THE JOURNAL OF COMPARATIVE NEUROLOGY 321:501-514 (1992)

GABAergic Neurons in the Mammalian Inferior Olive and Ventral Medulla Detected by Glutamate Decarboxylase Immunocytochemistry BARBARA J. FREDETTE, JOE C. ADAMS, AND ENRICO MUGNAINI Laboratory of Neuromorphology, The University of Connecticut, Storrs, Connecticut 06269-4154 (B.J.F., E.M.); Massachusetts Eye and Ear Infirmary, Eaton-Peabody Laboratory, Boston, Massachusetts 02114 (J.C.A.)

ABSTRACT Neurons containing glutamic acid decarboxylase (GAD) (presumed GABAergic neurons) were mapped by immunocytochemistry in the ventral medulla of rat, rabbit, cat, rhesus monkey, and human, with emphasis on the inferior olive. In all species, three categories of GABAergic neurons were identified: periolivary neurons in the gray matter and the white matter surrounding the inferior olive, internuclear neurons located in the white matter between the subnuclei of the inferior olive, and intranuclear neurons located within the olivary gray matter. The intranuclear GABAergic neurons of the inferior olive had a characteristic morphology which differed from non-GABAergic olivary neurons; they were usually smaller, and, wherever their processes were stained, they had radiating, sparsely branching dendrites. They were also usually distinguished from the other GABAergic neurons by their smaller size. The intraolivary GABAergic neurons constituted only a minor proportion of the total olivary neuronal population, but they were concentrated in regions of the olive that varied by species. In the rat, they were situated in the rostral tip of the medial accessory olive and in the caudal subdivision of the dorsal accessory olive, while in the rabbit, they were located in the caudal two-thirds of the medial accessory olive, in the dorsal cap, and in the ventral lateral outgrowth. Such neurons were extremely rare in the cat; only a few were found in the rostral parts of the principal olive, the medial accessory olive, and the dorsal accessory olive. In the rhesus monkey, the principal olive and the lateral region of the rostral medial accessory olive contained most of the intranuclear GABAergic neurons, but some were also present in the dorsal accessory olive. In the human, such neurons occurred in the principal olive, the dorsal accessory olive and the rostral medial accessory olive, but as in the rhesus monkey, most were observed in the principal olive. c-1992 Wiley-Liss, Inc. Key words: GAD, inhibition, local circuit neurons, human inferior olive

The entire mammalian inferior olive (10) contains a remarkably dense GABAergic innervation (Nelson and Mugnaini, '88; Nelson et al., '89; Oertel et al., '82; Sotelo et al., '861, which is clearly revealed by immunostaining with an antiserum to the synthetic enzyme for GABA, glutamate decarboxylase (GAD) (Oertel et al., '81b). Thus, GABA neurotransmission, which is generally inhibitory (Kmjevic and Schwartz, '76), must have a global effect on climbing fiber activity. The GAD immunostaining pattern in the I 0 is heterogeneous, suggesting that GABAergic neurons of several brain sources project to different olivary regions. Indeed, GABAergic projections to the I 0 from the cerebellar nuclei in rat and cat (Angaut and Sotelo, '87, '89; Buisseret-Delmas et al., '89; deZeeuw et al., '88a,b; Fre-

o 1992 WILEY-LISS, INC.

dette and Mugnaini, '91; Nelson and Mugnaini, '85, '89; Nelson et al., '84, '86) and the vestibular nuclei and the nucleus prepositus hypoglossi in rat and rabbit (Barmack et al., '89; deZeeuw et al., in preparation; Fredette and Mugnaini, '91; Nelson and Mugnaini, '89) have been experimentally demonstrated. The protocols of most previous immunocytochemical studies were designed to discern the pattern of GABAergic innervation and did not reveal GABAergic cell bodies in the I 0 (Nelson and Mugnaini, '88; Nelson et al., '89; Sotelo et al., '86). However, a number of Accepted March 24,1992. Address reprint requests to Enrico Mugnaini, Biobehavioral Sciences, Box U-154, The University of Connecticut, Storrs, CT 06269-4154.

B.J. FREDETTE ET AL.

502

Golgi, Nissl, and electron microscopic studies have uncovered some heterogeneity of neuronal size and dendritic morphology within the gray matter and the intervening white matter of the I 0 territory (Cajal, ’11; Foster and Peterson, ’86;Gwyn et al., ’77; Ramon-Moliner, ’62; Rutherford and Gwyn, ’80; Scheibel and Scheibel, ’55;Sotelo et al., ‘74; Szteyn, ’88; Taber, ’61). This raises the possibility that populations of local or regional interneurons exist, and some pilot studies from our laboratory and others support the notion that some of these interneurons are GABAergic, and therefore, presumably provide an additional inhibitory input to the principal neurons of the I 0 (Nelson and Mugnaini, ’89; Nelson et al., ’88; Walberg and Ottersen, ’89). This study demonstrates the distribution and cytological characteristics of a relatively small number of GAD-positive neurons within the boundaries of the I 0 in rat, rabbit, cat, rhesus monkey, and human. The findings are consistent with the notion that extraolivary sources supply most of the GAl3Aergic innervation in the 10.

MATERIALS AND METHODS Under deep sodium pentobarbital anesthesia, a group of adult animals consisting of 11 Sprague-Dawley rats, two Dutch-belted rabbits, and 4 domestic cats received 0.5-4.0 p1 topical injections of colchicine (10 pg/ p*.1saline) into the brain stem near the 10. The injections were made by a dorsal approach with a stereotaxically guided microsyringe or by a ventral approach under visual guidance. After a 24-48 hour survival, the animals were re-anesthetized and perfused transcardially with a vascular rinse of 0.9% NaCl followed by a formaldehyde-zinc salicylate fixative (pH 6.0-6.5), as detailed elsewhere (Mugnaini and Dahl, ’83). One hour after termination of the perfusion, the brain stems were dissected out of the cranium, cryoprotected in 30%sucrose in saline, and sectioned coronally at 20 pm on a freezing microtome. A group of untreated adult animals (consisting of 1rat, 2 rabbits, 1cat, and 1rhesus monkey) was fixed by perfusion as described above. The rat, rabbit, and cat brain stems were sectioned coronally on a Vibratome at 15-20 pm immediately after fixation. The monkey brain stem was cryoprotected in 30% sucrose in saline and sectioned coronally on a freezing microtome at the 25 pm setting. Brain stem sections from the colchicine treated and untreated groups of subjects were processed for GAD immunocytochemistry by a previously described method Abbreviations aMA0

beta bMAO CMAO DAO dc dfDA0 dmcc dlPO I0 MA0 PO rMAO

RF RP vfDA0 vlo VlPO

subnucleusa of MA0 beta nucleus subnucleusb of MA0 subnucleuse of MA0 dorsal accessory olive dorsal cap dorsal fold of DAO dorsomedial cell column dorsal lamella of PO inferior olivary complex medial accessory olive principal olive rostral lamella of MA0 reticular formation raphe pallidus ventral fold of DAO ventrolateral outgrowth ventral lamella of PO

TABLE 1. Sizes’ of GAD-Positiveand GAD-NegativeSomata in the Mammalian I 0 GAD-positive

(w2)

Rat2 Rabbit Cat2 Monkey2 Human2

190.8 f 193.0 ? 199.0 ? 201.8 259.7 2

*

10.69 ( 7 ) 6.42 ( 8 ) 4.29 (4) 11.71 ( 7 ) 21.93 (9)

GAD-negative bm? 247.1 ? 18.17 (71 207.7 ? 8.55 ( 8 ) 469.8 ? 42.01 (4) 561.6 42.13 17) 704.2 i 53.11 (9)

‘A relative measure of area was obtained from the product of long and short somata1 diameters. Values are means 2 standard errors. N in parentheses. Values of GAD-positive neurons are significantly different from immunonegative neurons in the same species (P= 0.02 for rat, P < 0.002 for cat, monkey, and human; Student’s t-test), with the exception of the rabbit.

(Mugnaini and Oertel, ’85; Nelson and Mugnaini, ’88). Briefly, an antiserum raised in sheep against rat brain GAD (Oertel et al., ’81b) was used at a 1:2,000 dilution with the double cycle-PAP method (Ordronneau et al., ’81; Sternberger, ’86). Tris-HC1 (0.5 M, pH 7.6) was the diluent and rinsing solution between steps throughout the procedure. Human brain stems were obtained at autopsy from three patients who had died of non-neurologic diseases. Two of the human medullas included the entire inferior olive: one medulla was obtained from a 58 year old male 10 hours after death during heart transplant surgery, and the other was obtained from an 83 year old male 3%hours after death resulting from a myocardial infarction. In the third case, only the rostral portion of the medulla was obtained from a 30 year old female who had died of uterine hemorrhage 12 hours earlier. In sections from this last case, illustrated in Figure 6, GAD-like immunoreactivity in the neuropil was minimal, compared to other human brain stems utilized in another study (Nelson et al., ’89). Moreover, the silverenhanced immunoreaction product highlighted not only the cell bodies and main stem dendrites of small neurons in the 10, which is commonly obtained, but also their distal dendrites, which is afforded only occasionally. Presumably, the interval between death and fixation, the size of the tissue blocks, and other uncontrolled variables of the stainingprocedure play a role in the degree of immunostaining of GAD-positive cells in the human brain. Thick slices of the human tissue were fixed by immersion for 1hour in cold formaldehyde-zinc salicylate (pH 6.5) and then overnight in formaldehyde-zinc dichromate (pH 5.0) at 4°C. The slices were cryoprotected in 30% sucrose in saline, and cut coronally at 40 pm on a freezing microtome. Sections containing the I 0 were processed for GAD immunocytochemistry with the GAD antiserum (1:2,000) by the avidin-biotin method (Vectastain ABC Kit, Vector Laboratories), and then they were treated with a Protargol-based silver enhancement method, which remarkably increased

Fig, 1. GAD immunostaining in the ventral medulla of a colchicinetreated rat. A Scattered GAD-positive neurons are present within the boundaries of the I 0 in the olivary gray matter (“intranuclear” neurons, closed arrows) and in the white matter between olivary subnuclei (“internuclear” neurons, arrowhead), while many more immunostained cells are observed in the surrounding reticular formation and in the raphe (“periolivary” neurons, open arrows). Inset: Internuclear GAD-positive neuron (arrowhead), located between DAO and PO, extends processes (crossed arrows) into the olivary gray matter. B: Small GAD-positive neurons are present at high density in the rostral tip of the MAO. Two of these neurons are indicated by closed arrows. Some GAD-positive neurons (open arrows) are also present in the raphe pallidus.

GARAERGIC! NEURONS IN MAMMALIAN I 0

503

Figure I

504

Fig. 2. GAD-positive neurons in the rabbit 10.A: Low magnification view of the rostra1 I 0 from a colchicine-treated rabbit shows GAD-positive neurons in the periolivary reticular formation (open arrows), and in the white matter between olivary subnuclei (arrowheads). Inset: An internuclear GAD-positive neuron (arrowhead)

BJ. FREDETTE ET AL.

extends dendrites (crossed arrows) towards the gray matter of the MAO. B: The caudal subnuclei of the MA0 contain several GADpositive neurons, some of which are indicated by closed arrows. Arrowhead points t o an internuclear neuron located between DAO and MAO.

Fig. 3. Periolivary GAD-positiveneurons in the rabbit (A, colchicinetreated; B, C, untreated). A: Immunostained neurons (open arrows) located dorsally in the reticular formation, ventrally in the white matter of the pyramidal tract, and laterally in a cluster beside the PO. B: High magnification photomicrograph of GAD-positiveneuron (closed

arrow) and GAD-negative neuron (asterisk) in the gray matter of the MAO. Some of the immunostained boutons abut the GAD-positive neuron. C: Large, elongated internuclear neuron (arrowhead) situated between DAO and PO. Note the immunostained boutons closely apposed to the cell body and main stem dendrites.

506

B.J. FREDETTE ET AL.

Figure 4

GABAERGIC NEURONS IN MAMMALIAN I 0 the signal in GABAergic somata (for details, see Adams and Mugnaini, ’87). Characterization of the GAD antiserum and control experiments demonstratingthe failure of nonimmune sheep serum to stain brain sections of species used in this study have been presented in several previous publications (Chang and Gottlieb, ’88; Kaufman et al., ’86; Mugnaini et al., ’82; Nelson et al., ’89; Oertel et al., ’81a). After the immunoreaction, sections were either wetmounted and coverslipped in a glycerol-phosphate buffer solution (1:3,pH 8.5), or dry-mounted and coverslipped in Permount. The locations of immunostained neurons were mapped with a microscope drawing tube. In order to compare the relative sizes of GAD-positive and GADnegative somata, an estimate of somata “area” was obtained from the product of the longest and shortest diameters. GAD-negative somata were viewed in immunostained sections by background contrast enhancement with an interference filter. The diameters were measured in the microscope with a calibrated ocular micrometer and a x40 objective lens, and the resulting sizes of GAD-positive and GAD-negative somata were compared with a Student’s t-test (Table 1). Neuronal diameters were obtained only from specimens not treated with colchicine, since abnormal somatal sizes may result from disruption of microtubules.

RESULTS Technical aspects In an effort to visualize GAD-positive neurons in the inferior olive, we have adopted three techniques known to enhance somatal immunostaining: a fixation protocol which favors cell body immunostaining, topical colchicine injections, and a reduced silver enhancement of the immunoreaction product. All specimens were fixed with a zincaldehyde solution at pH 6.0-6.5, a procedure which has been shown to favor somatal GAD-immunostaining in rat (Mugnaini and Dahl, ’83).Most of the rats and cats used in this study were pretreated with topical colchicine injections 24-48 hours prior to perfusion in order to block axonal transport and hence increase somatal concentration of GAD (Ribak et al., ’78). The intensity of GAD-positive neurons was greatly enhanced by colchicine pretreatment. However, the densities and distributions of immunostained neurons were similar in the olives of the treated and untreated rats and cats. Therefore, results from these two groups of animals will be described jointly. Colchicine did result in a decrease of GAD-positive boutons in the 10,due to disrupted anterograde transport of GAD in GABAergic fibers. Immunostained sections from the human brain stem were treated with a silver enhancement procedure, which in one case produced Golgi-like staining of somata, main stem, and peripheral dendrites of GABAergic neurons in the 10,

Fig. 4. GAD-positive neurons in the cat 10. A: Two GAD-positive neurons in the rostral DAO (closed arrows), and one in the reticular formation (open arrow). Inset: Higher magnification of one of the immunostained neuron in DAO (closed arrow). The thick process on the right side presumably represents a dendrite and the thin process on the left side, the initial axon segment. Two immunostained boutons (double arrow) abut the thick process. B: The reticular formation of a colchicinetreated cat contains many large GAD-positive neurons (open arrows), but the portions of the I0 comprised in this section are free of immunoreactive cells. C: Internuclear, GAD-positive,elongated neuron (arrowhead) located between MA0 and PO.

507 accompanied by a fortunate virtual elimination of immunostaining in axonal boutons. Furthermore, some nonspecific staining of nerve cell bodies was produced by the protargol treatment, and this made GAD-negative somata visible (Fig. 6).

Distribution of GAD-positiveneurons within the I 0 Nomenclature for the parcellation of the I 0 used in this study has been presented in separate publications, to which the reader is referred for details (Nelson and Mugnaini, ’88; Nelson et al., ’89). The distribution of GAD-positive cells in the gray matter of the 10,also termed intranuclear GABAergic neurons, varied by species. In the rat I 0 (Fig. l),the majority of these neurons were present in the caudally located dorsal fold of the dorsal accessory olive (DAO), and especially in the rostral tip of the medial accessory olive (MAO) (Fig. 1).A rare cell was also observed in the principal olive (PO) and the rostrally located ventral fold of the DAO. Immunostained neurons were never observed in the beta nucleus, the dorsomedial cell column, the dorsal cap, the ventrolateral outgrowth, or the caudal subdivisions of the MA0 (subnuclei a, 6, and c). In the rabbit I 0 (Figs. 2, 3), most of the GAD-positive cells were contained in the caudal and midolivary regions of the MAO. Labeled cells were also located in the dorsal cap and the ventrolateral outgrowth, but they were rarely observed in other subnuclei. The cat I 0 (Fig. 4) contained an extremely sparse number of GABAergic neurons, most of which were located in the caudal subdivision of the DAO; there were also occasional immunostained neurons in the rostral PO and MA0 (Fig. 4A). As in the rat, the smaller subnuclei (beta nucleus, dorsomedial cell column, dorsal cap, and ventrolateral outgrowth) did not contain GAD-positive somata, nor did the caudal subdivisions of the MA0 (Fig. 4B). In the I 0 of the rhesus monkey (Fig. 5), GAD-positive somata were intensely stained, and their frequency of occurrence varied in the different subnuclei. Most of the immunostained cell bodies were located in the PO and in the lateral aspect of the rostral MA0 at approximately midolivary levels (Fig. 5). Some immunostained cells were also observed in the rostral subdivisions of the DAO. Only an occasional GAD-positive neuron occurred in the remainder of the 10,including the small subnuclei. In the human I 0 (Fig. 61, GABAergic neurons were scattered throughout the PO, the DAO, and the MAO, and as in the rhesus monkey, the caudalmost portion of the I 0 and the small subnuclei were devoid of such neurons. We cannot exclude, however, that GABAergic neurons may also be revealed in these areas by analysis of additional human material. The density of intranuclear GABAergic neurons throughout the human olive was low and even in the PO amounted only to approximately 3% of the large neurons visible by their background staining (Fig. 6A). The distribution of GAD-positive neurons in each species is depicted in Figure 7 . It is apparent from this map that in each species except the cat there are regions of relatively high density of intranuclear immunostained cells, although their overall number in each section remains small. Moreover, it seems obvious that the predominant localization of these cells varies among species.

Cytological characteristics of intranuclear GABAergic neurons The intranuclear GAD-positive neurons were cytologically similar across species. They were invariably small

Figure 5

GABAERGIC NEURONS IN MAMMALIAN I 0

509

neurons. While their long diameters varied among species lar formation. The raphe pallidus of all the species investiwithin a range of 15-20 km, their short diameters were gated also contained many GAD-positive neurons, which most constant and measured approximately 12 krn in all varied considerably in shape and size. In untreated animals, species. The immunostained cytoplasm of these GAD- GAD-positive boutons were closely apposed to cell bodies positive cell bodies formed a thin rim around the pale and emerging dendrites of both internuclear (Fig. 3C) and nucleus (Figs. lB, 3B, 4A inset, 5A inset, 5B, 6A,B). As periolivary GABAergic neurons (not illustrated). shown in Table 1,the GAD-negative I 0 neurons in rat, cat, and monkey were substantially larger than the GABAergic DISCUSSION neurons. These immunonegative cells were visible as ghosts Distribution of GAD-positive neurons in the by light diffraction, or, in the human case, they were weakly stained by nonspecific argyrophilia. In the rabbit 10, howventral medulla ever, no statistically significant difference was observed By GAD immunocytochemistry, we have demonstrated between the sizes of GAD-positive and GAD-negative cells the presence of three groups of GABAergic neurons in the (Table 1).The dendrites of the GABAergic neurons were ventromedial region of the medulla: intranuclear olivary labeled in some of the animal specimens, and these radiated away from the parent somata with few branches (Figs. 4 neurons located in the gray matter of the 10,internuclear inset, 5B). In untreated animals, GAD-positive boutons neurons located in the white matter between the olivary were closely apposed to the somata and, more frequently, to subnuclei, and periolivary neurons located in the adjacent the radiating dendrites of the immunostained neurons reticular formation, the raphe, and the white matter sur10. (Figs. 3B, 4A inset, 5B). In the human 10, the immunoreac- rounding the Some of the dendrites of the periolivary and internuclear tive dendrites extended away from the soma for distances of GABAergic neurons penetrate into the olivary gray matter, more than 200 km (Fig. 6B,C). These arbors had a predomand thus may share some innervation with neurons of the inantly radial configuration, and appeared less ramified 1 0 . These neurons are usually distinguished from the than the curved dendrites of olivary neurons described by intranuclear olivary GABAergic neurons by their larger Cajal('11) and others. size. The morphological features of the GABAergic periolivary and internuclear neurons correspond to those deGABAergic neurons in the olivary white scribed with the Golgi method (Bowman and King, '73; matter and periolivary regions Cajal, '11; Rutherford and Gwyn, '80; Scheibel and Scheibel, Large GAD-positive cells, predominantly elongated in '55; Scheibel et al., '56). Cajal ('11) considered both the shape, were observed in the white matter which separates periolivary and the internuclear neurons as belonging to the olivary subnuclei (Figs. lA, 2A, 4 0 . We defined these the reticular formation, and described their axons as entercells as internuclear olivary neurons, because of their ing the neighboring white matter or the anterior funiculus, situation in the olivary white matter. Moreover, some either directly or after crossing the midline raphe. It is well immunolabeled dendrites of these internuclear neurons known, however, that myelinated axons usually are not extended into the gray matter of the I 0 (Fig. 1A and 2A, well impregnated by the Golgi method, and one cannot insets). Such neurons were particularly well stained after exclude the possibility that these neighboring GABAergic colchicine pretreatment. Colchicine injection in the rat, neurons contribute some axon collaterals to the olivary rabbit, and cat brain stem also revealed a high density of neuropil. Recurrent inhibitory responses recorded from the GAD-positive neurons in the reticular formation dorsal to I 0 have previously been attributed to the activation of the I 0 (Figs. lA, 2A, 3A, 4B) and also in dorsal regions of neurons present in the neighboring reticular formation the pyramidal tract. GAD-positive cells in the pyramidal (Crill, '70; Llinas et al., '74; Sedgwick and Williams, '67). tract territory were particularly evident in rabbit (Fig. 3A). Periolivary and internuclear serotonin-containingneurons, These variously situated cells, which we have classified as with features similar to those of GAD-positive periolivary periolivary neurons, were fusiform or multipolar in shape and internuclear neurons, have been observed in rat, cat, and usually had larger somata than the intranuclear olivary and rhesus monkey (Chan-Palay, '77; Dahlstrom and Fuxe, neurons. Fewer of these neurons were observed in noncolch- '64; Hokfelt et al.,'78; our unpublished observations). In icine-treated brain stems. Also included in the category of the rat, some of these serotoninergic neurons project to the periolivary neurons is a group of large GAD-positive neu- I 0 (Bishop and Ho, '86). It is possible that serotonin and rons located in the white matter ventral and lateral to the GABA are colocalized in some of the periolivary and PO and the DAO (Figs. lA, 3A). However, these periolivary internuclear neurons. neurons were easily distinguished from the GAD-positive The GAD-positive neurons situated within the gray neurons by their larger size and abundant cytoplasm which matter of the I 0 are the smallest of the three groups, and, are characteristic of most neurons in the periolivary reticu- in all species examined except the rabbit, they are clearly distinguishable in somata1 size (Table 1) from the larger, GAD-negative olivary neurons. The GAD-negative neurons presumably correspond to excitatory projection neurons, Fig. 5. GAD-positiveneurons of the monkey 10.A: The PO contains several scattered GAD-positive somata (closed arrows). Lacunae in the the parent cell bodies of cerebellar climbing fibers. The GABAergic terminal fields contain the ghosts of GAD-negative somata intranuclear GABAergk neurons sparsely populate the (lined asterisk). Inset: The cell bodies of olivary immunonegative nuclear complex of all species investigated (Fig. 7). The neurons (asterisks) are larger and contain more cytoplasm than an locations of these neurons within the I 0 vary by species and intranuclear GAD-positive cell body (closed arrow). B: High magnifica- do not correlate in any obvious way with the regional tion of an intraolivary GAD-positive neuron (closed arrow) with a thin rim of cytoplasm surrounding the pale nucleus. The labeled dendrites variations of GABAergic innervation that were described in a companion paper (Nelson et al., '89). Furthermore, in radiate away from the soma and are in close relation to immunostained most regions of the olive, the densities of such cells seem too boutons (double arrows) along their course.

510

Fig. 6. GAD-positive neurons in the human inferior olive. The immunoreaction product was intensified by a reduced silver method. A: GAD-positive somata (closed arrows) are scattered throughout the PO with a much lower frequency than the GAD-negative somata (background staining). B: Small GAD-positive neuron (closed arrow) in the

B.J. FREDETTE ET AL.

PO has several radiating, sparsely branched dendrites. Two of the larger, GAD-negative neurons are indicated by lined asterisks. C: High magnification view of a small, irregularly shaped, GAD-positive neuron (closed arrow) situated in the PO.

rost rat

dmcc

beta

caudal

,DAO

RAT

/OA0

,DAO

PO

/DAO

beta

CAT n

RHESUS MONKEY

Fig. 7. Semischematic representation of the distribution of intranuclear GAD-positivesomata in the I 0 of rat, rabbit, cat, rhesus monkey, and human, represented in serial coronal sections. Dots indicate the location of GABAergic neurons in representative cases, one dot representing one cell.

MA0

dmcc

MA0

RABBIT HUMAN

512

B.J. FREDETTE ET AL.

The mode and frequency of distribution of the olivary GAD-positive neurons in the different animals requires some comment, especially since other studies have noted a striking increase in the number of interneurons from Classification of intranuclear rodents to carnivores and primates. Such a trend has been GAD-positiveneurons described, for example, for local circuit neurons in the Although the predominant view may be that neurons of cerebellar cortex (Lange, '82), for GABAergic neurons in the I 0 represent a homogeneous population characterized the body of the thalamus (Houser et al., '80; Ottersen and by profusely branching, spinous dendrites which curl around Storm-Mathisen, '84b; Penny et al., '83; Rinvik et al., '87) the parent cell body (Cajal, 'll),a number of Golgi studies and in the pontine nuclei (Brodal et al., '88; Nelson et al., have emphasized the existence of some variability in the '89). Although, in general, GABAergic neurons appeared features of olivary neurons. With different protocols of the more widespread in the I 0 of rhesus monkey and human Golgi method, the occurrence of a second type of large than in rat and rabbit, no distinct trend could be estabneuron with radiating, less ramified dendrites has been lished because the modes of distribution of these cells were reported in the cat, monkey, and human I 0 predominantly highly heterogeneous. Although many olivary regions in rat in the accessory nuclei (Rutherford and Gwyn, '80; Scheibel and rabbit were completely or nearly devoid of GADand Scheibel, '55; Sotelo et al., '74). In the rat 10, similar positive neurons, certain restricted territories of the MA0 neurons are distributed throughout the nucleus (Gwyn et in both species contained relatively high densities of such al., '77). Thus, there are at least two subclasses of large cells, In the I 0 of the cat, GAD-positive neurons were olivary projection neurons. In addition, Scheibel and usually absent, and no such "hot spots" of GABAergic cells Scheibel ('55) illustrate one cell with unramified dendrites were present in any of the subnuclei. In the primate 10,the and small soma in the macaque dorsal accessory nucleus. PO contained the highest concentration of GAD-positive Unfortunately, the axon of this cell was not well impreg- cells, although their density may be lower than that obnated. This type of cell may correspond to our small served in the "hot spots" of the rat and rabbit MAO. The differences in the frequency of occurrence of GABAergic intranuclear GABAergic neurons. Nissl staining and electron microscopy show that all large small neurons throughout the I 0 suggest the existence of olivary neurons are provided with sizable Nissl bodies in all modifications of the olivary circuit, which may be related to mammals (Bowman and King, '73; Gwyn et al., '77; Scheibel species-specific signal processing. It is noteworthy that and Scheibel, '55;Taber, '61; Walberg, '63; our unpublished intranuclear, internuclear, and periolivary GABAergic neuobservations). In addition, these methods have revealed a rons are contacted by GAD-positive boutons. These neusmall number of cells of an additional category that are no rons may be subjected, therefore, to GABAergic inhibition, larger than 13 p,m in diameter and stain only weakly with the source of which is yet to be determined. In conclusion, the mammalian I 0 contains GAD-positive aniline dyes (Scheibel and Scheibel, '55;Taber, '61). Ultrastructurally, these cells have a thin rim of cytoplasm that is cells whose features suggest that they represent local poor in polyribosomes and surrounds a crenated nucleus circuit neurons, and thus provide an intrinsic source of (Rutherford and Gwyn, '80). The scarcity of granular GABAergic boutons. Another contingent of GABAergic endoplasmic reticulum in the small cells suggests that they boutons in the I 0 may arise from axon collaterals of do not sustain either long axons or dense axonal plexuses. GAD-positive neurons located in the olivary white matter Such cells, presumably a class of short axon neurons, may and along the I 0 boundaries. However, the bulk of the correspond to the small cells described by the Scheibels, and olivary GABAergic innervation must originate from projecto the small intranuclear, GAD-positive neurons of this tion neurons outside of the olivary territory, because tract-tracing methods combined with immunocytochemisaccount. try indicate that there are substantial GABAergic projections to the I 0 from the cerebellar nuclei, the lateral and General considerations descending vestibular nuclei, the parasolitary nucleus, and All authors agree that the pale small neurons are either the nucleus prepositus hypoglossi (Angaut and Sotelo, '87; rare or comprise a small minority of the olivary neurons. deZeeuw et al., '88a,b, '89a,b, '90, in preparation; Fredette This suggests that our immunocytochemical study has and Mugnaini, '91; Nelson and Mugnaini, '85, '89; Nelson provided a fair description of the distribution of these et al., '84, '86). Lesions of these nuclei substantially deplete intraolivary small cells, although a rigorous statistical GAD-immunoreactivity in the respective olivary target estimate remains to be performed. Our data on the distribu- regions. tion of GABAergic neurons in the I 0 of rat, rabbit, and cat are most likely accurate, since colchicine was used to enhance somatal staining. On the other hand, it is possible ACKNOWLEDGMENTS that more GABAergic neurons than are described here This work was supported by USPHS grants NS21307 could be revealed in the rhesus monkey and human I 0 (not colchicine pretreated) by antibodies raised against GABA, (J.C.A.) and NS09904 (E.M.). since these antibodies do not require intensification methods for somatal staining (Ottersen and Storm-Mathisen, '84a). Yet, a recent investigation of parts of the baboon I 0 LITERATURE CITED by such a method has shown that approximately 5% of the Adams, J.C., and E. Mugnaini (1987) Patterns of glutamate decarboxylase neurons contained GABA-like immunoreactivity (Walberg immunostainingin the feline cochlear nuclear complex studied with silver and Ottersen, '89). This low density is similar to our own enhancement and electron microscopy.J. Comp. Neural. 262:375-401. estimates for the PO of the rhesus monkey and the human. Angaut, P., and C. Sotelo (1987) The dentato-olivary projection in the rat as

small to account for the extremely profuse GABAergic innervation of the 10.

GABAERGIC NEURONS IN MAMMALIAN I 0 a presumptive GABAergic link in the olivo-cerebello-olivary loop. An ultrastructural study. Neurosci. Lett 83:227-231. Angaut, P. and C. Sotelo (1989) Synaptology of the cerebello-olivary pathway. Double labelling with anterograde axonal tracing and GABA immunocytochemistry in the rat. Brain Res. 479:361-365. Barmack, N.H., E. Mugnaini, and B. Nelson (1989) Vestibularly-evoked activity of single neurons in the beta nucleus of the inferior olive. In P. Strata (ed): The Olivocerebellar System in Motor Control. Berlin: Springer-Verlag. Exp. Brain Res. Ser. 17:313-322. Bishop, G.A., and R.H. Ho (1986) Cell bodies of origin of serotoninimmunoreactive afferents to the inferior olivary complex of the rat. Brain Res. 399:369-373. Bowman, M.H., and J.S. King (1973) The conformation, cytology and synaptology of the opossum inferior olivary nucleus. J. Comp. Neurol. 148:491-524. Brodal, P., G. Mihailoff, B. Border, O.P. Ottersen, and J. Storm-Mathisen (1988) GABA-containing neurons in the pontine nuclei of rat, cat and monkey, a n immunocytochemical study. Neuroscience 25.27-45. Buisseret-Delmas, C., C. Bantini, C. Compoint, H. Daniel, and D. Menetrey (1989) The GABAergic neurones of the cerebellar nuclei: Projection to the caudal inferior olive and to the bulbar reticular formation. In P. Strata (4):The Olivocerebellar System in Motor Control. Berlin: Springer-Verlag. Exp. Brain Res. Ser. 17.108-110. Cajal, R.S. (1911) Histologie du Systeme Nervew de 1’Homme e t des Vertebres, Vol. I. Paris: Maloine. Chang, Y.-C., and D.J. Gottlieb (1988) Characterization of the proteins purified with monoclonal antibodies to glutamic acid decarboxylase. J. Neurosci. 8.2123-2130. Chan-Palay, V. (1977) Cerebellar Dentate Nucleus. Berlin: Springer-Verlag. Crill, W. (1970) Unitary multiple-spiked responses in cat inferior olive nucleus. J. Neurophysiol. 33:199-209. Dahlstrom, A,, and K. Fuxe (1964) Evidence for the existence of monoamine neurons in the central nervous system-I. Demonstration of monoamines in the cell bodies of brain stem neurons. Acta Physiol. Scand. 62(Suppl. 232): 1-55. DeZeeuw, C.I., J.C. Holstege, F. Calcoen, T.J.H. Ruigrok, and J. Voogd (1988a) A new combination of WGA-HRP anterograde tracing and GABA immunocytochemistry applied to afferents of the cat inferior olive a t the ultrastructural level. Brain Res. 447.369-375. DeZeeuw, C.I., J.C. Holstege, T.J.H. Ruigrok, and J. Voogd (1988b) The GABAergic, the cerebellar and the mesodiencephalic innervation of the rostra1 medial accessory olive of the cat. A quantitative comparison a t the ultrastructural level. Suppl. Eur. J. Neurosci. 10.9:25. DeZeeuw, C.I., J.C. Holstege, T.J.H. Ruigrok, and J. Voogd (1989a) An ultrastructural study of the GABAergic,the cerebellar and the mesodiencephalic innervation of the cat medial accessory olive: Anterograde tracing combined with immunocytochemistry. J. Comp. Neurol. 284.12-35. DeZeeuw, C.I., J.C. Holstege, T.J.H. Ruigrok, and J. Voogd (1989b) The GABAergic, cerebellar and mesodiencephalic innervation of the glomeruli in the cat inferior olive. A comparison at the ultrastructural level. In P . Strata (ed): The Olivocerebellar System in Motor Control. Berlin: Springer-Verlag. Exp. Brain Res. Ser. 17:lll-117. DeZeeuw, C.I., J.C. Holstege, T.J.H. Ruigrok, and J. Voogd (1990) Mesodiencephalic and cerebellar terminals end upon the same dendritic spines within the glomeruli of the cat and rat inferior olive: An ultrastructural study using a combination of PHIleucine and WGA-HRP anterograde tracing. Neuroscience 34:645-655. Foster, R.E., and B.E. Peterson (1986) The inferior olivary complex of the guinea pig: Cytoarchitecture and cellular morphology. Brain Res. Bull. 17:785-800. Fredette, B.J., and E. Mugnaini (1991) The GABAergic cerebello-olivary projection in the rat. Anat. Embryol. 184:225-243. Gwyn, D.G., G.P. Nicholson, and B.A. Flumerfelt (1977) The inferior olivary nucleus of the rat: A light and electron microscopic study. J. Comp. Neurol. 174:489-520. Hokfelt, T., A. Ljungdahl, H. Steinbuscb, A. Verhofstad, G. Nilsson, E. Brodin, B. Pernow, and M. Goldstein (1978) Immunohistochemical evidence of substance P-like immunoreactivity in some 5-hydroxytryptamine-containing neurons in the rat central nervous system. Neuroscience 3r517-538. Houser, C.R., J.E. Vaughn, R.P. Barber, and E. Roberts (1980) GABA neurons are the major cell type of the nucleus reticularis thalami. Brain Res. 200:341-354. Kaufman, D.L., J.F. McGinnis, N.R. Krieger, and A.J. Tohin (1986) Brain

513 glutamate decarboxylase cloned in X g t - 1 1 : Fusion protein produces gamma-aminobutyric acid. Science 232: Il38-1140. Kmjevic, K., and S. Schwartz (1976) Inhibitory action of GABA and GABA-mimetics on vertebrate neurons. In E. Roberts (ed): GABA in Nervous System Function. New York: Raven Press, pp. 269-281. Lange, W. (1982) Regional differences in the cytoarchitecture of the cerebellar cortex. In S.L. Palay and V. Chan-Palay (eds): The Cerebellum, New Vistas. Berlin: Springer-Verlag. pp. 93-105. Llinas, R., R. Baker, and C. Sotelo (1974) Electrontonic coupling between neurons in cat inferior olive. J. Neurophysiol. 37:560-571. Mugnaini, E., N.H. Barmack, and W.H. Oertel (1982) GABAergic innervation of the rabbit inferior olive studied by GAD-immunocytochemistry. Soc. Neurosci. Abstr. 8.445. Mugnaini, E., and A.-L. Dahl (1983) Zinc-aldehyde fixation for lightmicroscopic immunocytochemistry of nervous tissue. J. Histochem. Cytochem. 31.1435-1438. Mugnaini, E., and W.H. Oertel (1985) An atlas of the distribution of GABAergic neurons and terminals in the rat CNS as revealed by GAD immunocytochemistry. In A. Bjorklund and T. Hokfelt (eds): Handbook of Chemical Neuroanatomy, Vol. 4: GABA and Neuropeptides in the CNS, Part I. New York Elsevier Publishers, pp. 436-608. Nelson, B.J., J.C. Adams, N.H. Barmack, and E. Mugnaini (1989) A comparative study of glutamate decarbowlase immunoreactive boutons in the mammalian inferior olive. J. Comp. Neurol. 286:514-539. Nelson, B.J., N.H. Barmack, and E. Mugnaini (1984) A GABAergiccerebelloolivary projection in the rat. SOC.Neurosci. Abstr. 10:539. Neison, B.J., N.H. Barmack, and E. Mugnaini (1986) GABAergic projection from vestibular nuclei to rat inferior olive. SOC. Neurosci. Abstr. 12:225. Nelson, B.J., and E. Mugnaini (1985) Loss of GABAergic nerve terminals in the inferior olive of cerebellectomized rats. SOC. Neurosci. Abstr. 11:182. Nelson, B.J., and E. Mugnaini (1988) The rat inferior olive as seen with immunostainjng for glutamate decarboxylase. Anat. Embryol. 179:109-127. Nelson, B.J., and E. Mugnaini (1989) Origins of GABAergic inputs to the inferior olive. In P. Strata (ed): The Olivocerebellar System in Motor Control. Berlin: Springer-Verlag. Exp. Brain Res. Ser. 17:8&107. Nelson, B.J., E. Mugnaini, J.C. Adams, and N.H. Barmack (1988) GABAergic components of the mammalian 10.SOC.Neurosci. Abstr. 14.494. Oertel, W.H., E. Mugnaini, D.E. Schmechel, M.L. Tappaz, and 1.J. Kopin (1982) The immunocytochemical demonstration of GABAergic neuronMethods and application. In V. Chan-Palay and S.L. Palay (eds): Cytochemical Methods in Neuroanatomy. New York: Alan R. Liss, pp. 279-329. Oertel, W.H., D.E. Schmechel, E. Mugnaini, M.L. Tappaz, and I.J. Kopin (1981a) Immunocytochemical localization of glutamate decarboxylase in rat cerebellum with a new antiserum. Neuroscience 62715-2735. Oertel, W.H., D.E. Schmechel, M.L. Tappaz, and I.J. Kopin (1981b) Production of a specific antiserum to rat brain glutamic acid decarboxylase by injection of an antigen-antibody complex. Neuroscience 6:2689-2700. Ordronneau, P., P.B.-M. Lindstrom, and P. Petrusz (1981) Four unlabeled antibody bridge techniques. J. Histochem. Cytochem. 29:1397-1404. Ottersen, O.P., and J. Storm-Mathisen (1984a) Glutamate- and GABAcontaining neurons in the mouse and rat brain as demonstrated with a new immunocytochemical technique. J. Comp. Neurol. 229.374-392. Ottersen, O.P., and J. Storm-Mathisen (198413)GABA-containingneurons in the thalamus and pretectum of the rodent. Anat. Embryol. 170:197-207. Penny, G.R., D. Fitzpatrick, D.E. Schmechel, and I.T. Diamond (1983) Glutamic acid decarboxylase-immunoreactive neurons and horseradish peroxidase-labeled projection neurons in the ventral posterior nucleus of the cat and Gulugo sengalensis. J. Neurosci. 3:1863-1887. Ramon-Moliner, E. (1962) An attempt at classifying nerve cells on the basis of their dendriticpatterns. J. Comp. Neurol. 119211-227. Ribak, C.E., J.E. Vaughn, and K. Saito (1978) Immunocytochemical localization of glutamic decarboxylase in neuronal somata following colchicine inhibition of axonal transport. Brain Res. 140:315-332. Rinvik, E., O.P. Ottersen, and J.Storm-Mathisen (1987)Gamma-aminobutyrate like immunoreactivity in the a t ’ s thalamus. Neuroscience 21:781-806. Rutherford, J.G., and D.G. Gwyn (1980) A light and electron microscopic study of the inferior olivary nucleus of the squirrel monkey Suimiri sciureus. J. Comp. Neurol. 189:127-155. Scheibel, M.E., and A.B. Scheibel(1955) The inferior olive: A GoIgi study. J. Comp. Neurol. 102.77-132. Scheibel, M.E., A.B. Scheibel, F. Walberg, and A. Brodal (1956) A real distribution of axonal and dendritic patterns in inferior olive. J. Comp. Neurol. 106.21-49.

514 Sedgwick, E.M., and T.D. Williams (1967) Responses of singie units in the inferior olive to stimulation of the limb nerves, peripheral skin receptors, cerebellum, caudate nucleus and motor cortex. J. Physiol. 189:261-279. Sotelo, C., T. Gotow, and M. Wassef (1986) Localization of glutamic-aciddecarboxylase-immunoreactive axon terminals in the inferior olive of the rat, with special emphasis on anatomical relations between GABAergic synapses and dendro-dendritic gap junctions. J. Comp. Neurol. 25232-50. Sotelo, C., R. Llinas, and R. Baker (1974) Structural study of inferior olivary nucleus of the c a t Morphological correlates of electronic coupling. J. Neurophysiol. 3 7 5 4 1 6 9 9 .

B.J. FREDETTE ET AL. Sternberger, L.A. (1986) Immunocytochemistry, 3rd Edition. New York: John Wiley and Sons. Szteyn, S. (1988) Types of neurons in nucleus olivaris inferior of the European bison. J. Hirnforsch. 29;353-356. Taber, E. (1961) The cytoarchitecture of the brain stem of the cat I: Brain stem nuclei of cat. J. Comp. Neurol. 116:27-70. Walberg, F. (1963) An electron microscopic study of the inferior olive of the cat. J. Comp. Neurol. 12Orl-17. Walberg, F., and O.P. Ottersen (1989) Demonstration of GABAimmunoreactive cells in the inferior olive of baboons (Papio pupio and Pupio unubis). Neurosci. Lett. 10lr149-155.