OVERDUE FINES: 25¢ per day per item RETURNING LIBRARY MATERIALS: Place in book return to remove charge from circulation record: MDRPHOLOGICAL ANALYSIS OF RAT NEOSTRIAIAL NEURONS By Howard Tien—Haw Chang A DISSERTATION Submitted to Michigan State University in partial fullfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Anatomy 1981 r.n.m.\..4n: .0 ABSTRACT MORPHOLOGICAL ANALYSIS OF RAT NEOSTRIATAL NEURONS BY Howard TienrHaw Chang The aim of the present study was to investigate the morphological characteristics and the synaptic relationships of individual neurons of the rat neostriatum. Both light and electron microscOpy were employed. Neurons were impregnated with the rapid Golgi technique and subsequently processed by the Golgi gold-toning method. Additional material included some efferent neurons labeled retrogradely with extracellulary injected horseradish peroxidase (HRP). Some neurons labeled intracellularly with HR? were also studied. At least two types of large neurons (somatic cross-sectional areas, SA > 300 square um) and five types of medium neurons (SA between 100 and 300 square um) were distinguished in Golgi preparations of the adult rat neostriatum. Type I large cells have aspinous sonata with long radiating, sparsely-spined dendrites which may be varicose distally, whereas type II large cells have spines on both somatic and dendritic surfaces. Medium type I cells have aspinous somata and proximal dendrites, but their distal dendrites are densely covered with spines. Medium type II cells have somatic spines, and their radiating dendrites are sparsely-spined. Other medium cells have no somatic spines: Type III cells have poorly-branched and sparsely-spined dendrites. Type IV cells have profusely-branched, sparsely-spined dendrites. Type V cells have Howard T. Chang radiating and varicose dendrites which may also be sparsely-spined. At the electron microscopic level, the presence of many axosomatic synapses on the type II large neurons distinguishes their somata from those of type I large cells. Deeply indented nuclei, dense bodies, well developed rough endOplasmic reticulum, and a perinuclear Golgi complex are found in both types of large neurons. For the medium neurons, only type I neurons have round, unindented nuclei, with the other types having varying degrees of nuclear indentations. Intranuclear rods are found in the type II medium neurons. Axosomatic synapses are found on all types of medium neurons, but are most numerous on the type II cells. Numerous ribosomes but only a few stacks of rough endoplasmic reticulum are present in the cytOplasm of the type I medium neurons. Other medium neurons have fewer free ribosomes, but have more stacks of rough endOplasmic reticulum. The dendrites of the type I medium neurons receive asymmetrical axospinous synapses with terminals which contain small to medium vesicles. The few synapses found on their dendritic shafts are mostly symmetrical and the presynaptic terminals contain large vesicles. Dendrites of other neurons form synapses with terminals which contain mainly small to medium vesicles. No serial synapses have been observed. Although the main axons of both types of large cells and one type IV medium neuron are myelinated as are those of type I medium cells, only the latter have been demonstrated unequivocally to be efferent neurons. Their axons have been traced into the globus pallidus. The distribution patterns of the intrinsic axon collaterals Howard T. Chang as well as the efferent axons of individual type I medium cells vary greatly, indicating that these neurons may have heterogenous functions. The axon terminals of medium type I cells contain large vesicles and make mainly symmetrical axodendritic and axosomatic synapses with their target neurons. Except for the type II large cells, intrinsic axon collaterals have been observed for every type of neuron. The unmyelinated axon of a type V medium neuron arborizes extensively near the parent dendritic field and make symmetrical synapses with somata and dendrites of neighboring neurons, most of which have the characteristics of type I medium neurons. Several small neurons (SA mostly less than 100 square nm) are also found in the rat neostriatum: some have aspinous soma with sparsely-spined dendrites, others have somatic spines. Rarely, medium neurons with dendrites exceeding 500.nm in length are found, but it is premature to classify them as separate types of neurons. To my parents, who made everything possible. ii ACKNOWLEDGEMENTS I thank.my advisor, Dr. S.T. Kitai, for his guidance and encouragement throughout my graduate career. I am also grateful to the advice and support from the following teachers and friends: W.M. Falls, I. Grofova, M.R. Park, P. Patterson, R.J. Preston, D. Steindler, C.A. Tweedle and C.J. Wilson. I thank Dr. J.A. Rafols of Wayne State University for introducing me to the art of both Golgi and electron microscOpic analysis of the nervous system. I also thank Dr. Karen Baker and her skillful assistants of the Center for Electron Optics at MSU in develOping my initial skills in transmission electron microscOpy. I thank Dr. M.R. Park and Ross Doil for their assistance in developing the computer program for measuring the cell body cross-sectional areas. I thank Dr. C.J. Wilson for the use of his wonderful Hasselblad camera. I am also grateful to the timely encouragements from my classmates and staffs on the 5th floor of West Fee Hall: Jim Lighthall, Kevin Phelan, Salman Afsharpour, Bill Bitzinger, Julie King and Carol Heister. Finally, I thank Pat Kowalski, without her patient support and companionship, all of those long hours in the darkroom and in front of the microtome, the light microsc0pe, the electron microscOpe, the typewriters and the computer terminal would have been unbearable. iii FOREWORD Owing to the complex nature of some of the eXperimental procedures employed in the intracellular labeling studies, these investigations were carried out in collaboration with Dr. G.A. BishOp and Dr. S.T. Kitai in 1979, and with Dr. C.J. Wilson and Dr. S.T. Kitai in 1980. Without their patient directions and skillful assistance, these studies would have been impossible. Portions of this study have been reported in the following papers and communications: BishOp, G.A., H.T. Chang and S.T. Kitai (1979) Light and electron microscOpic analysis of various neostriatal neurons intracellularly labeled with HRP: II. Patterns of axonal distribution. Soc. Neurosci. Abstr. 5, 67. Chang, H.T., G.A. BishOp and S.T. Kitai (1979) Light and electron microsc0pic analysis of various neostriatal neurons intracellularly labeled with HRP: I. Sons-dendritic morphology. Soc. Neurosci. Abstr. S, 69. Chang, H.T., C.J. Wilson and S.T. Kitai (1980) Collateral arborization of axons of rat striatal neurons in the globus pallidus studied by intracellular horseradish peroxidase labeling. Soc. Neurosci. Abstr. 6, 269. Chang, H.T. and S.T. Kitai (1981) Rat neostriatal neurons: A Golgi study. Anat. Rec. 199, 48A. BishOp, G.A., H.T. Chang and S.T. Kitai (1981) Morphological and physiological properties of neostriatal neurons: An intracellular HRP Study in the rat. Neurosci. (in press). Chang, H.T., C.J. Wilson and S.T. Kitai (1981) Single neostriatal efferent axons in the globus pallidus: A light and electron microscpic study. Science (in press). iv TABLE OF CONTENTS List Of TableSCOOOOOOCOO...O00....OOOOOOOOOOOOOOOOOOO0.00......OOOOOOOOOVi List Of Figur6800000000000000OOOOOOOOOOOOOOO0.00.0.0...OOOOOOOOOOOOOOCOVii Chapter 1. A. B. C. F. G. Chapter 2. A. B. C. D. Chapter 3. A. B. C. D. E. Chapter 4. A. B. C. D. Chapter 5. INTRODUCTION.................................................1 Gross Anatomy of Neostriatum.................................1 Historical Background3 Neostriatal Afferent Connections.............................7 a) The Cerebral Cortex.......................................7 b) The Thalamus9 c) The Substantia Nigra.....................................11 d) The Dorsal Raphe Nucleus.................................13 e) Other Minor Afferents....................................14 f) Physiology of Neostriatal Afferents......................14 Neostriatal Efferent Connections............................16 a) The Pallidum.............................................16 b) The Substantia Nigra.....................................17 c) Physiology of Neostriatal Efferents......................17 Neostriatal Intrinsic Organization..........................19 a) Light Micros00pic Studies of Nissl Preparations..........19 b) Golgi Studies............................................19 c) Electron Microscopic Studies of Normal Material..........25 d) Golgi Gold-toned Neostriatal Neurons.....................26 e) Extracellular Labeling Studies...........................27 f) Intracellular Labeling Studies...........................28 g) Biochemical, Histochemical & Immunocytochemical Studies..29 Summary of Backgrounds......................................32 Research Objectives.........................................34 METHODS.....................................................35 Golgi.Studies...............................................35 Extracellular Horseradish Peroxidase Labeling...............37 Intracellular Horseradish Peroxidase Labeling...............39 Method of Analysis..........................................41 RESULTS.....................................................42 Golgi Studies...............................................42 Electron Microsc0pic Analysis of Golgi Material.............83 Extracellular HRP Studies..................................118 Rat Neostriatal Neurons Intracellularly Labeled with Horseradish Peroxidase.....................................125 Somatic Sizes of Rat Neostriatal Neurons...................159 DISCUSSION.................................................162 Comparison of Classification Systems.......................162 Light MicroscOpic Morphology of Rat Neostriatal Neurons....169 Electron Microsc0pic Morphology of Rat Neostriatal Neurons.182 The Intrinsic Organizaion of the Neostriatum...............194 SUMMARY OF NEW FINDINGS....................................199 References...OCOOOOOOOOOOOOCOOOOOIOOOOOOOOO...0.0......0.0.0.0000000000202 LIST OF TABLES 1. Summary of rat neostriatal neuronal types.............164 2. Comparison of present classification with previous Golgi studies of the neostriatal neurons..............168 vi 11. 12. 13. 14. 150 16. 17. 18. 19. LIST OF FIGURES Photomicrographs of type I large neurons..............44 MicrosCOpic tracings of type I large neuron...........46 Photomicrographs of type II large neurons and type I and type II medium neurons.....................SO Microscopic tracings of type II large and type II medium neurons...OOOOOOOOOOOOOOOOOOOOOO0.0.00052 MicrosCOpic tracings of type I medium neurons.........54 Photomicrographs of type III medium neurons and a medium neuron with long dendrites.....................59 Microscopic tracings of type III medium neurons and a small neuron with intrinsic axons...............61 Photomicrographs of type IV medium neurons............65 Microscopic tracings of type IV medium neurons........67 Photomicrograph of a type V medium neuron.............69 Microscopic tracings of a type V medium neuron.. . . . . . .71 Photomicrographs of small neurons.....................76 Microscopic tracings of small neurons.................78 Microscopic tracings of medium neurons with long dendriteSOCOOOOOOOOOOOOOOOOOO0.0000000000000000I00.00080 Photomicrographs of_a neurogliaform cell..............82 Electronmicrograph of a type I larage neuron..........86 Electronmicrographs of soma and dendrites of a type I large neurODOCOOOOOOOOO...OIOOIOOOOOOOOOOOOOIOOOOO....88 Electronmicrographs of soma and axon of a type II large neuronIOOOOOOOO000......OIOOOOOOIOOOOOOOOOOIOOO0% Electronmicrographs of a somatic spine and dendrites Ofatype II large neuronOOOOOOOOOOOOOOOOOO0.00.0.0.0092. vii 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. Electronmicrographs of dendrites of a type II large neuron, showing synapses and partial impregnations....94 Electronmicrographs of a type I medium neuron........1OO Electronmicrographs of a type II medium neuron.......102 Electronmicrographs of dendrites of a type II medium neuronOOOOOOOOOOOOOOOOOOOO0.0.00.0...00.0.00...0.0.0.104 Electronmicrographs of a type III medium neuron......106 Electronmicrographs of soma and axon of a type IV medi‘m neuronOOOIIOOOOOOOOOOOOOOOOOO0.00.00...00.00.0109 Electronmicrographs of dendrites of a type IV medium neuronOOOOO00.0.0000...OOOOOOOOOOOOOOOOOOOOOO0.0.0.00111 Electronmicrographs of a type V medium neuron........115 Electronmicrographs of a small neuron................117 Photomicrographs of the site of extracellular horseradish peroxidase injection and the retrogradely labeled neostriatal projection neurons...............120 Electronmicrographs of a neostriatal projection neuron retrogradely labeled with horseradish peroxidase.....124 Microscopic tracings of aspinous neurons labeled intracellularly with horseradish peroxidase..........128 Photomicrograph and electronmicrographs of a type I large neuron labeled intracellularly with horseradish perOXidaseOOOOOOOOOOOOOOOOO0.00000000000000.0.0.00000131 Electronmicrographs of dendrites of the intracellularly labeled type I large neuron..........................133 Photomicrograph and electronmicrographs a medium aspinous neuron intracellularly labeled with horseradish peroxidase...............................136 Microscopic tracings of type I medium neurons labeled intracellularly with horseradish peroxidase..........14O Microscopic tracings of the axons of a type I medium neuron which projects into the globus pallidus.......144 Electronmicrographs of the soma of a type I medium neuron intracellularly labeled with horseradish perOXidaseOOOOIOOOOO0.0.0.0...00.00.00.000...0.0.00.0147 viii 38. 39. 40. 41. 42. Electronmicrographs of synapses on soma, dendrites and axon initial segment of a type I medium neuron intracellularly labeled with horseradish peroxidase..152 Electronmicrographs of the axons and intrinsic terminals of a type I medium neuron intracellularly labeled with horseradish peroxidase..................154 Photo- and electronmicrographs of the efferent axon of a type I medium neuron and its terminal arborization in the BIObuS filliduSOOOOOOOOOOOOOOOOO0.0.0.0.0.....156 The distribution of the diameters of the pallidal dendrites which are postsynaptic to a neostriatal typelmwi‘m neuronOOOOO0.0.000...0.0.0.000000000000158 The distribution of somatic sizes of rat neostriatal' projections neurons labeled retrogradely with horseradish peroxidase extracellularly injected in the globus pallidus, and the somatic sizes of neostriatal neurons impregnated by the rapid Golgi method........161 ix Chapter 1. INTRODUCTION Although previous anatomical studies have described a number of neostriatal neuronal types in various species, the classification and the morphological descriptions of these neurons have remained controversial. The aim of the present study was to investigate the morphological characteristics of various types neostriatal neurons in the rat at both the light and electron microsCOpic levels. The rapid Golgi method was used to establish a classification system for the neostriatal neurons based on their light microsc0pic morphology. The Golgi gold-toning procedures were used to analyze the ultrastructural features of these different types of neurons. Attemps were also made to identify the morphological characteristics of the neostriatal projection neurons which were labeled retrogradely with horseradish peroxidase (HRP) fellowing extracellular injection in the neostriatal target areas. Additionally, both light and electron microsc0pic analysis were carried out on physiologically characterized neostriatal neurons which have been intracellularly labeled with HRP. Background: A. Gross Anatomy of Neostriatum The neostriatum is a large telencephalic structure entirely covered by the cerebral cortex and the subcortical white matter dorsally and laterally. Mediodorsally it borders the lateral 1 ventricles and rostroventrally it is contiguous with the nucleus accumbens septi (the fundus striati). Medioventrally and caudally it is Contiguous with the external segment of the globus pallidus. Bundles of thinly myelinated fibers within this gray matter converge radially toward the paleostriatum as the "spokes of a wheel" (Papez, 1942) and are responsible for its striated appearance. The neostriatum together with the paleostriatum constitute the corpus striatum, generally regarded as the principal component of the so-called "extrapyramidal system" (Wilson, 1912). Frequently, the neostriatum is simply refered to as the striatum, whereas the paleostriatum, consisting of both the external and internal segments of the globus pallidus, is often refered to as the pallidum. The globus pallidus of the rodents is homologous to the external segment of the globus pallidus of the primates, and their entopeduncular nucleus is homologous to the internal segment of the globus pallidus of the primates. The corpus striatum arises from the basal plate of the telencephalon as a single gray mass at the level of the interventricular feramen (Hamilton & Mossman, 1972). As the cerebral cortex expands over the diencephalon of primates and carnivores during development, part of the striatal ridge is carried in the wall of lateral ventricle into the temporal lobe. Developing corticoPedal and corticofugal fibers course through this gray matter to join the internal capsule and the cerebral cortex. Thus, the two parts of the neostriatum, caudate nucleus medially and putamen laterally, are continuous rostrally and incompletely separated caudally by the internal capsule in primates and carnivores, but are essentially indistinguishable in rodents in which the thickely myelinated fiber bundles of the internal capsule fan out within the mass of the neostriatum before reaching the subcortical white matter. Since the putamen and the globus pallidus lie adjacent to each other and together they have a lens-shape appearance, they form the lentiform nucleus described in many older literatures (see review by Nauta & Mahler, 1966). The corpus striatum is sometimes refered to as the basal ganglia by clinicians (see review by Graybiel & Ragsdale, '79). However, "basal ganglia" has also been used to designate most or all of the non-cortical telencephalic gray matter, including the claustrum, the amygdaloid complex (archistriatum) and other gray matter ventral to the neostriatum and paleostriatum (e.g. the nucleus accumbens septi, the olfactory tubercle and the substantia innominata) which has been suggested to be the ventral extension of the striatum and pallidum (Heimer & Van Hoesen, 1979). More recently, the substantia nigra and the subthalamus have been included in the basal ganglia due to their intimate anatomical connections with the corpus striatum (Graybiel & Ragsdale, 1979). B. Historical Background Prior to the development of modern empirical scientific research methods in the 19th century, speculations of neostriatal functions were based primarily on its proximity to other brain structures. By virtue of its position near the cerebral ventricles approximately midway between the cerebral cortex and brain stem, the corpus striatum was described as "internode by which the cerebrum coheres with the medulla oblongata" by Willis in 1667, who believed that all incoming sensations were received by the striatum, which was therefore called "sensorium commune" (cited by Wilson, 1914). This sensory relaying role was also promoted by Swedenborg who declared, in 1740, "The royal road of the sensations of the body to the soul is thorugh the corpora striata, and all determinations of the will also descend by that road.", and that it was "the Mercury of the Olympus, it announces to the soul what is happening to the body, and it bears the mandate of the soul to the body." (cited by Wilson, 1914). Speculations of striatal functions found certain restriction as new experimental brain research methods were developed to test these theories. In 1820, Flourens failed to elicit movements in the rabbit by mechanical stimulation of either the striatum, or the cortex (cited by Wilson, 1914). Magendie, in 1841, fOund similar negative results with mechanical stimulation of the striatum with a needle, but he found that rabbits with large bilateral striatal lesions lept forward as if under an irresistible impulse. Magendie, therefore, concluded that the striatum controlled backward movements which normally would be in balance with the cerebellum which controlled the forward movement. Thus, if the striatum were destroyed, the animal would run ferward due to unchecked cerebellar actions. However, Longet in 1842, failing to reproduce similar results, declared that Magendie's theory was "purement imaginaire" (cited by Wilson, 1914). Nevertheless, the possible motor function of the striatum and the belief that it was a station between the cerebral cortex and the spinal cord stimulated numerous experiments in which the striatum was pinched, pricked, pierced, electrified or injected with various chemicals to see whether or not such treatments would produce movements (see review by Wilson, 1914). As both positive and negative results were obtained in these rather crude experiments, the role of the striatum in motor functions remained controversial. Moreover, as anatomical studies demonstrated that the fibers of the pyramidal system were uninterrupted on their way from the cortex through the striatum, various functions previously assigned to the striatum were shown to be the properties of the adjacent cortico-spinal paths (i.e. the internal capsule), the striatum "almost at once fell from its high estate and depreciated in physiological significance" (Wilson, 1914). The motor role of the striatum received a major boost from the clinico-pathological and experimental lesion studies by Wilson (1912, 1914) who was also one of the earliest to use stereotaxic methods in the experimental investigation of striatal connections with the help of Sir Victor Horsley (Wilson, 1914). Behavioral changes manifested mainly in terms of motor functions were observed in patients with intact pyramidal tracts but with either hepato-lenticular degeneration (Wilson, 1912) or other lesions involving the corpus striatum (Vogt & Vogt, 1920; McLardy, 1948). Using the classical but not very sensitive Marchi method of staining the degenerating axons, Wilson did not observe cortico-striatal connections. This result together with the assumption that the corpus striatum gave rise to an independent descending pathway to the spinal cord (his lenticulo-rubro-spinal projecting system) led him to believe that corpus striatum was completely independent of the cerebral cortex, hence independent of the pyramidal tracts. Although this view is no longer valid in light of later studies using more sensitive methods which enabled the visualization of the massive cortico-striate projections, Wilson’s terminology, the "extrapyramidal system" (Wilson, 1912), is still in continuous use. Perhaps this is for the lack of a better term describing the functions and/or the clinical manifestation of the dysfuncitons of the basal ganglia and its related nuclei. In addition to the classical clinico-pathological and experimental lesion methods used by Wilson (1912, 1914), other experimental approaches have been used in the investigation of neostriatal functions. These methods include various versions of electrical stimulation, neurochemical tests and electrOphysiological single unit recording. Often, more than one approach was pursued simultaneously. From the results of these experiments, the neostriatum has been ascribed a number of functions such as motor, sensory, inhibitory and cognitive (see review by Divac & Oberg, 1979). It is clear that interpretations of data obtained in these experiments have been based on the contemporary knowledge of the anatomical and physiological characteristics of the neostraital neurons and their intrinsic as well as extrinsic connections. Since these theories have remarkably broad definitions, they are not mutually exclusive. Thus, the functional role of the neostriatum remains as enigmatic as the definition of "sensorimotor integration" (Ungerstedt, Ljungberg & Ranje, 1977). C. Neostriatal Afferent Connections The majority of the afferent inputs to the corpus striatum have been demonstrated to terminate in the neostraitum. These afferents arise from the cerebral cortex, the intralaminar thalamic nuclei, the substantia nigra and the dorsal raphe nucleus. (For more detailed review on the neostriatal connections, see Graybiel & Ragsdale, 1979; Grcfova, 1979; Dray, 1979; and Carpenter, 1976). a) The Cerebral Cortex Although Ramon y Cajal (1911) observed the corticostriate connections by the axon collaterals of pyramidal cells which "transmit voluntary motor excitation" in his Golgi preparations, neither he nor his contemporaries could demonstrate any direct corticostriate connections (i.e. cortical axons which terminate mainly or exclusively in the striatum). With the development of the sensitive silver staining technique fer degenerating axons (Nauta & Gygax, 1954), however, the corticostriate projection was conclusively demonstrated (webster, 1961; Carmen, Cowan & Powell, 1963; Carman, Cowan, Powell & Webster, 1965; Webster, 1965; Kemp & Powell, 1970). In all species studied, the neostriatum receives axons from all parts of the cerebral cortex in a topographical pattern. The corticostriate fibers, mainly ipsilateral but also some contralateral projections (Carmen et al., 1965), are orderly arranged, showing distinct mediolateral, dorsoventral and anteroposterior organization with some degrees of overlap. Studies with recently developed axon pathway tracing techniques (i.e. HRP and autoradiography) revealed further details about the corticostriate connections. The cortical terminal fields in the striatum have a patchy or strip appearance suggesting clustering of terminals (Kfinzle, 1975). The primary motor and sensory areas project almost exclusively to the entire rostrocaudal extent of the putamen (Kfinzle, 1975, 1977; Jones et al., 1977) whereas the prefrontal cortex project almost exclusively to the caudate nulceus (Goldman & Nauta, 1977). On the other hand, premotor area (Kfinzle , 197a) and temporal cortex (Yeterian, 1978) project extensively to both caudate and putamen. After analyzing the intricate patterns of the corticostriate projections from their autoradiographic material in the monkey, Yeterian and van Hoesen (1978) suggested that a given region of the striatum receives input not only from a particular area of cortex, but also from all other cortical areas reciprocally interconnected with that area. Whether this is also true for all cortical areas or for other species remains unclear. The cells of origin of corticostriate fibers have been labeled retrogradely with HRP extracellularly injected in the striatum and are found in the layer III and layer VI of the cat (Kitai et al., 1976; Oka, 1980) or superficial part of the cortical layer V in the monkey (Jones et al., 1977) and the rat (Hedreen, 1977, Wise & Jones, 1977). Based on the soma size and the location of these neurons, most of these authors suggest that the corticostriate pathway is an independent cortical projection system different from those of corticospinal or corticobulbar fiber systems. This is in apparent contrast to the observations by both Ram6n y Cajal (1911) and Webster (1961) in their Golgi preparations in which many collaterals arising from the fibers of the internal capsule terminated within the 9 neostriatum. Ram6n y Cajal (1911) also observed that "In the rat and mouse, pyramidal cell axons have occasionally been observed to emit collaterals at intervals in the passage through the corpus striatum ... and to all appearances transmit voluntary motor excitation". Recent electrOphysiological studies indicate that at least some cortical neurons project to both the striatum and the brain stem (Endo, Araki & Yagi, 1973; Oka & Jinnai, 1978; Jinnai & Matsuda, 1979). Additionally, several electrophysiologically identified pyramidal-tract cortical neurons in the rat were labeled intracellularly with HRP and the distribution of their axons analyzed with the light microsc0pe (Donoghue & Kitai, 1980). Collaterals arising from the HRP labeled main axons in the internal capsule were observed to branch within the neostriatum and appeared to have terminal boutons. Therefore, the corticostriate pathway is likely composed of fibers both independent from those of corticospinal or corticobulbar pathways as well as collaterals from these fiber systems. The cortical afferent boutons contain small to medium round vesicles and make asymmetrical synapses mainly with dendritic spines of neostriatal neurons (Kemp & Powell, 1971c; Hassler et al., 1978; Hattori et al., 1979; Frotscher et al., 1981; Somogyi et al., 1981). b) The Thalamus The thalamostriatal connection was first detected in human neuropathological material (Vogt & Vogt, 1941; McLardy, 1948). Subsequently, this pathway was investigated in eXperiments in which 10 lesions were placed in the striatum in order to produce retrograde cellular degenerations in the thalamus (Droogleever—Fortuyn & Stefens, 1951; Powell & Cowan, 1954; Cowan & Powell, 1955; Powell & Cowan, 1956, 1967). Lesions were also placed in various regions of the thalamus in order to study the degenerating axons, some of which were followed to the striatum (Nauta & Whitlock, 1954; Jones, 1961; Mehler, 1966). Retrograde axonal transport of HRP extracellularly injected in the striatum (Jones & Leavitt, 1974; Nauta et al., 1974; Royce, 1978b; Sato et al., 1979) demonstrated that this pathway is strictly ipsilateral and organized tOpographically in the anteroposterior direction. These studies showed that the intralaminar thalamic neuclei, mainly the centromedian-parafascicular complex and the medial part of the centrolateral nucleus were the principle sources of this pathway. Additionally, some neurons in the ventroanterior, mediodorsal, anteromedial and rhomboid thalamic nuclei also contribute to this projection (Royce, 1978b; Sato et al., 1979). Studies with the autoradiographic techniques (i.e. anterograde axoplasmic transport of radioactively labeled proteins) showed that intralaminar nuclei project densely to the striatum and diffusely to the cerebral cortex (Herkenham, 1978; Kalil, 1978; Royce, 19788). It is interesting to note that intralaminar nuclei project mainly to the putamen in the monkey (Kalil, 1978) but mainly to the caudate nucleus in the cat (Royce, 1978a). Patchy distribution of thalamostriatal tenninal fields are noted by these authors, but their} relationships to those of corticostriatal terminal fields remain unclear. It is also not clear whether there are papulations of thalamic neurons projecting exclusively to the striatum or groups of 11 neurons which project to both striatum and cortex. The thalamostriatal terminal boutons contain small round vesicles and make asymmetrical synapses mainly with dendritic spines of striatal neurons (Kemp & Powell, 1971c; Chung et al., 1977). c) The Substantia Nigra The nigrostriatal connection was demonstrated before either the the corticostriatal or the thalamostriatal pathways by the observations that striatal lesions caused retrograde cellular. degenerations in the substantia nigra (von Monakow, 1895; Holms, 1901; Fox & Schmitz, 1944; Rosegay, 1944). However, this pathway could not be demonstrated by either the Marchi method (Ranson, Ranson & Benson, 1941; Rosegay, 1944) or the Nauta technique (Afifi & Kaelberg, 1964; Carpenter & McMasters, 1964; Carpenter & Strominger, 1967; Cole, Nauta & Mehler, 1964; Faull & Carmen, 1968). The degenerating nigral axons were traced to the pallidum but none could be traced into the striatum. The nigrostriatal projection, however, was convincingly demonstrated from biochemical and histochemical (i.e. catecholamine histofluorescence) techniques in the early 1960’s (Andén et al., 1964; 1965; 1966; Bédard, Larochelle & Parent, 1969; Fuxe, 1965; H8kfelt & Ungerstedt, 1969; Poirier & Sourkes, 1967; Ungerstedt, 1971; Lindvall & Bjorklund, 1974). Thereafter, the nigrostriatal pathway was revealed by the Fink-Helmer and Wiitanen silver techniques (Carpenter & Peter, 1972; Ibata, Nojyo, Matsuura & Sano, 1973; Maler, Fibiger & McGeer, 1973; Usunoff, Hassler, Romansky, Usunova & Wagner, 1976) and by retrograde transport of HRP extracellularly injected into the 12 striatum (Carter & Fibiger, 1977; Fallon & Moore, 1978; Faull & Mehler, 1978; VanderMaelen, Kocsis & Kitai, 1978; Kuyper, Kievit & GroeneKlevant, 1974; Miller, Richardson, Fibiger & McLennan, 1975; Nauta, Pritz & Lasek, 1974; Royce, 1978b; Sotelo & Riche, 1974; Sato et al., 1979). The nigrostriatal fibers ascend via the medial forebrain bundle (MFB) and project tapographically to various parts of the neostriatum with a distinct mediolateral, rostrocaudal and reversed dorsoventral organization (Fallon & Moore, 1978). Although this pathway is considered to be ipsilateral, a few contralateral nigral neurons have been labeled from extracellular injections of the neostriatum in the cat (Royce, 1978b). The nigrostriate fibers arise largely from the pars compacts of the substantia nigra (SNpc) where a concentration of dopamine containing neurons are located. Additionally, neurons in the pars reticulata of the substantia nigra (SNpr) and midbrain tegmental areas (e.g. retrorubral area) also project to the neostriatum (Fallon & Moore, 1978; Royce, 1978; VanderMaelen et al., 1978). On the other hand, the existence of a minor non-catecholaminergic projection from the substantia nigra to the neostriatum is suggested from studies with retrograde labeling combined with either selective lesion (Fibiger, Pudritz, McGeer & McGeer, 1977) or histofluorescence techniques (Guyenet & Craine, 1980; van der Kooy, Coscina & Hattori, 1981). Electron microscOpic studies indicate that the nigral inputs make mainly asymmetrical axodendritic and axospinous synapses on striatal neurons (Kemp & Powell, 1971; Hattori et al., 1973; Bak, Choi, Hassler, Usunoff & Wagner, 1975). Labeling with 5-hydroxydopamine (Arluison, Agid & Javoy, 19788; Groves, 1980) or 13 tritiated dopamine (Arluison et al., 1978b) or tritiated norepinephrine (Aghajanian & Bloom, 1967) indicate that dOpaminergic terminals make mainly asymmetrical synapses. These results agree with those derived from lesions with 6-hydroxydopamine which have additionally demonstrated that the terminals of non-catecholaminergic nigrostriatal afferents contain small pleomorphic vesicles and form symmetrical axodendritic synapses (Hattori, Fibiger, McGeer & Malwr, 1973). d) The Dorsal Raphe Nucleus Projection from dorsal raphe nucleus to the neostriatum has been demonstrated recently with retrograde HRP transport technique (Couch & Goldstein, 1977; Miller, Richardson, Fibiger & McLennaan, 1975; Nauta, Pritz & Lasek, 1974; Jacob, Foote & Bloom, 1978) and the autoradiographic axonal tracing technique (Azimita & Segal, 1978; Bobillier, Petitjean, Salvert, Ligier & Seguin, 1975; Bobillier, Seguin, Petitjean, Salvert, Touret & Jouvet, 1976; Conrad, Leonard & Pfaff, 1974). Both biochemical and histofluorescence studies indicate that the raphe-striatal fibers convey serotonin (5-hydroxytryptamine, or 5-HT) to the neostriatum (Kotowski, Giacalone, Garattim & Valzelli, 1968; Kuhar, Aghajanian & Roth, 1972; Kuhar, Roth, Aghajaninan, 1971; Poirier, McGeer, Larochelle, McGeer, Bedard, & Boucher, 1969; Poirier,Singh, Boucher, Bouvier, Olivier & Larochelle, 1967; Geyer, Puerto, Dawsey, Knapp, Bullard & Mandell, 1976; Holman and Vogt, 1972). Using microdissection and microassay techniques, it is found that the 5-HT terminals are more concentrated in the ventrocaudal neostriatum (Ternaux et al., 1977) which is consistent with the data 14 obtained from autoradiographic axoplasmic transport studies (Bobillier et al., 1975; Conrad et al, 1974). Recent immunocytochemical studies suggest that 5-HT terminals make asymmetrical axospinous synapses in the cat and monkey neostriatum (Pasik, Pasik, Saavedra & Holstein, 1981). e) Other Minor Afferents Several studies based on retrograde HRP axonal tracing techniques have described various other neostriatal afferent areas, including the locus coeruleus (Couch & Goldstein, 1977) and the globus pallidus (Staines, Atamadja & Fibiger, 1981). However, since these pathways have not been confirmed by independent studies, their true nature remains unclear. f) Physiology of Neostriatal Afferents The available data agrees that the afferents from the cerebral cortex and the thalamus exert predominantly excitatory synaptic actions on neostriatal neurons (Purpura & Malliani, 1967; Buchwald, Price, Vernon, & Hull, 1973; Kitai, Kocsis, Preston & Sugimori, 1976; Kocsis, Sugimori & Kitai, 1977; VanderMaelen & Kitai, 1980). The transmitters of the corticostriatal pathways appear to be glutamate (Spencer, 1976; Divac, Fonnum & Storm-Mathisen, 1977; McGeer, McGeer, Sherer & Singh, 1977; Fonnum & Walaas, 1979; Walaas, 1981), whereas those of the thalamostriatal fibers remain unclear (Simke & Saelens, 1977; Fonnum & Walaas, 1979). The physiological effect of the substantia nigra as well as the action of dapamine on the neostriatal neurons have remained 15 controversial. Studies with extracellular recordings and microiontOphoresis techniques have shown that nigral stimulation or dopamine iontophoresis produced either inhibitiory (Ben-ARi & Kelly, 1976; Bloom, Costa & Salmoiraghi, 1965; Connor, 1970; Gonzalez-Vegas, 1974; Herz & Zieglgansberger, 1968; Siggins, 1978) or exitatory actions (Norcross & Sphelmann, 1977 & 1978; Richardson, Miller & McLennan, 1977; Spehlmann & Norcross, 1978). Studies with intracellular recording techniques, however, demonstrate that nigral stimulation produces excitatory postsynaptic potentials (EPSPs) in neostriatal neurons and that iontophoretically applied dopamine has a depolarizing effect on the neostriatal neurons (Bernardi, Marciani, Morocutti & Giacomini, 1978; Herring & Hull, 1980; Kitai, Sugimori & Kocsis, 1976; Kitai & Kocsis, 1978; Kocsis & Kitai, 1977). The nature of a possible non-catecholaminergic nigrostriatal projection (Fibiger et al., 1972; Feltz & deChamplain, 1972; Hattori et al, 1973) remains unclear. Extracellular recording investigations have reported that dorsal raphe stimulation has an inhibitory effect on neostriatal neurons (Miller et al., 1975; Olpe & Koella, 1976). However, iont0phoretically applied 5-HT is reported to have either an excitatory (Bevan, Bradshaw & Szabadi, 1975) or inhibitory effect (Herz & Zieglganzberger, 1968) on neostriatal neurons. Recently, studies employing intracellular recording techniques have shown that stimulation of dorsal raphe results in mainly excitation of neostriatal neurons (VanderMaelen et al., 1980) and that this excitation is positively correlated to the levels of available 5-HT (Park et al., 1980). 16 D. Neostriatal Efferent Connections The major neostriatal efferent targets are both segments of the globus pallidus and the substantia nigra. a) The Pallidum The striato-pallidal fibers are t0pographically organized in the mediolateral and dorsoventral directions (Cowan and Powell, 1966; Nauta & Mehler, 1966; Szabo, 1967, 1970; see review in Grofbvé, 1979). The striatal efferent terminals contain large vesicles and form symmetrical axodendritic synapses with their target cells (Kemp, 1970; Fox et al., 1974a). The striato-pallidal axons are thinly myelinated such that groups of these axons converge radially toward the globus pallidus with the "spokes of wheel" appearance in myelin-stained sections, and are also called "Wilson’s pencils" (Wilson, 1914; Fox & Rafols, 1976). The striatal efferent fibers, after traversing both segments of the globus pallidus, pierce the cerebral peduncle forming the "comb system" (Edinger, 1911, cited in Fox & Rafols, 1976) to reach the substantia nigra. Most of the myelinated striatal axons become unmyelinated before reaching the substantia nigra (Fox et al., 1975). Electrophysiological studies (Yoshida, Rabin & Anderson, 1972) and Golgi studies (Fox et al., 1975) have suggested that the striato-pallidal and striato-nigral pathways are collateral fiber system. Recent studies with intracellular injection of HRP in the rat (Preston et al., 1980) have presented evidence supporting this view: HRP labeled axons are observed to give rise to collaterals in the globus pallidus before fading from view in the internal capsule, 17 presumably coursing toward either the entopeduncular nucleus or the substantia nigra or both. b) The Substantia Nigra The striato-nigral fibers are also topographically organized in the mediolateral, rostrocaudal and reversed dorsoventral directions (see review in Grofova, 1979), with most of the fibers terminating in the pars reticulata of the substantia nigra (SNpr) (Kemp, 1970; Grofbvé & Rinvik, 1970; Schwyn & Fox, 1974; Bunny & Aghajanian, 1976). Since neurons in the SNpc have dendrites that extend into the SNpr (Schwyn & Fox, 1974; Juraska et al., 1976), this pattern of strio-nigral projection is significant in that it may indicate the existence of a direct feedback loop between the neostriatal projection cells and the dopaminergic neurons of the SNpc, without an intervening interneuron. The striato-nigral terminals, like the striato-pallidal terminals, also contain large vesicles and make mainly symmetrical axodendritic synapses (Grofbva & Rinvik, 1970; Kemp, 1970; Hajdu, Hassler & Bak, 1973). c) Physiology of the Neostriatal Efferents Studies do not fully agree on the nature of the synatic influence of neostriatal projection cells on their target cells in the globus pallidus (GP) and the substantia nigra (SN). Mainly inhibitory but some excitatory responses are observed in GP neurons following stimulation of the neostriatum (Malliani & Purpura, 1967; Yoshida, Rabin & Anderson, 1972; Levine, Hull & Buchwald, 1974). Similarly, mainly inhibitory and some excitatory responses were observed in the 18 SN neurons following stimulation of the neostriatum (Frigyesi & Purpura, 1967; Yoshida & Precht, 1971; Precht & Yoshida, 1971; McNair et al, 1972; Frigyesi & Purpura, 1967; Preston et al., 1981). The neostriatal efferent fibers appear to convey both enkephalin and gamma-aminobutyric acid (GABA) to the pallidum (Obata & Yoshida, 1973; Cuello & Paxinos, 1978; Brann & Emson, 1980; Emson, Arregui, Clement-Jones, Sandberg & Rossor, 1980; Fonnum & Walaas, 1979) whereas those fibers terminating in SN contain both GABA and substance P (Precht & Yoshida, 1971; Brownstein, Mroz, Tappaz & Leeman, 1977; Gale, Bird, Spokes, Iversen & Jessel, 1978; Hong, Yang, Racagni & Costa, 1977; Jessell, Emson, Paxinos & Cuello, 1978; Hattori et al., 1973a; Kanazawa, Bird, O’Connell & Powell, 1977; Emson et al., 1980; Pickle, Joh, Reis, Leeman & Miller, 1979; Ribak, Vaughn & Roberts, 1979, 1980; Ljungdahl, Hokfelt, Goldstein & Park, 1975;). At least some of the striopallidal and strionigral axons are thought to be collaterals of the same neostriatal neuron (Yoshida et al., 1972; Fox & Rafols, 1976; Szabo, 1981), however it is unclear whether the same individual neostriatal efferent axon contain both GABA and enkephalin and/or substance P. 19 E. Neostriatal Intrinsic Organization a) Light Microscopic Studies of Nissl Preparations Light microscopic analysis of Nissl stained sections have recognized that the neostriatal neurons are composed of numerous medium sized neurons (somatic diameters range from 10 to 20 pm) and a few large (somatic diameters range from 20 to 50 um) and small (somatic diameters generally less than 10 um) neurons distributed in a seemingly random and homogeneous pattern. Most of the medium cells have large, pale nuclei surrounded by a narrow rim of cytOplasm with little or no Nissl substance (i.e. they are achromatic or hypochromatic). In contrast, the large neurons and a very few medium neurons have a lot of’cytoplasm and large number of Nissl bodies (Ram6n y Cajal, 1911; Kemp & Powell, 1971a; also see reviews by Fox & Rafols, 1976; Pasik, Pasik & DiFiglia, 1979). Although clustering of somata of neostriatal neurons have been reported (Mensah, 1977), the significance of this observation remains unclear since neither the dendritic nor the axonal morphology of the neurons in a cluster are revealed in Nissl preparations. b) Golgi Studies Studies of neostriatal neurons using the Golgi techniques have described a number of neuronal types in several species. However, the description and the classification of these neurons have remained controversial despite attempts at classifying them according to a unifying system (Pasik et al., 1979). 20 Ramén y Cajal (1911) placed considerable emphasis on the distribution of the axons of neostriatal neurons in his Golgi study of rabbit and human infant and divided the neostriatal neurons into long- and short-axon cells. The short-axon cells included large, small and dwarf cells whereas the long-axon cells included medium and large neurons. With the exception of the dwarf cell, most of these neurons have various amount of dendritic spines. Since no magnification scales has been provided for the drawings of Ramén y Cajal, it has been difficult to evaluate the sizes of these neurons. For instance, his dwarf or neurogliform neuron (his Figure 325D) has somatic and dendritic size comparable to those of the "small" short axon neurons (his figure 325A), which are considered to be equivalent to the medium spiny neurons by many modern investigators (Kemp & Powell, 1971a; Pasik et al., 1976; 1979). Considerable controversies have thus been generated with respect to the identification of similar neurogliform cells in later Gogli studies in other species (Fox & Rafols, 1971/1972; Fox, Lu Qui & Rafols, 1974; DiFiglia, Pasik & Pasik, 1976). Recent investigations of the neostriatum by the Golgi techniques began with the study of Leontovich (1954). Her extensive study, based on materials from hedgehogs, rabbits, mice, dog, monkey and human demonstrated the presence of numerous long-axon medium spiny neurons. In addition, both long- and short—axon large neurons as well as short-axon medium neurons with sparsely-spined dendrites have been described. However, since the illustrations in her report were of very low magnification, it has been difficult for the present author to discern their detailed morphology. Although both intrinsic 21 and projection neurons are shown in her semi-diagramatic summary illustration (her Figure 7), the lack of high magnification camera-lucida or photomicrographic documentation of the efferent axons arising from the presumed projection neurons may have contributed to the skepticism held by some investigators about her conclusion (Kemp & Powell, 1971a). Other modern Golgi investigations include the extensive and influential studies on the cat caudate nucleus by Kemp and Powell (1971a) and the monkey neostriatum by Fox and his colleagues (Fox et al., 1971/19728, b). Both of these groups described the presence of large number of short-axon medium spiny neurons which have extensive local collateral arborizations. Additionally, these cells are, distinguished by their aspinous somata and proximal dendrites and densely spined distal dendrites. The long-axon neurons were proposed to be the large neurons with either aspinous or sparsely-spined long dendrites and a myelinated axon (Fox et al., 1971/1972b) and the medium neurons with sparsely-spined dendrites (Kemp & Powell, 1971a). In addition to the medium spiny neurons, other short-axon neurons included sparsely—spined medium neurons (i.e. medium smooth cell of Kemp & Powell, and spidery neuron of Fox et al.). These authors thus prOposed that the neostriatal neurons consisted of large numbers of medium sized intrinsic neurons which received direct afferent inputs, and a small number of large efferent neurons responsible for most or all of neostriatal outputs. This theory was challenged by the Pasiks and DiFiglia in their Golgi studies of the monkey neostriatum (DiFiglia et al., 1976; Pasik et al., 1976; Pasik et al., 1977; Pasik et al., 1979). Drawing 22 support from the study of Leontovich (1954) and from new retrograde HRP transport studies (Grofové, 1975; Bunny & Aghajanian, 1976) which indicated that a majority of neostriatal neurons are medium efferent neurons, the Pasiks and DiFiglia concluded that the most frequently impregnated medium spiny neurons were projections neurons (their Spiny I cells). Their Spiny II cells, consisted of both large and medium neurons with somatic spines and sparsely-spined dendrites, were also considered to be long-axon cells. These authors further described the morphology of the short-axon neurons as consisting of large aspinous neurons with varicose dendrites (Aspiny II cell), and medium aspinous neurons with either varicose (Aspiny I cells) or straight dendrites (Aspiny III cells). Additionally, a neurogliform neuron without identifiable axon was described. A classification scheme similar to those of the Pasiks and DiFiglia has been used for the dog neostriatal neurons (Tanaka, 1980). However, both the large and medium Spiny II cells in the dog have no somatic spines. The Aspiny III cells in the dog have very long dendrites whereas those in the monkey have shorter dendrites. Recent Golgi studies in the rat have also described a number of neostriatal neuronal types (Mensah & Deadwyler, 1974; Lu 8 Brown, 1977; Danner & Pfister, 1979a, b). However, there are many discrepancies with respect to the morphological descriptions and the classification of these neurons: from 5 to 7 types of neurons have been illustrated in either free-hand drawings or partial photomicrographs. All of these studies described the presence of numerous medium spiny neurons which they considered to be a type of intrinsic neurons based on their dense local axon arborizations. 23 Several other types of neurons included the so—called "medium long-axon" cells which have either few (Mensah & Deadwyler, 1974; Danner & Pfister, 1979a, b) or many dendritic spines (Lu & Brown, 1977). Medium neurons with either smooth dendrites or varicose dendrites were considered to be short-axon neurons whereas large aspinous neurons were considered to be long-axon neurons despite the fact that long axons have never been illustrated. A recent study has provided camera-lucida tracings of 5 types of rat neostriatal neurons (Dimova et al., 1980). The medium spiny neurons are classified as the type I neurons. Their type II neurons are medium neurons with sparsely-spined dendrites, including some spines on the primary dendrites. Type III cells are medium neurons with varicose dendrites and type IV cells are medium neurons with sparsely-spined long dendrites. Type V neurons are giant neurons with varicose and sparsely-spined dendrites. However, neither photomicrographs of the neurons nor descriptions of their axonal distribution patterns have been provided in this recent study. Thus, the morphological characteristics of the rat neostriatal neurons as well as their classification relationships to those described in other species remain unclear. Although no definitive evidence have been obtained in the monkey, studies in the rat have produced unequivocal proof that the medium spiny neuron (i.e. the Spiny I of the monkey) projects to the substantia nigra (Somogyi & Smith, 1979) and the globus pallidus (Preston et al., 1980). However, comparable evidence is lacking for other types of neurons which may be projection neurons in the rat or other species. 24 In summary, most of previous Golgi studies have attempted to classify the neostriatal neurons into projection neurons and intrinsic neurons and then subdivide these categories into several types according to their somatic and dendritic morphology. However, since in Golgi material, the axons are usually only partially impregnated, the identifications of projection neurons (i.e. long-axon cells) have not been accompanied by micrographic evidence. Consequently, naming certain cells as long-axon neurons have been educated guesses at best. Confusion in comparing the results of these various Golgi studies also arises from the fact that either very restrictive or very broad naming systems have been used without clear and consistent definitions. The descriptive terms such as "large", "medium", "spiny", "aspiny", "varicose", "smooth", "stellate", "spidery", "slender", "spindle" and so on have been used loosely and in different combinations by different authors. Thus, considerable controversy has arissen from attempts to correlate the results of these studies. For example, in a recent review (Pasik et al., 1979), the Spiny II neurons of the monkey (DiFiglia et al., 1976) have been correlated to the medium and large long-axon cells of the cat (Kemp & Powell, 1971a), the medium spindle and large cells of the rat (Lu & Brown, 1977), the large long-axon cell of human (Ram6n y Cajal, 1911), the large aspiny long-axon cell of the monkey (Fox et al., 1971/1972b) and so on. The critical similarities among these various neurons are their sparsely-spined dendrites and their putatively long axons. However, the presence of long axons has not been convincingly demonstrated fer any of these neurons, and none of these cells have somatic spines, 8 critical characteristic of the Spiny II neurons in 25 the original study of DiFiglia et al., (1976). c) Electron Microscopic Studies of Normal Material Electron microsc0pic studies of the neostriatum have revealed several types of neurons in the rat (Mori, 1966; Hattori et al., 1973a; Bak et al., 1975), the cat (Adinolfi & Pappas, 1968; Adinolfi, 1971; Kemp & Powell, 1971a) and the monkey (Fox et al., 1971/1972a, b; Pasik, Pasik & DiFiglia, 1976). These studies generally agree that the most frequently encountered somatic profiles are medium sized (or small, as called by Mori, 1966) neurons with large and pale, unindented nuclei with sparse cytoplasmic organelles. These somata have been correlated with the achromatic medium (or small) neurons observed in Nissl preparations (Kemp & Powell, 1971a, Fox, Andrade, Hillman & Schwyn, 1971/19728). The large neostriatal neurons are less frequently found, and usually have indented nuclei surrounded by a large amount of well organized rough_endopalsmic reticulum, Golgi complexe, and numerous dense bodies. Various rarely-found small to medium neurons have also been described in several studies as to have indented nuclei surrounded by varying amount of cytoplasmic organelles. Since only limited portions of any individual neuron could be analyzed in these studies, and distal neuronal processes usually could not be traced back to the parent somata, correlation of these ultrastructural observations with the light microsc0pic findings (i.e. Golgi studies) has remained tentative until recently (see below). Analysis of the neostriatal neurOpil has also resulted in controversial classifications of various types of axon terminals and 26 synaptic connections based on sets of different criteria (Mori, 1966; Adinolfi & Pappas, 1968; Adinolfi, 1971; Kemp & Powell, 1971b; Rafols & Fox, 1971/1972; Kawana, Akert & Bruppacher, 1971; Bak et al., 1975; Pasik et al., 1976; Chung et al., 1977; Hassler et al., 1978). In general, the most prevalent synaptic profiles are axon terminals containing small round vesicles making asymmetrical axospinous synapses. These are considered to have arised mainly from the cerebral cortex, the thalamus and the midbrain areas (Kemp & Powell, 1971b, c; Chung et al., 1977; Hassler et al., 1978; Bak et al., 1975). Other axon terminals contain either large or small pleomorphic vesicles and make mainly symmetrical synapses with somata and dendritic shafts, and are considered to be terminals of intrinsic collaterals (Hassler et al., 1977; Tennyson & Marco, 1973). d) Golgi Gold-toned Neostriatal Neurons Recently, direct electron microscopic analysis of cytoplasmic details of Golgi impregnated neurons has become possible through gold-toning procedures (Fairén et al., 1977). With this technique, the ultrastructural morphology of two types of monkey neostriatal neurons, the Spiny I and the Aspiny I cells, have been analyzed (DiFiglia et al., 1980). The somatic and dendritic morphology of several types of rat neostriatal neurons has also been studied (Dimova et al., 1980). In general, the Spiny I cells of the monkey and the medium type I cells of the rat are characterized by round, unindented nuclei whereas the other cells have indented nuclei (Dimova et al., 1980). The Golgi gold-toning technique has also been combined with extracellular uptake of HRP to demonstrate that the medium spiny 27 neurons project to the substantia nigra in the rat (Somogyi & Smith, 1979) and that these cells receive direct cerebral cortical afferents which have been labeled by anterograde degeneration (Somogyi, Bolam & Smith, 1981). e) Labeling with Extracellularly Injected Markers The properties of axon terminal uptake of extracellularly injected markers such as HRP molecules and retrograde transport of these markers back to the soma have been widely used in recent investigations of connections in the nervous system. HRP injected extracellularly into the substantia nigra has resulted in the labeling of a large number of medium sized neostriatal neurons (Grofova, 1975; Bunny & Aghajanian, 1976). Recent studies combining the technique of retrograde transport of HRP with acetylcholinesterase histochemistry have shown that the neostriatal efferent neurons appear to be grouped in geometrically complex compartments which partially overlap the cholinesterase-rich areas (Graybiel a Ragsdale, 1979; Graybiel, Ragsdale a Moon-Edley, 1979). It is unclear how these patchy arrangements of the efferent neurons are related to the patchy distribution of the neostriatal afferent terminal fields formed by the cerebral cortex and the thalamus observed in autoradiographic studies (Kfinzle, 1975, 1977, 1978; Kalil, 1978). Identification of large neostriatal projection neurons have been described in studies in which the HRP has been injected into the midbrain (Grofova, 1979) or the auditory cortex (Jayaraman, 1980). Herpes virus injected into the midbrain has labeled both large and medium neurons (Bak et al., 1977; 1978). However, since only partial 28 morphology of the somata and proximal dendrites is revealed by these retrograde labelings techniques, the morphological identification of these neostriatal projection neurons has remained tentative. Recently, combining the technique of retrograde transport of HRP with Golgi impregnation and Golgi gold-toning procedures, some of the rat neostriatal projection neurons have been demonstrated to be medium spiny neurons (Somogyi & Smith, 1979). f) Labeling with Intracllularly Injected Markers The recently deve10ped technique of intracellular labeling with either fluorescent or enzyme (i.e. HRP) markers have enabled direct analysis of the morphology of neurons which have been physiologically characterized through intracellular recording techniques (Kitai, Kocsis, Preston & Sugimori, 1976). Studies in the cat (Kitai et al., 1976; Kitai, KOcsis & Preston, 1978; Kocsis & Kitai, 1977; Kocsis, Sugimori & Kitai, 1977) and the rat (Preston, BishOp & Kitai, 1980; VanderMaelen & Kitai, 1980) have shown that these cells receive convergent excitatory inputs from the cerebral cortex, the thalamus and the substantia-nigra. Additionally, they receive excitatory inputs from the dorsal raphe nucleus (VanderMaelen, Bonduki & Kitai, 1979). Light microscopic analysis has furtherly confirmed and extended the Golgi studies in revealing the full extent of the elaborate intrinsic axonal collateral arborizations of the medium spiny neurons (Preston et al., 1980; Wilson & Groves, 1980). The HRP labeled main axons have been traced into the globus pallidus and the internal capsule and a few collaterals have been observed to terminate 29 in the globus pallidus (Preston et al., 1980). It is interesting to note that only medium spiny neurons have been labeled intracellularly in either the cat or the rat, suggesting that the predominant neostriatal neuron type is the medium spiny neuron. Electron microsGOpic analysis has revealed that the intrinsic collaterals of the HRP labeled medium spiny neurons synpase with the spine necks, dendritic shafts, somata and axonal initial segments of neighboring neurons (Wilson & Groves, 1980). g) Biochemical, Histochemical & Immunocytochemical Studies Biochemical studies have shown that the putative neurotransmitters in the neostriatum include glutamate, gammapaminobytric acid (GABA), acetylcholine, depamine, serotonin (5-HT), substance P and enkephalin. Of these, afferent transmitters are considered to be glutamate from the cerebral cortex, dopamine from the substantia nigra, and 5-HT from the dorsal raphe nucleus. The transmitter of the thalamostriatal pathway remains unclear. Acetylcholine and GABA have been considered to be the transmitters of neostriatal intrinsic neurons. At the same time, both GABA and substance P are involved in the striato-nigral projections (Fonnum, Grofova, Rinvik & Storm-Mathisen, 1974; Brownstein, Mroz, Tappaz & Leeman, 1977; 6818, Hong & Guidotti, 1977; Hong, Yang, Racagni & Costa, 1977; Jessel, Emson, Paxinos & Cuello, 1978; Kanazawa, Bird, o’Gonnel & Powell, 1977; Fonnum, Gottsfeld & Grofova, 1978) whereas both GABA and enkephalin are involved in the striato-pallidal pathway (Fonnum et al., 1978; Cuello & Paxinos,1978; Brann & Emson, 1980; Ribak et al., 1980; also see review by Fonnum & Walaas, 1979). 30 Histochemical studies indicate that the large neostriatal neurons have the highest acetylcholinesterase (AChE) synthesizing activity (Butcher & Bilezikjian, 1975). Combined AChE histochemistry with HRP retrograde transport studies indicated that neostriatal projection cells are mostly AChE-negative medium neurons (Henderson, 1981), however AChE-positive medium neurons have been found to project to the substantia nigra by another laboratory (Kaiya et al., 1979). Since AChE is considered to be a necessary but not a sufficient marker for cholinergic neurons, it is unclear how these cells are related to the true cholinergic neurons which would contain the acetylcholine synthesizing enzyme-choline acetyltransferase (CAT). A recent study reported that AChE is involved in hydrolyzing substance P molecules in the central nervous system (Chubb, Hodgson & White, 1980). Since the neostriatum has a very high content of substance P, the AChE neurons are probably involved in interactions with both the cholinergic interneurons and the substance P projection neurons. Immunocytochemical studies with anti-choline acetyltransferase (anti-CAT), however, resulted in labeling of either medium (Hattori, Singh, McGeer & McGeer, 1976) or large neurons (Kimura, McGeer, Peng & McGeer, 1980). One study at the electron microsc0pic level shows that the CAT-containing medium neurons have unindented nuclei and have dendritic spines (Hattori et al., 1976). These discrepancies arising from the same laboratory have contributed much to the unabated controversies with respect to the identity of cholinergic neostriatal neurons. The GABA-synthesizing enzyme, glutamic acid decarboxylase (GAD), has been used as a marker for GABAergic neurons. Studies with 31 anti-GAD antibodies have indicated that the GABAergic neostriatal neurons are medium neurons and have dendritic spines (Ribak et al., 1979). Similar findings have been obtained from studies with anti-enkephalin antibodies (DiFiglia, Aronin, Liotta & Martin, 1980; Pickel, Sumal, Beckley, Miller & Reis, 1980). The results from these immunocytochemical studies, therefore, indicate that medium neurons with dendritic spines may contain three or four different types of putative neurotransmitters: acetylcholine, GABA, enkephalin and possibly substance P. Whether these transmitters are contained in the same neuron or in biochemically distinct neurons which are morphologically similar remain unclear. This difficulty is furtherly compounded by the fact that virtually all of these studies have been conducted on the rat neostriatum, in which the morphological descriptions of different types of neurons have many discrepancies, as has been mentioned above. 32 F. Summary of Background The neostriatum receives tapographical projections from the cerebral cortex, intralaminar thalamic nuclei, substantia nigra and the dorsal raphe nucleus. The neostriatum projects tapographically to the paleostriatum and the substantia nigra. The putative neurotransmitters of the afferents are glutamate, dapamine and serotonin (5-HT) arising from the cerebral cortex, substantia nigra and dorsal raphe nucleus respectively. The transmitter involved in thalamostriatal projection remains unclear. Electron microsGOpic studies indicate that the majority of the afferent terminals contain small to medium sized round vesicles and make asymmetrical axospinous synapses in the neostriatum. The neostriatal efferent terminals contain large synaptic vesicles and make symmetrical axodendritic and axosomatic synapses with targets cells in the globus pallidus and substantia nigra. The transmitters of neostriatal afferents are considered to consist of gammaeaminobutyric acid (GABA) and enkephalin in the striatOpallidal fibers, and substance P and GABA in the striatonigral fibers. It is not clear whether individual neostriatal efferent neurons contain more than one putative transmitter. Acetylcholine is considered to be the transmitters of the neostriatal interneurons. Immunocytochemical studies in the rat indicate that glutamate decarboxylase (GAD) containing neurons are medium neurons with dendritic spines, and are similar to the enkephalin containing neurons as well as substance P containing cells. Choline acetyltransferase containing neurons have been reported to be either large or medium 33 neurons, with the large neurons having the highest acetylcholinesterase synthetic activity. Evidence derived from retrograde transport of extracellularly injected markers indicates that the overwhelming majority of neostriatal efferent neurons are medium neurons, and rarely, a few large neurons. However, more than six types of neostriatal neurons have been described in Golgi and electron microsc0pic studies in several species. Of these cells, only the medium spiny neuron has been demonstrated conclusively to be a projection neuron. This neuron, in addition, has intrinsic integration functions through its elablorate intrinsic axonal collateral arborization. The medium spiny neuron receives convergent excitatory afferent inputs and exert recurrent inhibitory actions which are mediated by GABA. The morphological characteristics and the classification of other neurons, both large and medium, remain controversial. Some of these neurons have sparsely-spined dendrites, others have varicose dendrites. Although myelinated axons have been observed to arise from some of these cells, no photomicrographs or camera lucida tracings are available demonstrating that these axons have been traced out of the neostriatum. Since the main axons frequently are not impregnated in Golgi preparations, dividing these neurons into Golgi type I (i.e. efferent cells) or Golgi type II (i.e. intrinsic interneurons) remains controversial. 34 G. Research Objectives Since the rat has been the most widely used subject in recent experimental investigations of neostriatal functions, elucidation of the morphology of various types of rat neostriatal neurons at both light and electron microsc0pic levels should serve to clarify many of the controversies with respect to the intrinsic anatomical organization of the neostriatum, and should enable better understanding of the structural basis of various neostriatal functions. The specific aims of this study are: 1) To determine the detailed light microsc0pic morphological characteristics of various types of rat neostriatal neurons by the classical Golgi technique. 2) To directly analyze the ultrastructural morphology of Golgi impregnated neurons by the Golgi gold-toning procedure. 3) To investigate the light and electron microsc0pic morphology of the neostriatal projection neurons which have been labeled following the extracellular injection of HRP into the globus pallidus, and to compare with that from the Golgi preparations. 4) To investigate the light and electron microscOpic morphology of physiologically characterized neostriatal neurons which have been labeled intracellulary with HRP, and to compare with that from the Golgi preparations. Chapter 2. METHODS A. Golgi Studies Male adult (250-400 gm) Long-Evans hooded rats and Sprague-Dawley albino rats were used in this study. After anesthetizing the animals with i.p. injections of either Nembutal (50 mg/kg body weight) or Urethane (2 gm/kg), they were perfused through the aorta with 300 ml of 0.9% saline solution followed by 1 liter of fixtive containing 2% glutaraldehyde and 2% fermaldehyde (prepared freshly from paraformaldehyde powder) in a 0.1 M phosphate buffer (pHi7.3). The temperature of all perfusing solutions was maintained at 4 degrees Celsius. After perfusion, the brains were removed and kept overnight in a refregerator in the fixative solution. Brains were then cut into 3 to 4 mm thick blocks in either coronal or sagittal planes, rinsed briefly with 0.1 M phosphate buffer (pHd7.3, 353 mOsm.) and immersed in the first solution of the rapid Golgi technique (Valverde, 1970) consisting of 2.4% potassium dichromate and 0.2% osmium tetroxide (OsO4). Vials containing at least 20 ml of this solution per brain block were stored at room temperature and away from direct light for 3 to 7 days. The brain blocks were then briefly rinsed with 0.75% silver nitrate solution and immersed in fresh 0.75% silver nitrate solution, 20 ml per block, for 2 to 7 days at room temperature and away from direct light. The vials were occassionally agitated to promote even exposure of tissue surfaces to the solutions during both of these periods. 35 36 At the completion of the rapid Golgi reaction, the blocks were either sectioned immediatedly or temporarily stored in 100% glycerine in the refregerator. The longest storage was four months. The blocks were cut with an Oxfbrd Vibratome into 100 um thick sections. The sections were either dehydrated through ethanol and cleared in xylene for light microsc0pic analysis only, or cleared in graded series of glycerine solutions (20%-100%) for both light microscOpic analysis and possible further gold-toning treatments preparatory to electron microscopic analysis. The Golgi impregnated neurons were examined and photographed with a Leitz Orthoplan microsGOpe and detailed tracings of neurons were made with the aid of a Leitz drawing tube. Golgi gold-toning procedures: After light microscOpic analysis of desired neurons in glycerine cleared sections, the sections were rehydrated through a descending series of glycerine soultions (100%—20%) to distilled water. Sections were then immersed in a 0.05% solution of yellow gold chloride (hydrogen tetrachloroaurate (III), HAuCl4-4H20) for 15 minutes in an ice bath, with frequent agitation. The sections were rinsed briefly three times with ice cold distilled water before being reacted with 0.05% oxalic acid for 2 minutes in the ice bath to reduce the gold chloride to metallic gold. The sections were then rinsed again with distilled water and returned gradually to room temperature. The de-impregnation of Golgi reaction products was accomplished by reacting with freshly prepared 1% sodium thiosulfate solution for 30 to 60 minutes at room temperature with frequent agitation. 37 Upon completion of the above reactions, the tissue sections were rinsed in distilled water and postfixed in phosphate buffered 2% 0804 (pHd7.3) fer 1 hour. The sections were then dehydrated through a graded series of ethanol and propylene oxide and flat embedded in either Epon-Araldite or Spurr’s plastic on Teflon-coated glass slides and coverslips (Wilson & Groves, 1979). Light microscopic examination of polymerized plastic sections was conducted to select the optimally gold-toned neurons for electron microscopic analysis. These neurons were photographed again and their surrounding landmarks traced with a Leitz drawing tube to facilitate later search for the gold-toned profiles under the electron microscope. The Teflon-coated coverslips were removed and squares (0.5 mm per side) containing the desired neurons were cut off the slide and re—embedded on blank plastic blocks. Serial thin sections were cut with a Sorvall MT-2 Ultramicrotome. Blocks were often reexamined with a light microscope fitted with a metalurgical objective to check the progress of the thin sectioning. The thin sections were picked up on Formvar coated slot grids, stained with methanolic uranyl acetate and lead citrate, and examined with a Philips 201 transmission electron microsc0pe. B. Extracellular Horseradish Peroxidase Labeling Adult male Long-Evans hooded rats (250-400 gm) were anesthetized with Nembutal (50 mg/kg body weight) admininstered intraperitoneally and placed in a stereotaxic head holder (David Kopf Instruments). The skull was Opened and 0.1 to 0.2 microliters (ml) of a solution of 20 to 40 % horseradish peroxidase (HRP) in water was 38 pressure injected into the globus pallidus (GP) using a glass micropipette (tip diameter 50 um) fitted on a Hamilton microsyringe. The micropipette was intoduced into the rat brain at an angle to avoid leakage of HRP solution into the neostriatum. The animals were sacrificed after two days by an overdose of Nembutal (60 mg/kg body weight) followed by perfusion through the aorta with 300 ml of 0.9% saline and 1 liter of fixative consisting of 2% glutaraldehyde and 2% formaldehyde (prepared freshly from paraformaldehyde) in a 0.1 M phosphate buffer (pHs7.3). Following the perfusion, the brains were removed and stored in the fixative overnight. Sagittal 100 or 50 um thick sections were cut on an OxfOrd Vibratome and reacted for demonstration of HRP activity (Graham & Karnovsky, 1966): Sections were immersed in 0.5% cobalt chloride in 0.1 Tris buffer (pHi7.3) for 10 minutes (Adams, 1977), rinsed briefly with the Tris buffer and placed in 0.1 M phosphate buffer (pH=7.3) to react with 0.05% 3,3-diaminobenzidene tetrahydrochloride (DAB) and 0.06% hydrogen peroxide for 30 minutes. The sections were then cleared with either glycerine for immediate light microsc0pic analysis to be followed by postfixation with phosphate buffered 2% 0804 and plastic embedding (Christensen & Ebner, 1978), or were scanned directly in the phosphate buffer under a light microscope to locate the HRP labeled profiles and followed by postfixation and flat-emmbedding in Spurr’s plastic on Teflon-coated glass slides and coverslips (Wilson & Groves, 1979). Light and electron microscopic analysis of HRP labeled neurons were identical to those of Golgi gold-toned neurons mentioned earlier. 39 C. Intracellular Horseradish Peroxidase Labeling Adult male Long-Evans hooded rats (250-400 gm) were anesthetized with Urethane (1.5 gm/kg body weight) administered intraperitoneally and placed in a stereotaxic head holder (David Kopf Instruments). Following dorsal craniotomy, bipolar or monopolar stimulating electrodes (insulated stainless steel insect pins, size "00", bared at the tips) were stereotaxically positioned in the substantia nigra, thalamus or cortex for orthodromic and antidromic (from substantia nigra only) activation of neostriatal cells. Glass micropipettes pulled to a fine tip on a Narishige electrode puller and beveled to tip diameters of 0.5 to 1.0 um under microscopic observation were used for recording intracellular potentials and injecting horseradish peroxidase (HRP) into the neurons being recorded. The electrolyte solution within the glass electrode consisted of 4% HRP in 0.05 M Tris buffer (pH=7.6) and 0.5 M potassium chloride, potassium acetate or potassium methylsulfate. After placement of the microelectrode below the cortical surface, the exposed brain surface was covered with soft paraffin wax to reduce brain pulsation. Recordings were made with a conventional active bridge amplifier. Neurons recorded intracellularly were labeled with HRP by passing positiive DC current pulses (3-15 nA, 100 to 300 msec in durations, 3 to 5 Hz) for 5 t0 5 minutes through the recording microelectrode. Extracellular field responses were recorded to compare data obtained intracellularly. Following intracellular HRP injections, the rats were perfused through the aorta with 0.9% saline followed by a fixative consisting of 2% formaldehyde and 2% glutaraldehyde in 0.1 M phosphate 4O buffer (pH=7.3). The brains were removed and stored in the fixative overnight. Sagittal 100 or 50 um thick sections were cut on an Oxford Vibratome and reacted for demonstration of HRP activity (Graham & Karnovsky, 1966): Sections were immersed in 0.5% cobalt chloride in 0.1 Tris buffer (pH=7.3) for 10 minutes (Adams, 1977). rinsed briefly with the Tris buffer and placed in 0.1 M phosphate buffer (pHfi7.3) to react with 0.05% 3,3-diamin0benzidene tetrahydrochloride (DAB) and 0.06% hydrogen peroxide for 30 minutes. ‘The sections were then cleared with either glycerine for immediate light microscopic analysis, followed by postfixation with phosphate buffered 2% Os04 and plastic embedding (Christensen & Ebner, 1978), or were scanned directly in the phosphate buffer under a light micr0800pe to locate the HRP labeled profiles and followed by postfixation and flat-emmbedding in Spurr’s plastic on Teflon-coated glass slides and coverslips (Wilson & Groves, 1979). Light and electron micr0800pic analysis of HRP labeled neurons were identical to those of Golgi gold-toned neurons mentioned earlier. 41 D. Method of Analysis Light Microsc0py: The somatic cross-sectional areas were measured from the camera lucida (drawing tube) tracings with the aid of a graphic analyzer attached to a PDP-11 computer. The presence or absence of somatic spines and dendritic varicosities as well as the distribution of dendritic spines were noted. The dendritic branching characteristics as determined by the number of dendritic stems versus the number of dendritic tips were compared for neurons with similar somatic and dendritic surface morphology. Where possible, the distribution of the axons was also analyzed to determined the extent of its collateral arborizations. Electron Micr0800py: The somata of individual neurons previously characterized by light microsc0pic analysis were examined to compare the morphology of their nucleus and the distribution of various organelles. The somatic and dendritic synapses were analyzed to determined the symmetry of the synaptic contacts and the size and shape of the synaptic vesicles. The axons were also examined to determine the presence of synapses on the axon hillock or the initial segment, and-whether the axons were myelinated. Where possible, the axon tenninals were examined to determine the size and shape of the synaptic vesicles, the symmetry of the synaptic contacts, and the nature of the post-synaptic elements. Chapter 3. RESULTS A. Light MicrosCOpic Analysis of Golgi Material As have been commonly practiced in previous studies, neurons in the rat neostriatum are preliminarily divided into three groups on the basis of their somatic sizes. These groups are: a) Large cells, with somatic maximal cross-sectional areas (abbreviated as SA) larger than 300 square um, approximately equivalent to somatic diameters larger than 20 nm. b) Medium cells, SA between 100 and 300 square um, equivalent to diameters between 10 and 201nm. 0) Small cells, SA less than 100 square um, 0r diameters less than 10 um. After this initial arbitrary division, additional morphological characteristics such as somatic and dendritic surface appearance (presence or absence of spines, varicose vs. straight dendrites), dendritic length, dendritic branching patterns and axonal distribution patterns are analyzed. LARGE NEURONS : Large neurons in the rat neostriatum may be divided into at least two types, those with and those without somatic spines. Type I Large Neurons: This type of large neuron (n = 12) has a smooth soma and 3 to 5 thick primary dendrites (i.e. the number of dendritic stem, 3 = 3 t0 5) some of which branch near the soma into long secondary dendrites (Fig. 1A, 1B & 2A) while others extend over long distances before branching (Fig. 1C). The number of dendritic tips (F) ranges from 11 to 35. Cells within this group have either polygonal (Fig. 1A, 1B) or fusiform shaped somata (Fig. 1C). The dendrites often arise from 42 Figure 1. A. 43 Photomicrographs of type I large neurons. Type I large neuron with polygonal soma. Somatic cross-sectional area (SA) = 344 square um. Calibration scale (CAL) - 50 um. Higher magnification of A, small arrow points to a distal dendrite which has varicose appearance; blocked arrow points to the axonal branching point. CAL = 20 um. Type I large neuron with fusiform soma and radiating non-varicose, long dendrites; SA = 595 square um. Same magnification as in A. 44 Figure 1 Figure 2. A. 45 MicrOSCOpic tracings of type I large neuron. CAL = 50 um. Somato-dendritic morphology of a neuron shown also in Figure 1A & 18. Note some of the sparsely-spined dendrites become varicose distally. Arrowhead points the axon which is illustrated separately in B. Axonal distribution of the neuron shown in A. Note that the beaded collaterals overlapps extensively the caudal (to the right of the figure) parent dendritic field. 47 the poles of the soma radiating into a bipolar-shaped dendritic field which may be oriented along the course of the internal capsule fiber bundles (Fig. 1C, photomicrograph of fusiform large neuron). Other cells, however, have dendrites which are distributed without regard to the orientation of the fibers of passage, and give rise to multipolar-shaped dendritic fields. (Fig. 1A, 1B and 21). The dendritic surface of the primary dendrites and the proximal secondary or tertiary dendrites are usually smooth and straight without varicose appearance, but a few dendrite spines are occasionally found (Fig. 21). Usually only a few of these long dendrites can be analyzed throughout their entire length because they often become unimpregnated or are cut into segments located within adjacent sections where they are frequently entangled with other Golgi impregnated profiles. Nevertheless, some dendrites have been reconstructed from serial sections and have been traced up to 400 an away from the soma (Fig. 2A). Some dendrites become varicose or beaded in appearance distally (Fig. 1B, small arrowhead; Fig. 2A) and give rise to thin beaded processes or sparsely-spined processes. Other long dendrites, however, terminate with gradual tapering. I The axons of these large neurons are usually unimpregnated except for a portion of the initial segment. One type I large neuron, however, has an axon which gives rise to a local collateral near the soma (Fig. 1B, large arrowhead; Fig. 2B). The parent axon, showing no sign of tapering, becomes unimpregnated a short distance from the collateral branch point. The lone local collateral gives rise to many beaded processes (Fig. 2B) extending up to 400 am away from the soma 48 and appears to arborize mainly within the caudal quadrant of the parent dendritic domain. Type II Large Neurons: The somata of the second type of large neurons (n 8 7) usually have at least 5 visible somatic spines (Fig. 38) It is, however, impossible to evaluate accurately the number of somatic spines under the light microscope due to difficulties involved in visualizing spines hidden by the Opaque some. The somata may be polygonal (Fig. 3A, 3B & 4A), round (Fig. 3D) or fusiform (Fig. 4B) in shape and have cross-sectional areas ranging from 328 up to 624 square pm. The dendrites (S from 3 to 9, F from 13 to 42) of these large neurons are multipolar in orientation and often appear to be shorter than those of type I large cells (compare Fig. 2A with Fig. 4A and 4B). Whether this reflects actual dendritic length difference or results from partial impregnation of dendrites of the type II cells remains unclear. The density of the dendritic spine distribution usually decreases with increasing distance from the soma. In comparison to the spiny dendrites of the medium spiny neurons (Fig. 3B, curved arrow), the dendrites of some of these large neurons are virtually aspinous. Distally, the dendrites are either smooth or sparsely studded with a few patches of spines, and occassionally, a few long and thin dendritic appendages. Varicose dendritic swellings or beaded dendritic appendages are not observed on these neurons. The axons of these cells are unimpregnated except for a portion of the initial segments (Fig. 4A & 4B, arrowhead). Figure 3. A. B. 49 Photomicrographs of type II large neurons and type I and type II medium neurons. Type II large neuron, from a gold-toned Golgi section; SA = 542 square um. CAL 3 50 um. Higher magnification of A, straight arrow points to the somatic spines. Note that in comparison to a distal dendrite (curved arrow) of a type I medium neuron, the dendrites of the type II large neuron are virtually aspinous. CAL = 20 um. Medium type I neuron, from a gold-toned Gogli section; SA = 168 square um, same magnification as in A. Type II large neuron, rapid Golgi preparation; SA = 379 square um. Note that its somatic size is intermediated between the large cell shown in A, and the medium cell shown in E. Same magnification as in A. Type II medium neuron, rapid Golgi preparation; SA = 257 square um, arrow points to the axon. Same magnification as in A. Higher magnification of E, showing the presence of somatic spines; same magnification as in B. 50 Figure 3 51 Figure 4. Micr0800pic tracings of type II large and type II medium neurons. A & E: Type II large neurons, SA ' 452 and 374 square nm respectively, arrowheads point to the axons. C, D & E: Type II medium neuron, also in Figure 3E & 3F. Arrowhead point to the axon with a short segment of a collateral. Opened-block arrow points to the somatic region illustrated at higher magnification in D. Closed-block arrow point to the polymorphic terminal dendritic branches within the fiber bundles of the internal capsule, shown at higher magnification in E. F: Another type II medium neuron with relatively even distribution of spines over both soma and dendrites. SA = 150 square um, arrowhead points to the axon. CAL = 50 mm for A, B, C and F; 25 mm for D and E. 52 Figure 5. A. 53 Microscopic tracings of type I medium neurons. This cell exemplifies the most frequently encountered neostriatal neurons in our Golgi preparations, having densely spine-studded distal dendrites, resembling bottle brushes; SA 8 159 square nm. A type I medium neuron with lower spine density on its dendrites; SA 8 105 square pm. A possible type I medium neuron with uneven dendritic spine distribution. This cell also has a smaller soma (SA 8 88 square um) than most of the other type I medium neurons. Arrowheads point to the axons, CAL = 50 um. 54 Figure 5 55 MEDIUM NEURONS: At least five types of medium neurons are found in the rat neostriatum. Type I Medium Neurons: This is the most frequently impregnated neuronal type in our Golgi preparation (Fig. 30; Fig. 5A). The morphology of this cell type appears to be identical to the medium spiny neurons previously described in the Golgi studies of the cat (Kemp & Powell, 1971), the monkey (Fox et al., l97l/l972; DiFiglia et al., 1976), the dog (Tanaka, 1980), and the rat (Mensah & Deadwyler, 1974; Lu & Brown, 1977; Banner & Pfister, 1979a & b; Dimova et al., 1980). These cells are characterized by their spine-free soma and proximal dendrites and densely spine-studded distal dendrites, which often resemble bottle brushes. Three to four local axon collaterals are emitted near the soma and ramify extensively within or near the parent dendritic domain. The main axons, some taking rather tortuous course, join the fiber bundles of the internal capsule. In addition to these observations, we have also found several neurons which have similar but slightly different morphology as those of the more common medium spiny neurons. The distal dendrites of these neurons either do not have as many spines as the more commonly seen medium spiny neurons (compare Fig. 5A with 5B) or have a noticeably uneven distribution of spines on their dendrites (Fig. 50). Yet, the characteristic distribution of many spines on the distal dendrites and the spine-free nature of the proximal dendrites and the soma indicate that these are neurons closely related to the medium spiny neurons. Since the sample size of these relatively spine-poor medium spiny neurons is much smaller than the spine-rich type, we do not feel that they constitute a completely separate cell type. It is S6 likely that they represent cells at the lower end of a continuous spectrum of spine density distribution of all the neostriatal medium spiny neurons. Type II Medium Neurons: The second type of medium neostriatal neurons (n 8 8) have somatic spines (Fig. 3E & 3F; Fig. 4C, 4D & 4F). The overall appearance of these neurons is similar to that of the type II large neurons mentioned previously. A variable number of spines are visible on the soma and the proximal dendrites as illustrated in the high magnification photomicrograph (Fig. 3F) and tracing (Fig. 4D). Frequently, the spine density on the dendrites appears to decrease with increasing distance away from the soma (Fig. 4C & 4D). Occasionally, the spines appear to be evenly distributed over the soma and the dendrites (Fig. 4F). However, none of the neurons of this type have as many dendritic spines as the type I medium spiny neurons (Fig. 5). Some of the radiating and multipolarly distributed dendrites (S from 4 to 6, F from 19 to 36) extend into fascicles of internal capsule fibers of passage and terminate with elaborate branches of dendritic varicosities and appendages (Fig. 4C, closed arrow; Fig. 4E). The axons of these neurons have not been impregnated in our preparation beyond the initial segment (Fig. 3E, 4C, 4D & 4F, arrowheads), one of which gives rise to a branch which also quickly becomes unimpregnated (Fig. 4D, arrowhead). 57 Type III, IV & V Medium Neurons The remaining types of medium neostriatal neurons share several similar morphological characteristics. Namely, these cells have aspinous somata and very sparsely-spined dendrites. However, visual inspections indicate that there are two pOpulations of neurons with non-varicose dendrites, one has poorly-branched dendrites (type III) and the other has profusely-branched dendrites (type IV). The mean values of F and S for type III cells are, respectively, 19.6 (s.d.=5.2, n-10), 4.3 (s.d.=1.5) and those for type IV cells are 39.7 (s.d.=5.8, n=7) and 5.3 (s.d.=0.8). The difference in their F values is statistically significant (t= 2.188, of: 15, p< 0.05). Another group of medium neurons have varicose dendrites and is classified as type V medium neurons. Type III Medium Neurons: These cells (n = 10) have relatively poorly branched dendrites (F from 10 to 24, S from 3 to 7) which may radiate up to 250 um from the soma. The sparsely-spined dendrites are usually smooth without varicose appearance. Some dendrites have a few long and thin appendages. Many of these cells in frontal section have stellate dendritic fields (Fig. 6A, 7A, 7D) comparable to those of the type I cells (medium spiny projection cells) in appearance. Similarly, type III cells in sagittal section have either stellate (Fig. 6B) or slightly polarized dendritic fields without any apparent specific orientation (Fig. 6D). Since many type I cells are also fully impregnated in these same sections (Fig. 6C, arrowhead pointing to a spiny distal dendrite of a type I neuron), it is unlikely that the type III cells are results of partial 58 Figure 6. Photomicrographs of type III medium neurons and a medium neuron with long dendrites. A, B & C: Examples of type III medium neurons with sparsely-spined, non-varicose dendrites. SA = 176, 137 and 177 square um for A, B and C respectively. Arrow in A points to the axon of that cell. Arrow in C points to a distal dendrite of a neighboring type I medium neuron, illustrating that dendritic spines are well impregnated in these sections, therefore the type III medium neurons are probably not medium neurons with unimpregnated spines. D & E: A rarely found medium neuron with a dendrite which extends up to 500 pm from the soma (curved arrow). Straight arrow points to the partially impregnated axon. SA = 150 square pm. Some of its dendrites form polymorphic terminal branches in the fiber bundles of the internal capsule, one of which is shown at higher magnification in E. 59 Figure 6 60 Figure 7. Microscopic tracings of type III medium neurons and a small neuron with intrinsic axons. CAL = 50 pm. A: Somato-dendritic morphology of a type III medium neuron, also shown in Figure 6A; arrowhead points the axon. B: A type III medium neuron, arrowhead points to a well—impregnated axon which arborize extensively near the parent dendritic field; SA = 160 square um. C: A small neuron, arrowhead points to its axon which has a similar distribution as that in B; SA = 91 square pm. D: A type III medium neuron with one somatic spine, arrowhead points to the axon; SA 3 138 square pm. Note its similarity to the type III neuron shown in A, and compare with the tyep II medium neurons shown in Figure 4C and 4F. 61 h whfimflm 62 impregnation of type I cells. Rarely, one or two somatic spines or appendages may be visible on the soma or the proximal dendrites of these neurons as shown in Figure 7D and 12C. HOwever, since the overall appearance of these cells is closer to that of type III cells (Fig. 7A) than to that of type II cells (Fig. 40, 4F), They are classified as type III cells in this study. The axons of most of these neurons are unimpregnated except for the initial segments. One of the type III medium cells has a locally arborizing axon (Fig. 7B, arrowhead). Type IV Medium Neurons: These cells (n = 7) have spine-free somata and smooth or very sparsely—spined dendrites which branch repeatly, forming an elaborate and interwoven dendritic field around the soma (F from 31 to 49, S from 4 to 6; dendritic range up to 250 um) (Fig. 8A, 9A, 9C & 9D). The smooth dendrites generally are not varicose. Occasionally, long and thin dendritic appendages (Fig. 9A, closed block arrow) or patches of dendritic spines or appendages (Fig. 9B) are seen dotting the bare dendrite. Some dendrites of this cell also extend into a fiber bundle of the internal capsule to form specialized endings consisting of branching polymorphic processes with varicosities and appendages (Fig. 8B & SC and Fig. 9A a 9B) The axons of these cells are generally unimpregnated except for a portion of their initial segments. One cell of this type has an axon whcih gives rise to a few beaded local collateral fibers arborizing near the dendritc domain of the parent cell (Fig. 9C, arrowhead). The main axon divides into two equal branches which 63 subsequently give rise to more collaterals. A bulb-like enlargement (Fig. 9D, block arrow) is seen on one of the main branchs near the axon bifurcation. It is interesting to note that even though the neuron shown in Figures 8D has a soma size at the lower limit of the large size range (SA 8 318 square um), its profusely branching dendrites (F = 40 and S = 6) are shorter than the type I large cells (dendritic ranges of 200,nm vs. 400 an of large type I cells). Therefore it appears to be more closely related to the type IV medium cells than other larger neostriatal neurons. Type V Medium Neurons: These cells (n = 5) have smooth somata. The smooth primary dendrites divide near the soma into secondary dendrites which branch infrequently (F from 25 to 33, S from 3 to 5) and become varicose in appearance (Fig. 10, closed-block arrows; Fig. 11B, opened-block arrow) as they radiate into stellate dendritic field extending up to 250 am away from the soma. A few spines or dendritic appendages (Fig. 11B, closed-block arrows) are found on the dendrites. The most striking feature of this type of neuron is the distribution of its axon (Fig. 10, opened-block arrow). Arising from either the soma or the primary dendrite, the axon branches repeatly near the soma ferming a dense network of axon collaterals consisting mainly of beaded thin fibers (Fig. 10, 11A & 11B, arrowhead points to the first branch point of the axon). There is a preferential orientation of the axon arborization toward one side of the dendritic field. There is no indication of a main axon leaving the domain of Figure 8. A: B & 64 Photomicrographs of type IV medium neurons. A type IV medium neuron, SA 3 222 square pm. The arrowhead points to the partially impregnated axon which becomes myelinated as showin in Figure 25C. Arrows point to the polymorphic terminal dendritic branches within the fiber bundles of the internal capsule, also shown in B and C. CAL = 50 pm. C: Higher magnification of the terminal dendritic branches in the internal capsule fiber bundles. Note their spine-like appendages and varicose appearances. CAL 3 10 mm. A neuron with profusedly-branching dendrites (range about 200 mm in radial length) similar to other type IV neurons but its SA is slightly over 300 square um (SA 8 318 square um). Same magnification as in A. 65 Figure 8 Figure 9. A: 66 Microscopic tracings of type IV medium neuron. Same neuron as shown in Figure 8A. Arrowhead points the axon. Closed-block arrow points to one of the long dendritic appendages of this cell. Opened-block arrow points to the terminal dendritic branches within the fiber bundles of the internal capsule, also illulstrated at higher magnification in B. Terminal dendritic branches at higher magnification. A type IV medium neuron, SA = 247 square pm. Arrowhead points to its partially impregnated axon which gives rise to several local collaterals. Higher magnification of C, showing the axons in relation with withthe dendrites. The closed-block arrow points to an enlargement on one of the major axonal branches. A & C are at the same mangnification, so are B & D. CAL = 50 mm. 67 Figure 10. 68 Photomicrograph of a type V medium neuron. Note that the main axon (opened-block arrow) branches first at the arrowhead, and then branches repeatly to form an elaborate collateral arborization near the parent dendrites, many of which have pronounced varicose varicose appearance (closed-block arrow). SA = 178 square um. CAL = 20 um. 69 Figure 10 Figure 11. A. 70 Microscopic tracings of a type V medium neuron. Same neuron as that in Figure 10, shown at the same magnification as other types of neurons described previously. Arrowhead points to the first branch point of the axon. CAL 8 50 um. Higher magnification of A, showing the presence of a few dendritic spine or appendages (closed-block arrows) and the varicose nature of the dendrites (Opend-block) arrow. Additionally, the elaborate local axonal collateral arborization within the parent dendritic domain is illustrated in detail. CAL = 50,nm. 71 Figure 11 72 the local axon arborization. This pattern of axon arborization suggests that this is a type of neostriatal interneuron. SMALL NEURONS: Since the small neurons in the rat neostriatum are rarely impregnated, they are not subdivided into different types at this time. One of the small cells has a round soma (Fig. 7C, SA = 91 square um), and three primary dendrites which branch into a small stellate shaped dendritic field, extending up to 150,nm from the soma. A few spines and headed appendages are observed on the dendrites. The axon (Fig. 7C, arrowhead) arises from the soma and arborizes extensively within the parent dendritic field. The overall appearance of this cell is similar to the medium neuron shown in Figure 8B (SA = 160 square um). Several cells have smooth soma and a few sparsely-spined dendrites which usually arise from the poles of the soma (Fig. 13A & 13E). Rarely, a cell would have somatic spines or appendages (Fig. 12E & 13B). It is unclear how long their dendrites may extend because most of them become unimpregnated a short distance from the soma. The' dendrites in sagittal sections usually align themselves with the course of the internal capsule fibers of passage. Axons often arise from a primary dendrite (Fig. 13A, 13B & 13E, arrowheads), some give rise to collaterals near the soma. Two neurons with somatic sizes at the lower limit of the medium cells or the upper limit of the small cells (i.e. SA value is near 100 square um) have no apparant axons. One of these cells has a maximal somatic cross-sectional area of 105 square um (Fig. 12A, 12B & 73 13C). Its bipolarly distributed dendrites are sparsely-spined (Fig. 12A, arrowhead). The other cell has a maximal somatic cross-sectional area of 120 square mm (Fig. 12C & 13F), and it has aspinous dendrites some of which are varicose and have long and thin appendages (Fig. 12D, and 13H closed-block arrow). No axon is seen to arise from either soma or dendrites, but a thin process resembling a cilium is visible arising from the soma (Fig. 13H, opened-block arrow). This process has similar appearance to the cilia of the impregnated ependymal cells lining the ventricular surface in the same section. RARELY-FOUND CELL TYPES The neuron shown in Figure 13D (SA = 146 square pm) has profusely branching dendritic processes (S = 4, F = 33) which are sparsely covered with short and stubby appendages. One of the processes, however, has no surface appendages and appear to be extending away from other processes (Fig. 13D, arrowhead), indicating that this may be the axon of this cell. Two medium cells have long dendrites. One (Fig.14A) has a fusiform aspinous soma (SA 8 180 square pm) with four primary dendrites, one of which is cut off into an adjacent section. The dendrites are poorly branched and may radiate up to 450 nm in length. The dendritic surface is essentially aspinous with some varicose appearance. Only one or two spines or appendages are observed on the dendrites. The axon (Fig. 14A, arrowhead) of this cell gives rise to a few collaterals near the soma, but there is no indication that the main axon terminates locally near the soma. 74 The other unique medium neostriatal neuron with long dendrite (Fig. 14B, SA = 150 square pm) is also shown in photomicrogaph in Figure 6D. It has a few somatic spines resembling the type II medium neuron (Fig. 4F). Its radiating dendrites are sparsely-spined, resembling those of type III medium neurons (Fig. 6A, 7A, and 7D), but most of its dendrites terminate with elaborate polymorphic processes within the fiber bundles of the internal capsule (Fig. 6E; Fig. 14B, Opened-block arrows) comparable to those of a type IV neuron mentioned earlier (Fig. 83, so a 93). On the other hand, it is similar to the cell shown in Figure 14A because it has a dendrite extending over 500 pm in length. The axon of this cell is unimpregnated except for its initial segment (Fig. 14B, arrowhead). NEUROGLIAFORM CELLS Although these cells are very rarely found in most of our preparations, they are well represented in some rat brains. They have small cell bodies (SA = 62 square mm, or equivalent to diameters less than 10 pm in diameter) and numerous short processes projecting radially from the soma, resembling a fuzz ball with radius of about 30 ,nm (Fig. 15). Many of the processes are varicose or beaded in appearance (Fig. 15B, arrow). However, none of the processes can be identified as either axon or dendrites. Similar cells have been found in both the cerebral cortex and the globus pallidus. 75 Figure 12. Photomicrographs of small neurons. A & B: C & D: A neuron shown in B is at the same magnification as the other types of neurons described earlier. as well as that shown in Figure 12C, CAL = 50 um. It is shown at higher magnification in A to demonstrate the presence of a few spines (arrow) on its dendrites. This cell is also shown in Figure 130. CAL in A . 20 um. A neurons with SA = 120 square pm. This cell has sparsely-spined dendrites, some of which have long dendritic appendages. One of the long dendritic appendage (arrow) is shown at higher magnification in D. This cell is also shown in Figure 13F. CAL in C = 50 um. CAL in D = 20 um. E: A fusiform neuron with a few somatic spines and a few sparsely-spined dendrites. This cell is also shown in Figure 13B. SA = 95 square pm. Same magnification as in D. 76 Figure 12 Figure 13. P] 77 Microscopic tracings of small neurons. CAL = 50 pm. These are neurons with somatic cross-sectional areas near or below 100 square pm. Some of which are very rarely-found uniques cells which are not classified as separate types of neurons at this time. A small neuron with smooth soma and sparsely—spined dendrites. Arrowhead points to the axon. SA = 66 square pm. A small neuron with somatic spines and sparsely—spined dendrites. Arrowhead points to the axon. This is the same cell shown in Figure 12E. A neuron with smooth soma and two primary dendrites which branch into sparsely-spined distal dendrites. No axon has been identified. This is the same cell shown in Figure 12A & 12B. A neuron with many thin dendrites some of which have short appendages. Arrowhead points to a possible axon-like process. SA 3 146 square um. Another small neuron with sparsely-spined dendrites and smooth soma. The axon (arrowhead) arises from a dendrite. SA = 94 square pm. H: The same neuron as shown in Figure 12C. This cell has no identifiable axon. Opened-block arrow points to the somatic region shown at higher magnification in H. Small closed-block arrow points to the long dendritic appendage also shown in Figure 12D. Small Opened-block arrow in H points to a cilium-like process on the soma. No axon has been identified for this cell. A type III medium neuron with one somatic spine. Arrowhead points to the partially impregnated axon. SA = 129 square pm. 78 Figure 13 Figure 14. A: 79 Microsc0pic tracings of medium neurons with long dendrites. CAL = 50 pm. This neurons has virtually aspinous dendrites, some show slightly varicose appearance and extend up to 450 pm from some (about twice the dendritic length of a type I medium neuron). Arrowhead points to its axon which gives rise to a few local collaterals before becoming unimpregnated. SA = 180 square pm. This is the same neuron shown in Figure 6D. One of its dendrite extends up to 500 pm from the soma. Arrowhead points to the partially impregnated axon. Opened—block arrows points to the polymorphic terminal dendritic branches within the fiber bundles of the internal capsule. One of these terminal branches is also shown in Figure 6E. “1 FJ gure 15. 81 Photomicrographs of a neurogliaform cell. This small cell, SA = 50 square pm, is shown at the same magnification as those of other types of neurons described previously. Higher magnification of A, showing its slender and varicose short processes (arrow) which extend no farther than 50 pm from the soma. CAL = 10 mm. 82 Figure 15 83 B. Electron Microscopic Analysis of Golgi Material At least one of each type of rat neostriatal neurons has been gold-toned and analyzed at the electron microscopic level. General Observations & Comments Since the gold-toned Golgi materials have undergone much chemical and physical treatments before postfixation in OsO4 solution and plastic embedding, the preservation of the tissue was often less than ideal for critical comparative analysis of samples from different rats. Frequently the gold particles would accumulate around the synaptic densities and would obscure the synaptic membrane specialization. Thus the symmetry of synaptic contacts in most Golgi gold-toned materials often remained unclear, and the analysis of synaptic relationships of gold-toned neurons was limited mostly to descriptions of the relative size of the synaptic vesicles in the presynaptic terminals. Occasionly, when the morphology of the presynaptic boutons were very poorly preserved, only indications of synaptic contacts could be resolved. Rarely, the opposing synaptic lmembranes are revealed as pale lines against the dark gold particles and the synaptic gap may be measured. In the case shown in Figure 17D, the synaptic gap measures 17 nm. In general, the cytOplasm of the gold-toned neurons appears to be washed-out in comparison to the neighboring neurons which were not impregnated by Golgi reaction product. This appearance may have resulted from the dissolution of the Golgi reaction product by the sodium thiosulfate solution during the last step of the gold-toning 84 procedure, leaving empty spaces behind. In some cases, the deposition of gold particles was insufficient to provide an adequate light microsc0pic image after de-impregnation by the sodium thiosulfate solution. In these cases, locating the gold-toned profiles under the electron microsope depended on prior detailed tracings of surrounding landmarks (blood vessels or fiber bundles) and the recognition of this washed-out or extracted appearance of the de—impregnated cytoplasm. Where feasible, the morphology of the axons was analysed to determine whether the initial segments receive synaptic inputs and whether the axons were myelinated. The morphology of the intrinsic collateral terminals and their postsynaptic targets were also studied. Type I Large Neuron: The light micrOSCOpic morphology of the type I large neuron selected for this electron microsc0pic analysis is shown in Figure 1C. The soma of this large neuron is characterized by the presence of a deeply indented nucleus surrounded by a large amount of organelles, including well organized perinuclear Golgi complexes and abundant Nissl substance (i.e. rough endoplasmic reticulum) (Fig. 16 & 17A). Numerous dense bodies resembling either lysosomes or lipofuscin granules are scattered in the cyt0plasm. Despite the large size of its soma, very few axosomatic synapses are observed on the type I large neuron (Fig. 17A, white arrow). The presynaptic terminals contain either medium (40 to 45 nm in diameters) (Fig. 173) or small vesicles (30 to 35 nm) (Fig. 170 & 17D). Figure 16. 85 Electronmicrograph of a type I larage neuron. This is the same neuron shown in Figure 10 after it has been gold-toned. Note that its nucleus is invaginated, and the cytoplasm contain large amount of Nissl substance (rough endOpalsmic reticulum and free ribosomes), dense bodies, mitochondria and Golgi complex. The last organelle is better illustrated in Figure 17A. CAL = 5 um. 86 Figure 16 87 Figure 17. Electronmicrographs of soma and dendrites of a type I large neuron. A: A tangential sections of the soma, illustrating the perinuclear Golgi complex (g). Only a small portion of the nucleus (n) is visible. Large amount of Nissl substance (rer) and dense bodies are also found throughout the cytOplasm. White arrow points to a axosomatic synapse also shown in C and D. CAL - 5 um. B: An axosomatic synapse (arrow), the presynaptic terminal contain medium vesicles. CAL ' 0.5 mm. C: Another axosomatic synapse (arrow), the presynaptic terminal contain small vesicles. Same magnification as in B. D: Higher magnification of C, showing the opposing synaptic membranes contrasted by the dark gold particles. The synaptic gap 8 17 nm. CAL = 0.5,nm. E: A primary dendrite forms a synpase with two active sites (arrows) with a presynaptic terminal containing medium vesicles. Same magnification as in B. F: A distal secondary dendrite is contacted by several terminals (arrows) containing medium vesicles. The dense deposit of gold particles prevents detail analysis of the synaptic junctions. Same magnification as in B. 88 Figure 17 Figure 18. A: 89 Electronmicrographs of soma and axon of a type II large neuron. The soma of the type II large cell which is also shown in Figure 3A and 3B. Note that its nucleus is deeply invaginated, and the cytoplasm contains large number of mitochondria, microtubules and Golgi complexes (g). Although not apparent here, adjacent sections also contain abundant Nissl substance. CAL 8 5 pm. The axon of this neuron becomes unimpregnated at the point of myelination. CAL = 1 pm. 90 Figure 18 Figure 19. A: 91 Electronmicrographs of a somatic spine and dendrites of a type II large neuron. CAL = 0.5 mm. A somatic spine is contacted by terminal # 5. Also shown are numerous axosomatic synapses made by terminals #1, 2, 3 & 4. Terminal #6 also synapses with the somatic spine on an adjacent section. A primary dendrite with a spine which is contacted by terminals #1 & 2. Terminals #4, 5 & 6 synapse with the dendritic shaft. Terminal #3 contain larger vesicles and fOrm symmetrical contacts with an adjacent dendrite as well as the gold-toned dendrite on an adjacent section. A distal secondary dendrite is contacted by terminals #1, 2 & 3 containing medium vesicles, and #4 containing small vesicles. Terminal #1 (curved arrow) also makes symmetrical synapse with a neighboring spine (straight arrow). A spine on a distal tertiary dendrite is contacted by terminal #1. Terminal #2 synapses on the dendritic shaft at the base of the spine. Terminal #4 synapses on the dendritic shaft opposite to the spine. All of these contain medium vesicles. 92 Figure 19 Figure 20. 93 Electronmicrographs of dendrites of a type II large neuron, showing synapses and partial impregnations. CAL = 0.5,nm. This proximal secondary dendrite is contacted by terminals #1, 2 & 3 containing medium vesicles but has different vesicle packing densities. This tertiary dendrite becomes unimpregnated. The terminals #1, 2 & 3 synapses on the impregnated part, whereas #4 & 5 synapse on the unimpregnated part of the dendrite. These terminals contain _medium to small vesicles. In comparison, a terminal containing large vesicles (arrow) synapses on an adjacent dendrite. 94 Figure 20 95 Synapses are also found on both proximal and distal dendrites. A presynaptic bouton containing medium vesicles is shown making a synpatic contact on a primary dendrite in Figure 17E with two active zones (arrows) and 22 nm synaptic gap. Terminals containing medium vesicles also make synaptic contacts with distal secondary dendrite (Fig. 17F, white arrows). The nature of the axonal morphology of the gold-toned type I large neurons at the electron microscopic level remains unclear. However, a type I large neuron has been labeled intracellularly with HRP, and it has a myelinated main axon (Fig. 32). Type II Large Neurons: The somata of these large cells are also characterized by the presence of deeply indented nuclei, some containing intranuclear filaments, surrounded by a large amount of cytoplasmic organelles (Fig. 18A). However, the presence of many axosomatic synapses and synapses on somatic spines (Fig. 19A) distinguishes this type of large cell from type I large cells. The primary dendrites (Fig. 19B) also appear to receive more synaptic contacts than those of type I large cells. Often more than one presynaptic terminal synapses with one somatic or primary dendritic spine and most of these terminals contain medium vesicles. It is unclear whether these spines contain a spine apparatus. Many terminals containing medium vesicles make synaptic contacts with distal dendrites. Some of these terminals also make asymmetrical synapses with spines of neighboring neurons (Fig. 19C, terminal #1). Occasionally, terminals containing large vesicles 96 (diameter >50 nm) make synaptic contact with the dendritic shaft at the base of the dendritic spine (Fig. 19D, terminal #2). Although different sized presynaptic boutons containing varying amount of synaptic vesicles make synaptic contacts with the type II large cells (Fig. 20A), neither of these two parameters is reliable for classifying axon terminals. This is because the size of the presynaptic terminals and the packing density of the synaptic vesicles_ often vary according to the planes of sections. Some of the dendrites of the type II large neuron have been revealed to be only partially impregnated by the Golgi reaction products (Fig. 20B), illustrating the capriciousness of the Golgi technique and the need for critical interpretation of dendritic length based on Golgi data. The main axon of the large type II neurons becomes myelinated at the point where the Golgi impregnation stOpped (Fig. 18B). No collaterals arise from the unmyelinated initial segment. Type I Medium Neurons: The somata of these cells are characterized by a round, unindented nucleus surrounded by a rim of cytOplasm which contains many free ribosomes (in rosettes), mitochondria and microtubules (Fig. 21A). These cells appear to have the least amount of rough endoplasmic reticulum and the most poorly developed Golgi complexes of all the neostriatal neuronal types. The somatic profiles of the medium type I cells vary from small (7 or 8.um in diameters) up to large (up to 20 mm or more in diameters) depending on the orientation of the plane of the section. Most of the microtubules appear to be aligned along the long axis of the neuron. 97 Axon terminals containing either large or medium vesicles have been fbund making synaptic contacts with the somata of the type I medium neurons (Fig. 21B). Similar boutons also make synaptic contacts with the primary dendrites. Although some axodendritic synapses have been observed on distal dendrites (Fig. 210), these latter are mainly asymmetrical axospinous synapses with boutons containing medium vesicles. Further details relating to the ultrastructural morphology of the medium type I neurons are described in the intracellular HRP labeling studies in which the tissue is better fixed. Type II Medium Neurons: The somata of the type II medium neurons are characterized by the presence of a large nucleus which has indented nuclear membrane and often contains intranuclear rod (Fig. 22A). The intranuclear rod consists of aggregates of filaments aligned in the same direction (Fig. 22B). These neurons have synaptic contacts on both the somatic surface and the somatic spines (Fig. 22C). Often two or more boutons containing mainly medium vesicles terminate on the same somatic spine. Since the somatic spines occupy only-a small portion of the somatic surface area, the somatic spines are not observed in most sections through the soma. Terminals containing either medium or small vesicles make synaptic contacts with either dendritic shafts or dendritic spines of both the primary and distal dendrites (Fig. 22D, 23). It is unclear whether the spines of these cells have spine apparatus because the interior of the spinous processes have often been obscured by either 98 gold particles or dark, amorphous cytOplasmic derivatives (Fig. 23). Although the axon of one type II medium neuron has been traced through serial sections until the stopping point of Golgi impregnation, no indication of myelination has been observed. Type III Medium Neuron: The soma of the type III medium neuron contains a nucleus with relatively shallow nuclear membrane invaginations (Fig. 24A). Small stackes of rough endOplasmic reticulum and Golgi complexes are found in the surrounding cytOplasm. The axosomatic terminals contain large, medium or small vesicles (Fig. 24D, 24E, 24F). Primary as well as distal dendrites make synaptic contacts mainly with boutons containing medium vesicles (Fig. 24G, 24H, 24:). The rarely found dendritic spines or appendages are also postsynaptic (Fig. 24H). It is interesting to note that an entire primary dendrite and its branches was not impregnated with the Golgi reaction product (Fig. 24B). The stopping point of impregnation at the base of the dendrite is marked by a gap less than 0.5 um in width. Whether this gap was a cause of the failure of impregnation of this dendrite remains unclear. Boutons containing large vesicles make synaptic contacts with the axon initial segment (Fig. 240). However, this axon, like that of the type II medium neuron, became unimpregnated before any indication of myelination. Figure 21. A: 99 Electronmicrographs of a type I medium neuron. The soma of type I medium neuron contains a round, unindented nucleus. The surrounding cytoplasm contains abundant free ribosomes, mitochondria and microtubules. CAL - 5‘um. Axosomatic synapses on the type I medium neurons. Terminal #1 contains larger vesicles than terminal #2. CAL = 0.5 mm. A distal dendrite with spines (3). Terminals #1 d 3 containing mediuum vesicles synapse with the spines, whereas terminal #2, containing medium to large vesicles, synapses on the dendritic shaft. CAL = 0.5 pm. 100 Figure 21 Figure 22. A: 101 Electronmicrographs of a type II medium neuron. The nucleus (n) of the type II medium neuron is invaginated and contains intranuclear rod. Arrow points to the head of a somatic spine also shown at higher magnification in C. CAL - 5 um. Higher magnification of the intranuclear rod showing its filamentous nature. Arrows points to the indented nuclear membrane. CAL ‘ 0.5 um. Higher magnification of the somatic surface and the somatic spine. Terminals #1, 2 d 3 synapse on the head of the spine, and #4 d 5 synapse on the soma. Same magnification as in B. The primary dendrite contains numerous microtubules and is contacted by terminals #1, 2, 3, 4, 5 & 6. Same magnification as in B. 102 b',!,\1-4~. >. I Figure 23. A: 103 Electronmicrographs of dendrites of a type II medium neuron. CAL = 0.5 pm. A cross-sectioned primary dendrite is contacted by terminals #3, 4, 5, 6 & 7. An emerging spine (3) is synapsed upon by terminals #1 & 2. Terminal #1 synapses at the base of the spine (3) on a distal dendrite. Terminals #2 & 3 synapse on the shaft. The spine also make synapses on adjacent sections. Another distal dendrite makes symmetrical synapse (arrow) with terminal #1 which contains small vesicles. A thin distal dendrite is contacted by terminals #1 and #2, both contain medium vesicles. 104 Figure 23 Figure 24. A: C: 105 Electronmicrographs of a type III medium neuron. The nucleus of this cell has shallow invaginations (arrows). The surrounding cytoplasm contains a few stacks of rough endoplasmic reticulum (rer) and Golgi apparatus (g). CAL 8 5 mm. One of the primary dendrites has not been impregnated by the Golgi reaction products. The cisterns of a Golgi apparatus (g) extend from the soma into the dendritic cytOplasm. n: nucleus. CAL 8 1 mm. A terminal synapsing on the axon initial segment (arrow) contains large vesicles. D, E & F: Axosomatic terminals contain vesicles ranging G: H from large (D), medium (E) to small (F). A large terminal, containing medium vesicles, synapses on a primary dendrite. Terminals #1 & 2 synapse on the dendritic shaft, and #3 on the spine (s) of a distal dendrite. All contain medium to small vesicles. On a distal dendrite, the terminal #1 contains medium to small vesicles while #2 contains large vesicles. C to I are at the same magnification, CAL 8 0.5 um. 106 Figure 24 107 Type IV Medium Neuron: The nucleus of the type IV neuron also has shallow indentations (Fig. 25A). This appearance together with the presence of small stacks of rough end0plasmic reticulum and Golgi complexes resemble the type III medium neuron described above. Somatic synapses similar to those on the type II and type III medium neurons have also been found. The soma of this type IV medium neuron is juxtaposed next to another neuron which has not been impregnated by the Golgi reaction product. No intervening glial processes have been observed at the point of apposition. This neighboring neuron has an unindented nucleus and little, if any, rough endoplasmic reticulum and Golgi complexes. The axon of the type IV medium neuron has been found to be myelinated at 25 um away from the soma (Fig. 25B, 25C). The Golgi impregnation stopped near the beginning point of myelination as indicated by the location of gold particles (Fig. 250). The proximal and the distal dendritic processes synapse mainly with boutons containing medium vesicles (Fig. 26), some of these boutons appear to be dark, and fully packed with vesicles and mitochondria (Fig. 26F, bouton #5). Spine-like dendritic appendages are also postsynaptic, with terminals synapsing mainly with the heads of the appendages (Fig. 26A, B, C). These dendritic appendages often contain mitochondria interconnecting with those of the dendritic shafts. Some of the very thin dendritic appendages observed in the light microsc0pic analysis are postsynaptic (Fig. 26D). The processes of the polymorphic terminal dendritic branches within the internal capsule fiber bundles also has been found to be postsynaptic (Fig. 26F). Figure 25. A: 108 Electronmicrographs of soma and axon of a type IV medium neuron. This is the same cell shown in Figure 8A and 9A. The some of this type IV neuron is apposed to another neuron (arrowheads) with no intervening glial processes. Its nucleus (N) has a few shallow indentations (arrow). The surrounding cytoplasm contains Golgi apparatus (g) and a few stacks of rough endOplasmic reticulum (rer). CAL 8 5 am. A low magnification micrograph showing the soma of of the same type IV neuron (n) in relationship to its partially impregnated axon (white arrows). The long arrow indicates the trajectory of the axon initial segment which has been cut into adjacent sections. CAL 8 10 pm. A higher magnification of the partially impregnated axon. A mitochondria is seen crossing between the impregnated part of the initial segment and the unimpregnated myelinated axon (block arrow). CAL 8 1 pm. \.‘ “Va" “3 .-,, > ‘ c 1' *1 v f r ' x‘ . , ‘ ' 1- -J 4". o I. A b“ t “ ' , ..W' .. .~ ,,..-r~ AN di‘v .7 fif .31.... . < «I»: ~.' ' ' . Fwy-r. ‘3 .- ; . '. ' "' fl fia- -' .. . .l ‘ 'i' ( . '5'". r'«‘ Figure 25 . .-. I’m ¢I."y .1 '. ‘4'. .4 . . " , ' .5 - . . ‘ l. I ‘ .'_’ ' .L ... Q '_ ; - p . . . . ' . " , -. ,'. .l " . . P' ,. -» .- , . , _. II" J ' , . . , ,. , 4 , . 1 .. '1 c . ‘l . . ‘ . 110 Figure 26. Electronmicrographs of dendrites of a type IV medium neuron. CAL = 0.5 um. A, B & C: These are adjacent sections through a spine-like appendage (s) on a distal dendrite. Terminal #1 contains medium vesicles and synapse on the head of the appendage, which appears to contain a mitochondrion. The thin spine-shaft is about 3,um in length. Terminal #2 synapses on the dendritic shaft. D: A thin dendritic appendage with a small bulbous head is encircled by an axon terminal (small white arrows). E: One of the terminal dendritic branches within the fiber bundles of the internal capsule is shown to be postsyanptic (small white arrows). F: The distal secondary dendrite is contacted by terminals #1, 2, 3, 4 & 5, containing mainly medium vesicles. Figure 26 111 112 Type V Medium Neuron: This neuron has a deeply invaginated nucleus surrounded by well developed Golgi complexes and several stacks of rough endoplasmic reticulum (Fig. 27A). Boutons containing either large or medium vesicles make synaptic contacts on both soma and dendrites, including the dendritic varicosities, have been observed (Fig. 27E, 27F). However, few cytOplasmic details could be resolved from this particular neuron because of the accumulation of electron dense material within the cytOplasm. The axon previously identified light microsc0pically has been traced in serial sections to many branching points where the diameters of the daughter branches gradually decreased without any indication of myelination. The initial segment or rather, the proximal portion of the main axon appears to receive no synaptic contacts (Fig. 27D). Distally, the boutons of the axon collaterals make either axosomatic (Fig. 27B) or axodendritic (Fig. 270) synapses. At high magnification, the boutons appear to contain elongated small vesicles (Fig. 270) and make symmetrical synapses with a 24 nm synaptic gap. The postsynaptic somata often have unindented nuclei, very little rough endoplasmic reticulum but many free ribosomes. Small Neuron The small neuron selected fer electron microscopic analysis is that shown in Figure 12A. This neuron has a large, deeply indented nucleus surrounded by a narrow rim of cytOplasm (Fig. 28A). The soma is juxtaposed between two other neurons with similar somatic morphology, and there is an area where no glial processes intervene 113 between this small cell and its adjacent neighbor (Fig. 28A, arrowheads). Some ribosomes and a few cisterns of rough endoplasmic reticulum are found within the perikarial cytoplasm. A few somatic synapses have been observed (Fig. 28B), the presynaptic boutons contain medium vesicles. Boutons containing small to medium vesicles make synaptic contacts on the primary dendrites (Fig. 280), whereas the boutons contacting the distal dendrites and dendritic spines contain mainly medium or medium to large vesicles (Fig. 28D). Figure 27. A: D D: E: vs 114 Electronmicrographs of a type V medium neuron. The soma conatins a nucleus with deeply invaginated nuclear membrane (small white arrows). A very well deveIOped Golgi apparatus (g) occupies a perinuclear position. A few stacks of rough endoplasmic reticulum (rer) are also found in the cytOplasm. CAL 8 5 mm. Two of the teminals (T) of the intrinsic collaterals of this type V neuron make symmetrical axosomatic synapses (arrows) on the same neighboring neuron which has unindented nucleus (n). CAL 8 0.5 pm. A higher magnification view of an intrinsic terminal (T) making a symmetrical axodendritic synapse (arrow) on a neighboring dendrite. Note the presence of small elongated vesicles within the amorphous dark terminal. CAL 8 0.5 nm. , E d F are at the same magnification as in B. The axon (AX) of this type V neuron is not myelinated. The soma of this type V cell is contacted by a terminal containing large vesicles (arrow). The dendrite is contacted by terminals #1, 2 & 3, containing mainly medium vesicles. 115 Figure 27 Figure 28. A: 116 Electronmicrographs of a small neuron. This is the same cell shown in Figure 12A, 12B and 13C. The soma of this small cell lies between the somata of two other neurons (1 a 2) both of which have indented nucleus. The nucleus (n) of the small cell is deeply invaginated. Few organelles are observed in the surrounding narrow rim of cytoplasm. Its soma is apposed to the some of a neighbor cell without any intervening glial processes (arrowheads). CAL 8 5 mm. C & D are at the same magnification. A terminal containing medium vesicles makes axosomatic synapse on this small cell (arrow). CAL 8 0.5 pm. A terminal containing small to medium vesicles synapses on the primary dendrite (arrow). Terminals #1 synapses on a spine-like appendages of a distal dendrite (d). Terminal #2 synapses on the dendritic shaft. Both terminals contain medium vesicles. 117 Figure 28 118 C. Extracellular HRP Studies Light Microsc0pic Analysis The site of extracellular HRP injections has been analyzed histologically, and only those cases where the HRP was restricted outside of the neostriatum were used fer further light and electron microsc0pic analysis. A typical injection site placed in the globus pallidus is shown in Figure 29A. Note that the injection micropipette was introduced into the brain at an angle, penetrating the hippocampus, the thalamus and the internal capsule before reaching the globus pallidus. The neostriatal projection neurons labeled retrogradely from this type of injection are shown in Figure 29B and 29C. Typically, these neurons have cell body sizes at the lower range of the medium size (i.e. somatic cross-sectional areas near 150 square pm). The somata are often filled diffusedly with HRP reaction product together with some dark granules which may also be feund in the proximal dendrites (Fig. 29B, 290). Distally, the density of HRP reaction product gradually fades with increasing distance from the somata, making analysis of distal dendrites virtually impossible. Nevertheless, some of the dendritic spines were barely visible under the highest-power oil objectives. In thick (100 um) sections, the neostriatal projection neurons can be seen to have mainly multipolar dendritic trees (Fig. 29B). In semithin (2,um) plastic section, the projection neurons appear to have a large round nucleus surrounded by a rim of HRP positive cytoplasm. In no case was a large neuron found to be labeled with HRP from extracellular injections placed outside of the Figure 29. 119 Photomicrographs of the site of extracellular horseradish peroxidase injection and the retrogradely labeled neostriatal projection neurons. Horseradish peroxidase (HRP) injecxtion site is in the globus pallidus (G). Leakage of HRP into the neostriatum (caudate-putamen, or C-P), if any, is minimal. A retrogradely filled neostriatal projection neuron. Both diffuse and granular HRP labeling are observed. SA 8 113 square,um. The presence of dendritic spines on the distal dendrite (arrow) cannot be convincingly demonstrated by light microsc0py. CAL 8 20 mm. A semi-thin plastic section through two neostriatal projection cells. Note that the upper cell has a round, unindented nucleus surrounded by a rim of HRP-positive cytOplasm with both diffuse and granular labelings. Dark HRP-positive granules are also clearly visible on the lower cell. Same magnification as in B. 120 Figure 29 121 neostriatum. Although many medium neurons have been labeled by this study, a larger number of medium neostriatal neurons were not labeled, and may reflect the fact that the small doses of individual extracellular HRP injections in this study were able to reach only small number of neostriatal projection axons. Although the axons of the neostriatal projection neurons serve as conduits for the HRP molecules, the axons and their collateral branches are the least labeled comparing to the somata and the dendrites. Thus the axonal distribution pattern of individual neostriatal projection neurons remains unclear with this method. However, the axons of passage, perhaps cortical or thalamic in origin, have often been well labeled in this material, giving rise to both boutons "en passant" and "terminaux" within the neostriatum. Electron MicroscOpic Analysis The soma of a neostriatal projection neuron contains a large round, unindented nucleus surrounded by a thin rim of cytOplasm which often appears darker than the surrounding neurOpil due to its content of HRP reaction product (Fig. 30). A few dense bodies containing very dense HRP reaction product are also found in both the perikarial and primary dendritic cytoplasm. Many free ribosomes and some distended cisterns resembling damaged Golgi complexes have been feund in the soma. Little, if any, rough endoplasmic reticulum is found in the soma. Boutons forming axosomatic synapses with the neostriatal projection neurons contain either medium or large vesicles, many of which are pleomorphic (Fig. 30B, 30C). Similar boutons also make synaptic contacts with the primary dendrites. Distally, the HRP 122 labeled dendrites give rise to spines which form asymmetrical synapses with boutons containing medium vesicles (Fig. 30D, 30E). The HRP reaction product within the spines is very faint, explaining the difficulties in visualizing them light microsc0pically. Figure 30. A: 123 Electronmicrographs of a neostriatal projection neuron retrogradely labeled with horseradish peroxidase. The soma of a neostriatal projection neuron. Note that its nucleus is round and unindented. Surrounding rim of cytOplasm contains free ribosomes but few rough endoplasmic reticulum. CAL 8 5 um. B, C, D & E are at the same magnification. B: C: Terminals #1 & 2, containing medium to large vesicles, make symmetrical axosomatic synapses on the projection neuron. A neighboring terminal #3, containing medium to small vesicles, makes asymmetrical synapse with a HRP-negative spine. CAL 8 0.5 mm. Terminal #1, containing mainly large vesicles, makes symmetrical axosomatic synapses on the projection cell. Terminal #2 contains mainly small vesicles and does not synapse on the some of the projection cell. D & E: Adjacent sections showing the distal dendrite (ds) of a projection neuron. Two dendritic spines (white #1 & 2) make asymmetrical synapses with two terminals (black #1 d 2) containing medium to small vesicles. Their spine-shafts are oulined by the arrowheads. In comparison, a terminal containing larger vesicles makes symmetrical axodendritic synapse with a neighboring dendrite. 124 Figure 30 125 D. Rat Neostriatal Neurons Intracellularly Labeled with Horseradish Peroxidase General Comments In this study, we have analyzed the physiology and the morphology, at both light and electron microsc0pic levels, of the type I medium (medium spiny) neurons and two aspinous neurons corresponding to the type I large and type III medium neurons described in the preceding Golgi analysis. The sampling of two aspinous neurons out of more than one hundred intracellularly labeled neostriatal neurons in the present study is consistent with the results obtained in previous intracellular HRP labeling experiments in the cat and in the rat, where only type I medium neurons have been recovered. This sampling is also consistent with Golgi studies which show that the majority of neostriatal neurons are type I medium neurons (Kemp & Powell, 1971a, Fox et al., l97l/l972a, b; DiFiglia et al., 1976; also see preceding section A). With the exception of the large aspinous neuron, whose response to CI stimulation was not tested, all of these intracellularly labeled cells had similar excitatory response properties fellowing stimulation of GK and SN. This similarity in responses to two afferent sources is not surprising in light of previous anatomical studies which have shown that the various neostriatal neurons receive a similar range of afferent terminals (Kemp, 1968; Kemp & Powell, 1971b, 0). Whether there are subtle differences in response prOperties of these neurons requires further investigation. 126 The morphology of the two lightly labeled aspinous neurons will be described first, subsequently the morphology of the medium type I neurons will be analyzed in detail since their sample size is much larger. Aspinous Neurons: Light Microsc0pic Analysis Two aspinous neurons have been lightly labeled in this study. One is a large aspinous neuron with smooth soma (SA 8 435 square pm, or 34 X 20 mm in diameters) and five smooth primary processes which become varicose at various distances from the soma, some extend up to 450 um in length (Fig. 313). Because of the faint HRP labeling of this cell, it was difficult to distinguish the axon from dendritic process with light microsc0pic examination. However, electron microsc0pic analysis has revealed that the ventral process becomes myelinated (Fig. 32A, inset) at a distance of 50.um from the soma. We have been unable to follow this axon to its termination, and no collateral fibers have been observed in the short length (200.nm) over which it could be followed. The somatic and dendritic morphological characteristics of this neuron indicate that it is one of the type I 7 large neurons as defined in the preceding Golgi study. This cell responded with EPSPs to stimulation of SN (Fig. 3131). The other aspiny neuron found in this study (Fig. 31A and 34A) has a fusiform soma (SA 8 115 square mm, or 15 X 8,um in diameters) and three faintly labeled aspinous processes which divide into secondary and tertiary branches. It was not possible to identify the axon of this cell with either light or electron microsc0pic examination. This cell responded with EPSPs to stimulation of SN 127 Figure 31. Microsc0pic tracings of aspinous neurons labeled intracellularly with horseradish peroxidase. CAL 8 50 um. A: Medium aspiny neuron, note its sparsely branched dedrites and their smooth dendritic surface. B: Large aspiny neuron, note its long and aspinous varicose dendrites and the smooth soma. A1 & A2: Excitatory responses of the medium aspiny neuron to substantia nigra and cerebral cortex stimulation respectively. A3: Low gain record of A2, showing the action potential elicited by cerebral cortex stimulation. B1: Excitatory response of the large aspiny neuron to substantia nigra stimulation. Dashed lines are extracellular controls. Scale in A1 also applies to A2. Arrowheads point to the onset of the stimulus. 128 50 pm Figure 31 129 (Fig. 31Al) and CK (Fig. 31A2 and 31A3). Since its somatic cross-sectional area is larger than 100 square um, it is tentatively identified as a member of the type III medium neurons as defined in the preceding Golgi study based on the appearance of its aspinous soma and poorly-branched dendrites. Electron Microsc0pic Analysis: Large Aspinous Neuron (Type I Large Neuron) The large aspinous neuron (Fig. 32B) has an eccentrically located nucleus. Since the nucleus appears to be densely filled with HRP reaction products, the distribution of the heterochromatin remains unclear. It is nevertheless possible to see many deep nuclear membrane invaginations (Fig. 32B arrows), a characteristic of large neostriatal neurons well noted by the present and previous authors (Mori, 1966; Adinolfi & Pappas, 1968; Kemp & Powell, 1971a; Fox et al., l97l/l972b, Pasik et al., 1976; Dimova et al., 1980; see also preceding electron microsc0pic anlaysis of Golgi material). The surrounding perikarial cytoplasm has not been obscured by the HRP reaction product, thus revealing abundant rough ER and free ribosomes distributed throughout the somatic and proximal dendritic cytoplasm. A very well developed Golgi complex occupies a perinuclear position, and is continuous with the widely distributed smooth ER within the soma. Numerous multivesicular bodies and dense bodies resembling lipofuscin granules are found in both soma and dendrites. Elongated mitochondria, some of which are branched, have loosely packed cristae. The proximal portion of the large primary dendrites of the large aspiny neurons contains all the organelles found in the Figure 32. 130 Photomicrograph and electronmicrographs of a type I large neuron labeled intracellularly with horseradish peroxidase. Photomicrograph of the large aspiny neuron, the arrow points to the axon. Inset shows an eletron micrograph of this axon taken at about 60 um away from the soma, note that it is thinly myelinated. CAL 8 40,um for photomicrograph and 1 um for inset. Electron micrograph of the soma of the cell shown in A. Note that its nucleus (N) is deeply invaginated (arrows) and that most of the dark HRP reaction products are restricted within the nuclear envelOpe. The large cytOplasmic space is full of organelles, including the perinuclear Golgi complex (G), numerous free ribosomes (ri) and stacks of rough endOplasmic reticulum (HER). CAL 8 2 um. 131 32 Figure Figure 33. A: D: 132 Electronmicrographs of dendrites of the intracellularly labeled type I large neuron. A distal dendrite with a vacuolated swelling (*) makes slightly asymmetrical synapse with a terminal containing medium to small vesicles (arrows). CAL 8 0.5 um. Varicose distal dendrite makes slightly asymmetrical synapses (arrows) with terminals containing medium to small vesicles. Note that the multivesiclular body (*) is present in the varicosity of the dendrite along with other organelles. CAL 8 0.5 um. The large primary denrite of the large aspiny neuron. Note the abundance of organelles in this dendritic process, and especially the series of Golgi cisterns (arrows) extending along its long axis. CAL 8 1 um. Lightly asymmetrical synapse (arrows) on a relatively straight portion of a distal dendrite. The preterminal contains medium to small vesicles. CAL 8 0.5,um. i.__—___.___‘__ - V Figure 33 134 perikarial cytoplasm (Fig. 33C), including series of longitudinally oriented Golgi cisterns which are continuous with the perinuclear Golgi complex. Distally, the dendrites are characterized by their irregular and aspinous appearance. Occasionally, the dendrites appear to lie directly adjacent to the somatic or dendritic membrane of another neuron, however, no membrane specializations have been observed in these cases. Ribosomes, mitochondria and multivesicular bodies are found within the intermittent varicose swellings (Fig. 33B). Vaculated swellings (Fig. 33A) are frequently observed, whether these are artifacts caused by tissue preparation or from a possible exocytotic discharge activity remains unclear. Since the surrounding neuropil has not been optimally preserved, no distinct differences could be determined with respect to the morphology of synapses on the soma and dendrites (i.e. the size and shape of the vesicles, or the symmetry of the synaptic density). Most of the preterminals contain medium vesicles, and make slightly asymmetrical synapses (Fig. 33A, 33B a 33). The axon of this large aspiny neuron becomes myelinated (Fig. 32A, inset) at a distance of 50 um from the soma. No collateral branches have been observed. The initial unmyelinated portion of this. axon contains bundles of both microtubules and neurofilaments and is contacted by a few terminals making symmetrical synapses. Medium Aspiny Neuron The medium aspiny neuron has a large and centrally located nucleus within a narrow rim of cytoplasm (Fig. 343). The nucleus is characterized by very deep nuclear membrane invaginations, some of which appear to bisect the long axis of the nucleus. The scanty Figure 34. 135 Photomicrograph and electronmicrographs a medium aspinous neuron intracellularly labeled with horseradish peroxidase. Photomicrograph of the medium aspiny neuron. CAL 8 10 um. Electronmicrograph of the same cell shown in A. Note that its nucleus has deeply invaginated nuclear membrane and the narrow rim of cytOplasm has many short mitochondria and a few distended rough endOplasmic reticulum. * marks the enlarged astroglial process. CAL 8 2,um. A dendrite of the HRP labeled medium aspiny neuron synapses with a terminal containing medium to small vesicles (arrow). Note that the dendrite also contains multivesicular body (*). CAL 8 0.5,um. An unlabeled neostriatal neuron which is similar to the HRP-labeled medium aspiny neuron in that it has a deeply invaginated nucleus and little of other organelles. Magnification is the same as in B. 136 .{‘ ~ 51-71: , my . Figure 34 137 perikarial cyt0plasm contains numerous short mitochondria with loosely arranged cristae. A small amount of rough ER has distended cisternae near the base of the dorsal pole of the soma. Free ribosomes, if any, are obscured by the granular HRP reaction product. A sparse smooth ER and a few small Golgi complexes are observed in the soma. Neurons with similar somatic morphology are rarely observed in normal neostriatal material. A comparable unlabeled neuron is shown in Figure 34D. The dendrites of the medium aspiny neuron (Fig. 34C) are smooth, and contain mitochondria, ribosomes and some multivesicular bodies. Axon terminals containing medium to small vesicles are seen contacting soma and dendrites (Fig. 340, arrow), however, a critical analysis of the synaptic relationship of this neuron is not possible because of the poor preservation of the surrounding neurOpil. No myelinated profiles containing HRP reaction product has been observed in the nearby neuropil and none of the labeled processes can be conclusively demonstrated as the axon of this cell. Light MicroscOpic Analysis of Intracellularly Labeled Type I Medium Neurons The most frequently encountered HRP labeled cell type in the rat neostriatum is the type I medium neuron (Fig. 35). These neurons respond with monosynaptic excitatory postsynaptic potentials to stimulation of cerebral cortex, substantia nigra and the intralaminar thalamus as has been described fully in previous intracellular recording and intracellular labeling studies in the cat (Kitai et al., 1976; Kocsis & Kitai, 1977; Kocsis et al., 1977; Kitai & Kocsis , 138 1978; Kitai et al., 1978) and the rat (Park et al., 1979; VanderMaelen et al., 1979; Preston et al., 1980). These neurons have a spine-free some (10-20 am in diameter) and 4 to 6 spine-free primary dendrites (Fig. 35A, 35C). The dendrites become densely covered with spines at a distance of approximately 20 um from the soma as they branch into secondary and tertiary segments. The majority of spines are pedunculated, with a round head on a stalk of variable length. Some sessile and some branched spines are also observed. The dendritic field of these medium spiny neurons covers a spherical domain with approximately 250 um radius. Although the somato-dendritic morphology of the type I medium neurons are more-or-less the same, the distribution pattern of their axons has been found to vary greatly. The axons of many of these neurons often give rise to 3 to 4 collateral branches which arborize entensively within the dendritic domain of the cells of origin (Fig. 35B). Less frequently, however, the intrinsic axon collaterals would ramify extensively outside the parent dendritic domain. An extreme example of this type of intrinsic axon distribution is shown in Figure 350. This neuron has a main axon which courses ventrally and medially and gives rise to six main collateral branches which branch infrequently and extend beyond its own dendritic domain. These collaterals occupy a much larger area than those of the previously described medium spiny neuron (compare 35B to 35C). The parent axon of this spiny neuron has not been followed out of the neostriatum, as the HRP reaction products within the axon faded from view as it coursed within a fascicle toward GP. igure 35. 139 Microscopic tracings of type I medium neurons labeled intracellularly with horseradish peroxidase. Somatic and dendritic profile of a typical type I medium neuron, * mark the emergence points of the axon collaterals. Axon collateral distribution pattern of the same cell shown in A. The dendrites are shown in dashed lines to demonstrate the relationship between the axon collaterals and the dendritic field of the same cell. A less frequently encountered type I medium neuron with long intrastriatal collaterals. Note the similarities of the soma and dendritic appearances of the these two medium spiny neurons (A d C) and the differnce of their axon collateral distribution patterns. 140 I... c > "r 4...- / .. .\ ‘ w» §.\111H 4., Qt Wfivm r 9‘ WAS.” a? 1 4‘ q xii-.1 / ...: Figure 35 141 In addition to the differences observed with respect to the distribution of their intrinsic collaterals, it is also posssible to subdivide the medium type I neurons into two groups based on the differences in their efferent axonal morphology (e.g. pattern of projection to GP). In both groups of neurons, the main axon emitted several local collaterals in the striatum to form a complex intrinsic arborization near the dendritic domain of the neuron of origin. In one group of cells (N 8 7), the main efferent axons were unbranched throughout their course in the neostriatum, and gave rise to either none or only one or two thin collateral fibers along their course in GP. These observatios are in agreement with a previous light microsc0pic study of strio-pallidal axons (Preston et al., 1980). The main axons did not end in GP, but were observed to course caudally into the internal capsule. Some of the collateral fibers in GP became beaded in appearance, resembling "boutons en passant", but no elaborate axonal arborizations were seen. HRP reaction product within the thin collateral fibers often faded quickly beyond detection a short distance from the main axons. In another group of neostriatal projection neurons (N 8 6), the main axons divided within the neostriatum into two or three primary branches and these branches descend in a parallel course toward GP (Fig. 36A & 40A). Branching occured either close to the soma, or took place within the fiber bundles of the internal capsule. Upon entering GP, these branches immediately gave rise to a very localized but highly elaborate and overlapping axonal arborization in the rostral GP (Fig. 36B & 40B) Some primary axonal branches ended in this area while others continued caudally and then fbrmed another 142 separate collateral tenninal plexus in the caudal GP. These caudal axon arborizations also overlapped extensively. In the case shown in Figure 36A, one primary branch still continued caudally beyond GP into the internal capsule, but the HRP labeling within this axon became too faint to be traced any farther. The intrinsic axonal collateral arborization in the neostriatum and the terminal arborizations in both the rostral and the caudal GP appear to be similar in size, being approximately 400 um in diameter. Although all neurons of the second group showed similar basic pattern of axonal arborizations, the locations of the terminal fields in GP appeared to vary with location of the neuron of origin in the neostriatum. That is, more medially placed neostriatal cells gave rise to more medially placed fields in GP, and dorsally placed neurons arborized more dorsally in GP. This result agrees well with previous studies on the topographical organization of the neostriatal efferent system, However, no indication of a rostro-caudal organization was obtained, since all adequately labeled neurons of this group showed both rostral and caudal terminal fields in GP. No apparent differences were observed with respect to the spatial distribution of the cell bodies of these two groups (i.e. those with branched main axons vs. those with unbranched main axons) of medium type I neurons in the neostriatum. Electron Microscopic Analysis of Intracellularly Labeled Type I Medium Neurons Some of the type I medium neurons injected with HRP were not suitable for ultrastructural analysis because large amounts of HRP Figure 36. A. 143 Microscopic tracings of the axons of a type I medium neuron which projects into the globus pallidus. The axon of a projection neuron is traced into the globus pallidus. The dendrites are omitted from the tracings to clarify the distribution of local intrinsic collaterals. Note the presence of two discrete terminal arborizations in the globus pallidus. CAL 8 50 um. A higher magnification view of the rostral terminal field in the globus pallidus. The three main axonal branches are shown as solid, dashed and dotted lines to demonstrate the extent of their overlaps. CAL 8 50 um. Condom- Putomen Pollidus Figure 36 144 145 reaction product obscured the organelle morphology. When the nucleus was not penetrated by the intracellular electrode, the nucleus of the spiny I cell was found to be round and pale, without any noticeable nuclear membrane invaginations. On the other hand, in cases where the injected HRP was mostly restricted to the nucleus, the cytOplasmic detail could be analyzed (Fig. 37A & 37B). The relatively small perikarial cytOplasmic space fermed a rim around the nucleus and contained abundant amount of free ribosomes, many in rosettes, and many mitochondria having densely packed cristae (Fig. 37D). The rough endoplasmic reticulum (ER) and smooth ER were not very numerous, and were rarely seen in stacked fermations. Few Golgi complexes were present, but some appeared to be distended, possibly as a result of the HRP injection. A number of microtubules and some dense bodies resembling lipofuscin granules were distributed within the cytOplasm. Some cells also had a cilium (Fig. 370) as had been observed by Rafols and Fox (1971/1972) in the monkey medium spiny neuron. Axosomatic synapses were rarely seen, but when found were usually symmetrical synapses (Fig. 37D). At least two types of axosomatic terminals can be distinguished by the size of their vesicles. One type had large vesicles (Fig. 38A ), the other had smaller vesicles (Fig. 38B). Subsurface cisterns next to the axosomatic synapses as described by Rafols & Fox (1971/1972) were rarely seen in our materials. Enlargement of astroglial processes around the soma of an injected cell was frequently observed (Fig. 37B) and this may reflect the physical damage caused by electrode penetration. 146 Figure 37. Electronmicrographs of the soma of a type I medium neuron intracellularly labeled with horseradish peroxidase. A: Photomicrograph showing the soma and primary dendrites of the same cell shown in B. Magnification is approximately 1/4 of that in B. B: Electronmicrograph of the soma of the same cell. The dark appearance of the nucleus (N) is probably due to accumulation of the HRP reaction products , within the nuclear envelope. The slightly distended appearancce of the Golgi complex (G) and the perineuronal astrocytic processes (AS) may have been resulted from the recording microeletrode. The cytOplasm contain little rough or smooth endOplasmic reticulum, but has many free ribosomes (ri) and mitochondria. Block arrow points to a cilium of this neuron also shown in C. Arrow points to areas of axosomatic synapses and are also shown in D. CAL 8 2 um. C: Higher magnification of the cilium (arrow). CAL 8 1 um. D: Higher magnification of the axosomatic synapses (arrows) with symmetrical synaptic densities at all three synapses and the preterminals contain large pleomorphic vesicles. Also note that the rough endOplasmic reticulum (R) is not in stacked formation, but is surrounded with many free ribosomes (ri) and mitochondria (M) in the soma. CAL 8 0.5 um. 147 ' turf-VJ" 3".” h. "'wn “8 ~' )3 W .*0 N ‘V ‘9. 1...), ; Jfl;_:l »~.’$,f ‘ Figure 37 148 The HRP labeled spiny distal dendrites were usually easily located in the neurOpil by the charateristic appearance of numerous dark spine appendages lined up along a dark dendritic shaft, especially when the dendrite was cut in longitudinal section (Fig. 38F). The primary dendrites, on the other hand, were smooth and were contacted mainly by preterminal containing large vesicles (Fig. 38D) making symmetrical synapses similar to those seen on the soma. Symmetrical axodendritic synapses were also occassionally seen on the dendritic shafts of the distal dendrites which bear spinous processes (Fig. 38E). However, the vast majority of the synapses on the distal dendrites were on the spines, which were contacted by preterminals containing medium round vesicles (Fig. 38G) and which probably originated from extrinsic sources as identified in the previous studies (Kemp & Powell, 1971b; Chung, Hassler & Wagner, 1978; Hassler, Chung, Rinne & Wagner, 1978; Hattori, Fibiger, McGeer & Maler, 1973). These terminals made asymmetrical synaptic contacts with the spines, which usually contained a spine apparatus. The dendritic shafts of the spiny I neurons were also characterized by their numerous microtubules and mitochondria (Fig. 38D, 38E, 38F & 38G). The axon initial segments of the type I medium neurons were contacted by terminals containing large pleomorphic vesicles (Fig. 380), and these ususally made symmetrical synapses. The intrinsic collaterals of the type I medium neurons in the neostriatum were unmyelinated, and made synaptic contacts of both "en passant" and ”tenminal" types. These boutons were usually filled with medium to large vesicles and appeared to make symmetrical contacts on dendritic shafts (Fig. 390) and possibly the base of the spine stalk (Fig. 39D). 149 Occasionally, the synaptic vesicles could also be found within the narrow part of the collateral fibers, but no presynatic specializations were observed in these cases. The terminals of the intrastriatal collateral fibers of the type I medium neurons which ramify mainly outside the parent dendritic domain (Fig. 35C) also contained medium to large vesicles and appeared to make symmetrical contacts on the dendritic shafts of the neighboring neurons (Fig. 39E and 39F). The main efferent axons of the type I medium neurons generally were thinly myelinated throughout their trajectory in the neostriatum and or (Fig. 39A & 393), and the collateral fibers within the GP were unmyelinated. Some of these neostriatal projection neurons responded antidromically following stimulation of the substantia nigra (SN), a known neostriatal target area distal to GP. These axons may thus represent the anatomical substrate fer the inhibitory neostriatal axons projecting to both GP and SN. However, the branched efferent axons as shown in Figure 36A and 40A were thinly myelinated befbre branching in the neostriatum, but were unmyelinated subsequent to branching, both in neostriatum and in GP. The neostriatal efferent plexus in GP consisted of thin unmyelinated fibers and boutons of both the "en passant" and "terminaux" types (Fig. 36B & 40B). These boutons, containing large and moderately pleomorphic vesicles (Fig. 400), formed symmetrical synaptic junctions with pallidal dendrites of all sizes, including the occasional dendritic spines found on some dendrites (Fig. 41). Most of the synapses were fermed onto the small to medium sized dendrites which predominate in the neurOpil of GP. The ultrastructural 150 morphology of these HRP labeled boutons in GP appeared to be identical to the labeled intrinsic terminals of medium type I neurons in the neostriatum. Analysis of serial thin sections through individual boutons demonstrated that every labeled bouton formed only one synapse with one postsynaptic element. On the other hand, two or more labeled boutons arising from the same axon collateral were often seen contacting the same pallidal dendrite, thus confirming a degree of the "longitudinal axo-dendritic" organization of the strio-pallidal connections as postulated from light microsc0pic analysis of Golgi impregnated materials (Fox & Rafols, 1975, 1976). However, even in areas of their highest density as shown in light microscopic reconstructed drawings, the labeled boutons represented only a minute fraction of the total number of boutons with similar strio—pallidal morphology, indicating a high degree of convergence of neostriatal axons upon pallidal neurons. Figure 38. 13100 151 Electronmicrographs of synapses on soma, dendrites and axon initial segment of a type I medium neuron intracellularly labeled with horseradish peroxidase. A, B, C, D & E are at the same magnification. CAL in E 8 0.2 um. B: Axosomatic synapses (arrows) on the cell shown in Figure 37 showing the difference in synaptic vesicle sizes of these two terminals. Synapse (arrow) on the axon initial segment (is). Synapse (arrow) on the primary dendrite (d). Synapse (arrow) on the dendritic shaft of a spine-bearing distal dendrite(d). Emergence of the spine (3) is shown at the lower right hand corner. Low magnification view of a longitudinally sectioned spiny distal dendrite darkly labeled with HRP. CAL 8 1 um. Higher magnification of a lightly labeled spiny dendrite (d) showing the spine (3) containing stacks of spine apparatus. The axospinous synapses (arrows) are asymmetrical and the preterminal contains medium vesicles. For comparison, the blocked arrow points to another axospinous synapse on an unlabeled spine head. The arrow head points to an axodendritic contact which is out of the plane of this section. CAL 8 0.25 am. 152 Figure 38 Figure 39. 153 Electronmicrographs of the axons and intrinsic terminals of a type I medium neuron intracellularly labeled with horseradish peroxidase. All at the same magnification, CAL 8 0.5 pm. The myelinated axon of a type I medium neuron, within the neuropil of the neostriatum. The myelinated axon of a type I medium neuron in the fiber bundles at the lateral part of the globus pallidus. Local axon collateral of a type I neuron makes an axodendritic synapse (arrow) on the dendritic shaft (d) of a neighboring neostriatal neuron. Note that the synaptic complex is symmetrical, and the preterminal contains large oval vesicles. Another axodendritic synapse by the axon collateral of a type I medium neuron, the synapse appears to be at the base of a spine neck (arrow) of a dendrite (d). The block arrow points to the unmyelinated axon collateral fiber. E & F: HRP labeled axon collateral terminals of the type I neuron shown in Figure 35C. Arrows point to the symmetrical contacts with the dendritic shafts (d) of neighboring neurons. 154 Figure 39 155 Figure 40. Photo- and electronmicrographs of the efferent axon of a type I medium neuron and its terminal arborization in the globus pallidus. A: Photomicrograph of the first branching of the main axon of the neuron shown in Figure 36A. Although the branch forms at right angles to the parent axon, it immediately turns to take a parallel course toward the globus pallidus. The varicosity near the branch point is not presynaptic. CAL 8 2 um. B: Photomicrograph of a portion of the rostral terminal arborization in the globus pallidus shown in Figure 36B. Arrows indicate the positions of electronmicros00pically identified presynaptic terminals such as shown in C. Same magnification as in A. C: Electronmicrograph of a presynaptic bouton from an intracellularly labeled axon of a strio-pallidal neuron. The synaptic junctional complex appears to be symmetrical. The postsynaptic target is a small pallidal dendrite which also receives other unlabeled boutons, some of which (Open arrows) exhibit morphology similar to that of the labeled terminal, while others (closed arrow) contain smaller vesicles and form asymmetrical synapses. CAL 8 0.5 um. 156 Figure 40 Figure 41. 157 The distribution of the diameters of the pallidal dendrites which are postsynaptic to a neostriatal type I medium neuron. Dendritic diameters were measured from single sections at the site of contacts with the labeled axon terminals, and for obliquely sectioned dendrites were taken perpendicular to the axis of elongation as indicated by the orientation of the microtubules. Most synapses are formed on small dendritic shafts less than 2 pm in diameters, although all areas of somato-dendritic surface of pallidal neurons can receive strio-pallidal synapses. 158 20'} 7 _1 '6. U) 0) F1 10 '1 a c: 12« > (I) d u— <> 9‘ C5 4 1 Z 4- ‘ rfln r1 .-LCL. 1. . . . . . ...rrlrthm SPINE . .7 .9 LI L3 LS LT l.9 2.| 2.3 2.5 2.7 2.9 3.1 SOMA Figure 41 Dendritic Diameter (um) 159 E. Somatic Size of Rat Neostriatal Neurons Although somatic size has often been used as the main criteria for preliminary classification of neostriatal neurons in many clinical as well as eXperimental studies, analysis of the somatic size distribution of different types of neostriatal neurons as defined in this study reveals that somatic size alone is not a reliable parameter for predicting the dendritic or axonal morphology of any individual neostriatal neuron. Since the capriciousness of the Golgi impregnation prevents quantitative analysis of the population of different types of neurons within any nucleus, a distribution histogram showing the somatic size of all neostriatal neurons impregnated in our Golgi material would be skewed unavoidably to that of the type I medium neurons, the most frequently impregnated neurons in our preparation. Thus the somatic size of type I medium neurons (Fig. 42B) have been plotted separately from that of all other types of neostriatal neurons (Fig. 42C). The somatic size distribution of the neostriatal projection neurons labeled retrogradely with HRP extracellularly injected in the globus pallidus is also shown in Figure 42A. These histograms indicate that there is considerable overlap between the somatic size of the projection neurons with that of the type I medium neurons. However, the somata of many non-type I medium neurons also are within the size range of the projection neurons. Since the light microscope could not resolve adequately the distribution of dendritic spines of the HRP back-filled projection neurons, as mentioned earlier, it is premature to rule out the other medium neurons as projection neurons. Most large neurons, however, are probably not projection neuron (compare Fig. 42A and 42C). 160 Figure 42. The distribution of somatic sizes of rat neostriatal projections neurons labeled retrogradely with horseradish peroxidase extracellularly injected in the globus pallidus, and the somatic sizes of neostriatal neurons impregnated by the rapid Golgi method. A. Distribution of somatic size of neostriatal projection neurons labeled retrogradely from HRP injections in the globus pallidus. The tissue sections have been dehydrated in alcohol and cleared in xylene. The ordinate scale is relative, total cells 8 68. B. Distribution of somatic size of type I medium neurons impregnated by the rapid Golgi method. The tissue sections have also been dehydrated in alcohol and cleared in xylene. The ordinate is relative, total cells measured 8 100. C. Distribution of somatic size of rapid Golgi impregnated neostriatal neurons which are not type I medium neurons. The tissue sections have been cleared in glycerin. Total number of cell analyzed 8 69. , A Neostriatal Projection Neurons B 4 Medium Type I Neurons C IO Neostriatal Neurons (not including Medium Type 1 Neurons) 8 S u z n. o 3‘; 5 E 100 200 300 400 500 2 SOHATIC CROSS-SECTIONAL AREAS, 1n um Figure 42 600 Chapter 4. DISSCUSSION A. Comparison of Classification Systems Naming the rat neostriatal neuron types as type I, II and so on in this study has elimated the ambiguous and often misleading terms such as "stellate", "spidery", "aspiny", "less-spiny", "smooth" and so on in the designation of neuron types which existed in most of the previous Golgi studies (e.g. Mensah & Deadwyler, 1974; Lu & Brown, 1977; Danner & Pfister, 1979a, b). Since most of these reports provided only free-hand drawings (Mensah & Deadwyler, 1974; Lu & Brown, 1977) or partial photomicrographs (Danner & Pfister, 1979a, b), their published results were inevitably inadequate for comparison with each other and with those reports on other species which provided detailed camera lucida drawings (Kemp & Powell, 1971a, DiFiglia et al., 1976; Tanaka, 1980). A recent Golgi study of the rat neostriatal neurons has classified neurons into type I, II III and so on with accompanying camera lucida tracings (Dimova et al., 1980). However, no photomicrographs were provided in that particular study and the morphology of the axons was not described. The present study, therefore, extends the findings of Dimova et a1. (1980) by providing both photomicrographic and camera lucida illustrations of different types of rat neostriatal neurons. Additionally, the ranges of quantitative values for several morphological parameters (e.g. somatic cross-sectional areas, number of dendritic stems vs. dendritic tips, etc.) have been described in the text. Since several new cell types as well as the axon morphology of various types of neurons have been found in this study, I have modified the classification system 162 163 provided by Dimova et al. (1980) in order to accomodate these new observations. Idealy, anatomical classification of neurons should be based on the morphological characteristics of all parts of a neuron. However, since the axons are often incompletely impregnated in Golgi preparations, the somata-dendritic features have been the most important and reliable criteria used in analysis of neuronal morphology in the present and most of the other Golgi studies. Detailed somato-dendritic morphological descriptions are especially important fer interpretating the results of the recently developed anatomical techniques, such as immunocytochemistry or retrograde labelings, which enable partial labeling of somata and dendrites. The axonal distribution patterns of indidvidual neurons have been analyzed whenever possible with the understanding that neurons with similar somato-dendritic morphology may have different axonal distribution patterns as illustrated by the type I medium neurons labeled intracellularly with horseradish peroxidase. A summary of the morphology of the different types of rat neostriatal neurons observed in this study is listed in Table 1. As has been mentioned earlier in the INTRODUCTION, previous Golgi studies in other species have used rather descriptive terms for classifying the neostriatal neurons. Thus, found in the cat are medium spiny neurons with short axons, large and medium neurons with long axon, medium neurons with varicose or smooth dendrites and so on (Kemp a Powell, 1971a). However, since the presence of long axons has never been clearly demonstrated in that study, little distinction can be made between the medium long-axon cells and the medium smooth cells 164 . a: 00. meg» mmoa um n N .Hmfiuowme coaonma manuasaaoomsuaw so copmcmopmefi HmHou poguwo mo mwmhamcm oaeoomOhOHs :oupooao use unwed Anon so eommn one mama sew .maom on» scum harm 5: ON ems» mama mopfipvcmc omega Ham menace“ N a: 00m a co. accuses 1: a N a§.oon can» somata "A "mmeHmnzmn A .nmszmINAmmm1zoz ammoz1zoz amzozamm-»qmoom .amszm1summmamm .mADmm oz .mmeHmmzmm nmszmuwammmaam .saq mm was . mmBH ~5sz 020A sz Hm mlwqmmmedm A¢9WHQ szHmmIMAmmmdmm no Faoozm szHmmINAmwmH HHH HH SDHQWS HH 165 based on their somato-dendritic morphology. Moreover, since the medium spiny neurons have been shown to be projection neurons, the "short-axon" title is no longer applicable. Fox and his colleagues (1971/1972a, b, 1974) described the presence of spiny neurons and large aspiny neurons, spidery (medium) aspiny neurons and aspiny neurons with somatic spines in the monkey neostriatum. However, what exactly constitutes spidery remains unclear. A recent review by Rafols & Fox (1979) reclassified the monkey neostriatal neurons into six types: 1) medium neurons with many dendritic spines, 2) medium neurons with fewer dendritic spines, 3) large neurons with many dendrites, 4) large neurons with fewer dendrites, 5) small aspiny neurons, and 6) medium neurons with radiating beaded dendrites. Although this division of neurons based on their somato-dendritic morphology is similar to our present approach, naming each type with a descriptive title has the aforementioned disadvantage of ambiguity and can be misleading. For example, the "small aspiny neurons" included subgroups of neurons with medium somata, as well as neurons with either smooth or varicose dendrites, or dendrites with long-stalked beaded appendages and axons ' indicative of "local circuit neurons" (Rafols a Fox, 1979). However, since the medium neurons with radiating beaded dendrites also have varicose dendrites and locally arborizing axons, the distinction between these two types of cells becomes unclear. On the other hand, the Pasiks and DiFiglia classified the monkey neostriatal neurons into Spiny I, Spiny II, Aspiny I, Aspiny II and Aspiny III cells, and they considered the Aspiny cells to be intrinsic cells and the Spiny I and II cells to be projcetion neurons 166 (DiFiglia et al., 1976; Pasik et al., 1979). However, since spines or dendritic appendages have been found on both Aspiny II and Aspiny III cells (Pasik et al., 1979), are these, therefore, Aspiny cells with spines? Since Fox’s large aspiny cell is considered to be a large Spiny II cell (Pasik et al., 1979), then it must be a Spiny II cell without spines! By their classification criteria, then, if a sparsely-spined neuron had a long axon, it would be a Spiny II cell, otherwise it would be an Aspiny III cell. However, no definition has been provided for "long axon" other than the unwritten understanding that it is synonymous with efferent axon. Additionally, neostriatal efferent axons with accompanying somata and dendrites have never been clearly demonstrated in a Golgi study. The above discussion demonstrates that most descriptive terms are not suitable fer naming neuron types. On the other hand, since the somatic size is a measurable parameter and can be easily correlated with histological studies from different laboratories, the descriptive terms such as "large", "medium" and "small" have been used in the present study and their values have been clearly defined in order to better correlate with previous histological studies of the neostriatum. Since these definitions have artificial boundaries, several neurons with somatic sizes at the boundary regions have been found in this study. The classification of these neurons would then rely more on their other morphological characteristics. The relationships of the present neuronal classification to those of previous Golgi studies is summarized in Table 2. In addition to establishing the morphological characteristics of different neuron types from the Golgi material, both extracellularly and 167 intracellularly labeled neurons have been included in this study in order to determine the morphology of the neostriatal projection neurons and to analyze the morphology of neurons which have been characterized physiologically by intracellular recordings. Further detailed comparisons of these observations are discussed below. 168 oVPme 66—“ Q ewmbm ..\ 950— SCH.“ @hw mQOCQHOMQH .Hm no xom w . crop HHH HH H HH scaam HH .Hn so w 8H hcfimm< wH hewmm< hefigm< hcfimm hcfimm omsmg hewmm< mflamflmwn moafinccov mosfimm omoofium> oflpfiuczoe mocfimm mum? .xoh mnfiumfiemu hcflam< soM\3 owumEOm\3 heflmm< m mHomem w \3 suave: huocwmm suave: hcflmms hcwmm w owned #.Hm as non sHH scam HH scaam suave: a suave: w eH sH HHH a HH sasam HH one. m hswmm< hswmm< hefimm< w H hewmm omumA hswmm< excess mopfisesoc wsoxm meoaxs no pump omeoaum>\: “mouwpecoc spoosm hqwmm w omsma daemon e assess a \s asses: c asses: a case mopwsecoc macaw msoa chm. omoowum>\3 no spooem opmaaoum w owns; uoahsemon wHHmsm esflcos w suave: w suave: a gonzo: “cows wcoq mopfiuecoo «caveamm omoOfinm>\3 moufluvnoe mcoa «huoefimm bum. waaosm suave: w \3 suave: w asses: w owned ssosm a :4 em a n .mmbm. a as case em easy N easy a _ case s n case success a eczema we a ommp .Hm as e n case em mass m case a . case a m easy assess AA >H HHH HH H HH H stnmz mumonm nus: coaumowmfimmmao peommhm mo confinemeoo wmxzo: com e