I N T E G R A T E D SYSTEMS OF THE CNS PART III
This Page Intentionally Left Blank
HANDBOOK OF CHEMICAL NEUROANATOM...
25 downloads
515 Views
49MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
I N T E G R A T E D SYSTEMS OF THE CNS PART III
This Page Intentionally Left Blank
HANDBOOK OF CHEMICAL NEUROANATOMY Series Editors" A. Bj6rklund and T. H6kfelt
Volume 12
INTEGRATED SYSTEMS OF THE CNS, PART III
Cerebellum, Basal Ganglia, Olfactory System Editors."
L.W. S W A N S O N Department of Biological Sciences, University of Southern California, Los Angeles, CA, U.S.A. oo
A. B J O R K L U N D Department of Medical Cell Research, University of Lund, Lund, Sweden to
T. H O K F E L T Department of Neuroscience, Histology, Karolinska Institute, Stockholm, Sweden
1996
ELSEVIER A m s t e r d a m - Lausanne - New York - O x f o r d - S h a n n o n - Tokyo
9 1996 Elsevier Science B.V. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the Publisher, Elsevier Science B.V., Copyright and Permissions Department, P.O. Box 521, 1000 AM Amsterdam, The Netherlands. No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein. Because of the rapid advances in the medical sciences, the Publisher recommends that independent verification of diagnoses and drug dosages should be made.
Special regulations for readers in the USA. This publication has been registered with the Copyright Clearance Center, Inc. (CCC), 222 Rosewood Drive, Danvers, MA 01923. Information can be obtained from the CCC about conditions under which photocopies of parts of this publication may be made in the USA. All other copyright questions, including photocopying outside the USA, should be referred to the Publisher.
ISBN 0-444-82451-0 (volume) ISBN 0-444-90340-2 (series) This book is printed on acid-free paper.
Published by: Elsevier Science B.V. RO. Box 211 1000 AE Amsterdam The Netherlands
Printed in The Netherlands
Dedicated to J/mos Szentfigothai and Walle J.H. Nauta
This Page Intentionally Left Blank
List of contributors MATTHEW ENNIS Department of Anatomy The University of Maryland School of Medicine Baltimore, MD 21201 U.S.A.
MICHAEL T. SHIPLEY Department of Anatomy The University of Maryland School of Medicine Baltimore, MD 21201 U.S.A.
CHARLES R. GERFEN Laboratory of Systems Neuroscience National Institute of Mental Health Bldg 36 Room 2D-10 Bethesda, MD 20892 U.S.A.
J. V O O G D
D. JAARSMA Department of Anatomy Erasmus University Medical Center RO. Box 1738 3000 DR Rotterdam The Netherlands
CHARLES J. WILSON Department of Anatomy and Neurobiology University of Tennessee School of Medicine Memphis, TN U.S.A.
E. MARANI Department of Physiology Leiden University Rijnsburgerweg 10 2300 RC Leiden The Netherlands
Department of Anatomy Erasmus University Medical Center P.O. Box 1738 3000 DR Rotterdam The Netherlands
LEE A. ZIMMER Department of Anatomy The University of Maryland School of Medicine Baltimore, MD 21201 U.S.A.
JOHN H. MCLEAN Division of Basic Medical Sciences Memorial University of Newfoundland St. John's, Newfoundland Canada A 1B 3V6
vii
This Page Intentionally Left Blank
Preface It is with a mixture of pleasure and sadness that we dedicate this third volume of the
Integrated Systems series of the Handbook of Chemical Neuroanatomy to the memory of two outstanding structural neuroscientists, J~mos Szentfigothai and Walle J.H. Nauta, who are widely regarded as having led the Romantic School of neuroanatomy through the Twentieth Century. Szentfigothai was born on October 31, 1912, in Budapest, and passed away on September 8, 1994 in his native city. He was a student of Cajal's friend von Lenhoss6k, and like Cajal made enduring contributions to our understanding of many components of the nervous system, including (roughly in chronological order) the autonomic system, spinal cord, vestibulo-ocular and stretch reflex circuitry, neuroendocrine system, cerebellum, thalamus, and cerebral cortex. What sets his work apart from many of his contemporaries was the ability to generalize sensibly. This led, for example, to the concepts of synaptic glomeruli and neuronal modules, and to the synthesis for which he will always be remembered, The Cerebellum as a Neuronal Machine, published in 1967 with his collaborators John Eccles and Masao Ito. Nauta was born on June 8, 1916 in Medan, Indonesia; received the M.D. and Ph.D. degrees at the University of Utrecht; served the last 30 years of his career at the Massachusetts Institute of Technology; and died on March 24, 1994. He perhaps will be remembered longest for the 'Nauta method', the first selective silver impregnation technique for degenerating axons. It was introduced in 1950 and variants were the method of choice for tracing axonal connections for about 25 years, until the use of more sensitive intraaxonal transport techniques became widespread. However, Nauta was a brilliant writer and an inspiring lecturer; and he published very influential experimental analyses of many forebrain systems in a variety of mammals. The limbic system and basal ganglia were his specialties, and indeed his work with Mehler on the lentiform nucleus of the cat and monkey was the first paper published in Brain Research (I :3-42, 1966) and is a classic with regard to both style and content. We are profoundly grateful to the authors who have committed so much time and thoughtfulness to the chapters in the third part of the Integrated Systems component of the Handbook. When planning began in 1983, we had hoped to review each of the major sensory and motor systems, along with parts of the broader system that controls motivated and emotional behavior. Furthermore, each chapter was to be written from a dual perspective- a classical functional neuroanatomical overview, combined with what has been learned more recently about neurotransmitters and receptors within the circuitry. For the usual reasons familiar to editors, all of the planned chapters were not written, and it proved impossible to devote single volumes to an internally consistent theme. Nevertheless, the series as a whole does survey the major sensory systems (retina by Ehinger and Dowling, part I; central visualpathways by Parnavelas, Dinopoulos, and Davies, part II; auditory system by Aitkin, part II; somatosensory system by Rustioni and Weinberg, part II; gustatory and related chemosensory systems by Kruger and Mantyh, part II; and olfactory system by Shipley, McLean, Zimmer, and Ennis, part III); two important parts of the motor system (cerebellum by Voogd, Jaarsma, and Marani, part III; basal ganglia by Gerfen and Wilson, part III); and three key parts of
ix
the limbic system (hypothalamusby Swanson, part I; amygdalaby Price, Russchen, and Amaral, part I; hippocampusby Swanson, K6hler, and Bj6rklund, part I). The literature in the field as a whole continues to explode. Keeping pace is a challenge that we hope will be facilitated by the imminent revolutions in electronic publishing, database management, and computer graphics. Los Angeles, Lund and Stockholm in June 1995 LARRY W. SWANSON
ANDERS BJORKLUND
TOMAS HOKFELT
Contents THE CEREBELLUM, CHEMOARCHITECTURE AND A N A T O M Y J. VOOGD, D. JAARSMA AND E. MARANI ~
2. 3.
Introduction Cytology of the cerebellar cortex Chemical anatomy of the cerebellar cortex 3.1. Purkinje cells 3.1.1. Gamma-aminobutyric acid (GABA), glutamic acid decarboxylase (GAD) and the GABA-transporters in Purkinje cells 3.1.2. Motilin and taurine in Purkinje cells 3.1.3. Calcitonin gene-related peptide (CGRP), acetylcholinesterase (ACHE), somatostatin and tyrosine hydroxylase in Purkinje cells 3.1.4. The localization of the IP3 receptor and the intracellular calcium stores of Purkinje cells 3.1.5. Protein kinase C in Purkinje cells 3.1.6. cGMP, cGMP-dependent protein kinase and nitric oxide synthase in Purkinje cells 3.1.7. Calcium-binding proteins in Purkinje cells 3.1.8. Other specific biochemical markers for Purkinje cells 3.1.9. Cytoskeleton and metabolism of Purkinje cells 3.1.10. Nerve growth factor and nerve growth factor-receptor protein in Purkinje cells 3.1.11. Immunoreactivity of Purkinje cells in paraneoplastic diseases 3.2 Excitatory pathways 3.2.1. Mossy fibers 3.2.2. Climbing fibers 3.2.3. Granule cells and parallel fibers 3.3. Localization of glutamate receptors 3.3.1 Ionotropic glutamate receptors 3.3.2. Metabotropic glutamate receptors 3.4. Nitric oxide: the cerebellar localization of nitric oxide synthase, guanylate cyclase and cyclic GMP 3.5. Adenosine, 5'-nucleotidase and adenosine desaminase 3.6. Interneurons of the cerebellar cortex 3.6.1 Stellate and basket cells 3.6.2. Golgi cells and Lugaro cells 3.6.3. Unipolar brush cells 3.7. Localization of GABA receptors and glycine receptors 3.7.1. GABAA receptors
1 1
17 17
17 21
23 24 32 34 36 38 43 44 47 49 51 55 57 60 60 72 76 77 81 84 85 89 93 93
xi
4. 5.
6.
xii
3.7.2. GABAB receptors 3.7.3. Glycine receptors 3.8. Monoaminergic afferent systems and receptors 3.9. Hypothalamocerebellar connections and histaminergic projections 3.10. Cholinergic systems and acetylcholinesterase (ACHE) in the cerebellum 3.10.1. Distribution of choline acetyltransferase 3.10.2. Cholinergic receptors 3.10.3. Acetylcholinesterase 3.11. Neuroglia Gross anatomy of the mammalian cerebellum The cerebellar nuclei 5.1. Subdivision of the cerebellar nuclei 5.1.1. The cerebellar nuclei of the cat 5.1.2. The cerebellar nuclei of primates 5.1.3. The cerebellar nuclei of the rat 5.2. The GABAergic nucleo-olivary projection neurons of the cerebellar nuclei 5.3. Nucleocortical and intrinsic neurons of the cerebellar nuclei 5.4. Non-GABAergic projection neurons of the cerebellar nuclei 5.5. Afferent connections of the cerebellar nuclei: Purkinje cell axons 5.6. Extracerebellar afferents of the cerebellar nuclei: collaterals of mossy and climbing fibers 5.7. Extracerebellar afferents of the cerebellar nuclei: serotoninergic, noradrenergic, dopaminergic and peptidergic projections Efferent and afferent connections of the cerebellar cortex: corticonuclear, olivocerebellar and mossy fiber connections and cytochemical maps 6.1. Compartments and corticonuclear projection zones: Correlations with cytochemical maps 6.1.1. Corticonuclear projection zones in the cat. Correlation with white matter compartments and cytochemical zones 6.1.2. Compartments and corticonuclear projection zones in monkeys 6.1.3. Parasagittal zonation in the cerebellar cortex: Antigenic compartmentation for Zebrin and other markers 6.1.4. The corticonuclear projection of the cerebellum of the rat. Correlations with Zebrin-antigenic compartmentalization 6.1.5. The corticovestibular and corticonuclear projections of the flocculus and the caudal vermis. Correlations with cytochemical zones and compartments 6.2, Regional differences in the development of the cerebellum 6.3. The organization of the olivocerebellar projection 6.3.1. Configuration and ultrastructure of the inferior olive 6.3.2. Afferent connections of the inferior olive
100 101 102 111 113 113 121 127 128 133 138 140 146 148 151 154 158 160 164
165 167
170 177
177 184 189
201
207 217 225 225 233
6.3.3. 6.3.4. 6.4.
Mossy 6.4.1. 6.4.2. 6.4.3. 6.4.4.
.
,
9.
The connections between the inferior olive and the cerebellum The distribution of peptides and calcium binding proteins in climbing fibers and cells of the inferior olive fiber systems Concentric and discontinuous, lobular arrangement of mossy fiber systems Zonal arrangement in the termination of mossy fibers: Correlations with cytochemical maps The somatotopical organization in mossy fiber pathways Collateral projections of mossy fiber systems to the cerebellar nuclei. The nuclear projection of the red nucleus The chemoarchitecture of mossy fibers
6.4.5. Postscript 7.1. Biochemical correlates of cell types and fiber systems 7.2. Neurotransmitters and their receptors 7.3. Lobules and zones 7.4. The role of biochemically defined systems in cerebellar motor control Acknowledgements References
242 275 284 284 293 299
302 303 305 305 307 307 309 310 311
II. THE BASAL G A N G L I A - C.R. GERFEN AND C.J. WILSON 1. 2. 3. 4.
5.
6.
7.
Introduction Organizational overview 2.1. Comparisons between rodents and primates Cerebral cortex input to striatum Striatum 4.1. Spiny projection neuron 4.1.1. Cortical input 4.1.2. Thalamic input 4.1.3. Nigrostriatal dopamine input 4.1.4. Spiny cell local collaterals inputs (GABA and peptide) 4.1.5. Cholinergic input 4.1.6. Striatal GABA interneuron inputs 4.1.7. Somatostatin interneuron inputs 4.1.8. Other inputs 4.2. Striatal interneurons 4.2.1. Cholinergic neurons Globus pallidus (external segment) 5.1. Synaptic input 5.2. Output Subthalamic nucleus 6.1. Synaptic input 6.2. Output Substantia nigra/entopeduncular nucleus
371 372 376 377 379 380 382 382 386 388 389 389 389 390 390 394 396 397 399 400 400 402 402
xiii
10.
11.
12. 13.
7.1. Synaptic input to pars reticulata neurons 7.2. Synaptic input to pars compacta neurons 7.3. Projections of pars reticulata neurons Connectional organization of basal ganglia Relationship between cortex and basal ganglia 9.1. Topographic organization 9.2. Overlap of inputs: cortico-cortical organization 9.3. Striatal output systems: topography/convergence/divergence 9.4. Striatal outputs in relation to nigral outputs: dual output systems 9.5. Summary of organization of cortico-basal ganglia circuits Striatal patch/matrix compartments 10.1. Nigrostriatal dopamine system 10.2. Striatal outputs 10.3. Cortical inputs 10.4. Thalamic afferents 10.5. General patch-matrix organization 10.6. Cortical organization related to striatal patch-matrix compartments Direct/indirect striatal output systems 11.1. Connectional basis 11.2. Peptide basis 11.3. Dopamine receptor-mediated regulation 11.4. Other (non-dopaminergic) regulatory receptor systems in striatum 11.5. Cellular interactions within the striatum 11.6. Functional significance 11.7. Regional differences Acknowledgements References
403 404 407 409 409 410 413 418 421 425 426 427 429 431 435 435 437 439 439 443 447 449 451 453 455 457 457
III. THE OLFACTORY S Y S T E M - M.T. SHIPLEY, J.H. MCLEAN, L.A. ZIMMER AND M. ENNIS 1.
2.
xiv
469 Introduction 470 1.1. The olfactory epithelium 473 1.2. Two olfactory systems 473 1.3. Human diseases and the olfactory system 474 The main olfactory bulb 474 2.1. Laminar organization 474 2.1.1. Olfactory nerve layer 475 2.1.2. Glomerular layer 486 2.1.3. External plexiform layer 488 2.1.4. Mitral cell layer 490 2.1.5. Internal plexiform layer 491 2.1.6. Granule cell layer 2.1.7. Mitral-granule cell interactions: Anatomical considerations 492 493 2.1.8. Subependymal zone 493 2.2. Transmitter receptors in the MOB 493 2.2.1. Excitatory amino acids (EAAs) 493 2.2.2. GABA receptors
2.3.
Influence of the olfactory nerve on transmitter expression in MOB neurons 2.4. Functional organization of the MOB 2.4.1. Organization of olfactory nerve inputs to MOB 2.4.2. Broad topographic mapping 2.4.3. Neural processing in the glomerular layer 2.4.4. The mitral/granule cell inhibitory system 2.4.5. Glomerular versus infraglomerular inhibition 2.5. Outputs of the MOB 2.5.1. Intrabulbar collaterals 2.5.2. Mitral/tufted cell projections beyond the MOB 2.5.3. Projections to olfactory cortex 2.5.4. Transmitter(s) mediating MOB to PC monosynaptic excitation 2.6. Centrifugal afferents to MOB 3. Primary olfactory cortex 3.1. Anterior olfactory nucleus (AON) 3.1.1. Architecture of AON 3.1.2. Inputs to AON 3.1.3. Outputs of AON 3.1.4. Organization of AON circuitry 3.1.5. Transmitters of AON 3.1.6. Transmitter receptors in AON 3.1.7. Functions of AON 3.2. Rostral olfactory cortex 3.2.1. Indusium griseum 3.2.2. Anterior hippocampal continuation 3.2.3. Taenia tecta 3.2.4. Infralimbic cortex 3.2.5. Olfactory tubercle 3.2.6. Nucleus of the lateral olfactory tract (NLOT) 3.3. Lateral olfactory cortex 3.3.1. Architecture of the lateral olfactory cortex 3.3.2. Neuron types in the piriform cortex 3.3.3. Connections of the lateral olfactory cortex 3.3.4. Transmitter receptors in the lateral olfactory cortex 3.3.5. Piriform cortex is a seizurogenic focus 3.3.6. Modeling of olfactory network function 4. Integration of the main olfactory system with other functions 4.1. Odors and cognition 4.2. Olfaction and taste/visceral integration 4.3. Olfaction and motor activity 4.4. Olfaction and memory 5. The accessory olfactory system 5.1. Accessory olfactory bulb 5.1.1. Neurotransmitters in the AOB 5.1.2. Transmitter receptors in the AOB 5.1.3. Outputs of the AOB 5.1.4. Centrifugal afferents to AOB
493 496 496 496 498 501 503 504 504 504 505 506 507 507 509 509 509 509 510 514 514 515 516 516 516 516 518 518 519 519 519 522 524 529 529 532 532 532 534 534 536 536 536 537 538 539 539
XV
5.2.
,
7. 8. 9.
Higher order connections of the accessory olfactory system and reproductive functions 539 541 5.3. Sexual dimorphism of AOB and its target structures 541 'Non-olfactory' modulatory inputs to the olfactory system 541 6.1. Cholinergic innervation of the olfactory system 541 6.1.1. Cholinergic inputs to the MOB 544 6.1.2. Cholinergic inputs to the piriform cortex 546 6.2. Noradrenergic (NE) innervation of the olfactory system 546 6.2.1. NE innervation of the olfactory bulb 548 6.2.2. NE inputs to the piriform cortex 550 6.3. Serotonin (5-HT) innervation of the olfactory system 6.3.1. 5-HT innervation of the MOB 550 551 6.3.2. 5-HT inputs to the piriform cortex 553 6.4. Dopamine (DA) innervation of the olfactory system 6.4.1. Dopamine (DA) innervation of the piriform cortex 553 Comparison of NE, 5-HT and DA inputs in the rat piriform cortex 553 6.5. 553 6.6. Differential innervation of MOB and AOB 555 Acknowledgments 555 Abbreviations 556 References
SUBJECT INDEX
xvi
575
CHAPTER I
The cerebellum: chemoarchitecture and anatomy J. VOOGD, D. JAARSMA AND E. MARANI
......... but the Spirits inhabiting the Cerebel perform unperceivedly and silently their Work of Nature without our Knowledge or Care. Thomas Willis. Of the Anatomy of the Brain. Englished by Samual Pordage, Esquire, London. Printed for Dring, Harper, Leigh and Martyn, 1681. Facsimile Edition, McGill University Press, Montreal, 1965. p. 111.
1. INTRODUCTION During the last 150 years the morphology of the cerebellum attracted numerous histologists. Its relatively simple structure, with its three-layered cortex and clearly defined afferent and efferent connections made it one of the favourite sites in the brain to test out new hypotheses on the connectivity, the development and chemical interaction in nervous tissue. We have attempted to review present knowledge about the external and internal morphology of the cerebellum and to relate the 'classical' topography of the cerebellum to the more recently discovered chemical specificity of its neurons and afferent and efferent pathways. Not all what is new in the histochemistry of the cerebellum is relevant to a better understanding of its chemoarchitecture. This review, therefore, does not pretend to be complete. It is focussed on afferent and intrinsic connections of the cerebellum. The efferent connections of the cerebellum to the brain stem and the spinal cord have not been systematically covered.
2. CYTOLOGY OF THE CEREBELLAR CORTEX A complete description of the histology of the cerebellar cortex was given by Ramon y Cajal (1911) (Figs 1 and 4). More recently the anatomy of the cortex including its ultrastructure was reviewed by Braitenberg and Atwood (1958), Eccles et al. (1967), Fox et al. (1967), Mugnaini (1972), and Palay and Chan-Palay (1974). Three layers are distinguished in the cortex (Fig. 3). The granular layer borders on the central white matter of the cerebellum. The Purkinje cell layer contains the cell bodies of the Purkinje cells, that are arranged in a single row. The perikarya of the Bergmann glia (the Golgi epithelial cells) are intercallated between the larger Purkinje cells (Fig. 9A). The molecular layer has a low cell content. It contains the dendritic arbors of the Purkinje cells and the Bergmann glial fibers, which run up to the pial surface where they constitute the external glial limiting membrane. The morphology of the cerebellar cortex can be characterized as a lattice: '... it can only be represented in two planes perpendicular to each other and having definite relations to the longitudinal and transversal axes of the
Handbook of Chemical Neuroanatomy, Vo112. Integrated Systems of the CNS, Part IH L.W. Swanson, A. Bj6rklund and T. H6kfelt, editors 9 1996 Elsevier Science B.V. All rights reserved.
Ch. I
J. Voogd, D. Jaarsma and E. Marani
,J-
I htll l,t'Mi,~/.~
_ "
F
1
d
/c
t
B
A
I
Fig. 1. Cerebellar cortical circuits. Top. Diagram showing the main mossy fiber-granule cell-Purkinje cell circuit and the innervation of the granule cells by the axonal plexus of the Golgi cell. A: mossy fiber; a: granule cell; B: Purkinje cell axon; b: parallel fiber; c: Golgi cell; d: Purkinje cell. Bottom. Similar diagram showing the main cortical circuit and the connection of the basket cell with the Purkinje cell somata. A: mossy fiber; a: granule cell; B: Purkinje cell axon; b: basket cell; C: climbing fiber; c: Purkinje cell soma. Redrawn from Ramon y Cajal (1911).
The cerebellum." chemoarchitecture and anatomy
Ch. I
cf
pi
pcc
I L-. GR mf
CN
NC
A
Fig. 2. D i a g r a m s of the cerebellar circuit. Inhibitory neurons are indicated in black. A. M a i n circuit. B. Cortical interneurons and recurrent pathways. Abbreviations: B = basket cell; cf = climbing fiber; G = Golgi cell; G R = granule cell; IO = inferior olive; m f = mossy fiber; nc = nucleocortical axons; no = nucleo-olivary axons; pcc = recurrent Purkinje cell axon collaterals; P cell = Purkinje cell; P C N = precerebellar nuclei; p f - parallel fiber; pi = pinceau of basket cell axons; S = stellate cell; U B C = unipolar brush cell; 1 = extracerebellar mossy fiber; 2 - nucleo-cortical mossy fiber; 3 - mossy fiber collateral of uni-polar brush cell.
animal. The whole three dimensional structure, therefore, cannot be obtained by rotation but by translation in two directions, thus producing a lattice' (Braitenberg and Atwood, 1958, p. 1). The elements of the main cerebellar circuit were discovered by Ramon y Cajal (1888, 1911). The electrophysiological properties of the circuit were established by Eccles et al. (1967). The main circuit (Figs 1 and 2) consists of the mossy fiber afferent system, that terminates on the granule cells; the granule cell axons that ascend to the molecular layer and bifurcate into parallel fibers, that run in the long axis of the folium and terminate on the Purkinje cells and the projection of the Purkinje cells to the cerebellar or vestibular nuclei. Each Purkinje cell is innervated by a single climbing fiber (Ramon y Cajal, 1911; Eccles et al., 1966a) that takes its origin from the contralateral inferior olive. The synaptic connections of mossy fibers, parallel fibers and climbing fibers are excitatory. The Purkinje cells are inhibitory and use gamma aminobutyric acid (GABA) as a transmitter (Ito and Yoshida, 1964). Small interneurons of the cerebellar cortex (stellate, basket and Golgi cells) receive a parallel fiber input and constitute inhibitory feed back and feed forward loops terminating on the granule cells and the Purkinje cells (Figs 1, 2 and 4). The main determinant of the firing rate of Purkinje cells is the mossy fiberparallel fiber system. Excitatory coupling between climbing fibers and Purkinje cells is very strong, but the frequency of the complex spikes evoked in Purkinje cells by the climbing fiber is too low to contribute significantly to its firing rate. The function of the climbing fibers, therefore, is one of the main problems in cerebellar neurobiology. Purkinje cells project to the cerebellar or the vestibular nuclei, where their axons terminate with inhibitory synapses. The cerebellar nuclei receive their excitatory drive from collaterals of the mossy and the climbing fibers.
Ch. I
J. Voogd, D. Jaarsma and E. Marani
/
i
Fig. 3. A. Nissl-stained section of the cerebellar cortex of the cat. G = Golgi cell; Gr = granule cells; P = Purkinje cell, asterisks: protoplasmatic islands of Held. Bar = 20 r Purkinje (1837).
B. diagram of the cerebeUar cortex of
Granule cells are small neurons located in cell nests in the granular layer. Cell-free spaces in the granular layer, that are known as the protoplasmatic islands of Held, contain the terminals of the mossy fibers (Fig. 3A, asterisks). Mossy fibers originate from many different sites in the spinal cord and the brain stem and constitute the main afferent system of the cerebellar cortex. Mossy fibers are myelinated fibers that branch extensively within the cerebellar white matter and the granular layer. They terminate with large irregular swellings (the mossy fiber rosettes, Figs 1, 5 and 6) that are located along or at the end of the axon. Each rosette forms the center of a complex synapse (cerebellar glomerulus) between the mossy fiber rosette, the dendrites of several granule cells and the terminals of one type of short axon (Golgi) cell of the cerebellar cortex. More than one mossy fiber rosette may be present within a protoplasmatic island. Granule cells possess 3-4 short dendrites, terminating in claw-like excrescenses (Fig.7). The thin, unmyelinated axon ascends towards the molecular layer, where it bifurcates in the form of a T. The two branches, that are known as the parallel fiber, pursue a straight course in the long axis of the folia, parallel to the thousands of other parallel fibers that constitute the bulk of the molecular layer. Parallel fibers synapse with dendrites of Purkinje cells and short axon cells in the molecular layer. Both the ascending portion of the granule cell axon and the parallel fiber are beaded. These varicosities probably correspond to the synaptic sites (Fig. 7D-E). Parallel fibers are very long. In monkeys their length varied between 0.8 and 5 mm. (Fox and Barnard, 1957). Maximal lengths of parallel fibers of 4.6-5.0 mm were reported for the rat (Brand et al., 1976; Schild, 1980; Mugnaini, 1983). The mean length
The cerebellum." chemoarchitecture and anatomy
9 9 . . . . .
.'..I
..
,:f
"..,"
.....:...'...:;.';
:...;
...
Ch. I
.4~O;o 9.
.'
~ " "":'2.':,- :'.
. ..
'.'.
".-.:.'...'~. ;..;,.
,2..o ,..:../. ;..,.."
" :
. 9
."
...,
: . , ......, ..
;
. .',, 9
": ; :
.
-.
.
.
,
,
B.
""..:-.
Fig. 4. Semidiagrammatic parasagittal section through a folium of the mammalian cerebellum, based on data
from Golgi-stained material 9 A: molecular layer; B: granular layer; C: white matter; a: Purkinje cell; b: basket cells of the lower molecular layer; d: terminal basket formation of the basket cell axon; e: superficial stellate cells; f: Golgi cell; g: granule cells with their ascending axons; h: mossy fibers; i: the bifurcation of the granule cell axons; j: epithelial glial cell; m: astrocyte of the granular layer; n: climbing fiber; o: branching point of Purkinje cell recurrent axon collaterals. Redrawn from Ramon y Cajal (1911).
of parallel fibers of 4.4 mm, measured after microinjections of biocytin in the granular layer in the rat (Pichitpornchai et al., 1994) is close to the mean length of these fibers of 5 mm, estimated with stereological techniques by Harvey and Napper (1988). The two branches of the parallel fiber are of equal length (Pichitpornchai et al., 1994). Shorter parallel fibers are located at the base of the molecular layer (mean branch length 2.08 mm), they become progressively longer as they approach the pial surface (mean branch length 2.35 mm: Pichitpornchai et al., 1994). Parallel fibers in the superficial molecular layer are of a smaller calibre than deep parallel fibers (Fox and Barnard, 1957, monkey). A similar increase in size of the parallel fibers from superficial to deep laminae of the molecular layer was noticed by Pichitpornchai et al. (1994) in the rat. They also observed proximo-distal tapering of parallel fibers. Van der Want et al. (1985a,b) observed corresponding differences in synaptic size in superficial and deep layers of the molecular layer in the cat. The size and the spacing of the varicosities along the parallal fibers was found to be correlated with their caliber. The mean interval between two varicosities was 5.2 ~tm for the parallel fibers, 4.02 ~tm for the ascending axon of the granule cell (Pichitpornchai et al., 1994). The lamination in the molecular layer may be the expression of a deep to superficial gradient in the development of the parallel fibers
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 5. Mossy fiber rosettes in the granular layer. Left. Mossy fiber rosettes from neurons of the lateral reticular nucleus, labelled with antegrade transport of Phasaeolus vulgaris lectin. Bar = 25/lm. Right: Mossy fibers, Golgi impregnation. Cajal (1911). Abbreviations: a = large, terminal rosettes; b = rosettes 'en passage'; c = small rosette 'en passage'; G = granular layer; M - molecular layer; W = white matter. Courtesy of Dr. T.J.H. Ruigrok.
(Pellegrino and Altman, 1979). A population of thick, short parallel fibers was noticed by Pitchitpornchai et al. (1994) in the deep parts of the molecular layer. Deep lying parallel fibers may be myelinated and are one of the constituents of the supraganglionic plexus located above the Purkinje cells (Mugnaini, 1972). The mossy fiber-parallel fiber-Purkinje cell pathway is characterized by a large divergence. Each mossy fiber terminates on a great number of granule cells and each granule cell contacts hundreds of Purkinje cells along its parallel fiber. An average parallel fiber with a length of 6 mm forms approximately 1100 boutons (Brand et al., 1976). A portion of the molecular layer 6 mm wide contains approximately 750 Purkinje cell dendritic sheets (Brand and Mugnaini, 1976). This number is somewhat lower than the number of available boutons, when a parallel fiber would synapse once with each Purkinje cell it meets on its way (Brand et al., 1976). It is higher than the estimate of Napper and Harvey (1988b) in the rat that 15% of the boutons on parallel fibers synapse with non-Purkinje cells and that the rest synapses once with half of the Purkinje cell dendritic sheets it meets on its way. The granule cell/Purkinje cell ratio was estimated at 274/1 by Harvey and Napper (1988) and at 350-500/1 for different lobules of rat vermis by Drfige et al. (1986). Napper and Harvey (1988) concluded that there are some 175.000 parallel fiber synapses on a single Purkinje cell of the rat. Fox et al. (1967) arrived at a number of 120.000 in monkeys. The actual strength of the convergence of individual mossy fibers to Purkinje cells depends on the distribution of their mossy fiber rosettes. Electrophysiological studies of Bower and Woolston (1983) in the rat demonstrated that Purkinje cells are most responsive to mossy fiber input that reaches the granule cells located immediately below them. Llinas (1982) explained this strong radial connectivity by the greater number of
The cerebellum." chemoarchitecture and anatomy
Ch.l
A
13
Fig. 6. Drawing of horizontal section through rat cerebellum showing orientation of mossy fibers. A. Elliptical segment or stripe of mossy fiber terminals in the medial portion of the anterior lobe showing the strong caudal-rostral organization of the terminal neuropil. Note the small cluster of granule cell bodies at the open arrow. B. Single mossy fiber from the next adjacent section showing the almost linear caudal-rostral pattern of the related terminals and small groups of parallel fibers (pf). a: View of rat cerebellum from the above showing approximate position of the field illustrated (note square and arrow), b: Medial sagittal section through cerebellum showing approximate location and plane of section. Abbreviations: fp = fissura prima; Isim = lobulus simplex; crI = crus I; fsp = fissura superior posterior; fpl = fissura posterolateralis; pf - parallel fiber. Golgi modification; 21-day-old rat. Scheibel (1977).
synapses with Purkinje cells on the ascending portion of the parallel fiber. However, according to Napper and Harvey (1988) the synapses on ascending portions of parallel fibers would account for only 3% of the total number of synapses of these fibers. Pichitpornchai et al. (1994), who observed a closer spacing of varicosities on the ascending axon and the proximal branches of the parallel fibers than on their distal branches, concluded that parallel fibers will exert a graded synaptic influence on their target Purkinje cells, with the most powerful influence occurring on cells located around the proximal regions of the fibers where they bifurcate. Mossy fiber terminal branches in the granular layer are oriented longitudinally, in the same plane as the Purkinje cells (Scheibel, 1977), (Fig. 6) (see also Section 6.4.2.). Mossy fibers, therefore, preferentially activate longitudinally oriented patches of Purkinje cells. Different types of mossy fiber rosettes were described by Brodal and Drablos (1963) with the Glees and Rheumont-Lhermitte silver impregnations and the Golgi method in rat and cat. Highly branching mossy fibers terminating in small, relatively simple rosettes, located along or at the end of the fiber, occur in all parts of the cerebellum. Large rosettes, consisting of aggregations of larger and smaller argyrophilic particles, interconnected by fiber fragments occur exclusively in nodulus and adjoining uvula, lingula and flocculus. The dendritic tree of the Purkinje cell is flattened in a plane perpendicular to the long
Ch.I
J. Voogd, D. Jaarsma and E. Marani
@
k @ !y,.
@
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 7. Granule cells and parallel fibers after an injection of biocytin in lobule X of the cerebellum of the rat. A. Biocytin labelled granule cell. B. Golgi-impregnated granule cells and parallel fibers in transverse section. Cajal (1911). C. biocytin injection site in granular layer and labelled parallel fibers in molecular layer. D. bifurcation site of labelled parallel fibers. E. labelled varicose parallel fibers. Abbreviations: A: molecular layer; B: granular layer; C: white matter; a: granule cell axon; b: bifurcation of granule cell axon; d: Purkinje cell; f: Purkinje cell axon; g: granular layer; I: injection site; m: molecular layer. Bars in A = 12/~m, in C = 500 /~m, in D and E - 50/~m. Courtesy of Dr. T.J.H. Ruigrok. (
axis of the folia (Figs 8 and 9). The soma and the proximal dendrites of the Purkinje cell are relatively smooth, the distal dendrites (spiny branchlets) bear long-necked spines (Fox and Barnard, 1957). When the parallel fibers traverse the Purkinje cells they terminate with boutons en passage on the spines of their spiny branchlets. Climbing fibers terminate on short, stubby spines on the proximal dendrites of the Purkinje cells (Larramendi and Victor, 1967; Palay and Chan-Palay, 1974) (Figs 10, 11 and 14). The axon of the Purkinje cell is myelinated (Fig. 9) and gives rise to recurrent collaterals (Bishop, 1982, 1988; Bishop and O'Donoghue, 1986; Bishop et al., 1987; O'Donoghue and Bishop, 1990). The collaterals form a plexus of beaded axons, mainly at the level of the Purkinje cell layer (Fig. 8a and b). They terminate on neighbouring 9
!
99
i!i 9
i;~ ~,~
..
9
9~
~,~...
~
"
.,.~N
9
.
Fig. 8. Purkinje cell of rat cerebellum. Intracellular injection with lucifer yellow and staining with anti-lucifer yellow antibody. PAP method, cresyl violet counterstained. Note plexus of beaded axon collaterals in A and B and spiny branchlets in C. Courtesy of Dr. T.J.H. Ruigrok. Bars in B = 50 ~m, in C = 5/~m. Abbreviations: a = Purkinje cell axon; cp = collateral plexus.
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 9. Purkinje cell in sagittal section. H~iggqvist stain. A. The small, densely stained nuclei in the Purkinje cell layer belong to the Bergmann glial cells. B. Initial segment of Purkinje cell myelinated axon (A) surrounded by pinceau of terminal basket cell axons. Abbreviations: A = Purkinje cell axon; B = Bergmann glial fiber; D = Purkinje cell dendrite. Bar - 25 pm.
10
The cerebellum: chemoarchitecture and anatomy
Ch. I
s
(-
L
Climbing
Ds ~
'9-Ss
Stellate
Parallel
\ Basket
Fig. 10. Synapses on mouse Purkinje cell. Climbing fibers terminate on short, stubby spines of proximal dendrites (Ds), parallel fibers terminate on spiny branchlets (Bs). Stellate cell and basket cell axons terminate on proximal dendrites and soma. Larramendi and Victor (1967).
Purkinje cells. The recurrent collaterals extend into the molecular layer where they contact basket cells (O'Donoghue et al., 1989). The whole collateral arborization is oriented perpendicular to the long axis of the folia, i.e. in the same plane as the dendritic tree of the Purkinje cell. In the cat it measures 300-700 #m in the sagittal and 100-400 #m in the transverse direction (Bishop, 1988). The width of the arborization and its penetration in the molecular and granular layers varies for different parts of the cerebellum. Recurrent collaterals of Purkinje cell axons are constituents of the infra- and supraganglionic plexus. The main Purkinje cell axon enters and traverses the white matter to terminate on cells of the cerebellar or the vestibular nuclei. Climbing fibers (Fig. 14) innervate the Purkinje cells, each Purkinje cell receiving only one climbing fiber (Ramon y Cajal, 1911). The olivocerebellar parent fibers of the climbing fibers branch extensively in the cerebellar white matter. For the adult rat the ratio of climbing fiber innervated Purkinje cells to neurons of the inferior olive is approximately 10:1 (Schild, 1970; Delhaye-Bouchaud et al., 1985). During their development the Purkinje cells receive more than one climbing fiber, it is not known how these supernumary climbing fibers are eliminated. Branching of olivocerebellar fibers occurs 11
Ch. I
J. Voogd, D. Jaarsma and E. Marani
sb
Pf Pf
. ii.~: -
.
,
.
Fig. 11. Diagram of the interaction of Purkinje cell dendrite with a climbing fiber and several parallel fibers. A proximal Purkinje cell dendrite (pd) shows stubby thorns contacted by a climbing fiber (cf), whereas parallel fibers (pf) synapse on spines protruding from a spiny branchlet (sb). Rossi et al. (1991).
in the parasagittal plane (Armstrong et al., 1973; Brodal et al., 1980; Rosina and Provini, 1983). This is one of the reasons for the longitudinal, strip-like organization of the olivocerebellar projection (see Section 6.3.3.). Transverse branching is limited to climbing fibers terminating in certain longitudinal strips (Ekerot and Larson, 1982). Climbing fibers take their origin from the contralateral inferior olive. For a long time the origin of the climbing fibers remained obscure. Their ultrastructure and their mode of termination were first recognized by Larramendi and Victor (1967) (Figs 10 and 11) in the mouse as beaded fibers, with boutons en-passage, filled with rounded vesicles terminating on short spines on Purkinje cell proximal dendrites. The clear intervesicular axoplasm distinguishes climbing fibers from the neurofilamentous basket cell axons. Earlier Scheibel and Scheibel (1954) had reviewed Ramon y Cajal's (1888) original description of the morphology of the climbing fiber. They concluded that climbing fibers emit collaterals in the granular and molecular layer, that terminate in glomeruli, on somata of Golgi, basket and stellate cells and on neighbouring Purkinje cells. Szentagothai and Rajkovits (1959) subsequently identified climbing fibers in axonal degeneration studies from their 'Scheibel-collaterals' and concluded that the climbing fibers originate from the inferior olive. Hamori and Szentagothai (1966b) described the climbing fibers as packed with neurofilaments and making synaptic contacts with few vesicles on the smooth parts of the dendrites. They probably mistook ascending collaterals of basket cell axons for the climbing fibers. The origin of the climbing fibers from the inferior olive was finally settled by Desclin (1974), who observed their degeneration with axonal silver impregnation methods after lesioning the inferior olive of the rat with 3-acetylpyridin (3-AP) administrated intra-peritoneally. In an exhaustive analysis of the normal light- and ultrastructural morphology of the climbing fiber, Palay and Chan-Palay (1974) observed the existence of climbing fiber glomeruli and synapses with Golgi cells in the granular layer and synaptic contacts of climbing fiber tendrils with basket and stellate cells. Desclin and Colin (1980) were unable to confirm these types of collateral contacts, outside the Purkinje cells, in an 12
The cerebellum." chemoarchitecture and anatomy
Ch. I
//
Fig. 12. Purkinje cells from the cerebellum of, from left to right, birds (Gallus domesticus, Feirabend, 1983); mammals (cat, Cajal, 1911) and fish (Gnathonemus petersii, Nieuwenhuys, 1969). Note different length, orientation and position in the molecular layer of the spiny dendritic branchlets.
ultrastructural study of the cerebellar cortex of 3-AP-treated rats. O'Donoghue et al. (1989) found intracellularly stained basket cells of the cat to lack climbing or mossy fiber terminals on their somata. During postnatal maturation of the cerebellum of the mouse, Mason and Gregory (1984) found many axons that combine the morphology and synaptic connections of both climbing and mossy fibers. These combination fibers are rare in the adult. Purkinje cell dendritic trees in the molecular layer remain oriented perpendicular to the parallel fibers irrespective of the changes in direction of the folial chain. Their dendrites share this orientation with the climbing fibers terminating on them. This type of spatial organization is found in all vertebrates and is the main condition which determines the morphology of the cerebellum. Purkinje cells in fish and amphibians are not arranged in a monolayer, but can be clustered in specific parts of the cortex, reminiscent of the clustering of the Purkinje cells during early stages of cerebellar development in all vertebrates (Nieuwenhuys, 1967). Purkinje cells in lower vertebrates differ from the mammalian type by the disposition of their smooth, proximal branches and their spine-loaden terminal branches in the molecular layer (Fig. 12). In fish the proximal smooth branches are found at the same level as the somata of the Purkinje cells, and the distal spiny branchlets extend as straight spikes into the molecular layer. This condition was extensively studied by Nieuwenhuys and Nicholson in the cerebellum of mormyrid fish (Nieuwenhuys and Nicholson, 1969a,b). As a consequence the climbing fibers, that synapse with the smooth proximal part of the dendrites, do not 'climb' 13
Ch. I
J. Voogd, D. Jaarsma and E. Marani
into dendrites within the molecular layer, but terminate at the same level as the perikarya of the Purkinje cells (Kaiserman-Abramof and Palay, 1969). In reptiles and birds the smooth, proximal dendrites with their climbing fiber terminals do not extend beyond the lower third of the molecular layer (Mugnaini, 1972; Freedman et al., 1977; Kfinzle 1985). Only in mammals the smooth branches and the climbing fiber arborizations reach the pial surface (Fig. 12). Cerebellar nuclei that contain the target cells of the Purkinje cell axons have been described in species of all vertebrate classes. In some species of fish target cells of the Purkinje cell axons are also located within the cortex among the Purkinje cells (the 'eurodendroid' cells of Nieuwenhuys et al., 1974). The cells of the 'fourth cortical layer' in some aquatic mammals, that are located below the granule cells in the white matter, can be considered as displaced cerebellar nuclear cells (Ogawa, 1934). Interneurons in the cerebellar cortex are inhibitory and constitute various feed-back and feed-forward circuits between parallel fibers, granule cells and Purkinje cells (Figs 1, 2 and 4). Their dendrites are located in the molecular layer, where they are contacted by parallel fibers. Golgi cells are most numerous in the upper part of the granular layer. Some of their dendrites ramify in the granular layer, where they are contacted by mossy fiber terminals in the glomeruli. The dendritic tree of Golgi cells is not oriented in a specific plane. Recently it was shown by De Zeeuw et al. (1994c) that axons of Golgi cells course for some distance in the supra- or infraganglionic plexus in the direction of the long axis of the folia, before they branch into a dense telodendrion in the granular layer. Their terminals are located at the periphery of the glomeruli, where they synapse with granule cell dendrites (Fox et al., 1967). Their ratio was estimated in the rat as 4-6 Golgi cells for each Purkinje cell. The number of Golgi cells is about three times higher in lobule X than in other lobules (Drfige et al., 1986). However, unipolar brush cells (see below and Section 3.6.3.) may have been mistaken for Golgi cells by these authors.
Fig. 13. Orthogonal arrangement of basket cell axons (thick horizonal fibers oriented in the plane of the Purkinje cells in A and B) and parallel fibers (thin, vertical fibers in A and B). A. Drawing from Golgiimpregnated section, Cajal (1911). B. Bodian-stained section of rat cerebellar cortex. Abbreviations: A and B = stellate cells; C = basket cell axon; E = pericellular basket; F = Purkinje cell dendritic tree; G = climbing fiber; Pb = pericellular baskets. Bar = 100/~m.
14
The cerebellum." chemoarchitecture and anatomy
Ch.I
Fig. 14. Phaseolus vulgaris lectin-labelled climbing fibers of rat cerebellum. A. Sagittal section. B. Coronal section. Abbreviations: G = granular layer; M = molecular layer; P = Zebrin- labelled Purkinje cells. Bar = 100/lm. Courtesy of Dr. T.J.H. Ruigrok.
Apart from the parallel fiber boutons on the dendrites, the soma of Golgi cells is contacted by Purkinje cell recurrent collaterals (Hamori and Szentagothai, 1966a, 1968, Palay and Chan Palay, 1974). Mossy- and climbing fiber terminals on Golgi cells, that were mentioned by several authors (Hamori and Szentagothai, 1966a, 1968, Palay and Chan Palay, 1974) have not yet been confirmed in experimental axonal tracing studies. Myelinated fibers, indicated as mossy and climbing fibers, and recurrent collaterals of Purkinje cell axons, terminate on Golgi cell somata with large, crenelated synapses ('synapse en marron': Palay and Chan-Palay, 1974). The synapse en marron recently was identified by Mugnaini and Floris (1994) as a synapse of the mossy fibers with the unipolar brush cells of the cerebellar cortex. Stellate cells are located in the entire molecular layer, basket cells constitute a special population located in its lower one third. Dendrites of stellate and basket cells are oriented in a direction perpendicular to the long axis of the folium. Axons of stellate cells terminate on Purkinje cell dendrites. The basket cell axon increases in thickness after it emerges from its soma (Figs 1 and 13). It runs, perpendicular to the long axis of the folium, above the perikarya of the Purkinje cells and gives off descending and ascending collaterals. The descending collaterals branch and surround and synapse with the somata of Purkinje cells. The axons of these pericellular baskets of the Purkinje cell terminate in a periaxonal plexus (the pinceau) surrounding the initial segment of the Purkinje cell axon. Ascending collaterals of the basket cell axon terminate on the smooth surface of the proximal dendrites of Purkinje cells. O'Donoghue et al. (1989) who studied the connections of intracellularly stained basket cells and Purkinje cells in the cat concluded that each basket cell soma received input from recurrent collaterals from a single Purkinje cell. Other afferents of the basket cell include parallel fibers, climbing fibers and stellate and basket cell axons (Palay and Chan Palay, 1974). The infra- and supraganglionic plexus, are located on either side of the layer of Purkinje cell somata. They contain myelinated Purkinje cell collaterals. Most myelinated fibers in the supraganglionic layer are oriented in the long axis of the folia and, therefore, represent myelinated granule cell axons or, possibly, axons of candelabrum cells (Lain6 and Axelrad, 1994) or Golgi cells (De Zeeuw et al. 1994c). In silver impregnations these axons are distinctly smaller than the basket cell axons, that cross them at right angles 15
Ch.I
J. Voogd, D. Jaarsma and E. Marani
(Fig. 13). The so-called multi-layer fibers also contribute to the plexus surrounding the Purkinje cells. These fibers were traced from several sources, including the noradrenergic, serotoninergic and cholinergic cell groups of the brain stem. They ramify in all layers of the cortex, but constitute the densest plexus at the level of the Purkinje cells. They are discussed more fully in Sections 3.8., 3.9. and 3.10.1. Several other neuronal cell types have been identified in the cerebellar cortex. The Lugaro cell is a relatively rare fusiform neuron, located just below the Purkinje cell layer (Lugaro, 1894; Fox, 1959; Palay and Chan-Palay, 1974). Its dendrites stretch out along the boundary of the granular and the Purkinje cell layer, the destination of its axon is not known. Lugaro cells can be discriminated from Golgi cells immunocytochemically with specific antibodies (Section 3.6.2., Fig. 67). The candelabrum cell has been recognized in Golgi-impregnated sections from rat cerebellum by Lain6 and Axelrad (1994). The neuron is rather frequently encountered in all lobules of the cerebellum. Its medium sized perikaryon is sqeezed in between the somata of Purkinje cells. Its dendritic tree is somewhat flattened and mainly extends in the parasagittal plane. One or two dendrites course through the molecular layer, dividing into few, slightly oblique branches, that are covered with irregularly distributed spines. A few slender dendrites branch in the upper granular layer. The axon courses in the direction of the long axis of the folium in the Purkinje ceil - or the supraganglionic layer, and gives off terminal, beaded branches that ascend in the molecular layer at regular parasagittal intervals. The chemical anatomy of the candelabrum cell has not yet been studied (Section 3.6.2). The unipolar brush cells were first identified in the rat by Altman and Bayer (1977) as the 'pale cells' of the granular layer. These cells are intermediate in size between the granule and the Golgi cells, and possess a typical, pale nucleus. They are concentrated in the nodulus, the ventral uvula, the flocculus and parts of the paraflocculus. They are born after the Purkinje cells, but before the stellate, basket and granule cells. The cells were sporadically recognized as monodendritic neurons in a number of immunocytochemical studies (see Section 3.6.3), but have been characterized with Golgi impregnation and electron microscopic methods only recently (Floris et al., 1994; Mugnaini and Floris, 1994; Mugnaini et al., 1994). The name 'unipolar brush cell' was given by Mugnaini and colleagues (Mugnaini and Floris, 1994) after the tip of the stubby dendrite, that forms a tightly packed group of branchlets resembling a paint brush (Fig. 68). The soma of unipolar brush cells is spherical to oval and carries thin appendages. The axon only can be impregnated for a short distance, suggesting that its distal, unimpregnated part is myelinated. Side branches of the axon terminate in rosette-like formations in the granular layer (Fig. 2) (Berthi6 and Axelrad, 1994; Floris et al., 1994; Rossi et al., 1995), the main stem of the axon may enter the white matter. Unipolar brush cells are innervated by one or two mossy fiber rosettes, in the form of particularly extensive contacts. Mossy fibers end on the perikaryon as well as on the dendritic brush (Mugnaini et al., 1994). These large synapses correspond to the 'synapse en marron' of Palay and Chan-Palay (1974), originally identified as a mossy fiber-Golgi cell synapse (see also Monteiro et al., 1986). Unipolar brush cells also receive symmetrical synapses from boutons containing pleomorphic vesicles, presumably originating from Golgi cells or Purkinje cell recurrent axons. Some of the dendritic branchlets may be presynaptic to dendrites of other cells in the granular layer (Floris et al., 1994). Pale cells, monodendritic and unipolar brush cells are all more frequent in the vestibulocerebellum. The chemical identity of the unipolar brush cell will be discussed in Section 3.6.3.
16
The cerebellum." chemoarchitecture and anatomy
Ch. I
3. CHEMICAL ANATOMY OF THE CEREBELLAR CORTEX By virtue of its laminated and relatively simple structure the cerebellar cortex has served as the playground for every student who wanted to test a histochemical reaction or a new antibody on the brain. From this large body of data we have selected those which are important for the understanding of the morphology and the connections of the cerebellum. The localization in the cerebellar cortex of neurotransmitters, peptides, second-messenger systems, calcium-binding proteins and other biochemical markers is reviewed separately for each cell type of the cortex and for the mossy and climbing fibers. Glutamate and GABA receptors, nitric oxide, adenosine, the monoamine afferent systems and receptors, the hypothalamo cerebellar and histaminergic afferents and the cholinergic systems and acetylcholinesterase are discussed in separate sections. The chemoarchitecture of the cerebellar cortex has been reviewed by Schulman (1983), Nieuwenhuys (1985) and Oertel (1993). 3.1. PURKINJE CELLS
3.1.1. Gamma-aminobutyric acid (GABA), glutamic acid decarboxylase (GAD) and the GABA-transporters in Purkinje cells Purkinje cells use gamma-aminobutyric acid (GABA) as their main neurotransmitter and exert a postsynaptic inhibitory effect on cells of the cerebellar and vestibular nuclei (Ito and Yoshida, 1964; Obata et al., 1967; Obata, 1969, 1976; Obata and Takeda, 1969). GABA in rabbit Purkinje cells was first demonstrated using a histochemical method, demonstrating the conversion of GABA into succinic acid (Van Gelder, 1965). In selective uptake studies of [3H]GABA in cerebellar slices, only a low activity was present over the Purkinje cells (H6kfelt and Ljungdahl, 1970, 1971; Schon and Iversen, 1972). Minimal uptake of [3H]GABA was also observed for Purkinje cell axon terminals (Storm-Mathisen, 1975). All Purkinje cell somata of the cerebellum of the rat and their primary and secondary dendrites were immunoreactive for antisera against glutamic acid decarboxylase (GAD), the synthesizing enzyme of GABA (Fig. 62D,E). Varicose fibers and terminals in the cerebellar nuclei were densely stained (Saito et al., 1974; McLaughlin et al., 1974; Oertel et al., 1981b; Perez de la Mora et al., 1981; Somogyi et al., 1985). Immunoreactivity in Purkinje cell somata was generally found to be weak, or to be dependent on blocking of axonal transport by colchicine (Ribak et al., 1978). Strong immunoreactivity in Purkinje cell somata was, however, reported by Mugnaini and Oertel (1985) with an anti-GAD antiserum produced by Oertel et al. (1981 a). The presence of GAD mRNA in Purkinje cells has been demonstrated with in situ hybridization histochemistry in rodents and primates resulting in dense labelling over somata of Purkinje cells (Wuenschell et al., 1986; Julien et al., 1987; Ferraguti et al., 1990; Herrero et al., 1993). Two forms of GAD with apparent molecular weights in the range of 59-67 kDa, that differ by 2-4 kDa, were identified by Chang and Gottlieb (1988) and Martin et al. (1991). In situ hybridization histochemistry with probes for the high molecular weight form, GAD67, and the low molecular weight form, GAD65, showed a prevalent localization of GAD67 over GAD65 in Purkinje cell bodies of rat cerebellum. The reverse localization was reported for Golgi cells (Esclapez et al., 1993; Feldblum et al., 1993). A differential distribution of GAD67 and GAD65 in Purkinje cells was found in immunocytochemical studies with specific antibodies for GAD67 and GAD65. Antibody K2, which is specifc 17
Ch. I
J. Voogd, D. Jaarsma and E. Marani
...., ..
,. ~... ..:i."./~i;~.".!...i. .... ~
.
.
..: .,.. .~.,....
for GAD67 , strongly immunoreacted with Purkinje cell perikarya, their proximal dendrites and their axon terminals in rat cerebellum (Kaufman et al., 1991; Moffett et al., 1994). The monoclonal antibody GAD-6, which is specific for GAD65 (Chang and 18
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 15. Localization of GABA-like immunoreactivity in semithin sagittal sections of rat (A,B) and chick (C,D) cerebellum. A. Low magnification of cell bodies and neuronal processes reacting with monoclonal anti-GABA antibodies. B. Higher magnification of an area partially included in the frame in A. Note strong immunoreactivity in stellate cell bodies (open arrow), in Golgi neurons (thick arrow), in the basket terminals surrounding Purkinje cell bodies, in puncta at glomeruli (dotted line) and in axons in the white matter. Immunoreactivity in Purkinje cells is weak. C and D. Two typical patterns of GABA-like immunoreactivity observed in Vibratome slices of the chick cerebellum. C: Intensely (thick arrow) and weakly (open arrow) immunoreactive Purkinje cells together with the staining in their dendritic arborization (thick arrowhead). D. Basket terminals around two weakly stained Purkinje cells (open arrowheads). Molecular layer (MO); Purkinje cell layer (P); granular cell layer (GL); white matter (WM). Bar in A = 100 r bar in B, C and D = 25 r (Matute and Streit, (1986). (
Gottlieb, 1988), i m m u n o r e a c t e d with axon terminals of Purkinje cells, but p o o r l y imm u n o s t a i n e d the p e r i k a r y a of Purkinje cells ( K a u f m a n et al., 1991). Antibodies against conjugates of G A B A were first applied to d e m o n s t r a t e specific G A B A - l i k e i m m u n o r e a c t i v i t y in Purkinje cells by S t o r m - M a t h i s e n et al. (1983). Imm u n o r e a c t i v i t y of the cell b o d y and the dendrites with antibodies against conjugates of G A B A was generally f o u n d to be weak or absent, but strong in the a x o n and the myelinated axon collaterals in the infraganglionic, but especially in the supra-ganglionic plexus, and in their terminals in rat (Ottersen and S t o r m - M a t h i s e n , 1984a,b, 1987; Ottersen et al., 1987; M a d s e n et al., 1985; Sdgudla et al., 1985; G a b b o t t et al., 1986; M a t u t e and Streit, 1986; S o m o g y i et al., 1986; A o k i et al., 1986) cat (Somogyi et al., 1985) and m o u s e ( T a k a y a m a , 1994). Staining in Purkinje cell s o m a t a in the chicken was stronger t h a n in m a m m a l s ( M a t u t e and Streit, 1986) (Figs 15, 62, 63). Several G A B A t r a n s p o r t e r proteins, that are active in the high-affinity u p t a k e of GABA
GABA/MOTILIN
~r
~
-....
GABA/TAURINE
MOTILIN
TAURINE
I SINGLE SPECTRUM II SEPARATE POOLS III DYNAMIC SHIFTS IV GENETIC CONTROL V FUNCTIONAL INFLUENCE
TAURINEIMOTILIN
Fig. 16. Schematic diagram to illustrate the concept of dynamic interrelationships between taurine, motilin, and gamma-aminobutyric acid (GABA) in a single neuron. A neuron with both substances in coexistence may have fluctuating levels of one or both substances depending upon parameters of rhythm, time, and physiologcal demands for one or another mediator during specific types or phases of activity. Chan-Palay (1984). 19
Ch. I
J. Voogd, D. Jaarsma and E. Marani
-,:
T
. . . . . . . . . . .
i
20
:
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 17. Immunoreactive staining with taurine (Tau2) antibody in rat cerebellum fixed with 4% paraformaldehyde. A. Tau2-immunoreactive staining on a coronal section though lobule 6 of the vermis. Purkinje cells and dendrites exhibiting Tau-LI were separated by bands of unstained Purkinje cells. M, molecular layer; P, Purkinje cell layer; G, granule cell layer; W, white matter; P", Purkinje cell layer out of plane focus. Location, bregma = 11,5 mm. B. High-power photomicrograph of area indicated in A demonstrating taurine-like immunoreactivity in Purkinje cells (short solid arrows) and dendrites (long solid arrows) separated by unstained Purkinje cells (open arrows). C. Tau2-immunoreactive staining in a horizontal section through lobule 3 of the vermis. D. Adjacent section to (C) demonstrating that taurine-like immunoreactivity was completely absorbed by incubation of Tau2 (1 : 40) with original antigen, taurine conjugated to KLH using glutaraldehydeborohydride. Bars in A, C, and I) = 100/.tm, in B = 50 r Magnusson et al. (1988).
GABA, have been cloned (GAT1-4: Guastella et al., 1990; Lopez-Corcuera et al., 1992; Borden et al., 1992; Liu et al., 1993; and GAT-B: Clark et al., 1992). All transporters occur in brain tissue. The regional distribution of GAT1 was studied by Rattray and Priestly (1993) with in situ hybridization in rat cerebellum. GAT1 mRNA is not expressed by Purkinje cells, but strongly by Bergmann glial cells. GAT-2 may be confined to glia (Liu et al., 1993), but detailed studies of their localization have not been published thus far. 3.1.2. Motilin and taurine in Purkinje cells
Certain inconsistencies in the results on the localization of GABA in Purkinje cells were discussed by Chan-Palay (1984). She concluded that GABA is present in varying amounts in different Purkinje cells and that it may co-exist with other neuroactive substances, notably with motilin and taurine, that also produce an inhibitory action on postsynaptic cells (Fig. 16). The presence of motilin in Purkinje cells was demonstrated with an antibody directed against conjugates of motilin (Chan-Palay et al., 1981; Nilaver et al., 1982). More than half of the Purkinje cells of the rat are immunoreactive for this antibody and in human cerebellum their proportion was even higher (Nilaver et al., 1982). Chan-Palay et al. (1981) found coexistence of GAD and motilin in 10-20% of the Purkinje cells of the rat. The presence of motilin in Purkinje cells has, however, been disputed by Lange (1986), who was unable to demonstrate the presence of motilin using radioimmuno-assay and reversed phase HPLC in extracts of rat cerebellum. Only one of Lange's anti-motilin antibodies, all of which had been demonstrated to be effective in demonstrating motilin-like immunoreactivity in rat duodenum, was found to immunoreact with Purkinje cells in immunocytochemical studies with rat cerebellum. Taurine has been proposed as a neurotransmitter in certain fiber systems. In the guinea pig cerebellum it was found to exert a hyperpolarising effect on Purkinje cell dendrites and was proposed as a neurotransmitter in stellate cell-Purkinje cell synapses (Okamoto et al., 1983). [3H]Taurine was found to accumulate in Purkinje cells. Immunocytochemical studies with antibodies specific for cysteine-sulfonic acid decarboxylase (CSADCase), the enzyme involved in taurine synthesis, by Chan-Palay et al. (1982a,b), showed that CSADCase immunoreactivity was present in most, but not all the Purkinje cells of rat cerebellum, and was more prominent in the main dendritic arbor than in the perikarya and the axon. CSADCase, motilin and GAD-like immunoreactivities were found to co-exist in Purkinje cells located near the midline. In contrast to the observations of Chan-Palay et al. (1982a,b), Almarghini et al. (1991) found CSADCase immunoreactivity to be localized in Bergmann glia and interfascicular oligodendrocytes and to be absent from Purkinje and stellate cells. Most authors who used antisera against conjugates of taurine to localize taurine-like 21
Ch. I
J. Voogd, D. J a a r s m a a n d E. M a r a n i
J..i
J..i i iJJ .
~,.
~
~
it
"
~
:~i ! : ;~i~
~ ~:'i~
9
2
,,~i~,
"~,
~,~.,'~.
",J
.
~~....
~
"
E
..
MO ..
In,
GC
lii
....
6 :.
...
immunoreactivity found staining in all Purkinje cells of the cerebellum of the rat (Madsen et al., 1985; Campistron et al., 1986b; Tomida and Kimura, 1987; Ida et al., 1987; Ottersen et al., 1988b; Ottersen, 1988, 1989). Magnusson et al. (1988) recognized a zonal distribution of taurine-immunoreactivity similar to the zonal labelling of CSADCase immunoreactivity observed by Chan-Palay et al. (1982a,b) (Fig. 130) in paraformaldehyde-fixed brain tissue (Fig. 17), but not in glutaraldehyde-fixed tissue. Analysis of semithin sections and immunogold electron microscopy indicated that taurine-immunoreactivity is selectively enriched in the somata, proximal and distal dendrites and axon terminals of the Purkinje cells (Fig. 18). Stellate and basket cell somata and their axon terminals are only weakly immunolabelled (Ottersen et al., 1988b; Ottersen 1988, 22
The cerebellum." chemoarchitecture and anatomy
Ch.I
Fig. 18. Photomicrographs showing the distribution of taurine-like immunoreactivity in the rat cerebellum, and the results of different control experiments. A,B. Semithin (0.5 r sagittal sections through vermis posterior treated with taurine (Tau) antiserum 20 diluted 1:3000 and subsequently processed according to the peroxidase-antiperoxidase procedure. A. Intense labelling of the somata, dendrites (small arrowheads), and axons (crossed arrow) of the Purkinje cells. The neurons (large arrowheads) and glial processes (small arrows) of the molecular layer appear immunonegative. Small asterisks indicate pial surface, large asterisk indicates Purkinje cell enlarged in B. B. Note staining of Purkinje cell dendritic spines (large arrow heads). Inset: Semithin test section mounted on the same slide as the tissue section shown in A and B and incubated in the same drops of sera. The test section contains brain protein-glutaraldehyde conjugates of different amino acids, separated by brain tissue that appears as darkly stained zones. Code: 1, GABA; 2, glutamate; 3, taurine; 4, glycine; 5, none (i.e., no amino acid in the reaction mixture); 6, aspartate; 7, glutamine (only small part of section represented in this particular view). The taurine conjugate is selectively stained. C. Transverse semithin section through nucleus interpositus anterior showing intense staining of axons (arrows) and nerve terminal-like dots, some of which (large arrowheads) appear to contact unstained cell bodies (asterisks). Same procedure as in A. D. Thin layer chromatograms (5 mm width) of soluble brain extracts fixed with glutaraldehyde and subsequently stained with an antiserum against glutamate (left strip) or taurine (right strip). The taurine antiserum reveals a single spot which has comigrated with free taurine and which is separate from the spot labelled by the glutamate antiserum. E, T and G indicate the application sites of the brain extract, taurine and glutamate, respectively. E. Adjacent section to that shown in A and accompanying test section (inset) incubated with the taurine antiserum after preabsorption with glutaraldehyde-taurine complexes (200/tM with respect to taurine). There is virtually no staining. Abbreviations: MO, molecular layer; GC, granule cell layer. Bars 25/~m. Ottersen (1988). (
1989). Following hypo-osmotic stress there is a transient shift of taurine from the Purkinje cells to the B e r g m a n n glial c o m p a r t m e n t , where taurine-immunoreactivity now becomes a p p a r e n t (Nagelhus et al., 1993).
3.1.3. Calcitonin gene-related peptide (CGRP), acetylcholinesterase (ACHE), somatostatin and tyrosine hydroxylase in Purkinje cells Some neuroactive substances have been reported to be present in Purkinje cells only for a certain period during development. Calcitonin gene-related peptide ( C G R P ) is almost undetectable in adult rat cerebellum, but a transient immunoreactivity in i m m a t u r e Purkinje cells in rat cerebellum has been detected with antisera against C G R P ( K u b o t o et al., 1987, 1988; Chedotal and Sotelo, 1992). In adult rats, however, C G R P - l i k e immunoreactivity, co-localized with G A D immunoreactivity, can be detected in m a n y Purkinje cells near injections of colchicine (Kawai et al., 1985, 1987). C G R P receptors measured autoradiographically with [125I]CGRP as the ligand, are a b u n d a n t in adult rat and h u m a n cerebellum. [125I]CGRP-binding is dense over the molecular and Purkinje cell layers and low over the granular layer and the cerebellar nuclei (Inagaki et al., 1986). Binding to the molecular layer occurs in a pattern of longitudinal stripes (Kruger et al., 1988) and increases after intraperitoneal administration of harmalin (Rosina et al., 1990, 1992). Choline acetyltransferase (CHAT) and acetylcholinesterase (ACHE) have been found to be transiently expressed in Purkinje cells during development: Purkinje cells in certain parts of the i m m a t u r e rat and guinea pig cerebellum, including the lobules IX and X of the caudal vermis, display a transient reactivity for A C h E (Csillik et al., 1963, 1964; A l t m a n and Das, 1970; Odutola, 1970; Brown et al., 1986). The authors suggested that this transient A C h E activity in Purkinje cells is due to a transient cholinoceptive stage, when they are contacted by cholinergic mossy fiber afferents. A similar, transient expression of C h A T was observed in Purkinje cells of the rat vestibulocerebellum (Gould and Butcher, 1987). Pseudo-cholinesterase was localized in adult Purkinje cells of the 23
Ch. I
J. Voogd, D. Jaarsma and E. Marani
lobules IX and X (Robertson et al., 1991). These cells are arranged in multiple, sagittal bands (Gorenstein et al., 1987). Robertson et al. (1991) were unable to confirm the transient staining with AChE in rat Purkinje cells. Somatostatin was located in rat Purkinje cells using polyclonal and monoclonal antibodies against conjugates of somatostatin (Johansson et al., 1984; Vincent et al., 1985; Villar et al., 1989). Reactive Purkinje cells were especially numerous in parts of the vermis and paraflocculus and flocculus during early postnatal stages, but mostly disappeared later on (Figs 19 and 20). In part of the vermis they were located in bands. In the adult rat Purkinje cells can be stained on the ventral aspect of the paraflocculus (Gonzalez et al., 1988) and in the vermis, after interventricular administration of colchicine (Villar et al., 1989). Somatostatin-like immunoreactivity was also observed in climbing fibers, that were correlated with the patches of immunoreactive Purkinje cells and, more diffusely distributed, in Golgi cells (Villar et al., 1989). The presence of somatostatin in adult rat Purkinje cells of the paraflocculus was confirmed with nonradioactive in situ hybridization for somatostatin mRNA (Kiyama and Emson, 1990). Specific binding of iodinated agonists of somatostatin was studied in rat, using ligands for short, 14 amino-acid ([125I]SS-14) and long forms ([125I]SS-28). Binding in the cerebellar cortex was found to be low, but strong binding of both ligands was observed over the cerebellar nuclei (Uhl et al., 1985). Binding to somatostatin receptors in the human cerebellar cortex was higher. Different distribution patterns were noted among the patients studied, with higher densities over the granular layer (Laquerriere et al., 1994). Leroux et al. (1985) and Gonzalez et al. (1988) failed to demonstrate specific binding over the cerebellar nuclei of the rat of a different SS-14 ligand, but confirmed binding of SS-28 (Leroux et :al., 1985). Binding of an octopeptide somatostatin analogue was reported to be almost absent in rat cerebellum (Reubi and Maurer, 1985) and low over the cerebellar cortex of the human cerebellum, with intermediate values in the molecular layer (Reubi et al., 1986). Tyrosine hydroxylase, the synthesizing enzyme of dopamine, is expressed by Purkinje cells of the ventral vermis (lobules I and X) and the hemisphere (ansiform lobule, paraflocculus) of rat cerebellum (Takada et al., 1993). Expression of tyrosine hydroxylase by Purkinje cells is increased in the mutant tottering and leaner mice (Austin et al., 1992). 3.1.4. The localization of the IP3 receptor and the intracellular calcium stores of Purkinje cells
The phosphoinositide system is a second messenger system coupled to metabotropic, G protein-linked receptors (see Ross et al. (1990), Mayer and Miller (1990), Ferris and Snyder (1992) and Berridge (1993), for reviews). Receptor-mediated hydrolysis of phosphatidylinositol (PIP2) is catalyzed by phospholipase C and leads to the formation of inositol-l,4,5-triphosphate (IP3) and diacylglycerol (DAG), two second messengers that function in a bifurcating signal pathway. Other inositol phosphates (inositol 1,3,4,5tetrakiphosphate, IP4; inositol 1,3,4,5,5-pentakiphosphate, IPs; and inositol hexakiphosphate, IP6) have been localized in rat cerebellum (Vallejo et al., 1987; Theibert et al., 1987, 1991). Phosphorylation of IP 3 by the enzyme IP 3 3-kinase leads to the formation of IP4. IP3, through activation of IP3 receptors, causes Ca 2+ mobilization from intracellular sources, whereas DAG, together with Ca 2+, activates the enzyme protein kinase C that phosphorylates regulatory proteins. The localization of phospholipase C, IP 3 receptors and protein kinase C has been extensively studied in Purkinje cells. 24
The cerebellum: chemoarchitecture and anatomy
Ch. I
B
C
f
Fig. 19. Schematic illustration of the zonal distribution of somatostatin immunoreactive Purkinje cells at
different levels of the cerebellum of a 21 day old rat. Drawings have been made from frontal, cresyl-violet stained sections. Each dot represents 2-5 cells. Abbreviations: 5-9, cerebellar lobules V-IX; 4V, 4th ventricle; COP, copula pyramis; CR2, crus 2, ansiform lobule; FL, flocculus; PFL, paraflocculus; PM, primary fissure; SF, secondary fissure. Villar et al. (1989).
IP 3 3-kinase, the enzyme that produces IP 4 from IP3, was exclusively localized in Purkinje cells of the rat using immunohistochemistry (Mailleux et al., 1991 a, Mizuguchi et al., 1992) and in situ hybridization in rat and human cerebellum (Mailleux et al., 1991 b, 1992). Immunoreactivity was present in Purkinje cell dendrites more than in the perikarya. Intense immunolabelling of the dendritic spines was observed in the rat (Yamada et al., 1992; Go et al., 1993) (Fig. 21) but a specific role of IP 4 in Purkinje cell dendritic spines has not been disclosed. A similar localization in Purkinje cell dendritic spines was described for the mGluR1 subunit of the metabotropic glutamate receptor (Section 3.3.2., Fig. 52). Different isoenzymes of the phospholipase C (PLC) family, belonging to three major groups (fl, ~ and d), have been identified (Rhee et al., 1989; Rhee and Choi, 1992). PLC-fll, PLC-y and PLC-~ have been localized with in situ hybridization in the brain of the rat. Moderate activity was found for PLC-fll in the granular layer and strong activity in Purkinje cells and granule cells for PLCT'. The activity of PLC-d is low and may be localized in glial cells (Choi et al., 1989). PLC-A m R N A that was localized in rat Purkinje cells by Ross et al. (1989b), probably codes for a thiol-protein disulphide oxido-reductase and not for a PLC (Berridge, 1993). The IP 3 receptor has been found to be identical to the Purkinje cell-specific P400 25
Fig. 20. Examples of somatostatin-immunoreactive elements in the cerebellar cortex of the paraflocculus of adult colchicine treated rats. Patch of Purkinje cells and an immunostained Golgi cell (arrow head) are present in this section. Somatostatin-imrnunoreactive climbing fibers are observed. Calibration bar 50 prn. Villar et al. (1989).
5
The cerebellum." chemoarchitecture and anatomy
Ch. I
protein (Mignery et al., 1989) (Fig. 22). The P400 protein was originally isolated by Mallet et al. (1976) as a Purkinje cell-specific protein, that was reduced in homozygous Purkinje cell-deficient (pcd, Mullen et al., 1976) and 'staggerer' (Sidman et al. 1962) mice, but relatively enriched in the cerebella of 'reeler' and 'weaver' mutant mice, with a loss of granule cells (Mikoshiba et al., 1979). Immunocytochemical studies with a monoclonal antibody specific for P400 protein, indicated that the protein was localized in somata, dendrites and axons of Purkinje cells in rodents (Maeda et al., 1988; Nakanishi et al., 1991; Rodrigo et al., 1993). The development of Purkinje cells could be traced with P400-immunostaining of staged cerebella of mouse embryos (Maeda et al., 1989) (Fig. 24). At the ultrastructural level it was identified on the plasma-membranes and the endoplasmatic reticulum, including the subsurface cisterns (Maeda et al., 1989). Notably Purkinje cells of'staggerer' mice, that are defective in synaptic contacts of parallel fibers and lack dendritic spines, do not express P400-immunoreactivity, whereas P400-immunoreactivity was found at 'normal' levels in ectopic Purkinje cells of 'reeler' cerebellum (Mariani et al., 1977; Mikoshiba et al., 1980; Maeda et al., 1989) (Fig. 23), and in the few remaining Purkinje cells of 'pcd' mutant mice. Cloning of the P400 protein cDNA revealed that it was identical to the IP 3 receptor protein, as well as the Purkinje cell-specific PCPP-260 protein isolated by Walaas et al.
Fig. 21. IP3-3-kinase immunoreactivity in the rat cerebellum. Electron micrograph showing intense immunoreactivity in the dendritic spines of Purkinje cells. Bar 2 r Yamada et al. (1992).
27
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 22. Localization of inositol 1,4,5-triphosphate receptor with PCD6 antibody in frozen sections of rat cerebellum by immunofluorescence. Sagittal section of the cerebellar cortex. Small arrows in the granule cell layer (GL) point to segments of immunoreactive axons which represent recurrent collaterals of Purkinje cell axons. Mignery et al. (1989).
(1986) and the PDC6 protein of Nordquist et al. (1988). The localization of P400 (= IP3 receptor) mRNA in Purkinje cells was confirmed by in situ hybridization (Furuichi et al., 1989) (Fig. 25). The IP3 receptor was purified from rat cerebellum as a protein with a molecular weight of 260 kDa (Supattapone et al., 1988). The primary structure of the mouse IP3 receptor protein, and its partial homology to the skeletal muscle ryanodine receptor were elucidated by Mignery et al. (1990). The IP3 receptor is composed of four identical subunits of a molecular weight of 320 kDa, and forms a calcium-permeable channel (Maeda et al., 1991). Three additional cDNAs encoding for the IP3 receptor, 28
The cerebellum." chemoarchitecture and anatomy
Ch. I
named IP~R-II, III and IV, were identified by Sfidhof et al. (1991) and Ross et al. (1992), but were not found to be expressed at significant levels by Purkinje cells. The presence of the IP~ receptor in Purkinje cells was confirmed immunocytochemically. In immunocytochemical studies with gold-conjugates, that allow precise ultrastructural localization of the immunoreactivity, it was shown that gold particles were located on membranes of the endoplasmatic reticulum in somata, dendrites, dendritic spines and axons of the Purkinje cells (Mignery et al., 1989; Ross et al., 1989a; Sharp et al., 1993a,b) (Fig. 26). Immunolabelling predominated in the smooth-surfaced endoplasmatic reticulum, including the subsurface cisterns, but was also found on portions of the perinuclear and rough endoplasmatic reticulum, and on the cis-cisternae, but not the intermediate and trans-cisternae, of the Golgi apparatus. IP~-receptor immunoreactivity was also observed in a subpopulation of spherical or elongated, membrane-bound structures, named calciosomes (Volpe et al., 1989), that are present throughout the cytoplasm of the Purkinje cells (Volpe and Villa, 1991; Nori et al., 1993). Strong immunoreactivity for the IP~ receptor was found on stacks of flattened cisternae of the endoplasmatic reticulum (Otsu et al., 1990; Satoh et al., 1990; Takei et al., 1992, 1994). The labelling on the cisternal stacks was mostly located in the spaces between the cisternae and between the cisternae and the plasmalemma or mitochondria (Satoh et al., 1990; Takei et al., 1992, 1994). It should be noted that the amount of cisternal stacks in Purkinje cells may depend on the conditions of perfusion fixation. The presence of cisternal stacks in healthy Purkinje cells, therefore, has been disputed (Takei et al., 1994).
9 .:.
,~.~.
9~ .
...,.:.'.,,y'-:L.
-~,
'
..
' ,,.~
9. ~
~-
.
~:-~
.
.
..
.,
.~
"...~
',."
.'~ . . . .
;
~
9
-,~--~i.i'.,~
.
.... ..,
.
/
~e,'"
9~ - ' ~ ' ; C. ~
.
-., -,,,.~
.
. .r#:-[,#'.~':,.
9 -;-
--~,.2,.:
~
~ ...~,
.
.
.,,: ..
..,-
.
. . . . , : ' , , ,.:
.
~,]-~_.~-~:
..,. r.:,.-
.
....- ~ i ~ '
q~.
.~
-
:.
..
.,, . . . . .
'
..~.
~ ~,~-,.,:.
Fig. 23. Section of reeler m u t a n t m o u s e cerebellum stained with m o n o c l o n a l a n t i b o d y 4C11 against the P400 protein. N o t e stained Purkinje cells in the cortex (CX) and in the central mass o f dislocated cells (DP). Bar = 200/~m. M a e d a et al. (1989).
29
Ch. I
J. Voogd, D. Jaarsma and E. Marani
9 ~~
-~A~
~
'
~'~-
E
~,:.. ~....
., t.
..r. .... .
~!~
.
.
.
i~ ~i!,,~~............,~ ~i~ii~ 'D
.
9
~
~.
~
"
~~ H
"
Fig. 24. Sagittal sections of mouse cerebella of various ages stained with a monoclonal antibody (4C 11) against the P400 protein.The samples were from (A) postnatal day O (PO), (B) P3, (C) P5, (D) P7, (E) P10, (F) P15, and (G) P20 cerebellum. A section from a P20 old mouse cerebral cortex did not react with same antibody (H). Magnification 100• Maeda et al. (1989).
30
The cerebellum." chemoarchitecture and anatomy
Ch. I
Nevertheless the formation of stacks of cisternae of the endoplasmatic reticulum could be induced by overexpression of IP3 receptors in fibroblasts, which indicates that cisternal stacks may exist as special organelles related to the IP3 receptor (Takei et al., 1994). The localization of the IP3 receptor has been compared to the localization of other luminal or membrane components of the endoplasmatic reticulum related to Ca 2+ homeostasis. The membrane pump CaZ+-ATPase, immunolabelled with antibodies against cardiac CaZ+-ATPase, was found to be located in regular cisternae of the endoplasmatic reticulum, the lateral tips of cisternae of the Golgi complex and in calciosomes of Purkinje cells (Kaprielian, 1989; Michelangeli et al., 1991; Villa et al., 1991; Takei et al., 1992) (Fig. 27). Distal axons of Purkinje cells, however, lacked CaZ+-ATPase immunoreactivity (Takei et al., 1992). Appreciable levels of calsequestrin, the main intraluminal calcium-binding protein of muscle, were present in Purkinje cells of the chicken. Calsequestrin-immunoreactivity was present over the lumen (Villa et al., 1991) and membranes (Takei et al., 1992) of stacked and isolated cisternae of the endoplasmatic reticulum and in a subpopulation of calciosomes (Volpe et al., 1988; Volpe and Villa, 1991). Mammalian Purkinje cells do not have calsequestrin but, instead, express calretic-
d
b
9
~..
ML PL
]cL " ;7-'.-11
Fig. 25. Localization of the P400-specific mRNA by in situ hybridization, a. Autoradiograph of a sagittal section of mouse cerebellum, b. Higher magnification of a. ML, molecular layer; PL, Purkinje cell layer; GL, granular layer. Furuichi et al. (1989).
31
Ch. I
~.
J. Voogd, D. Jaarsma and E. Marani
:.5.. ,~ -.
Fig. 26. Electron-microscope immunocytochemical localization of InsP3 receptor in Purkinje cells of rat cerebellum using pre-embedding avidin-biotin labelling and InsP3 receptor antiserum. A, B. Nuclear membrane and some, but not all, portions of endoplasmic reticulum (ER) are labelled. C. Higher magnification of the dendritic pole of a labelled Purkinje cell. Note the unlabelled ER very close to labelled ER. D. Labelled portions of endoplasmic reticulum (L-ER) immediately subjacent to an unlabelled presynaptic terminal (U-T). Cell membrane indicated with triangles in (D) and (E). E. InsP3 receptor antiserum. Labelled portions of ER near plasma membrane, but not directly subjacent to presynaptic terminal. F. Preimmune serum. No specific label is present, even though the section is very close to the surface of the vibratome section. Abbreviations: L-ER, labelled endoplasmic reticulum; L-G, labelled Golgi apparatus; L-NM, labelled nuclear membrane; U-ER, unlabelled endoplasmic reticulum; U-G, unlabeled Golgi apparatus; U-M, unlabelled mitochondrion; U-NM, unlabelled nuclear membrane. Scale bars for all panels 1 r (Ross et al., 1989a).
ulin (Treves et al., 1990). Calreticulin-immunoreactvity was located in stacks of rough and smooth endoplasmatic reticulum in rat Purkinje cells (Nori et al., 1993). Calsequestrin and calreticulin are not exclusively present in Purkinje cells, but also in other cell types of the cerebellar cortex. 3.1.5. Protein kinase C in Purkinje cells
Protein kinase C (PKC) plays an important role in the control of several cellular processes, such as the short-term modulation of membrane excitability and transmitter release, positive or negative interaction with the conductance through various ion channels and the regulation of gene expression and cell proliferation (Shearman et al., 1989, 1991; Farago and Nishizuka, 1990; Nishizuka et al., 1991). PKC, that through phosphorylates multiple target protein including neurotransmitter receptors, and has been implicated in long-term depression (LTD) of glutamate sensitivity of Purkinje cells (Cr~pel and Krupa, 1988). Breakdown of PIP2 by phospholipase C (see Section 3.1.4) in Purkinje 32
The cerebellum." chemoarchitecture and anatomy
Ch. I
a
b
C
Fig. 27. Immunofluorescence localization of the cerebellar Ca2+-ATPase in a transverse cryosection of adult chicken cerebellum. CaS/CI-IgG localizes the Ca2+-ATPase to the Purkinje cell bodies in the Purkinje layer (b), and the dendritic trees in the molecular layer (a). Very faint immunofluorescence was detected in the granule cell layer (c). Bar 50/~m. Kaprielian et al. (1989).
cells can activate PKC through the production of DAG and the mobilization of C a 2+ from the endoplasmatic reticulum. Alternative routes for the production of DAG and the mobilization of C a 2+ from extracellular sources are available (Nishizuka et al., 1991). Three isoenzymes of PKC have been distinguished on the basis of the analysis of the sequence homology of complementary DNA clones from different sources. The PKCtypes I, II and III of Huang et al. (1987a,b) are the products of the 7', fl and ~ genes respectively (Ono et al., 1987; Nishizuka, 1988). The PKC fl isoenzyme occurs in two forms, flI and flII, generated through alternative splicing (Ono et al., 1987; Nishizuka, 1988; Saito et al., 1989; Shimohama et al., 1990; Farago and Nishizuka, 1990). PKC ~, fl and 7' are calcium-dependent forms. In addition, calcium-independent isoenzymes of PKC have been identified: ~, e, e' and ( (Ono et al., 1988). Non-specific antibodies against PKC were found to strongly immunostain Purkinje cell perikarya, dendrites and axons (Mochly-Rosen et al., 1987; Kitano et al., 1987; Saito et al., 1988). Immunocytochemical studies with subtype specific antibodies and in situ hybridisation histochemistry have shown that several PKC subtypes are located in Purkinje cells (Figs 28 and 29, Table 1) (Brandt et al., 1987; Huang et al., 1987a,b, 1988, 1991; Ase et al., 1988; Hashimoto et al., 1988; Hidaka et al., 1988; Kose et al., 1988; Shimohama et al., 1990; Wetsel et al., 1992; Chen and Hillman, 1993a; Garcia et al., 1993; Merchenthaler et al., 1993). PKC~' immunoreactivity occurs at high levels in both the somatodendritic and axonal domains of Purkinje cells, and is absent from other cell types of the cerebellar cortex. Immunoreactivity for PKC ~ is also present in Purkinje 33
Ch. I
J. Voogd, D. Jaarsma and E. Marani
cells. PKC d-immunoreactive Purkinje cells are distributed in immunopositive and immunonegative columns (Fig. 133) (Chen and Hillman, 1993a). According to Wetsel et al. (1992) Purkinje cells were stained with antisera against PKC e, but Chen and Hillman's (1993a) found Purkinje cells to be unlabelled for PKC e. PKC/6 and e' were not located in Purkinje cells (Table 1). 3.1.6. cGMP; cGMP-dependent protein kinase and nitric oxide synthase in Purkinje cells Purkinje cells are the only cerebellar cell type containing cyclic guanosine 3',5'-monophosphate (cGMP)-dependent protein kinase (cGK) (Walter et al., 1981; Walter, 1984; Lohmann et al., 1981; De Camilli et al., 1984; Wassef and Sotelo, 1984). cGK-immunoreactivity is present throughout the entire Purkinje cell, including its dendrites and its axon (Fig. 30). During development Purkinje cells display a heterogeneity in their expression of immunoreactivity for cGK (Wassef and Sotelo, 1984, rat; Levitt et al., 1984, monkey) (see Section 6.2.). A 23 kD protein, which is likely to be a substrate of cGK was found to be concentrated in Purkinje cells (Walter, 1984; Nairn and Greengard, 1983). Immunoreactivity for guanylate cyclase, the synthesizing enzyme of cGMP, was
Fig. 28. Developmental expression of protein kinase C (PKC) isoenzymes in rat cerebellum. Immunofluorescent staining of cerebellar cortex by antibodies specific for PKC 1, corresponding to PKC~" (panels A, B and C), PKCfl (panels D, E and F) and PKC~ (panels G, H and I). Sagittal sections of cerebellum of 1-week-old (A, D and G), 2-week-old (B, E and H) and 3-week-old (C, F and I) rats were used. PKCz- antibody stained mainly the Purkinje cell bodies and dendrites throughout the development. PKCfl antibody stained the cerebellar granule cells in the external germinal layer (EGL) of the 1- and 2-week-old rats and mainly the granular layer of the 3-week-old rats. PKC~ antibody stained both granule cells and Purkinje cells throughout the development. Huang et al. (1991).
34
The cerebellum." chemoarchitecture and anatomy
Ch. I
....
.~:
.,
~
?~
.~.~'
, . .
?:.:
T".~":... ~ ~
....
9
.
.
.
.
..
.
9
"
Fig. 29. Immunostaining for different isozymes of PKC in the rat cerebellum. PKC0~ (A) is present in Purkinje cells (P). The dendrites of these cells can be followed as far as the top of the molecular layer (m). The granular layer (g) and the white matter (w) are not stained. To avoid crowding, the abbreviations for layers of the cerebellum are indicated only in (A); however, the layer of Purkinje cells (P) is indicated in each figure for orientation. PKCfl (B) and PKCflI I (C) are present only in cells of the granular layer. PKCg is present in Purkinje cells and Bergmann glial cells in the molecular layer (D). Not only the perikarya but also the dendrites of Purkinje cells in the molecular layer and their axons in the granular layer are immunopositive. The antiserum for PKC~ stained Purkinje cells (E) and presently unidentified cells below the unstained Purkinje cells (F). The dorsally located folia contain mainly unstained Purkinje cells. Their axonal origin is surrounded by immunopositive cells. In the basal folia, the Purkinje cells are immunostained. PKCe is present in Purkinje cell (G), whereas PKCe' is present in cells in the molecular and granular layers and in the nerve fibers surrounding the unstained Purkinje cells (H). Antiserum against PKC~" stained only Purkinje cells in the cerebellum (I). Bar 100 •m. Wetsel et al. (1992).
35
Ch. I
J. Voogd, D. Jaarsma and E. Marani
localized in Purkinje cells, but does also occur in other cell types of the cerebellar cortex (Zwiller et al., 1981; Ariano et al., 1982; Nakane et al., 1983; Poegge and Luppa, 1988). cGMP was, however, found to be absent from rat Purkinje cells, using antibodies against conjugates of cGMP in combination with sodium nitroprusside-stimulation of cGMP synthesis. Prominent cGMP-immunoreactivity within the molecular layer was detected in Bergmann glial cells (Fig. 56) (Berkelmans et al., 1989; De Vente et al., 1989, 1990). Soluble guanylate cyclase is activated by nitric oxide (NO) (see Section 3.4). NO has been implicated in the generation of long term depression (LTD) of parallel fibermediated EPSP's in Purkinje cells. LTD can be prevented by the application of haemoglobin that absorbs NO, or by the inhibition of NO synthesis (CrGpel and Jaillard, 1990; Shibuki and Okada, 1991; Ito, 1991). However, nitric oxide synthase, the synthesizing enzyme of NO, appears to be absent from the Purkinje cell (Section 3.4.).
3.1.7. Calcium-binding proteins in Purkinje cells Calbindin-D28K, parvalbumin and calmodulin are cytosolic, calcium-binding proteins of the EF-hand family (see Baimbridge et al., 1992 and Andressen et al., 1993 for reviews), that are present in high amounts in Purkinje cells. Calretinin is a calciumbinding protein closely related to calbindin-D28K, that is absent from the Purkinje cells, but present in other neurons and in afferent mossy and climbing fibers of the cerebellar cortex (Rogers, 1989; Arai et al., 1991; RGsibois and Rogers, 1992; Floris et al., 1994). One of the calcium-binding proteins, the 28 kDa vitamin-D-dependent calcium-binding protein (calbindin-D28K), occurs in most, if not all, Purkinje cells in rat and chicken cerebellum (Lawson, 1981; Roth et al., 1981; Jande et al., 1981a,b; Baimbridge and Miller, 1982; Legrand et al., 1983; Schneeberger et al., 1985; Kosaka et al., 1993; Amenta et al., 1994). Its presence in soma, dendrites and axon was demonstrated with polyclonal and monoclonal antibodies raised against calbindin-D28K (Fig. 31A). Its exclusive presence in the cerebellum in Purkinje cells was confirmed with in situ hybridization with cDNA probes in rat and mouse (Nordquist et al., 1988; Iacopino et al., 1990; Abe et al., 1992a; Kadowaki et al., 1993). According to Garcia-Seguera et al. (1984) only 74% of the rat Purkinje cells was immunoreactive for a polyclonal antibody raised against chick duodenal calbindin-28K. This antibody also stained Golgi cells in the granular layer in rat and human cerebellum (Fournet et al., 1986). Developmental gradients in the expression of immunoreactivity for calbindin-28K by Purkinje cells were studied by Legrand et al. (1983) and Wassef et al. (1985) (see Section 6.2.). TABLE
1. Immunoreactivities o f P K C in cerebellar neurons
Isoenzymes
P u r k i n j e cells
Basket &
G r a n u l e cells
Cerebellar nuclei
+
++
+
s t e l l a t e cells Alpha
++
Beta
-
+
++
+
Gamma
+++
-
-
-
Delta
+++
+++
-
-
Epsilon
-
+
++
++
Zeta
++
+
++
++
Chen and Hillman (1993a)
36
The cerebellum." chemoarchitecture and anatomy
Ch. I
P 21
Fig. 30. A. Frontal section through the cerebellum and attached brainstem of an adult rat. All the Purkinje cells are stained by cyclic 3',5'-guanosine monophosphate-dependent protein kinase (cGK) antiserum, including their dendrites in the molecular layer and their axon terminals in the deep nuclei and in the brainstem (arrow). Bar = 1 mm. B. Higher magnification of the neurons indicated by an arrow head in A. Like a few other isolated labelled cells found in variable locations, these cells are considered as ectopic Purkinje cells. Bar = 50/lm. C. cGK immunoreactive neuron in the cerebellum of 1 day-old rat. This ectopic Purkinje cell is located in the white matter and its appearance mimics that of 1-day-old Purkinje cells as visualized in Golgi impregnated material. Bar = 25 ~m. Wassef and Sotelo (1984).
Calmodulin-immunoreactivity was observed both in Purkinje cells and in cells of the cerebellar nuclei of the rat (Lin et al., 1980; Means and Dedman, 1980; Seto-Oshima et al., 1983, 1984). During postnatal development calmodulin-immunoreactivity was transiently present in the inner part of the external germinative layer and in fibers in the white matter of P3-P11 rat pups (Seto-Oshima et al., 1984). Polyclonal antibodies against parvalbumin stain all Purkinje cells and stellate and basket cells in the molecular layer of rat and avian cerebellum (Figs 31B and 32) (Celio and Heizmann, 1981; Heizmann, 1984; Braun et al., 1986; Endo et al., 1985; Schneeberger et al., 1985; Seto-Oshima et al., 1983; Rogers, 1989; Kosaka et al., 1993). The localization of parvalbumin in Purkinje, stellate and basket cells was confirmed in the rat with non-radioactive in situ hybridization (Kadowaki et al., 1993). Parvalbumin 37
Ch. I
J. Voogd, D. Jaarsma and E. Marani
. .
Fig. 31. A. Calbindin-D28k immunoreactivity. B. Parvalbumin-immunoreactivity in rat cerebellar cortex. Purkinje cells react with both antibodies; arrows in B indicate parvalbumin- immunoreactive stellate and basket cells. Bar - 50/zm. Courtesy of Dr. M.P.A. Schalekamp.
supposedly occurs in GABAergic neurons (Celio and Heizmann, 1981) and/or neurons with characteristically high firing rates (Karmy et al., 1991). Karmy et al. (1991) studied the co-localization of parvalbumin and cytochrome oxidase, as an indicator of metabolic activity, in many regions of the brain. They found only weak immunoreactivity with antibodies against cytochrome oxidase in parvalbumin immunoreactive Purkinje cells of the rat. A developmentally regulated polypeptide (PEP-19), that is a presumptive neuronspecific calcium binding protein, was identified in adult and neonatal rat cerebellum and its amino acid sequence was determined (Ziai et al., 1986). PEP-19-like immunoreactivity is expressed by Purkinje cells and by the cartwheel cells of the dorsal cochlear nucleus of the mouse (Mugnaini et al., 1987). Berrebi et al. (1991) drew attention to the expression of PEP-19, CaBP and other Purkinje cell markers (cerebellin, L7: see below) by bipolar cells and other neurons of the retina.
3.1.8. Other specific biochemical markers for Purkinje cells Several polypeptides, that are present in all Purkinje cells, but not in other cells of the cerebellum, have been mentioned in the previous sections of this chapter. They include the IP3 receptor (identical to the P400 protein and to the PCPP-260 protein of Walaas et al., 1986) (see Section 3.1.4), IP3-3-kinase (Section 3.1.4), cGMP-dependent protein kinase (Section 3.1.5), PEP-19 and calbindin-D28K (Section 3.1.7). Two other 38
The cerebellum." chemoarchitecture and anatomy
Ch.I
proteins, cerebellin and L-7 that occur in all Purkinje cells, are dealt with in this section. Other proteins only occur in certain subpopulations of Purkinje cells. Zebrin I and II (Hawkes et al., 1985) are the prototypes of this group. The restriction of the Zebrins to a subpopulation of Purkinje cells is the more remarkable because they are originally present in all Purkinje cells of rat neonates (Leclerc et al., 1988). The developmental histories of cGMP-dependent protein kinase, calbindin-D28K and L-7 are quite different, in that these proteins make their first appearance in subpopulations of fetal Purkinje cells and only in later stages become expressed by all Purkinje cells of the cerebellum (Wassef and Sotelo, 1984; Smeyne et al., 1991) (Section 6.2.). Purkinje cell-specific markers include several glyco- and phosphoproteins, peptides, antigenic determinants that have not been identified or determinants that Purkinje cells share with other, non-cerebellar cell types. One of the first sera specific for rat Purkinje cells was obtained, using immunohistochemical screening, by Woodhams et al. (1979), but the antigen corresponding to this antibody has not been identified. Reeber et al. (1981) isolated a Purkinje cell specific 24 kDa glycoprotein from rat, that was present (Reeber et al., 1981) throughout the whole somatodendritic extent of the Purkinje cells, associated with the plasma membrane, as well as with the rough endoplasmatic reticulum and polysomes, the cytoplasmic side of the nuclear envelope and subsurface cisterns (Langley et al., 1982). Visinine, a soluble, 24 kDa protein, isolated from chicken retina, was found to be an exclusive marker for Purkinje cells in rat cerebellum (Yoshida et al., 1985). Specific staining of Purkinje cells was also found with monoclonal antibodies directed against human T cells (Garson et al., 1982), against certain cytoplasmic antigens in Purkinje cells (Weber and Schachner, 1982) and against antigenic determinants on trypanosomes (Wood et al., 1982). One of the antibodies (UCHT 1), isolated by Garson et al. (1982) is remarkable because its antigen is not present in Purkinje cells from 'staggerer' mutant mice (Caddy et al., 1982), a property the UCHT 1 antigen shares with the IP3 receptor protein (Section 3.1.4). One group of Purkinje cell-specific markers, the cerebellins, has been studied in more detail. A Purkinje cell-specific hexadecapeptide called 'cerebellin' and its metabolite, des-Serl-cerebellin were isolated and sequenced by Slemmon et al. (1984). Cerebellin immunoreactivity as studied with polyclonal antibodies in rat was found in soma and dendrites of nearly all Purkinje cells, but was absent beyond the initial axon segment (Slemmon et al., 1984). Cerebellin-immunoreactivity could also be demonstrated in cerebella of different species, including human and chick (Morgan et al., 1988), and in cartwheel cells and basal dendrites of pyramidal neurons of the dorsal cochlear nucleus (Fig. 33) (Mugnaini and Morgan, 1987). Cerebellin differs from most other markerproteins of Purkinje cells in being absent from other sites in the CNS, including the retina (Berrebi et al., 1991). Slemmon et al. (1988) and Morgan et al. (1988) concluded from an analysis of cerebellin immunoreactivity in Purkinje cells of different mutant mice with a varying loss of the granule cells, that the amount of cerebellin is correlated with the formation and the number of parallel fiber-Purkinje cell synapses. L-7 is a protein specific for Purkinje cells. Labelling with polyclonal antibodies against predicted L-7 sequences was present in somata, including the nucleus, in dendrites and dendritic spines, and in axon and axon terminals of Purkinje cells. All Purkinje cells, but no other types of cerebellar neurons appeared to be labelled (Berrebi and Mugnaini, 1992). The expression of the L-7 gene by all adult Purkinje cells of the rat cerebellum was reported by Nordquist et al. (1988, their PCD5 clone), Oberdick et al. (1990) Vandaele et al. (1991, their Purkinje cell protein-2) and Smeyne et al. (1991). According to Oberdick et al. (1990) and Berrebi et al. (1991) the L-7 gene is also expressed by retinal 39
Ch.I
J. Voogd, D. Jaarsma and E. Marani
Fig. 32. Parvalbumin immunoreactivity in the developing cerebellar cortex of the zebra finch. A. Incubation day D 16: Clusters of labelled Purkinje cells of varying staining intensity. Stained Purkinje cells axons are seen in the internal granular layer (IGL). Note the areas containing unstained or only slightly stained cells and the dot-like staining pattern in the external granular layer (EGL). B. Adult: The dendrites of the Purkinje cells have reached the cerebellar surface and are now fully branched. Between them many immuno-stained basket and stellate cells are visible. Parvalbumin immunoreactivity in Purkinje cell axons is no longer visible except for a few fragments lying in the internal granular layer (IGL). The layer of Purkinje cells is still interrupted by parvalbumin immunonegative areas. Calibration bar in A - 50/lm, in B = 100/~m. Braun et al. (1986).
bipolar cells. The initial expression of the L-7 gene by zonally distributed Purkinje cells during prenatal and early postnatal development was studied by Vandaele et al. (1991), Smeyne et al. (1991) and Oberdick et al. (1993) (see also Section 6.2.). 40
The cerebellum." chemoarchitecture and anatomy
Ch. I
Several other markers are only present in zonally distributed subpopulations of Purkinje cells (see also Section 6.1.3.). The monoclonal antibody B1 of Ingram et al. (1985) was raised against rat embryonic forebrain membranes. Purkinje cells in broad parasagittal bands, alternating with B 1-negative zones, were immunoreactive in the cerebellum of Macaca fascicularis. Other neurons in the molecular layer and cells of the cerebellar nuclei were also stained by this antibody. A similar pattern of B l-immunoreactivity was present in the cerebellum of the rat. The monoclonal antibody mabQ 113 was developed, specified and used in anatomical studies by Hawkes et al. (1985), Hawkes and Leclerc (1986, 1987), Hawkes and Gravel (1991), Hawkes (1992) and Leclerc et al. (1992). It is directed against a 120 Kda protein (Zebrin I); the function of this protein is still unknown. A specific subpopulation of Purkinje cells displays immunoreactivity for Zebrin I in their dendrites, soma, axon and axon terminals. Zebrin I-positive and negative Purkinje cells are distributed in parasagittal bands (Fig. 34) (see also Section 6.1.3.). Ultrastructurally Zebrin I-immunoreactivity in rat Purkinje cells is localized in the cytosol. It is absent from membrane-bound organelles such as the mitochondria and the synaptic vesicles. In large dendrites reaction product is associated with microtubuli, in spines it is located at the postsynaptic densities. An antibody raised against the cerebellum of the weakly electric fish Apteronotus (anti-Zebrin II: Brochu et al., 1990) recognizes the same Purkinje cells as anti-Zebrin I in the cerebellum of the rat, and is effective in staining these neurons in a large number of other species such as the opossum (Fig. 137). The epitope of the Zebrin II antibody is associated with a 36 kDa polypeptide identified as the glycolytic enzyme aldolase C. In situ hybridization of Zebrin II mRNA showed a strong signal in mouse Purkinje cells with normal regional heterogeneity (Hawkes, 1992; Ahn et al., 1994). Rat Purkinje cells containing low affinity nerve growth factor receptor protein (Sotelo and Wassef, 1991; Dusart et al., 1994) (see Section 3.1.10 and Fig. 38C,D), PKC delta (see Section 3.1.10 and Fig. 133), or the monoclonal antibody B30 of Stainier and Gilbert (1989), that recognizes two minor gangliosides, show the same distribution as Zebrin-stained Purkinje cells. Although the distribution of the enzyme 5'-nucleotidase in the molecular layer of rat and mouse cerebellum (Scott, 1963; Marani, 1982a,b) is identical to that of the Zebrins (Eisenman and Hawkes, 1993) (Fig. 135), it may be located in Bergmann glial and not in Purkinje cells (see Section 3.5.). Several proteins are distributed in more or less complementary patterns, either in Zebrin-negative Purkinje cells (Ppath, HNK, cytochrome oxidase) or in Bergmann glia (3a-fucosyl-N-acetyl lactosamine [FAL], glycogen phosphorylase). The antibody P-path is directed against acetylated gangliosides (Edwards et al., 1989, 1994; Leclerc et al., 1992) and reacts with Zebrin-negative Purkinje cells in mouse cerebellum (Fig. 134). The localization of cytochrome oxidase was described by Hess and Voogd (1986), Leclerc et al. (1990) and Harley and Biejalew (1992) in the cerebellum of macaques, the squirrel monkey and the rat. The localization of HNK was studied by Eisenman and Hawkes (1993) in the mouse. The FAL-epitope (Fig. 94; Bartsch and Mai, 1991) and the enzyme glycan phosphorylase (Marani and Boekee, 1973; Harley and Bielajew, 1992) have been located in subsets of mouse Bergmann glial cells, that are distributed in a complementary manner with respect to the Zebrin pattern. Gangliosides are glycolipids, concentrated in the outer layer of neural plasma membranes. Biochemical analysis showed a correlation between the selective degeneration of Purkinje cells in pcd and nervous mutant mice with the loss of the ganglioside GT~A. GT~A was more concentrated and the ganglioside GD~A was diminished in weaver mutant mice with a selective loss of the granule cells (Seyfried et al., 1983, 1987; Marani 41
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 33. Light photomicrographs showing cerebellin immunoreactivity in rat cerebellum (A-C) and the dorsal cochlear nucleus (D-F) in parasagittal sections. A. Cerebellar hemisphere with part of the underlying dorsal cochlear nucleus (DCoN, arrowhead). CN, cerebellar nuclei. B. Immunostaining in DCoN. The cell bodies of cartwheel neurons in the superficial layers (layers 1 and 2) of the DCoN and the plexus in the deeper region (layer 3) predominate. The plexus is most dense in the upper portion of the deep region, which may correspond to layer 3 of the feline nuclei, a zone that contains the basal dendritic arbors of the bipolar pyramid neurons, one of which is indicated by an arrow. C. Immunoreaction product is present in Purkinje cell body and main dendrites. D. In the axon, immunostaining is restricted to the initial axon segment (arrowhead). E. Three subependymal displaced Purkinje cells in DCoN. Smaller cell bodies of several cartwheel neurons (arrowheads) are also shown. F. Portion of the ventral cochlear nucleus in which immunostaining is restricted to rare cartwheel cell bodies (arrowheads) displaced in the superficial granular layer. Bars in A and B = 0.5/~m, in C-F = 50 r Mugnaini et al. (1987).
a n d M a i , 1992). A n o t h e r g a n g l i o s i d e , GD3, was localized in i m m a t u r e P u r k i n j e cells o f the rat, u s i n g a m o n o c l o n a l a n t i b o d y (Fig. 35). I m m u n o r e a c t i v i t y d i s a p p e a r e d f r o m the cell b o d y in the adult, b u t r e m a i n e d p r e s e n t in the m o l e c u l a r layer ( R e y n o l d s a n d Wilkin,
42
The cerebellum." chemoarchitecture and anatomy
Ch. I
.,.....
.".: ,, ~
, ~ ;~,
'
; .
,
. -i:
;~:57
'";'
:'~'~":
'2., ~
Fig. 34. 50/lm horizontal sections through the cerebellar cortex of the rat at postnatal day 25 to show the distribution of mabQ113 (Zebrin I) immunoreactivity. A. The peroxidase reaction product is confined exclusively to a subset of Purkinje cells that are distributed symmetrically into parasagittal compartments in both the vermis and hemispheres. Labelling of the bands of Zebrin I-immunoreactive Purkinje cells P l+ to P7+ according to Hawkes and Leclerc (1987). Scale bar = 500/~m. B. A higher-power view of P5 + and P6 + of the posterior lobe hemisphere, in the lobules bordering the intercrural fissure. Immunoreactivity is seen to extend throughout the Purkinje cell, and no other cell types in the cerebellum are stained. Scale bar = 200/zm. C. In addition to the regular band display, additional narrow 'satellite' bands are also common. The arrowheads indicate two such satellites in the posterior lobe vermis. Scale bar = 100/lm. Leclerc et al. (1988).
1988). Levine et al. (1986), who used another monoclonal antibody against GD3, found immunoreactivity of reactive astrocytes in mouse mutants, but failed to observe a reaction within the Purkinje cells. These different results probably are due to differences in fixation (Reynolds and Wilkin, 1988). 3.1.9. Cytoskeleton and metabolism of Purkinje cells The DNA content of mature Purkinje cells is high. Feulgen-DNA or propidiumiodideDNA reveal hyperdiploid values (Bernocchi, 1986; Bernocchi et al., 1986). Purkinje cells stand out by their high content of enzymes, mostly dehydrogenases (Adams, 1965). Their content of the glycolytic enzyme enolase is low (Pelc et al., 1986; Vinores et al., 43
Ch. I
J. Voogd, D. Jaarsma and E. Marani
1984). However, Purkinje cells of the human cerebellum stand out from other nerve cells by their high content of aldolase-C (Royds et al., 1987). Purkinje cells do not react with antibodies against the phosphorylated forms of the 70, 150 and 200 kDa neurofilament proteins (Pelc et al., 1986; Matus et al., 1979; Marc et al., 1986; Langley et al., 1988). The phosphorylated form of the 200 kDa protein is present in axons in the granular layer, that were identified as Purkinje cell axons by Marc et al. (1986) and as mossy fibers by Langley et al. (1988), both in the rat (Fig. 36). The non-phosphorylated form of the neurofilament proteins was found to be present in the entire Purkinje cell with the exception of distal dendrites. According to Marc et al. (1986) the protein is present as filamentous aggregates. Langley et al. (1988) stated that a monoclonal antibody against the non-phosphorylated form of the 200 kDa protein is present in soma and dendrites as patches of diffuse immunoreactivity without a filamentous substructure. In Friedreich's ataxia neurofilament, mainly the phosphorylated form, is expressed by human Purkinje cells within their soma and dendrites (Marani, unpublished results) (Fig. 37). The process of endocytosis in Purkinje cell has been studied in relation to synaptogenesis of the Purkinje cell dendrites. Glycoproteins located on the parallel fiber are also pinocytosed into the Purkinje cell. Lysosomal action degradates these glycoproteins. In this process alpha-D-massosidase plays an important role, which is selectively present in the Purkinje cell dendrites (Dontenwill et al., 1983). Other glycoproteins, like K+Na+ATP-ase are not taken up, indicating a receptor-mediated recognition of some glycans of the glycoproteins. The specificity of the pinocytosis for certain molecules suggests that this recognition is the preliminary event in the establishment of Purkinje cell synapses.
3.1.10. Nerve growth factor and nerve growth factor-receptor protein in Purkinje cells Nerve growth factor-like immunoreactivity was present in Purkinje cell somata and dendrites, with dense labelling in the paraflocculus, and in neurons of the cerebellar nuclei and the lateral vestibular nucleus of rat cerebellum (Nishio et al., 1994). All but a few of the Purkinje cells of the adult rat cerebellum stain with an antiserum against basic fibroblast growth factor. Staining was observed in all cellular compartments (Matsuda et al., 1992). P75 nerve growth factor-receptor protein (NGF-R) is present in developing and adult Purkinje cells. Yan and Johnson (1988) and Cohen-Cory et al. (1989) described and reviewed the development of NGF-R in rat cerebellum. Low affinity NGF-R immunoreactivity has been demonstrated with species-specific monoclonal antibodies in Purkinje cells of adult rats (Pioro and Cuello, 1988, 1990; Pioro et al., 1991; Fusco et al., 1991; Dusart et al., 1994), monkey and human brain (Mufson et al., 1991). Immunoreactivity was present in the somata, dendrites and the proximal axon of the Purkinje cells. Additional immunoreactivity in granule cells was reported by Vega et al. (1994), using Bouin's fixative. NGF-R mRNA is expressed during early development in neurons of the rat external granular layer and in Purkinje cells. It peaks at postnatal day 10 and declines afterwards (Cohen-Cory et al., 1989; Lu et al., 1989) but also can be demonstrated in Purkinje cell somata in adult rodents (Fig. 38) (Koh et al., 1989) and primates (Mufson et al., 1991). NGF-R immunoreactivity was found to be highest in the flocculonodular lobe (Pioro and Cuello, 1988, 1990; Fusco et al., 1991). A distribution with strong expression in the flocculonodular lobe, the ventral parts of the anterior lobe and the lobules VII, VIII and 44
The cerebellum." chemoarchitecture and anatomy
Ch.I
Fig. 35. Double-immunofluorescent staining of 20-day rat cerebellar sections with antibodies to GD 3 ganglioside and glial acidic fibrillary protein (GFAP). Purkinje cell dendrites are intensely GD3-immunoreactive (A) but do not extend to the pial surface, unlike the Bergmann glial fibers (B), which project brightly GFAPimmunoreactive end-feet onto the pial membrane. Scale bar is 35 ~tm. Reynolds and Wilkin (1988).
45
Ch. I
J. Voogd, D. Jaarsma and E. Marani
..
9i:!?~iii~ &
1
F
.
-
.g:
!i . . . .
Immunocytochemical staining patterns of two monoclonal anti-bodies directed against nonphosphorylated and phosphorylated neurofilaments were studied in the cerebellum of developing normal rats. A. Non-phosphorylated neurofilaments on postnatal day 11. B. Day 21. Basket cell axons form a characteristic brush-like plexus around the initial segment of the Purkinje cell axon. C. Phosphorylated neurofilaments on postnatal day 13. D. Postnatal day 21. Stained filaments are restricted to Purkinje cell and basket cell axons and are absent from the Purkinje cell cytoplasm. Calibration bars in A and C 30/lm, in B and D 10 ~tm. Marc et al. (1986).
Fig. 36.
46
The cerebellum." chemoarchitecture and anatomy
Ch. I
IX of the caudal vermis and low activity in the hemisphere, was described by Mufson et al. (1991) for primates and man. The administration of colchicine results in the expression of N G F - R in most cerebellar Purkinje cells (Pioro and Cuello, 1988, 1990; Pioro et al., 1991). Koh et al. (1989) and Fusco et al. (1991) found N G F - R mRNA expression and NGF-R immunoreactivity in adult rat~ to be present in alternating Purkinje cell zones of strong and weak activity (Fig. 38C,D). This zonal pattern was also observed by Pioro and Cuello (1990). Its correspondence to the pattern of mabQ113 (Zebrin) immunoreactive zones (Hawkes and Leclerc, 1987) was noticed by Sotelo and Wassef (1991) and verified by Dusart et al. (1994) in adult rats. Lesions of the white matter, or knife cuts isolating the dorsal portion of the vermis of the rat cerebellum induces NGF-R immunoreactivity in previously unstained Purkinje cells (MartinezMurillo et al., 1993; Dusart et al., 1994).
3.1.11. Immunoreactivity of Purkinje cells in paraneoplastic diseases Specific forms of immunoreactivity of Purkinje cells have been discovered in human paraneoplastic conditions. Subacute cortical cerebellar degeneration in man may be associated with several types of carcinoma (see Vecht et al., 1991 for review). It has been most frequently observed in association with ovarian or endometrial carcinoma, but it also occurs as a rare sequal of small-celled bronchial carcinoma. It is generally characterized by a diffuse or patchy loss of Purkinje cells; granule cells also can be affected (Brain et al., 1951; McDonald, 1961; Brain and Wilkinson, 1965; Schmid and Riede, 1974; Steven et al., 1982). Strong labelling of Purkinje cells and weak staining of the granular layer was observed in sections of human cerebellum with a serum of patient with cerebellar degeneration with Hodgkin's disease using the indirect fluorescent staining procedure (Trotter et al., 1976). Sera of patients with carcinoma of the ovary were found to react with human Purkinje cells and neurons of the cerebellar nuclei using the same method (Greenlee and Sun, 1985). Jaeckle et al. (1985) distinguished a granular cytoplasmic and a diffuse form
9
.
.
.
. .
..
Fig. 37. Expression of phosphorylated neurofilament localization in normal human cerebellar cortex (A) and in Friedreich's disease (B). Note the strong positivity of the white matter and the molecular layers in a case of Friedreich's ataxia. No expression was found in the normal folium that was Nissl counterstained to demonstrate the granular and Purkinje cell layer. M = molecular layer, P = Purkinje cell layer, G = granular layer, F = fiber layer. Marani, unpublished.
47
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 38. A. Nerve growth factor-R (NGF-R) transcripts are localized within Purkinje cells in the paraflocculus of rat cerebellum. B. NGF-R immunocytochemistry shows the perikarya of the Purkinje cells as well as the dense staining of the molecular layer, where the dendritic trees of the Purkinje cells arborize. Arrows in C and D point to parasagittal zones of intense labelling interdigitated with weaker labelling. Bar = 90 r Koh et al. (1989).
of immunoreactivity of human Purkinje cells with sera from patients with cerebellar degeneration suffering from ovarian or breast cancer. The diffuse form of Purkinje cell staining also was observed at higher concentrations with some sera of normal controls. Moreover the diffuse staining is not restricted to Purkinje cells, but also involves stellate, basket and some granule cells (Andersson et al., 1988). Cunningham et al. (1986) further analysed the sera causing granular deposits in the Purkinje cell cytoplasm with immunoblotting of extracts of human Purkinje cells. This so-called anti-Yo serum recognizes a 62 kDa and a 34 kDa protein. Antibodies raised against both proteins react with Purkinje cells in tissue sections (Fig. 39). The strongest reaction was observed for the antibody against the 62 kDa protein. The specificity of this reaction and the presence of anti-Yo immunoreactivity in tumor tissue was demonstrated by Furneaux et al. (1990). The 34 kDa antigen was found to correspond to the c D N A sequence of a clone recognized from a cerebellar expression library by a serum from a patient with paraneoplastic cerebellar degeneration (Dropcho et al., 1987; Furneaux et al. 1989). Other forms of immunoreactivity, with different cerebellar epitopes and a different localization of the immunoreactivity have been described (Tanaka et al., 1986; Smith et al., 1988; Rodriguez et al., 1988; Tsukamoto et al., 1989; Szabo et al., 1991). Differences in the localization of the immunoreactivity and in the characterization of the epitopes may be due to the use of rat cerebellum instead of human cerebellum in testing the sera by Tanaka et al. (1986), Smith et al., (1988) and Tsukamoto et al. (1989). Szabo 48
The cerebellum." chemoarchitecture and anatomy
Ch. I
et al. (1991) isolated the (NuD) neuronal antigen recognized by sera from patients with paraneoplastic encephalomyelitis associated with small-celled bronchus carcinoma. This serum, also designated as anti-Hu, reacts with nuclei of neurons in the CNS, including the cerebellum (Andersson et al., 1988). 3.2. EXCITATORY PATHWAYS The cerebellar cortex is innervated by two types of excitatory afferents, the mossy and climbing fibers, and an intrinsic excitatory fiber system, the parallel fibers. An additional excitatory intrinsic pathway may be formed by unipolar brush cells, that give rise to mossy fiber-like fibers. The excitatory amino acid glutamate is the most likely neurotransmitter candidate for these pathways. An inherent problem in the localization of glutamate as a neurotransmitter is that there is no unequivocal marker for glutamatergic neurons and fibers since glutamate also participates in several metabolic pathways of nerve cells (Van den Berg and Garfinkel, 1971; Fonnum, 1984; Erecinska and Silver, 1990). The identification of glutamatergic pathways, therefore, is based upon a combination of anatomical, biochemical and physiological techniques (Fonnum, 1984). Immunocytochemistry with antibodies against glutamate (Storm-Mathisen et al., 1983) and physiological studies have proven to be particularly fruitful in the identification of glutamate as the neurotransmitter of the cerebellar excitatory pathways. These methods, however, do not totally exclude the possibility that other excitatory amino acids, such as aspartate or homocysteate, also participate as excitatory neurotransmitters. This holds in particular for the climbing fibers that have been frequently proposed to use aspartate as their primary neurotransmitter (see below). A major problem with 'nonglutamate' excitatory neurotransmitter candidates is that, as yet, no vesicular uptake
Fig. 39. Immunofluorescence of rat Purkinje cells with anti-Yo serum of a patient suffering from a cerebellar syndrome with ovarian carcinoma. Courtesy Dr. Ch. J. Vecht and Dr. J.W.B. Moll, Department of Neurology, Erasmus University Medical Center, Rotterdam.
49
Ch. I
9
~-t-
J. Voogd, D. Jaarsma and E. Marani
.,
o~,
~
i ~ ,'.~ ~ , ~ .
o:
p
.
~,-.L,~,~,~ ~ .~,"
~
.
~
.
,~,
~..-~.
.."-.~
system has been isolated for these compounds. Instead, glutamate has been shown to accumulate in synaptic vesicles by a proton-driven vesicle transporter. This vesicle transporter is highly specific for glutamate, and in contrast to the cytoplasma membrane 50
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 40. Immunostaining in rat cerebellar cortex produced by anti-glutamate(Glu) mAb 2D7 (A,C,C',E and F) or by 'anti-GABA' mAb 3A12 (B and D). A, B (overview) and C, D (details) are from a pair of consecutive semithin sections. C and C' are enlargements of areas indicated in A and C, respectively. A and B. Note drastic difference in labelling patterns obtained with the two antibodies, gr, granule cell layer; P, Purkinje cell layer; mol, molecular layer. C. Frame indicating part of area shown enlarged in C'. C and D. Complementary labelling in stellate cells (stars), Golgi cell (arrows), pinceau formed by basket cell terminals (double arrow heads). Mossy fiber terminal-like structures in C (arrow head) fit into glomerular arrangements outlined by dots in D (arrow head). C'. Densely packed puncta probably represent parallel fiber terminals in molecular layer. E. Numerous strongly stained patches (arrow heads) and some fibers (arrow) are reminiscent of mossy fiber terminals. Granule cells with unlabelled nuclei appear less immunoreactive than those in C. F. Large mossy fiber terminal with several synaptic contacts (arrows) shows higher surface density of gold granules (EM immunogold procedure) than another terminal nearby (stars). Bars 100 ~tm in B, D and E, 1 r in F. Liu et al. (1989). (
transporter, does not transport aspartate (reviewed by Nicholls and Atwell, 1990; Jahr and Lester, 1992). 3.2.1. Mossy fibers Glutamate-like immunoreactivity in mossy fibers
Although subpopulations of mossy fibers may be peptidergic or cholinergic (see Sections 3.10. and 6.4.5.), it is now generally accepted that most if not all of the mossy fibers use L-glutamate as their principal neurotransmitter. The glutamatergic nature of mossy fibers has been evidenced with immunocytochemistry with antibodies against glutamateglutaraldehyde (Storm-Mathisen et al., 1983) or carbodiimide-glutamate conjugates (Madl et al., 1986). The rationale of this method is that glutamate, although ubiquiteously present throughout the neuronal cytoplasm at relatively high concentrations (~ 10 mM; Van den Berg and Garfinkel, 1971; Nichols and Attwell, 1990), is particularly enriched in glutamatergic nerve terminals, because of the presence of synaptic vesicles that concentrate glutamate to at least 60 mM. When electron microscopic post-embedding immunogold protocols are employed, quantitative and statistical analysis of the distribution of immunolabelling can be performed (e.g. see Ottersen, 1989). Glutamate immunoreactivity is widely distributed throughout the granular layer, but is enriched over mossy fiber rosettes in rat (Figs 40 and 41) (Ottersen and Storm-Mathisen, 1984a,b, 1987; Ottersen et al., 1987, 1990; Liu et al., 1989; Ji et al., 1991), cat (Somogyi et al., 1986) and monkey (Zhang et al., 1990). Mossy fiber rosettes contained significant higher levels of immunoreactivity than Golgi cell terminals and granule cell dendrites. Enriched glutamate-like immunoreactivity was also demonstrated in anterogradely horseradish peroxidase-wheat germ agglutinin (WGA-HRP) labelled spinocerebellar mossy fiber terminals. Notably, the density of glutamate-like immunoreactivity showed a strong positive correlation with the density of synaptic vesicles in these mossy fiber terminals (Ji et al., 1991). The anterogradely labelled mossy fiber terminals had a similar density of glutamate-like immunoreactivity as other mossy fiber rosettes. Mossy fiber terminals were not enriched in aspartate- or GABA-like immunoreactivities (Ji et al., 1991; Zhang et al., 1990). Data from physiological studies including recent patch-clamp studies are in line with the assumption that glutamate is the neurotransmitter of mossy fibers (Garthwaite and Brodbelt, 1989, 1990; Silver et al., 1992; D'Angelo et al., 1993; Rossi et al., 1995). 51
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Glutamine and glutaminase
Ottersen et al. (1992) quantified the compartmentalization of glutamate and glutamine in the cerebellar cortex of the rat, using post-embedding immunogold immunocytochemistry. They found the highest ratios of glutamate/glutamine in parallel fibers, high ratios in mossy and climbing fibers, low ratios in Purkinje and granule cells and in basket cell and Golgi cell terminals and the lowest ratios in Bergmann glia and astrocytes. This distribution is in accordance with uptake of glutamate from the extracellular space by glial cells, and its conversion into glutamine by the enzyme glutamine synthase, that is exclusively present in glia (Van den Berg and Garfinkel, 1971; Norenberg and MartinezHernandez, 1979; Fonnum, 1984; Erecinska and Silver, 1990). The glutaminase-glutamine loop is closed by diffusion of glutamine into neurons, that contain glutaminase, the enzyme that catalyzes the hydrolytic cleavage of glutamine to form glutamate. Wenthold et al. (1986) and Kaneko et al. (1987, 1989) used antibodies against glutaminase as an alternative approach to determine the cellular localization of glutamate. In the granular layer glutaminase-like immunoreactivity was present in granule cell somata (Wenthold et al., 1986) and in in small clusters, that probably represent mossy fiber rosettes (Fig. 42e) (Wenthold et al., 1986; Kaneko, 1987, 1989). Intense glutaminase-like immunoreactivity was also detected in several precerebellar nuclei, that give rise to mossy fibers, such as the pontine nuclei, the reticular nucleus of the pons, the lateral reticular nucleus, the vestibular nuclei and the external cuneate nucleus (Fig. 42a-d). Neurons in some of these nuclei have also been shown to react with antibodies against conjugates of glutamate (Beitz et al., 1986; Clements et al., 1986; Raymond et al., 1984). Glutamate transporters
The major mechanism by which synaptically released glutamate is inactivated is by highaffinity, sodium-dependent transport (Fonnum, 1984; Nicholls and Attwell, 1990). The sodium-dependent glutamate transporters are present in both neurons and astroglial cells, and have been assumed to be enriched on nerve terminals of glutamatergic axons. [3H]D-aspartate, a metabolically inert substrate of the glutamate transporter with very low affinity for glutamate receptors, has been widely used to locate glutamate or aspartate using fiber systems in the brain (Fonnum, 1984). Autoradiographic studies on cryostate sections indicate that [3H]D-aspartate binding sites are particularly enriched in the molecular layer, but are also present in the granular layer (Greenamyre et al., 1990; Anderson et al., 1990). Studies in cerebellar slices, however, show that [3H]Daspartate is not taken up by mossy fiber terminals (Garthwaite and Garthwaite, 1988). Accordingly, [3H]D-aspartate is not retrogradely transported by mossy fibers, allthough it is efficiently transported by climbing fibers (Wiklund et al., 1984). Three high-affinity sodium-dependent glutamate transporters have been cloned in rat: GLT-1 (Pines et al., 1992; Tanaka, 1993), EAAC1 (Kanai and Hediger, 1992), and GLAST (Storck et al., 1992). Recently, also four subtypes of human glutamate transporters, EAAT1-EAAT4, have been cloned with similar properties as their rat counterparts (Arriza et al., 1994; Fairman et al., 1994). In situ hybridisation and immunocytochemistry showed a differential distribution of the three transporters throughout the cerebellum. GLT1 is concentrated in the Bergmann glial fibers, but also occurs in the glial processes of the granular layer and in the cerebellar nuclei (Danbolt et al., 1992; Rothstein et al., 1994). High levels of GLAST are present in Bergmann glial fibers, but it is essentially absent from the granule cell layer. In the cerebellar nuclei it is mostly 52
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 41. Electron micrographs of serial sections through a glomerulus in the granular layer of the cat cerebellar cortex. The section shown in (A) was reacted with antiserum to glutamate (GLU), the section in (B) with antiserum to GABA. The electron-dense gold particles show immunoreactive sites. For GLU the highest density of gold appears to be over the mossy fiber terminal (mt) and the lowest over glial processes and Golgi cell terminals (1-3). This was confirmed by statistical comparison of the populations. The same Golgi cell terminals are strongly reacting for GABA, while other processes have only a low surface density of gold. Scale (A and B) 0.5/Ira. Somogyi et al. (1986).
53
Ch. I
J. Voogd, D. Jaarsma and E. Marani
I
Fig. 42. Phosphate-activated glutaminase-like immunoreactivity (PAG-LI) in the precerebellar nuclei and the cerebellar cortex of the rat. Intensely labelled neuronal somata are seen in the pontine tegmental reticular nucleus of Bechterew (a), pontine nuclei (b), external cuneate nucleus (c), and lateral reticular nucleus of the medulla oblongata (d). Small clusters of grains, possible axon terminals, with PAG-LI are seen in the granular layer of the cerebellar cortex (e). Fine grains with PAG-LI are densely distributed, but no cell bodies are seen in the inferior olivary nucleus (f). CM, cerebellar medulla; G, granular layer; M, molecular layer; ML, medial lemniscus; P, pontine longitudinal fibers; Py, pyramidal tract; R, raphe. Scale bar 200 pm in a-d, f, 50 pm in e. Kaneko et al. (1987).
54
The cerebellum." chemoarchitecture and anatomy
Ch. I
associated with neurons (Rothstein et al., 1994). EAAC1 has been exclusively localized in neurons, with high densities in Purkinje cells and granule cells. Interestingly, immunocytochemical data show that EAAC1 is enriched in axon terminals of Purkinje cells, indicating that EAAC1 is not selective for glutamatergic nerve terminals. In accordance with the biochemical data, there was no immunocytochemical evidence for the presence of EAAC 1 or one of the other glutamate transporter proteins in mossy fiber terminals. Taken together the above data indicate that mossy fiber terminals are not provided with high-affinity glutamate transporters. Consequently, glutamate released by mossy fibers is likely to be predominantly cleared through glial cells (Wilkin et al., 1982; Garthwaite and Garthwaite, 1988). However, since glial processes do not enter the glomeruli (e.g. see Palay and Chan-Palay, 1974), an exclusive glial uptake implies that 'mossy fiber glutamate' molecules have to travel throughout extracellular space of the glomeruli before being inactivated. The clearance of 'mossy fiber glutamate' may be particularly slow at the giant mossy fiber-unipolar brush cell synapses, that may extend over 12-40 ,um2 with multiple clusters of presynaptic vesicles apposed to continuous regions of postsynaptic densities (Mugnaini and Floris, 1994). In fact, unusually long excitatory postsynaptic responses have been observed in unipolar brush cells following mossy fiber stimulation, consistent with a slow clearance of synaptically released glutamate (Rossi et al., 1995). 3.2.2. Climbing fibers Aspartate and glutamate
Several observations have led to the assumption that L-aspartate is the principal neurotransmitter of climbing fibers. (1) Destruction of the inferior olive in the rat with 3-acetylpyridine resulted in a small decrease in cerebellar aspartate concentration in total tissue homogenate (Nadi et al., 1977) and synaptosomal fractions (Rea et al., 1980). However, these observations were not confirmed by Perry et al. (1976). (2) It was demonstrated that after 3-acetylpyridine treatment Ca2+-dependent and K+-induced release of aspartate was significantly decreased (Toggenburger et al., 1983). Glutamate release was more dramatically decreased (e.g. see Cu6nod et al., 1989). (3) It was observed that climbing fibers but not mossy fibers in rat (Wiklund et al., 1984) and monkey (Matute et al., 1987) retrogradely transported [3H]D-aspartate. These experiments, however, only showed that climbing fibers are provided with high-affinity sodium-dependent glutamate transporter protein, and did not give information about the kind of transmitter used by the climbing fibers (see 3.2.1.). It should be noted that high affinity glutamate transporters have not yet been located at synapses of climbing fibers in immunocytochemical studies with antibodies against high-affinity glutamate transporters, although this possibility is still open since a detailed electron microscopical analysis of the cerebellar molecular layer has not yet been done (Rothstein et al., 1994). (4) Physiological studies suggested that the distal region of the Purkinje cell dendrites was relatively less sensitive towards aspartate as compared to glutamate than the proximal dendrites (Cr6pel et al., 1982). Since climbing fibers chiefly innervate the proximal two-thirds of the Purkinje cell dendritic tree (Palay and Chan-Palay, 1974), these data would be consistent with the proposal that aspartate is a climbing fiber transmitter, whereas glutamate is the transmitter of the parallel fibers (see Cu6nod et al., 1989). Voltage-clamp studies of Purkinje cells in slices, however, suggest that climbing fibers 55
Ch. I
56
J. Voogd, D. Jaarsma and E. Marani
The cerebellum." chemoarchitecture and anatomy
Ch.I
Fig. 43. Photomicrographs of consecutive semithin sections from rat inferior olive stained with antisera to aspartate, glutamate and GABA, respectively. All neurons (arrows) in this field are labelled for aspartate and glutamate, but unlabelled for GABA. Glial cells (arrowheads; identity established on the basis of electron microscopic analysis of adjacent sections) contain little or no GABA and glutamate immunoreactivities, but are moderately stained with the aspartate antiserum. Asterisks indicate fiber bundles. Scale bar = 50 r Insets show test sections incubated together with the respective tissue sections. The test antigens are GABA (1), glutamate (2), taurine (3), glycine (4), 'none' (5), aspartate (6), and glutamate (7). Note selective staining of the respective amino acid conjugates. Zhang et al. (1990). (
and parallel fibers activate the same type of glutamate receptors (Llano et al., 1991). Summarizing, one may conclude that the case for aspartate as the principal neurotransmitter of climbing fibers is far from being conclusive. Zhang et al. (1990), who compared glutamate- and aspartate-like immunoreactivities in the neurons of the inferior olive and climbing fibers in rat and baboon (Papaio anubis), showed that glutamate and aspartate-like immunoreactivities were co-localized in all neurons of the inferior olive, with a slightly heavier staining in the principal olive (Fig. 43). Significant glutamate-like, but little aspartate-like labelling, however, was recognized over climbing fiber profiles and, therefore, it was concluded that glutamate and not aspartate is the most likely transmitter of the climbing fibers (see also Zhang and Ottersen, 1993). It was also concluded that the presence of aspartate-like immunoreactivity in cell bodies is an unreliable indicator of transmitter identity.
Homocysteate Cu6nod et al. (1989) reported on the results of a series of experiments on K+-induced release of different transmitters by the cerebellum of the rat, after previous destruction of the inferior olive by 3-acetylpyridine. Release of aspartate was found to be decreased compared to the controls, with the main decrease occurring in the hemisphere. Values for the vermis were only slightly lower than in normal rats. This difference might be explained by a relative sparing of neurons in the caudal inferior olive, that project to the vermis. Decreased values after 3-acetylpyridine treatment were also found for adenosine (see Section 3.5) and for homocysteic acid. For the release of these substances no differences were noticed between vermis and hemisphere. Homocysteic acid was originally considered as a transmitter of the climbing fibers (Grandes et al., 1989), but proved to be located in Bergmann glia (Figs 44 and 45) (Cu6nod et al., 1990; Grandes et al., 1991). Climbing fibers, therefore, interact with Bergmann glia, both in the release of homocysteic acid and in 5'-nucleotidase-regulated adenosine release (see Section 3.5). Immunocytochemical studies have shown that subpopulations of climbing fibers may use peptides as a neurotransmitter, including somatostatine, corticotrophin-releasing factor and enkephalin. Their distribution and characteristics will be discussed in Section 6.3.4.
3.2.3. Granule cells and parallel fibers In early studies it was found that in 'staggerer', 'weaver' and 'reeler' mutant mice which have almost complete or partial loss of their granule cells (McBride et al., 1976a; Hudson et al., 1976) and in rats or mice that lost their granule cells by a viral infection or 57
Ch.I
J. Voogd, D. Jaarsma and E. Marani
postnatal X-irradiation (McBride et al., 1976b; Rohde et al., 1979), glutamate was depleted. However, the interpretation of this finding is not immediately clear, because mossy fiber terminals and inhibitory interneurons of the cerebellar cortex also may have been affected. Furthermore, it proved difficult to exclude aspartate as a transmitter of granule cells (Rohde et al., 1979; Roffler-Tarlov and Turey, 1982). Also the demonstration of Garthwaite and Garthwaite (1985) that granule cells in slices accumulate [3H]Daspartate did not provide conclusive evidence about the nature of the neurotransmitter used by parallel fibers. Immunocytochemical studies strongly support glutamate as the neurotransmitter of the parallel fibers. Thus, glutamate-like immunoreactivity but no other amino acids were enriched over parallel fiber terminals in rat (Ottersen and Storm-Mathisen, 1984a,b, 1987; Ottersen et al., 1987, 1990; Liu et al., 1989) (Fig. 40), cat (Somogyi et al., 1986) and monkey (Zhang et al., 1990). Also electrophysiological experiments are in favour of glutamate as the neurotransmitter at the parallel fiber-Purkinje cell synapse (Barbour, 1993 and references therein).
~
[!2s .
Fig. 44. Immunocytochemical localization of homocysteate (HCA) in Bergmann glia with polyclonal antiHCA antibodies. A. Test system mimicking immunocytochemical procedure. Conjugates are assembled in 'sandwich' construction with tissue as spacer and contain the following compounds (from top to bottom): HCA, Glu (glutamate), Asp (aspartate), Tau (taurine), Gly (glycine), GABA (~,-aminobutyric acid), L-Ala (L-alanine), fl-Ala (fl-alanine), Htau (homotaurine), Hypotau (hypotaurine), Gline (glutamine), Ca (cysteate), CSA (cysteine sulphinate), HCSA (homocysteine sulphinate), Cys (cysteine), Cyt (cystine), Met (methionine), carnosine, Hcys (homocysteine), cystathionine, gluta-thione, homocarnosine, y-Glu-Glu (y-glutamyl glutamate), fl-L-Asp-Gly (fl-L-aspartyl glycine), no AA (no amino acid conjugated to glutaraldehyde-treated rat brain protein). B. Pattern of HCA-like immunoreactivity in low-power view of rat cerebellar cortex in semithin section. Double arrow: fibrous, radially oriented immunoreactive element. Arrowheads, stained varicosities in association with Purkinje cell dendrites. C. Pattern of HCA-like immunoreactivity in rat cerebellar section pretreated with 3-acetylpyridine 10 days previously and degeneration of the inferior olive. No changes in the distribution of HCA are apparent. Bars 50/lm. Grandes et al. (1991).
58
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 45. Immunocytochemical localization of homocysteate (HCA) with polyclonal anti-HCA antibodies. A. Staining pattern in section close to that in Fig. 44B at higher magnification. Cell (asterisk) and capillary (circle) used as landmarks in A and B. B and C. Electron micrographs from ultrathin section immediately preceding semithin section in A. The HCA-immunoreactive varicosities indicated with arrows in (A) were identified as parts of the glial sheath surrounding Purkinje cell dendrites (d) in B and C. Bars: 10/lm in A, 5/lm in B, 1 j~m in C. Grandes et al. (1991).
Specific markers for granule cells are few. Seyfried et al. (1983), concluded from biochemical analysis in 'weaver' mutant mice that the ganglioside GDIA was more concentrated in granule cells. Webb and Woodhams (1984) developed three monoclonal 59
Ch. I
J. Voogd, D. Jaarsma and E. Marani
antibodies (G-l-3; 7-8D2 and 8-20-1), that recognize cell surface antigens expressed by rat granule cells and their axons (see also Reynolds and Wilkin, 1988). Calcium-binding proteins, with the exception of calretinin (Rogers, 1989; Arai et al., 1991; Kadowaki et al., 1993; Floris et al., 1994) have not been localized in granule cells. Proteine kinase C (PKC) e, flI and II, e and ~"are expressed by rat granule cells (Ase et al., 1988; Wetsel et al., 1992; Chen and Hillman, 1993) (Table 1). 3.3. LOCALIZATION OF GLUTAMATE RECEPTORS
3.3.1. lonotropic glutamate receptors Glutamate activates two main classes of glutamate receptors, the ionotropic and metabotropic glutamate receptors. The ionotropic receptors are receptor/channel complexes that can be categorized into three groups according to their differential sensitivity to agonist ligands, as ~-amino-3-hydroxy-5-methyl-4-isoxazole proprionic acid (AMPA) receptors, formerly known as quisqualate receptors, kainate receptors, and N-methylD-aspartate (NMDA) receptors (Monaghan et al., 1989; Mayer and Miller, 1990; Westbrook, 1994). The non-NMDA (AMPA and kainate) receptors display rapid kinetics. They are typically inhibited by 7-cyano-7-nitroquinoxaline-2,3-dione (CNQX), are permeable to monovalent cations (Na+, K+), but mostly impermeant to Ca 2+, and have been implicated in fast excitatory synaptic transmission (Mayer and Westbrook, 1987; Jahr and Lester, 1992). NMDA receptor channels, instead, have relatively slow kinetics, are also permeable to Ca 2+ ions, and are typically inhibited by D-2-amino-5-phosphonovalerate (APV). NMDA receptors are characterized by a voltage-dependent channel block by MgZ+-ions. They are dependent on, and are equipped with a coagonist site for glycine. Apart from their role in excitatory synaptic transmission, NMDA receptors have been implicated in synaptic plasticity and in developmental processes like cell migration and synaps formation (Collingridge and Singer, 1990). AMPA receptors
AMPA receptors have been autoradiographically labelled with [3H]AMPA and the antagonist [3H]CNQX: [3H]AMPA binding is moderately high over the rodent (Rainbow et al., 1984b; Monaghan et al., 1984; Nielsen et al., 1990; Garcia-Ladona et al., 1991; Makowiec et al., 1991) and human (Jansen et al., 1990) cerebellum, and is higher over the molecular than over the granular layer. [3H]CNQX binding sites are preferentially localized over the molecular layer, but cerebellar [3H]CNQX binding is relatively higher than [3H]AMPA binding, when the two are compared to binding levels of both ligands in other brain areas (e.g. see Fig. 6 in Nielsen et al., 1990). This difference is not due to [3H]CNQX binding to kainate receptors since these receptors are preferentially localized in the granular layer. Both [3H]AMPA and [3H]CNQX binding in the molecular layer was decreased in Purkinje cell deficient (pcd) mutant mice, but strongly upregulated in granuloprival mice (Makowiec et al., 1991). These observations favour a primary localization of AMPA receptors on Purkinje cells and an upregulation of the number of AMPA receptors on Purkinje cells as a consequence of deafferentation (Makowiec et al., 1991). Originally, AMPA receptors were assessed as quisqualate-sensitive [3H]glutamate binding sites (Cha et al., 1988, and references therein). Quisqualate-sensitive [3H]glutamate binding is strongly increased by the presence of CaC12, and is relatively high in the cerebellar molecular layer. CaC12-dependent quisqualate-sensitive [3H]glutamate bind60
The cerebellum." chemoarchitecture and anatomy
Ch. I
ing over the molecular layer, however, is largely insensitive to AMPA (Cha et al., 1988). These sites most likely correspond to the quisqualate-sensitive metabotropic glutamate receptors (Young et al., 1991), that have been recently demonstrated to be expressed at high levels by Purkinje cells (see Section 3.3.2.). Kainate receptors
High-affinity [3H]kainate binding sites predominate in the granular layer in rat (Monaghan and Cotman, 1982; Olson et al., 1987; Cambray-Deakin et al., 1990; Bahn et al., 1994) and man (Jansen et al., 1990). Low to moderate levels of [3H]kainate binding occur in the rat cerebellar nuclei. [3H]Kainate binding is not affected in Purkinje cell deficient (pcd) or 'nervous' mutant mice, but is decreased in granuloprival mice (Griesser et al., 1982). This decrease concerns the granular but not the molecular layer (Olson, 1987; Makowiec et al., 1991). Henke et al. (1981) noted a high level of low-affinity [3H]kainate binding sites in the molecular layer of pigeon cerebellum. Similar [3H]kainate binding sites were also labelled in the chicken cerebellum (Henley and Barnard, 1990), in fish (Maler and Monaghan, 1991) and in amphibian cerebellum, although in the amphibian kainate-binding sites seems to have somewhat different pharmacological and functional properties (reviewed in Henley, 1994). The chicken kainate binding sites could also be labelled by [3H]CNQX (Henley and Barnard, 1990). Several non-mammalian vertebrate kainate-binding proteins have been purified and cloned. These proteins display some homology towards mammalian ionotropic AMPA and kainate receptor subunits (see below), but are smaller (40-50 kDa instead of 100 kDa), and do not form functional receptors channels (reviewed by Hollman and Heinemann, 1994; Henley, 1994). In situ hybridisation and immunocytochemistry has shown that avian kainate-binding protein is localized in Bergmann glia (Fig. 95) (Somogyi et al., 1990; Gregor et al., 1992 and others). Somogyi et al. (1990) showed that immunostaining with a monoclonal antibody (IX-50) against chicken kainate-binding protein, was also localized in Bergmann glia in the cerebellum of fish. Frog kainatebinding protein, however, is widely distributed throughout the frog brain. High receptor densities were found in cerebellum, but their cellular distribution has not yet been reported (Dechesne et al., 1990; Wenthold et al., 1990). N M D A receptors
The distribution of NMDA receptors has been autoradiographically determined as NMDA-replaceable [3H]glutamate binding sites. In rat (Greenamyre et al., 1985; Monaghan and Cotman, 1985) and human cerebellum (Jansen et al., 1990), moderate densities of binding sites are found over the granular layer and in the cerebellar nuclei, whereas binding over the molecular layer is low. Olson et al. (1987) and Makowiec et al. (1991) reported that NMDA-sensitive [3H]glutamate binding is unchanged in Purkinje cell deficient (pcd) mutant mice, but that the density of binding sites is considerable reduced over the granular layer in granuloprival mice. These data suggest that NMDAbinding sites are absent on Purkinje cell dendrites and, instead, are present on granule cells and perhaps on stellate, basket and Golgi cells. Using different ligands including the competitive antagonist [3H]-2-carboxypiperazine-4-yl-propyl-l-phosphonic acid ([3H]CPP), [3H]glycine that specifically binds to the glycine coagonist site of NMDA receptors, and the non-competetive channel blockers [3H]MKS01 and [3H]-N-[1-(2-thienyl)cyclohexyl]-3,4-piperidine ([3H]TCP), it was found that the pharmacological properties of NMDA receptors in the cerebellar 61
Ch. I
J. Voogd, D. Jaarsma and E. Marani
cortex were different from those in other brain areas (reviewed in Monaghan and Anderson, 1991). Notably, cerebellar NMDA receptors label poorly with the noncompetetive channel blockers [3H]MK801 and [3H]TCP (Maragos et al., 1988; Monaghan and Anderson, 1991). Monaghan and coworkers recognized at least 4 pharmacologically distinct NMDA receptor types throughout the brain and recently demonstrated that their pharmacological heterogeneity reflects differences in subunit composition (see below; Buller et al., 1994). They identified two populations of NMDA receptors in the cerebellar cortex. One population of 'antagonist-prefering' sites, that can be labelled by [3H]CPP, is present throughout the brain, and represents NMDA receptors containing NR2A subunits. The second population consists of the 'cerebellar-like' sites, where competitive antagonists and the agonist homoquinolinate are relatively ineffective in displacing the NMDA-sensitive [3H]glutamate binding. They reflect the presence of NR2C subunit, that is uniquely expressed by cerebellar granule cells (Buller et al., 1994). The low level of [3H]MK801 and [3H]TCP binding in the cerebellum remains to be explained. Distribution of subunits
Like other classes of ionotropic receptors functional glutamate receptor channel complexes are multimeric proteins. Recent molecular cloning studies have revealed families of AMPA (GluR1-GluR4, also named GluRA-GluRD), kainate (GluR5-GluR7, and KA1 and KA2), NMDA (NR1, named ~'1 in mice, and NR2A-NR2D, named el-e4 in mice) and orphan (~1 and ~2) glutamate receptor subunits (reviewed in Nakanishi, 1992; Sommer and Seeburg, 1992; Hollman and Heinemann, 1994). The diversity of glutamate receptor subunits is further increased through alternative splicing that primarily involves the AMPA receptor subunits GluR1-GluR4, each of which exists in two versions, i.e. flip or flop, and the NR 1 subunit, that has eight splice variants. Combinatorial expression studies have demonstrated that the subunits aggregate into functional receptor channels in the homomeric as well as the heteromeric configuration. Thus multiple functionally distinct forms of each receptor type can be formed through different combinations of subunits (see below). In situ hybridisation (KeinS.nen et al., 1990; Monyer et al., 1991, 1994; Araki et al., 1993; Sato et al., 1993; Wisden and Seeburg, 1993; Akazawa et al., 1994; Laurie and Seeburg, 1994; Watanabe et al., 1994; and others) and immunocytochemical studies with antibodies for specific subunits (Martin et al., 1992, 1993; Petralia and Wenthold, 1992; Brose et al., 1993; Baude et al., 1994; Nusser et al., 1994; Petralia et al., 1994a,b,c; Jaarsma et al., 1995b) have shown that subunits are heterogeneously distributed throughout the cerebellum, each cell type expressing a characteristic set of subunits (see Table 2). AMPA subunits
AMPA receptor subunits are not only expressed by cerebellar neurons, but also by Bergmann glia, that express high levels of GluR 1 (GluRA) and GluR4 (GluRD) mRNA (Table 2, Fig. 46). GluR1 subunit mRNA is also expressed by Purkinje cells but not by other cerebellar cells (Kein~inen et al., 1990; Monyer et al., 1991; Sato et al., 1993). GluR4 mRNA in addition to Bergmann glial cells, is produced by granule cells and neurons of the deep nuclei. Granule cells express a GluR4 splice variant exclusively found in the cerebellum, GluR4c, consisting of GluR4 with the flop module and a truncated C-terminus (Gallo et al., 1992). GluR2 (GluRB) mRNA is found over the granular and molecular layers, in Purkinje cells, and in cells of the deep nuclei (Fig. 46). 62
The cerebellum." chemoarchitecture and anatomy
Ch. I
GluR3 (GluRC) mRNA is not expressed in granule cells, but occurs in Golgi cells, stellate and basket cells, Purkinje cells and cells in the deep nuclei (Fig. 46) (KeinS.nen et al., 1990; Monyer et al., 1991; Sato et al., 1993). The distribution of AMPA subunits has been studied immunocytochemically with antibodies specific for GluR1, GluR2/3, and GluR4 (Martin et al., 1992; Petralia and Wenthold, 1992; Baude et al., 1994; Nusser et al., 1994; reviewed in Jaarsma et al., 1995b). The processes of the Bergmann glia are densely immunostained for GluR1 and GluR4 (Fig. 47) confirming in situ hybridisation. Electron microscopy showed that GluR1 immunoreactivity was localized throughout the cytoplasma membrane (Baude et al., 1994). Dense immunostaining was associated with the processes of Bergmann fibers ensheathing PC spines and the attached synaptic varicosities of parallel fibers and climbing fibers (Fig. 48). This indicates that AMPA receptors on Bergmann glia may be activated by glutamate released by parallel fibers or climbing fibers, allthough there is no clue as yet of the functional role of glial cell activation (see discussion Baude et al., 1994, but also Mfiller et al., 1992). Purkinje cells are weakly-to-moderately immunopositive for GluR1. Dense GluR1 immunolabelling was found at the post-synaptic membrane specialisations of the dendritic spines of Purkinje cells, facing parallel and climbing fiber boutons (Fig. 48). The post-synaptic membranes of the parallel fiber-Purkinje cell and the climbing fiberPurkinje cell synapses are also stongly immunoreactive for GluR2/3 (Nusser et al., 1994; Jaarsma et al., 1995b). The GluR2/3 antibodies immunoreact with all cerebellar neurons. The perikarya and dendritic arbors of Purkinje cells densely immunostain, whereas
TABLE Type
AMPA
Kainate
NMDA
orphan
2.
Distribution of glutamate receptor subunit mRNAs in rat cerebellum Subunit
Cell t y p e PC
GrC
GoC
GluR1
+ flip
-
-
-
+ + flip
-
GluR2
+ + flip/flop
+ flip
+
+
-
++
GluR3
+ flip
-
++
++
-
+
GluR4
-
+ 4c-flop
-
-
+ + flip
+
GluR5 GluR6
+ -
. ++
GluR7
-
-
-
+
KA1
+
.
.
BC/Stc
. .
Bg
.
.
.
.
-
.
+
.
.
DCN
.
.
KA2
-
++
-
-
-
+
NR1
+(NRI-b)
++(NRI-a)
+
+
-
++
NR2A
-
+
-
-
-
+
NR2B
.
NR2C
-
++
.
NR2D
-
-
+
-
+
delta 1
+
.
delta2
.
.
.
.
.
.
.
.
.
+
. .
.
. .
. .
. .
S y m b o l s : - , n o t d e t e c t e d ; +, p o s i t i v e ; + + , s t r o n g l y p o s i t i v e ; P C , P u r k i n j e cells; G r C , g r a n u l e cells; G o C , G o l g i cells; B C , b a s k e t cells; St, s t e l l a t e cells; Bg, B e r g m a n n
glia; D C N ,
deep cerebellar nuclei.
B a s e d o n d a t a f r o m K e i n ~ n e n et al., 1990; M o n y e r et al., 1991, 1994; L a m b o l e z et al., 1992; A r a k i et al., 1993; L o m e l i et al., 1992, 1993; S a t o et al., 1993; W i s d e n a n d S e e b u r g , 1993; A k a z a w a S e e b u r g , 1994; W a t a n a b e
et al., 1994; L a u r i e a n d
et al., 1994.
63
P
Fig. 46. In situ hybridization of AMPA glutamate receptor mRNAs in sections of rat cerebellum. A. GluRl (GluRA) mRNA distribution; arrow heads indicate continuous line of silver grains along the Purkinje-Begmann layer. B. GluR2 (GluRB) mRNA; arrow heads indicate labelled Purkinje cells. C. GluR3 (GluRC) mRNA; small arrow heads indicate clusters of silver grains in molecular layer over stellate-basket cells. D. GluR4 (GluRD); arrow heads as in (A). gr, granule cell layer; mol, molecular layer; p, Purkinje cells; wm, white matter. Scale bar 500 fim. Keinanen et al. (1990).
a
& h
The cerebellum." chemoarchitecture and anatomy
Ch. I
5',.2 uletl
Fig. 47. Sagittal sections of the rat cerebellar cortex immuno-labelled with antibodies to GluR1 (a), GluR2/3 (b,e), and GluR4 (c,d). As, astrocyte-like cells; BG, Bergmann glial processes; Go, Golgi cell; Gr, granular layer; L, Lugaro cell; Mo, molecular layer; Pj, Purkinje cell body; WM, white matter; small arrow, Purkinje cell dendrite; asterisks, Bergmann glial cell body; arrow head, basket/stellate cell. Petralia and Wenthold (1992).
light-to-moderate staining neurons occur in basket/stellate cells, Golgi cells and granule cells (Fig. 47) (Martin et al., 1992, 1993; Petralia and Wenthold, 1992; Jaarsma et al., 1995b). Unipolar brush cells are also strongly GluR2/3-immunopositive (Jaarsma et al., 1995b). Dense and moderate GluR2/3-staining was found in the perikarya and neuropil of the deep nuclei, respectively. Using electronmicroscopic immunogold protocols, that allow precise ultrastructural localization of the immunoreaction product, Nusser et al. (1994) obtained stong proof that GluR2/3 immunoreactivity is associated with postsynaptic membrane specialisations of excitatory synapses in the cerebellar cortex (Fig. 49B, C, F). Their data indicate that GluR2/3 immunoreactivity is considerably stronger at parallel fiber-Purkinje cell, climbing fiber-Purkinje cell and parallel fiber-stellate cell synapses than at mossy fiber-granule cell synapses (compare Figs 49B and C with F). Conventional peroxidase-DAB (3,3'-diaminobenzidine tetrahydrochloride)-immuno65
Ch. I
J. Voogd, D. Jaarsma and E. Marani
electron microscopy also indicates that GluR2/3 immunoreactivity is relatively weak at the mossy fiber-granule synapses (Jaarsma et al., 1995b). The post-synaptic membranes of the giant mossy fiber-unipolar brush cell synapses are, however, strongly GluR2/3 immunopositive (Jaarsma et al., 1995b). The GluR4 antibodies, in addition to the Bergmann glia, moderately immunostain the granular layer, and the neuropil and perikarya in the the deep nuclei (Fig. 47). It was originally assumed that GluR4-immunostaining in the granular layer was associated with granule cells (Martin et al., 1992, 1993; Petralia and Wenthold, 1992), but electron microscopy showed that GluR4 immunoreactivity is localized in the astroglia (Jaarsma et al., 1995b). Thus granular layer astroglia like Bergmann glia express AMPA receptor subunits, but unlike the Bergmann glia, do not have GluR1. The absence of GluR4-immunoreactivity in granule cells can be explained by the fact that granule cells primarily express an atypical form of GluR4, GluR4c (see above), that is not recognized by the GluR4 antibodies currently available. AMPA receptors are believed to mediate most of the fast excitatory neurotransmission in the brain, including the cerebellum. Concordantly the types of AMPA receptor subunits expressed by a cell largely determine the characteristics of the fast excitatory responses (see Jonas and Spruston, 1994). The GluR2 subunit dominate the AMPA receptor channel behavior, in that homomeric GluR2 channels as well as heteromeric
Fig. 48. Electron micrograph of the synaptic distribution of immunoreactivity for the GluR 1 subunit of the AMPA receptor in rat cerebellum as detected by an antibody against the carboxy-terminal (intracellular) region of GluR1. A. A spine (s) emerging from a Purkinje cell dendrite (Pd) establishes an immunopositive type 1 synapse (solid arrows) with a parallel fiber terminal (pft). Intra-cellular immunoreactivity is present inside Bergmann glial cell processes along dendritic elements (e.g., open arrow). B. The peroxidase reaction end-product labels the postsynaptic density (psd) at the intracellular face of the postsynaptic membrane (pom) and not the synaptic cleft between the presyaptic (pem) and postsynaptic (pom) membranes. Scale bars in A = 0.5 r in B = 0.1 r Baude et al. (1994).
66
The cerebellum." chemoarchitecture and anatomy
Ch. I
channels formed with the participation of GluR2 show the properties of 'typical' AMPA receptors, i.e. linear current-voltage relations and a relative impermeability to Ca 2+. Homo- or heteromeric channels without GluR2, instead, display inward rectification and are permeable to Ca 2+ and other divalent cations (for references see Sommer and Seeburg, 1992; and Hollman and Heinemann, 1994). These channels usually are not found in neuronal cells, concordant with the notion that most neuronal cells express GluR2, but are present in Bergmann glial cells (Mfiller et al., 1992; but see also Burnashev et al., 1992). In accordance with the presence of high levels of GluR 1 and GluR4 but absence of GluR2 in these cells. By combining two powerful methods, i.e. the patch-clamp technique to characterize the properties of native receptor channels in single cells in brain slices, followed by single cell PCR-amplification methods to analyse the mRNA contents of the respective cells semiquantitatively, Jonas et al. (1994) recently showed that the CaZ+-permeability of native AMPA receptor channels in cerebral cortical cells is related to the relative abundance of GluR2 subunit mRNA in the respective cells. Thus inhibitory interneurons of the cerebral cortex have low GluR2/non-GluR2 ratios (-- 0.3) and highly Ca 2+permeable AMPA receptors (which, however, display linear current-voltage relations unlike 'Bergmann-glial' AMPA receptors), whereas pyramidal cells, which have a relatively high GluR2/non-GluR2 mRNA ratio (-- 3), contain CaZ+-impermeable AMPA receptors. PCR-amplification analysis of AMPA subunit m R N A of Purkinje cells, indicates that GluR2 mRNAs are more abundant than GluR1 and GluR3 mRNAs (Lambolez et al., 1992), implying that Purkinje cells express weakly CaZ+-permeable AMPA receptors. Also in (pooled) granule cells GluR2 mRNA is more abundant than non-GluR2 (GluR4) mRNA. PCR-amplification analysis has not yet been done for other cerebellar cells. One might speculate that the basket, stellate and Golgi cells express CaZ+-permeable AMPA receptor like the inhibitory interneurons of the cerebral cortex, since according to in situ hybridisation data they seem to express relative high levels of GIuR3 compared to GluR2 (see Table 2). AMPA receptors made from different subunits may have different desensitization kinetics. Desensitization is particularly fast for AMPA receptors formed with GluR3flop or GluR4-flop (Mosbacher et al., 1994). Granule cells produce GluR4-flop and GluR2 (Table 2), and therefore are likely to have fast (submillisecond) desensitizing AMPA receptors. This could explain the very fast decay kinetics of non-NMDA component of the excitatory post-synaptic currents (EPSCs) at the mossy fiber-granule cell synapses (Silver et al., 1992; Rossi et al., 1995). If this holds true this would imply that the length of the excitatory responses at the mossy fiber-granule cell synapses is largely controlled by the desensitization properties of the AMPA receptors and does not depend upon the time course of transmitter removal, that may be relatively slow at these synapses (Jonas and Spruston, 1994) (see Section 3.2.1.). Purkinje cells express GluRl-flip, GluR2-flip and -flop, and GluR3-flip mRNA (Lambolez et al., 1992) that form AMPA receptors with desensitization time constants 3-5 times slower than 'GluR4-flop-GluR2' channels (Mosbacher et al., 1994). Concordantly, AMPA receptors in Purkinje cells appear to have relatively slow desensitization kinetics (Barbour et al., 1994). Also the decay phases of AMPA receptor-mediated EPSCs in Purkinje cells after parallel fiber or climbing fiber activation, have slow time constants (Perkel et al., 1990; Llano et al., 1991; Barbour et al., 1994). Interestingly stellate/basket cells, that express AMPA receptors with the same desensitization kinetics as Purkinje cell AMPA receptors, showed much faster decaying parallel fibers EPSCs (Barbour et al., 1994). Barbour et al. (1994) concluded that glutamate is rapidly cleared 67
Ch. I
J. Voogd, D. Jaarsma and E. Marani
at the parallel fiber-stellate/basket cell synapses, resulting in rapid deactivation of postsynaptic AMPA receptors, whereas synaptically released glutamate seems to be present during a prolonged time at Purkinje cell synapses. According to H/~usser (1994) EPSCs of climbing fiber-Purkinje cell synapses have decay time constants that are slower than parallel fiber-evoked EPSCs in Purkinje cells, which may be explained by the fact that clearance of glutamate at climbing fiber synapses is slower due to their larger size.
68
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 49. Electron micrographs showing the subsynaptic segregation of GluR2/3 AMPA receptor subunits (GluR B/C/4c) and the metabotropic mGluRl~ glutamate receptor (mGluR1; see Section 3.3.2) as revealed by post-embedding immunogold labelling. A and B. Consecutive sections of the same synaptic junctions showing that immunoparticles for mGluRl~ (double arrows in A) are concentrated at the edge, whereas immunoparticles for GluR2/3 (arrows in B) are concentrated in the main body of synaptic junctions established by parallel (pft) and climbing (cft) fiber terminals with spines (s) of Purkinje cell dendrites (Pd). Note that mGluRlcz is often localized extrasynaptically (double arrow heads in A). C. Immunoreactivity for GluR2/3 (arrows) was always very strong on basket and stellate (Stc) cells. D and E. Double immunolabelling of mGluRl~ (large particles, double arrows) and GluR2/3 (small particles, arrows) immunoreactivity in the synapses on spines (s) of Purkinje cells, confirming synaptic segregated subsynaptic localization of mGluR1 and GluR2/3. The synapse in E is from Triton treated material. F. Generally a lower density of immunoparticles for GluR2/3 (arrows) has been found in synapses between mossy fiber terminals (mt) and granule cell dendrites (d) than in the parallel fiber synapses (compare to B and C). Scale bars = 0.1 r in A,B,D,E, 0.2 r in C,F. Nusser et al. (1994). (
Kainate subunits
Kainate receptor subunits are most prominent in granule cells, which express high amounts of both GluR6 and KA2 mRNA (Fig. 50) (Wisden and Seeburg, 1993). Purkinje cells express moderate levels of GluR5 and low levels of KA1 mRNAs; basket and stellate cells express GluR7 mRNA, and neurons of the deep cerebellar nuclei produce GluR7 and KA2 mRNAs (Table 2, Fig. 50) (Wisden and Seeburg, 1993). The high level of kainate receptor mRNA expression by granule cells is in accordance with the preferential binding of [3H]kainate over the granular layer (see above, but also Bahn et al., 1994). Immunocytochemical studies with antibodies specific for GluR6 and GluR7 show that dense GluR6/7-immunostaining occurred over the granule cell layer, where it was associated with granule cell perikarya and dendrites (Petralia et al., 1994c; Jaarsma et al., 1995b). The post-synaptic membranes of the mossy fiber-granule cell synapses were strongly immunoreactive for GluR6/7 (Jaarsma et al., 1995b). Stellate and basket cells and cells in the deep cerebellar nuclei were also immunoreactive for GluR6/7. KA2 immunoreactivity was relatively low in the cerebellar cortex and was concentrated over the glomeruli and neurons in the deep nuclei (Petralia et al., 1994c; Jaarsma et al., 1995b). Recombinant expression studies have shown that GluR5 and GluR6 form glutamategated channels in the homomeric configuration as well as in the heteromeric configuration with KA1 and KA2, whereas KA1 and KA2 do not assemble into functional receptor channel complexes (reviewed in Wisden and Seeburg, 1993). Thus functional kainate receptors can be formed in several cerebellar cells, in particular in granule cells expressing significant levels of GluR6 and KA2. Unequivocal physiological evidence for the presence of kainate receptors in the cerebellum (as well as in other brain areas) is, however, still lacking (see discussion Wisden and Seeburg, 1993). It should be noted that kainate in spite of its low affinity for AMPA receptors, potently activates AMPA receptors, and that excitatory responses evoked by kainate in brain tissue are generally mediated through AMPA receptors. Recombinant kainate receptors have been demonstated to desensitize very rapidly, which in part may explain why kainate receptor responses have not been detected (Wisden and Seeburg, 1993). Another possibility is that kainate receptor responses are masked by AMPA receptors, which are assumed to be present in much higher concentrations in neurons (Wenthold et al., 1994 and references therein). Autoradiographic (see above) and immunocytochemical studies (Jaarsma et al., 1995b), however, suggest that kainate receptors predominate over AMPA recep69
Ch. I
~LuR-5
J. Voogd, D. Jaarsma and E. Marani
,~ .~LuR ::6
B .3LuR 7
C Mol
Gr P "If ;:' l"'.s,"''~,;,"\ T (
'
i
f
Po
Fig. 50. Distribution of GluR5 (A), GluR6 (B), GluR7 (C), KA-1 (D), KA-2 (E) of subunits in RNAs of high-affinity kainate receptor mRNAs in coronal sections at level of the cerebellum of the rat. Gr, granular layer; LC, locus coeruleus; Mol, molecular layer; P, Purkinje cell layer; Po, pontine nuclei. Scale bars: 2.3 mm. Wisden and Seeburg (1993).
tors in granule cells, and may significantly contribute to excitatory neurotransmission at the mossy fiber-granule cell synapses. N M D A subunits Functional NMDA receptors are believed to be generated as heteromeric assemblies of NR1 subunits with members of the NR2 subunit family. The pharmacological and kinetic heterogeneity of NMDA receptors seems to be primarily dependent upon the type of NR2 subunit (Monyer et al., 1992; Meguro et al., 1992; Nakanishi, 1992; Buller et al., 1994), although NR1 diversity generated through alternative splicing may also contribute to NMDA receptor heterogeneity (Buller et al., 1994; Hollman and Heineman, 1994). Essentially all cerebellar neurons seem to express significant levels of NR1 mRNA (Table 2, Fig. 51E) (Moriyoshi et al., 1991). The main splice variants produced in the cerebellum are NR 1-2 (with 3'-end deletion 1) and to a lesser extent NR1-4 (with 3'-end deletions 1 and 2; Laurie and Seeburg, 1994). There is a remarkable difference between the Purkinje cells and the other cells of the cerebellar cortex, in that Purkinje cells express high levels of the NRI-a forms (without 5'-insertion), whereas in the other cells the N 1-b splice variants (with 5'-insertion) predominate (Laurie and Seeburg, 1994). NR2 subunit mRNAs are heterogeneously expressed throughout cerebellar neurons (Table 2, Fig. 51) and show pronounced changes during development (Akazawa et al., 1994; Monyer et al., 1994; Watanabe et al., 1994). NR2 subunit mRNAs are most prominent in granule cells, that express high levels of NR2C mRNA and moderate levels of NR2A mRNAs in adult rodent cerebellum (Fig. 51). Interestingly, whereas NR2A mRNA expression in rodent granule cells begins early postnatally, NR2C first appears in later stages (postnatal day 10-11 in rat) in post-migratory cells of the internal granular layer. It apparently replaces NR2B, which is transiently expressed by cerebellar granule cells (Akazawa et al., 1994; Monyer et al., 1994; Watanabe et al., 1994). The expression of NR2C starts in granule cells of the caudal vermis (lobules VIII-X) and subsequently extends throughout the whole cerebellar cortex by postnatal day 13 (see Fig. 3N and O in Watanabe et al., 1994 and Fig. 7 in Akazawa et al., 1994). This pattern is compat-
70
The cerebellum." chemoarchitecture and anatomy
Ch. I
ible with the sequence of maturation of the granule cells (Altman, 1972). According to Akazawa et al. (1994) and Watanabe et al. (1994), but not Monyer et al. (1994), NR2C mRNA is also expressed in the external granular layer during the first postnatal days. NMDA receptors have been demonstrated to be critically involved in granule cell migration (Komuro and Rakic, 1993; Rossi et al., 1993). Since NR2B is transiently expressed by granule cells during the period of migration, one may speculate that receptors made with NR2B and NR1 (and possibly NR2A) may act as 'migration receptors'. The presence of multiple NMDA receptors in granule cells is consistent with the presence of multiple NR2 subunits and has recently been demonstrated with patchclamp methods (Farrant et al., 1994): Pre-migratory and migratory granule cells were shown to express NMDA receptor channels with conductancy properties of recombinant NMDA receptors formed by co-expression of NR 1 and NR2A or NR2B. Mature post-migratory cells, in addition, express 'low-conductance' NMDA receptor channels, which have the properties of NMDA receptors with NR2C (Monyer et al., 1994). Basket, stellate cells, Golgi cells and neurons in the cerebellar nuclei express NR2D mRNAs (Akazawa et al., 1994; Monyer et al., 1994; Watanabe et al., 1994). Neurons of the cerebellar nuclei also produce NR2A mRNA, but it is not clear whether NR2A and NR2D producing cells reflect distinct neuronal populations. It should be noted that NMDA receptors composed of NR1 and NR2D have very slow deactivation kinetics (roll = 4.8 s) (Monyer et al., 1994) and, therefore, may modulate the cell activity during many seconds even when the receptor channel has been briefly activated by glutamate (see discussion Monyer et al., 1994). Quinlan and Davies (1985) have provided indirect physiological evidence for the presence of NMDA receptors in stellate and basket cells, by showing that NMDA may induce inhibition of Purkinje cells. Also neurons of the deep cerebellar nuclei have been shown to display prominent NMDA responses in cerebellar slice cultures (Audinat et al., 1990).
,Sg~,'
.4
o ~
.._ ., /;.'_. . ;
,
.
_.
.
..
Fig. 51. Bright-field micrographs showing cellular distributions of the NMDA receptor channel subunit mRNAs in the cerebellar cortex of the adult mouse: (A) el (mouse homologue of NR2A) mRNA; (B) e2 (NR2B); (C) e3 (NR2C); (D) e4 (NR2D); and (E) ~'1 (NR1). Each photograph in the figure was taken from lobule V of the cerebellar vermis, and the expression patterns of the respective subunit mRNAs are identical to those in remaining regions of the cerebellum. Sections were counter-stained with toluidine blue. Arrows indicate cell bodies of the Purkinje cells. Gr, granular layer; Mol, molecular layer. Scale bar = 50 r Watanabe et al. (1994).
71
Ch. I
J. Voogd, D. Jaarsma and E. Marani
The presence of NMDA receptors on Purkinje cells has been disputed. Some studies have supported the presence of NMDA-receptors on Purkinje cells (Sekiguchi et al., 1987), but in most studies no evidence of NMDA-receptors on Purkinje cells has been found (e.g. Audinat et al., 1990; Perkel et al., 1990; Llano et al., 1991; Farrant and Cull-Candy, 1991). Studies of Krupa and Cr6pel (1990) and Rosenmund et al. (1992) have indicated that NMDA receptors are present on most Purkinje cells during early post-natal life, but disappear with age. Both in situ hybridisation in rat and mouse and immunocytochemical studies in rat have shown that the NR1 subunit is expressed at high levels by Purkinje cells (Brose et al., 1993; Akazawa et al., 1994; Monyer et al., 1994; Petralia et al., 1994a; Watanabe et al., 1994). Petralia et al. (1994a) further demonstrated that NRl-immunoreactivity occur at the post-synaptic membrane specialisations in Purkinje cell spines. With respect to the NR2 subunits, Akazawa et al. (1994) found that rat Purkinje cells may express NR2D mRNA until post-natal day 8 and thereafter express low levels of NR2A mRNA. Accordingly Petralia et al. (1994b) observed that Purkinje cells display a low level of NR2A/B immunoreactivity, also in the post-synaptic densities of Purkinje cell dendritic spines, indicating that low levels of 'NR1-NR2A' receptors may be present at parallel fiber or climbing fiber synapses. Watanabe et al. (1994), however, found that mouse Purkinje cells only express low levels of NR2B (indicated as e2, which is the mouse homolog of NR2B) until one day postnatally, but not at any later stage, whereas according to Monyer et al. (1994) Purkinje cells do not produce any NR2 mRNA at any age. One may conclude from the in situ hybridisation and immunocytochemical data, that in spite of the presence of high levels of NR1 subunit, Purkinje cells both during development and in adulthood are likely to express none or only low amounts of functional NMDA receptors, which is in line with the aut0radiographic data. NR1 subunits can also form receptor-channel complexes in the homomeric configuration, but these channels produce very small currents and are, therefore, unlikely to contribute significantly to the excitatory actions of glutamate in Purkinje cells (Moriyoshi et al., 1991). Orphan receptors
51 and 52 are two related subunits isolated by homology screening. 51 is not produced in the rodent cerebellum, but 52 is selectively expressed by Purkinje cells (Araki et al., 1993; Lomeli et al., 1993). The subunit protein is distributed throughout the somatodendritic domain of the Purkinje Cells, similar to other glutamate receptor subunits (Araki et al., 1993). The function of S1 and 52 is not yet understood. The subunit protein does not bind glutamate receptor agonists and does not aggregate into functional receptors (Lomeli et al., 1993).
3.3.2. Metabotropic glutamate receptors Metabotropic glutamate receptors are coupled to G-proteins and modulate intracellular second messenger systems. The metabotropic glutamate receptors consist of at least seven subtypes that can be subdivided into three subgroups on the basis of sequence homology, agonist selectivity, and second messenger system (Nakanishi, 1992; Tanabe et al., 1992): (1) mGluR1 and mGluR5, that are coupled primarily to activation of phosphoinositide hydrolysis and are activated by quisqualate (QA) and 1S,3R-aminocyclopentane dicarboxylate (1S,3R-ACPD); (2) mGluR2 and mGluR3, that are coupled to inhibition of the cAMP cascade, are sensitive to pertussis toxin, and are activated by 72
The cerebellum." chemoarchitecture and anatomy
Ch. I
1S,3R-ACPD, but are insensitive to QA; and (3) mGluR4, mGluR6 and mGluR7, which are also coupled to inhibition of the cAMP cascade, and are potently activated by L-2-amino-4-phosphonobutyrate (L-AP4), but are insensitive to QA and 1S,3R-ACPD. Metabotropic glutamate receptors have been implicated in multiple neuronal processes including modulation of transmitter release, plasticity phenomena such as long term potentiation and long term depression, and other long term changes of neuronal functions (see Schoepp, 1994 for a review). With the exception of mGluR6 that is expressed only in retina, all metabotropic receptors are expressed in the cerebellum. mGluR1 mRNA is expressed to some extent by most cerebellar neurons (Shigemoto et al., 1992), but is found at very high levels in Purkinje cells. Immunocytochemistry shows that the mGluR1 protein is localized in the spines of Purkinje cell dendrites (Fig. 52) (Martin et al., 1992; Baude et al., 1993; Shigemoto et al., 1994). Dense mGluR1 immunostaining is also associated with the brushes of unipolar brush cells (Jaarsma, Mugnaini, Shigemoto et al., in preparation). Interestingly, mGluR1 immunostaining is not associated with the post-synaptic region of the giant mossy fiber-unipolar brush cell synapses, but instead, occurs at very high levels in spiny appendages and small branchlets that emanate from the dendritic stem that do not have synaptic specialisations (see Mugnaini et al., 1994). Recently workers from Somogyi's group (Baude et al., 1993; Nusser et al., 1994) demonstrated, with immunogold techniques, that mGluR1 immunoreactivity in Purkinje cell spines (as well as in other neurons) was never localized to the postsynaptic membrane specialisations of the synapses, but was associated with perisynaptic and extrasynaptic regions. This is in marked contrast with ionotropic glutamate receptor subunits that are primarily located at the postsynaptic membrane (Fig. 49) (Nusser et al., 1994). It has been proposed that, as a consequence of its peri-and extrasynaptic localization, mGluR1 is only activated during high frequency stimuli, because low frequency stimuli may not release enough glutamate to reach the perisynaptic receptors at significant concentrations (Baude et al., 1993; Nusser et al., 1994). It was originally reported by Kano and Kato (1987) that a QA/transAPCD-sensitive glutamate receptor is critically involved in the induction of long term depression (LTD) of parallel-fiber-Purkinje cell synapses, a cerebellar paradigm of synaptic plasticity that is induced following repetitive stimulation of parallel fibers in conjunction with climbing fiber input (Ito, 1989; Linden and Connor, 1993). Recently strong evidence has been obtained that mGluR1 plays a major role in cerebellar LTD: (1) the induction of LTD could be inhibited with antiserum that inactivated mGluR1 in an in vitro model of LTD (Shigemoto et al., 1994); and (2) LTD could not be induced in a mutant mouse lacking mGluR1 (Aiba et al., 1994). In these animals the anatomy of the cerebellum was not overtly disturbed. The Purkinje cells showed some minor morphological alterations, but had normal excitatory responses upon parallel fiber and climbing activation. Interestingly, the animals showed characteristic cerebellar symptoms such as ataxic gait and intention tremor, which suggest that mGluR1, possibly through its role in LTD, is important in cerebellar function. The mGluR5 receptor is selectively localized to a subpopulation of Golgi cells with the receptor protein localized throughout the somato-dendritic domain of the cells (Abe et al., 1992b; Shigemoto et al., 1993). Also mGluR2 and mGluR3 mRNA's are selectively expressed by Golgi cells (Ohishi et al., 1993, 1994), although mGluR3 may also occur in glial cells (Ohishi et al., 1994; Tanabe et al., 1993). Using an antibody selective for mGluR2 and mGluR3, Ohishi et al. (1994) found that mGluR2/3 immunoreactivity was strongest in Golgi axon terminals in the glomeruli (Figs 53 and 54). The Golgi axon terminals are not in close contact with mossy fibers, but the distance between mossy fiber 73
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 52. A. Photomicrograph of semithin 3/lm thick plastic section of the nodulus of rat cerebellar cortex immunostained with an antibody against the carboxyterminus of the metabotropic glutamate receptor, mGluRl~z (antibody A52) (Shigemoto et al. 1994). Immunoreaction product in the molecular layer (ml) has a punctate distribution. Very little staining occur in the perikarya and primary dendrites of Purkinje cells (PC). In the granular layer moderate and dense immunoreactivity is localized to the perikarya and 'brushes' (open arrows) of unipolar brush cells (asterisks in cell nucleus), respectively. Bar = 20/lm. B. Electron micrograph of the molecular layer showing that puncta within the cerebellar molecular layer correspond to mGluRlctimmunoreactive spines (arrows) of Purkinje cells. Curved arrow point to an immunoreactive spine branching from an unlabelled Purkinje cell dendrite (PCd). pf, parallel fiber terminal. Bar = 0.5/lm. Courtesy of Jaarsma, Dino, Mugnaini, Ohishi and Shigemoto.
terminals and Golgi axon terminals is usually less than 1 ~tm, and it is possible that glutamate released from mossy fibers may diffuse into the intercellular space to activate mGluR2/3 on the Golgi cell axons (see Section 3.2.1.). mGluR2/3 in Golgi axon terminals may be involved in the regulation of inhibitory neurotransmitter release, which would imply that mossy fibers may directly influence inhibitory neurotransmission on granule cell dendrites (e.g. see discussion Ohishi et al., 1994). Both mGluR5 and mGluR2/3 antibodies immunostain subpopulations of Golgi cells (Shigemoto et al., 1993; Ohishi et al., 1994). mGluR2/3 immunoreactive Golgi cells constitute three-quarters of the total population of Golgi cells (defined as GABApositive, parvalbumin-negative cells of the granular layer), whereas only a small population of Golgi cells appears mGluR5 positive. Large mGluR2/3-positive Golgi cells were frequently encountered in the Purkinje cell layer and the superficial part of the granular layer (Fig. 53), and at least in part may represent the candelabrum cells as described by Lain6 and Axelrad (1994, see section 2). In contrast, large mGluR5-positive immunoreactive Golgi cells were mostly found deeper in the granular layer. This indicates that mGluR2/3 and mGluR5 positive Golgi cells represent different subpopulations of Golgi cells. It remains to be determined whether mGluR2/3 and mGluR5 positive cells are entirely exclusive or overlapping populations, and whether yet another 74
The cerebellum." chemoarchitecture and anatomy
Ch. I
b
a
_
-,.~
z~-
.
.
... :.-2: . '~i
-.:..,4 "-:',~":," 11:4 :!T%~:
W
"
.." 9 9 .-
,,. . .
... '
.
,~
J
/
9..~:~..
Fig. 53. a. Immunocytochemicallocalization of mGluR2/3 in a parasagittal section through the vermis of rat cerebellum. The most intense immunoreactivity is seen in the granular layer. Staining in the molecular layer is associated with Golgi cell dendrites. B. Drawing of the section shown in (a) showing presumed Golgi cell bodies with mGluR2/3 immunoreactivity (closed circles) and those without mGlu2/3 immunoreactivity (open circles). G, granular layer; M, molecular layer; P, Purkinje cell layer; W, white matter. Bar = 500 #m. Ohishi et al. (1994).
subpopulation exists that is both mGluR5 and mGluR2/3 negative. It is important to realize that mGluR2/3 as well as mGluR5 immunoreactivity is detected in both small and large Golgi cells, and that, therefore, the segregation of Golgi cells into mGluR2/3 respectively mGluR5-positive and negative cells does not correspond to previous classifications which were based on size (e.g see Palay and Chan-Palay, 1974; but also section 3.6.2.). The L-AP4-sensitive mGluR, m G l u R 4 is expressed at high levels by granule cells (Kristensen et al., 1993; Tanabe et al., 1993), whereas mGluR7 m R N A is produced by Purkinje cells (Okamoto et al., 1994; Saugstadt et al., 1994). Physiological data indicate that the L-AP4 sensitive mGluRs are predominantly located presynaptically, where they may act as autoreceptors to regulate glutamate release. Studies in turtle suggest the presence of a L-AP4-sensitive presynaptic glutamate receptor at the parallel fiber-Purkinje cell synapse (Larson-Prior et al., 1989). Thus, possibly, mGluR4 is located presynaptically on parallel fiber boutons. The ultrastructural localisation of m G l u R 4 and m G l u R 7 remained to be determined at the time of writing this manuscript. 75
Ch. I
J. Voogd, D. Jaarsma and E. Marani
3.4. NITRIC OXIDE: THE CEREBELLAR LOCALIZATION OF NITRIC OXIDE SYNTHASE, GUANYLATE CYCLASE AND CYCLIC GMP Nitric oxide (NO) (see Dawson et al., 1992 and Vincent and Hope, 1992 for reviews) has gained importance as an intracellular and diffusible intercellular messenger in the cerebellum, since the demonstration by Garthwaite et al. (1988, 1989) that N-methyl-Daspartate (NMDA) receptor activation caused an increase in cyclic guanosine 3',5'monophosphate (cyclic GMP) in the cerebellum by stimulating the release of a diffusible messenger with properties similar to a endothelium-derived relaxing factor which was identified as NO. They considered granule cells as the main source of NO and glial cells as the main target for the activation of soluble guanylate cyclase by NO and the production of cyclic GMP. The enzyme nitric oxide synthase (NOS), that produces NO and citrullin from arginine, occurs as several isoenzymes (Knowles et al., 1989). Type I NOS is a constitutive, calcium and calmodulin-dependent enzyme, present in neurons and, possibly, in glia. Type II NOS is calcium-independent and can be induced in macrophages and glial cells by exposure to bacterial lipopolysaccharide (Galea et al., 1992; Murphy et al., 1993). Type III NOS is the endothelial iso-enzyme. NOS-I, II and III are produced by different genes (Bredt et al., 1991; Lamas et al., 1992; Xie et al., 1992; Lowenstein et al., 1992; Lyons et al. 1992; Ogura et al., 1993). NOS displays NADPH-dependent diaphorase
Fig. 54. Ultrastructural localization of mGluR2/3 immunoreactivity in the granular layer of rat cerebellar cortex. Dense immunoreaction products accumulate in axon terminals of Golgi cells, which often make synaptic contacts (curved arrows) with possible granule cell dendrites around a mossy fiber terminal (MT) in the cerebellar glomerulus. Bar = 0.5 r Ohishi et al. (1994).
76
The cerebellum." chemoarchitecture and anatomy
Ch. I
activity and can be demonstrated in aldehyde-fixed tissue by NADPH-dependent reduction of tetrazolium salts to visible formazans (Hope et al., 1991). NOS-I has been localized with antisera to the purified enzyme (Bredt et al., 1990) and by in situ hybridization to NOS-I mRNA (Bredt et al., 1991) in basket cells and in granule cells and their axons, where NOS-I is co-localized with NADPH diaphorase (Bredt et al., 1991; Vincent and Kimura, 1992; Schmidt et al., 1992; Schilling et al., 1994). NADPH-diaphorase-positive granule cells are distributed in a symmetrical pattern of heavily stained clusters, separated by granule cells that were stained weakly, or not at all (Fig. 55) (Schilling et al., 1994). The NADPH-diaphorase-positive granule cell clusters were correlated with the Zebrin pattern in the overlying molecular layer by Hawkes and Turner (1994). A sparse axonal network and a few cells were stained in the cerebellar nuclei (Vincent and Kimura, 1992). Schmidt et al. (1992) also found weak NOS-I immunoreactivity in Bergmann glia and astrocytes where it co-localized with NADPH-diaphorase. NOS-II was expressed by astrocytes in lipopolysaccharide-stimulated cultures. These cells also double-label for NADPH-diaphorase (Galea et al., 1992). Guanylate cyclase, the enzyme responsible for the synthesis of cyclic GMP from guanosine triphosphate, was localized with immunofluorescence in Purkinje, granule stellate and Golgi cells and in oligodendrocytes, astroglia and Bergmann glial fibers of the cerebellar cortex of the rat (Zwiller et al., 1981). The localization in Purkinje and granule cells and in astrocytes was confirmed by Ariano et al. (1982), Nakane et al. (1983) and Schmidt et al. (1992). Bergmann glia and small cells in the molecular and granular layers were weakly stained. Expression of the soluble guanylyl cyclase mRNA in rat cerebellum was moderate in Purkinje, basket, stellate and Golgi cells, weak in granule cells, but could not be demonstrated in glial cells (Matsuoka et al., 1992, see also Burgunder and Cheung, 1994). Cyclic GMP was located with immunohistochemical methods in Bergmann glia (Cumming et al., 1977, 1979; Chan-Palay and Palay, 1979; Ariano et al., 1982) and in a subpopulation of stellate and basket cells (Chan-Palay and Palay, 1979). Its preferential localization in Bergmann glia and cerebellar astrocytes was stressed by Berkelmans et al. (1989) and De Vente et al. (1989, 1990), using antibodies against conjugates of cyclic GMP and activation of cyclic GMP by sodium nitroprusside in slices of rat cerebellum. They observed a patchy distribution of the reactive Bergmann glia in the molecular layer (Fig. 56). Purkinje and granular cells remained unstained. Immunoreactive varicose (mossy?) fibers and astrocytes and/or Golgi cells were observed in the granular layer. Owing to the differential localizations of NOS, guanylate cyclase and cyclic GMP, the cellular basis for the actions of cerebellar NO remains difficult to establish. Basket and stellate cells appear to be the only cell types that can be stimulated by NMDA receptors (Quinlan and Davis, 1985; Hussain et al., 1991) that contain both NOS-I, guanylate cyclase and cyclic GMP. It has been suggested that carbon monoxide (CO) is an activator of soluble guanylyl cyclase in Purkinje cells. Heme oxygenase-2, which degrades heme to biliverdin and releases carbon monoxide in the process, was shown to be co-localized with guanyl cyclase in rat Purkinje and granule cells with in situ hybridization histochemistry (Verma et al., 1993). 3.5. ADENOSINE, 5'-NUCLEOTIDASE AND ADENOSINE DESAMINASE Adenosine-like immunoreactivity was found in rat Purkinje cells, using polyclonal anti77
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 55. Coronal section through the copula pyramidis (lobule VIII). In the adult rat granule cells in the lateral tip of the copula pyramidis show strongly reduced staining intensity for NADPH-diaphorase, in contrast to the medial copula, where a cluster of heavily stained granule cells can be seen. g, granular layer; m, molecular layer. Scale bar = 200/lm. Schilling et al. (1994).
sera against a conjugate of the adenosine derivative laevulinic acid (Braas et al., 1986). Staining was present in the cell soma outside the nucleus, extending in the dendrites. Weaker staining was observed in the granular layer. Adenosine is co-released with adenosine triphosphate (ATP) and certain neurotransmitters (Richardson and Brown, 1987). High affinity uptake sites for adenosine are present in all layers of the cerebellar cortex (Marangos et al., 1982; Nagy et al., 1985; Biss6rbe et al., 1985). Steady state concentrations of adenosine are maintained through the activities of only three enzymes, 5'-nucleotidase (5'-N), adenosine kinase and adenosine deaminase. Adenosine kinase and adenosine deaminase were located mainly in the soluble fractions of rat cerebellar homogenates, whereas 5'-N was present in subcellular fractions (Philips and Newsholme, 1979), mainly in the synaptosomal fraction (Marani, 1977). Adenosine deaminase-immunoreactivity in rat cerebellum was present with one out of five polyclonal sera prepared by Nagy et al. (1988). Staining was present in most Purkinje cells with a variation in intensity. Staining was observed in the Purkinje cell axons and terminals in the cerebellar and vestibular nuclei. The localization of 5'-N will be discussed below. Adenosine blocks the parallel fiber-induced simple spike discharge in Purkinje cells (Kostopoulos et al., 1975) but not the climbing fiber-mediated synaptic transmission (Kocsis et al., 1984). The effect of adenosine is presynaptic and is mediated by A1adenosine receptors that are located on parallel fibers. A 1-adenosine receptors are coupled to pertussis toxin-sensitive G proteins and inhibit adenyl cyclase. Activation of A 1-adenosine receptors decreases transmitter release from the terminals (Dolphin and Prestwich, 1985, see Fredholm and Dunwiddie, 1988, for a review). The presence of Al-adenosine receptors on parallel fibers was demonstrated autoradiographically by 78
The cerebellum." chemoarchitecture and anatomy
Ch. I
Fig. 56. cGMP-immunostaining of adult rat cerebellum. A. Section of a cerebellar slice that was incubated with cyclic GMP antiserum, in the presence of 1 mM isobutyl-methylxanthineto inhibit phosphodiesterase activity and 10 r nitroprusside and post-fixed in paraformaldehyde. B. Same areas of the same section as shown in (A) after removal of cGMP-immunostaining using the methods of Tramu et al. (1978) and reincubation of the section with glial fibrillary acidic protein-antiserum. Note presence of cGMP-immunoreactivity in Bergmann glial fibers and in thin varicose fibers (arrows in A) and in astrocytes or Golgi cells (arrow head in A). Bars = 100 r De Vente et al. (1989).
Goodman and Snyder (1982) and Goodman et al. (1983) using specific binding of [3H]cyclohexyladenosine ([3H]CHA) and Weber et al. (1990), using the antagonist [3H]DPCPX (Fig. 57). Binding was highest over the molecular layer, with lower concentrations in the granular layer. Binding was absent in the granuloprival cerebellum of 'weaver' mice (Goodman et al., 1983; Wojcik and Neff, 1982, 1983). Al-adenosine receptors were present over the entire molecular layer; no bands of high activity, corresponding to the 5'-N pattern, were observed (Fastbom et al., 1987). Adenosine is released in a Ca2+-dependent manner by K + stimulation from rat cerebellar slices (Cu6nod et al., 1989; Do et al., 1990). The stimulated release of adenosine was decreased by 60-70% in vermis and hemisphere, in slices from 3-acetylpyridine-treated rats, which may indicate that the released adenosine, at least in part, is released by climbing fibers. The 'climbing fiber-dependent' adenosine release, however, occurs with some time delay after the K + stimulus. Adenosine, therefore, has been proposed to be derived from extracellular degradation of released nucleotides by ectonucleotidases. Inhibition of 5'-nucleotidase (5'-N) by ~,fl-methylene-ADP and GMP, indeed, decreased stimulated adenosine release by 50-60%. 5'-Nucleotidase (5'-N) is an integral glycoprotein of the cellular plasma membrane in a wide range of animal cells. Its functional role is still unclear. 'Possibilities .... include recovery of purines and pyrimidines from the extracellular space, the extracellular formation of neuromodular adenosine from released nucleotidases and non-enzymatic functions related to the interaction of 5'-nucleotidase with compartments of the cytoskeleton and extracellular matrix' (Schoen et al., 1987). 5'-N catalyses the production of adenosine by the hydrolytic cleavage of 5'-nucleotide monophosphates (i.e. adenosine5'-monophosphate). The development of 5'-N in the cerebellum was studied by Schoen et al. (1987, 1988, 1990). 5'-N in the molecular layer of mouse cerebellum is distributed in positive and negative parasagittal bands (Scott, 1963). The distribution of cerebellar 5'-N has been reviewed by Marani (1986). Its zonal distribution in mice is very similar to the distribution of the m a b Q l l 3 (Zebrin)-positive dendrites of Purkinje cells in the molecular layer (Marani, 1986; Eisenman and Hawkes, 1989) (Figs 58A, 130, 131, 135) (Section 6.1.4.). The 79
Ch. I
J. Voogd, D. Jaarsma and E. Marani ..
.'..'.." 9 .... 7 ~ - . . "
M[.
i':
.,.".,
" ":.-52-"." 9
~
.
, . ~ ' .
.
9
~ :,,.~.~
"~,'.'""
~." "~:. .'": - : , - ' ' : i ' : . . ""u : . . , ;
. . . .
.
~
I
:
..,
.
9
.
":
."
. - ~
~ -
..
.".,..g
-o.
9
~
i . 9
.
-
.
-,'.i.
g,.'5./i
~ . . i
,
""
.'..;
".- ~ : ' " ~,-.:i ~
~
~o
~'::
,
:
'. ".'.;"
".
.
"7:'"
~~
,
..
,,a.
.
.
,
..
.
9
--
.
.. .
.
.
.
~
..
.
.
9
.
. -
,
..
.
- .
,
.
, .
9
.~
: .
..
9 ~
"
,
.
.
.
.
.
.
-,
.
..,
9
.
,.
,
.
.
,
,
~
,
.
,
9 .,
, .
.
.
:
.
..
.,.
.~ ~
.
~~
9 . - .
9
"
.
.
.
,
.
.
9
.
.
:
~..
.
.
,
.
-,~
,
. . ,
t 9 I, 9
~.
.
.
"
9 . . ~
.
.
9
*
.
.
9
-'.. ,'".
~
~ .
,.-.
~,
,
,
. ,
: .
.-..~"
g
9
9
9 .
.
.
. .
9
.
9 ~
.
.
. ~ l w .
.
Fig. 57. Light microscopic autoradiographic localization of A l adenosine receptors using [3H]DPCPX binding
to a section of rat cerebellum. Left. Photomicrograph of a pyrorine Y-stained section of rat cerebellum showing the molecular layer (ML), the Purkinje cell perikarya (PC), the granule cell layer (GL), and some white matter ( W M ) . R i g h t . Darkfield photomicrograph of the tissue section incubated with 0.8 nM [3H]DPCPX and apposed to nuclear emulsion-coated coverslips. The silver grains are from the same area shown on the left. Note the high density of A~ adenosine receptors in the molecular layer and moderate labelling in the granule cell layer. The white matter and the Purkinje cell bodies showed background levels of labelling. Bar = 5 0 / l m . Weber et al. ( 1 9 9 0 ) .
reaction product in the histochemical procedure of Scott (1964, 1965, 1967) is uniformly distributed within the bands of high 5'-N activity in the molecular layer. In the hemisphere the staining in some of the 5'-N bands is less uniform and assumes the aspect of radially disposed striations. Hess and Hess (1986) tentatively identified these striations as the processes of the Bergmann glia. These authors found 5'-N in the molecular layer of Purkinje cell-deficient mice (pcd and nr strains) to be reduced and the residual enzyme activity to be localized in approximately the same position as the surviving Purkinje cells. This would imply that the expression of 5'-N in Bergmann glia is regulated by the adjoining Purkinje cells. Marani's electron microscopic enzyme-histochemical studies (Marani, 1981, 1982a,b, 1986), favoured a localization of the enzyme in the subsurface cisterns and the spine apparatus of the spines of the Purkinje cell dendrites and within boutons of parallel fibers (Fig. 58B-D). A non-zonal distribution of 5'-N in the somata of Purkinje cells and other large cells of the cerebellar cortex was observed when different substrates for the enzyme histochemical reaction for 5'-N were used (Scott, 1967; Marani, 1982a and b, 1986; Hess and Hess, 1986). According to Marani this represents a rest-activity of non-specific phosphatases, that disappears when the appropriate inhibitors are used. The presence of 5'-N in parallel fibers was disputed by Hess et al. (1983), who showed that 5'-N remains at significant levels in the molecular layer in agranular 'weaver' 80
The cerebellum." chemoarchitecture and anatomy
Ch. I
cerebellum. According to Kreutzberg et al. (1978) 5'-N is predominantly associated with glial membranes. Schoen et al. (1987, 1988), used monoclonal and polyclonal sera directed against rat liver 5'-N in the localization of cerebellar 5'-N in addition to the enzyme-histochemical techniques. They found the enzyme to be situated at the outer border of the plasma membranes of Bergmann glial fibers in the molecular layer, astroglial endfeet around blood vessels and glial processes surrounding Purkinje and granule cells (Fig. 59). They were unable to confirm Marani's observations of an intracellular localization of the enzyme. The study of Schoen et al. (1987) was done in rats, which do not have the longitudinal band pattern of 5'-N with their antibody directed against this enzyme. Balaban et al. (1984) observed an increase of cerebellar 5'-N in the P2 (synaptosome) fraction after climbing fiber activation with harmaline in rats (Fig. 60). Harmaline synchronizes the discharge in climbing fibers from certain parts of the inferior olive and induces a rhythmic tremor (Sj61und et al., 1977, 1980). Two different climbing fiber induced effects, therefore, may be involved in adenosine-mediated blockade of transmission in parallel fiber-Purkinje cell synapses: an increased release of nucleotides and an increase of cerebellar 5'-N. Loss of climbing fiber-induced 5'-N and/or adenosinemediated blockade of transmission in the parallel fiber-Purkinje cell synapses (see Marani, 1986) would explain the long-term increase of simple spike activity that occurs when complex spikes are suppressed by destruction or inactivation of the inferior olive in rats (Colin et al., 1980; Montarolo, 1982). Bloedel and Lou (1987), however, observed a short-term facilitation of transmission in the mossy fiber-parallel fiber-Purkinje cell pathway on stimulation of climbing fibers in the cat. This difference may be due to species-dependent differences in 5'-N mediated formation of adenosine or to a facilitation at the level of mossy fiber-granule cell synapse. If the formation of adenosine is largely dependent on the degradation of nucleotides by 5'-N, the zonal distribution of this enzyme in different species and of the climbing fibers which promote their release would be of crucial importance (see Marani (1986) and Section 6.1.4.). 3.6. INTERNEURONS OF THE CEREBELLAR CORTEX Stellate, basket and Golgi cells are inhibitory (Eccles et al., 1964a, 1966a,b,c,d, 1967). It was against this background that Uchizono (1965) (see also Uchizono, 1969 for a review) formulated and tested his hypothesis that excitatory and inhibitory axon terminals in aldehyde fixed tissue can be distinguished by the shape of their synaptic vesicles (Fig. 61). Inhibitory boutons contain flattened vesicles (F-type boutons) and excitatory boutons contain spherical vesicles (S-type boutons). Earlier Gray (1959) distinguished two types of synaptic junction, which were also supposed to represent the excitatory and inhibitory synapse (Landis and Reese, 1974). Gray type 1 junctions are characterized by a widening of the synaptic cleft that contains dense material and a distinct asymmetry caused by the presence of a dense undercoating of the postsynaptic membrane. It was considered to be excitatory. The thickening of the pre- and postsynaptic membranes in the Gray type 2 junction is symmetrical and the cleft is narrow; this type was supposed to be inhibitory. According to Uchizono (1969) there is an excellent correlation in the cerebellar cortex of the cat of S-type boutons with Gray's type 1 synaptic junctions and of F-types with a synapse of Gray's type 2. For the excitatory connections of the mossy and climbing fibers and for the parallel fiber-Purkinje cell synapse the correlation with S-type terminals and Gray 1 synaptic junctions still is valid. For the terminals of the inhibitory interneurons of the cerebellar cortex (Golgi cells: pleomorphic vesicles, synap81
Ch. I
J. Voogd, D. Jaarsma and E. Marani
Fig. 58. Light and electron micrographs of incubations for 5-nucleotidase according to Scott (1967). A. Detail of the light microscopic location of 5'-nucleotidase in uvula (IX) and pyramis (VIII). B. Electron microscopic location of 5'-nucleotidase reaction products in the subsurface cisternae of a Purkinje cell dendrite. C. Electron microscopic localization of 5'-nucleotidase in the spine apparatus of Purkinje cell dendritic spines (asterisks). D. Localization of reaction product in a parallel fiber bouton, synapsing on a Purkinje cell dendritic spine. Bars in A = 1 mm, in B,C = 0.5 ~tm, in D = 0.25 ~tm. Marani (1977).
82
The cerebellum. chemoarchitecture and anatomy
Ch.I
Fig. 59. 5'-Nucleotidase immunohistochemical staining of rat cerebellum. A. Immunofluorescence. B. PAPmethod. Enzyme activity is predominantly found within the molecular layer on Bergmann glial fibers (long arrows). Purkinje cells are surrounded by fine rims of reaction product (small arrows). Within the granular layer 5'-nucleotidase activity is diffusely scattered between granule cells (arrow heads). Vibratome sections. C. Longitudinally sectioned Bergmann glia cell processes (B) of the molecular layer of rat cerebellum. Fine DAB reaction product is located on adjacent membranes of these processes (arrows). Bars in A,B = 50 ~tm, in C = 0.5/lm. Schoen et al. (1987).
tic junction resembles Gray type 1; basket cells: ellipsoid, irregular and spherical vesicles, Gray type 2; stellate cells: flattened vesicles, Gray type 2) there is a greater variation in morphology (Palay and Chan-Palay, 1974). Cell bodies and terminals of the Golgi, basket and stellate cells can be labelled with selective uptake of [3H]GABA (H6kfelt and Ljungdahl, 1970, 1971; Schon and Iversen, 1972), immunostaining with antibodies against GAD (Saito et al., 1974; McLaughlin et al., 1974; Oertel et al., 1981b; Mugnaini and Oertel, 1985) and in situ hybridization for G A D 6 5 and G A D 6 7 (Wuenschell et al., 1986; Julien et al., 1987; Esclapez et al., 1993; Feldblum et al., 1993). They are also immunostained with antibodies against conjugates 83
Ch.I
J. Voogd, D. Jaarsma and E. Marani O----o 125
-
100
-
Intact
inj.)
olive (vehicle
3-AP olivectomy
r ~ (9
A
O t,.. Q.
_
_
75
-
> ..--l-., r~
/
1. ~ . ' . . ~ . , , . : : ,,:,;,,'. . . . . . ~ . . .".~ .. ~.~ ~,.~ :,_.~~~:.~:~..,..-/.:~.~ ,; ~:~;,.',
.... ,
... 9
":..,.,,..~...,, :,
:,:X'..
'..~-
:,'."~,,::'~
:,~\
'
/,' I'.-,
/
--';.,.~"
.-t..,.'...t t.~.,
, : : - ,,:: ~i!=~'i~iii!e' ==~i:i~=i~i',i!=!!i~:i! ~gi'~g=,~'=!ii~i
i ~ : 7 > =================================================,~;' ::i. .
....:= ~:,..............................
" ~ : ; . ~ i : : i ~ ............~'::~:...........
.
R
~;~',
~::::~:::. aa=:~: ....... ==.....~,,~,~::s::~:=, I!1~ : .....
..
.................................................... ==':~,='==:==~'===~=~',~i~!~!~i .....{~{{i{igi~=i
::==:=7 ~:~.......~
~
:E=:....... g=~==:)=g{{i>=~ :~.=~{ ==:E, "'. ,,',.',.,. .". .'' .~:':=: =: a =~=:= a:==:=~:~::"'~;~'~'":~ ...............'*::'~:~;:;~'~'~""'~"'":'"~'::~":~"~":~:;:'~:"~'';"~:~' ............... ?=,=g!=~=,qi====:==:=7,i::~:i=(,~.....=:===== '........... ::~'~::;: :'.: . :.......... . ............
~=:~,,~.~,~,,~,~:,=~,,.,,~,~.:: ~: .......... "~ ............ .........
-:~~:~:.~i~!;!~i, ~II! :~i~,{!i~}~iii~!ir :~:
'%i~,,~,~i:,!!~:!,,',:,':~:!!:~%~i~i~ ;'~ ~,,il~,:~:~:~:~'~'~''~ik~~ ::
:~i:!"!~.:~:ii::. ,ii{;~ili~,;
. . . . . . . . . .:...:./t:.~:~..::.
;;:;~i~4;~.:~ ........
,
~:,~
9 ,
,:' ~.
":"x::~ ! ~ "
.:: ~: : "~:::.'~
.~
:":~:
".
' ....
: i ~' 9
~.;~
~. .... :
:~'::;...:~
.,. :..:
~
''
~; :;; :,,,::.":::': ;:.;;:.
9 -e:..
..:~:
:~:..~.. ....~::-.. .......
:. ~; , ..
:::....: ..::~ ...... .~! ...."..:..:;'.'