synaptic relationships of serotonin-inmmunoreactive terminal baskets pm gaba neurons in the cat...

Synaptic Relationships ofSerotonin-lmmunoreactiveTerminal Baskets on GABANeurons in the Cat AuditoryCortex

J. DeFelipe,1-2 S. H. C. Hendry,1-2 T. Hashikawa,2 andE. G.Jones1-2

1 Department of Anatomy and Neurobiology,University of California, Irvine, California 92717and 2 Neural Systems Laboratory, Frontier ResearchProgram, Institute for Physical and ChemicalResearch, Wako-shi, Saitama, 351-01, Japan

Correlative light and electron microscopic immunocy-tochemical methods were used to analyze the 5-HTinnervation of the primary auditory area (Al) of the catcerebral cortex and to examine the synaptic relationshipsof 5-HT basket terminations on target neurons in thatarea. Three morphological types of 5-HT-immunoreac-tive fibers are present: type I, which is very thin andvery finely beaded; type II, which is thin and coarselybeaded; and type III, which has a relatively thick mainshaft and very few beads. Type I is the most abundant,type II is relatively less common, and type III is theleast abundant type. The 3 types of fibers are presentthrough the thickness of Al and in the subjacent whitematter, but the densest plexus is found in layers 1—III.One of the most characteristic features of type II fibersis that they commonly form small, dense clusters thatresemble baskets apposed to the somata and primarydendrites of unstained neurons. The basket formationsare more frequently found in layers I and II, and theyvary in complexity. Simultaneous immunostaining forGABA and 5-HT reveals that many 5-HT baskets sur-round the somata and dendrites of GABA neurons. In2-pm-thick plastic sections, each basket formation canbe seen surrounding 1 or a group of 2 or 3 cells. In thelatter case, one cell is much larger and at the electronmicroscope level is identified as a neuron, while theother cells are neuroglial cells. Reconstructions weremade from serial electron micrographs of 135 5-HT-immunoreactive boutons. Of these boutons, 110 be-longed to basket formations, 14 to type I axons locatedin the neuropil, and the remaining 11 to type II fiberslocated in the white matter. Only 4 of the 135 boutonsmade conventional synaptic contacts. These were of theasymmetrical type. Most of the boutons made very small,indistinct membrane specializations or none at all. Thepresent results therefore suggest a strong interactionbetween 5-HT axon terminals and specific GABA neu-rons, which may be mediated by release sites that arenot associated with morphologically distinct synapticcontacts.

Monoaminergic projections, when first visualized inthe cerebral cortex with histofluorescence methods,were described as diffuse and nonspecific: relativelyfew neurons located in the brainstem were found togive rise to widespread, diffuse axonal arborizationsthat innervated mainly the outer cortical layers in amanner similar to nonspecific thalamic afferents (e.g.,Fuxe et al., 1968). These observations, together withthe early electron microscopic studies of Descarriesand colleagues (see review by Beaudet and Descar-ries, 1984), which emphasized that monoaminergicfibers made few morphologically distinguishable syn-aptic contacts, gave support to the idea of a nonspe-cific monoaminergic cortical innervation. With theintroduction of immunocytochemical techniques, itwas shown that there are clear areal and laminar dif-ferences in the distribution of monoaminergic fibers(e.g., Levitt et al., 1984; Morrison et al., 1984), andstudies at the electron microscopic level have indi-cated that synaptic contacts are more frequent thanpreviously supposed (e.g., de Lima et al., 1988; Gold-man-Rakic et al., 1989; for a review, see Parnavelasand Papadopoulos, 1989).

There has been little evidence to indicate thatmonoamine-containing fibers terminate specificallyon particular types of cortical neurons. However, re-cent studies on the cerebral cortex of the cat (Mul-ligan and Tork, 1988) and marmoset (Hornung et al.,1990) have described dense, multiterminal arrange-ments of 5-HT-immunoreactive fibers ("baskets")around certain unidentified cells. These findings sug-gest that 5-HT axons selectively innervate single cellsin a manner not yet found for other afferent systemsto the cerebral cortex.

In their study of the cat cerebral cortex, Mulliganand Tork (1988) found that the 5-HT basket termi-nations were more common in the temporal lobe thanin other regions of the cortex and were particularlycommon in the primary auditory area (Al). This istherefore the area of choice for a study designed todetermine the nature of the neurons innervated bythe 5-HT baskets.

The terminations of 5-HT-immunoreactive fibersthat do not form baskets have been described in elec-tron microscope immunocytochemical studies of therat and monkey cerebral cortex (Papadopoulos et al.,1987; DeFelipe and Jones, 1988; de Lima et al., 1988;Seguela et al., 1989), but results differ regarding thefrequency with which morphologically distinguish-

Cerebral Cortex Mar/Apr 1991;M17-133; 1047-3211/91/J2.00

at OC

LC

on Novem

ber 5, 2014http://cercor.oxfordjournals.org/

Dow

nloaded from

http://cercor.oxfordjournals.org/

able synaptic contacts are made. Several authors claimthat nearly all (approximately 90%) of the boutonssampled make demonstrable synaptic contacts (seeParnavelas and Papadopoulos, 1989), while others,including ourselves, in serial-section analyses havefound that only a small proportion of varicosities (3 -46%) form distinct synaptic contacts (DeFelipe andJones, 1988; Seguela et al., 1989). De Lima et al. (1988)report that many contacts are made but that they areextremely small. The dense terminations of 5-HT ax-ons in the baskets thus offer an opportunity to reex-amine this problem.

Materials and Methods

Tissue PreparationEight adult cats of both sexes were used in this study.All were given an overdose of sodium pentobarbitaland were perfused through the heart with 4% para-formaldehyde and 0.1% or 0.2% glutaraldehyde in 0.1 Mphosphate buffer (pH, 7.4). The brains were rapidlyremoved and postfixed in 4%-buffered paraformal-dehyde for 6-12 hr at 4°C. Blocks including the firstauditory area (Al) were then transferred to either 7%or 20% sucrose in phosphate buffer. Those in 7% su-crose were cut on a Vibratome at 25 or 40 /zm andcollected in ice-cold phosphate buffer. Blocks in 20%sucrose were stored at 4°C until they had sunk andwere then frozen in dry ice. Frozen sections were cuton a sliding microtome at alternating thicknesses of15 and 30 Mm. Some were cut on a cryostat at 5 nm.All Vibratome-cut and cryostat-cut blocks and half thesliding-microtome-cut blocks were cut in the frontalplane. The other blocks cut on the sliding microtomewere cut parallel to the pial surface after flatteningthe blocks during freezing.

Itnmunocytochemistry and HistocbemtstryRegular series of Vibratome and frozen sections wereprocessed for immunocytochemistry with a previous-ly characterized rat monoclonal antibody to serotonin(5-HT; Consolazione et al., 1981; Sera Labs). Sectionswere preincubated in 3% normal horse serum over-night, incubated in the anti-5-HT antibody diluted1:500 in 0.1 M phosphate buffer with 3% normal horseserum for 36-48 hr, and processed by the avidin-bi-otin-peroxidase (ABC) method. All incubations weredone at 4°C, and for frozen sections, 0.3% Triton X-100was added to all solutions. The peroxidase signal wasdetected by reaction with 3,3'-diaminobenzidinetetrahydrochloride (50 mg/100 ml 0.1 M phosphatebuffer) and 0.01% hydrogen peroxide. For blocks cuton the sliding microtome, a regular series of the thick-er sections was mounted and stained with thionin orprocessed histochemically for acetylcholinesterase,which provides confirmation of the localization of Al.Cryostat sections were processed as above, except thatthe primary incubation included both the rat anti-5-HT monoclonal antibody and a previously character-ized rabbit anti-GABA antiserum (Chemicon), bothat dilutions of 1:1000. After the sections were incu-bated in the 2 primaries overnight, they were washed

and incubated simultaneously in fluorescein isothio-cyanate (FITC)-conjugated goat anti-rat IgG and rho-damine-conjugated goat anti-rabbit IgG (Chemiconand Cal-Tag). These sections were then washed andmounted on slides in phosphate-buffered glycerol (1:3) and examined and photographed in a Leitz Dialuxepifluorescence microscope equipped with FITC andrhodamine excitation filters. Many of the sections wereexamined and photomicrographed without a rhoda-mine barrier filter so that part of the rhodamine(GABA) signal could be detected simultaneously withthe FITC (5-HT) signal.

Electron MicroscopyVibratome sections, processed by immunocytochem-istry for 5-HT by the ABC method, were postfixed in1% osmium tetroxide in 0.1 M phosphate buffer for 1hr. They were dehydrated, infiltrated with Aralditeresin, and embedded flat between silicone-coatedslides and coverslips. Serial 2.5-Mm-tnick plastic sec-tions were then obtained with a Reichert Ultracutultramicrotome, examined with the light microscope,photographed, and then resectioned at 60-70 nm(DeFelipe and Fairen, 1982). The thin sections werecollected on Formvar-coated, single-slot grids, stainedwith uranyl acetate and lead citrate, and examined ina Phillips CM-10 electron microscope. By using thegoniometer tilt control of the electron microscope,specimens were rotated until an optimum tilt anglewas obtained for visualization of putative synapticcontacts (see Fig. 8D,E). Selected immunoreactiveboutons (see below) were reconstructed from trac-ings of serial electron micrographs.

Results

Light microscopy

Types o/5HT FibersThree morphological types of 5-HT-immunoreactivefibers are present in cat Al (Fig. 1). Type I axons arethe most common. They are very thin (<0.5 fim) andvery finely beaded (with swellings that are 0.5-1 nmin diameter). Over distances of several hundred mi-crons, these axons are seldom seen to divide. TypeII immunoreactive fibers are less common. They arealso thin but are very coarsely beaded with round orovoid varicosities of 1-3 nm in diameter. Type II fibersare very tortuous; they frequently bend at irregularangles and sometimes make several recurrent turnsover a distance of 100-200 iitn. Over that distance,they often give off 1 or 2 branches. Type III axons arethe least-abundant type of immunoreactive axon. Theyhave relatively thick main shafts (0.5-1 mm) and veryfew beads.

Laminar Distribution of FibersThe 3 types of 5-HT-immunoreactive fibers are pres-ent through the thickness of Al and in the subjacentwhite matter. Within Al, the fibers are unevenly dis-tributed: they form a dense plexus in layers I—III anda much less dense plexus in layers IV-VI (Fig. 2/4).

118 Serotonin Baskets on GABA Cells • DeFelipe et al.

at OC

LC

on Novem


Dow

nloaded from


wm

Figure 1 . Camera lucida drawings of 5-HT-immunoreactive fibers to illustrate the 3 morphological types (/. //. and imatter [wnt\. See Results for more details. Scale bar, 10 fim.

| of 5-HT fibers found in the cortex and subjacent white

All 3 types of fibers contribute to the superficial plex-us, but the density of the coarsely beaded (type II)fibers is much greater in layers I and II, and the densityof finely beaded (type I) fibers is greater in layers IIIand IV. In layers III and IV, the majority of the typeI fibers adopt no preferred orientation, while in layersI and II they join with type II fibers to form ascendingbundles. Evenly distributed throughout layers I-IVare a small number of thick (type III) immunostainedfibers, most of which are oriented parallel to the pialsurface.

The lighter plexus of fibers in layers IV-VI is madeup principally of type I fibers (Fig. 2A). These aresomewhat thicker in layers IV-VI and the subjacentwhite matter than in layers I—III. Most are orientedparallel to the laminar boundaries, and in single fron-tal sections they can be followed for several milli-meters. Type II fibers are represented in this deepplexus mainly by sharply twisting, isolated segmentsthat, in single sections, can be followed for only 50-100 pirn. Type III fibers form bundles that adopt ra-dially or obliquely ascending paths and, in single sec-tions, can be traced for 200-300 urn.

5HT-Immunoreactive BasketsShort segments of coarsely beaded (type II) axonscan be found in isolation through the thickness of AI.They commonly form dense clusters that consist ofas many as 40 boutonlike swellings (Fig. 3)- Mostimmunostained clusters surround a central round oroblong unstained region, 8-20 /«n in diameter, andfrom them emerge 1-3 beaded fibers that radiate fromthe central region, as though following dendrites (Fig.5A,C,D). The clusters thus resemble "basket" for-mations enclosing the somata and primary dendritesof unstained neurons. However, not all type II fibersin single sections are found to be pan of a basket

formation (Fig. 3B). The basket formations are pres-ent in all layers but, like the type II fibers from whichthey arise, are more common in the supragranularlayers, particularly layers I and II.

The immunostained basket formations vary in com-plexity. Some consist of 20-25 swellings intercon-nected by long, thin fiber segments without swellings;some show an almost continuous sequence of manytens of swellings (Figs. 3A, 4). Some baskets possessmore swellings in the segments that radiate alongpresumed dendrites from the central unstained regionsthan in the central region itself (Figs. 3D, AA,D). With-in layers I and II, the individual baskets are usuallymore complex, with more swellings and longer ra-diations than in the deeper layers. The size and shapeof the basket formations appear to be related to thesize and shape of the target cell they are presumedto enclose; the most common morphology of the un-stained region outlined by the baskets is rounded(Fig. AB,C,G), fusiform, or horizontal (Fig. 4A,D-F,H,t). The configuration of the basket formations inwhich radiating segments outline putative dendritessuggests that the enclosed cell is usually nonpyrami-dal, but a few with pyramidlike morphologies arealso found (Fig. 4B,C).

In single frontal sections through AI, 1-5 basketsare commonly seen, along with a number of varicose5-HT-immunoreactive type II fibers that appear tooutline only dendrites. The latter include some elon-gated type II fibers, 400-500 Mm in length, over whichseveral dozen swellings are given off.

At irregular intervals in layer V, 1 or 2 finely beaded(type I) fibers also form complex, tangled arrange-ments (Fig. 3E,F). These are small regions (20-40 nmin diameter) in which the immunoreactive fibers bi-furcate repeatedly, each branch making sharp turnsso that a solid mass of fibers results. Unlike the bas-

Cerebral Cortex Mar/Apr 1991, V 1 N 2 119

at OC

LC

on Novem


Dow

nloaded from


2

> : -4.

'• I V

B

. , 1VI

• V S • • • • • • ; • • • • : * *

WM/ • • * . - • ' ' • "

Hgura 2 . Phonnicroor^hs of S-HT-orvmnorBsctivs sxoni in cs Al. A Fronts! ssction throuyh ISYBT I to whitB mattsr. F snd C TanQsnus) ssction dwough tayen I and fi.Ths 2 obfiqus bands of axons in B [optn snwsft sni domvGntJBOy orantod in d£ anirnaL At Nyto nBoni&stnn (r)> tnB stained bands ore sasr to nctuds a QTBStai density oftype II fiber*. Solid antm indkate the same blood vetsets in A and fl Seals bar 120 M"I to A 500 M"I to fl 230 MTI far C.

kets, these formations do not enclose large unstainedregions.

Organization of Immunoreactive Fibers inTangential SectionsIn sections cut parallel to the pial surface of flattenedAI (i.e., in an approximately parasagirtal plane), thetype II immunoreactive fibers in layers I—III adopt apreferential, dorsoventral orientation (Fig. 25). In thetangential sections, baskets are distributed irregular-ly, but the varicose segments that radiate from thecentral core or traverse the space between baskets are

elongated dorsoventrally. Accompanying them arenumerous finely beaded (type I) fibers and, rarely,thick unbeaded (type III) fibers. The dorsoventrallyoriented fibers are also periodic, as densely innervat-ed bands and lightly innervated bands alternate withone another (Fig. 25). Both sets of bands are 500-700 iim wide.

The immunoreactive fibers in deeper layers displayno preferred orientation. In layers IV-VI, most fibersare cut in cross section or appear as sharply curvedsegments. As a result, the immunostained fibers inthese layers appear as unorganized tangles of threads.

12t Serotonin Baskets on GABA Cells • DeFellpe et al.

at OC

LC

on Novem


Dow

nloaded from


v

* » % • -- * - ' •>

• • (k

B

: . • > • • • < .

< • • .• • • • ; , ' :?•• •

• V

• / . ;

f 3. Differential i i i a fumce contrast plmaiiiuugiatJa of 54fTHmmunoreactive basket formatkns (A C D) and other complex arrangements of fibers | f i f , f ) in catAl. The baskets vary in orgarizauon from those that are dansest srotnd the presumpova somata of neurons [A. C) to those that are distributed preferentially along proximaldsndrites [0], B, A type II fiber with a duster of vBncosmes that do not appear to form a basket E and F, A type I fiber, photographed in 2 focal ptanes, that forms a densetangle. The tangle is without a carers! unstained region. Scale bar 15 /jm for A fl f . and F; 35 / im for C and 0.

Simultaneous Localization of 5-HT andGABA ImmunoreactivityThe appearance of the coarsely beaded type II fibersthat form the baskets suggests that they surround thesomata and dendrites of neurons. Simultaneous im-munostaining for GABA and 5-HT reveals that manyof the regions enclosed by the 5-HT fibers are thesomata and dendrites of GABA neurons (Fig. 5). Inlayers I and II, the more complex baskets enclose notonly the cell bodies and primary dendrites of GABA-immunoreactive cells (Fig. bE.P), but also secondaryand tertiary branches of the dendrites (Fig. 5/1,5). In

layers III—IV, the simpler baskets surround GABA-immunostained somata and primary dendrites. Ex-amination of serial frozen sections through AI of 3cats reveals that throughout AI most 5-HT-immuno-stained baskets surround the somata and processes ofGABA neurons. However, extremely rare examplesexist in which 5-HT baskets surround regions that arenot stained for GABA. As an indication of the fre-quency of GABA-unstained neurons enclosed by 5-HTbaskets, of 117 baskets identified in a series of 10sections from 1 cat, 115 surrounded GABA-immuno-reactive somata, and 2 did not. Both were in layer III.


at OC

LC

on Novem


Dow

nloaded from


Figure 4 . Camera lucida drawings of 54fT-immunoreactive basket formations from layer II | f l ) , layer III (A 0 - f ) , layer IV (C), layer V (G). and layer VI \H. I) that were firstanalyzed in serial semithin plastic sections, and then in selected series of thin sections. All except baskets A. B, and H were resectioned for electron microscopy. The enclosedsomata of the target cells (verified in semithin sections) are represented by stippled profiles. Scale bar, 20 ^m.

In all layers of AI, the number of GABA neuronssurrounded by 5-HT-immunostained baskets is muchsmaller than that unapposed by 5-HT-immunoreac-tive axons. In many cases, a cluster of 3 or 4 GABA-immunoreactive somata includes one that is sur-rounded by a type II 5-HT axon while the others arenot (Fig. 5A,B). Less commonly, 2 closely adjacent

immunoreactive somata are apposed to one 5-HT axon(Fig. 5CD).

Although less numerous than baskets, isolated seg-ments of the type II 5-HT axons are also apposed tothe surfaces of relatively thick, GABA-immunoreac-tive processes resembling dendrites. Examples of iso-lated type II 5-HT axons apposed to GABA-immu-


at OC

LC

on Novem


Dow

nloaded from


H g n n 6. Pain of ftuoretcsnca photomicrographs showing GABA and 5-HT atocaliflmun. A and B. A type II fiber (B) that forms a basket along the soma and a pnxdmaldendrite ( a rms) of a GABA-immrareactive neuron in layer IL Several adjacent GABA-immunoreactive toman are not endosed by 5-HT MOTS. In this and all other pairs ofmicrographs, the 5-HT FTTC immunoflunestence was plmutf aphed with rhodsnine barrisr fflter removed to alow simJtaneous detection of the GABA immunoftuorescence. CandD. A single 5-HT a m comes in dose apposition to 2 GABA neurons in layer III. Patches of 5-HT imnuweactivtty on the Iowa red are terminals. E and F. A comptei 5-HT-iinmnntaimd baslcet that is in proximity to the soma and proximal dendrites of a GABA-iinmunoreaciive neuron in layer II. Anvwhesds imfcaie the same Mood vessel m the 2lukaugraplft. G and H. Type II 5-HT-iiinmiudUive fibers that are in dose prtnimity to 2 GABA neurons in tever I and also give rise to vancosilies \snvws\ that are not associatedwith GABA-immunoreactive elements. Scale bar 25 /im for A and B. 15 iim for C-H

nostained processes are common throughout layers Iand II and, though relatively rare in the deeper layers,can be found even in the white matter subjacent toAI. In the white matter they abut on the somata ordendrites of the scattered GABA-immunoreactiveneurons found there. Other isolated segments haveno apparent association with GABA neurons (Fig.5G,fi).

Electron MicroscopyAll basket formations illustrated in Figure 4 were an-alyzed first in serial semithin (2.5-Mm-thick) plasticsections (Figs. 6, 7), and then selected semithin sec-tions (n = 10) were resectioned for electron micros-copy. Baskets A, B, and H (Fig. 4) were studied onlyat the light microscopic level.

In semithin sections, it is clear that each basket


at OC

LC

on Novem


Dow

nloaded from


• • • ' • * 1 » . '•'

"•&2^&* -W \

• i i - .-•>.. - > * < *:•

R g a r e 6. Correlative Tight and electron microscopy of the 5-HT-immunoreaaive basket formation of Rgtn 41 A Photomicrograph of a 2.5-ftnHrack plastic section throughthe middle portion of the basket Immunoreactive boutons are seen around a principal cell (n) and stnaBer. companon caD|s) (5). B, Electron micrograph after resectioning thesemrtrun section illustrated in A The principal cefi [rii is a neuron, and the companion celj are dqadendrocytes [g. g1). C. Higher nimpiilitaiion of B tram a dHlerent sectorshowing 3 (1-3) immunoreacovt boutons around the rdiuudaaJjocytes. D, Higher magnification of bouton 3 in C showing a dose membrane apposition (arrow) tupsrfjoalty rejernbfmga symmetrical synaptc contact f, LovHtagnification electron micrograph from a serial section, showing boutons 4 and 5 around one of the ougodsndrocytes. F. Higher magnificationof boutons 4 and 5 from another sera) section, showing bouton 5 making a d o » apposition [mow] similar to that in D with the oligodendrocyte. Scale bars: A 10 nm B, 5/un; C and £ 1 jim; D and F. 0 2 5 /im.

formation surrounds a single cell or a group of 2 or3 cells (which can be identified as faint backgroundghosts; Fig. 6). In the latter case, 1 cell is always muchlarger, and we call it the "principal" cell. At the elec-tron microscope level, the principal cell can alwaysbe identified as a neuron, and the other cells of thegroup as neuroglial cells, usually oligodendrocyr.es(Fig. 6). The somata of all the principal cells receive

both symmetric and asymmetric synapses formed byunstained terminals. They are therefore consideredto be nonspiny, nonpyramidal neurons (see Petersand Jones, 1984a). They could not, however, be fur-ther identified as a particular morphological type ofinterneuron on the basis of ultrastructural character-istics of their somata alone.

Reconstructions were made from serial electron

124 Serotonin Baskets on GABA Cells • DeFellpe et al.

at OC

LC

on Novem


Dow

nloaded from


F t g o n 7. Correbtive Hghi and dectnn mcroscopy of the S-HT-uuiuuuuive basket tnmaton shown in Figure 4G. 4 flimjiuuugrapti of the basket taken through thebbck. Arrows infic8ie the sanv portion of the basket seen in B. 8 and b are two 5-HT-uinuuBdUfve boutors of the basket that ara also mdcated in B snd £ B, Photomicrograph81 the same rnsgrufcatjon of a samnhn section through approximately the same focai plane as in A The apparently larger size of the biood vessel [by] in A in companson whhB \s due to the optical affect of photographing through the block. C, Uiw-rnaQniDcaiion electron nucrograph after fgw?ctioiw^| the stuiidui section B. 0, Electron mcrographcorresponding approxnTBtely to the region nficated between the ffwws in A snd B. Note that the tmrnunoreacuve boutons are rfi^pffffrt aromj a dendntic segment \d) belongingto the c d enclosed by the basket. £ Higher magnification of boutons » and b. Note the large size of the immunaeauive boutons in m n u ' s o n whh adjacent nonimmunoreactiveboutons (fiar). Anvw nficates an intervaricose i i i iuuuacove segment. Scale bars: A and B. 20 iinc C. S M T 0. 2.5 im; E. 0.5 nm.

micrographs of 135 5-HT-itnmunoreactive axon ter-minals. Series consisted of 20-30 thin sections witha thickness of 60-70 nm. The axon terminals of thebasket formations are packed with numerous roundvesicles 40-60 nm in diameter (Figs. 6-8). Many axonterminals contain, in addition, 1, 2, or more large(100-120-nm) dense-cored vesicles. Some of the im-munoreactive axon terminals are very large (up to 3

urn), 2 or 3 times larger than the other nonimmu-noreactive, perisomatic boutons. Of the 135 immu-noreactive boutons that were serially reconstructed,110 belonged to basket formations, 14 to type I fiberslocated in the cortex, and the remaining 11 to typeII fibers located in the white matter.

Approximately 85-90% of the nonimmunoreactiveperisomatic boutons on cells enclosed by the 5-HT


at OC

LC

on Novem


Dow

nloaded from


-It- •- *

JtV V... --y3..% -. •*

Rgnra B. Electron micror/aphs of 54ff-<mnunraactive boutons belonging to basket formations. A and B, Two nonconsecutiw sections from a serial showing a 5-HT-immunoreacnw bouton establishing in A B dose membrsrt apposition that superficudy resembles a symmetrical synaptk contact [smm] with the soma (n) of the enclosed neuronIn A the same boinon is seen making a similar apposition [snow] with a nonimmunoresarvB aion terminal (sr). C, A higher magnification from a serial section of Figure 70.showing an imraunoreactive bouton forming a dose contaa that resenttes a symmetrical contact (arrow) with an amn terminal (or). 0 and £ Micrographs of the same boutonin the same section viewed with a different t i t angle using the gonioraeter stage of the electron nioouape. In D the trh angle of the specimen is 0°. and in f h is 35°. In 0the contact zone {am*) appears like a posBynapn: density, while in f and in the higheHimgniStation cnser the apparent contact zone is seen to be formed by an SMeSnedtubular structure. Scale bars: A-C. O 2 n m ; 0 8 n d £ 0,4 jure imst 0.1 pm.

baskets form overt synaptic contacts. The membranethickenings of these synapses could invariably be seenin at least 3-7 sections of a series (average, 4.8). Bycontrast, of the total of 135 5-HT-immunoreactiveboutons reconstructed, only 4 made synaptic contactsof comparable morphology and size (Figs. 9B, 1OD-F). All others were equivocal (see below). The 4 overtsynapses were of the asymmetrical type. Three of these

synaptic contacts belonged to the same basket for-mation (Fig. 95), and the other synaptic contact be-longed to a type I fiber in the neuropil (Fig. 1OD-/0-

Nonconventional Membrane ContactsVery frequently in the baskets, the apposed mem-branes of the 5-HT-immunoreactive boutons and the

12t Serotonin Baskets on GABA Cells • DefeUpc et al.

at OC

LC

on Novem


Dow

nloaded from


fl 9 . A Etectron rttotJgraph of an mn'munal iw bouton of a basket formation faming a dose mentrara apposition {maw} with da ce8 body of the andossd csfl. Theapparent ^Jticuriostion appsaftxl only in dvs saction. A Etectron ihuuyiuph of an urnijnonJacnve bouton of a basket forroatHn formihg a convanoonal asyinniBtncai tynripttccomaci (arrow) with a dendrita of de a r t e e d cell. Thk contact was viable in 5 cortseanivfl terial seoiora. Scale bars. 0 2 5 jim in A snd fl

adjacent nonlmmunoreactive profiles are darkened ina manner that superficially resembles a synaptic con-tact (Figs. 6,8,10). The most common of these mem-brane irregularities resembles the symmetrical typeof synaptic contact (Figs. 6'D, 8A~d). They are char-acterized by a parallel, close membrane appositionwith an intermembrane deposit of electron-dense ma-terial. However, there are major differences betweenthese and the conventional symmetrical synapticcontacts seen to be made by nonlmmunoreactive ter-minals in the same sections: (1) they lack any elec-tron-dense material on the cytoplasmic face of themembrane of the putative postsynaptic element (Fig.8D,E); (2) they are rarely seen in more than 1 or 2serial sections; and (3) they are formed on any adja-cent element, including neuroglial cells (Fig. 6C-F)and other axon terminals (Fig. 8.6,(7). A smaller num-ber of membrane appositions superficially resemblingasymmetrical synaptic contacts are often seen to bepuncta adherentia, but in the other cases it is difficultto characterize them as any form of known junctionalspecialization. Most consist of a small deposit of elec-tron-dense material overlying the cytoplasmic face ofthe membrane of the nonimmunoreactive profile (Figs.9A, \0A-C). This is usually visualized in only 1 or 2serial sections and often resembles artifactual HRPreaction product (Fig. KM-C1). In a number of cases,the goniometer tilt control of the electron microscopewas used, and the putative membrane thickeningswere found to be ill-defined, tubular structures lo-cated near the cytoplasmic face of the membrane ofthe nonimmunoreactive profile (Fig. 8D,E). Three ofthe 110 serially sectioned 5-HT-immunoreactive ter-minals in the baskets made unambiguous asymmet-rical contacts (Figs. 9B, 11); 39 of the 110 terminalsshowed some evidence of ephemeral or ill-definedcontacts that were not conventional synaptic contacts;68 of the 110 terminals showed no evidence of anymembrane specialization.

DiscussionThe present study revealed 4 major findings: (1) each5-HT-immunoreactive basket formation encloses thecell body and proximal dendrites of a principal neu-ron and, frequently, 1 or 2 neuroglial cells; (2) theenclosed neuron is almost invariably a GABA-im-munoreactlve cell; (3) serial-section analysis of thebasket formations showed that the axon terminalsrarely form conventional synaptic contacts (3 out of110 boutons); and (4) when a rare membrane spe-cialization could be unquestionably identified as aconventional synaptic contact, it was of the asym-metrical type.

Target CeOs of the Basket FormationsThe principal cells enclosed by the 5-HT-immuno-reactive baskets possess light and electron micro-scopic features of nonpyramidal neurons, which thedouble-labeling immunofluorescence observationsshow to be GABA neurons. GABA neurons in the ce-rebral cortex represent a variety of morphological typesthat differ in their axonal ramifications and synapticterminations (see Peters and Jones, 1984b). The pres-ent study was confined to the somata and dendritesof the GABA neurons in the baskets, which does notallow us to identify a specific morphological class thatis innervated by the basket formations. However, thevariety of shapes, sizes, and laminar positions of theneurons innervated by the baskets suggests that morethan one morphological type of GABAergic neuroncould be the target of the 5-HT baskets. However, thesmall percentage of the total population of GABA-immunoreactive somata that are enclosed by the 5-HTbaskets makes it doubtful that all the members of anyone class would be surrounded by the immunoreac-tive axons.

The close apposition of the 5-HT baskets and theGABA somata and dendrites occurred with very fewobvious synaptic specializations, raising the possibil-


at OC

LC

on Novem


Dow

nloaded from


- • •#»

B

R g m 1 0 . Serial electron mieroarapht of bouton a shown in Figure 1A-C and of a second immmeaaive bouton belonging to a type I fiber (0. F). A-C, Single arrow inA and B nftcste a doss apposhton between the immunoreactive bouton and the soma of the enclosed cell, which superficiafiy resembles an asymmetncai synsptic contact butwas visualized only n these 2 sections. Conventional asymmetrical synaptic contacts [douiis arrows), however, are more dscrete and visualized over several sections of the series.O-f. Nonconsecutive sections from a series to show a conventional asymmetrical synaptic contact (arrow in 0 end F] formed between a 5-HT bouton and a dendritic profile. Notethat the posuynaptic density is discrete and cteariy seen in all sections. The arrowhead in f indicates a typical subsynaptic bar (Peters et al.. 1976). Scale bar, 0.5 imtot A-F.

ity that nonsynaptic release and binding of 5-HT couldoccur in the vicinity of the baskets.

The basket formations also surround neuroglialcells, which are thus in close apposition to the vesicle-filled, 5-HT-immunoreactive terminals. Glial cells inother regions of the brain have been shown to possessreceptors for several neurotransmirters, including 5-HT(Hertz et al., 1979; Tardy et al., 1982; Cambray-Deakin

et al., 1985), and may be targets for these transmitters(e.g., Murphy and Pearce, 1987; Stone and Ariano,1989). Application of 5-HT leads to a change in Ca2+

conductance (Lazarewicz and Kanje, 1981; Sugino etal., 1984) in cultured neuroglial cells and to hyper-polarization of glioma cell membranes (Ogura andAmano, 1984). In cat Al, the numerous close appo-sitions between the 5-HT baskets and glial cells would

l i t Serotonin Baskets on GABA Cells • DeFelipe et al.

at OC

LC

on Novem


Dow

nloaded from


n 3 9 •o < z

Figu

re 1

1.

Reco

nstru

ctio

n of

the

5-HT

bas

ket f

orm

atio

n sh

own

in F

igure

s 4

end

7 m

ade

from

3 c

onse

cutiv

e se

ries

each

of

10se

rial

thin

sec

tions

. Th

e bo

uton

s sh

owin

g re

cons

truct

ion

lines

are

imm

unor

eact

ive,

whi

le t

he o

pen

bout

ons

are

noni

mm

unor

eact

ivepe

ricel

lula

r te

rmin

als.

/,

Imm

unor

eact

ive b

outo

ns t

hat

mak

e co

n-ve

ntio

nal a

sym

met

rical

syn

aptic

con

tact

s w

ith th

e en

close

d ce

ll; 2

.im

mun

orea

ctive

bou

tons

tha

t fo

rm n

onco

nvem

iona

l m

embr

ane

ap-

posit

ions

with

the

enc

lose

d ce

ll; 3

, im

mun

orea

ctive

bou

tons

tha

tfo

rm n

onco

nvem

iona

l ap

posit

ions

with

in t

he n

euro

pil;

4, i

mm

uno-

reac

tive

bout

ons

that

for

m n

onco

nvem

iona

l ap

posit

ions

with

bot

hth

e en

close

d ce

ll an

d th

e ne

urop

il. T

he s

oma

of t

he e

nclo

sed

cell

rece

ives

16 c

onve

ntio

nal s

ynap

tic c

onta

cts:

5 o

f th

e as

ymm

etric

alty

pe, 8

of t

he s

ymm

etric

al t

ype,

and

3 s

ynap

tic c

onta

cts

that

wer

eno

nide

ntifi

ed. S

cale

bar

, 5

itm.

at OCLC on November 5, 2014http://cercor.oxfordjournals.org/Downloaded from


appear to make these also likely targets for 5-HT ac-tions.

The Nature of Contacts Formed by theBasket FormationsChemical synapses are distinguished by the presenceof oven morphological specializations at points ofmembrane contact between neurons (Peters et al.,1976). Several well-established morphological crite-ria are required before a vesicle-containing profile canbe regarded as forming a conventional synaptic con-tact (Colonnier, 1968; Peters et al., 1976): (1) pres-ence of parallel membrane thickenings on the cyto-plasmic faces of both the vesicle-containing profileand its target element; (2) clustering of vesicles atthe membrane specialization of the vesicle-contain-ing profile; (3) the presence of intermembrane (syn-aptic cleft) electron-dense material; and (4) wideningof the intercleft space. These morphological featuresare usually seen consistently in each of a series ofultrathin (60-70-nm) sections (Figs. 10 and 11) overdistances of 200-400 nm. In the cerebral cortex, thereare 2 morphological types of synaptic contacts (Gray,1959; Colonnier, 1968): type I or asymmetrical, whichis characterized by thick postsynaptic densities, andtype II or symmetrical, which is characterized by thinpostsynaptic densities. In both cases, the postsynapticdensity consists of a filamentous or "fuzzy" electron-dense material.

It is generally anticipated that any vesicle-contain-ing varicosity or bouton will make at least 1 morpho-logically distinct synaptic contact; this was true forthe great majority of the nonimmunoreactive, periso-matic boutons in the present study. Results from thepresent and previous studies (Papadopoulos et al.,1987; DeFelipe and Jones, 1988; de Lima et al., 1988;Seguela et al., 1989) have shown conclusively that acertain number of 5-HT-immunoreactive boutonsmake conventional synaptic contacts of the asym-metrical type. However, a concerted attempt by serial-section analysis revealed that the 5-HT-immunoreac-tive boutons of the basket formations made very fewcontacts that could be characterized as conventionalsynapses. The lack of obvious synaptic contacts in themajority of the varicosities that make up the basketformations was surprising. Other forms of cortical ter-minals when stained immunocytochemically (e.g.,GABAergic terminals) are invariably found to formsynaptic contacts, even though they are of the sym-metrical type, which is often more difficult to visualize(Freund et al., 1983; DeFelipe et al., 1986). The easewith which a few 5-HT axons in the present studycould be demonstrated to form synapses strongly sug-gests that the lack of conventional contacts arisingfrom the 5-HT baskets is not an artifactual result ofthe fixation, processing, or histochemical reaction usedin the immunostaining procedure.

Although many 5-HT terminals in the baskets couldnot be found to make distinguishable contacts of anykind, commonly in 1 or 2 sections of a series througha 5-HT terminal, a membrane structure was seen thatcould be interpreted as a symmetrical synaptic con-

tact. This, in itself, is unusual because the few distinct5-HT contacts seen were definitely asymmetrical.Moreover, close examination of the putative mem-brane specializations revealed that they could beformed between 5-HT axons and any element of theneuropil, including neuroglial cells and other, un-stained axon terminals. Furthermore, the membranespecializations were not typical of symmetrical syn-apses and often appeared to be deposits of reactionproduct that are commonly observed in sections im-munostained by peroxidase-based localization meth-ods (see also DeFelipe and Jones, 1988). The non-synaptic nature of these deposits was frequentlydetermined with the use of the goniometer stage ofthe electron microscope. We conclude that, in thepresent material from cat Al, most boutons in thebasket formations and in the surrounding neuropildo not form conventional synaptic contacts. Differ-ences among reports in the proportion of axon ter-minals forming conventional synapses may arise fromthe differences in the methods used to localize 5-HTand from the criteria used to define synapses.

The release of 5-HT is regulated by several mech-anisms (Balfour, 1980; Bowery et al., 1980), and thecontents of boutons and other elements may be rap-idly depleted by activity or under other conditions(Shaskan and Snyder, 1970). If sufficiently reduced,the contents of certain axons or components of axonscould escape detection by immunocytochemistry.Thus, even minor differences in the handling andpreparation of the experimental material could elim-inate certain populations of elements from subse-quent analyses and bias the results toward examina-tion of either those elements that form conventionalsynapses or those that do not. However, when atten-tion is restricted to the 5-HT axons forming basketformations onto cortical GABA neurons, the situationis one in which all components of the axons sur-rounding the somata are immunoreactive, yet very fewform what is, by conventional criteria, a synaptic con-tact.

The question of the extent to which monoaminer-gic terminals make recognizable synaptic contacts inthe cerebral cortex is controversial. Some workers havenoted the almost ephemeral nature of membrane den-sities formed at points of close apposition betweenterminals and other profiles and have stressed thelikelihood of a nonsynaptic form of transmitter re-lease (see Beaudet and Descarries, 1984). Others haveconcluded that small points of contact with mem-brane thickenings that appear in only 1 or 2 serialthin sections should be regarded as release sites andhave therefore stressed the ubiquity of monoaminesynapses (e.g., de Lima et al., 1988). Others havestressed that all monoamine terminals form overt,conventional synapses (Parnavelas and Papadopou-los, 1989) • The small size and even lack of membranethickenings at the points of close apposition between5-HT-immunoreactive terminals and the GABA cellsin the baskets do not resemble other cerebral conicalsynapses. However, the intimate nature of such ap-positions would support the idea that localized re-


at OC

LC

on Novem


Dow

nloaded from


lease sites could occur without overt membrane spe-cializations.

Functional SignificanceAlthough the predominant effect of 5-HT is inhibitionof cell activity in neurons of the cerebral cortex (e.g.,Olpe, 1981), both inhibitory and excitatory corticalcell responses to iontophoretically applied 5-HT orelectrical stimulation of the midbrain raphe nucleihave been reported (e.g., Krnjevic and Phillis, 1963a,b;Roberts and Straughan, 1967; Sastry and Phillis, 1977;Reader, 1978; Olpe, 1981). Olpe (1981) suggested,as one possibility, that the excitatory effects obtainedafter microiontophoretic administration of 5-HT mightbe mediated indirectly through inhibition of an in-hibitory interneuron. Recently, Sheldon and Aghaja-nian (1990) hypothesized that the IPSPs produced by5-HT in layer II pyramidal cells in an in vitro prep-aration of rat piriform cortex were mediated by acti-vating a subpopulation of putative GABA interneu-rons located at the border of layers II and III. Thesefindings suggest a close coupling between 5-HT af-ferents and GABA interneurons. An anatomical cor-relate became apparent with the first demonstrationof 5-HT baskets surrounding nonpyramidal neuronsin the cat (Mulligan and Tork, 1988) and marmoset(Hornungetal., 1990) cerebral cortex; the GABAergicnature of those neurons was confirmed in the presentstudy (see also Tork et al., 1988). Thus, 1 of the 3classes of 5-HT axons in the cerebral cortex has GABAneurons as its major target, and if transmitter releaseoccurs at the points of close membrane apposition,then this could provide a basis for the indirect inhib-itory effects of 5-HT produced through these axons.

Regional and laminar variations in the distributionof the fine or small varicose axons (our type I axons)and beaded or large varicose axons (our type II axons)have been reported in the cortex of rats and monkeys(e.g., Morrison et al., 1982,1984; Morrison and Foote,1986; Campbell et al., 1987; O'Hearn et al., 1988;Wilson et al., 1989; Hornung et al., 1990). In rats these2 types of fibers originate from different raphe nucleiand appear to differ in their sensitivity to certain psy-chotropic amphetamine derivatives (Kosofsky andMolliver, 1987; Mamounas and Molliver, 1988; O'Hearnet al., 1988; Wilson et al., 1989; Molliver and Molliver,1990). Both the original description of the 5-HT baskets (Mulligan and Tork, 1988) and the present studyfound that the baskets are formed by a single class oflarge varicose or type II axons. These axons may arisefrom cells in one of the raphe nuclei, while the othertypes of 5-HT axons may originate from different nu-clei. We do not know if the type III axons arise froma third site or are simply the parent fibers of one ofthe others.

Not all species possess cortical 5-HT basket for-mations. They have not been found, for example, inthe neocortex of the rat (Seguela et al., 1989), andthough encountered in the macaque monkey cortex(DeFelipe and Jones, 1988), they rarely form axonalarborizations as complex as the baskets of the cat(Wilson et al., 1989). Even within the cerebral cortex

of cats (Mulligan and Tork, 1988) and marmosets(Hornung et al., 1990), the number of 5-HT basketsdiffers greatly among cortical areas. Some cortical ar-eas, such as those in the temporal lobe of the cat(Mulligan and Tork, 1988; present results) and thefrontal cortex of the marmoset (Hornung et al., 1990),are very rich in 5-HT baskets, while other corticalareas possess very few. It appears, then, that the in-teractions between 5-HT axons and GABA neuronswould be particularly strong within specific areas, suchas cat Al, and not in others.

Within cat Al, GABA neurons are present in alllayers (Winer, 1986), form the total population ofcells in layer I, and are more concentrated in layersII-IV (Hendry and Jones, 1991). That the 5-HT bas-kets target preferentially the GABA cells of layers Iand II and are more elaborate in those layers indicatesthat the close interaction between 5-HT axons andGABA cells varies not only by area, but also by layer.

The preferred dorsoventral orientation of the typeII 5-HT fibers in layers I and II suggests that theseaxons and their basket formations may run parallel toisofrequency bands in cat Al (Merzenich et al., 1975).However, because the 5-HT-immunoreactive peri-odicities are significantly wider than a single isofre-quency band, it appears that both the densely im-munoreactive and lightly immunoreactive regionsoverlap 2 or more bands of cells displaying the samebest-frequency responses. Whether the greater den-sity of 5-HT fibers and baskets in some isofrequencybands and the lesser density in others are correlatedwith physiologically distinct responses other than best-frequency tuning is unknown.

NotesThis work was supported by Grant DCOO45O from NIH, U.S.Public Health Service, and by the Frontier Research Program,Japan.

Dr. DeFelipe is on leave from Instituto Cajal, 28002-Ma-drid, Spain.

Correspondence should be addressed to Dr. E. G.Jones,Department of Anatomy and Neurobiology, California Col-lege of Medicine, University of California at Irvine, Irvine,CA 92717.

ReferencesBalfour DJK (1980) Effects of GABA and diazepam on 3H-

serotonin release from hippocampal synaptosomes. EurJ Pharmacol 68:11-16.

Beaudet A, Descarries L (1984) Fine structure of mono-amine axon terminals in cerebral cortex. In: Monoamineinnervation of cerebral cortex (Descarries L, Reader TR,Jasper HH, eds), pp 77-93. New York: Liss.

Bowery NG, Hill DR, Hudson AL, Doble A, Middlemiss DN,ShawJ.TurnbullM (1980) (-)Baclofen decreases neu-rotransmitter release in the mammalian CNS by an actionat a novel GABA receptor. Nature 283:92-94.

Cambray-Deakin M, Morrow C, Murphy S (1985) Corticalastrocytes in primary culture: presence of neurotrans-mitter binding sites. Biochem Soc Trans 13:231-232.

Campbell MJ, Lewis DA, Foote SL, Morrison JH (1987) Dis-tribution of choline acetyltransferase-, serotonin-, do-pamine-beta-hydroxylase, and tyrosine hydroxylase-im-munoreactive fibers in monkey primary auditory cortex.J Comp Neurol 261:209-220.

Colonnier M (1968) Synaptic patterns on different cell typesin the different laminae of the cat visual cortex. An elec-tron microscope study. Brain Res 9:268-287.


at OC

LC

on Novem


Dow

nloaded from


Consolazione A, Milstein C, Wright B, Cuello AC (1981)Immunocytochemical detection of serotonin with mono-clonal antibodies. J Histochem Cytochem 29:1425-1430.

DeFelipe J, Fairen A (1982) A type of basket cell in su-perficial layers of the cat visual cortex. A Golgi-electronmicroscope study. Brain Res 244:9-16.

DeFelipe J, Jones EG (1988) A light and electron micro-scopic study of serotonin-immunoreactive fibers and ter-minals in the monkey sensory-motor cortex. Exp BrainRes 71:171-182.

DeFelipe J, Hendry SHC, Jones EG (1986) A correlativeelectron microscopic study of basket cells and largeGABAergic neurons in the monkey sensory-motor cortex.Neuroscience 17:991-1009.

de Lima AD, Bloom FE, Morrison JH (1988) Synaptic or-ganization of serotonin-immunoreactive fibers in primaryvisual cortex of the macaque monkey. J Comp Neurol274:280-294.

Freund TF, Martin KAC, Smith AD, Somogyi P (1983) Glu-tamate decarboxylase-immunoreactive terminals of Gol-gi-impregnated axo-axonic cells and of presumed basketcells in synaptic contact with pyramidal neurons of thecat's visual cortex. J Comp Neurol 221:263-278.

Fuxe K, Hamberger B, Hokfelt T (1968) Distribution ofnoradrenaline nerve terminals in conical areas of the rat.Brain Res 8:125-131.

Goldman-Rakic PS, Leranth C, Williams SM, Mons N, GeffardM (1989) Dopamine synaptic complex with pyramidalneurons in primate cerebral cortex. Proc Natl Acad SciUSA 86:9015-9019.

Gray EG (1959) Axo-somatic and axo-dendritic synapsesof the cerebral cortex: an electron microscopic study. JAnat 93:420-433.

Hendry SHC, Jones EG (1991) GABA neuronal subpopu-lations in cat primary auditory cortex: co-localization withcalcium binding proteins. Brain Res 543:45-55.

Hertz L, Baldwin F, Schousboe A (1979) Serotonin recep-tors on astrocytes in primary cultures: effects of meth-ysergide and fluoxetine. CanJ Physiol Pharmacol 57:223-226.

HornungJP, FritschyJM, TorkI (1990) Distribution of twomorphologically distinct subsets of serotoninergic axonsin the cerebral cortex of the marmoset. J Comp Neurol297:165-181.

Kosofsky BE, Molliver ME (1987) The serotoninergic in-nervation of cerebral cortex: different classes of axonterminals arise from dorsal and median raphe nuclei.Synapse 1:153-168.

Krnjevic K, Phillis JW (1963a) Iontophoretic studies ofneurones in the mammalian cerebral cortex. J Physiol(Lond) 165:274-304.

Krnjevic K, Phillis JW (1963b) Actions of certain amineson cerebral cortical neurones. Br J Pharmacol 20:471-490.

Lazarewicz JW, Kanje M (1981) Effects of neurotransmitterson calcium efflux from cultured glioma cells. J NeurosciRes 6:283-291.

Levitt P, Rakic P, Goldman-Rakic P (1984) Region-specificdistribution of catecholamine afferents in primate cere-bral cortex: a fluorescence histochemical analysis. J CompNeurol 227:23-36.

Mamounas LA, Molliver ME (1988) Evidence for dual se-rotonergic projections to neoconex: axons from the dor-sal and medial raphe nuclei are differentially vulnerableto the neurotoxin p-chloroamphetamine (PCA). ExpNeurol 102:23-36.

Merzenich MM, Knight PL, Roth GL (1975) Representationof cochlea within primary auditory cortex in the cat. JNeurophysiol 38:231-249.

Molliver DC, Molliver ME (1990) Anatomic evidence fora neurotoxic effect of (±)-fenfluramine upon serotoner-gic projections in the rat. Brain Res 511:165-168.

Morrison JH, Foote SL (1986) Noradrenergic and seroto-ninergic innervation of cortical, thalamic, and tectal vi-sual structures in old and new world monkeys. J CompNeurol 243:117-138.

Morrison JH, Foote SL, Molliver ME, Bloom FE, Lidov HGW(1982) Noradrenergic and serotoninergic fibers inner-vate complementary layers in monkey primary visual cor-tex: an immunohistochemical study. Proc Natl Acad SciUSA 79:2401-2405.

MorrisonJH, Foote SL, Bloom FE (1984) Regional, laminar,developmental, and functional characteristics of nor-adrenaline and serotonin innervation patterns in monkeycortex. In: Monoamine innervation of cerebral cortex(Descarries L, Reader TR, Jasper HH, eds), pp 61-75.New York: Liss.

Mulligan KA, Tork I (1988) Serotoninergic innervation ofthe cat cerebral cortex. J Comp Neurol 270:86-110.

Murphy S, Pearce B (1987) Functional receptors for neu-rotransmitters on astroglial cells. Neuroscience 22:381-394.

OguraA.AmanoT (1984) Serotonin receptor coupled withmembrane electrogenesis in a rat glioma clone. BrainRes 297:387-391.

O'Hearn E, Battaglia G, De Souza EB, Kuhar MJ, MolliverME (1988) Methylenedioxyamphetamine (MDA) andmethylenedioxymethamphetamine (MDMA) cause se-lective ablation of serotonergic axon terminals in fore-brain: immunocytochemical evidence for neurotoxiciry.J Neurosci 8:2788-2803.

Olpe H-R (1981) The conical projection of the dorsal ra-phe nucleus: some electrophysiological and pharmaco-logical properties. Brain Res 216:61-71.

Papadopoulos GC, Parnavelas JG, Buijs RM (1987) Lightand electron microscopic immunocytochemical analysisof the serotonin innervation of the rat visual cortex. JNeurocytol 16:883-892.

Parnavelas JG, Papadopoulos GC (1989) The monoamin-ergic innervation of the cerebral cortex is not diffuse andnonspecific. Trends Neurosci 12.315-319.

Peters A, Jones EG (1984a) Classification of cortical neu-rons. In: Cerebral cortex, Vol 1, Cellular components ofthe cerebral cortex (Peters A, Jones EG, eds), pp 107-121. New York: Plenum.

Peters A, Jones EG, eds (1984b) Cerebral cortex, Vol 1,Cellular components of the cerebral cortex. New York:Plenum.

Peters A, Palay SL, Webster HD (1976) The fine structureof the nervous system. Philadelphia: Saunders.

Reader TA (1978) The effects of dopamine, noradrenalineand serotonin in the visual cortex of the cat. Experientia34:1586-1588.

Roberts MHT, Straughan DW (1967) Excitation and de-pression of cortical neurones by 5-hydroxytryptamine. JPhysiol (Lond) 193:269-294.

SastryBSR, Phillis JW (1977) Inhibition of cerebral corticalneurones by a 5-hydroxytryptaminergic pathway frommedian raphe nucleus. Can) Physiol Pharmacol 55:737-743.

Seguela P, Watkins KC, Descarries L (1989) Ultrastructuralrelationships of serotonin axon terminals in the cerebralcortex of the adult rat. J Comp Neurol 289:129-142.

Shaskan EG, Snyder SH (1970) Kinetics of serotonin ac-cumulation into slices from rat brain: relationship to cat-echolamine uptake. J Pharmacol Exp Ther 175:404-418.

Sheldon PW, Aghajanian GK (1990) Serotonin (5-HT) in-duces IPSPs in pyramidal layer cells of rat piriform conex:evidence for the involvement of a 5-HT2-activated inter-neuron. Brain Res 506:62-69.

Stone EA, Ariano MA (1989) Are glial cells targets of thecentral noradrenergic system? A review of the evidence.Brain Res Rev 14:297-309.

Sugino H, Ogura A, Kudo Y, Amano T (1984) IntracellularCa2* elevation induced by a neurotransmitter in a glialcell clone. Brain Res 322:127-130.

Tardy M, Costa MF, Fages C, Bardakdjian J, Gonnard P (1982)Uptake and binding of serotonin by primary cultures ofmouse astrocytes. Dev Neurosci 5:19-26.

Tork I, Hornung JP, Somogyi P (1988) Serotonergic in-nervation of GABAergic neurons in the cerebral conex.Neurosci Lett [Suppl] 3O:S131.


at OC

LC

on Novem


Dow

nloaded from


Wilson MA, Ricaurte GA, Molliver ME (1989) Distinct mor-phologic classes of serotoninergic axons in primates ex-hibit differential vulnerability to the psychotropic drug3,4-methylenedioxymethamphetamine. Neuroscience 28:121-137.

Winer JA (1986) Neurons accumulating [3H]gamma-ami-nobutyric acid (GABA) in supragranular layers of cat pri-mary auditory cortex (AI). Neuroscience 19:771-793.

Cerebral Cortex Mar/Apr 1991, VI N 2 133

at OC

LC

on Novem


Dow

nloaded from


synaptic relationships of serotonin-inmmunoreactive terminal baskets pm gaba neurons in the cat...

Documents