattori T., Takada M., Moriizumi T. and Van Der Kooy D. (1991) Single dopaminergic nigrostriatal neurons form two chemically distinct synaptic types: possible transmitter segregation within neurons. J. comp. Neurol. 309, 391401. 17. Johnson M. D. (1994) Synaptic glutamate release by postnatal rat serotonergic neurons in microculture. Neuron 12, 433-442. 18. Johnson M. D. and Ma P. M. (1993) Localization of NADPH diaphorase activity in monoaminergic neurons of the rat brain. J. comp. Neurol. 332, 391406. 19. Johnson M. D. and Potter D. D. (1989) Co-culture of dissociated hippocampal and sympathetic ceils from the neonatal rat. Q. J. exp. Physiol. 74, 1043-1050. 20. Kapadia S. E., de LaneroUe N. C. and LaMotte C. C. (1985) Immunocytochemical and electron microscopic study of serotonin neuronal organization in the dorsal raphe nucleus of the monkey. Neuroscience 15, 729-746. 21. Leger L. and Descarries L. (1978) Serotonin nerve terminals in the locus coeruleus of adult rat: a radioautographic study. Brain Res. 145, 1-13.

Synaptic structure and function of cultured 5-HT neurons

623

22. Levine E. S. and Jacobs B. L. (1992) Neurochemical afferents controlling the activity of serotonergic neurons in the dorsal raphe nucleus: microiontophoretic studies in the awake cat. J. Neurosci. 12, 4037-4044. 23. Nicholas A. P., Pieribone V. A., Arvidsson U. and Hokfelt T. (1992) Serotonin-, substance P-, and glutamate/aspartatelike immunoreactivities in medullo-spinal pathways of rat and primate. Neuroscience 48, 545-559. 24. Pan Z. Z., Colmers W. F. and Williams J. T. (1989) 5-HT-mediated synaptic potentials in the dorsal raphe nucleus: interactions with excitatory amino acid and GABA neurotransmission. J. Neurophysiol. 62, 481-486. 25. Pan Z. Z. and Williams J. T. (1989) GABA- and glutamate-mediated synaptic potentials in rat dorsal raphe neurons in vitro. J. Neurophysiol. 61, 719-726. 26. Park M. R., Gonzales-Vegas J. A. and Kitai S, T. (1982) Serotonergic excitation from dorsal raphe stimulation recorded intracellularly from rat caudate-putamen. Brain Res. 243, 49 58. 27. Peters A., Palay S. L. and Webster H. (1976) Synapses. In The Fine Structure of the Nervous System, pp. 118-180. Saunders, Philadelphia. 28. Pickel V. M., Joh T. H., Chart J. and Beaudet A. (1984) Serotonergic terminals: ultrastructure and synaptic interaction with catecholamine-containing neurons in the medial nuclei of the solitary tracts. J. comp. Neurol. 225, 291-301. 29. Sah D. W. Y. and Matsumoto S. G. (1987) Evidence for serotonin synthesis, uptake, and release in dissociated rat sympathetic neurons in culture. J. Neurosci. 7, 391-399. 30. Segal M. (1979) Serotonergic innervation of the locus coeruleus from the dorsal raphe and its action on responses to noxious stimuli. J. Physiol. Lond. 286, 401-415. 3 I. Segal M. M. and Furshpan E. J. (1990) Epileptiform activity in microcultures containing small numbers of hippocampal neurons. J. Neurophysiol. 64, 1390-1399. 32. Silva N. L., Mariani A. P., Harrison N. L. and Barker J. L. (1988) 5,7-Dihydroxytryptamine identifies living dopaminergic neurons in mesencephalic cultures. Proc. natn. Acad. Sci. U.S.A. 85, 7346-7350. 33. Soghomonian J.-J., Beaudet A. and Descarries L. (1988) Ultrastructural relationships of central serotonin neurons. In Neuronal Serotonin (eds Osborne N. N. and Hamon M.), pp. 57 92. Wiley, New York. 34. Sossin W. S., Fisher J. M. and Scheller R. H. (1989) Cellular and molecular biology of neuropeptide processing and packaging. Neuron 2, 1407-1417. 35. Sprouse J. S. and Aghajanian G. K. (1986) (-)-Propranolol blocks the inhibition of serotonergic dorsal raphe cell firing by 5-HT~A selective agonists. Eur. J. Pharmae. 128, 295-298. 36. Wang Q.-P., Ochiai H. and Nakai Y. (1992) GABAergic innervation of serotonergic neurons in the dorsal raphe nucleus of the rat studied by electron microscopy double immunostaining. Brain Res. Bull. 29, 943 948. 37. Wang R. Y. and Aghajanian G. K. (1978) Collateral inhibition of serotonergic neurones in the rat dorsal raphe nucleus: pharmacological evidence, Neuropharmacology 17, 819-825. 38. Yoshimura M. and Higashi H. (1985) 5-Hydroxytryptamine mediates inhibitory postsynaptic potentials in rat dorsal raphe neurons. Neurosci. Lett. 53, 69-74. 39. Yoshimura M., Higashi H. and Nishi S. (1985) Noradrenaline mediates slow excitatory synaptic potentials in rat dorsal raphe neurons in vitro. Neurosci. Lett. 61, 305-310. (Accepted 13 December 1994)

NSC 67/3 .-E