Quantitative ultrastructural analysis of the periaqueductal gray in the rabbit.код для вставкиСкачать
THE ANATOMICAL RECORD 236573-585 (1993) Quantitative Ultrastructural Analysis of the Periaqueductal Gray in the Rabbit S.T. MELLER AND B.J. DENNIS Department of Physiology, The University of Adelaide, South Australia, Australia ABSTRACT The fine structure of the periaqueductal gray (PAG) of the rabbit was examined using the transmission electron microscope. On the basis of synaptic polarity, vesicle size, and the nature of the pre- and postsynaptic elements, 10 essentially different synaptic types could be discerned (6 axo-dendritic, 2 axo-somatic, 1 axo-axonic, and 1 dendro-dendritic). Synaptic contacts on the soma of PAG neurons were small and covered, on average, only 1.6%of the soma surface. The most striking feature of the synaptic structure of the PAG was that more than 94.1%of all synapses were axo-dendritic. Of these, 83.5% were of the symmetrical type. Most of these contacts occurred on buttons of small to medium size, and contained either round vesicles of medium size or pleomorphic vesicles of medium size. Boutons containing only flattened vesicles were quite rare. Boutons contacting larger dendrites were generally small-to-medium in size, made asymmetric-type synaptic contacts, and contained pleomorphic vesicles of medium-to-large size. Medium-sized dendrites were contacted principally by small boutons exhibiting either symmetrical or asymmetrical junctions containing medium-sized pleomorphic vesicles, and in addition a few of these boutons contained both large, and small, round vesicles. Dendritic spines were generally provided with only one synaptic contact, stretching the entire width of the spinous process. Boutons and the spines on dendrites were approximately the same size. Synapses between two vesicle-containing structures (axo-axonicor dendro-dendritic synapses) were rare (1.4%).They were generally asymmetric and contained round vesicles of medium size. Complex synapses, where a glial sheet enclosed the synapse, were occasionally observed. Also seen were multiple synapses, with up to 11 contacts on a single dendritic profile. Large dense-core vesicle were seen in approximately 40% of all synapses, whereas small dense-core vesicles were only found in about 3%. Data is provided on how different synaptic features relate to ventral, lateral, dorsal, and medial PAG. Principally this is in relation to neuron size, glia cell content, axonal characterization, and vesicular type. o 1993 Wiley-Liss, Inc. Key words: Neuropil, Synapse, Quantitative, Neuron, Vesicle The periaqueductal gray (PAG) is a region of recent interest due to the great variety of functional roles which have been attributed to it (see Beitz, 1985; Meller and Dennis, 1986; Meller and Dennis, 1991). There have been a number of reports on the ultrastructure of the PAG in the cat and rat (Hamilton et al., 1973; Liu and Hamilton, 1976; Liu, 1978; Gioia et al., 1983; Bianchi and Gioia, 1984; Moss and Basbaum, 1983; Williams and Beitz, 1989, 1990a,b). Although most of these reports have focused on individual features of structures in the PAG, to date, there has been no systematic analysis of the PAG in any species. In addition, while there are detailed descriptions of the neuronal types, subdivisions, and afferent and efferent connections of the rabbit PAG (Meller and Dennis, 1986, 1990a,b, 1991), and there is good reasons to use the rabbit as an experimental animal in neuroscience 0 1993 WILEY-LISS, INC. research (Dennis and Meller, in press), there is no ultrastructural analysis of the rabbit PAG. Considering the large range of neurotransmitters, neuromodulators, and peptides found in neurons of the CNS, the manner in which synapses should be classified to account for functional differences is an obvious challenge still to be met. It has long been suggested that synapses with a symmetrical arrangement of preand post-synaptic membrane thickenings may serve as Received July 2, 1992; accepted January 13, 1993. Address reprint requests to Dr. B.J. Dennis, Department of Physiology, The University of Adelaide, South Australia, Australia. Dr. S.T. Meller’s present address is Department of Pharmacology, College of Medicine, University of Iowa, Iowa City, IA 52242. 574 S.T. MELLER AND B.J. DENNIS inhibitory synapses, and the asymmetric type synapse may be excitatory (Gray, 1959; Eccles, 1964). Also, it has been proposed that round vesicles are associated with excitatory synapses, and that flattened and pleomorphic vesicles are associated with inhibitory synapses (Uchizono, 1965). Further, dense-core vesicles, of different sizes, have been implicated as containing a variety of neuroactive peptides (Pickel et al., 1977a; Beaudett and Descarries, 1981; Gabella, 1976; Pickel et al., 197713; Johansson et al., 1980). Most recently, immunocytochemicalanalysis of ultrathin sections has provided further analysis of the excitatory and inhibitory transmitters localized within different vesicle and synapse types. Whereas there is an increasing number of reports concerning the cellular and functional organization of the PAG (see Beitz, 1985; Meller and Dennis, 1986, 1990a,b), as yet there is no general agreement on ultrastructural details of the PAG. Therefore, the present study was initiated to provide a detailed ultrastructural analysis of the entire PAG on which to base subsequent functional studies. Portions of these data have been previously reported (Meller and Dennis, 1987). MATERIALS AND METHODS Three adult male lop-eared rabbits (OrctoEagus cuniculus) weighing between 2.1 and 2.7 kg were used in order to examine the ultrastructural characteristics of previously described PAG subdivisions (Meller and Dennis, 1986, 1990b):dorsal, lateral, ventral, and medial. Rabbits were housed on a 12/12hr light/dark cycle and provided with food and water ad libitum for 1-2 weeks prior to use. Rabbits were deeply anaesthetized by an intravenous injection of sodium pentobarbital (Ciba-Geigy, Pendle Hill, New South Wales, Australia). The descending aorta was clamped, and 2,500 IU heparin was injected intracardiacally before transcardiac perfusion via a cannula tied into the ascending aorta. A prefixation rinse of 500 ml phosphate buffered heparinized saline was followed by fixation, using 800 ml, 0.05M 3% glutaraldehyde, 2% paraformaldehyde, pH 7.2. In each case, the brain was removed, trimmed of cortex, and the regions rostra1 to the third ventricle and caudal to the fourth ventricle were also removed. The remainder of each brain was sectioned into 3-mm-thick coronal blocks. The resultant blocks encompassed the entire length of the PAG surrounding the aqueduct. Block 1 contained the PAG from the level of the posterior commissure to the level of the red nucleus, block 2 extended from the level of the red nucleus to the level of the caudal oculomotor nucleus, and block 3 extended from the level of the caudal oculomotor nucleus to the level of the caudal inferior colliculus. In each block, at least 3 core samples were taken from each of the 4 PAG subdivisions previously described (Meller and Dennis, 1986, 199Ob). This ensured that the samples did not overlap boundaries between divisions. The dissected blocks were washed in 0.1M phosphate buffer, pH 7.2 (2 x 10 min), postfixed in 1%osmium tetroxide (1hr) and embedded in a TAAB plastic resin (TAAB Laboratories Equipment Ltd., Reading, UK). Thin sections (1-2 pm) from each block were cut using an ultramicrotome, mounted on glass slides, and stained with toluidine blue to permit light microscopic structural examination of each region. At least 10 (and up to 30) ultrathin sections from each block were subsequently cut, placed on a metal grid, and stained with a saturated alcoholic solution of uranyl nitrate solution in the dark for 15-20 min, followed by a lead citrate solution for 12 min. The grids were washed, dried, and subsequently viewed with a JEOL transmission electron microscope (TEM) a t 60 Kv. Soma, dendrites, axons, synapses, and glia from more than 360 cut sections within the PAG were examined quantitatively with a light microscope and TEM. Soma, dendrite, and axonal shape and size, the arrangement and size of all cytoplasmic organelles, and nuclear details were recorded. All synaptic contacts were counted and classified according to: location of the contact (soma, axon, or dendrite), vesicle type (round, flattened, or pleomorphic), vesicle size (small, medium, or large), vesicle density (low, medium, high, or crystalline), bouton size (small, medium, or large), type of synaptic contact (symmetrical or asymmetric), whether they made single or multiple contacts, the presence of large and small dense-core vesicles, and whether subsynaptic dots were found. Vesicle size (small = 20-28 nm, medium = 34-42 nm, large = 50-60 nm) and density (low = 1-10 vesicles, medium = 10-20 vesicles, high = >20 vesicles) and bouton size (small = <1 pm, medium = 1-2 pm, large = > 2 pm) were arbitrarily classified according to the sizes indicated. All size measurements were made directly on the TEM. All measured and descriptive parameters for all synapses analyzed were subjected to cluster analysis using BioMedical Data Package (BMDP, Statistical Software Inc., Los Angeles, California) whereby, through progressive levels of clustering, significant relationships can be extracted. Basically, this permitted the identification of synapses having similar features to be grouped together. RESULTS General Cytologica/ Features Soma Some shape varied from round to spindle to polygonal. The only regional difference in the distribution of cells was a predominance of round and ovoid cells in the medial PAG and more polygonal cells in lateral and ventral PAG. Neuronal size varied from the smallest cells (11pm in diameter), generally found within medial PAG, through to the largest cells (44 pm in diameter), found within the outer limits of both lateral and ventral regions of the PAG. Some cells (located predominantly in the medial division, but present also in smaller numbers in other PAG regions) appeared to have a much denser cytoplasm due t o the predominantly greater number of ribosomes and rough endoplasmic reticulum (RER) within them. The nucleus, which was generally found to occupy a central position in the cell, contained large clumps of chromatin accumulated on the inner surface of its membrane, with small interchromatic granules and perichromatic particles scattered throughout (Fig. 1). Nuclear inclusions were found in about half of the cells examined. The majority of these inclusions, up to 1.6 pm in diameter, had a crystalline-like appearance (Fig. lb). Fibrillar-like inclusions were also detected in some ULTRASTRUCTURE OF THE RABBIT PAG Fig. 1. Nuclear inclusions in PAG cells. (A), Low-power magnification of the general cytological features of the soma of a PAG neuron and its surrounding neuropil. The nucleus within the soma shows clearly examples of both crystalline (C) and fibrillar (F) nuclear inclusions. x 4,000. Bar = 1 pm. (B), The most common type of inch- 575 sion is one with a crystalline appearance, seen here (arrowed) immediately below the nucleolus (N) and its nuclear satellites (NS). X 18,000. Bar = 1 pm. (C), Occasionally, the nuclear inclusion had a distinctly fibrillar appearance (F). x 24,000. Bar = 1 pm. These examples are enlargements of the inclusions in a. 576 S.T. MELLER AND B.J. DENNIS Fig. 2. This electron micrograph illustrates the general ultrastructural features of the organelles to be found within the soma and neuropil of PAG neurons. Note within the soma randomly-dispersedchromatin (C) internal to the nuclear envelope (E). The cytoplasm contains several Golgi complexes (G), lysozomes (L),rough endoplasmic reticulum (RER), and RER aggregated into well-defined Nissl bodies (N), plus numerous mitochondria (M).x 26,000. Bar = 1 Km. nuclei (Fig. lc). The nucleolus was roughly spherical, contained numerous dense granules, and had a latticelike appearance. Occasionally, one nuclear satellite (two on rare occasions) was attached to the nucleolus. Ribosomes, observed around the cytoplasmic surface of the nuclear envelope, tended to accumulate in the cytoplasm within nuclear invaginations. The intermembranous distance of the nucleus was 22-33 nm. Nuclear pores were seldom observed. In the cytoplasm, Nissl substance was observed to be either dust-like, with numerous ribosomes appearing either singly or in rosettes, or clumped with a dense accumulation of RER giving rise to a Nissl body (Fig. 2). At least one, and sometimes up to 4 Nissl bodies were observed in the cytoplasm of larger cells. The Golgi apparatus was often located in close proximity to the nucleus. It generally consisted of 4-5 flattened cisternae with no ribosomes, and numerous associated small vesicles (about 30-100 nm diameter). Occasionally up to 3 Golgi apparatuses were observed within a cell. Spherical or ovoid shaped lysozomes (100-370 nm) were observed both near the Golgi apparatus and at distant sites within the cytoplasm (Fig. 2). Large multivesicular bodies (500-600 nm) which contained small round and elongated vesicles were observed in all cells, often in close proximity to the Golgi apparatus. Numerous round (120-400 nm) and elongated (120-400 X 270-520 nm) mitochondria, which displayed a dense matrix, were found distributed at random throughout the cytoplasm. Microtubules (20-25 nm) and neurofilaments (10-12 nm) were also found scattered throughout the cytoplasm. At the junction of soma and primary dendrites, neurofilaments and microtubules were very prominent where they funnelled into the dendrite. Neuropil Dendrites varied in diameter from small (< 2 Fm), medium (2-4 Fm) to large (>4 Fm) (Table 1).The dendrites (especially the larger ones) were ultrastructurally similar to the soma, except that microtubules were far more prominent in dendritic profiles. Mitochondria were observed in most dendritic profiles, as were neu- ULTRASTRUCTURE OF THE RABBIT PAG TABLE 1. Representation of the total numbers of synapses (expressed as a %) contacting small (<2 am), medium (2-4pm), and large (>4 pm) diameter dendrites in each subdivision of the PAG 577 PAG (Fig. 3):round, elongated, and flattened (Table 2). Variations in vesicle shape within boutons located in dorsal, lateral, ventral, and medial PAG are indicated in Table 2. Although larger dendrites tended to have larger boutons contacting them, there was no signifiDendrite size PAG region Small Medium Large cant difference in the size or types of boutons contacting a proximal dendrite, a distal dendrite, or a dendritic Dorsal 58.1 36.1 5.8 spine. In addition to the vesicle types described above, Lateral 57.7 28.2 14.1 large and small dense-core vesicles (which were generVentral 50.6 29.1 20.3 ally round with an electron-dense core) were distribMedial 61.5 29.4 9.1 uted throughout boutons of the PAG (Fig. 4;Table 3). Average 57.4 31.0 11.6 Large dense-core vesicles were found to be present in approximately 40% of all synapses, always in combination with electron-lucent vesicles. There were more rofilaments. Other features, not always apparent boutons containing large dense-core vesicles in ventral within a dendrite, were the endoplasmic reticulum, and medial PAG than in dorsal and lateral PAG (Table Golgi apparatus, multivesicular bodies, and lysozomes. 3). Small dense-core vesicles were only observed in Generally, the smaller the dendrite, the fewer the about 3% of PAG synapses and never in boutons within number of organelles observed within them. The num- ventral PAG (Table 3). ber of spines on the soma of neurons in the PAG were Vesicle size: The size of electron-lucent vesicles very low, especially when compared with the numbers ranged from 30-120 nm in diameter (Fig. 3).The large observed on dendrites. Spines were about 1-2 pm in majority of these vesicles fell into two main groups length, had variable diameters and shapes, and typi- ranging between 34-38 nm or 40-42 nm in diameter cally contained a spine apparatus. No microtubules or (Table 4).A third group of round vesicles were much neurofilaments were observed in dendritic spines. smaller, averaging between 20 and 28 nm in diameter. Unmyelinated axons contained mitochondria, many These vesicles were not prevalent and tended to be loneurofilaments, a few microtubules, smooth endoplas- cated (along with other vesicles) in boutons which mic reticulum, small vesicles, and multivesicular bod- showed a relatively greater density of vesicles, and also ies. No ribosomes or RER were found in axons. Mito- in those containing flattened vesicles. A fourth group chondria were thinner and more elongated than in the of round vesicles were quite large, with diameters in cytoplasm, and were generally orientated parallel to the range of 50-60 nm. Very occasionally, an excepthe length of the axon. At the site of synaptic boutons, tionally large round electron-lucent vesicle was found, mitochondria were smaller and associated with many approximately 90-120 nm in diameter. Large elonsmall, round, or flattened vesicles (0.4-2.5pm). My- gated vesicles were also observed which were in the elinated axons had the same characteristic structure as range of 73-77 x 40-42 nm, whereas medium sized unmyelinated axons except for their larger size and the elongated vesicles were about 50-60 x 25-35 nm. It presence of myelin sheaths surrounding them. was a consistent finding that the neuropil contained Glia outnumbered neurons by 4.2:l.Glia generally more than twice as many boutons exhibiting large contained a well-developed Golgi apparatus and large round vesicles and twice as many small vesicles of both numbers of granules, ribosomes, and microtubules round and elongated types, as did axo-somatic synwhich gave them a dense appearance. Glia were repre- apses. sented as astrocytes, oligodendrocytes, and microglia. Approximately 40% of all somatic synapses and Astrocytes did not have a very extensive RER and con- 30.2% of all neuropil synapses contained a t least 1 tained relatively few ribosomes. Consequently, they large dense-core vesicle and, quite frequently, several were generally lighter in appearance than other glia were present. Some large dense-core vesicles were apand neuronal cells. Also, their cell bodies were quite proximately 70-85 nm in diameter while another usually irregular in outline. They were found t o send group of even larger vesicles measured about 100-120 out sheets of cytoplasmic processes which on occasions nm in diameter, with the dense core occupying a t least enclosed almost an entire neuronal soma or many syn- 75% of the vesicle. Smaller dense core vesicles were aptic contacts. Oligodendrocytes were usually darker also observed in which the average diameter was than other cells. In contrast to astrocytes, they pos- 40-50 nm, with a dense core occupying about 50% of sessed a well-developed RER, and each had a large the vesicle. number of free ribosomes near their outer perikaryal Vesicle density: Vesicle density varied markedly membrane surface. The nucleus of these cells was usu- within different boutons (Table 5). Density was considally found in an eccentric position within the cyto- ered medium where there was an accumulation of vesplasm. Oligodendrocytes were usually associated in a icles at the site of the synaptic contact with a moderate satellite position with larger neurons in the PAG, with number of vesicles distributed throughout the rest of the membranes of the two cells in direct apposition. the bouton. This was the case in 50.9% of synapses on There were no specializations observed on the mem- the soma and 47.8% of synapses in the neuropil. Lowbranes between glia and neuronal cells. density boutons were those containing only a few vesicles at the synaptic contact and few or no vesicles in Synaptic characteristics the rest of the synaptic bouton (15.9% of synapses on Presynaptic profiles. Vesicle shape: Three different the soma and 7.9% of synapses in the neuropil). A highshapes of electron-lucent vesicles were distinguished in density bouton was evaluated where the vesicles over presynaptic boutons of axons in the neuropil of the the synaptic contact were moderate to dense in number 578 S.T. MELLER AND B.J. DENNIS Fig. 3.The most common synaptic types observed in the rabbit PAG. (A), Two adjacent axo-somatic synapses of the symmetric type, containing predominantly round vesicles. x 44,000. (B), Two axo-dendritic synapses of the asymmetrical type both containing round vesicles. The one in the center of the micrograph has subsynaptic dots below the post-synaptic membrane. X 22,000. (C), Two axo-axonal profiles (both labelled A) on either side of a dendritic spine (D) show- ing a n axodendritic synaptic contact. X 29,000. (D), An example of a complex synapse in the PAG, with four well-defined synapses contacting a single dendritic profile. Two of the synapses (small arrows) are symmetrical. The other two (large arrows) are asymmetrical. All synapses contain round vesicles with the one at the lower right containing, additionally, several dense-core vesicles. X 22,500. 579 ULTRASTRUCTURE OF THE RABBIT PAG TABLE 2. Summary of the distribution of vesicles by shape (expressed as a %) in boutons contacting either the soma or elements in the neuropil within the four subdivisions of the PAG Vesicle shape Round Pleomorphic Flattened Medial Soma Neuropil 9.7 37.3 86.7 61.0 3.6 1.7 Dorsal Soma Neuropil 12.0 46.6 84.0 52.3 4.0 1.1 and the rest of the bouton was filled with vesicles, as was the case for 32.5%of synapses on the soma and for 42.9% of synapses in the neuropil. Occasionally, boutons were observed where the vesicles were packed very closely together so that their structure took on a crystalline-like appearance (0.7% of synapses on the soma and 1.4%of synapses in the neuropil). Bouton size: A number of different-sized presynaptic boutons could be recognized within the PAG (Table 6): small boutons (<1 pm on their longest axis), medium boutons (1-2 pm), and large boutons (>2 pm). Of the boutons contacting the soma, 75.5%fell into the small- and medium-sized range, and over 85%of these were between 0.8 and 1.3 pm in length. In the neuropil, 92.4% of boutons were of the small and medium size, again with over 85% between 0.8 and 1.3 pm in size. Thus, the large majority of boutons fell into a relatively narrow range of sizes. Those contacting somas tended to be slightly larger than those contacting dendrites. Postsynaptic specializations. Both symmetrical and asymmetrical thickening of the synaptic membrane up to 60 nm in thickness were observed, but they generally averaged about 35 nm (Fig. 3). Although length of synapses generally varied between 230 nm and 450 nm, some were occasionally observed to be up to 700 nm in length. Of synapses contacting the soma, 83.5% were symmetrical and 16.5% were asymmetrical, whereas in the neuropil 46.4%were symmetrical and 53.6%were asymmetrical junctions. Occasionally, rows of subsynaptic dots were observed beneath the postsynaptic thickening of asymmetric synapses. These dots were 19 nm in diameter, evenly spaced a t 18 nm, and they were continuous across the entire length of the synapse. Synaptic clefts varied from about 13 to 28 nm in width; the majority being between 20 and 24 nm. Dense areas of granular or filamentous material were also associated with the cytoplasm at the synapse. PAG region Lateral Soma Neuropil 16.7 57.5 76.6 39.8 6.7 2.7 Ventral Soma Neuropil 32.4 41.9 64.7 56.9 2.9 1.2 Soma 17.0 78.7 4.3 Total Neuropil 42.2 56.2 1.6 soma surface was covered by synapses; never more than 4.9% was covered. There were usually only 1-4 synaptic contacts on the soma. Most axo-somatic contacts (83.5%)were of the symmetrical type and the boutons associated with these were one of several different types. The most common type of synapse involved boutons of small and intermediate size containing round vesicles of medium size. Boutons of small and medium size containing pleomorphic vesicles of medium size were also quite prevalent. Boutons with pleomorphic vesicles contained only a relatively small percentage of elongated vesicles where ellipticity (length divided by width) of these pleomorphic vesicles varied from 1.23.4. Boutons containing only elongated vesicles were very rare (4.3%).Of all somatic synapses, 37.1% contained large dense-core vesicles and only 0.7% of somatic synapses contained small dense-core vesicles. Soma-soma synaptic contacts were never observed. Axo-dendritic synapses. Dendrites were classified as: proximal dendrites (when they were seen to leave the soma), large dendrites which were probably proximal dendrites but could not be positively identified as such, medium-sized dendrites, and small dendrites (which were actually classified as spines because of their small size or the fact that they were quite clearly seen to be spine appendages on dendrites). Boutons contacting larger dendrites were generally small to medium in size, of the asymmetric type, and contained pleomorphic vesicles of medium-to-large size. The medium sized dendrites were contacted principally by small boutons exhibiting both symmetrical and asymmetrical junctions containing pleomorphic vesicles of medium size. In addition a few of these boutons showed both large, and small, round vesicles. Dendritic spines generally showed only one synaptic contact stretching the entire width of the spinous process. Spines on dendrites, and the bouton, were approximately the same size, although sometimes the bouton had a larger diSynaptic types ameter than the dendritic spine. Basically the same All PAG synapses were assumed to be chemical, al- types of synaptic contacts were observed on small and though nonsynaptic junctions were also found. These large dendrites. Of synapses on dendrites, 57.4% occontacts, the puncta adherens, showed membrane curred on small dendrites, 30.7% were on mediumthickenings in a symmetrical arrangement, yet not as- sized dendrites, and only 11.9% were on large dendrites. The characteristic features of the 6 major sociated with vesicles. Using the criteria of type of contact, bouton size, and axo-dendritic synapses are listed in Table 7 and the vesicle size, density and shape, it proved possible to distribution of synapses on different regions of the dencategorize synapses into a number of groups (Table 7). drite are listed in Table 1. Axo-axonic and dendro-dendritic synapses. Synapses Ten distinctly different synaptic types were distinguished: 2 axo-somatic synaptic types, 6 axo-dendritic between two vesicle-containing structures were rarely synaptic types, 1 axo-axonic synaptic type, and 1 den- observed (1.4%).Those that were found appeared to be of the asymmetric type with the presynaptic profile dro-dendritic synaptic type (Table 7). Axo-somatic synapses. Axo-somatic synapses covered containing round vesicles of medium size. The postsynonly a small area of the soma. On average, 1.6%of the aptic profile generally contained fewer vesicles, and 580 S.T.MELLER AND B.J. DENNIS Fig. 4. Illustration of some of the various types of vesicles observable in synapses within the PAG. (A), An axodendritic asymmetric synapse showing a wide variety of round vesicles ranging from small to the largest. x 60,000. (B), An axon synapsing on a dendritic spine containing a spine apparatus (large arrow). Nearby, to the left (small arrow) is an axonal profile containing many dense-core vesicles, though not at a synaptic junction. x 27,000. (C), Two asymmetric, axodendritic synapses (arrowed) on a single dendrite. The axonal bouton at the large arrow contains mainly round, clear vesicles. The other contains fewer, flattened vesicles. x 27,000. (D), An example of two neighbouring axo-dendritic, asymmetric synapses containing small, round vesicles, in contact with the same dendritic spine 6). x 68,000. these were almost always of the flattened or pleomorphic type. Occasionally, the presynaptic bouton contained pleomorphic vesicles and the postsynaptic bouton contained round vesicles. These synaptic junctions were either obvious axo-axonic connections, specified by their morphology, or, more infrequently, presumed to be reciprocal dendro-dendritic synapses. Complex synapses. Synapses were very occasionally 581 ULTRASTRUCTURE OF THE RABBIT PAG TABLE 3. Summary of the total numbers of boutons of different sizes, within each subdivision of the PAG, that contain large or small dense-core vesicles (1arge:small)' ~~ PAG region Dorsal Lateral Ventral Medial Average Small 43.1:2.6 28.6:4.8 48.7:O.O 44.3:7.1 42.1:3.9 Bouton Medium 34.4:8.3 40.9:O.O 37.5:O.O 43.8:O.O 39.1:1.8 size Large 20.0:6.3 77.7:O.O 50.0:O.O 50.0:O.O 50.0:2.5 Average 37.0:5.0 38.4:7.1 45.6:O.O 45.5:4.5 42.2:3.1 'Within each division and within each bouton size the remainder of the percentage indicates vesicles that were electron-lucent and did not contain dense-core vesicles. TABLE 4. Summary of the distribution of vesicle size (expressed as a %) in boutons contacting either the soma or elements in the neurovil within the four subdivisions of the PAG' ~ Vesicle size Medium Small Large Small & med Med & lge Small, med, & lge Medial Soma Neuropil 48.1 43.0 11.5 17.8 0.0 2.8 19.2 23.4 15.4 6.5 5.8 6.5 'Small = 20-28 nm diameter, medium (med) Soma 72.0 16.0 0.0 8.0 4.0 0.0 = Dorsal Neuropil 38.0 16.2 4.5 24.2 12.6 4.5 PAG region Lateral Soma Neuropil 47.5 54.0 7.5 16.1 2.5 4.1 27.5 14.9 10.0 9.5 5.0 1.4 34-42 nm, and large (lge) = Ventral Soma Neuropil 55.9 41.9 5.9 23.5 5.9 7.4 11.7 16.0 14.7 8.7 5.9 2.5 Soma 53.7 9.9 1.9 18.0 11.9 4.6 Total Neuropil 43.6 18.2 4.2 20.8 8.6 4.6 50-60 nm. TABLE 5. Summary of the distribution of vesicle density (expressed as a %) in boutons contacting either the soma or elements in the neuropil within the four subdivisions of the PAG' Vesicle density Medium High Low Crystalline Medial Soma Neuropil 48.5 44.5 40.7 47.0 10.8 6.8 0.0 1.7 Dorsal Soma Neuropil 40.0 53.5 40.0 42.0 20.0 4.5 0.0 0.0 PAG region Lateral Soma Neuropil 55.0 48.6 20.0 36.5 22.5 14.9 2.5 0.0 Ventral Soma Neuropil 58.8 50.6 29.4 38.3 11.8 8.6 2.5 0.0 Soma 50.9 32.5 15.9 0.7 Total Neuropil 47.8 42.9 7.9 1.4 lSee text for explanation of density criteria. observed where astrocytic processes within the neuropil, either completely or nearly completely, enveloped synaptic profiles. Such an arrangement suggests that glia may serve, as one of their functions, to separate regions of the neuropil. What functional significance is reflected in this remains to be seen. Variations With Respect to PAG Subdivisions Obvious differences between neurons in different subdivisions of the PAG were: neurons in medial PAG were the smallest; lateral and ventral PAG contained the greatest number and diversity of cells; dorsal PAG contained the largest number of glia; dendrites in the medial region were of the smallest diameter. Within medial PAG, axons were generally unmyelinated, though additionally, isolated myelinated fibres (0.51.2 pm) were observed. In the other regions of the PAG, myelinated fibers more often were grouped in bundles and were larger, varying from 0.4-2.6 pm in diameter. Synapses also showed some slight variation in differ- ent regions of the PAG. Synapses on the soma of neurons in ventral PAG contained more round vesicles and proportionally fewer pleomorphic-shaped vesicles. Also, a greater number of synaptic boutons containing large vesicles contacted the soma of cells in ventral PAG than in other regions. As a general rule, there was a lower density of vesicles at soma synaptic contacts than a t contacts in the neuropil. This was particularly the case for cells in dorsal and lateral PAG. Medial and dorsal PAG showed far more boutons containing densely-packed vesicles on the soma. Medial PAG contained more axo-somatic synapses than other regions of the PAG. In the neuropil, the medial PAG showed most boutons with more pleomorphic vesicles, and lateral PAG the least. Vesicle size was relatively constant in synapses on the neuropil, except in ventral PAG where a slightly larger number of both smaller and larger vesicles were found within synaptic boutons. Similar proportions of small, medium, and large boutons were found throughout the neuropil of all 582 S.T.MELLER AND B.J. DENNIS TABLE 6. Summary of the distribution of bouton size (expressed as a %) in boutons contacting either the soma or elements in the neuropil within the four subdivisions of the PAG' Bouton size Small . medium Large ~~ ISmall = Medial Soma Neuropil 23.1 42.7 51.9 50.9 25.0 6.4 i l km diameter, medium = Dorsal Soma Neuropil 36.0 60.4 52.0 35.1 12.0 4.5 1-2 pm, and large = PAG region Lateral Soma Neuropil 27.5 44.6 40.0 47.3 32.5 8.1 Ventral Soma Neuropil 20.6 39.5 55.9 45.7 23.5 14.8 Soma 25.8 49.7 24.5 Total Neuropil 46.9 45.5 7.6 >2 pm. TABLE 7.A summary of the features of the 10 synaptic types which were discerned within the PAG' Synaptic type Axo-somatic Axo-somatic Axo-dendritic Axo-dendritic Axo-dendritic Axo-dendritic Axo-dendritic Axo-dendritic Axo-axonic Dendro-dendritic Membrane Symmetrical Asymmetrical Symmetrical Symmetrical Symmetrical Asymmetrical Asymmetrical Asymmetrical Asymmetrical Svmmetrical Vesicle sham Round Pleomorphic Round Pleomorphic Pleomorphic Round Round Pleomorphic Round Round Vesicle size Medium Medium Medium Medium Small Medium Small Medium Medium Medium Vesicle density Medium Medium Medium Medium High Medium High Medium Low Low Bouton size Medium & large Medium & large Small Small & medium Small & medium Small & medium Small & medium Small Small Small 'See text for an explanation of the vesicle size and density and bouton size. regions, except dorsal PAG which showed far greater difference between the descriptions for the rabbit and numbers of small boutons and fewer medium-sized cat was the presence, in the rabbit, of soma profiles which had an electron-dense appearance characterized ones. by a large accumulation of a granular-type of material in the cytoplasm. A similar appearance was also noted DISCUSSION in isolated dendritic profiles which would This study has provided a concise description of the occasionally presumably have belonged to cells whose soma would ultrastructure of PAG neurons, dendrites, axons, glia, have a similar structure. Another group of small neuand synaptic contacts in the rabbit. While there have were observed, predominantly in the medial divibeen a number of ultrastructural investigations of the rons which was characterized by a large amount of PAG in other species, most have focused on one feature sion, RER in their cytoplasm. Neither of these cell types of the PAG and none have incorporated the extent of were as having been observed in the cat PAG the analysis provided here. In addition, there has not (Gioiareported et al., 1983; and Gioia, 1984; Moss and been any ultrastructural analysis of the rabbit PAG. Basbaum, 1983). InBianchi other experiments demThe most striking feature of the synaptic ultrastruc- onstrated that some axons and dendritesweofhave PAG neuture of the rabbit PAG was that over 95% of the con- rons enter the aqueduct (Meller and Dennis, 1990a), tacts between neuronal elements occurred within the the increased amount of RER in some cells of the neuropil with axo-dendritic contacts providing 94.1%of thus medial could conceivably be associated with an the total synapses indicating that integration takes increasedivision in synthetic which might therefore be place at the level of the neuropil. The variations in the involved somehow in activity secretion into or absorption from size of synaptic boutons, and the density and size of the CSF. vesicles observed, presumably reflect functional differThe appearance of cytoplasmic organelles, the arences. rangement of the nucleus and nucleolus, and the high nucleus/cytoplasmic ratio were similar to that deFine Structure of the Soma of PAG Neurons and scribed for the cat (Gioia et al., 1983). Earlier studies in the Neuropil the cat indicated the presence of a large number of The soma of PAG neurons were generally round, cytoplasmic inclusions in the nucleus of PAG neurons spindle, or polygonal in shape. Soma sizes of rabbit (Hamilton et al., 1973; Liu and Hamilton, 1976; Liu, PAG neurons were found to be larger than those ob- 1978; Gioia et al., 1983; Bianchi and Gioia, 1985). In served for the cat (Gioia et al., 1983), falling within the contrast, the present study has shown that cells within range of 12-45 p,m diameter. There did not appear to the PAG generally have only one crystalline-like inbe a clear cut-off point which permitted groups of cells clusion; on rare occasions, a fibrillar inclusion was also to be differentiated as either small or large neurons, as observed. Whether these structures play a role in celGioia et al. (1983) reported for the cat. One significant lular control as suggested by Liu and Hamilton (1976) ULTRASTRUCTURE OF THE RABBIT PAG and Bianchi and Gioia (1985) remains to be determined. The number of synaptic contacts on the soma of PAG neurons was found to be quite low when compared with those made on dendrites. On the average there were only 3-4 contacts per soma per section which accounted for 8-10% of the total number of synaptic contacts. There was very little variation in the numbers of synaptic contacts on the soma with respect to cells in dorsal, lateral, ventral, or medial PAG. Gioia et al. (1983) and Bianchi and Gioia (1984) came to a similar conclusion for the cat PAG. Moss and Basbaum (19831, also in the cat, described the number of synaptic contacts on the soma of caudal PAG neurons to be even lower (about 2-3% of total synapses). The number of soma synaptic contacts is probably limited in relation to the area available for synaptic contact on the receptive surfaces. A similar suggestion has been made for the pyramidal cells of the somatosensory cortex where the soma accounts for 4% of the surface area of the cell and 43%of the total area is taken up solely by dendritic spines (Mungai, 1967). In contrast, the soma surface area of large cells in the reticular formation accounts for 20% of the total surface area of neurons (Mannen, 1966). Within the neuropil, the large number of dendrites, cut in any of a number of planes, were observed to range from very large to very small in size and, as the findings of our Golgi study also suggested, dendritic spines were not very prevalent (Meller and Dennis, 1990b). In contrast, studies on the ultrastructure of the PAG in the cat have suggested that dendritic spines are quite numerous (Gioia et al., 1983; Bianchi and Gioia, 1984). Myelinated and unmyelinated axons, which made up a large percentage of the neuropil, are more numerous and larger in both number and size with distance from the aqueduct. The number and size of the myelinated axons was far greater in dorsal, lateral, and ventral PAG regions than in medial PAG, whereas the proportion of unmyelinated axons to myelinated axons was greater in medial PAG, suggesting that neurons of the small-sized medial PAG may be predominantly locally-projecting cells or interneurons. The types and ultrastructural analysis of glia within the PAG has not previously been reported. In the present study, numerous glia were observed in the neuropil, with oligodendrocytes frequently being found in direct apposition to the soma of PAG neurons. Whether they perform a purely nutritive or supportive role is not yet clear, although astrocytes and their processes were observed penetrating the neuropil, where they were occasionally found to either encapsulate a synaptic contact or to contact and surround blood vessels. It has been suggested that this arrangement may limit the free diffusion of neurotransmitters, or alter the distribution of ions around synapses (Schubert, 19841, or control the distribution and density of synapses on the soma and dendrites of cells (Meshul et al., 1987). Synaptic Features of the PAG The major characteristic ultrastructural feature of any synapse is the presence of synaptic vesicles within the presynaptic bouton associated with varied thickenings of the pre- and post-synaptic membranes. On the 583 soma, the most prominent synaptic contacts were symmetric or multiple symmetric synapses containing pleomorphic vesicles (80%) which agrees quite well with the number (88%)assessed by Bianchi and Gioia (1984) for the cat PAG. The size of the synaptic boutons varied from small to large with the most prominent types being the intermediate and large boutons. In the neuropil in the cat, Bianchi and Gioia (1984) reported that 88% of the synapses contained pleomorphic vesicles whereas our estimate for the rabbit was 53.2%, more closely agreeing with the 59.3%proposed for caudal PAG in the cat by Moss and Basbaum (1983). Both of these reports in the cat and our results for the rabbit agree that, in the neuropil, approximately half of the synapses have symmetrical and half asymmetrical junctions. Small and medium-sized boutons were the most prominent contacts, and these were mostly on the middle and distal dendrites. The synaptic density on proximal dendrites is less than that found on more distal dendrites, and the type of synapses found on proximal dendrites are similar in size and morphology to those found at soma contacts. The presence of a large number of axo-dendritic synapses, together with smaller numbers of presumed axo-axonic, dendro-dendritic, and complex synapses suggest that the neuropil is indeed the main target of afferent input to the PAG. Four general categories of vesicles have been found in the presynaptic profiles of the PAG: round, flattened, pleomorphic, and dense-core vesicles. In the PAG of the cat, Gioia et al. (1983) and Bianchi and Gioia (1984) found these same four basic vesicle types, while Moss and Basbaum (1983) recognized only round, pleomorphic, and dense-core vesicles. Not only were four categories recognized in this study, but a range of different sizes within each population has been described. Different vesicle shapes and sizes have been proposed to reflect different functional properties of synapses. Symmetrical contacts have been suggested to be associated with inhibitory synapses and asymmetrical contacts associated with excitatory synapses (Gray, 1959; Eccles, 1964). Further, a correlation between round vesicles and excitation, as against pleomorphic or flattened vesicles and inhibition, has also been proposed (Uchizono, 1965; Bodian, 1970). Of the axo-dendritic synapses in the neuropil (which account for 94.1% of all synapses), a clear majority of them displayed symmetrical contacts containing pleomorphic vesicles or asymmetrical contacts containing round vesicles. This is in line with there being basically inhibition a t the level of the soma, whereas in the neuropil half the synapses result in inhibition and half in excitation. A report by Penny et al. (1984) demonstrated that glutamic acid decarboxylase activity (an indicator of the presence of the inhibitory transmitter, gamma amino butyric acid) is spread relatively evenly throughout the PAG of the rabbit. This present account indicates that presumed inhibitory synapses are also spread evenly throughout the PAG. Although this situation may provide a good correlation between morphological features of a synapse and its supposed functional role, it remains to be seen whether vesicle shape and the polarity of the synapse support the concept of the nature of inhibitory and excitatory synapses. Such features should be closely examined in subsequent 584 S.T. MELLER AND B.J. DENNIS studies where immunocytochemical evidence for known transmitters is available at the TEM level. Boutons containing flattened vesicles and displaying symmetrical contacts were few, though found consistently in the rabbit. This was also the case for the PAG of the cat (Gioia et al., 1983; Bianchi and Gioia, 1984; Moss and Basbaum, 1983). Another consistent finding was the presence of a few large dense-core vesicles of varying sizes spread amongst the usual population of vesicles in a presynaptic profile. These were found in approximately 40% of all synaptic profiles. A smaller number of synapses (3%) contained small dense-core vesicles which appeared similar to those identified a s catecholamine-containing profiles in the cerebellum (Chan-Palay, 19771. Although dense-core vesicles were also reported in the cat (Moss and Basbaum, 1983) and rat (Clements et al., 1985), they were not nearly as prevalent as we have found here in the rabbit. A number of additional endings containing up to a dozen large dense-core vesicles were also observed in the neuropil of the PAG of the rabbit, but these were never observed to form synaptic contacts in the PAG. These profiles resembled the serotonergic axons found by Beaudett and Descarries (1981) and could conceivably be derived from the raphe nuclei. A similar feature was also reported by Clements et al. (1985) who suggested that such axons in the rodent PAG were either travelling through the PAG without releasing transmitters or that they were involved in some type of non-synaptic transmitter release. That, together with the finding by Clements et al. (1985) that very few synapses in the PAG showed serotonin immunoreactivity, may explain why investigators such as Aghajanian et al. (1972) were unable to observe much of a n effect with iontophoretically-applied serotonin. Synapses in the neuropil were found to consist largely of axodendritic synapses (sometimes up to 11 synapses on one dendrite); however, a small number of presumed axo-axonic and dendro-dendritic synapses were also found. The presence of these, together with other complex synapses also observed, would indicate that the PAG is intricately involved in complex processing of afferent information within the neuropil. The existence of axo-axonic connections, though small in number, is indicative of presynaptic inhibition, which has already been proposed to be active in the PAG in pharmacological and physiological studies (Hara et al., 1961; Sakuma and Pfaff, 1980; Soper and Melzack, 1982; Handwerker and Sack, 1982; Reichling et al., 1984). In summary, the present study has provided a comprehensive ultrastructural analysis of the rabbit PAG. The PAG of the rabbit shows qualitatively and quantitatively similar synaptic types to those shown for the cat. However, there also are differences shown between these two species. For example, the PAG of the rabbit shows larger neurons that only contain one cytoplasmic inclusion; other neurons in the PAG are more electron-dense and contain large amounts of RER (typically in medial PAG). In addition, neurons and dendrites have far fewer dendritic spines that neurons and dendrites in the cat. Further, this study has provided descriptions of the ultrastructure of glia in the PAG and also of variations in different subdivisions of the PAG. In general, though, many of the cytological features are similar between the cat and rabbit, further validating and justifying the use of the rabbit in neuroscience research (Dennis and Meller, 1993). These results have provided a clear picture of normal synaptology at all levels, and in each subdivision, of the rabbit PAG. This information, together with burgeoning knowledge of neurotransmitters present within the PAG (see Moss and Basbaum, 1983; Beitz, 1985; Clements et al., 1985) and of the ultrastructural localization of excitatory and inhibitory neurotransmitters to different synapse types (Williams and Beitz 1989, 1990a,b), should be of assistance in determining the specifics of neuronal circuitry whereby the PAG performs its numerous functional roles. ACKNOWLEDGMENTS This study was made possible with support from The Australian Brain Foundation. LITERATURE CITED Aghajanian, G.K., H.J. Haigler, and F.E. Bloom 1972 Lysergic acid diethylamide and serotonin: Direct actions on serotonin-containing neurons in rat brain. Life Sci., 11:615-622. Beaudett, A,, and L. Descarries 1981 The fine structure of central serotonin neurons. J. Physiol. (Paris), 77:193-203. Beitz, A.J. 1985 The midbrain periaqueductal gray in the rat: I. Nuclear volume, cell number, density, orientation, and regional subdivisions. J . Comp. Neurol., 237:445-459. Bianchi, R., and M. Gioia 1984 Ultrastructural features of the synapses of the periaqueductal gray matter (PAG) of the cat. J. Hirnforsch., 25:275-283. Bianchi, R., and M. Gioia 1985 Morpho-quantitative analysis of nuclear inclusions in periaqueductal gray matter neurons in the cat. J. Anat., 140:159-164. Bodian, D. 1970 An electron microscope characterization of classes of synaptic vesicles by means of controlled aldehyde fixation. J . Cell Biol., 44:115-124. Chan-Palay, V. 1977 Cerebellar dentate nucleus, organization, cytology and transmitters. Springer, New York, pp. 416-433. Clements, J.R., A.J. Beitz, T.F. Fletcher, and M.A. Mullet 1985 Immunohistochemical localization of serotonin in the rat periaqueductal gray: A quantitative light and electron microscopic study. J . Comp. Neurol., 236:60-70. Dennis, B.J., and S.T. Meller, Investigations of the periaqueductal gray (PAG) of the rabbit with consideration of experimental procedures and functional roles of the PAG. Progress in Neurobiol., in press. Eccles, J.C. 1964 The physiology of synapses. Springer-Verlag, Berlin. Gabella, G. 1976 Structure of the autonomic nervous system. Chapman and Hall, London. Gioia, M., G. Tredici, and R. Bianchi 1983 The ultrastructure of the periaqueductal gray of the cat. J . Submicrosc. Cytol., 15:10131026. Gyra, E.G. 1959 Axo-somatic and axo-dendritic synapses of the cerebral cortex: An electron microscope study. J . Anat. (Lond.), 93: 420-433. Hamilton, B.L., P.A. Fenner-Crisp, and J.C. Mazziotta 1973 Nuclear inclusions in neurons of the periaqueductal gray matter in the cat. Anat. Rec., 175.504. Handwerker, H.O., and R. Sack 1982 Single unit activity in the rat’s midbrain during nocifensive tail flick reaction. Neurocci. Lett., 30:79-84. Hara, T., E. Favale, G.F. Rossi, and G. Sacco 1961 Responses in mesencephalic reticular formation and central gray matter evoked by somatic peripheral stimuli. Exp. Neurol., 4:297-309. Johansson, O., T. Hokfelt, S.L. Jeffcoate, N. White, and L.A. Sternberger 1980 Ultrastructural localization of TRH-like immunoreactivity. Exp. Brain Res., 38:l-10. Liu, R.P.C. 1978 Electron and light microscopic correlation of intranuclear bodies in neurons of PAG. Anat. Rec., 190:462. Liu, R.P.C., and B.L. Hamilton 1976 Intranuclear bodies in neurons of the periaqueductal gray in the cat. J . Anat., 147:139-145. Mannen, H. 1966 Contribution to the morphological study of dendritic arborization in the brain stem. In: Correlative Neurosciences. Part A. Fundamental Mechanisms. T. Tokizane and J.P. Schad6, eds. Elsevier, Amsterdam, pp. 131-162. ULTRASTRUCTURE OF THE RABBIT PAG Meller, S.T., and B.J. Dennis 1986 Afferent projections to the periaqueductal gray in the rabbit. Neurosci., 19:927-964. Meller, S.T., and B.J. Dennis 1987 Ultrastructural characteristics of the periaqueductal gray matter in the rabbit. Proc. Aust. Physiol. Pharmacol. SOC.,18:58P. Meller, S.T., and B.J. Dennis 1990a A multiple Golgi analysis of the Deriaaueductal may in the rabbit. J . ComD. Neurol.. 302:66-86. Melier, S.?., and B.J. Dennis 1990b QuantitiGive Nissl'study of the neuronal types, and recognition of cytoarchitectural subdivisions, within the rabbit periaqueductal gray. J. Comp. Neurol., 302:8799. Meller, S.T., and B.J. Dennis 1991 Efferent projections of the periaqueductal gray in the rabbit. Neurosci., 40:191-216. Meehul, C.K., F.J. Seil, and R.M. Herndon 1987 Astrocytes play a role in regulation of synaptic density. Brain Res., 402:139-145. Moss, M.S., and A.I. Basbaum 1983 The fine structure of the caudal periaqueductal gray of the cat: Morphology and synaptic organization of normal and immunoreactive enkephalin-labeled profiles. Brain Res., 289:27-43. Mungai, J.M. 1967 Dendritic patterns in the somatic sensory cortex of the cat. J. Anat., 101:403-418. Penny, G.R., M. Conley, D.E. Schmechel, and I.T. Diamond 1984 The distribution of glutamic acid decarboxylase immunoreactivity in the diencephalon of the opossum and rabbit. J . Comp. Neurol., 228:38-56. Pickel, V.M., T.H. Joh, and D.J. Reis 1977a A serotonergic innervation of noradrenergic neurons in nucleus locus coeruleus: demonstration by immunocytochemical localization of the transmitter specific enzyme tyrosine and tryptophan hydroxylase. Brain Res., 131:197-214. Pickel, V.M., D.J. Reis, and S.E. Leeman 197713 Ultrastructural lo- 585 calization of substance P i n neurons of rat spinal cord. Brain Res., 122534-540. Reichling, D.B., S.F. Lakos, and A.I. Basbaum 1984 Intracellular electrophysiological and Golgi analysis of the midbrain periaqueductal grey (PAG) of the rat. SOC.Neurosci. Abstr., 1O:lOl. Sakuma, Y., and D.W. Pfaff 1980 Convergent effects of lordosis-relevant somatosensory and hypothalamic influences on central gray cells in the rat mesencephalon. Exp. Neurol., 70:269-281. Schubert, D. 1984 Developmental Biology of Cultured Nerve, Muscle, and Glia. John Wiley and Sons, New York. Soper, W.Y., and R. Melzack 1982 Stimulation-produced analgesia: Evidence for somatotopic organization in the midbrain. Brain Res., 251 :301-311. Uchizono, K. 1965 Characteristics of excitatory and inhibitory synapses in the central nervous system of the cat. Nature (Lond.), 207:642-643. Williams, F.G., and A.J. Beitz 1989 A quantitative ultrastructural analysis of neurotensin-like immunoreactive terminals in the midbrain periaqueductal gray: Analysis of their possible relationship to periaqueductal gray-raphe magnus projection neurons. Neurosci., 29:121-134. Williams, F.G., and A.J. Beitz 1990a Ultrastructural morphometric analysis of GABA-immunoreactive terminals in the ventrocaudal periaqueductal grey: Analysis of the relationship of GABA terminals and the GABA, receptor to periaqueductal grey-raphe magnus projection neurons. J. Neurocytol., 19:686-696. Williams, F.G., and A.J. Beitz 1990b Ultrastructural morphometric analysis of enkephalin-immunoreactiveterminals in the ventrocaudal periaqueductal gray: Analysis of their relationship to periaqueductal gray-raphe magnus projection neurons. Neurosci., 38:381-394.