Sunteți pe pagina 1din 853

The Hippocampus Book

This page intentionally left blank

The Hippocampus Book


Per Andersen, Richard Morris,

David Amaral, Tim Bliss, & John OKeefe

Oxford University Press, Inc., publishes works that further
Oxford Universitys objective of excellence
in research, scholarship, and education.

Oxford New York

Auckland Cape Town Dar es Salaam Hong Kong Karachi
Kuala Lumpur Madrid Melbourne Mexico City Nairobi
New Delhi Shanghai Taipei Toronto

With offices in
Argentina Austria Brazil Chile Czech Republic France Greece
Guatemala Hungary Italy Japan Poland Portugal Singapore
South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright 2007 by Oxford University Press, Inc.

Published by Oxford University Press, Inc.

198 Madison Avenue, New York, New York 10016

Oxford is a registered trademark of Oxford University Press

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 permission of Oxford University Press.

The hippocampus book / Per Andersen ...[et al.], editors.

p. ; cm.
Includes bibliographical references and index.
ISBN-13: 978-0-19-510027-3
ISBN-10: 0-19-510027-1
1. Hippocampus (Brain)Physiology. I. Andersen, Per.
[DNLM: 1. Hippocampusanatomy & histology. 2.
Hippocampusphysiology. 3. Memoryphysiology. 4. Models,
Theoretical. 5. Synaptic Transmission. WL 314 H666 2007]
QP383.25.H57 2007
612.825dc22 2006007088

Printed in the United States of America
on acid-free paper

Kari Andersen
Hilary, Louise, and Josephine Morris
David Joseph and Jennie Amaral
Isabel Vasseur
Eileen OKeefe
This page intentionally left blank


The hippocampus is one of the most widely studied regions of card-carrying hippocampologists. Inevitably, our conversa-
the brain and is of interest to a wide spectrum of neuroscien- tion turned to our favorite brain structure and in particular to
tists, ranging from those who study its normal structure and the lack of reviews and surveys to which we could direct edg-
function to others who study its malfunction in various dis- ling hippocampologists. Eventually, Per Andersen threw out
eases and pathological conditions. Information about it has the challenge: Could the four of us write a treatise on the hip-
accrued across a considerable span of time and is spread over pocampal formation that captured the essentials, derived as
a wide variety of publications. It seems all the more remark- they were from so many branches of neuroscience? He won-
able, therefore, given the explosion of texts on every conceiv- dered whether, among us, we could cover the entire eld from
able topic in neuroscience, that there is no single, synapse to behavior by way of the cells and the nerve nets in
comprehensive source of information on the hippocampal between. The original crew had a not inconsiderable breadth
formation. It would clearly be helpful to the community of of experience from physiology, psychology, and behavioral
hippocampologists to have the essential information about science, but we quickly realized the essential need for struc-
hippocampal neurobiology gathered together in one volume. tural information as one of the foundations of such a book.
This book is our attempt at a reasonably comprehensive sur- We were happy to bring on board a neuroanatomist with great
vey of hippocampal research, as viewed through many eyes knowledge and experience of the hippocampus, and the gang
and collected with a wide variety of methods. We hope that of four became ve.
the book will be useful at several levels. For those new to the Over the next 10 years, the Editors met in numerous
arena of hippocampal research, we hope it will act as a primer delightful venues in the United States and Europe to read each
pointing to the important advances of the past decades and others work and to confer on the book. Although much writ-
suggesting important research paths to be pursued; for hip- ing was done during this phase, it gradually dawned on us that
pocampal veterans, we hope that by laying bare the many two formidable obstacles made our objective signicantly
uncertainties and open questions that remain, it will provide more difficult than we had originally anticipated. First, the
ample stimulus to see old problems with new eyes and to pur- material on the hippocampus was even more extensive than
sue them with renewed vigor. we had expected and, with the explosion of neuroscience, was
The hippocampus is a structure eminently worthy of study growing rapidly day by day. A recent PubMed search elicited
in its own right, but it has also been seen by many as a model more than 73,000 references to the target word hippocam-
structure for the study of cortical function and plasticity in pus. More daunting still, the breadth of the eld had
general. The twin aims of the book therefore are to discuss expanded with the introduction of new methods and infor-
hippocampal structure and function in its own terms and to mation, not least from the rapidly developing elds of molec-
highlight how study of the hippocampus has revealed ideas ular biology, genetics, and development. To cover these
and advances that are of wider signicance and generality for important elds adequately, we decided to ask a set of col-
neuroscience. leagues with the appropriate specialist knowledge to join us in
The inception of The Hippocampus Book occurred many the effort. We were fortunate to attract a powerful team, and
years ago, at the end of a symposium in the beautiful, stimu- the Editors are extremely grateful to these colleagues for their
lating city of Palermo in Sicily. We ended up at a street-side important contributions.
cafe, each with a glass of birra glowing in the setting We hope that The Hippocampus Book will prove useful to
Mediterranean sun. We were four old friends, all long-term its readers. We are condent that many share our affection for
viii Preface

this uniquely fascinating structure. If the book helps future ory and its involvement in a host of psychological and psychi-
neuroscientists in their planning and execution of hippocam- atric conditions, the hippocampal formation will undoubt-
pal studies, our project will have been a success. If, in addition, edly continue to be an active focus of energetic research. Our
the book facilitates new lines of thought and experimental hope is that The Hippocampus Book will act as a springboard
approaches that challenge existing ideas and concepts, our and a guide for some of that research.
reward will be even greater. With its important role in mem-
Per Andersen
Richard Morris
David Amaral
Tim Bliss
John O Keefe


Over the years, several individuals and institutions have Many dozens of our colleagues have been badgered with
helped us in various ways with the creation of this book. An questions over the years and have replied with patience
important planning session took place at The Institute for and dispatch. Per Andersen particularly thanks Theodor
Neuroscience, then at The Rockefeller University. We thank its Blackstad, Gyorgyi Buzsaki, Leif Gjerstad, Vidar Jensen,
Director Gerald Edelman and Research Director Einar Gall for Edward Jones, Bruce Piercey, Geoffrey Raisman, Eric Rinvik,
their generous support. Twice we were accommodated for Thomas Sears, and Jon Storm-Mathisen. Finally, we express
writing sessions at the Centre for Advanced Studies at the our particular gratitude to Fiona Stevens, at Oxford University
Academy of Science and Letters in Oslo, and we thank the Press, New York for keeping faith with our project through too
director of the Centre Vigdis Ystad and the President of the many fallow periods and, also at OUP to Joan Bossert and
Academy Bjarne Waaler for their hospitality. At other times Mallory Jensen for the graceful pressure they applied during
we met in La Jolla, California, Overstrand, Norfolk and in the nal months while shepherding the book to publication.
London at the Royal Society and the Novartis Institute. We Finally, we are deeply grateful for the efficient, yet gracious
enjoyed several editorial meetings in Edinburgh at the Centre manner, in which Berta Steiner at Bermedica Production
for Neuroscience. To all these institutions we express our helped us to bring the book to fruition.
This page intentionally left blank


Chapter 1 2.4.3 New Anatomical Techniques that Revolutionized

The Hippocampal Formation 3 Connectivity Studies 18
2.4.4 Predominantly Unidirectional Connectivity Between
Per Andersen, Richard Morris, David Amaral, Tim Bliss,
Cortical Strips 20
and John OKeefe
2.4.5 New Tracing Studies Using Axonal Transport 20
1.1 Overview 3 2.4.6 Electron Microscopy Offers New Opportunities 21
1.2 Why Study the Hippocampal Formation on its 2.4.7 Hippocampal Synapses Are Highly Plastic: Early
Own? 4 Studies of Sprouting 21
1.3 Dening the Contemporary Era 5 2.4.8 Hippocampal Neurons: Transplantable with
1.4 Organization and Content of the Book 6 Retention of Many Basic Properties 22
2.4.9 Hippocampal Cells Grow Well in Culture 22
Chapter 2
2.4.10 Development of Hippocampal Slices: From Seahorse
Historical Perspective: Proposed Functions, Biological to Workhorse 22
Characteristics, and Neurobiological Models of the 2.5 Several Neurophysiological Principles Have Been
Hippocampus 9 Discovered in Hippocampal Studies 23
Per Andersen, Richard Morris, David Amaral, Tim Bliss, 2.5.1 Identication of Excitatory and Inhibitory
and John OKeefe Synapses 23
2.1 Overview 9 2.5.2 Gray Type 2 Synapses are Inhibitory and are Located
2.2 The Dawn of Hippocampal Studies 9 on the Soma of Pyramidal and Granule Cells 23
2.2.1 A Famous Dispute About the Signicance of the 2.5.3 Gray Type 1 Synapses are Excitatory and are Located
Hippocampus 9 on Dendritic Spines 24
2.3 Early Ideas About Hippocampal Function 10 2.5.4 Long-lasting Alterations of Synaptic Efficiency After
2.3.1 The Hippocampal Formation and Olfactory Physiological Stimulation 24
Function 10 2.5.5 Hippocampal Systems: Exhibiting Several Types of
2.3.2 The Hippocampal Formation and Emotion 11 Oscillatory Behavior 25
2.3.3 The Hippocampal Formation and Attention 2.5.6 Studies of Epileptiform Behavior 25
Control 12 2.6 Development of Methodological Procedures for
2.3.4 The Hippocampal Formation and Memory 13 General Use 26
2.3.5 More Direct Evidence for Hippocampal Involvement 2.6.1 Hippocampus as a Test Bed for Microelectrode
in Memory 13 Work 26
2.3.6 The Hippocampus as a Cognitive Map 16 2.6.2 Pioneers of Intracellular Recording 26
2.3.7 Conclusions 16 2.6.3 Tetrode Development 27
2.4 Special Features of Hippocampal Anatomy 2.6.4 Field Potential Analysis 27
and Neurobiology 16 2.6.5 Histochemistry: Pioneered in the Hippocampus 30
2.4.1 Early Neuroanatomical Studies of the 2.6.6 Pharmacological Analysis of Cellular Properties 30
Hippocampus 17 2.6.7 Development of Computational Models of Neural
2.4.2 New Fiber Tracing Methods 18 Networks 31
xii Contents

2.6.8 The Hippocampal Formation: A Test Bed for 3.7.3 Functional Implications of the Transverse Topography
Several Types of Neural Dysfunction and of Connections 108
Neuropathology 31 3.7.4 Serial and Parallel Processing in the Hippocampal
References 31 Formation 109
3.8 Conclusions 109
Chapter 3 References 110
Hippocampal Neuroanatomy 37 Chapter 4
David Amaral and Pierre Lavenex Morphological Development of the Hippocampus 115
3.1 Overview 37 Michael Frotscher and Lszl Seress
3.1.1 Hippocampus: Part of a Functional Brain System 4.1 Overview 115
Called the Hippocampal Formation 37 4.2 Neurogenesis and Cell Migration 115
3.1.2 Similarities and Differences Between the Hippocampal 4.2.1 Pyramidal Neurons 116
Formation and other Cortical Areas 37 4.2.2 Granule Cells 116
3.1.3 Hippocampal Formation: With A Unique Set of 4.2.3 Local Circuit Neurons and Hilar Neurons 118
Unidirectional, Excitatory Pathways 38 4.2.4 Determinants of Neuronal Migration in the
3.1.4 Hippocampus of Humans and Animals: Same or Hippocampus 120
Different? 39 4.3 Development of Hippocampal Connections 120
3.1.5 Synopsis of the Chapter 39 4.3.1 Entorhinal Connections 121
3.2 Historical Overview of Hippocampal Nomenclature 4.3.2 Commissural Connections 123
Whats in a Name? 39 4.3.3 Septal Connections 125
3.2.1 Denition of Hippocampal Areas: Denition 4.3.4 General Principles Underlying the Formation of
of Terms 42 Synaptic Connections in the Hippocampus 126
3.2.2 Subdivision of Hippocampal Areas 42 4.4 Development of the Primate Hippocampal
3.2.3 Major Fiber Bundles of the Hippocampal Formation 127
Formation 44 4.4.1 Neurogenesis 127
3.3 Three-dimensional Organization and Major Fiber 4.4.2 Neuronal Differentiation 127
Systems of the Hippocampal Formation 44 References 128
3.3.1 Rat Hippocampal Formation 44
3.3.2 Major Fiber Systems of the Rat Hippocampal
Chapter 5
Formation 47
3.3.3 Monkey Hippocampal Formation 48 Structural and Functional Properties of Hippocampal
3.3.4 Human Hippocampal Formation 49 Neurons 133
3.4 Neuroanatomy of the Rat Hippocampal Nelson Spruston and Chris McBain
Formation 51 5.1 Overview 133
3.4.1 Dentate Gyrus 55 5.2 CA1 Pyramidal Neurons 133
3.4.2 Hippocampus 67 5.2.1 Dendritic Morphology 134
3.4.3 Subiculum 76 5.2.2 Dendritic Spines and Synapses 136
3.4.4 Presubiculum and Parasubiculum 81 5.2.3 Excitatory and Inhibitory Synaptic
3.4.5 Entorhinal Cortex 83 Inputs 137
3.5 Chemical Neuroanatomy 94 5.2.4 Axon Morphology and Synaptic Targets 138
3.5.1 Transmitters and Receptors 94 5.2.5 Resting Potential and Action Potential Firing
3.5.2 Steroids 95 Properties 139
3.6 Comparative Neuroanatomy of the Rat, Monkey, and 5.2.6 Resting Membrane Properties 141
Human Hippocampal Formation 95 5.2.7 Implications for Voltage-Clamp Experiments in CA1
3.6.1 Neuron Numbers 96 Neurons 144
3.6.2 Comparison of Rat and Monkey Hippocampal 5.2.8 Attenuation of Synaptic Potentials in CA1
Formation 96 Dendrites 144
3.6.3 Comparison of Monkey and Human Hippocampal 5.2.9 Mechanisms of Compensation for Synaptic
Formation 104 Attenuation in CA1 Dendrites 144
3.7 Principles of Hippocampal Connectivity and 5.2.10 Pyramidal Neuron Function: Passive Versus Active
Implications for Information Processing 107 Dendrites 145
3.7.1 Highly Distributed Three-Dimensional Network of 5.2.11 Dendritic Excitability and Voltage-Gated Channels in
Intrinsic Connections 107 CA1 Neurons 146
3.7.2 Functional Implications of the Septotemporal 5.2.12 Sources of CA2 Elevation in CA1 Pyramidal Neuron
Topography of Connections 107 Dendrites 148
Contents xiii

5.2.13 Distribution of Voltage-Gated Channels in the Chapter 6

Dendrites of CA1 Neurons 149 Synaptic Function 203
5.2.14 Functional Implications of Voltage-Gated Channels
Dimitri Kullmann
in CA1 Dendrites: Synaptic Integration and
Plasticity 152 6.1 Overview 203
5.2.15 General Lessons Regarding Pyramidal Neuron 6.2 General Features of Synaptic Transmission:
Function 152 Structure 204
5.3 CA3 Pyramidal Neurons 153 6.2.1 Transmitter Release and Diffusion 205
5.3.1 Dendritic Morphology 154 6.2.2 Receptors and Receptor Activation 207
5.3.2 Dendritic Spines and Synapses 155 6.2.3 Quantal Transmission 209
5.3.3 Excitatory and Inhibitory Synaptic Inputs 155 6.2.4 Short-term Plasticity 210
5.3.4 Axon Morphology and Synaptic Targets 156 6.3 Glutamatergic Synaptic Transmission 211
5.3.5 Resting Potential and Action Potential Firing 6.3.1 AMPA Receptors 213
Properties 156 6.3.2 Kainate Receptors 215
5.3.6 Resting Membrane Properties 158 6.3.3 NMDA Receptors 215
5.3.7 Active Properties of CA3 Dendrites 158 6.3.4 Co-localization of Glutamate Receptors 216
5.4 Subicular Pyramidal Neurons 159 6.3.5 Metabotropic Glutamate Receptors 216
5.4.1 Dendritic Morphology 159 6.3.6 Receptor Targeting and Anchoring 218
5.4.2 Dendritic Spines and Synaptic Inputs 159 6.4 GABAergic Synaptic Transmission 218
5.4.3 Axon Morphology and Synaptic Targets 160 6.4.1 GABAA Receptors 218
5.4.4 Resting and Active Properties 160 6.4.2 GABAB Receptors 220
5.4.5 Mechanisms of Bursting 162 6.5 Other Neurotransmitters 220
5.4.6 Membrane Potential Oscillations 162 6.6 Special Features of Individual Hippocampal
5.5 Dentate Granule Neurons 162 Synapses 222
5.5.1 Dendritic Morphology and Spines 163 6.6.1 Small Excitatory Spine Synapses 223
5.5.2 Excitatory and Inhibitory Synaptic Inputs 163 6.6.2 Mossy Fiber Synapses 226
5.5.3 Axon Morphology and Synaptic Targets 164 6.6.3 Other Glutamatergic Synapses on
5.5.4 Resting Potential and Action Potential Firing Interneurons 228
Properties 164 6.6.4 Inhibitory Synapses 229
5.5.5 Resting Membrane Properties 165 6.7 Unresolved Issues 231
5.5.6 Active Properties of Granule Cells 166 References 231
5.6 Mossy Cells in the Hilus 166
Chapter 7
5.6.1 Dendritic Morphology and Spines 168
5.6.2 Excitatory and Inhibitory Synaptic Molecular Mechanisms of Synaptic Function in the
Inputs 168 Hippocampus: Neurotransmitter Exocytosis and
5.6.3 Axon Morphology and Synaptic Targets 168 Glutamatergic, GABAergic, and Cholinergic
5.6.4 Resting and Active Properties 168 Transmission 243
5.6.5 Other Spiny Neurons in the Hilus 168 Pavel Osten, William Wisden, and Rolf Sprengel
5.7 Pyramidal and Nonpyramidal Neurons of Entorhinal 7.1 Overview 243
Cortex 169 7.2 Neurotransmitter Exocytosis 243
5.7.1 Stellate Cells of Layer II 170 7.2.1 Introduction: Proteins Involved in Synaptic
5.7.2 Pyramidal Cells of Layer II 172 Release 243
5.7.3 Pyramidal Cells of Layer III 172 7.2.2 Reserve Pool of Synaptic Vesicles 244
5.7.4 Pyramidal Cells of Deep Layers 174 7.2.3 Synaptic Vesicle Docking and Priming at the Active
5.8 Pyramidal and Nonpyramidal Neurons of Zone: Role of the SNARE Complex, Munc18, and
Presubiculum and Parasubiculum 176 Munc13 244
5.9 Local Circuit Inhibitory Interneurons 176 7.2.4 Ca2-triggered Synaptic Release: Role of
5.9.1 Understanding Interneuron Diversity 179 Synaptotagmin 246
5.9.2 Dendritic Morphology 181 7.2.5 Ca2-triggered Synaptic Release: Role of Rab3A and
5.9.3 Dendritic Spines 183 RIM1 247
5.9.4 Excitatory and Inhibitory Synapses 184 7.2.6 NSF-mediated Disassembly of the SNARE
5.9.5 Axon Morphology and Synaptic Targets 185 Complex 248
5.9.6 Resting Membrane Properties 186 7.2.7 Synaptic Vesicle Endocytosis, Recycling, and
5.9.7 Voltage-Gated Channels in Inhibitory Relling 248
Interneurons 186 7.3 Glutamate Receptors: Structure, Function,
References 188 and Hippocampal Distribution 249
xiv Contents

7.3.1 Introduction: Ionotropic and Metabotropic 8.1.3 Subcellular Domain Specicity in Hippocampal
Receptors 249 Circuits 299
7.3.2 AMPA Receptors 251 8.1.4 Patterns of Local Circuit Connectivity 299
7.3.3 NMDA Receptors 253 8.1.5 Circuit Specic Receptor Distribution 299
7.3.4 Kainate Receptors 254 8.1.6 Convergence and Divergence 301
7.3.5 Metabotropic Glutamate Receptors 257 8.2 Dentate Gyrus 301
7.4 Trafficking of Glutamate Receptors and Hippocampal 8.2.1 Inputs to the Dentate Gyrus 301
Synaptic Plasticity 258 8.2.2 Granule Cell Projection to Area CA3 302
7.4.1 Synaptic Transport of AMPA Receptors in LTP 8.2.3 Granule Cell Interneuron Connections 303
and LTD 259 8.2.4 Interneuron Granule Cell Connections 303
7.4.2 NMDA Receptor-associated Cytoskeletal and 8.2.5 Granule Cell Local Excitatory Neuron
Signaling Proteins 261 Connections 305
7.4.3 Proteins Regulating Transport and Function of 8.2.6 Interneuron Interneuron Connections 305
mGluRs 263 8.3 Areas CA3 and CA1 305
7.5 Glutamate Receptor Mutant Mice: Genetic Analysis of 8.3.1 Inputs to CA3 and CA1 305
Hippocampal Function 263 8.3.2 Pyramidal Cell Interneuron Connections 306
7.5.1 Introduction: Building of Hippocampus-specic 8.3.3 Interneuron Pyramidal Cell Connections 307
Genetic Models 263 8.3.4 Pyramid Pyramid Local Connections 312
7.5.2 NMDA Receptor Mutant Mice 264 8.3.5 Interneuron Interneuron Connections 312
7.5.3 AMPA Receptor Mutant Mice 269 8.3.6 Gap Junction Connections 313
7.5.4 Kainate Receptor Mutant Mice 272 8.4 Summary 314
7.5.5 mGluR Mutant Mice 273 References 315
7.5.6 Synopsis of the Section 273
7.6 GABAergic Receptors: Structure, Function, and Chapter 9
Hippocampal Distribution 274 Structural Plasticity 321
7.6.1 Introduction: Synaptic and Extrasynaptic GABAergic Elizabeth Gould
Receptors Mediate Tonic and Phasic Inhibition in the
9.1 Overview 321
Hippocampus 274
9.2 Dendritic and Synaptic Plasticity in the Hippocampal
7.6.2 GABAA Receptors 274
Formation 321
7.6.3 GABAB Receptors 280
9.2.1 Naturally Occurring Structural Plasticity 321
7.7 Trafficking of GABA Receptors and Hippocampal
9.2.2 Hormones and Dendritic Architecture 322
Synaptic Function 282
9.2.3 Experience and Dendritic Architecture 322
7.7.1 Role of Gephyrin in GABAA Receptor Localization
9.2.4 Structural Plasticity Following Damage 323
9.2.5 Transplantation 324
7.7.2 Role of Dystrophin-associated Protein Complex in
9.3 Adult Neurogenesis 324
GABAA Receptor Function 282
9.3.1 Turnover of Dentate Gyrus Granule Cells 326
7.7.3 Plasticity of GABAA Receptor Expression at
9.3.2 Hormones and Adult Neurogenesis 326
Hippocampal Synapses 283
9.3.3 Experience and Adult Neurogenesis 328
7.8 Genetic Analysis of GABA Receptor Function in the
9.3.4 Neurogenesis Following Damage 331
Hippocampus 283
9.3.5 Unusual Features of Adult-Generated
7.8.1 GABAA Receptor Mutant Mice 283
Neurons 332
7.8.2 GABAB Receptor Mutant Mice 284
9.4 Possible Functions of New Neurons 332
7.9 Cholinergic Receptors 284
9.4.1 A Possible Role in Learning? 332
7.9.1 Introduction: Muscarinic and Nicotinic
9.4.2 A Possible Role in Endocrine Regulations? 334
Receptors 284
9.4.3 A Possible Role in the Etiology and Treatment of
7.9.2 Hippocampal Muscarinic Receptors 285
Depression? 335
7.9.3 Hippocampal Nicotinic Receptors 285
References 335
References 286
Chapter 10
Chapter 8
Synaptic Plasticity in the Hippocampus 343
Local Circuits 297 Tim Bliss, Graham Collingridge, and Richard Morris
Eberhard Buhl and Miles Whittington 10.1 Overview 343
8.1 Overview 297 10.1.1 LTP: The First Two Decades 344
8.1.1 Neuronal Classication Issues 297 10.2 Transient Activity-dependent Plasticity in
8.1.2 Input Specicity of Extrinsic Afferents 298 Hippocampal Synapses 347
Contents xv

10.2.1 Short-term Activity-dependent Changes in Synaptic 10.4.8 Membrane Spanning Molecules Contribute to
Efficacy in the Hippocampal Formation 347 Signaling Between Presynaptic and Postsynaptic
10.2.2 Single Stimuli in Hippocampal Pathways Produce Sides of the Synapse 384
Two Transient Aftereffects: Facilitation and 10.4.9 Late LTP: Persistent Potentiation Requires Gene
Depression 347 Transcription and Protein Synthesis 386
10.2.3 Post-tetanic Potentiation Is the Sum of Two 10.4.10 Structural Remodeling and Growth of Spines Can
Exponential Components: Augmentation and Be Stimulated by Induction of LTP 393
Potentiation 350 10.5 LTP at Mossy Fiber Synapses 398
10.3 NMDA Receptor-dependent Long-term Potentiation: 10.5.1 Mossy Fiber Synapses Display Striking Short-term
Properties and Determinants 350 Plasticity 398
10.3.1 Long-term Potentiation: Tetanic Stimulation Induces 10.5.2 Basic Characteristics of NMDA Receptor-
a Persistent Increase in Synaptic Efficacy 350 independent LTP at Mossy Fiber Synapses 398
10.3.2 Time Course of LTP: Rapid Onset and Variable 10.5.3 Induction Mechanisms of Mossy Fiber LTP 398
Duration 353 10.5.4 Expression of Mossy Fiber LTP Is Presynaptic 401
10.3.3 Three Distinct Temporal Components of 10.5.5 E-LTP and L-LTP at Mossy Fiber Synapses Can Be
Potentiation: STP, Early LTP, Late LTP 353 Distinguished by the Effects of Protein Synthesis
10.3.4 Input-Specicity of LTP: Potentiation Occurs Only at Inhibitors 402
Active Synapses 354 10.5.6 Summary 403
10.3.5 Associativity: Induction of LTP Is Inuenced by 10.6 LTP Can Be Modulated by Other Neurotransmitters,
Activity at Other Synapses 355 Neuromodulators, and Effectors and by Endogenous
10.3.6 Requirement for Tight Coincidence of Presynaptic and Circadian Rhythms 403
and Postsynaptic Activity Implies a Hebbian 10.6.1 Modulation by Other Neurotransmitters and
Induction Rule 356 Neuromodulators 403
10.3.7 Molecular Basis for the Hebbian Induction Rule: 10.6.2 Cyclical Inuences Modulate Induction
Voltage Dependence of the NMDA Receptor Explains of LTP 406
Cooperativity, Input Specicity, and Associativity 10.6.3 Neurogenesis and LTP 407
357 10.7 Long-term Depression and Depotentiation:
10.3.8 Spike Timing-dependent Plasticity (STDP) 358 Properties and Mechanisms 407
10.3.9 Ca2 Signaling in LTP 359 10.7.1 Overview 407
10.3.10 Metabotropic Glutamate Receptors Contribute to 10.7.2 NMDAR-dependent LTD: Properties and
Induction of NMDA Receptor-dependent LTP 360 Characteristics 408
10.3.11 Role of GABA Receptors in the Induction of 10.7.3 NMDAR-dependent LTD: Induction
NMDAR-dependent LTP 361 Mechanisms 410
10.3.12 E-S Potentiation: A Component of LTP That 10.7.4 NMDAR-dependent LTD: Expression
Reects Enhanced Coupling Between Synaptic Drive Mechanisms 412
and Cell Firing 362 10.7.5 mGluR-dependent LTD 414
10.3.13 Metaplasticity: The Magnitude and Direction of 10.7.6 Homosynaptic Depotentiation 416
Activity-dependent Changes in Synaptic Weight Are 10.7.7 Heterosynaptic LTD and Depotentiation:
Inuenced by Prior Activity 364 Activity in One Input Can Induce LTD in
10.3.14 Synaptic Scaling and Long-term Changes in Another 417
Intrinsic Excitability 368 10.7.8 LTD and Depotentiation at Mossy Fiber CA3
10.4 NMDA Receptor-dependent LTP: Expression Pyramidal Cell Synapses 419
Mechanisms 369 10.8 Synaptic Plasticity and Inhibitory Pathways 420
10.4.1 From Induction to Expression of LTP 369 10.8.1 LTP and LTD at Glutamatergic Synapses on
10.4.2 STP Is a Transient Presynaptic Form of Plasticity Interneurons 420
369 10.8.2 LTP and LTD at GABAergic Synapses 422
10.4.3 Early LTP Involves Multiple Protein Kinase- 10.9 LTP and LTD in Development and Aging and in
dependent Mechanisms 369 Animal Models of Cognitive Dysfunction 422
10.4.4 Site of Expression of Early LTP: Experimental 10.9.1 Hippocampal Synaptic Plasticity During
Approaches 374 Development 422
10.4.5 E-LTP: Presynaptic Mechanisms of Expression 376 10.9.2 Synaptic Plasticity and the Aging
10.4.6 E-LTP: Postsynaptic Mechanisms of Hippocampus 423
Expression 377 10.9.3 Animal Models of Cognitive Decline 425
10.4.7 Retrograde Signaling System Is Required for 10.10 Functional Implications of Hippocampal Synaptic
Communication Between the Postsynaptic Site of Plasticity 427
Induction and the Presynaptic Terminal 383 10.10.1 Synaptic Plasticity and Memory Hypothesis 427
xvi Contents

10.10.2 Detectability: Is Learning Associated with the 11.4.7 Olfactory Stimulation Can Elicit Hippocampal
Induction of LTP? 429 Gamma and Beta Waves 485
10.10.3 Anterograde Alteration: Do Manipulations That 11.5 Single-cell Recording in the Hippocampal Formation
Block the Induction or Expression of Synaptic Reveals Two Major Classes of Units: Principal Cells
Plasticity Impair Learning? 432 and Theta Cells 486
10.10.4 Retrograde Alteration: Does Further Induction or 11.5.1 Distinctive Spatial Cells Complex-spike Place
Reversal of LTP Cause Forgetting? 441 Cells, Head-direction Cells, and Grid Cells Are
10.10.5 Mimicry 443 Found in Various Regions of the Hippocampal
10.10.6 Synaptic Plasticity, Learning, and Memory: The Formation 487
Story So Far 443 11.6 Theta Cells 490
References 444 11.6.1 Theta Cells Fire with a Consistent Phase Relation to
EEG Theta 490
Chapter 11 11.6.2 Pharmacology of Theta Cells 490
11.6.3 Hippocampal Theta Cells Have Behavioral
Hippocampal Neurophysiology in the Behaving
Correlates Similar to Those of the Hippocampal
Animal 475
EEG 490
John OKeefe 11.7 Complex-spike Cells and Spatial Processing 491
11.1 Overview 475 11.7.1 Place Cells Signal the Animals Location in an
11.2 Hippocampal Electroencephalogram Can Be Environment 491
Classied into Distinct Patterns, with Each Providing 11.7.2 Basic Properties of Place Fields 492
Information About an Aspect of Hippocampal 11.7.3 Place Fields are Nondirectional in Unrestricted
Function 477 Open-eld Environments but Directional When
11.2.1 Hippocampal EEG Can Be Classied into Four Types Behavior is Restricted to Routes 495
of Rhythmical and Two Types of Nonrhythmical 11.7.4 What Proportion of Complex-spike Cells Are Place
Activity 477 Cells? 497
11.2.2 Each EEG Pattern Has Distinct Behavioral 11.7.5 Frame of Reference of Place Fields 498
Correlates 477 11.7.6 Place Fields Can Be Controlled by Exteroceptive
11.3 Hippocampal Theta Activity 479 Sensory Cues 499
11.3.1 Hippocampal Theta Activity: Historical 11.7.7 Idiothetic Cues Can Control Place Fields 502
Overview 479 11.7.8 Are Place Cells Inuenced by Goals, Rewards, or
11.3.2 Hippocampal Theta Activity Is Comprised of Two Punishments? 504
Components, a-Theta, and t-Theta, Which Can Be 11.7.9 Temporal Patterns of Place Cell Firing 505
Distinguished on the Basis of Behavioral Correlates 11.7.10 Place Fields in Young and Aged Animals 507
and Pharmacology 480 11.7.11 Hippocampal Place Cell Firing Is Inuenced by
11.3.3 Both Types of Theta Activity Are Dependent on the Other Areas of the Brain 509
Medial Septal/DBB but Only t-Theta Is Dependent 11.7.12 Primate Hippocampal Units also Exhibit Spatial
on the Entorhinal Cortex 480 Responses 510
11.3.4 t-Theta Occurs During Movement 11.8 Place Cells Are Memory Cells 511
Through Space 481 11.8.1 Hippocampal Place Cells Remember the Animals
11.3.5 a-Theta Occurs During Arousal and/or Attention as Location for Several Minutes During a Spatial
well as Movement 481 Working Memory Task 512
11.3.6 Theta and Sleep 481 11.8.2 Place Field Plasticity During Unidirectional
11.3.7 Theta Activity in Nonhippocampal Areas 481 Locomotion 512
11.3.8 Does the Hippocampal EEG in Monkeys and 11.8.3 Cue Control over Hippocampal Place Cells Can
Humans Have a Theta Mode? 482 Change as a Function of Experience 513
11.3.9 Functions of Theta 482 11.8.4 Control of the Angular Orientation of Place
11.4 Non-theta EEG Patterns in the Hippocampal EEG: Cells in a Symmetrical Environment Can
LIA, SIA, Ripples, Beta, and Gamma 483 Be Altered by the Animals Experience of Cue
11.4.1 Sharp Waves, Ripples, and Single Units During Large Instability 514
Irregular Activity 483 11.8.5 Complex-spike Cell Firing and Connectivity During
11.4.2 Dentate EEG Spikes During LIA 484 Sleep Is Modulated by Prior Spatial Learning
11.4.3 Pharmacology of LIA 484 Experiences 516
11.4.4 Behavioral Correlates and Functions of LIA 484 11.8.6 NMDA Receptor Confers Mnemonic Properties on
11.4.5 Small Irregular Activity 485 Place Cell Firing 516
11.4.6 Beta/Gamma Activity in the 11.8.7 Summary of Place Cell Plasticity 517
Hippocampus 485 11.9 Head Direction Cells 517
Contents xvii

11.9.1 Head Direction Cells are Controlled by Distal Chapter 12

Sensory Cues 518 Functional Role of the Human Hippocampus 549
11.9.2 Angular Distance Between any Given Pair of Head
Craig Stark
Direction Cells Always Remains Constant 519
11.9.3 Head Direction Cells Can also Be Controlled by 12.1 Overview 549
Idiothetic Cues 519 12.2 Patient H.M. 550
11.9.4 Head Direction Cells Are Found in Different 12.3 Methods for Studying Human Hippocampal
Anatomically Connected Brain Areas 520 Function 552
11.9.5 Dorsal Tegmental Nucleus of Gudden Provides 12.3.1 Behavioral Tasks and Terms 552
Information About the Direction and Angular 12.3.2 Behavioral Measures 553
Velocity of the Animals Head Rotation 521 12.3.3 Anoxia and Bilateral Hippocampal Lesions 554
11.10 Interactions Between Hippocampal Place Cells and 12.3.4 Depth Electrode Recordings 556
Head Direction Cells 521 12.3.5 Neuroimaging 557
11.11 Hippocampal Complex-spike Cells Have Been 12.3.6 Technical Challenge: Alignment of MTL Regions
Implicated in Nonspatial Perception and Across Participants 559
Learning 523 12.4 Dissociating Hippocampal Function 561
11.11.1 Hippocampal Cells Have Been Implicated in 12.4.1 Explicit Versus Implicit 561
the Processing of Nonspatial Sensory 12.4.2 Encoding Versus Retrieval 564
Information 523 12.4.3 Time-limited Role in Declarative Memory 565
11.11.2 Hippocampal Unit Activity May Show 12.4.4 Spatial Memory 567
Correlations with Different Aspects of Nonspatial 12.4.5 Associations, Recollections, Episodes,
Learning Tasks 524 or Sources 569
11.11.3 Hippocampal Unit Activity During Aversive 12.5 Conclusions 573
Classical Conditioning 524 References 575
11.11.4 Nictitating Membrane Conditioning in the Rabbit:
Chapter 13
Role of Theta 525
11.11.5 Single-unit Recording in the Hippocampus Theories of Hippocampal Function 581
During Nictitating Membrane Conditioning Richard Morris
of Rabbits 526 13.1 Overview 581
11.11.6 Hippocampal Unit Recording During Operant 13.2 Cognitive and Behavioral Neuroscience,
Tasks 528 Interventional Techniques, and the
11.11.7 Comparison of Hippocampal Cells During Hippocampus 582
Operant Conditioning and Place Tasks in 13.2.1 Value of Interventional Studies to Identify
Rats 532 Function 582
11.11.8 Hippocampal Units During Nonspatial Learning 13.2.2 Lesions, Functional Hypotheses, and
in Nonhuman Primates 533 Behavioral Tasks 583
11.11.9 Hippocampal Units During Nonspatial Learning 13.2.3 Contemporary Lesion Techniques: Pharmacological
in Humans 533 and Genetic Interventions 585
11.11.10 Conclusions 534 13.2.4 Biological Continuity of Hippocampal
11.12 Other Distinctive Cells in the Hippocampal Function 588
Formation and Related Areas 534 13.3. Declarative Memory Theory 591
11.12.1 Subicular Region Has Fewer Place Cells than 13.3.1 Outline of the Theory 591
the Hippocampus Proper, and Their Properties 13.3.2 Development of a Primate Model of
Differ 535 Amnesia 596
11.12.2 Presubiculum Contains Several Classes of 13.3.3 Domains of Preserved Learning Following Medial
Spatial Cell 536 Temporal Lobe Lesions in Primates 601
11.12.3 Parasubiculum 537 13.3.4 Selective Lesions of Distinct Components of the
11.12.4 Spatial Cells in the Entorhinal Cortex 537 Medial-temporal Lobe Reveal Heterogeneity of
11.12.5 Cells in the Perirhinal Cortex Code for the Function 602
Familiarity of Stimuli 537 13.3.5 Remote Memory, Retrograde Amnesia, and the
11.12.6 Cells in the Medial Septum Are Theta Time Course of Memory Consolidation in
Cells 538 Primates 606
11.12.7 Summary of Extrahippocampal Place Field 13.3.6 Critique 608
Properties 538 13.4 Hippocampus and Space: Cognitive Map Theory
11.13 Overall Conclusions 538 of Hippocampal Function 617
References 540 13.4.1 Outline of the Theory 618
xviii Contents

13.4.2 Representing Spatial Information, Locale Processing, 14.5.5 Consolidation and Cross-modal Binding of
and the Hippocampal Formation 623 Events in Memory 738
13.4.3 Using Spatial Information: Spatial Navigation and 14.5.6 Hippocampal Contributions to Various Types
the Hippocampal Formation 627 of Memory and Retrieval 739
13.4.4 Comparative Studies of Spatial Memory and the 14.6 Reconciling the Hippocampal Roles in Memory
Distinction Between Spatial and Associative and Space 740
Learning 640 14.7 Conclusions 744
13.4.5 Storage and Consolidation of Spatial References 744
Memory 650
13.4.6 Critique 654 Chapter 15
13.5 Predictable Ambiguity: Congural, Relational, Stress and the Hippocampus 751
and Contextual Theories of Hippocampal Richard Morris
Function 656 15.1 Overview 751
13.5.1 Congural Association Theory 658 15.2 Glucocorticoid Receptors and Hippocampal
13.5.2 Relational Processing Theory: Renement of the Function 753
Declarative Memory Theory 662 15.2.1 Glucocorticoid Receptors Are Present in the Animal
13.5.3 Contextual Encoding and Retrieval 668 and Human Hippocampus 753
13.5.4. Critique 676 15.2.2 There is an Inverted U-Shape Function Between
13.6 Episodic Memory, Hippocampus, and Neurobiology Level of Stress and Memory 754
of Rapid Context-specic Memory 677 15.2.3 Stress Modulates Intrinsic Hippocampal Excitability
13.6.1 Concept of Episodic Memory 677 and Activity-dependent Synaptic Plasticity Associated
13.6.2 Scene Memory as a Basis for Episodic Memory and with Learning and Memory 756
Top-down Control by the Prefrontal Cortex 679 15.3 Stress and Hippocampal Structure 759
13.6.3 What, Where, and When: Studies of Food-caching 15.3.1 Chronic Exposure to High Levels of Stress or Stress
and Sequence Learning 682 Hormones Is Associated with Structural Changes in
13.6.4 Problem of Awareness 686 the Hippocampus 759
13.6.5 Elements of a Neurobiological Theory of the Role of 15.3.2 Stress or Stress Hormones Can Impair Neurogenesis
the Hippocampus in Episodic-like Memory 687 in the Hippocampus 759
References 694 15.3.3 Fetal Programming of GC Regulation 760
15.4 Other Higher Brain Structures Implicated in Stress
Chapter 14
and Their Interaction with the Hippocampus 761
Computational Models of the Spatial and Mnemonic 15.5 How the Hippocampus Orchestrates Behavioral
Functions of the Hippocampus 715 Responses to Arousing Aversive Experiences 762
Neil Burgess References 764
14.1 Overview 715
14.2 Introduction 715 Chapter 16
14.3 Hippocampus and Spatial Representation 716 Hippocampus and Human Disease 769
14.3.1 Representing Spatial Location and Orientation: Matthew Walker, Dennis Chan, and Maria Thom
Data 716 16.1 Overview 769
14.3.2 Representing Spatial Location: Feedforward 16.2 Mesial Temporal Lobe Epilepsy and Hippocampal
Models 719 Sclerosis 770
14.3.3 Representing Spatial Location and Orientation: 16.2.1 Introduction 770
Feedback Models 721 16.2.2 Clinical Features 771
14.3.4 Modeling Phase Coding in Place Cells 727 16.2.3 Etiology 774
14.4 Hippocampus and Spatial Navigation 729 16.2.4 Pathophysiology 777
14.4.1 Spatial Navigation: Data 729 16.2.5 Conclusion 789
14.4.2 Spatial Navigation: Feedforward Models 729 16.3 Alzheimers Disease 789
14.4.3 Spatial Navigation: Feedback Models 732 16.3.1 Introduction 789
14.5 Hippocampus and Associative or Episodic 16.3.2 Clinical Features 789
Memory 733 16.3.3 Genetics 795
14.5.1 Hippocampus and Memory: Data 733 16.3.4 Pathophysiology 796
14.5.2 Marrs Hippocampo-neocortical Model of 16.3.5 Treatment Options 801
Long-term Memory 734 References 803
14.5.3 Associative Memory and the Hippocampus 734
Index 813
14.5.4 Hippocampal Representation, Context,
and Novelty 737


David Amaral, PhD Michael Frotscher, Dr Med

The M.I.N.D. Institute Institute of Anatomy
Department of Psychiatry and Behavioral Sciences University of Freiburg
University of California-Davis Freiburg, Germany
Davis, California, United States
Elizabeth Gould, PhD
Per Andersen, MD PhD Department of Psychology
Department of Neurophysiology Princeton University
University of Oslo Princeton, New Jersey, United States
Oslo, Norway
Dimitri Kullmann, MBBS DPhil FRCP
Tim Bliss, PhD Institute of Neurology
Division of Neurophysiology University College London
MRC National Institute for Medical Research, London London, United Kingdom
London, United Kingdom
Pierre Lavenex, PhD
Eberhard Buhl, Dr Med* Institute of Physiology
School of Biomedical Sciences Department of Medicine
Leeds University University of Fribourg
Leeds, United Kingdom Fribourg, Switzerland

Neil Burgess, PhD Chris McBain, PhD

Institute of Cognitive Neuroscience Laboratory of Cellular and Synaptic Neurophysiology
University College London National Institute of Child Health and Human Development
London, United Kingdom Bethesda, Maryland, United States

Dennis Chan, PhD MRCP Richard Morris, DPhil

Dementia Research Group Laboratory for Cognitive Neuroscience
Institute of Neurology College of Medicine and Veterinary Medicine
University College London University of Edinburgh
London, United Kingdom Edinburgh, Scotland

Graham Collingridge, PhD John OKeefe, PhD

MRC Centre for Synaptic Plasticity Department of Anatomy & Developmental Biology
University of Bristol University College London
Bristol, United Kingdom London, United Kingdom

xx Contributors

Pavel Osten, PhD Maria Thom, MRCPath

Department of Molecular Neurobiology Department of Clinical and Experimental Epilepsy
Max-Planck-Institute for Medical Research Institute of Neurology
Heidelberg, Germany University College London
London, United Kingdom
Lszl Seress, MD PhD
Central Electron Microscopic Laboratory, Faculty of Medicine Matthew Walker, PhD MRCP
University of Pcs Department of Clinical and Experimental Epilepsy
Pcs, Hungary Institute of Neurology
University College London
Rolf Sprengel, PhD
London, United Kingdom
Department of Molecular Neurobiology
Max-Planck-Institute for Medical Research Miles Whittington, PhD
Heidelberg, Germany Department of Neuroscience
Newcastle University
Nelson Spruston, PhD
Newcastle upon Tyne, United Kingdom
Department of Neurobiology and Physiology
Institute for Neuroscience William Wisden, PhD
Northwestern University Department of Neuroscience
Evanston, Illinois, United States University of Aberdeen
Aberdeen, Scotland
Craig Stark, PhD
Department of Psychology
Johns Hopkins University
Baltimore, Maryland, United States
The Hippocampus Book
This page intentionally left blank
1 Per Andersen, Richard Morris, David Amaral, Tim Bliss, and John OKeefe

The Hippocampal Formation

rons in the brain. As a result, we now know a great deal about

1.1 Overview their development, synaptogenesis, neurotransmitter recep-
tors and ion channels, micro-circuitry, and cell biological
Buried deep within the medial temporal lobe of the human machinery. No less impressive has been the molecular analysis
brain lies a group of many millions of neurons organized into of neurotransmission in hippocampal cells; what has been
a network quite different from that found anywhere else in the learned from this has revealed general properties that seem to
nervous system. It is a structure whose bulb-like shape, pro- apply in other areas of the central nervous system. We also
truding into the lateral ventricles, has captivated anatomists know something about why and when these cells are activated
since the rst dissections took place in classical Egypt. The in the living brain. Recordings from freely moving animals as
hippocampal formation is a group of brain areas consisting of they navigate around a space with which they have become
the dentate gyrus, hippocampus, subiculum, presubiculum, familiar have shown that individual hippocampal pyramidal
parasubiculum, and entorhinal cortex. The basic layout of cells re in particular locations. This nding led to the devel-
cells and ber pathways of the hippocampal formation is opment of new behavioral tools to study the neural mecha-
much the same in all mammals (Fig. 11). nisms of memory in animals. Along the extensive dendritic
There are several reasons the hippocampus has attracted arborizations of these pyramidal cells, there can be many
the interest of scientists in the many disciplines that now thousands of miniature dendritic spines. These are the sites
characterize modern neurosciencethe hippocampus has where most of the excitatory synapses are to be found. An
something for everyone. Whether you are a psychologist important nding is that the efficiency with which these
interested in memory, a synaptic physiologist investigating excitatory synapses transmit messages can vary as a function
neuronal and synaptic plasticity, or a computational neurosci- of neural activity. The circumstances bringing about this
entist wanting to build a neural network model, the hip- synaptic plasticity are now well understood, and we are begin-
pocampus and its associated structures are an attractive set of ning to understand some of the underlying biochemical
brain structures on which to work. In parallel, clinicians con- mechanisms. Many suspect that synaptic plasticity is a key
cerned with the basis of neurological conditions such as mechanism of memory. Another exciting feature of the hip-
epilepsy or Alzheimers disease had their attention drawn to pocampal formation is that the dentate granule cell is one of
the hippocampal formation because of the pathological the rare types of nerve cell that multiplies throughout life.
processes observed to occur there and the opportunities that Adult neurogenesis is likely to yield important lessons about
scientic study of this area of the brain offers for novel thera- neuronal growth and survival. This is an area of obvious
peutics. The hippocampus has been a neural Rosetta Stone. potential importance for neuronal repair and therapeutic
The striking discovery, 50 years ago, that patient H.M. had a intervention.
relatively pure memory decit after surgical excision of the These observations have emerged from the widespread fas-
medial temporal lobe for the relief of epilepsy had a profound cination with this beautiful cortical structure that so attracted
effect on the study of memory and, through that, on our the early anatomists. As we shall see, certain structural princi-
understanding of the functions of the hippocampus itself. ples that are characteristic of the hippocampal formation have
During the last 30 years, the pyramid-shaped cells of the greatly facilitated structural, physiological, and behavioral
hippocampus have become the most intensively studied neu- analyses.

4 The Hippocampus Book

Figure 11. Nissl-stained coronal sections

through the rat, monkey, and human hippocam-
pal formation. Although there are obvious dif-
ferences in certain regions, the most striking
feature is the general similarity of this brain
region across phylogeny.

only some of them. There is also general agreement that some

1.2 Why Study the Hippocampal Formation types of learning and memory are propositionalthat is, they
on Its Own? relate to memories that can be described as a series of propo-
sitional statementswhereas other types, such as the process
Can one protably focus on one group of brain structures and of acquiring motor skills, are not. The hippocampus is
appear to ignore the rest of the brain? The brain is so heavily engaged in remembering information that can be described in
interconnected that it is misleading to think about the func- a propositional or declarative manner. That much is clear.
tion of one small part of it in isolation. What it does and how Beyond that, there remain many areas of disagreement and
it does it might be understood better by comparing it with debate that continue to fuel imaginative new research in
other brain areas and by thinking about how it works in con- behavioral and cognitive neuroscience. A striking feature of
junction with them to realize the seamless fabric of normal this strand of research has been the constructive convergence
brain function. We have much sympathy with this view, not of both top-down and bottom-up approaches to the study
least because the general signicance of the many exciting dis- of brain function. Work on the function of the hippocampal
coveries that have been made about the hippocampus can formation, arguably more than any other brain area, has led to
only become clear with a more inclusive approach in which it the development of current concepts about the organization
and other parts of the brain are considered together. We of memory systems in the brain.
believe, nonetheless, that there is still value in our focused Some traditionalists still hold that the various subjects that
approach. There are two primary reasons for singling out the make up the biomedical sciencessuch as psychology, physi-
hippocampal formation as a brain area worthy of a book in its ology, biochemistryeach has a self-contained level of analy-
own right. They have to do with trying to understand its func- sis that is logically independent of other levels of analysis.
tion(s) within the brain and the way in which work on hip- With this view, a psychological problem requires an answer in
pocampal tissue has revealed general principles in strictly psychological terms and a physiological problem a
neuroscience. physiological approach. The emerging view at the start of the
First, contemporary research points strongly to the hip- twenty-rst century, however, is that many years of a strictly
pocampus having a highly specic role in memory. There has psychological analysis of learning and memorybeginning
been a great deal of debate about how that role should be with the discovery of Pavlovian conditioning, the ensuing
characterized; and after the explosive period of research dur- dark phase of behaviorism, and the subsequent enlightenment
ing the last quarter century, a consensus is beginning to of the cognitive revolutionfailed to unravel the existence of
emerge. It is now generally recognized that there are multiple multiple memory systems. It was not until a different brain-
types of memory and that the hippocampus plays a part in systems approach was taken during the last part of the twen-
The Hippocampal Formation 5

tieth century, an approach that emerged from studies of the other experiments tractable in a manner that remains difficult
memory problems faced by patient H.M. and other amnesic in other brain areas and would be wholly impracticable in
patients, that the major breakthroughs in our understanding vivo.
of the cerebral organization of memory came about. Such Beyond its key role in the development of in vitro tech-
thinking led to research that married brain and behavior in niques, the hippocampus has also been the area of choice
single-unit recording and lesion studies in awake primates for many other types of brain research. It is involved in a num-
and freely moving rodents, as well as the development of new ber of disparate neurological disorders, including epilepsy,
behavioral tests of learning and memory. There is a lesson Alzheimers disease, and cerebrovascular disease. Thus, the
here that we believe is generally applicable to other branches abnormal electrical activity that is at the root of seizures in
of functional neuroscience. Specically, the brain-systems epileptic patients is often easily detected in the hippocampus.
approach is one in which there is analytical value in juxtapos- Moreover, a hallmark feature of the neuropathology of tem-
ing the biomedical sciences with psychology and, in particu- poral lobe epilepsy is loss of neurons in several hippocampal
lar, in thinking carefully about what bits of the brain do what. elds. For example, the characteristic pathological changes of
The second justication for writing this book is that the Alzheimers disease manifest initially in the entorhinal cor-
hippocampus has preeminently been the structure in which texone of the components of the hippocampal formation
many of the general principles of modern neuroscience have and the disease spreads from there to involve the hippocampus
been studied and established. Thirty years ago, the most proper and ultimately the entire cerebral cortex. Such ndings
widely studied cell in the nervous system was the alpha have led to the development of model systems in which patho-
motoneuron of the ventral horn of the spinal cord. Today it is physiological events such as these may be studied and, hope-
the pyramidal cell of the hippocampus. One reason for this fully, alleviated by treatment.
development has been the peculiar anatomy of the hippocam- These two main reasons for writing the book therefore
pus, with all principal cells in a single layer and synaptic provide us with two intersecting themes that run through
inputs to well dened dendritic lamina. This simplied archi- many of its chapters. One perspective is functional, the other
tecture facilitated the recording of both synaptic signals and heuristic.
population discharges. Field potential recording became pos-
What does the hippocampus do?
sible because the well dened somatic and dendritic laminae
What can we learn about general principles of neuro-
allowed identication of current sources and sinks in extra-
science from studying the hippocampus?
cellular recordings made in vivo. It was through these studies
that the basic principle of unidirectional excitatory transmis- To achieve these twin aims, we have broken up our task
sion was rst described and the phenomenon of long-term into chapters that are largely organized along conventional
potentiation was discovered. Of particular importance for the disciplinary lines but where, when appropriate, both strands
study of synaptic function in the context of learning and of thought are intertwined.
memory was the fact that eld-potential recording of synaptic
responses could be made as easily in the freely moving animal
as in the hippocampal slice. The pyramidal cell also became a
popular cell to study because of the analytical potential of the 1.3 Dening the Contemporary Era
in vitro brain slice. The rst brain slices were made from neo-
cortex and piriform cortex with limited synaptic activation. Our starting point when preparing this book was to dene the
With its better identication of cells and input bers, the contemporary era of research as beginning in the 1960s and to
development of the transverse hippocampal slice revolution- regard discoveries made before that as modern and those
ized neurophysiology and neuropharmacology. Of course, made afterward as contemporary. Any division of this kind
many fundamental concepts had been worked out before- is arbitrary, but the period of the 1960s and 1970s was a water-
handin the axon of the giant squid, spinal cord, and cere- shed for both functional and mechanistic studies of the hip-
bellumbut the hippocampal slice rendered analytical pocampus. That decade saw the rst intracellular analysis
extra- and intracellular studies of mammalian cells and iden- of synaptic, antidromic, and epileptiform activities of hip-
tied synapses feasible on an unprecedented scale. Work in the pocampal neurons; the characterization and interpretation of
hippocampus has been a major contributor to our under- eld potentials signaling excitation and inhibition; and elec-
standing of the actions and mechanisms of various types of tron micrographs of synapse types and their distribution
synapse and the various classes of receptors for excitatory and along the cell bodies and their extensions. It was also at this
inhibitory amino acids, the many transmitter uptake mecha- time that several of the new tract-tracing techniques became
nisms, activity-dependent synaptic plasticity, and the deleteri- available to anatomists, replacing the classic degeneration
ous consequences of excitotoxicity for brain cells. It is the techniques hitherto used to identify regional connectivity.
two- or three-layered architecture of the hippocampus and its These advances made it possible to map the extrinsic connec-
capacity to survive in vitro for long periods of time coupled tions of the hippocampal formation at a previously unachiev-
with a strict layering of synapses in the dendritic tree that has able level of sensitivity and detail. During the same decade,
rendered the design and analysis of electrophysiological and hippocampal place cells were discovered and the phenomenon
6 The Hippocampus Book

of long-term potentiation was rst described in detail. Chapter 5 takes us forward to a detailed analysis of indi-
Methodological developments of the 1970s included the hip- vidual cells in which we present, side by side, both anatomical
pocampal slice, new tests of recognition memory for primates, and physiological ideas. To understand the principal cell types
and the open-eld watermaze, a behavioral technique to study of the hippocampal formationpyramidal and granule
learning in rodents that has since been used extensively to cellsand the several types of interneuron, one must know
analyze the role of the hippocampus in spatial navigation. what these cells look like, know about their inputs and out-
Mathematicians also started thinking about the hippocampus puts, and grapple with the biophysical principles that govern
as a neural network and so set in motion a theoretical neuro- how current injected into the cell at a dendritic synapse con-
science that has attracted increasing attention. tributes to cell ring. It is necessary to understand how
synapses at various dendritic locations differ in their synaptic
effects and how these individual effects sum to control the cell
discharge pattern. The large group of after-hyperpolarizing
1.4 Organization and Content of the Book responses is essential for cellular behavior. The important
modulatory effects of a set of controlling systems comprise a
During the discussions about when the contemporary era major topic.
actually started, we recognized the enormous contributions The intricacies of neurotransmission are considered in
made beforehand and upon which so much of modern re- Chapters 6 and 7 from physiological, pharmacological, and
search rests. Accordingly, we have devoted Chapter 2 to a his- molecular biological perspectives. Chapter 6 considers how
torical discussion of the key discoveries and concepts of the release of glutamate from presynaptic terminals in the
hippocampal neurobiology of the modern era up to about the hippocampus and its action on various glutamatergic recep-
1970s. In it, we also identify some of the key gures whose tors mediate excitatory synaptic transmission. It also discusses
work helped to usher in the contemporary era. In addition, we the essential role of inhibitory synaptic transmission at
highlight key concepts in neurobiology that were rst estab- GABAergic synapses. Because the hippocampus is one of the
lished through work on the hippocampal formation. few regions where it is possible to have experimental control
In Chapter 3, we outline the anatomy of the hippocampal over a large set of converging inputs to an individual cortical
formation, beginning with issues of nomenclature and def- cell, most of our ideas on cellular integration are derived from
inition. Here we present the three-dimensional organization studies on hippocampal tissue. This chapter describes the
of the rodent, monkey, and human hippocampal formation, intricate molecular machinery responsible for the mobiliza-
the detailed architecture of each of its component structures, tion of glutamate-containing vesicles, the exocytotic release
and the patterns of cellular interconnectivity. Straightaway, we machinery for glutamate, and its postsynaptic actions. It also
found that we disagreed about certain conceptual issues, such outlines the various ligand-binding and modulatory sites on
as the status of the lamella hypothesis that was proposed these receptors, the insights coming from the relatively new
nearly 30 years ago on the basis of electrophysiological data. work on receptor subunits afforded by molecular biology and
Modern neuroanatomical tract-tracing studies are thought the mechanisms responsible for transmitter uptake. In many
not to support this concept in its original form, although of these studies, genetically altered animals have been essential
the electrophysiological data remain on the table. After long tools. The chapter touches on the important and developing
debate, we arrived at a description of the extent to which con- topic of neuromodulatory transmitters in the hippocampus,
nectivity is primarily in the transverse plane and the extent to such as acetylcholine, norepinephrine, 5-hydroxytryptamine,
which longitudinal and other more diffuse projections are a dopamine, and peptides such as dynorphin and somatostatin.
characteristic feature of hippocampal anatomy. Similarly, The transmitters inuence hippocampal excitability and
although recognizing that the original concept of the tri- second-messenger effector mechanisms in an orchestrated
synaptic circuit may be too narrow, we continue to stress the cascade of formidable complexity. Chapter 6 also discusses
sense in which unidirectional connectivity is a distinctive fea- how the many neuromodulatory systems inuence the tradi-
ture of the hippocampus compared to the neocortex, includ- tional synaptic processes and the mechanisms by which they
ing the parallel connections made from the entorhinal cortex are mediated. These studies often are models for the general
to the dentate gyrus, and to areas CA3 and CA1 of the hip- neuronal activity seen in other parts of the nervous system.
pocampus proper. Chapter 7 discusses the molecular biology of hippocampal
Chapter 4 considers how this intricate structure develops cells with particular emphasis on synaptic function. Here,
both from the perspective of overall organization and from much of the pioneering work on glutamate receptor catego-
that of individual cells. Here the interplay moves from pure rization was made in the spinal cord, whereas the important
anatomy to genes and molecular biology. Modern studies of discovery of the voltage-dependent magnesium block of the
the development of the hippocampus are characterized by N-methyl-D-aspartate (NMDA) receptor channel was discov-
molecular studies in which the temporal expression of the ered in hippocampal tissue. Notably, the hippocampal slice
genes responsible for migration, neuronal specication, and has been of critical importance for developing the modern
axonal guidanceand for major cellular constituents such as understanding of ionotropic and metabotropic glutamate
receptor subunitsare being mapped out. receptors.
The Hippocampal Formation 7

Having described the three-dimensional organization of and power. However, the extent to which bidirectional
the hippocampus, its cells, synapses, and their transmitters changes can be seen in the hippocampus in freely moving
and receptors, we need next to consider the local circuits that adult animals remains unclear. Chapter 10 also takes on the
these structures produce and the possible types of neuronal task of explaining the various behavioral studies that have
processing they make possible. Chapter 8 takes on this task attempted to identify the role LTP might play in learning,
and endeavors to explain how an impressive multitude of memory, or other cognitive processes. We outline studies that
inhibitory interneurons shape the number and pattern of have examined correlations between the physiological proper-
the participating cells in a near-physiological situation. These ties of synaptic plasticity and behavioral learning and the
interactions between principal cells and interneurons are par- effects of blocking it, saturating it, and erasing it. We offer no
ticularly evident during periods of rhythmical theta and denitive answer to the question of the purpose of these vari-
gamma activity. ous types of plasticity, but the tenor of our account is sympa-
The chapter on local circuits also provides an introduction thetic to a role in the encoding and storage of certain kinds of
to Chapters 9 and 10, where we confront activity-dependent memory.
neuronal and synaptic plasticity. One of the more extraordi- Chapter 11 continues the behavioral theme with a survey
nary discoveries of recent times has been that new neurons of studies in which electrical activity in structures of the hip-
can be formed in the adult brain. This is contrary to estab- pocampal formation has been recorded during various kinds
lished dogma but, interestingly, the dogma that had been chal- of behavior. The pioneering work of this kind was conducted
lenged in hippocampal studies nearly 30 years ago but was in rats and was heralded by the discovery of place cells noted
unappreciated at the time. The dentate gyrus is one of two earlier (neurons that re when rats occupy particular posi-
major sites of adult neurogenesis. In Chapter 9, we discuss the tions in space). The further discovery of a different group of
extent to which new cells and connections can be made, a eld cells that red during certain kinds of movement and in phase
of acute interest for basic and practical neurobiology. Chapter with slow-wave electroencephalographic activity was also
10 focuses on processes to explain the fact that, once formed, important, as were ndings of cells that code for the direction
hippocampal circuits are not immutable, as synapses through- in which an animals head is pointing or show grid-like pat-
out the various components of the hippocampal formation terns as an animal traverses a familiar space. These categories
show both short-term and long-term changes. The latter of cell are found in distinct areas of the hippocampal forma-
includes the intensively studied phenomenon of long-term tion and in different cell classes within a single area. Work in
potentiation (LTP). Research on this form of synaptic plastic- nonhuman primates has also identied a class of cells that are
ity has revealed a range of properties that are highly suggestive responsive to the animals view. The properties of these cells,
of a potential cellular memory mechanism. The effort to their role in mapping space, and, according to some, other
understand LTP mechanistically led to passionate debates that cognitive functions as well are gradually being unraveled.
have stimulated a host of ingenious experiments. Research on The process of discovery has been accompanied by the
synaptic plasticity, perhaps more than any other area of hip- development of a new single-cell methodology that permits
pocampal neurobiology, spans the twin themes of the book. simultaneous recording from large numbers of neurons. This
What is the purpose of LTP? Are its underlying neural mech- tetrode-recording technique enables the properties of several
anisms applicable to other areas of cortex? Research on the individual neurons in a population to be mapped simultane-
induction of this phenomenon has helped unravel one of the ously and has applications that extend well beyond hip-
most satisfyingly elegant biophysical mechanisms in the nerv- pocampal neurobiology.
ous system: the role of the NMDA receptor as a coincidence Chapter 12 begins with the discovery, almost 50 years ago,
detector. Having detected the conjunction of presynaptic that surgical damage to the hippocampal formation and
activity and postsynaptic depolarization, this receptor signals related structures on both sides of the brain causes a
the conjunction by an inux of the divalent cation calcium profound, lasting global amnesia. Once this unexpected
(Ca2), which then sets in train a cascade of biochemical consequence had been observed, neurosurgeons took care
effects. The discussion of the NMDA receptor complements not to perform bilateral medial temporal lobectomies. Such
the earlier description of its molecular basis in Chapter 7. It memory-impaired patients are therefore rare, but their very
activates signaling proteins associated with the receptor itself existence spurred the development of animal models through
and calcium-dependent kinases. It may also help trigger the which, it is hoped, the role of the hippocampus in memory
release of calcium from intracellular stores, thereby sustaining can be worked out. Modern functional imaging techniques are
the synaptically evoked Ca2 transient beyond the period that also providing new insights into the differential activation of
the NMDA receptor-associated ion channel remains open. various brain areas for different types of memory and differ-
The resulting biochemical cascade eventually results in an ent stages of memory processing. Such work has led to new
increase in synaptic strength. It should also be recognized concepts, such as the extent to which neuronal activation in a
that activity-dependent synaptic plasticity is not restricted to memory-related area reects the effort devoted to encoding or
increases in synaptic efficacydecreases can also be induced. retrieval or to the success of achieving either. Such conceptual
This bidirectional feature of synaptic plasticitythat what dissociations are almost impossible to draw on the basis of
goes up must come downadds computational exibility lesion studies alone.
8 The Hippocampus Book

Chapter 13 takes up the theme set by these human studies and patches that will eventually enable a comprehensive neu-
to lay out a number of prominent theories of hippocampal ral network account of hippocampal function to be woven
function that have been developed from work on animals. together are beginning to become apparent.
One theory concerns the role of the hippocampus in remem- Last come Chapters 15 and 16. The rst of these chapters
bering facts and events that can be consciously recalled, and picks up from a set of intriguing observations suggesting that
another has to do with its role in mapping and navigating the hippocampus, together with the amygdala and prefrontal
through space. Other theories have been developed to help cortex, play a role in regulating the hypothalamic-pituitary-
address some of the issues and problems that have arisen in adrenal axis, which is responsible for the release of stress hor-
connection with these two ideas, particularly in situations of mones. The hippocampus contains a particularly high density
predictable ambiguity. We also outline current interest in the of corticosteroid receptors of two major types, and this neu-
idea that the hippocampus plays a critical role in contextual roendocrine detection of mild or severe stress has a dramatic
encoding and retrieval, including episodic memory. This impact on hippocampal physiology and memory function,
section attempts to link behavioral studies on rapidly learned including effects on LTP. Glucocorticoid receptors act as tran-
forms of associative memory to the neurobiological observa- scription factors, and it has long been thought that the pri-
tions on circuitry and synaptic plasticity outlined earlier mary expression mechanism was an intracellular regulation of
in the book. Although one of the authors of this book is genes. However, it is now becoming clear that there are a vari-
an architect of the second theory discussed, we attempt to ety of more rapid effects of corticosteroid action in the hip-
present as objective an account of the state of the eld as pocampus. Chapter 16 considers hippocampal pathologies.
possible. This is, of course, a large subject in its own right, and we can-
Chapter 14 takes us to computational models of hip- not here do justice to a vast eld of clinical research. To
pocampal function. Some models focus on its global function, exclude it, however, would be wrong because all of us have an
but even when they work they often involve assumptions that interest in alleviating the impact of human neurological dis-
seem far removed from real neurobiology. These models are ease. The importance of these conditions is underscored by
the articial intelligence and idealized neural network models the prevalence of disease: Alzheimers disease alone affects
that can solve interesting cognitive problemsincluding nav- around 20 million people worldwide. Substantial advances
igation around the world or remembering associations in have been made in understanding the involvement of vari-
memorybut using processing units and learning algorithms ous components of the hippocampal formation in epilepsy
that may not exist among real hippocampal cells. Other mod- and Alzheimers disease, on which we focus; and this infor-
els incorporate more realistic assumptions about the hip- mation has underpinned improvements in diagnosis and
pocampal network and its neuronal components, the different treatment.
classes of cells, and their phase relationships; but their pro- This book is our tribute to the thousands of scientists who
cessing tends to be more limited. Ultimately, the value of these have contributed to our understanding of this beautiful and
models is to provide a predictive framework for understand- enigmatic part of the brain. We hope that it will help design
ing the mass of detail we have covered in other chapters of new, revealing experiments and thereby increase our insight
this book. To date, no model incorporates information realis- into its cellular, molecular, physiological, and behavioral
tically at each of the levels of the anatomical circuit, cellular processes. Should it also facilitate clinical studies, our satisfac-
architecture, and synaptic transmission. However, the pieces tion will be so much the greater.
2 Per Andersen, Richard Morris, David Amaral, Tim Bliss, and John OKeefe

Historical Perspective: Proposed Functions,

Biological Characteristics, and Neurobiological
Models of the Hippocampus

many functions, ranging from olfaction to motor function

2.1 Overview and even reason, stubbornness, and inventiveness. The
Bolognese anatomist Giulio Cesare Aranzi (circa 1564) was
This chapter has two major sections. In the rst, we provide a the rst to coin the name hippocampus, undoubtedly
selective overview of some of the historically important pro- because of its similarity to the tropical sh (Fig. 21).
posals concerning the function of the hippocampus. The cur- After the advent of microscopy, the hippocampus was per-
rent view, that the hippocampus plays a prominent role in haps even more impressive, with its characteristic, neatly reg-
memory, is dealt with in detail in Chapters 12 and 13. In the imented cellular arrangement. Although the hippocampal
second major section, we highlight the historically signicant formation is organized in a very different way from other cor-
advances in neurobiology that have been made by using the tical areas, it has still played an important role in unraveling
unique structure of the hippocampus for studies ranging some of the basic principles of cortical organization.
from dening the morphology of excitatory and inhibitory Its stunning structure, with cell populations apparently
synapses to developing computational models of memory. condensed into single layers, has attracted the attention of
This second section is itself divided into subsections that many investigators of the central nervous system. A striking
emphasize rst the biological characteristics of the hippocam- example is Camillo Golgis picture from 1886, where he used
pus and then the advantages of the hippocampus as a model his revolutionary new technique to illustrate the unique
system for neurobiological research. Many of the most recent organization of the hippocampus, here reproduced from
advances are dealt with in appropriate chapters, as we intend Golgi et al. (2001) (Fig. 22).
this chapter to be primarily a historical overview. As noted in Chapter 1, the two themes of this bookthe
overall functions of the hippocampus and how its study has
shed light on a range of general principles in neuroscience
can be highlighted by reference to the history of research on
the hippocampus. Early work inevitably focused on anatomi-
2.2 The Dawn of Hippocampal Studies cal issues, whereas later research successively brought in phys-
iological, behavioral, and other technologies in an effort to
From the very start of brain investigations, the hippocampus fathom the function of the hippocampus and through it the
has played a central role. No doubt this is due to the striking workings of the brain.
appearance of the large, bulging structure impressing itself
into the lateral ventricle and visible even to the ancient 2.2.1 A Famous Dispute About the Signicance
anatomists. Members of the Alexandrian school of medicine of the Hippocampus
were impressed by the elegant, curved structure, which when
seen with its contralateral half strongly resembles the coiled Interest in the hippocampus was particularly high during the
horns of a ram. Hence, the ancient scholars named the hip- eighteenth and nineteenth centuries, a notable example being
pocampus cornu ammonis, Latin for horn of the ram. This its curious role in the famous debate between Richard Owen
terminology survives in the acronyms for the hippocampal and Thomas Huxley at the British Association in Oxford in
subelds CA1 (cornu ammonis), CA2, and CA3. Over the cen- June 1860 following the publication of Charles Darwins book
turies, this part of the brain has been proposed as the seat of On the Origin of Species in 1859 (Gross, 1993). The superin-

10 The Hippocampus Book

Figure 21. Human hippocampus dissected free (left) and com- Figure 23. Hippocampus major and minor as seen in a human
pared to a specimen of Hippocampus leria (right). (Source: Courtesy brain after the dorsal cortex and corpus callosum have been
of Professor Laszlo Seress, University of Pecs.) removed.

tendent of the natural history collections of the British objected strongly, not least because of his own dissection
Museum, Richard Owen, claimed that only human brains results. Today, it is difficult to see how this particular bulge
contain a structure he called hippocampus minor, this being could have any bearing on the question of human evolution.
one essential distinguishing feature between Man and Apes. The name hippocampus minor is really a misnomer because
This structure is a relatively small bulge on the medial wall of it has nothing to do with the hippocampal formation per se
the posterior horn of the lateral ventricle (Fig. 23). Huxley but is a fold of white matter around the calcarine ssure, in
other words a visual cortical structure.
Figure 22. Section of a rabbit hippocampus stained with the origi-
nal Golgi method (1886). (Source: Golgi et al., 2001).

2.3 Early Ideas About Hippocampal Function

2.3.1 The Hippocampal Formation

and Olfactory Function

Although the hippocampus had been casually implicated in a

number of functions during the nineteenth and early twenti-
eth centuries, until the 1930s the hippocampal formation was
considered by neuroanatomists to be part of the olfactory sys-
tem. Perhaps this was because macroscopic investigations left
the impression that the hippocampus is especially large in
macrosmatic animals (with large olfactory mucosa and bulbs)
such as nocturnal insectivores and rodents. However, in these
macrosmatic animals, the absolute sizes of the olfactory parts
of the brain are quite small. In fact, it was claimed that the
ratio of the volume of the hippocampus to the olfactory bulb
is largest in humans (Rose, 1935). Indeed, as observed by
Brodal (1947):

From a functional point of view it is worth emphasiz-

ing that the development not only of the fornix, but
also of the mammillary body, the mammillo-thalamic
tract, the anterior (more particularly the antero-
ventral) nucleus of the thalamus and the cingular
gyrus runs roughly parallel with the degree of develop-
Historical Perspective 11

ment of the hippocampus. All these structures reach pocampal formationhas subsequently proven incorrect.
their peak of development in man. More sensitive anatomical techniques have shown that in the
rodent the entorhinal cortex receives a massive direct projec-
A possible basis for the idea of an olfactory function for the tion from the olfactory bulb as well as secondary olfactory
hippocampus (an idea that was repeated in numerous text- inputs from the piriform and periamygdaloid cortices
books during the rst half of the twentieth century) is the (Shipley and Adamek, 1984). Even in the monkey, the anterior
appearance of a few early behavioral and clinical observations. portion of the entorhinal cortex is directly innervated by the
For example, David Ferrier (1876) observed movements of lateral olfactory tract (Amaral et al., 1987). Thus, the olfactory
the lip and nostrils on stimulation of the hippocampal lobe in system retains a privileged position in relation to the entorhi-
monkeys. John Hughlings Jackson and Charles Beevor (1890) nal cortex as none of the other sensory channels from which
reported a patient who had subjective olfactory sensations it receives information originates in primary or even higher
during seizures that originated in the olfactory cortex of the order unimodal sensory cortices; rather, it comes from poly-
periamygdaloid gyrus. Later, Wilder Peneld and Theodore sensory association cortices. Thus, although Brodals review
Erickson (1941) reported a patient who had olfactory sensa- was a milestone in the evaluation of hippocampal function
tions as part of his epileptic seizure and where it seems to be and is partially responsible for the currently held view that the
the hippocampus which must be the site of the discharge (p. hippocampal formation is not a major component of the
56). olfactory system, olfactory information certainly must con-
In a scholarly and comprehensive review, Alf Brodal (1947) tribute to the functions in which the hippocampus is engaged.
summarized the evidence for and against such a role for the Interestingly, olfactory learning and memory tasks are now
hippocampal formation. He noted that phylogenetic and widely used in studies of hippocampal function in rodents.
comparative neuroanatomical studies suggested that the hip-
pocampus develops in parallel with the olfactory portions of
2.3.2 The Hippocampal Formation and Emotion
the brain and that it is particularly prominent in macrosmatic
mammals such as rodents and insectivores. Moreover, even in Another inuential neuroanatomical hypothesis was pro-
the gross brain or in normal histological preparations, bers posed by James W. Papez (1937), who suggested that the hip-
from the lateral olfactory tract can easily be traced to brain pocampus was part of a circuit that provides the anatomical
regions surrounding the hippocampal formation. Even so, substrate of emotion. He was inuenced by the work of Walter
Brodal raised a number of arguments against an olfactory role B. Cannon (1929) and Philip Bard (1934), which indicated
for the hippocampus, suggesting that the association of the that the hypothalamus was essential for evocation of the
hippocampal formation with olfactory function was largely autonomic and visceral aspects of emotional behavior. He
based on circumstantial evidence. accepted the view of Cannon and Bard that emotion has two
Central to his thesis was the claim that bers arising in the component processes: emotional behavior and the cognitive
olfactory bulb did not, in fact, directly innervate any portion appreciation of emotion. In an attempt to explain certain
of the hippocampus. In particular, although he agreed that aspects of emotion, he proposed a circuit that intercon-
some olfactory bers innervated the anterior portions of the nected cortical and subcortical structures (Fig. 24). In this
parahippocampal gyrus, they did not extend back into the now famous circuit, which bears his name (Papez circuit), he
caudal portion of this gyrus where the entorhinal cortex viewed the hippocampus as a collector of sensory informa-
resides. He did not entirely dismiss an olfactory role for the tion; this information, in turn, would develop an emotive
hippocampal formation and suggested that the entorhinal state that would be transferred to the mammillary nuclei. In
cortex should be considered as concerned mainly with the addition to mediating the appropriate behavioral response to
association and integration of olfactory impulses . . . with this emotive state, the mammillary nuclei would also relay
other cortical inuences (p. 206). Among other negative evi- information to the anterior cingulate cortex via the anterior
dence, Brodal cited several studies that attested to the fact that thalamic nuclei where conscious appreciation of the emotion
the hippocampal formation was present in anosmatic and would be achieved.
microsmatic animals such as dolphins and whales (Ries and
Langworthy, 1937) and that there was substantial regional dif- The central emotive process of cortical origin may
ferentiation in microsmatic humans. Furthermore, he cited then be conceived as being built up in the hippocam-
the data of William F. Allen (1940) in which lesions of the pal formation and as being transferred to the mammil-
temporal lobe had no effect on the ability of dogs to perform lary body and thence through the anterior thalamic
olfactory discrimination tasks. Brodal (1947) concluded, No nuclei to the cortex of the gyrus cinguli. The cortex of
decisive evidence that the hippocampus is concerned in olfac- the cingular gyrus may be looked on as the receptive
tion appears to have been brought forward (p. 180). region for the experiencing of emotion as the result of
Brodals review was highly inuential in raising doubts impulses coming from the hypothalamic region, in the
concerning the role of the hippocampus in olfaction. same way as the area striata is considered the receptive
However, at least one of the pieces of evidence cited by cortex for photic excitations coming from the retina.
Brodalthat there are no olfactory projections to the hip- (Papez, 1937, pp. 725743)
12 The Hippocampus Book

objects by sight. In the emotional sphere, they were sexually

Gyrus cinguli anterior aberrant and apparently without fear of stimuli that usually
are frightening to them. It was this latter observation that
appeared to conrm the role of the hippocampus in emotion
predicted by Papez. However, it was subsequently shown that
the loss of fear and other behavioral alterations could be
Anterior attributed to damage to the amygdala and its connections
cingulum with the visual regions of the inferotemporal cortex (Mishkin,
1954; Schreiner and Kling, 1956; Weiskrantz, 1956; Aggleton,
1993; Hayman et al., 1998). The hippocampus was again in
Vicq search of a function.
d'Azyr's Unbeknown to Papez, and probably also to Klver and
bundle Bucy, a report appearing nearly a half century earlier by
Sanger Brown and Hans Schfer (1888) described the results
of a large bilateral temporal lobe lesion on a ne, large, active
Corpus Hippocampus rhaesus monkey. In addition to describing virtually all of the
mammillare fornix behavioral and emotional alterations of the Klver-Bucy syn-
drome, they also reported disturbances in memory. They
commented on the apparent forgetfulness of the monkey in
Figure 24. Original Papez circuit. Later versions included the
the following way: And even after having examined an object
amygdala, hypothalamus, prefrontal cortex, and some association
cortices. in this way with the utmost care and deliberation, he will, on
again coming across the same object accidentally, even a few
minutes afterwards, go through exactly the same process, as if
Again, a comment from Brodal (1981) is skeptical. After he had entirely forgotten his previous experience (p. 262).
having scrutinized the anatomical literature, he stated: There This little known observation of Brown and Schfer presaged
is no sound biological basis for selecting a few brain regions what has become the most enduring notion of hippocampal
as being involved in these complex functions (emotions), functionthat it plays an essential role in memory.
and today the theory of Papez is of historical interest only
(p. 672). 2.3.3 The Hippocampal Formation
However, we may conclude that the Papez circuit was and Attention Control
important historically because it shifted emphasis away from
the idea of olfactory functions of the hippocampus. Moreover, As early as 1938, Richard Jung and Alois Kornmller had
it was one of the early ideas about how the brain might have noted that desynchronization of the neocortical electroen-
more autonomy from the environment than suggested by the cephalogram (EEG) was temporally linked to a large-ampli-
then current stimulusresponse behaviorism. As we shall see, tude, sinusoidal wave pattern in the rabbit hippocampus at
there is only moderate evidence to support the specic idea between 4 and 7 Hz, which they named theta activity. Later
that the hippocampus is involved in emotion, although the work suggested that this particular activity was related to
idea has occasionally been championed, most notably in enhanced attention (Green and Arduini, 1954). John Green
recent years by Jeffrey Gray (1982, 2001). and W. Ross Adey (1956) found that arousal caused an inverse
Papezs speculations about the role of the hippocampus in relation between the electrocortical waves of the hippocampus
emotion appeared to receive support from the experiments of and neocortex. Endre Grastyn and colleagues (1959) pro-
Heinrich Klver and Paul Bucy, which were carried out at the posed that theta activity could be coupled to specic learning
same time (1937). Klver was a psychologist interested in the states (Fig. 213). Both hippocampal and entorhinal theta
neural locus of action of psychotropic drugs such as mesca- waves showed distinct changes during the acquisition of con-
line. Together with Bucy, a neurologist, he searched for the ditioned response (Holmes and Adey, 1960).
locus of action by removing brain tissue from monkeys and Other signs pointing to a role for the hippocampus in the
looking for changes in their behavioral response to drugs. The control of attention came from electrical stimulation of the
changes that followed temporal lobe removal were interesting anterior part of the temporal lobe, including the hippocam-
enough to warrant extensive description, and the constellation pus, which led to widespread and long-lasting desynchroniza-
of changes is now known as the Klver-Bucy syndrome. There tion of the EEG in both anesthetized and awake cats (Kaada et
are two important aspects to the syndrome: changes in visu- al., 1949; Sloan and Jasper, 1950). Such responses were taken
ally guided behavior and changes in emotional expression. as evidence of a general activation of attention. Stimulation of
The lesioned monkeys appeared to have visual agnosia but the hippocampus in awake cats produced a peculiar reaction
would compulsively follow stimuli brought into their visual with pupillary dilatation and slow turning of the head and
eld; they continually placed all objects encountered in their neck to the contralateral side, as if the animal experienced a
mouths, perhaps owing to their failure to recognize the weird sensory stimulus in the contralateral visual eld (Kaada
Historical Perspective 13

et al., 1953). The reaction was associated with enhanced respi- Independently, Sergei S. Korsakov (1889, 1890), a Russian psy-
ration and blood pressure. Similar reactions were elicited from chiatrist, described a similar clinical picture he called psy-
the anterior cingulate gyrus, later to be associated with higher chosis polyneuritica and emphasized that the main symptoms
analysis of emotional signals. Modern imaging methods have were memory decits. Finally, Johann Bernhard Aloys von
revealed that this latter area is deeply engaged in the analysis Gudden (1896) associated the condition with pathology of the
of emotional aspects of sensory stimuli, suggesting that it is a mammillary bodies and the mediodorsal nucleus of the thal-
device for estimating potential conicts between old and new amus, both important target stations for hippocampal efferent
experiences (Bush et al., 2000; Botvinick et al., 2001; Bishop et pathways (see Chapter 3). After the rst description of the
al., 2004; Kerns et al., 2004). Wernicke-Korsakov syndrome, a similar condition with an
In summary, this evidence suggests that together with the emphasis on memory impairment was described by another
anterior cingulate cortex some portion of the hippocampus Russian neurologist, Vladimir Bekhterev (1900). He reported
takes part in a certain form of general attention control. two patients with a prominent memory decit who were
later found at autopsy to have softening of the hippocampus
2.3.4 The Hippocampal Formation and Memory and neighboring cortical areas on both sides. This study
appears to be the rst hint of a hippocampal localization for
An early effort to analyze memory functions was that by memory.
Thodule-Armand Ribot, a French philosopher and psychol-
ogist. He proposed that memory loss was a symptom of pro- 2.3.5 More Direct Evidence for Hippocampal
gressive brain disease. His Les Maladies de la Mmoire Involvement in Memory
[Diseases of Memory] (1881) constitutes an inuential early
attempt to analyze abnormalities of memory in physiological The idea that the hippocampal formation is intimately associ-
terms. He even proposed that a plausible memory mechanism ated with memory is due to observations made on brain-
could be an alteration in the activity of engaged cells in the damaged patients by William Scoville and Brenda Milner in
cortex. This seems to be the rst hypothesis to suggest a direct 1957. Scoville removed the mesial aspects of the temporal
role of nerve cells for memory functions. Strongly inuenced lobes from several patients in an attempt to relieve a variety of
by Paul Brocas reports on language localization (Broca, neurological and psychiatric conditions. The most famous
1861a,b), Ribots views on localization of function in general of these, H.M., was a severely epileptic patient whose seizures
were further supported by the clinical evidence of John were resistant to antiepileptic drug treatment. Following sur-
Hughlings Jackson (1865), which showed memory loss in gery, his seizures were reduced, but he was left with a pro-
selected lesions, and by the experimental evidence from David found global amnesia that has persisted to this day. H.M.s
Ferrier (1876), who reported on lip and nostril movements memory decit is observable as an inability to remember
upon stimulation of the hippocampal lobes in monkeys. material or episodes experienced after the operation (antero-
Richard Semon, much inuenced by Ribot but working in grade amnesia); it also includes an inability to recall informa-
Berlin (1908), was responsible for coining the term engram tion experienced for some period of time prior to the
to signify a memory trace in the form of a physical change operation (retrograde amnesia). H.M. can remember items
in the participating nerve cells. for brief periods, provided he is allowed to rehearse and is not
Remembering that the hippocampal efferent impulses distracted. Upon distraction, however, H.M. rapidly forgets.
inuence many hypothalamic nuclei, including the mammil- Thus, he never remembers for any length of time the doctors
lary bodies, the rst link between memory functions and part who test him, his way around the hospital, or the story he has
of the hippocampal system was made by three scientists been reading. He reports his conscious existence as that of
during the 1880s who sequentially and semi-independently constantly waking from a dream and everything looking
described an affliction causing amnesia. This condition was unfamiliar (see Chapter 12). The temporal characteristics of
commonly associated with heavy drinking of alcohol and was H.M.s retrograde amnesia continue to be controversial.
later to be named Wernicke-Korsakoff psychosis. Subse- Originally it was thought by Milner to extend up to 2 years
quently, this disorder was associated with thiamine deciency, prior to the operation, with memories for events earlier than
which is often a correlate of dietary lack in the presence of that remaining relatively intact. This duration of retrograde
alcoholism. In 1881, Carl Wernicke rst described three amnesia was later extended to 11 years (Sagar et al., 1985),
patients whose illnesses were characterized by paralysis of eye with the recollection of more distant events preserved. The
movements, ataxia, and mental confusion. All three patients, interpretation of the extent of retrograde amnesia is compli-
two men with alcoholism and a woman with persistent vom- cated by the fact that this length of time was approximately
iting following sulfuric acid ingestion, developed coma and equivalent to the period of the preoperative epileptic condi-
died. In all three patients at autopsy, Wernicke detected punc- tion. We shall return to a more detailed discussion of the sub-
tate hemorrhages affecting the gray matter around the third sequent work on H.M. and other amnesic patients in Chapter
and fourth ventricles and near the aqueduct of Sylvius. He 12. Suffice it to say here that his memory disabilities are due to
believed them to be inammatory and therefore named the the removal of a large part of the hippocampal formation and
disease polioencephalitis hemorrhagica superioris (Fig. 25). surrounding cortical regions (Fig. 27).
14 The Hippocampus Book

Figure 25. Cerebral localiza-

tion and memory pioneers.
Upper left: Carl Wernicke.
Upper right: Sergei Korsakoff.
Lower left: Bernard v. Gudden.
Lower right: Vladimir Bekhterev.
(Source: Courtesy of www.

The global amnesia reported in H.M. spurred efforts to to the next) but not the delayed response task (where animals
nd an animal model of amnesia. Early studies by Jack choose a well learned response after a delay period). Even
Orbach, Brenda Milner, and Theodore Rasmussen (1960) and lengthening the delay failed to reveal a decit in the delayed
by R.E. Correll and William Scoville (1967) looked at the response task. Thus, these early attempts in the primate were
effects, in primates, of various temporal lobe lesions on mem- viewed as falling short of developing an animal model of
ory tasks such as delayed alternation and delayed response. medial temporal lobe amnesia.
Decits were seen with combined lesions of the amygdala and Work on rats also failed to nd a convincing memory
hippocampal formation in the delayed alternation task (where decit following hippocampal damage. Lesioned rats had no
animals are required to alternate their response from one trial difficulties learning simple sensory discriminations: to press
Historical Perspective 15

levers in a Skinner box or to run down alleys to obtain food or

avoid punishment. What emerged, however, was a loose con-
stellation of decits, including changes in exploration and
habituation to novelty. For example, hippocampus-lesioned
rats placed in novel environments were hyperactive (i.e., they
moved around more than control rats and failed to reduce
their activity progressively over time). They also failed to
alternate choices on two successive runs in a simple T-maze
when no rewards were available (spontaneous alternation).
Both changes in behavior were attributed to decits in
inhibitory processes; and consequently a role for the hip-
pocampus in response inhibition or Pavlovian inhibition was
postulated by Robert Isaacson and his colleagues Robert
Douglas and Daniel Kimble (e.g., Kimble, 1963, 1968; Douglas
and Isaacson 1964; Isaacson and Kimble, 1972). Another
change was a striking decit in the ability to withhold a pre-
potent behavioral response, a nding taken as support for the
response inhibition hypothesis. Hippocampectomized rats
learned to run down an alley or press a lever for food as
quickly as normals but were decient in changing this behav-
ior when the circumstances altered (e.g., when food was with-
drawn or when an alternate response was rewarded).
Interestingly, they were consistently better at learning one par-
ticular type of avoidance task: two-way active avoidance. In
this task, the animal must learn to run back and forth between
two compartments of a shuttle box to escape punishment.
This improved learning seemed particularly anomalous in
Figure 26. Drawings of the brain of patient H.M., made on the
light of the presumed role of the hippocampus in memory.
basis of an MRI examination performed 40 years after bilateral
resection of the medial temporal lobe. Drawings are of transverse A third decit lay in complex maze learning. Whereas they
MRI sections from various anterior-posterior levels: A through the had no trouble learning to make a turn in a simple T- or Y-
amygdala; D at the posterior tail of the hippocampus. (Source: maze, they failed miserably at more complex multichoice con-
Corkin et al., 1997.) The lesion was bilateral, but its right side is gurations (Kaada et al., 1961; Kimble, 1963; Kveim et al.,
here left intact to show structures that were resected. 1964).

Figure 27. Brenda Milner (left)

pioneered the analysis of the role
of the medial temporal lobes in
memory. Donald Hebb (right)
inuenced a generation of neuro-
scientists with his theories on
memory mechanisms. (Source:
Courtesy Brenda Milner and
16 The Hippocampus Book

This confused state of affairs remained until around 1970

when several developments took place that together changed
the approach to the hippocampal theory. The changes, which
have continued to be inuential, involved alterations in con-
cepts about memory, the development of new tests for assess-
ing memory, and the application of single-cell recording
techniques to behaving animals.
Around this time, it was realized that there might be more
than one type of memory, only one of which involved the hip-
pocampus (Gaffan, 1974a; Hirsh, 1974; Nadel and OKeefe,
1974; Olton et al., 1978). Two related ways of classifying
human memory proved particularly inuential. The rst was
Endel Tulvings contrast between memories for episodes set in
a spatiotemporal context as distinguished from those for
semantic items such as facts and other context-independent
material (Tulving, 1972). The second was the related but dif-
ferent distinction between declarative and procedural memo-
ries, which was originally introduced by Terry Winograd
(1975) in the eld of articial intelligence, subsequently
applied to studies of amnesia by Cohen and Squire (1980),
and extensively promulgated by Larry Squire and his col-
leagues (Squire, 1988, 2004). The former include both
Figure 28. Olga Vinogradova was among the rst who recorded
episodic and semantic memory, and the latter refers to skills
single units from the hippocampus of awake animals (rabbits) while
and other stimulus-response habits. Both memory schemes
testing their behavioral responses to natural stimuli. She interpreted
have inuenced research on amnesic patients and animal her recordings to mean that the hippocampus served as a novelty
models of amnesia (see Chapters 12 and 13). detector. (Source: Courtesy of Richard Morris.)
The second important development around 1970 owed
from the realization that the behavioral tests for animal mem- lesions had with avoidance learning. They concluded that
ory available at that time were not optimal for testing hip- decits were found only when the animals had to learn to
pocampal function. David Gaffan (1974b) introduced unique avoid places, not when they were given a clear prominent
object recognition tasks. Mortimer Mishkin and Jean object or cue to avoid. Selective tests for the spatial functions of
Delacour (1975) developed a version of this task that has the hippocampus, including the Olton radial arm maze and
become the standard test for recognition memory in the mon- the Morris watermaze, were subsequently developed (see
key (see Chapter 13). Following the introduction of these Chapter 13).
more powerful testing paradigms, efforts to establish memory
systems became increasingly more fruitful. 2.3.7 Conclusions

2.3.6 The Hippocampus as a Cognitive Map We have provided a brief overview of some of the historical
perspectives concerning the function of the hippocampal for-
An important development in the analysis of hippocampal mation. Although there is widespread acceptance of the view
function was the use of implanted microelectrodes to monitor that the hippocampal formation is involved in some aspects of
single-neuron activity in the hippocampus of the awake intact memory, there may still be some surprises as to other poten-
animal (Hirano et al., 1970; Vinogradova et al., 1970; OKeefe tial functions of the hippocampal formation. As an important
and Dostrovsky, 1971; Ranck, 1973). These experimenters example, the hippocampus has been implicated in the modu-
described the relations between cellular activity and a variety lation of stress responses through inhibitory projections to the
of sensory and behavioral parameters (Fig. 28). One corre- hypothalamus. As described in Chapter 15, this putative role
latethat of the animals location in the environmentgave of the hippocampus has been implicated in a variety of nor-
rise to the cognitive map theory, which has fostered research mal and abnormal stress responses such as post-traumatic
into the spatial functions of the hippocampus (OKeefe and stress disorder.
Dostrovsky, 1971, see Chapters 11 and 13). This theory sug-
gested that the hippocampus in animals was dedicated to spa-
tial memory and allowed the animal to navigate in familiar
environments. Extension of this theory to humans envisaged 2.4 Special Features of Hippocampal
the addition of a temporal signal, allowing the hippocampus to Anatomy and Neurobiology
act as a spatiotemporal context-dependent (episodic) memory.
Lynn Nadel, John OKeefe, and Abe Black (1975) proceeded to We now turn to an overview of some of the early research on
look carefully at the problems that rats with hippocampal the hippocampal formation and the important historical
Historical Perspective 17

gures associated with it. The outcome of some of this porting the reticular theory. He also pictured the dendrites of
research led to general principles of neuroscience (e.g., the granule cells as being continuous with blood vessels that occu-
neuron doctrine), whereas others have established the exis- pied the hippocampal ssure. Camillo Golgis most outspoken
tence of unique features of the hippocampal system (e.g., the adversary in the conict over the neuron doctrine versus the
largely unidirectional nature of the intrinsic excitatory con- reticular theory was Ramon y Cajal, whose support of the
nections). In some cases, the motivation for studying the neuron doctrine came, ironically, from preparations made
hippocampal formation was not so much to understand with Golgis method. Although he also had many examples
its function but to capitalize on some unique aspect of its from the hippocampal formation, particularly in newborn
anatomical and physiological organization to carry out an rodents, his favorite preparation was the cerebellum. After
experiment of more general interest more easily. examining thousands of preparations, he became convinced
A number of features have attracted scientists to the hip- that individual neurons did not form a syncytial network. One
pocampus for studies of general neuronal and systems proper- particularly important piece of evidence was that he observed
ties. A short list of the useful anatomical and neurobiological individual neurons that were in the process of dying. Neurons
features of the hippocampus includes the following. adjacent to these cells were entirely healthy. This was convinc-
ing evidence to Ramon y Cajal that neurons were individuals,
A single cell layer and strictly laminated inputs
not just members of interdependent groups. The debate lasted
Predominantly unidirectional connections between
a number of years with a large number of participants, and
a series of cortical regions
was notable for the vigorous exchanges between the two
Extrinsic and intrinsic bers making numerous en pas-
sage contacts with target neuronal dendrites, running
The nal resolution of this classic debate did not occur
orthogonal to the main dendritic axis
until the advent of electron microscopy, which conclusively
Synapses that are highly plastic
supported the neuron doctrine by showing the neurons as
Tissue that can be used in transplantation studies
individual entities with no connecting intercellular brils
Neurons that can be successfully grown in
(Shepherd, 1972).
Acute or cultured slices surviving for prolonged periods
in vitro 2.4.1 Early Neuroanatomical
Studies of the Hippocampus
The most striking difference between hippocampal and
neocortical cortices is the aggregation of the principal cells in Hippocampal neuroanatomy beneted from the pioneering
a single layer. Another major difference between the two cor- investigations of Camillo Golgi (1886), Luigi Sala (1891),
tical types is the direction of the afferent bers. In contrast to and Karl Schaffer (1892). Santiago Ramon y Cajal (1893)
the input to neocortical areas where most afferent bers are described the stratication of the various afferent systems
radially oriented, the extrinsic and intrinsic afferent nerve and drew a distinction between cells with long and short
bers of the hippocampal formation run in a horizontal direc- axons (Fig. 29).This observation made it clear that a hip-
tion (parallel to the pial surface) and orthogonal to the apical pocampal neuron could inuence a large number of target
dendritic axis. cells and areas. Even before the formal denition of synapses
Hippocampal studies were important during the nine- by Charles Sherrington in 1897, Ramon y Cajal saw the
teenth century controversy between the neuron doctrine and functional implication of lamination. He suggested that there
the reticular theory. The neuron doctrine proposed that each was a convergence of afferent input onto a single neuron, a
neuron was an individual cell that contacted but did not notion that we take for granted today. His monumental effort,
merge with other target neurons. The reticular theory, on the including an analysis of all portions of the hippocampal for-
other hand, posited that nerve cells form a syncytium where mation in a number of animal species, allowed him to propose
one cell emits a protrusion, or bril, that continues into other a functional circuit diagram of this region. Following his
cells, thus creating an interconnected network of bers from a principle of dynamic polarizationinput to the dendrites,
large number of neurons. This controversy continued for output through the axonhe placed arrows indicating his
some time because microscopes at the time were not able to view of the direction of impulse ow through the hippocam-
resolve the neuronal membranes and the spaces (e.g., the pal formation; many of these circuit characteristics have
synaptic clefts) that separated individual neurons. stood the test of time. He emphasized the size and variabil-
Camillo Golgi, who strongly supported the reticular the- ity of the various dendritic trees and gave a detailed descrip-
ory, used observations of the hippocampal formation to bol- tion of dendritic spines and their distribution (structures
ster his arguments. Employing the reazione nera (black that were dismissed by Golgi as artifacts of his staining
reaction), as he called the staining technique, later to be called method).
the Golgi method, he repeatedly pointed to the convergence of Simultaneous with Ramon y Cajals rst studies, the
ne axonal branches from a large number of dentate granule Hungarian anatomist Karl Schaffer, who gave his name
cells to form a dense bundle in the hilus of the dentate gyrus (Schaffer collaterals) to the axons from CA3 cells that termi-
(vertical bundle in Fig. 22). Here, he believed laments from nate in CA1 (1892), found by meticulous charting of well
various axons made a large intertwined feltwork, thus sup- impregnated Golgi material not only that axons had short
18 The Hippocampus Book

Figure 29. Santiago Ramon y Cajal

and his famous drawing of the hip-
pocampus in his 1911 book Histologie
de Systme Nerveux. in two volumes.
The arrows give his interpretation of
likely impulse direction. Later func-
tional studies have vindicated his ideas
on nearly all points. Portrait from Royal
Society, London, gratefully acknowl-
edged. Permission to reproduce draw-
ing from Cajal (1911) from Istituto
Cajal is gratefully acknowledged.

branches but that some axons could be very long, connecting degenerating bers could be stained better than intact bers.
neurons in neighboring cortical elds. Nearly 40 years later Temporal myelinization gradients were also useful. Unfor-
Rafael Lorente de N (1934), building upon the work of his tunately, these methods initially met with limited success in
compatriot Ramon y Cajal, greatly extended the analysis of the hippocampus. Although the hippocampus contains sev-
the many hippocampal cell types and their axonal and den- eral ber systems with moderately thick myelinated bers,
dritic patterns. He described many detailed networks of inter- most of its axons are much thinner than in other parts of the
connected neurons (Fig. 210). On the basis of their dendritic central nervous system (Shepherd et al., 2002). These features
tree and connections, he divided the hippocampal formation are probably one reason why the rst available methods for
into a set of clearly dened divisions and coined the well experimentally establishing pathways by degenerating bers
known terms CA1 to CA4. did not give satisfactory results. For example, even after
The pioneering neuroanatomists came impressively close appropriate lesions, Vittorio Marchis original method
to modern-day thinking in their proposed schemata for func- (Marchi and Algeri, 1886) stains only a small number of
tional connectivity in the hippocampus. More often than not, degenerating bers in the alveus and mbria. In addition, the
the diagrams of Ramon y Cajal (1893, 1911) indicated the cor- stained bers are exclusively among the thicker of those
rect direction of information ow. Likewise, the detailed present.
drawings of various neurons by Lorente de N (1934)in
which he used the number of dendritic spines to estimate the 2.4.3 New Anatomical Techniques that
relative efficiency of afferents terminating at various positions Revolutionized Connectivity Studies
on the dendritesallowed him to formulate a general rule
about summation. Even so, to understand the operational The situation improved greatly when certain variants of silver
rules of the hippocampus, specic details beyond the major impregnation were applied to lesioned hippocampal path-
sources and trajectories of the main afferent and efferent ways. Admittedly, the rst silver degeneration method devel-
pathways were needed. oped by Paul Glees (1946), was disappointing in the
hippocampus despite the success it had in pathways of the
2.4.2 New Fiber Tracing Methods brain stem. However, Nautas 1950 method, soon followed by
the Nauta and Gygax variant (1954) and Fink and Heimers
In parallel with the classic neuroanatomical studies with the method (1967), were all remarkably effective.
Golgi method just described, much work was carried out to Although similar in principle to other methods, Walle
trace major afferent and efferent ber systems in the central Nautas methods turned out to be highly useful in hippocam-
nervous system. The early studies exploited the fact that pal tissue. By employing the rst variant of Nautas tech-
Historical Perspective 19

Figure 210. Top: a rare picture of the pioneer Lorente de N, who had every reason to be
proud of his 1934 drawings shown below, signed along the margin. The diagram shows the
cell types and divisions and the terminology in general use today: fascia or gyrus dentata, CA1
to 4, subiculum, pre- and parasubiculum and the entorhinal area. Acknowledgment to Larry
Swanson for the portrait. Permission from Georg Thieme Verlag, Stuttgart to reproduce de
Ns drawing gratefully acknowledged.
20 The Hippocampus Book

niques, Theodor Blackstad (1956, 1958) (Fig. 211) showed were the basis for the so-called trisynaptic circuit (Andersen et
how commissural and ipsilateral afferents to several parts of al., 1966). With this unidirectionality, the hippocampal system
the hippocampus terminated in laminae oriented parallel to differs fundamentally from the reciprocal connectivity of
the cell layers (Fig. 211). These data not only veried the gen- nearly all neocortical areas. This principle of hippocampal
eral principles established by Ramon y Cajal on stratication neuroanatomy is discussed in detail in Chapter 3.
of afferent bers but in addition showed astonishing density When a small ber bundle of the Schaffer collaterals was
of presynaptic boutons in the innervated zone. stimulated, the amplitude of the synaptic potentials showed a
characteristic distribution over the CA1 region. Signals were
2.4.4 Predominantly Unidirectional detected in the whole transverse extension of CA1. However,
Connectivity Between Cortical Strips within the borders to CA3 and the subiculum, the largest
potentials were detected along a strip oriented nearly trans-
A key aspect of hippocampal connectivity was to emerge from versely to the longitudinal axis of the hippocampus, with
these analyses: unidirectionality. Because the Golgi technique gradually reducing amplitudes on either side, the range span-
stains a relatively small proportion of the total number of ning nearly half the length of the hippocampus. With stronger
neurons, this picture needed to be complemented by methods stimulation, the synaptic potentials generated a population
giving more quantitative data. Systematic study of each region spike (see Section 2.3), which had the same main orientation
of the hippocampal formation with suitable silver degenera- but with more restricted longitudinal distribution (Andersen
tion techniques, later supplemented by electron microscopic et al., 1971b). This general arrangement was found for all four
and anterograde and retrograde tracing methods, led to the pathways studiedperforant path, mossy bers, Schaffer col-
realization that each component of the hippocampal forma- laterals, CA1 axons in the alveusand the same orientation
tion projects to its neighboring region but generally does not was found with orthodromic as with antidromic testing. This
receive a return pathway from this target (Hjorth-Simonsen, arrangement, which was named the lamellar organization, has
1973). This was in good accord with previous physiological received considerable opposition, mostly on anatomical
studies in which activation of the perforant path led to grounds (see Chapter 3), probably because the name lamella
sequential activation of CA3 followed by CA1. A cut of CA3- invited vsualizing a set of thin slices with sharp borders, like a
to-CA1 bers removed activity caudal to the section in CA1 loaf of bread. Instead, the lamella should be seen as the main
and subiculum. On the other hand, stimulation in CA1 did orientation of a set of overlapping fan-shaped ber regions (Li
not give synaptic activation in the CA3 region. These ndings et al., 1994), with most branches and their synaptic effect
along the major orientation and gradually less inuence on
Figure 211. Theodor Blackstad, often called the father of modern either side. The efficiency of the transverse slice is an illustra-
hippocampal histology because of his ground-breaking studies of its tion of this organization, although the spread of ber orienta-
external and internal connectivity, with charting of degenerating tion allows signals to be elicited in slices cut at a less than
bers through silver staining and electron microscopy. (Source: optimal angle.
Courtesy of Jon Storm-Mathisen.)
2.4.5 New Tracing Studies
Using Axonal Transport

In addition to the connectivity of single cells, it was necessary

to determine how aggregations of neurons were connected so
early researchers could gain insight into how larger systems
could operate in concert. New anatomical techniques
exploited intra-axonal transport. Maxwell Cowan (Fig. 212)
and his colleagues introduced a new technique for ber trac-
ing based on the anterograde transport of injected radioactive
amino acids (Cowan et al., 1972) (Fig. 212). Here, they
exploited earlier information showing that radioactive pro-
teins could be transported inside axons in both anterograde
and retrograde directions (Grafstein, 1971; Lasek, 1975).
Cowan had an impressive lineage: His doctoral (PhD) studies
were carried out under the mentorship of Thomas Powell of
Oxford University (Fig. 212). Powell, Cowan, and Geoffrey
Raisman were responsible for many of the classic studies of
hippocampal and fornix connections (Raisman et al., 1965,
1966). Max Cowans laboratory, in turn, spawned a number of
hippocampal researchers, including Larry Swanson, Brent
Staneld, and David Amaral, who among them conducted a
Historical Perspective 21

Figure 212. Thomas (Tom) Powell (top left) and W. Maxwell (Max) Cowan (top right), the
former born in Britain and the latter in South Africa. They both had illustrious careers in neu-
roscience, starting their work by tracing ber systems in the hippocampal formation. (Source:
Courtesy of Geoffrey Raisman.)

substantial portion of the modern studies on hippocampal microscopy gave an unprecedented view of the details of indi-
connectivity. The autoradiographic method had the merit of vidual cells and the contacts of neural circuits of the hip-
allowing extremely small and discrete injections to be placed, pocampus and its main target nuclei (Raisman, 1969).
with the 3H-amino acids being taken up and transported by Moreover, the electron microscope also provided a new
cells of origin, not bers of passage. One important result of method for ber tracing. When a ber system was lesioned,
this study was that it became clear that direct hypothalamic the subsequent degeneration of both bers and their associ-
projections of the hippocampal formation arose not from the ated boutons darkened within 1 to 2 days, a process that could
hippocampus proper but from subicular regions (Swanson be used for both synapse indentication and connectivity
and Cowan, 1977). However, the hippocampal formation can studies (Alksne et al., 1966).
inuence a multitude of hypothalamic nuclei through multi-
synaptic pathways. 2.4.7 Hippocampal Synapses Are Highly
In addition to the autoradiographic method, new methods Plastic: Early Studies of Sprouting
that depended on axonal transport of various proteins or
plant lectins were developed (Kristensson and Olsson, 1971; The term plasticity is used to indicate several types of change.
Gerfen and Sawchenko, 1984). In addition to the well known activity-dependent alterations
Fibers containing monoamines were also found to inner- of synaptic efficiency (e.g., long-term potentiation, described
vate the hippocampus. The rst reports described noradrena- in Chapter 10) there are numerous examples of anatomical
line (norepinephrine)-containing axons from the locus changes in cell dimensions and number and in axonal length,
coeruleus (Segal and Bloom, 1974a,b). branching and connections, and biochemical content in
response to various stimuli or injury (see Chapter 9). The hip-
2.4.6 Electron Microscopy Offers pocampal formation was the rst brain region in which
New Opportunities axonal sprouting and reactive synaptogenesis were unequivo-
cally demonstrated. Convincing experimental evidence of
The rst steps toward analysis of the number and types of adult synaptic reorganization was described by Geoffrey
synapses in the central nervous system also proted from Raisman (1969). He studied two sets of afferent bersone
work on the hippocampus. Following the pioneering work of from the hypothalamus and the other from the hippocam-
Sanford Palay and George Palade in other brain regions (Palay pusthat converged on cells of the septal nuclei. After long-
and Palade, 1955), Lionel Hamlyn (1963) carried out the rst term lesions of the mbria, there was an increased number of
electron microscopic analysis of the hippocampus. He pro- multiple synapses (boutons in contact with several dendritic
vided the rst comprehensive description of all synapses con- spines), interpreted as being the result of residual axons
tacting the various portions of the dendritic tree in both CA3 sprouting. Conversely, following removal of hypothalamic
and CA1 pyramidal cells. The introduction of electron afferents, mbrial bers were found to contact somata of sep-
22 The Hippocampus Book

tal cells, which they rarely do normally. In another report, 1981, 1984). Lesions of the commissural or perforant path
Raisman and Pauline Field (1973) exploited the fact that m- bers did not induce such effects. The grafted sympathetic
bria bers innervate a segment of the lateral septal nucleus on axons respected the lamination borders and formed typical
both sides of the midline. A unilateral mbrial lesion gave rise norepinephrine-containing boutons with target neurons.
to remarkable sprouting from the contralateral ber system Similar results were obtained when cholinergically deaffer-
such that the total number of fimbrioseptal synapses ented hippocampi received implants of acetycholine-contain-
remained unchanged. Further work by the same group using ing cells placed in a cavity formed after the original septal
transplantation of embryonic tissue for reinnervation of den- region had been removed. Again, both peripherally and cen-
ervated hippocampal areas (Raisman and Field, 1990) sparked trally located donor neurons were effective, and the reinner-
interest in the plasticity of hippocampal afferent and efferent vation by transplanted cholinergic bers displayed homotypic
ber systems. However, the authors also warned that the reor- localization. Particularly impressive was the functional recov-
ganization was not necessarily adaptive or functionally ery of spatial learning ability and the partial restitution of hip-
important. A special case was the crossed entorhinal-dentate pocampal place cell activity (Shapiro et al., 1989), which
pathway, allowing a set of control experiments to determine paralleled the regrowth of cholinergic bers (Dunnett et al.,
factors of importance for reinnervation (Steward et al., 1974). 1982).
Nearly simultaneously, sprouting of hippocampal ber sys- Following these encouraging initial observations, trans-
tems was demonstrated with another approach by Gary Lynch plantation research in the hippocampal formation ourished.
and Carl Cotman and their associates (Lynch et al., 1972; Subsequent efforts were directed at determining if cerebellar
Mosko et al., 1973; Matthews et al., 1976; Cotman et al., 1977; neurons would survive when transplanted into the hippocam-
Goldwitz and Cotman, 1978). This group rst showed that the pus and if hippocampal tissue might be electrophysiologically
acetylcholinesterase-containing septohippocampal fibers active when grafted into the cerebellum and vice versa. In the
innervating the dentate gyrus increased in intensity and dis- latter situations, the transplanted hippocampal tissue
tribution if the perforant path bers to the same region were assumed its original cellular appearance and synaptic lamina-
removed. Electron microscopy conrmed that there was an tion, and the cells showed electrophysiological properties typ-
early postlesion loss of synapses followed by a partial recovery. ical of hippocampal neurons (Hounsgaard and Yarom, 1985).
The remarkably efficient and fast regeneration is an impres-
sive example of the plastic properties of hippocampal neu- 2.4.9 Hippocampal Cells Grow Well in Culture
rons. After lesions to the perforant path, for example, the
maximal reduction of associated boutons in the molecular The hippocampus has become a favorite source of neurons in
layer of the dentate gyrus was seen after about 5 days. several forms of tissue culture. Early successes at dening the
However, after 2 weeks there was substantial recovery, and parameters for successful survival and maturation of hip-
after 3 weeks the tissue had regained normal bouton density pocampal neurons came from the work of Gary Banker in the
(Matthews et al., 1976). laboratory of Max Cowan (Banker and Cowan, 1977; Banker
and Goslin, 1991). In such cultures both neurons and glial
2.4.8 Hippocampal Neurons: Transplantable cells are readily identiable and may be manipulated or
with Retention of Many Basic Properties recorded as individual elements. A disadvantage, however, is
that the original cytoarchitecture disappears, including such
Another remarkable example of neuronal plasticity is the abil- characteristic hippocampal features as the laminated input
ity of transplanted neuronal cells to emit axonal branches that with synaptic segregation. This problem led to the develop-
grow and connect to existing neuronal target cells. The hip- ment of the organotypic slice by Beat Ghwiler and colleagues
pocampal formation was the testing ground for analysis of (Ghwiler, 1981, 1988; Ghwiler and Brown, 1985). With this
incorporation of embryonic transplants. After the work by technique, thin slices of hippocampal tissue can be grown for
Raismans group, described above, Anders Bjrklunds group several weeks with an impressive capacity for neuronal growth
(Bjrklund et al., 1975, 1976) independently investigated the and differentiation (Zimmer and Ghwiler, 1984). With care,
regenerative ability of transplanted monoaminergic tissue most neuronal phenomena seen in vivo or in acute slices can
to reinnervate normal hippocampal tissue. They showed be demonstrated. In addition, the longevity of the preparation
that transplanted tissue containing catecholamine- or acetyl- makes possible a number of experiments, including rehabili-
choline-synthesizing neurons could extend axons and rein- tation after injury and analysis of the long-lasting effects of
nervate the hippocampus in a normal fashion. When a graft of growth factors.
norepinephrine-containing neurons, whether from a periph-
eral source such as the sympathetic superior cervical gland or 2.4.10 Development of Hippocampal Slices:
from a central location such as the locus coeruleus, was From Seahorse to Workhorse
allowed to grow into a noradrenergic denervated hippocam-
pus, grafted cells were observed to send out axons that entered A major technological advance for hippocampal research, and
the hippocampal formation but only if the cholinergic indeed for the eld of neurobiology, was the development of
septo-hippocampal bers were cut (Bjrklund and Stenevi, the in vitro hippocampal slice preparation. The pioneer in this
Historical Perspective 23

effort was the neurochemist Henry McIlwain. He set out to Role of oscillations in neuronal networks
develop a reliable and functional in vitro preparation to inves- Underlying mechanisms of epileptogenesis
tigate how various stimuli and compounds inuenced the bio-
chemistry of central nervous tissue (Li and McIlwain, 1957). 2.5.1 Identication of Excitatory
He used a variety of preparations from the central nervous sys- and Inhibitory Synapses
tem. Despite considerable success, most of the horizontally cut
neocortical slices did not allow a physiologically realistic affer- George Gray (1959) found that synapses in visual and
ent impulse pattern. More physiologically relevant data came frontal cortex could be divided into two distinct structural
when he employed slices from the piriform cortex. Here, stim- types, which he called type 1 and 2. Type 1 synapses were
ulation of the lateral olfactory tract elicited synaptic and cell more heavily stained, and the synaptic specialization was
activity in neurons of the piriform cortex (Yamamoto and asymmetrical in that the presynaptic density was less intensely
McIlwain, 1966a,b; Richards and McIlwain, 1967; McIlwain stained than the postsynaptic density. Type 1 synapses
and Snyder, 1970). Tim Bliss and Chris Richards (1971) were usually associated with dendritic spines. Type 2 synapses
showed that eld potentials similar to those found in the intact had pre- and postsynaptic densities with the same moderate
animal could be elicited in slices of the dentate gyrus cut along staining intensity, and they were found mainly in associa-
the septal-temporal axis of the hippocampus. With some tech- tion with dendritic shafts and neuronal somata. As to pos-
nical modications of the McIlwain procedure, Per Andersen, sible functional consequences of these differences, Gray (1959)
Knut Skrede, and Rolf Westgaard started to use transversally concluded that, At present there is no evidence to suggest that
sectioned hippocampal slices from guinea pigs and demon- type 1 and type 2 synapses are functionally different (p. 252).
strated that intrinsic pathways could be successfully activated When Grays colleague Lionel Hamlyn (1963) studied the
(Skrede and Westgaard, 1971). Single-cell activity could be hippocampus, he found the same two synapse types and a
recorded and showed short-term synaptic plasticity similar to similar distribution on dendritic spines and shafts. All
that in intact, anesthetized preparations (Andersen et al., synapses involving spines were of type 1, whereas synapses
1972). Rat and mouse hippocampal slices were useful as well, formed on the soma and thicker dendritic trunks or smooth
with the same type of signals as are seen in intact preparations. branches were always type 2. A simultaneous report by
A distinct advantage was the great stability of the isolated slice Theodor Blackstad and Per Flood (1963) corroborated these
preparation, which allowed long-lasting high-quality intracel- data. It was comforting that the hippocampal cortex showed
lular recordings to be made (Schwartzkroin, 1975). Important the same synaptic structures as those in neocortical areas.
also was the precision with which electrodes could be placed Conversely, this similarity made it possible that ndings in
and lesions created. The virtual lack of a blood-brain barrier is hippocampal tissue also could be generalized to other cortical
an experimenters dream. A new opportunity arose when areas. What was needed for such a purpose was a correlation
intracellular recording could be combined with iontophoretic of functional and structural data from the same tissue. Such
delivery of transmitter candidates or blockers from multiple an opportunity arose with the advent of intracellular record-
pipettes arranged to hit various parts of the dendritic tree ings from hippocampal neurons.
(Schwartzkroin and Andersen, 1975). Fortunately, acute
slices also supported well developed long-term potentia- 2.5.2 Gray Type 2 Synapses are Inhibitory
tion (Schwartzkroin and Wester, 1975), a prerequisite for and are Located on the Soma of Pyramidal
the wealth of analytical studies of this phenomenon (see and Granule Cells
Chapter 10).
In an inuential intracellular study, Eric Kandel, W. Alden
Spencer, and Floyd Brinley, Jr. (1961) reported on the ubiqui-

tous and long-lasting inhibitory postsynaptic potentials

2.5 Several Neurophysiological Principles Have
(IPSPs) following both orthodromic and antidromic activa-
Been Discovered in Hippocampal Studies
tion of cat CA2-CA3 pyramidal cells. Per Andersen, John
Eccles, and Yngve Lyning (1964a) used a combination of
Although at rst glance the hippocampal formation appears
intracellular recording of IPSPs and laminar analysis of the
to be a special part of the brain with many features not
associated eld potentials elicited by the same two activation
seen elsewhere, it shares enough general features with other
modes. They showed that the initial phase of the IPSP in both
cortical regions to make it remarkably useful for studies
CA1 and CA3 pyramidal cells and in dentate granule cells was
of a general nature. Neurobiological principles that have
associated with an extracellular positive eld potential with a
been discovered from work on hippocampal preparations
sharp maximum at the cell body layer, indicating a synaptic
current leaving the somata of many principal neurons. This
Identication of excitatory and inhibitory synapses and observation could be explained if the basket cell synapses,
their localization, transmitters, and receptors which terminate at the soma and initial axon, caused activated
Discovery of long-term potentiation and long-term hyperpolarizing chloride conductances, and were in fact
depression inhibitory interneurons. The basket cell synapses were exclu-
24 The Hippocampus Book

sively Gray type 2 (Hamlyn, 1963). Andersen, Eccles, and gave rise to a localized fEPSP carrying a population spike,
Lyning (1964b) looked for cells that were not antidromically proving the excitatory nature of the connection. Following
invaded from the alvear bers and thus were not pyramidal specic surgical lesions to each of these monosynaptic excita-
cells. They discovered a subset of such neurons that dis- tory pathways, the method of Alksne et al. (1966) showed
charged repetitively with a time course corresponding to the that boutons belonging to all four pathways were
rise time of synaptically activated IPSPs found in pyramidal electron-dense, closely associated with dendritic spines, and
and dentate granule cells, respectively. The location of such Gray type 1 (Andersen et al., 1966a). Thus, the widely
cells corresponded to the basket cells described by Ramon y accepted idea that cortical (and many other) excitatory
Cajal (1911) and Lorente de N (1934). Hippocampal IPSPs synapses are Gray type 1 and are usually located on dendri-
were found to be bicucculine-sensitive by David Curtis, John tic spines emerged from these studies of the hippocampal
Felix, and Hugh McLennan (1970) and therefore were medi- formation.
ated by -aminobutyric acid ionotropic receptor type A In conclusion, within a few years during the 1960s, both
(GABAA) receptors. inhibitory and excitatory synapses of the hippocampus were
Thus, the dentate and hippocampal basket cells were the identied with respect to location, anatomical type, and func-
rst interneurons for which both the functional role and the tion. These ndings turned out to be applicable to many,
identity of the effective synapses were revealed. Subsequently, albeit not all, parts of the central nervous system.
ndings from the hippocampal formation served as a tem-
plate for similar searches in the cerebellum, thalamus, and 2.5.4 Long-lasting Alterations of Synaptic
other parts of the central nervous system. Efficiency After Physiological Stimulation
However, the basket cell inhibition was far from exclusive.
Later work in the cerebellum, hippocampus, and neocortex Since the discovery of long term potentiation (LTP), hip-
showed that many more types of inhibitory interneurons and pocampal synapses have been the most studied exemplars of
synapses were present (Eccles et al., 1966). synaptic plasticity. These investigations are dealt with in detail
in Chapter 10. Prior to the discovery of LTP, many other
2.5.3 Gray Type 1 Synapses are Excitatory processes had been studied as potential models for learning.
and are Located on Dendritic Spines In particular, post-tetanic potentiation (PTP) was a common
candidate for a number of years. Originally described by
Identication of the morphological and functional character- Martin Larrabee and Detlev Bronk in the sympathetic ganglia
istics of excitatory synapses was also rst carried out in the (1939), PTP appeared as enhanced synaptic responses follow-
hippocampus. Kandel et al. (1961) saw examples of excitatory ing a period of high frequency tetanization of afferent fibers.
postsynaptic potentials (EPSPs) in cat CA3 and CA2 pyrami- The main weakness of PTP as an adequate mechanism for
dal cells in response to subiculum or mbria stimulation, but behavioral learning was its limited duration; in the spinal
they were surprised how relatively rare such responses were cord, PTP lasted only up to 7 minutes (Lloyd, 1949). The
compared to the ubiquitous IPSPs. This relative rarity made duration was increased to 30 minutes by Alden Spencer
them difficult to analyze in detail. In rabbits and rats, eld (Spencer et al., 1966) using long-lasting tetanization of poly-
potential recording allowed better identication of synapti- synaptic spinal reexes.
cally activated cells. Limited duration was also an obstacle in the usefulness of
Identication of excitatory synapses and their localization the early studies of synaptic plasticity in the hippocampus.
were the result of a combination of extra- and intracellular Several investigators had noted the remarkable growth of
responses to various excitatory afferent pathways and a new synaptic potentials during a period of high frequency stimu-
method for marking synapses. In the perforant path-to-den- lation (Cragg and Hamlyn, 1955; Kandel and Spencer, 1961b;
tate granule cells, the rst association between a eld EPSP Gloor et al., 1964). The critical question was: Would the
(fEPSP) and an intracellular EPSP was made by Per Andersen, enhanced excitability last for a signicant period following
Birgitta Holmqvist, and Paul Voorhoeve (1966b), proving that cessation of the tetanic stimulation? Andersen (1960) noted
the fEPSPs used were monosynaptic events and that the that a period of enhanced synaptic response of commissural
synapses in question were excitatory. responses of CA3 and CA1 cells did indeed follow a short
With a similar approach, several other excitatory synapses period of 10- to 20-Hz stimulation, but the duration was only
were identied including perforant path bers activating den- up to 8 minutes and was initially regarded as an example of a
tate granule cells, mossy bers activating CA3 cells, and com- special form of PTP. It soon became clear, however, that
missural and local intrinsic bers in the stratum oriens longer-lasting enhancement could be achieved when higher
activating CA1 pyramidal cells. Using intracellular recordings stimulation frequencies and, above all, repeated tetani were
of monosynaptic EPSPs in rabbits and cats, in response to applied, as was rst reported by Lmo (1966). This important
stimulation of four independent afferent ber systems to CA1 discovery led the way to a description of truly long-term
and the dentate granule cells, the activated synapses were synaptic plasticity (hours and weeks), which is the basis for
identied as excitatory. In each system, synaptic activation studies of LTP to this day (Bliss and Lomo, 1970, 1973). For a
Historical Perspective 25

full description of the processes and mechanisms underlying ranges, have been studied in hippocampal preparations (see
LTP, see Chapter 10. Chapter 8).

2.5.5 Hippocampal Systems: Exhibiting Several 2.5.6 Studies of Epileptiform Behavior

Types of Oscillatory Behavior
Richard Jung (1949) reported that electrical polarization (by
A few years after Hans Berger (1929) described the human long-lasting direct current) of the hippocampal formation
EEG, Alois Kornmller in Berlin began a search for possible readily gave rise to electrical after-discharges similar to those
electrical equivalents to the cytoarchitectonic parcellation of seen during epileptic seizures. By measuring the stimulus cur-
the cerebral cortex pioneered by Korbinian Brodmann (1909), rents necessary to elicit such seizures in various tissues, he
as mentioned in Section 2.1.3. concluded that compared to other cortical areas the hip-
Working with rats, Case Vanderwolf (1969) discovered that pocampal formation needed considerably less current to elicit
theta activity occurred each time the animal initiated a vol- epileptiform seizures. This observation was in good accord
untary movement, in contrast to movements initiated more with clinical experience gained from the EEG techniques
automatically or in a reex-like manner. He also identied introduced during the mid-1930s, suggesting that many
two types of theta activityatropine-sensitive and atropine- epileptic conditions originated in medial temporal structures
insensitivethe former related to attention and the latter (Fig. 2-13).
to movements (Kramis et al., 1975). Atropine is an alkaloid Further progression of an electrographic seizure is often
that blocks muscarinic cholinergic receptors. In elegant characterized by a set of depolarizing burst responses in which
analytical studies, Abe Black obtained evidence that the the extracellular responses are typically seen as a large local
higher-frequency hippocampal theta activity was related to negative wave carrying a number of high frequency spikes of
motor activity rather than to learning as such (Dalton and decreasing amplitude on its back, the so-called burst response
Black, 1968; Black, 1972; Black and Young, 1972) (see also (Ajmone-Marsan, 1961) at the onset of epileptiform dis-
Chapter 11). charges (von Euler et al., 1958; Kandel and Spencer, 1961b;
Theta activity is also of interest as a means of identifying Gloor et al., 1964). After a period with spontaneous high
hippocampal interneurons and their relation to behavior amplitude burst responses, the electrographic records dimin-
(see Chapters 8 and 11). In addition, other types of oscilla- ish and all cells cease ring because of a depolarization block.
tion at different frequencies, including beta and gamma Both in vivo and in vitro hippocampal preparations have been

Figure 213. Endre Grastyan (left), a Hungarian neuroscientist, was and named theta waves, measured the low hippocampal threshold
among the rst to investigate the role of the hippocampus in condi- for epileptic seizures, and gave the rst descriptions of visual corti-
tioned learning. (Source: Courtesy of Gyorgyi Buzsaki.) Richard cal neuronal responses to contours, contrasts, and colors. (Source:
Jung (right) had a number of discoveries to his credit. He discovered Courtesy of Volker Dietz.)
26 The Hippocampus Book

favorite tools in the search for cellular and network properties ing on the chorda tympani and lingual nerves (1926).
underlying the generation and spread of epileptiform activity Microelectrodes were developed during studies in cardiac
(see Chapter 16). Among these preparations are the epilepti- and skeletal muscle by Ralph Gerards group in Chicago
form activity provoked by kainic acid injections in CA3, (Ling and Gerard, 1949). In the spinal cord, Chandler McC.
GABA blockade by benzyl penicillin, and the effect of ouabain Brooks and John Eccles (1947) used microelectrodes to detect
intoxication. An important model for epileptiform activity is what they called a focal potential, a eld potential generated
the kindling phenomenon, described by Graham Goddard, by monosynaptic activation of motoneurons by volleys in
D.C. McIntyre, and C.K. Leech (1969), who elicited seizures by muscle afferent bers.
repeated low-strength stimulation of the amygdala. Despite The rst use of microelectrodes in the brain for extracellu-
the ease with which electrographic seizures may be recorded lar recording of single nerve cell discharges was in the hip-
from the hippocampus or the entorhinal area, these areas do pocampus by Birdsie Renshaw, Alexander Forbes and, Robert
not usually develop the kindling phenomenon. The role of the Morison (Renshaw et al., 1940). The researchers had greater
hippocampal formation in experimental and clinical epilepsy success detecting activity of single nerve cells in the hip-
is discussed in Chapter 16. pocampus than in neocortex, possibly because the latter tissue
was more depressed at the levels of pentobarbital anesthesia
used. In the hippocampus, these researchers recorded sponta-

neous and evoked discharges of what they called axon-like

2.6 Development of Methodological
spikes. Their description stated: what is evidently the dis-
Procedures for General Use
charge of the same unit - they recur repeatedly with the same
waveform (p. 90). When describing the individual elements
The hippocampal cortex has served as a test bench for the with a constant amplitude and predominantly negative polar-
development of a variety of general neuroscientic method- ity, they used what later became the classic critera for identi-
ologies. In many instances this has led to the realization that fying individual nerve cell discharges. Renshaw et al. localized
the apparently special features of hippocampal tissue is also such discharges to elements in the pyramidal layer and con-
found in other central nervous structures. For example, the cluded that the axon-like spikes were discharges of individ-
physiological properties of CA1 hippocampal pyramidal cells ual pyramidal cells.
have much in common with those of neocortical pyramidal
2.6.2 Pioneers of Intracellular Recording
Below are listed some areas where hippocampal neurobiol-
ogy has been instructive for neuroscience in general. The same experimental advantages that facilitated the analysis
First use of microelectrodes for extracellular neuronal of extracellular unit recording potentials also paved the way
studies for using the hippocampus for intracellular recording. The
Development of tetrodes for unit recording from behav- rst to succeed in this technically demanding enterprise were
ing animals Denise Albe-Fessard and Pierre Buser in 1952. They were sur-
Interpretation of eld synaptic potentials and popula- prised by the fact that nearly all penetrated cells showed a
tion spikes as tools for analysis of extracellular signals large, long depolarizing signal. Both its large amplitude and
Pioneering use of intracellular recording for central long duration made interpretation difficult. Initially, they took
nervous system neurons the response to be an EPSP. Later, it turned out to be an arti-
Isolated slices of cortical tissue for neuroscience studies factual recording of a large IPSP, which due to an electrode-
Development of histochemical methods for localization induced chloride infusion went in the depolarizing direction.
of neurotransmitters and receptor types They did not record EPSPs with their technique, however.
Transplantation studies The rst satisfactory intracellular recordings from cortical
Pharmacological studies of central neurons and neurons were made by Charles Phillips in 1956. He recorded
synapses from Betz cells in the precentral gyrus of cats and character-
Molecular biological analysis of synaptic function ized their response to both antidromic and orthodromic acti-
Formulation of computational models to explain ways vation. This too proved a technically difficult task, not least
in which neural networks can implement learning and because of the difficulty of nding monosynaptic afferent
memory bers to stimulate. Fortunately, the situation turned out to be
easier in hippocampal preparations. As described in Section
2.6.1 Hippocampus as a Test 2.3, Kandel, Spencer, and Brinley (1961) carried out the rst
Bed for Microelectrode Work comprehensive intracellular study of hippocampal pyramidal
cells. In addition to their analysis of EPSPs and IPSPs (see
Many of the standard methods used in neurobiology were above), they reported on prepotentials, which were short
developed in the peripheral or autonomous nervous systems depolarizing deections at the very start of the full action
or in muscular tissue. Recording from single nerve bers potential, interpreted as a sign of dendritically generated
was pioneered by Edgar Adrian and Yngve Zotterman work- action potentials. They also recorded major features associ-
Historical Perspective 27

ated with the transition from normal behavior to epileptiform In the hippocampal formation, the histological arrange-
behavior in the form of slow depolarization waves and the ment is highly favorable for eld potential studies of both
occurrence of giant depolarizations with superimposed burst orthodromic and antidromic activation (see Box 21). The
discharges (burst responses) (Kandel and Spencer, 1961a,b). dense packing of the cell bodies, the roughly parallel position
This was an essential step in epilepsy research. of the apical dendrites of hippocampal neurons, and the ease
In contrast to the wealth of inhibitory synaptic data, there with which they can be synchronously activated are three
was a surprising paucity of excitatory synaptic signals. This main reasons for the striking appearance of hippocampal eld
may be related to the barbiturate anesthesia with its facili- potentials.
tating effect on IPSPs, an effect also discovered in the hipp- The main advantage of using eld potentials is that an
ocampus by Roger Nicoll and Eccles (Nicoll et al., 1975). As extracellular potential recording may give an accurate index of
mentioned above in the discussion of IPSP and EPSP identi- synaptic activity with regard to amplitude, time, and polarity.
cation (see Section 2.3.1), intracellular recording from hip- First, the parallel orientation of a large number of principal
pocampal neurons was essential for the identication and cells may generate a eld potential of considerable magnitude.
localization of inhibitory synapses to the soma and of excita- The fEPSP has maximum negativity in the region with the
tory synapses to dendritic spines. highest concentration of activated excitatory synapses.
Neighboring activated synapses add their effect largely lin-
2.6.3 Tetrode Development early, justifying the use of the fEPSP amplitude as a measure
of synaptic strength. Conversely, synchronous and local acti-
Following the recording of single-cell activity in awake, behav- vation of inhibitory bers also gives rise to large currents, but
ing animals (OKeefe and Dostrovsky, 1971; Ranck, 1973) it in the opposite direction. The usefulness of the recording pro-
became clear that hippocampal cells signaled differently from, cedure coupled with its simplicity lies at the heart of the pop-
say, visual cortical cells. The latter could be classied as sensi- ularity of eld potential studies.
tive to contrast or color and with specic and repeatable pat- Brian Cragg and Lionel Hamlyn (1955) were the rst to
terns to a standardized stimulus delivered to a restricted part record signicant elements of the hippocampal eld poten-
of the visual eldthe cells receptive eld. In contrast, hip- tials. However, because they placed their stimulation electrode
pocampal cells showed a much lower rate of discharge, more very close to the recording site, the axonal conduction dis-
variability from one trial to the next, and a receptive eld tance was very short. Consequently, their records of the synap-
related to the aberrant concept of space (see Chapter 11). tic component of the eld potentials were brief and difficult to
Subsequently, it became clear that hippocampal neurons oper- isolate from the rest of the compound signal (action poten-
ated in ensembles and deciphering the full code might receive tial in their nomenclature). Nevertheless, they were able to
simultaneous recording from a number of neighboring cells. report the rst evidence of conduction along the apical den-
This led to the development of recording techniques with dou- drites and found that the minimal latency was associated with
ble electrodes, or stereotrodes (McNaughton et al., 1983), and the dendritic position of the activating synapses. With a com-
nally quadruple electrodes, or tetrodes (OKeefe and Recce, missural input, the shape and conduction of these action
1993; Wilson and McNaughton, 1993). Such electrode assem- potentials were similar to those found with close-range stim-
blies allow simultaneous recording of a large number of cells ulation (Cragg and Hamlyn, 1957).
in the freely moving and behaving animal by taking into The large compound action potential that followed stim-
account the constant shape and form of the potential gener- ulation at close range, later called the population spike,
ated by discharges of each of the active cells as seen by each of was rst noted by Cragg and Hamlyn (1955, 1957). Although
the four electrodes. The tetrode technique has revolutionized claims to dendritic conduction had been made on the basis
the study of hippcampal neuronal activity and spatial behav- of neocortical surface stimulation, the specialized hippocam-
ior. For a further account of this eld, see Chapter 11. pal histological arrangement made interpretation of the
records much more convincing. Today, a new and much more
2.6.4 Field Potential Analysis detailed view on dendritic conduction has emerged, largely
owing to whole cell patch recordings from dendrites. Again,
After the pioneering use of microelectrodes during the 1940s, much of the evidence for dendritic properties has been
several years passed before further progress was made in stud- obtained from hippocampal preparations (Spruston et al.,
ies of the hippocampal region using cell discharge recordings. 1995; Johnston et al., 1996; Stuart et al., 1997) (see details in
During the mid-1950s, however, several groups started to use Chapter 5).
eld potentials to understand the pattern of activation or Several other groups exploited the laminated synaptic
inhibition of hippocampal systems. arrangement to identify the eld potentials associated with
In an early study of eld potentials, Lorente de N (1947) restricted dendritic synapses. To make the hippocampus a
studied the signal sequence when a nerve volley traveled useful preparation for studying synaptic activation, it was nec-
antidromically into the hypoglossus motor nucleus. He essary to ascribe a standard extracellular signal to a standard-
offered the rst theoretical explanation for the generation of ized input. Recording from various dendritic positions,
such eld potentials. Andersen (1960) described how commissural bers to CA1
Box 21
Field potentials and current source density analysis

Field potentials are extracellular potentials recorded from groups of nerve cells in response to
synaptic or antidromic stimulation. In laminated structures such as the hippocampus, eld
potentials provide surprisingly detailed information on cellular activity.
A eld potential is generated by extracellular current owing across the tissue resistance
between the recording electrode and, in general, the ground electrode. Although measurable
extracellular voltages are generated by action potentials in a single neuron, and form the basis
of single unit recording, synaptic currents generated by single neurons are generally too small
to be detected. In the hippocampus, and other highly laminated structures such as the olfactory
cortex, the synchronous and localized currents generated by synaptic activation of a population
of pyramidal or granule cells gives rise to a characteristic and easily measured response called a
population or eld EPSP (fEPSP); with weak stimulation, the synaptic response is below
threshold for the generation of action potentials in the target cells, and a pure fEPSP is gener-
ated. With stronger stimulation, the cells discharge synchronously, giving rise to a population
spike which is superimposed on the rising phase of the fEPSP. Unlike the all-or-none action
potential generated by a single neuron, the population spike is a graded response; as stimulus
intensity increases, the number of neurons discharged becomes greater, and the population
spike becomes correspondingly larger.
In Box Fig. 21 stimulation of medial perforant path bers activates synapses in the middle
third of the molecular layer of the dentate gyrus (shaded gray). The eld responses recorded at
various positions along the soma and dendritic tree are shown on the left. Synaptically gener-
ated current ows into the dendrites in the activated region; inside the cells, current ows prox-
imally and distally away from the synaptic region, exiting where membrane area is greatest,
notably in the region of the soma. The current loop is completed extracellularly, with current
owing radially from distal and proximal sources towards the sink in the synaptically active
region. With a distal ground electrode, these synchronous extracellular currents give rise to
eld EPSPs that are negative in the region of current sinks and positive in regions of strong
current sources. Note that the current sink generated by synaptic activation is in the molecular
layer, while the passive source is in the cell body region and in distal dendrites; the converse sit-
uation obtains for the population spike; here, the active sink is in the cell body region while the
passive source is in the dendrites. The polarity of the fEPSP and the population spike are, as a
result, spatially out of phase.



Box Fig. 21. Field potentials. A. Field potentials generated by dentate granule cells in response
to stimulation of the medial perforant path. Activated synapses are located in the middle third
of the dendritic tree (gray band in B). In this region the eld potential (fEPSP) is negative,
reversing to a positive polarity near the cell body layer. The population spike (star) is negative
in the cell body layer, and positive in the dendritic region. (Source: Andersen et al., 1966a). C.
The direction of intracellular, extracellular and transmembrane current ow generated by
synaptic activation in an activated neuron.


Recording Distance from Border of


St.Pyramidale (m)



5 mV
A B 10 msec

Box Fig. 22. Current source density analysis. A. Field potentials evoked by weak activation of
bers projecting to stratum oriens in area CA1, evoking a pure synaptic response without a
superimposed population spike. B. CSD analysis reveals a sink (positive) in stratum oriens and
a source (negative) in the cell body layer and proximal dendrites. Arbitrary units. (Source:
Richardson et al., 1987).
A further renement called current source density (CSD) analysis allows the precise regions
of inward and outward current to be plotted. Box Fig 22A displays the fEPSPs recorded at dif-
ferent locations in area CA1 in vitro following weak synaptic activation. For a given latency, a
depth prole can be generated, showing how the amplitude and polarity of the response
changes with location.. By Ohms law, the current owing in the longitudinal direction, parallel
to the dendrites, is proportional to the rate of change of voltage in that direction:
IL  k.V/L
where IL is the longitudinal current density at a given point L along the longitudinal axis of the
cell, V is the potential at point L, and k is the conductivity of the extracellular space. Thus the
rst spatial derivative of the depth prole yields a plot of the longitudinal current as a function
of location at the selected latency. Current owing through the membrane at any point must be
equal to the rate at which longitudinal current changes at that point, i.e.
IM  -IL/L  -k.2V/L2
The second spatial derivative of the depth prole is thus proportional to the amplitude of the
membrane current density, Im. By repeating this analysis at a succession of latencies, the time
course of current sinks and sources at each location can be calculated. A full CSD analysis,
computed from the sequence of fEPSPs evoked at different locations along the dendritic tree by
weak stimulation of Schaffer commissural bers projecting to the basal dendrites of pyramidal
cells in area CA1 is shown in Box Fig. 22B. The voltage responses are on the left, and the cur-
rent source density analysis, revealing the development of sources and sinks over time, are plot-
ted on the right.
Field EPSPs have the same time course as the synaptic current that generates them, and are
therefore phase advanced with respect to intracellular EPSPs which are delayed by the time
required for the membrane current to charge membrane capacitance (Box Fig 23). Further
discussion of eld potential theory can be found in Stevens (1966), Rall and Shepherd (1968),
Nicholson and Freeman (1975) and Johnston and Wu (1994).

Intra -

Extra +

Box Fig. 23. The extracellular eld EPSP reects intracellular events, but is phase advanced
with respect to the intracellular EPSP. Recordings were made from the dentate gyrus in vivo
(Source: Lmo, 1971).

30 The Hippocampus Book

and CA3 gave a local negative eld potential in exactly the cholinesterases, and aldolase to lactic, malic, and glutamic
same strata where Theodor Blackstad (1956) had found that dehydrogenases. Lowrys group also determined the lipid con-
the relevant bers terminated. Such negative eld potentials tent and type as related to cortical layers. Third, it was the rst
were taken as a sign of excitatory synaptic activity and was instance when a neurochemist explicitly exploited a special
called a eld excitatory postsynaptic potential (fEPSP). Above histological arrangement for localizing biochemical elements,
a given strength the fEPSP was interrupted by a compound initially in the hippocampus and then in the cerebellum. His
action potential, called the population spike because it was group reported impressive homogeneity in a single stratum
interpreted as the synchronous discharge of a number of and often a considerable variation among various strata. In
pyramidal cells. This interpretation was supported by the particular, he noted the much higher ATP concentration in
observation that most of the synaptically activated unitary cell the dendritic areas than in the cell body layer.
discharges fell within the time envelope of the population Using a microdissection method, Storm-Mathisen and
spike (Andersen et al., 1971a). Fonnum (1972) were able to attain quantitative data for
Thus, the extracellular eld potentials can be used as a sen- GABA and glutamate concentrations in various hippocampal
sitive, quantitative measure of the intensity of excitatory and strata. The amount of GABA was particularly high in the
inhibitory synaptic effects and the efficacy of postsynaptic pyramidal layer, in accord with the concentration of
activation. Final conrmation of the eld potential interpreta- inhibitory boutons here. However, the high amount of GABA
tion given above came with intracellular recording of dentate even in the stratum lacunosum-moleculare pointed to new
granule cells (Andersen et al., 1966b; Lmo, 1971) and the use challenges.
of isolated slices, where intracellularly recorded EPSPs were Peter Lewis and Charles Shute (1967) extended the chemi-
associated with excitatory eld potentials in CA1 pyramidal cal dissection by performing the histochemical analysis
cells (Schwarzkroin, 1975; Andersen et al., 1980). directly on sections from the hippocampus. They showed that
the distribution of acetylcholinesterase was concentrated in
2.6.5 Histochemistry: Pioneered certain lamina, an approach that heralded a new approach to
in the Hippocampus chemical analysis of the CNS. This histochemical approach
also provided the groundwork for many of the studies on
We have become used to the hippocampus as a favorite prepa- synaptic sprouting that followed during the 1970s (see
ration for histological, electrophysiological, and behavioral Chapter 9).
studies. Less well known is its essential role in the develop- The later development of histochemical and immunohis-
ment of neurochemical methodology and histochemistry. tochemical studies have largely concentrated on microscopic
Once again, it was the packing of homogeneous cells into sections. For this approach, hippocampal tissue has also been
compact cell layers and the striking stratication of afferent a favorite choice. Among these developments, the electron
bers and synapses to specic parts of the dendritic tree of microsopic identication of bouton contents with antibodies
principal cells that appealed to the exploring scientist. raised against an aggregate of amino acids and bovine globu-
Until the middle of the twentieth century, virtually all neu- lin marked an entirely new, powerful approach (Storm-
rochemical analyses were conducted on homogenized tissue Mathisen et al., 1983). Thus, the hippocampus has played an
of relatively large samples from various parts of the brain. important role in the development of histochemical tech-
This procedure was thought to be necessary because of the niques for both fresh and xed tissue.
extreme complexity of central nervous tissue. In an attempt to
perform analysis of more discrete CNS elements, Oliver 2.6.6 Pharmacological Analysis
Lowry exploited the synaptic lamination of the hippocampus of Cellular Properties
to perform a detailed neurochemical dissection of cortical tis-
sue. He noted the following: The hippocampus has also been important for the develop-
ment of ideas in the neuropharmacology of synaptic trans-
Ammons horn is a region of the cerebral cortex which
mission in the CNS. Following the demonstration of
is organized in such a manner as to invite quantitative
glutamate sensitivity of spinal cord cells by David Curtis and
histochemical study.
Jeff Watkins (1960), Tim Biscoe and Donald Straughan (1966)
Enzyme concentrations and lipid type and concentration found that hippocampal pyramidal cells were sensitive to ion-
were measured for individual strata of the hippocampal cor- tophoretic application of glutamate. Together, these studies
tex. In a pioneering series of papers (Lowry et al. 1954a,b,c; suggested the radical (at the time) possibility that glutamate
1964), Lowry and his colleagues made several fundamental was an excitatory neurotransmitter in the CNS. Support came
advances. First, they introduced a number of methodological from an ingenious study on hippocampal slices in which
improvements. By dissecting freeze-dried tissue from thin Nadler et al. (1976) measured Ca2-dependent release of
slices of hippocampus under a microscope, they were able to aspartate and glutamate. Because the release of both com-
reduce the sample size to 5 to 10 g of tissue. Second, the pounds was grossly reduced after lesions of commissural or
accuracy of the enzymatic measurements greatly improved. entorhinal bers, the release is likely to have come from bou-
The enzymes of interest spanned a large range from acid and tons of the two fiber systems, thus supporting the idea that
alkaline phosphatases through adenosine triphosphatase, aspartate or glutamate could be excitatory transmitters.
Historical Perspective 31

Postsynaptic inhibition produced by Purkinje cells in the pocampus has been a focus of interest in epilepsy research
lateral vestibular, and deep cerebellar nuclei was rst shown to (Schwartzkroin, 1997).
be mediated by GABA by Obaka and colleagues (1967). As The hippocampal formation has proven to be particularly
mentioned above for hippocampal inhibition, David Curtis, vulnerable to various traumatic insults. Spielmeyer (1925)
John Felix, and Hugh McLellan (1970) showed that the large, noted that Sommers sector, which corresponds to the CA1
ubiquitous IPSPs in all principal cells of the hippocampus eld of the hippocampus, is a brain structure extremely vul-
could be blocked by localized application of bicuculline nerable to ischemic or hypoxic insults. This may be related to
methochloride, establishing that these IPSPs were therefore a lower level of mitochondrial oxidative enzymes in CA1
mediated by GABAA receptors. Thus, the studies of Curtis and pyramidal dendrites than in other hippocampal subregions
his colleagues extended the Obata et al. results to cortical (Davolio and Greenamyre, 1995; Kuroiwa et al., 1996). Early
inhibitory pathways. neuropathologists also noted that the hippocampal formation
is a major target for pathological changes in patients suffering
2.6.7 Development of Computational from senile dementia of the Alzheimers type (Alzheimer,
Models of Neural Networks 1909). These issues continue to challenge modern neuro-
science, as described in Chapter 16.
In a remarkable set of three papers, David Marr (1969, 1970,
1971) developed a theoretical proposal for the mode of
operation of three major components of the CNS. Arguably, REFERENCES
the triad inaugurated modern computational neuroscience.
Further description of this development is found in Chap- Adrian ED, Zotterman Y (1926) The impulses produced by sensory
ter 14. nerve-endings. Part 2. The response of a single end-organ. J
Marr started with the cerebellum, for which he proposed Physiol (Lond) 61:151171.
the main operation to be a comparison between an initial state Aggleton JP (1993) The contribution of the amygdala to normal and
of the motor system and the state after an elemental move- abnormal emotional states. Trends Neurosci 16:328333.
ment had taken place (Marr, 1969). The next article dealt with Ajmone-Marsan C (1961) Electrographic aspects of epileptic neu-
the neocortex and how the cortical matrix behaved during ronal aggregates. Epilepsia 2:2238.
perception and attention (Marr, 1970). The third paper con- Albe-Fessard D, Buser P (1952) tude de rponses neuroniques
laide de microlectrodes intrasomatiques. Rev Neurol (Paris)
tained a description of the mode of operation of the archicor-
tex (Marr, 1971), meaning the hippocampal formation. Basing
Alksne JF, Blackstad TW, Walberg F, White LE Jr (1966) Electron
his proposal on what was then known about its neural cyto- microscopy of axon degeneration: a valuable tool in experimen-
architecture, he hypothesized that the hippocampus per- tal neuroanatomy. Ergeb Anat Entwicklungsgesch 39:332.
formed a simple memory function which allowed the animal Allen WF (1940) Effect of ablating the frontal lobes, hippocampi and
to make use of previous experience to adapt its behavior. In occipito-parieto-temporal (excepting pyriform areas) lobes on
the process, he described how elemental neurophysiological positive and negative olfactory conditioned reexes. Am J
processes could perform certain basic mathematical proce- Physiol 132:8191.
dures. For example, inhibition exerted at a dendritic level Alzheimer A (1907) ber eine eigenartige Erkrankung der
would cause relatively little enhancement of the resting mem- Hirnrinde. Allg Z Psychiatr Psych Gerichtl Med (Berl) 64:
brane potential of the neuron, a procedure equivalent to sub- 146148.
Amaral DG, Insausti R, Cowan WM (1987) The entorhinal cortex of
traction. In contrast, inhibition at the soma level, produced by
the monkey. I. Cytoarchitectonic organization. J Comp Neurol
basket cells, would represent a much stronger change of the
membrane potential, removing a substantial part of the Andersen P (1960) Interhippocampal impulses. II. Apical dendritic
synaptic drive, equivalent to mathematical division. activation of CA1 neurons. Acta Physiol Scand 48:178208.
He also pointed to the recurrent axon collaterals of CA3 as Andersen P, Eccles JC, Lyning Y (1964a) Location of postsynaptic
a possible neural basis for attractor networks in which the inhibitory synapses on hippocampal pyramids. J Neurophysiol
excitatory interactions between networks of pyramidal cells 27:592607.
could form the basis of content-addressable memories and Andersen P, Eccles JC, Lyning Y (1964b) Pathway of postsynaptic
support retrieval of a complete memory from partial infor- inhibition in the hippocampus. J Neurophysiol 27:608619.
mation (see Section 5 in Chapter 14). Andersen P, Blackstad TW, Lmo T (1966a) Location and identica-
tion of excitatory synapses on hippocampal pyramidal cells. Exp
Brain Res 1:236248.
2.6.8 The Hippocampal Formation:
Andersen P, Holmqvist B, Voorhoeve PE (1966b) Excitatory synapses
A Test Bed for Several Types of Neural on hippocampal apical dendrites activated by entorhinal stimu-
Dysfunction and Neuropathology lation. Acta Physiol Scand 66:461472.
Andersen P, Bliss TVP, Skrede KK (1971a) Unit analysis of hip-
It is a well established clinical observation that epileptic pocampal population spikes. Exp Brain Res 13:208221.
patients exhibit a high frequency of temporal lobe foci, often Andersen P, Bliss TVP, Skrede KK (1971b) Lamellar organization of
coupled with the neuropathological nding of hippocampal hippocampal exitatory pathways. Exp Brain Res 13:222238.
sclerosis (Sommer, 1890; Gastaut, 1956). As a result, the hip- Andersen P, Bland BH, Skrede KK, Sveen O, Westgaard RH (1972)
32 The Hippocampus Book

Single unit discharges in brain slices maintained in vitro. Acta Broca P (1861a) Sur le principe des localisations crbrales. Bull Soc
Physiol Scand 84:1A2A. Anthropol 2:190204.
Andersen P, Sundberg, SH, Sveen O, Wigstrm H (1977) Specic Broca P (1861b) Perte de la parole, ramollissement chronique et
long-lasting potentiation of synaptic transmission in hip- destruction partielle du lobe antrieur gauche [sur le sige de la
pocampal slices. Nature 266:736737. facult du langage]. Bull Soc Anthropol 2:235238.
Andersen P, Silfvenius H, Sundberg SH, Sveen O (1980) A compari- Brodal A (1947) Hippocampus and the sense of smell. Brain
son of distal and proximal dendritic synapses on CA1 pyramids 70:179222.
in guinea-pig hippocampal slices in vitro. J Physiol 307:273299. Brodal A (1981) Neurological anatomy in relation to clinical medicine.
Aranzi GC (1564) De humano foetu opusculum. Bologna: Rubrii. New York: Oxford University Press.
Banker GA, Cowan WM (1977) Rat hippocampal neurons in Brodmann K (1909) Vergleichende Lokalisationslehre der Grosshirn-
dispersed cell culture. Brain Res 126:397425. rinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues.
Banker G, Goslin K (1991) Primary dissociated cell cultures of neu- Leipzig: Johann Ambrosius Barth Verlag.
ral tissue. In: Culturing nerve cells (Banker G, Goslin K, eds), pp Brooks CMcC, Eccles JC (1947) Electrical investigations of the
171. Cambridge, MA:MIT Press. monosynaptic pathway through the spinal cord. J Neurophysiol
Bard P (1934) On emotional expression after decortication with 10:251274.
some remarks on certain theoretical views. Part 1. Psychol Rev Brown S, Schfer EA (1888) An investigation into the function of the
4:309322; Part 2: 4:424449. occipital and temporal lobes of the monkeys brain. Philos Trans
Bekhterev V (1900) Demonstration eines Gehirns mit Zerstrung der R Soc B 179:303327.
vorderen und inneren Theile der Hirnrinde beider Bush G, Luu P, Posner MI (2000) Cognitive and emotional inuences
Schlafenlappen. Neurol Zentralbl 19:990991. in anterior cingulate cortex. Trends Cogn Sci 4:215222.
Berger H (1929) ber das Elektroenkephalogramm des Menschen. Cannon WB (1929) Bodily changes in pain, hunger, fear and rage. New
Arch Psychiatr Nervenkr 87:527570. York: Appleton-Century-Crofts.
Biscoe T, Straughan D (1966) Micro-electrophoretic studies of neu- Cohen NJ, Squire LR (1980) Preserved learning and retention of
rones in the cat hippocampus. J Physiol (Lond) 183:341359. pattern-analyzing skill in amnesia: dissociation of knowing how
Bishop S, Duncan J, Brett M, Lawrence AD (2004) Prefrontal cortical and knowing that. Science 210:207210.
function and anxiety: controlling attention to threat-related Corkin S, Amaral DG, Gonzalez RG, Johnson KA, Hyman BT (1997)
stimuli. Nat Neurosci 7:184188. H.M.s medial temporal lobe lesion: ndings from magnetic res-
Bjrklund A, Stenevi U (1981) In vivo evidence for a hippocampal onance imaging. J Neurosci 17:39643979.
adrenergic neuronotrophic factor specically released on septal Correll RE, Scoville WB (1967) Signicance of delay in the perform-
deafferentation. Brain Res 229:403428. ance of monkeys with medial temporal lobe resections. Exp
Bjrklund A, Stenevi U (1984) Intracerebral neural implants: neuro- Brain Res 4:8596.
logical replacement and reconstruction of damaged circuitries. Cotman C, Gentry C, Steward O (1977) Synaptic replacement in the
Annu Rev Neurosci 7:279308. dentate gyrus after unilateral entorhinal lesion: electron micro-
Bjorklund A, Johansson B, Stenevi U, Svendgaard NA (1975) Re- scopic analysis of the extent of replacement of synapses by the
establishment of functional connections by regenerating central remaining entorhinal cortex. J Neurocytol 6:455464.
adrenergic and cholinergic axons. Nature 53:446448. Cowan WM, Gottlieb DI, Hendrickson AE, Price JL, Woolsey TA
Bjorklund A, Stenevi U, Svendgaard N (1976) Growth of trans- (1972) The autoadiographic demonstration of axonal connec-
planted monoaminergic neurones into the adult hippocampus tions in the central nervous system. Brain Res 37:2151.
along the perforant path. Nature 262:787790. Cragg B, Hamlyn LH (1955) Action potentials of the pyrami-
Black AH (1972) The operant conditioning of central nervous system dal neurons of the hippocampus of the rabbit. J Physiol (Lond)
electrical activity. In: The psychology of learning and motivation, 129:608627.
vol 6 (Bower G, ed), pp 4795. San Diego: Academic Press. Cragg B, Hamlyn LH (1957) Some commissural and septal connec-
Black AH, Young GA (1972) Electrical activity of the hippocampus tions of the hippocampus in the rabbit: a combined histological
and cortex in dogs operantly trained to move and to hold still. J and electrical study. J Physiol (Lond) 135:460485.
Comp Physiol Psychol 79:128141. Curtis DR, Watkins JC (1960) The chemical excitation of spinal neu-
Blackstad TW (1956) Commissural connections of the hippocampal rones by certain acidic amino acids. J Physiol (Lond)
region in the rat, with special reference to their mode of termi- 150:656682.
nation. J Comp Neurol 105:417537. Curtis DR, Felix D, McLellan H (1970) GABA and hippocampal inhi-
Blackstad TW (1958) On the termination of some afferents to the bition. Br J Pharmacol 40:881883.
hippocampus and fascia dentata. Acta Anat (Basel) 35:202214. Dalton AJ, Black AH (1968) Hippocampal electrical activity during
Blackstad TW, Flood PR (1963) Ultrastructure of hippocampal axo- the operant conditioning of movement and refraining from
somatic synapses. Nature 198:542543. movement. Commun Behav Biol 2:267273.
Bliss TVP, Lmo T (1970) Plasticity in a monosynaptic cortical path- Davolio C, Greenamyre JT (1995) Selective vulnerability of the CA1
way. J Physiol (Lond) 207:61P. region of hippocampus to the indirect excitotoxic effects of mal-
Bliss TVP, Lmo T (1973) Long-lasting potentiation of synaptic onic acid. Neurosci Lett 192:2932.
transmission in the dentate gyrus of the anaesthetized rabbit. J Douglas RJ, Isaacson RL (1964) Hippocampal lesions and activity.
Physiol (Lond) 232:331356. Psychon Sci 1:187188.
Bliss TVP, Richards C (1971) Some experiments with in vitro hip- Dunnett SB, Low WC, Iversen SD, Stenevi U, Bjrklund A (1982)
pocampal slices. J Physiol (Lond) 214:79P. Septal transplants restore maze learning in rats with fornix-m-
Botvinick MM, Braver TS, Barch DM, Carter CS, Cohen JD (2001) bria lesions. Brain Res 251:335348.
Conflict monitoring and cognitive control. Psychol Rev Eccles JC, Llins R, Sasaki K (1966) The inhibitory interneurones
108:624652. within the cerebellar cortex. Exp Brain Res 1:116.
Historical Perspective 33

Ferrier D (1876) The functions of the brain. London: Smith, Elder. Hamlyn LH (1963) An electron microscope study of pyramidal neu-
Fink RP, Heimer L (1967) Two methods for selective silver impregna- rons in the Ammons horn of the rabbit. J Anat (Lond)
tion of degenerating axons and their synaptic endings in the 97:189201.
central nervous system. Brain Res 4:369374. Hayman LA, Rexer JL, Pavol MA, Strite D, Meyers CA (1998) Klver-
Gaffan D (1974a) Loss of recognition memory in rats with lesions of Bucy syndrome after selective damage of the amygdala and its
the fornix. Neuropsychologia 10:327341. cortical connections. J Neuropsychiatry Clin Neurosci
Gaffan D (1974b) Recognition impaired and association intact in the 10:354358.
memory of monkeys after transaction of the fornix. J Comp Hirano T, Best P, Olds J (1970) Units during habituation, discrimina-
Physiol Psychol 86:11001109. tion learning, and extinction. Electroencephalogr Clin
Ghwiler BH (1981) Organotypic monolayer cultures of nevous tis- Neurophysiol 28:127135.
sue. J Neurosci Methods 4:329342. Hirsh R (1974) The hippocampus and contextual retrieval of infor-
Ghwiler BH (1988) Organotypic cultures of neural tissue. Trends mation from memory: a theory. Behav Biol 12:421444.
Neurosci 11:484489. Hjorth Simonsen A (1973) Some intrinsic connections of the hip-
Ghwiler BH, Brown DA (1985) Functional innervation of cultured pocampus in the rat: an experimental analysis. J Comp Neurol
hippocampal neurons by cholinergic afferents from co-cultured 147:145161.
septal explants. Nature 313:577579. Holmes JE, Adey WR (1960) Electrical activity of the entorhinal cor-
Gastaut H (1956) Etude electroclinique des pisodes psychotiques tex during conditioned behavior. Am J Physiol 199:741744.
survenant en dehors des crises clinique chez les pileptiques. Rev Hounsgaard J, Yarom Y (1985) Intrinsic control of electroresponsive
Neurol (Paris) 94:587594. properties of transplanted mammalian brain neurons. Brain Res
Gerfen CR, Sawchenko PE (1984) An anterograde neuroanatomical 335:372376.
tracing method that shows the detailed morphology of neurons, Isaacson RL, Kimble DP (1972) Lesions of the limbic system:
their axons and terminals: immunohistochemical localization of their effects upon hypotheses and frustration. Behav Bull 7:
an axonally transported plant lectin Phaseolus vulgaris-leucoag- 767793.
glutinin. Brain Res 290:219238. Jackson JH (1865) In: Selected writings of John Hughlings Jackson
Glees P (1946) Terminal degeneration within the central nervous sys- (2 vols). (Taylor J, ed). London: Hodder & Stoughton, 1932.
tem as studied by a new silver method. J Neuropathol Exp Neurol Jackson JH, Beevor CE (1890) Case of tumor of the right temporo-
5:5459. sphenoidal lobe bearing on the localization of the sense of smell
Gloor P, Vera CL, Sperti L (1964) Electrophysiological studies of hip- and on the interpretation of a particular variety of epilepsy.
pocampal neurons. 3. Responses of hippocampal neurons to Brain 12:346357.
repetitive perforant path volleys. Electroencephalogr Clin Johnston D, Wu M-S (1994) Foundations of cellular neurophysiology.
Neurophysiol 17:353370. Cambridge, MA: Bradford Books, MIT Press.
Goddard GV, McIntyre DC, Leech CK (1969) A permanent change in Johnston D, Magee JC, Colbert CM, Cristie BR (1996) Active proper-
brain function resulting from daily electrical stimulation. Exp ties of neuronal dendrites. Annu Rev Neurosci 19:165186.
Neurol 25:295330. Jung R (1949) Hirnelektrische Untersuchungen ber den Elekt-
Goldwitz D, Cotman CW (1978) Induction of extensive mbrial rokrampf: Die Erregungsablufe in corticalen Hirnregionen bei
branching in the adult rat brain. Nature 275:6467. Katze und Hund. Arch Psychiatr Nervenkr 183:206244.
Golgi C (1886) Sulla na anatomia degli organi centrali del sistema Jung R, Kornmller AE (1938) Eine methodik der Ableitung
nervosa. Milan: Hoepli. lokalisierter Potentialschwankungen aus subcorticalen
Golgi C, Bentivogli M, Swanson L (2001) On the ne structure of the Hirngebieten. Arch Psychiatr Nervenkr 109:130.
pes hippocampi major. Brain Res Bull 54:461483. Translated Kaada BR, Pribram KH, Epstein JA (1949) Respiratory and vascular
from Golgi C (1886) responses in monkeys from temporal pole, insula, orbital sur-
Grafstein B (1971) Transneuronal transfer of radioactivity in the cen- face and cingulate. Fed Proc 8:8384.
tral nervous system. Science 172:177179. Kaada BR, Jansen J Jr, Andersen P (1953) Stimulation of the hip-
Grastyn E, Lissk K, Madarsz I, Donhoffer H (1959) Hippocampal pocampus and medial cortical areas in unanesthetized cats.
electrical activity during the development of conditioned Neurology 3:844857.
reexes. Electroencephalogr Clin Neurophysiol 11:409430. Kaada BR, Rasmussen EW, Kveim O (1961) Effects of hippocampal
Gray EG (1959) Axo-somatic and axo-dendritic synapses of the cere- lesions on maze learning and retention in rats. Exp Neurol
bral cortex: an electron microscope study. J Anat (Lond) 3:333355.
93:420433. Kandel ER, Spencer WA (1961a) Electrophysiology of hippocampal
Gray JA (1982) The neuropsychology of anxiety: an enquiry into the neurons. II. After-potentials and repetitive ring. J Neurophysiol
functions of the septo-hippocampal system. New York: Oxford 24:243259.
University Press. Kandel ER, Spencer A (1961b) Excitation and inhibition of
Gray JA (2001) Emotional modulation of cognitive control: single pyramidal cells during hippocampal seizure. Exp Neurol
approach-withdrawal states double-dissociate spatial from ver- 4:162179.
bal two-back task performance. J Exp Psychol Gen 130:436452. Kandel ER, Spencer WA, Brinley FJ (1961) Electrophysiology of hip-
Green JD, Adey WR (1956) Electrophysiological studies of hip- pocampal neurons. I. Sequential invasion and synaptic organi-
pocampal connections and excitability. Electroencephalogr Clin zation. J Neurophysiol 24:225242.
Neurophysiol 8:245263. Kerns JG, Cohen JD, MacDonald AW, Cho RY, Stenger VA, Carter CS
Green JD, Arduini AA (1954) Hippocampal electrical activity in (2004) Anterior cingulate conict monitoring and adjustments
arousal. J Neurophysiol 17:533557. in control. Science 303:10231026.
Gross CG (1993) Huxley versus Owen: the hippocampus minor and Kimble DP (1963) The effects of bilateral hippocampal lesions in
evolution. Trends Neurosci 16:493498. rats. J Comp Physiol Psychol 56:273283.
34 The Hippocampus Book

Kimble DP (1968) Hippocampus and internal inhibition. Psychol ral activity (Cohen MM, Snider RS, eds), pp. 178191. New
Bull 70:285295. York: Harper & Row.
Klver H, Bucy PC (1937) Psychic blindness and other symptoms fol- Lynch G, Matthews DA, Mosko S, Parks T, Cotman C (1972) Induced
lowing bilateral temporal lobectomy in rhesus monkeys. Am J acetylcholinesterase-rich layer in rat dentate gyrus following
Physiol 119:352353. entorhinal lesions. Brain Res 42:311318.
Korsakov SS (1889) tude mdico-psychologique sur une forme des McIlwain H, Snyder SH (1970) Stimulation of piriform and neo-cor-
maladies de mmoire. Rev Philos 28:501530. tical tissues in an in vitro ow-system: metabolic properties and
Korsakov SS (1890) Eine psychische Strung kombiniert mit multi- release of putative transmitters. J Neurochem 17:521530.
plen Neuritis. Allg Z Psychiatr 46:475485. McNaughton BL, OKeefe J, Barnes CA (1983) The stereotrode: a new
Kramis R, Vanderwolf CH, Bland BH (1975) Two types of hip- technique for simultaneous isolation of several single units in
pocampal rhythmical slow activity in both the rabbit and the the central nervous system from multiple unit records. J
rat: relations to behavior and effects of atropine, diethyl ether, Neurosci Methods 8:391397.
urethane, and pentobarbital. Exp Neurol 49:5885. Marchi V, Algeri EG (1886) Sulle degenerazioni discendenti consecu-
Kristensson K, Olsson Y (1971) Retrograde axonal transport of pro- tive a lesioni in diverse zone della corteccia cerebrale. Riv Sper
tein. Brain Res 29:4347. Freniatr Med Leg Alien Ment 14:149.
Kuroiwa T, Terakado M, Yamaguchi T, Endo S, Ueki M, Okeda R Marr D (1969) A theory of cerebellum. J Physiol (Lond) 202:437470.
(1996) The pyramidal layer of sector CA1 shows the lowest hip- Marr D (1970) A theory for cerebral neocortex. Proc R Soc B
pocampal succinate dehydrogenase activity in normal and 176:161234.
postischemic gerbils. Neurosci Lett 206:117120. Marr D (1971) Simple memory: a theory for archicortex. Philos Trans
Kveim O, Setekleiv J, Kaada BR (1964) Differential effects of hip- R Soc B 262:2381.
pocampal lesions on maze and passive avoidance learning in Matthews DA, Cotman C, Lynch G (1976) An electron microscopic
rats. Exp Neurol 9:5972. study of lesion-induced synaptogenesis in the dentate gyrus of
Larrabee MG, Bronk DW (1939) Prolonged facilitation of synaptic the adult rat. I. Magnitude and time course of degeneration.
excitation in sympathetic ganglia. J Neurophysiol 10:139154. Brain Res 115:121.
Lasek R (1975) Axonal transport and the use of intracellular markers Mishkin M (1954) Visual discrimination performance following par-
in neuroanatomical investigations. Fed Proc 34:16031611. tial ablations of the temporal lobe. II. Ventral surface vs. hip-
Lewis PR, Shute CCD (1967) The cholinergic limbic system: projec- pocampus. J Comp Physiol Psychol 47:187193.
tion to the hippocampal formation, medial cortex, nuclei of the Mishkin M, Delacour J (1975) An analysis of short-term visual mem-
ascending cholinergic reticular system and the subfornical ory in the monkey. J Exp Psychol Anim Behav Process 1:326334.
organ and supra-optic crest. Brain 90:521537. Mosko S, Lynch G, Cotman CW (1973) The distribution of septal
Li C-L, McIlwain H (1957) Maintenance of resting membrane poten- projections to the hippocampus of the rat. J Comp Neurol 152:
tials in slices of mammalian cerebral cortex and other tissues in 163174.
vitro. J Physiol (Lond) 139:178190. Nadel L, OKeefe J (1974) The hippocampus in pieces and patches: an
Li XG, Somogyi P, Ylinen A, Buzsaki G (1994) The hippocampal CA3 essay on modes of explanation in physiological psychology. In:
network: an in vivo intracellular labeling study. J Comp Neurol Essays on the nervous system: a festschrift for Prof JZ Young
339:181208. (Bellairs R, Gray EG, eds), pp 367390. Oxford: Clarendon
Ling G, Gerard RW (1949) The normal membrane potential of frog Press.
sartorius bers. J Cell Physiol 34:383396. Nadel L, OKeefe J, Black AH (1975) Slam on the brakes: a critique of
Lloyd DPC (1949) Post-tetanic potentiation of response in monosy- Altmann, Brunner and Bayers response inhibition model of
naptic pathways of the spinal cord. J Gen Physiol 33:147170. hippocampal function. Behav Biol 14:151162.
Lmo T (1966) Frequency potentiation of excitatory synaptic activ- Nadler JV, Vaca KW, White WF, Lynch GS, Cotman CW (1976)
ity in the dentate area of the hippocampal formation. Acta Aspartate and glutamate as possible transmitters of excitatory
Physiol Scand 68(Suppl 277):128. hippocampal afferents. Nature 260:538540.
Lmo T (1971) Patterns of activation in a monosynaptic cortical Nauta WJH (1950) ber die sogenannte terminale Degeneration in
pathway: the perforant path input to the dentate area of the hip- der Zentralnervensystem und ihre Darstellung durch
pocampal formation. Exp Brain Res 12:1845. Silberimprgnation. Schweiz Arch Neurol Psychiatr 66:353376.
Lorente de N R (1934) Studies on the structure of the cerebral cor- Nauta WJH, Gygax PA (1954) Silver impregnation of degenerating
tex. II. Continuation of the study of the ammonic system. J axons in the central nervous system: a modied technic. Stain
Psychol Neurol (Lpz) 46:113177. Technol 29:9193.
Lorente de N (1947) Action potential of the motoneurons of the Nicoll RA, Eccles JC, Oshima T, Rubia F (1975) Prolongation of hip-
hypoglossus nucleus. J Cell Comp Physiol 29:207288. pocampal inhibitory postsynaptic potentials by barbiturates.
Lowry OH, Roberts NR, Leiner KY, Wu M-L, Farr AL (1954a) The Nature 258:625627.
quantitative histochemistry of brain. I. Chemical methods. J Obata K, Ito M, Ochi R, Sato N (1967) Pharmacological properties of
Biol Chem 207:117. the postsynaptic inhibition of the Purkinje cell axons and the
Lowry OH, Roberts NR, Wu M-L, Hixon WS, Crawford EJ (1954b) action of -aminobutyric acid on Deiters neurons. Exp Brain
The quantitative histochemistry of brain. II. Enzyme measure- Res 4:4357.
ments. J Biol Chem 207:1927. OKeefe J, Dostrovsky J (1971) The hippocampus as a spatial map:
Lowry OH, Roberts NR, Leiner KY, Wu M-L, Farr AL, Albers RW preliminary evidence from unit activity in the freely-moving
(1954c) The quantitative histochemistry of brain. III. Ammons rat. Brain Res 34:171175.
horn. J Biol Chem 207:3949. OKeefe J, Recce ML (1993) Phase relationship between hippocampal
Lowry O (1964) In: Morphological and biochemical correlates of neu- place units and the EEG theta rhythm. Hippocampus 3:317330.
Historical Perspective 35

Olton DS, Walker JA, Gage FH (1978) Hippocampal connections and Schreiner L, Kling A (1956) Rhinencephalon and behavior. Am J
spatial discrimination. Brain Res 139:295308. Physiol 184:486490.
Orbach J, Milner B, Rasmussen T (1960) Learning and retention in Schwartzkroin P (1975) Characteristics of CA1 neurons recorded
monkeys after amygdala-hippocampus resection. Arch Neurol intracellularly in the hippocampal in vitro slice preparation.
3:230235. Brain Res 85:423436.
Palay SL, Palade GE (1955) The ne structure of neurons. J Biophys Schwartzkroin PA (1997) Origins of the epileptic state. Epilepsia
Biochem Cytol 1:6988. 38:853858.
Papez JW (1937) A proposed mechanism of emotion. Arch Neurol Schwartzkroin P, Andersen P (1975) Glutamic acid sensitivity of den-
Psychiatry 38:725743. drites in hippocampal slices in vitro. Adv Neurol 12:4551.
Peneld W, Erickson TC (1941) Epilepsy and cerebral localization: a Schwartzkroin P, Wester K (1975) Long-lasting facilitation of a
study of the mechanism, treatment, and prevention of epileptic synaptic potential following tetanization in the in vitro hip-
seizures. Springeld, IL: Charles C Thomas. pocampal slice. Brain Res 89:107119.
Phillips CG (1956) Intracellular records from Betz cells in the cat. Q Scoville W, Milner B (1957) Loss of recent memory after bilateral
J Exp Physiol 41:5869. hippocampal lesions. J Neurol Neurosurg Psychiatry 20:1121.
Raisman G (1969) Neuronal plasticity in the septal nuclei of the adult Segal M, Bloom FE (1974a) The action of norepinephrine in the rat
rat. Brain Res 14:2548. hippocampus. I. Iontophoretic studies. Brain Res 72:7997.
Raisman G, Field PM (1973) A quantitative investigation of the Segal M, Bloom FE (1974b) The action of norepinephrine in the rat
development of collateral reinnervation after partial deaf- hippocampus. II. Activation of the input pathway. Brain Res
ferentation of the septal nuclei. Brain Res 50:241264. 72:99114.
Raisman G, Field P (1990) Synapse formation in the adult brain after Semon R (1908) The mneme. English translation: London: 1908,
lesions and after transplantation of embryonic tissue. J Exp Biol 1911, and 1921. First German edition: 1904.
153:277287. Shapiro ML, Simon DK, Olton DS, Gage FH III, Nilsson O, Bjorklund
Raisman G, Cowan WM, Powell TPS (1965) The extrinsic, com- A (1989) Intrahippocampal grafts of fetal basal forebrain tissue
missural and association bres of the hippocampus. Brain alter place elds in the hippocampus of rats with mbria-fornix
88:963998. lesions. Neuroscience 32:118.
Raisman G, Cowan WM, Powell TPS (1966) An experimental analysis Shepherd GM (1972) The neuron doctrine: a revision of functional
of the efferent projection of the hippocampus. Brain 89:83108. concepts. Yale J Biol Med 45:584599.
Rall W, Shepherd GM (1968) Theoretical reconstruction of eld Shepherd MG, Raastad M, Andersen P (2002) General and vari-
potentials and dendrodendritic synaptic interactions in olfactory able features of varicosity spacing along unmyelinated axons in
bulb. J Neurophysiol 31:884915. the hippocampus and cerebellum. Proc Natl Acad Sci USA 99:
Ramon y Cajal S (1893) Estructura del asta de Ammon. Anal Sociedad 63406345
Espaol Historia Natural, vol 22. Translated to German by von Sherrington CS (1897) The central nervous system. In: A textbook of
Klliker A (1893). In Zeitschr Wiss Zool 56. physiology (Foster M, ed), vol 3, 7th ed. London: Macmillan.
Ramon y Cajal S (1911) Histologie du systme nerveux de lhomme et Shipley MT, Adamek GD (1984) The connections of the mouse olfac-
des vertbrs. Paris: A. Maloine. Reprinted (1955) Madrid: tory bulb: a study using orthograde and retrograde transport of
Instituto Ramon y Cajal. wheat germ agglutinin conjugated to horseradish peroxidase.
Ranck JB Jr (1973) Studies on single neurons in dorsal hippocam- Brain Res Bull 12:669688.
pal formation and septum in unrestrained rats. Exp Neurol 41: Skrede KK, Westgaard RH (1971) The transverse hippocampal slice:
461455. a well-dened cortical structure maintained in vitro. Brain Res
Renshaw B, Forbes A, Morison BR (1940) Activity of isocortex 35:589593.
and hippocampus: electrical studies with micro-electrodes. Sloan N, Jasper H (1950) Studies of the regulatory functions of
J Neurophysiol 3:74105. the limbic cortex. Electroencephalogr Clin Neurophysiol
Ribot T (1881) Les maladies de la mmoire. Paris: Ballire. 2:317327.
Richards CD, McIlwain H (1967) Electrical responses in brain sam- Sommer W (1890) Erkrankung des Ammonshorns als aetiologisches
ples. Nature 215:704707. Moment der Epilepsie. Arch Psychiatr Nervenkrank 10:631675.
Richardson TL, Turner RW, Miller JJ (1987) Action-potential dis- Spencer WA, Thompson RF, Neilson DR Jr (1966) Response decre-
charge in hippocampal CA1 pyramidal neuron: current source- ment of the exion reex in the acute spinal cat and transient
density analysis. J Neurophysiol 58:981996. restoration by strong stimuli. J Neurophysiol 29:221239.
Ries FA, Langworthy OR (1937) A study of the surface structure of Spielmeyer W (1925) Zur Pathogenese rtlich elektiver Gehirnvern-
the brain of the whale (Balaenoptera physalus and Physeter derungen. Z Ges Neurol Psychiatr 99:756776.
catodon). J Comp Neurol 68:147. Spruston N, Jonas P, Sakmann B (1995) Dendritic glutamate receptor
Rose M (1935) Cytoarchitektonik und Myeloarchitektonik der channels in rat hippocampal CA3 and CA1 pyramidal neurons.
Grosshirnsrinde. In: Handbuch der Neurologie (Bumke O, J Physiol (Lond) 482:325352.
Foerster O, eds), vol 1, pp 588778. Berlin. Squire LR (1998) Memory systems. C R Acad Sci III 321:153156.
Sagar JH, Cohen NJ, Corkin S, Growdon JH (1985) Dissociations Squire LR (2004) Memory systems of the brain: a brief history and
among processes in remote memory. Ann NY Acad Sci current perspective. Neurobiol Learn Mem 82:171177.
444:533535. Stevens CF (1966) Neurophysiology: a primer. New York: Wlley.
Sala L (1891) Zur Anatomie des grossen Seepferdfusses. Zeitschr Wiss Steward O, Cotman C, Lynch GS (1974) Growth of a new ber pro-
Zool 52. jection in the brain of adult rats: reinnervation of the dentate
Schaffer K (1892) Beitrag zur Histologie der Ammonshorn-forma- gyrus by the contralateral entorhinal cortex following ipsilateral
tion. Arch Mikr Anat 39:611632. entorhinal lesions. Exp Brain Res 20:4566.
36 The Hippocampus Book

Storm-Mathisen J, Fonnum F (1972) Localization of transmitter can- Von Gudden JBA (1896) Klinische und anatomische Beitrge zur
didates in the hippocampal region. Prog Brain Res 36:4158. Kenntniss der multiplen Alcoholneuritis nebst Bemerkungen
Storm-Mathisen J, Leknes AK, Bore AT, Vaaland JL, Edminson ber die Regenerationsvorgnge in peripheren Nervensystem.
P, Haug FM, Ottersen OP (1983) First visualization of glutamate Arch Psychiatr Nervenkr 28:643741.
and GABA in neurones by immunocytochemistry. Nature 301: Weiskrantz L (1956) Behavioral changes associated with ablation of
517520. the amygdaloid complex in monkeys. J Comp Physiol Psychol 49:
Stuart G, Spruston N, Sakmann B, Husser M (1997) Action poten- 381391.
tial initiation and backpropagation in neurons of the mam- Wernicke C (1881) Lehrbuch der Gehirnkrankheiten, vol 2. Berlin:
malian CNS. Trends Neurosci 20:125131. Theodore Fischer.
Swanson LW, Cowan, WM (1977) An autoradiographic study of the Wilson MA, McNaughton BL (1993) Dynamics of the hippocampal
organization of the efferent connections of the hippocampal ensemble code for space. Science 261:10551058.
formation in the rat. J Comp Neurol 172:4984. Winograd T (1975) Frame representations and the declarative/
Tulving E (1972) Episodic and semantic memory. In: Organization procedural controversy. In: Representation and understanding
and memory (Tulving E, Donaldson W, eds), pp 382403. San (Bobrow DG, Collins A, eds), pp 185210. San Diego: Academic
Diego: Academic Press. Press.
Vanderwolf CH (1969) Hippocampal electrical activity and volun- Yamamoto C, McIlwain H (1966a) Potentials evoked in vitro in
tary movement in the rat. Electroencephalogr Clin Neurophysiol preparations from the mammalian brain. Nature 210:
26:407418. 10551056.
Vinogradova O, Semyonova TP, Konovalov VP (1970) Trace phe- Yamamoto C, McIlwain H (1966b) Electrical activities in thin sec-
nomena in single neurons of hippocampus and mammillary tions from the mammalian brain maintained in chemically
bodies. In: Biology of memory (Pribram KH, Broadbent DE, dened media in vitro. J Neurochem 13:13331343.
eds), pp 191222. San Diego: Academic Press. Zimmer J, Ghwiler BH (1984) Cellular and connective organization
Von Euler C, Green JD, Ricci G (1958) The role of hippocampal den- of slice cultures of the rat hippocampus and fascia dentata. J
drites in evoked responses and after-discharges. Acta Physiol Comp Neurol 228:432446.
Scand 42:87111.
3 David Amaral and Pierre Lavenex

Hippocampal Neuroanatomy

century of neuroanatomical study and literally tens of thou-

3.1 Overview sands of research articles on the hippocampus, there is still no
consensus concerning certain facets of its nomenclature.
Where is the hippocampal formation and what does it look Hippocampal researchers tend to follow one of several
like? What types of neurons are located in this group of struc- implicit views about what the hippocampus is and what it is
tures, and what types of connections do they form? What is not. The view adopted in this book is that the hippocampus is
the difference between the hippocampus and the hippocam- one of several related brain regions that together comprise a
pal formation? Does the hippocampus look similar in the rat, functional system called the hippocampal formation.
monkey, and human brains? Are other components of the The hippocampus proper has three subdivisions: CA3,
hippocampal formation similar across species? Are the princi- CA2, and CA1. (CA comes from cornu ammonis; the deriva-
ples of neuroanatomical connectivity similar or different from tion of these terms is discussed shortly.) The other regions of
those found in other brain regions? These are only some of the the hippocampal formation include the dentate gyrus, subicu-
neuroanatomical issues that are explored in this chapter. The lum, presubiculum, parasubiculum, and entorhinal cortex.
elegant neuronal architecture of the hippocampal formation Thus, although the title for this book is The Hippocampus
and the simplicity and orderliness of its major connections Book, a more correct, albeit less melodious, title would have
have been seductive features to neuroscientists for decades. been The Hippocampal Formation Book. The rationale for
Yet, there are pragmatic reasons why a working knowledge of grouping these cytoarchitectonically distinct brain regions
its neuroanatomy is important. under the rubric hippocampal formation is developed
First, it is likely that certain peculiarities of the neural throughout this chapter.
organization of the hippocampus, such as its highly associa-
tional intrinsic connections, provide important clues to its 3.1.2 Similarities and Differences
function(s). Second, the design of functional, electrophysio- Between the Hippocampal Formation
logical, pharmacological, and behavioral studies require and other Cortical Areas
increasingly sophisticated knowledge of its boundaries, cell
types, connections, and chemical neuroanatomy. This chapter In some ways, such as the occurrence of large, pyramid-
lays the structural foundations for discussions of the func- shaped projection neurons and smaller interneurons, the neu-
tional organization of the hippocampal formation explored ral organization of some portions of the hippocampal
later in the book. formation resembles other cortical regions. Yet in important
ways, such as the largely unidirectional passage of information
3.1.1 Hippocampus: Part of a Functional Brain through intrahippocampal circuits and the highly distributed
System Called the Hippocampal Formation three-dimensional organization of intrinsic associational con-
nections, the neuroanatomy of the hippocampal formation is
One of the most captivating features of the hippocampus is its unique. It has often been touted as a heuristically simple
neuroanatomy. The relatively simple organization of its prin- model of neocortical organization, but this perspective
cipal cell layers coupled with the highly organized laminar dis- demeans both the hippocampal formation and the neocortex.
tribution of many of its inputs has encouraged its use as a It is more protable to highlight and investigate the distinctive
model system for modern neurobiology. Despite more than a neuroanatomical features of the hippocampal formation as

38 The Hippocampus Book

potential clues to its particular function(s) and the mecha- back to region A. As rst described by Ramn y Cajal (1893),
nisms by which these functions are realized. this is clearly not the case for the connections that link the var-
Our understanding of hippocampal neuroanatomy leads ious parts of the hippocampal formation (Fig. 31). The
to the prediction that whatever processes this group of struc- entorhinal cortex can, for convenience, be considered the rst
tures carries out they are likely to be quite different from those step in the intrinsic hippocampal circuit. The logic behind this
performed in other cortical regions. The hippocampal forma- is developed later in the chapter, but the priority afforded to
tion, for example, is one of only a few brain regions that the entorhinal cortex is based on the fact that much of the
receive highly processed, multimodal sensory information neocortical input reaching the hippocampal formation does
from a variety of neocortical sources. Moreover, its own sys- so through the entorhinal cortex. Cells in the supercial layers
tem of widely distributed intrinsic neuronal networks is ide- of the entorhinal cortex give rise to axons that project, among
ally suited for further mixing or comparing this information. other destinations, to the dentate gyrus. The projections from
This ability to integrate information from all sensory modali- the entorhinal cortex to the dentate gyrus form part of the
ties may thus be a unique attribute of the hippocampus con- major hippocampal input pathway called the perforant path.
ferred by the highly convergent-divergent organization of its Although the entorhinal cortex provides the major input to
connections. the dentate gyrus, the dentate gyrus does not project back to
the entorhinal cortex. This pathway is therefore nonrecipro-
3.1.3 Hippocampal Formation: With A Unique cated, or unidirectional.
Set of Unidirectional, Excitatory Pathways Likewise, the principal cells of the dentate gyrus, the gran-
ule cells, give rise to axons called mossy bers that connect
A common organizational feature of connections between with pyramidal cells of the CA3 eld of the hippocampus. The
regions of the neocortex is that they are largely reciprocal CA3 cells, however, do not project back to the granule cells.
(Felleman and Van Essen, 1991). If cortical region A projects The pyramidal cells of CA3, in turn, are the source of the
to cortical region B, region B often sends a return projection major input to the CA1 hippocampal eld (the Schaffer col-

Figure 31. The hippocampal formation.

A. Neurons in layer II of the entorhinal
cortex project to the dentate gyrus and the
CA3 eld of the hippocampus proper via
the perforant pathway. Neurons in layer III
of the entorhinal cortex project to the CA1
eld of the hippocampus and the subicu-
lum via the perforant and alvear pathways
(see text for a detailed description). The
granule cells of the dentate gyrus project
to the CA3 eld of the hippocampus via
mossy ber projections. Pyramidal neu-
rons in the CA3 eld of the hippocampus
project to CA1 via Schaffer collaterals.
Pyramidal cells in CA1 project to the
subiculum. Both CA1 and the subiculum
project back to the deep layers of the
entorhinal cortex. B. Projections along the
transverse axis of the hippocampal forma-
tion; the dentate gyrus is located proxi-
mally and the entorhinal cortex distally.
Hippocampal Neuroanatomy 39

lateral axons). Following the pattern of its predecessors, CA1 (Amaral et al., 1984). Another example is the more complex
does not project back to CA3. The CA1 eld of the hip- organization of the primate entorhinal cortex, which appears
pocampus then projects unidirectionally to the subiculum, to be associated with stronger interconnections with the asso-
providing its major excitatory input. Again, the subiculum ciational areas of the neocortex, which are more developed in
does not project back to CA1. primates. Myriad other subtle species differences (e.g., the
Once one reaches CA1 and the subiculum, the pattern of chemical neuroanatomy of the hippocampal formation) have
intrinsic connections begins to become somewhat more elab- been noted in the literature. Certain neuroanatomical differ-
orate. CA1, for example, projects not only to the subiculum ences have even been described in different strains of mice
but also to the entorhinal cortex. Furthermore, whereas the that seem too subtle to be worthy of note yet could be of enor-
subiculum does project to the presubiculum and the para- mous practical importance given the current interest in the
subiculum, its more prominent cortical projection is directed use of transgenic techniques. A major future challenge is to
to the entorhinal cortex. Through these connections both CA1 determine whether and how these structural and neurochem-
and the subiculum close the hippocampal processing loop ical alterations affect the functioning of the hippocampal for-
that begins in the supercial layers of the entorhinal cortex mation.
and ends in its deep layers. Although this cursory survey of the Thus, to address the question posed in the title of this sec-
intrinsic connections of the hippocampal formation leaves tionis the hippocampus of humans and animals same or
out many of the facts that make the system somewhat more differentthe answer is both.
complex, it does serve to emphasize that the hippocampal for-
mation is organized in a fashion that is distinctly different 3.1.5 Synopsis of the Chapter
from most other cortical areas.
Our rationale for focusing on the rodent (mainly the rat) hip-
3.1.4 Hippocampus of Humans pocampal formation in the rst half of this chapter is that
and Animals: Same or Different? much of the available neuroanatomical data have been derived
from studies carried out in this animal and because of the
Once one has gained a familiarity with the neuroanatomical prominent role the rat plays in current functional analyses of
appearance of the rat hippocampal formation, it is not diffi- the hippocampal formation. Unfortunately, far less work has
cult to identify each of the major subdivisions in the monkey been carried out in the mouse, the clear choice for molecular
or human hippocampal formation (Fig. 32). Although the biology studies. It is widely presumed that the neuroanatomy
volume of the hippocampus is about 10 times larger in mon- of the mouse hippocampal formation is similar to that of the
keys than in rats and 100 times larger in humans than in rats, ratthough some additional conrmation would certainly be
the basic hippocampal architecture is common to all three welcome. We next compare and contrast the picture of hip-
species. Yet there are some striking species differences. The pocampal anatomy obtained from studies of the rat with that
compact pyramidal cell layer in the CA1 region of the rat, for of the monkey. We also describe the extensive cortical connec-
example, becomes thicker and more heterogeneous in the tions of the entorhinal cortex in the macaque monkey and the
monkey and human. Whereas this layer is only about 5 cells emerging prominence of adjacent related brain regions, such
thick in the rat, it can be more than 30 cells thick in the as the perirhinal and parahippocampal cortices. We then
human. The other region that demonstrates striking species move on to compare information on the monkey hippocam-
differences is the entorhinal cortex. In the rat, the entorhinal pal formation with what is known about the human hip-
cortex is typically divided into two main, cytoarchitectonically pocampal formation. The chapter concludes with a summary
distinct subdivisions. In the monkey there are seven subdivi- of the principles of hippocampal intrinsic circuitry and a dis-
sions, and in the human brain the classical cytoarchitectoni- cussion of how they may govern the ow of information
cists dened as many as 27 subdivisions (although recent through the hippocampal formation.
descriptions recognize only 8 subdivisions, similar to what is
observed in monkeys). Thus, despite the fact that the hip-
pocampal formation is often portrayed as a phylogenetically
primitive brain region, it nonetheless demonstrates substan- 3.2 Historical Overview of Hippocampal
tial species differences. NomenclatureWhats in a Name?
Although the patterns of connectivity appear to be gener-
ally similar in the rodent and primate brains, there are again The hippocampal formation has been prodded and sliced and
striking species differences. One example is the organization stained by anatomists for nearly 400 years. Like many brain
of the commissural connections of the dentate gyrus. In the regions, it has fallen victim to the imposition of various and
rat, there is a massive commissural system that provides nearly often confusing terminologies to describe its gross anatomical
one-sixth of the excitatory input to the dentate gyrus and histological structure. The term hippocampus (derived
(Raisman, 1965; Gottlieb and Cowan, 1973). In the macaque from the Greek word for sea horse) was rst coined during the
monkey and presumably in humans, however, commissural sixteenth century by the anatomist Arantius (1587), who
connections in the dentate gyrus are almost entirely absent considered the three-dimensional form of the human hip-
40 The Hippocampus Book

Figure 32. Nissl-stained sections and line drawings illustrating the general organization and
similarities of the subdivisions of the hippocampal formation in the rat, monkey, and human.
Note the differences in the relative position of the mbria in the rat (located lateroventrally)
and in the monkey and human (located mediodorsally). Bar in A  1 mm and applies to all

pocampus, lying in the oor of the inferior horn of the lateral both the names sea horse and silk worm to this region. Others
ventricle, to be reminiscent of this sea creature (Fig. 33). likened the arched structure of the hippocampus to a rams
Arantius also likened the shape of the hippocampus to a silk horn, and De Garengeot (1742) named the hippocampus
worm, but his term bombycini or bombyx never caught on. cornu ammonis or Ammons horn after the mythological
Because Arantius did not take the trouble to stake claim to the Egyptian god Amun Kneph, whose symbol was a ram.
hippocampus by illustrating its location and structure, other According to the Nomina Anatomica, the term hippocampus
anatomists felt free to devise their own terms. The main, or is currently acknowledged to be the standard term for the
intraventricular, portion of the human hippocampus was rst bulge occupying the oor of the lateral ventricle in the human
illustrated by Duvernoy in 1729, and he continued to assign brain. The term Ammons horn is now only rarely used,
Hippocampal Neuroanatomy 41

Figure 33. Dorsolateral view of the human

hippocampus after removing the overlying
structures. The lateral ventricle has been
opened, exposing the shape of the hippocam-
pus. Bar  1 cm.

although a curious irony of current terminology is that, they could recruit or effect appropriate emotional responses
although hippocampus has become the standard term, subdi- (Papez, 1937). Despite the admirable job done by Papez in
visions of this structure are referred to using abbreviations of marshaling largely circumstantial evidence in support of his
cornu ammonis (CA3, CA2, CA1). Some of the terms men- view, there has been little modern substantiation of Papezs
tioned above, as well as many others, have found their way theory. The role of orchestrator of emotional expression is
into descriptions of the histological structure of the hip- now more closely linked with another prominent medial tem-
pocampal formation; hence it is helpful, at the start, to discuss poral lobe structure, the amygdaloid complex.
neuroanatomical terms currently associated with the hip- The hippocampus is also associated with the term limbic
pocampus. system. The origin of this term stems from the description by
A few terms are now of only historical interest. The hip- the neurologist Broca (1878) of le grande lobe limbique,
pocampus was for some time considered to be a prominent which comprised a series of contiguous cortical and subcorti-
part of the rhinencephalon, which, loosely speaking, referred cal structures located on the medial surface of the brain and
to a group of forebrain structures thought to be related espe- surrounding the ventricle at the border (limbus) of the corti-
cially to the sense of smell. In a classic article published in cal mantle. The limbic lobe was described initially as includ-
1947, however, Alf Brodal concluded that the hippocampus ing such circumventricular structures as the subcallosal,
could not be functioning solely as an olfactory structure cingulate, and parahippocampal gyri, as well as the underly-
because anosmic mammals such as dolphins had a substantial ing hippocampal formation. The number of structures
hippocampal formation (Brodal, 1947). It seemed unreason- included under the rubric limbic system has escalated dra-
able to Brodal that the hippocampus would be so robust in a matically since the 1950s, however, and has included such
nonsmelling creature if it functioned solely as a component of diverse structures as the amygdala, septal nuclei, and midbrain
the olfactory system (see Chapter 2). periaqueductal gray matter. The proliferation of brain regions
During the 1930s, the neurologist James Papez considered encompassed by the term limbic system has fueled a long-
the hippocampus to be a central component of a system for lasting debate about the utility of this term either as a neu-
emotional expression, and in many textbooks the hippocam- roanatomical or functional entity. Suffice it to say that in this
pus is still claimed to be the central component of the so- book the hippocampal formation is viewed as an independent
called Papez circuit. In Papezs view, the hippocampus was a functional system rather than part of the larger, ill-dened
conduit by which perceptions of emotionally salient situations group of structures collectively referred to as the limbic sys-
could be collected and channeled to the hypothalamus where tem. That said, the hippocampal formation obviously does
42 The Hippocampus Book

not function in isolation. Even as a functionally dened, inde- the main output projections and that these projections were
pendent system, it relies on interconnections with many other directed subcortically. With the discovery of robust projec-
brain systems to do its job effectively. tions from CA1 to the subiculum and entorhinal cortex, and
the major projections from the entorhinal cortex to the neo-
3.2.1 Denition of Hippocampal Areas: cortex, the trisynaptic circuit is now considered to be only a
Denition of Terms portion of the functional circuitry of the hippocampal forma-
tion. As it is now clear that the subiculum is the main source
When a histological section through the hippocampus is pre- of subcortical projections and the entorhinal cortex is the
pared and stained for cell bodies by a Nissl method, it is main source of projections to the neocortex, the concept of
immediately obvious that a number of cytoarchitectonically the trisynaptic circuit, although of great signicance in the
distinct structures are encompassed in the region grossly history of hippocampal research, is currently less inuential
dened as the hippocampus. A number of terminologies on theories of hippocampal function.
have been applied to some of these regions, and several syn- The reader who ventures from the relative safety of this
onymous terms are still commonly employed. Beyond the book into the primary hippocampal literature should be
problem of different terms being applied to the same region, aware that our usage of the term hippocampal formation is
the borders between several regions of the hippocampal for- widely, though not universally, accepted. Some authors
mation have not yet been rmly established. The terms include only the allocortical (a term applied to cortical regions
adopted in this book are based, in part, on converging evi- having fewer than six layers) regions as parts of the hip-
dence from cytoarchitectonic, histochemical, connectional, pocampal formation. Three-layered cortical regions typically
and functional data and in part on personal preference. have a single neuronal cell layer with ber-rich plexiform
The term hippocampus has historically been plagued by layers above and below the cell layer. In articles employing this
the fact that it is used as a term both for a gross anatomical usage, the hippocampal formation comprises the dentate
region of the brain (the bulge or protuberance in the oor of gyrus, hippocampus, and subiculum. The remaining elds
the human lateral ventricle) and for one of the cytoarchitec- presubiculum, parasubiculum, and entorhinal cortexare
tonically distinct entities that make up the region. As a result, then typically grouped together on the basis of their multil-
the meaning of the word hippocampus is context-dependent aminate structure under the term retrohippocampal (retro 
and may be ambiguous. We reserve the term hippocampus for behind) or parahippocampal (para  alongside or near) cor-
the region of the hippocampal formation that comprises the tex. In yet other variants, the terms hippocampus or hip-
CA elds (CA3, CA2, CA1) identied by the neuroanatomist pocampal complex are sometimes applied to the combination
Rafael Lorente de N (Lorente de N, 1934). The term hip- of the dentate gyrus and hippocampus proper. Clarication of
pocampal formation, in contrast, is applied to a group of the nomenclature is typically the rst sign of maturity for a
cytoarchitectonically distinct adjoining regions including the scientic enterprise, and we scrupulously adhere to the
dentate gyrus, hippocampus, subiculum, presubiculum, para- nomenclature in which the hippocampal formation com-
subiculum, and entorhinal cortex (Fig. 31). The adjective prises the six structures listed above.
hippocampal, by necessity, remains somewhat vague and Having outlined the major regions of the hippocampal for-
context-dependent. In general, it is used to refer to the larger mation, we now delve more deeply into the subdivisions of
area (as in hippocampal lesions) rather than to the cytoar- each of these regions. Before doing so, however, it is important
chitectonic region. If this seems confusing, the termi- to note that the terminology we apply to these regions is a
nological denitions of the next few paragraphs may provide hybrid derived from the analyses of classical and modern hip-
clarication. pocampal neuroanatomists. The two main contributors to the
The main justication for including the six regions named surviving hippocampal terminologies are Santiago Ramon y
above under the rubric hippocampal formation is that they Cajal (Ramn y Cajal, 1893) and his student Raphael Lorente
are linked, one to the next, by unique and largely unidirec- de N (Lorente de N, 1933, 1934). Some of their subdivi-
tional (functional) neuronal pathways. The early literature on sions, based solely on the analysis of Golgi-stained material,
the hippocampal formation emphasized the rst three links have not stood the test of time or the introduction of modern
in the hippocampal circuitry that were highlighted by apply- neuroanatomical methods. Many revisions of their nomencla-
ing the term trisynaptic circuit to the ensemble of pathways ture have been made as information has become available
(Andersen et al., 1971). concerning the connections and chemical architecture of the
hippocampal formation.
EC DG (synapse 1), DG CA3 (synapse 2),
CA3 CA1 (synapse 3)
3.2.2 Subdivisions of Hippocampal Areas
The denition of this powerful excitatory circuit was pro-
duced by a collaboration between early neuroanatomical and The dentate gyrus is a trilaminate cortical region with a char-
electrophysiological studies. It should be borne in mind, how- acteristic V or U shape. The dentate gyrus has a relatively sim-
ever, that the term trisynaptic circuit was coined during an era ilar structure at all levels of the hippocampal formation and is
when it was assumed that the hippocampus proper generated not typically divided into subregions. When discussing fea-
Hippocampal Neuroanatomy 43

tures of the dentate gyrus, however, it is often useful to refer to are somewhat larger and less compact than those in the pre-
a particular portion of the V- or U-shaped structure. The por- subiculum.
tion of the granule cell layer that is located between the CA3 The entorhinal cortex is the only hippocampal region that
eld and the CA1 eld (separated by the hippocampal ssure) unambiguously demonstrates a multilaminate appearance.
is called the suprapyramidal blade; and the portion opposite Based on differences in the organization of these layers and
this is the infrapyramidal blade. The region bridging the two more recently on differences in connectional attributes, the
blades (at the apex of the V or U) is the crest. entorhinal cortex has been divided into two or more subre-
The hippocampus, especially in rodents, can easily be gions depending on the species. We shall return to a descrip-
divided into two major regions: a large-celled region that tion of the subdivisions of the entorhinal cortex later in the
abuts the dentate gyrus and a smaller-celled region that fol- chapter.
lows from it. Ramon y Cajal called these two regions regio Before moving on to descriptions of the three-dimensional
inferior and regio superior, respectively. The terminology of organization of the hippocampal formation in the rat, mon-
Lorente de N has achieved more common usage and is key, and human brains, we must say a few more words con-
employed here. He divided the hippocampus into three elds: cerning nomenclature. We often want to refer to a specic
CA3, CA2, and CA1. His CA3 and CA2 elds are equivalent to portion of one of the hippocampal regions. Given the com-
the large-celled regio inferior of Ramon y Cajal, and his CA1 plex shape of the hippocampal formation, no reference system
eld is equivalent to the regio superior. In addition to the is wholly adequate, and any description inevitably involves
greater size of the pyramidal cells in CA3 and CA2 compared arbitrary decisions about where to start or nish, which direc-
to CA1, the inputs and outputs of these areas are also differ- tion is up or down, and so on. We have adopted a reference
ent. The pyramidal cells of CA3, for example, receive the system in which the dentate gyrus is considered to be the
mossy ber input from the dentate gyrus, whereas the CA1 proximal pole of the hippocampal formation and the entorhi-
pyramidal cells do not. nal cortex is the distal pole (Fig. 31). A portion of any hip-
The CA2 eld has been the subject of substantial contro- pocampal eld can therefore be dened in relation to this
versy. As originally dened by Lorente de N, it is a narrow proximo-distal axis. For example, the proximal portion of
zone of cells interposed between CA3 and CA1. CA2 has large CA3 is located closer to the dentate gyrus, and the distal por-
pyramidal cell bodies similar to those in CA3 but, like CA1, it tion is located closer to CA2.
is not innervated by the mossy bers from the dentate gyrus. We also often need to specify subregions within the thick-
Although the existence of CA2 has often been questioned, the ness of a particular hippocampal region. As in most other cor-
bulk of available evidence indicates that there is indeed a tical areas, this radial dimension is usually described along a
narrow CA2 eld that can be distinguished from the other supercial-to-deep axis. In six-layer structures, layer I is close
hippocampal elds using a variety of criteria, including neu- to the pial surface, and layer VI is located close to the subcor-
rochemical markers. Lorente de N also dened a CA4 eld. tical white matter. Consistent with this convention, regions
As originally claried by Theodor Blackstad (1956) and then closer to the pia or hippocampal ssure are considered super-
by David Amaral (1978), the region that Lorente de N called cial and those in the opposite directioncloser to the alveus
CA4 is actually the deep, or polymorphic, layer of the dentate or ventricle where applicable) are considered deep. The
gyrus. molecular layer of the dentate gyrus, for example, is super-
The subiculum, presubiculum, and parasubiculum are cial to the granule cell layer. This supercialdeep nomencla-
sometimes grouped under the term subicular complex. ture has the merit of being applicable to all portions of the
Because each of these regions has distinct neuroanatomical hippocampal formation and to all of its cytoarchitectonic
features, they are better thought of as independent cortical elds. It has the drawback, however, of being somewhat coun-
areas. The border between CA1 and the subiculum occurs pre- terintuitive to neuroscientists, especially electrophysiologists,
cisely at the point where the Schaffer collateral projection whose electrodes approach the hippocampus from the alveus
from the CA3 eld ends. In the rodent, this occurs approxi- in the septal (or dorsal) portion of the hippocampus. As an
mately where the condensed pyramidal cell layer of CA1 electrophysiologist advances an electrode from the dorsal sur-
begins to broaden into the thicker layer of the subiculum. face of the brain toward the hippocampus, it would rst hit
The presubiculum lies adjacent to the subiculum and is the alveus and then the pyramidal cell layer. Although the
typically thought to have more than the three layers that char- alveus might seem like the supercial portion of the hip-
acterize the dentate gyrus, hippocampus, and subiculum. pocampus because it is closer to the surface of the brain, it
However, the exact delimitation of the deep layers of the pre- is actually its deep portion. As the electrode continues its
subiculum and the differentiation of cells belonging to the advance toward the hippocampal ssure, it enters the super-
presubiculum from those that belong to the deep layers of the cial portion of the hippocampus. Crossing the hippocampal
entorhinal cortex has never been clearly established. The most ssure, it then enters the supercial portion of the molecular
distinctive feature of the presubiculum is the densely packed layer of the dentate gyrus, and a further advance would bring
external cellular layer, which is populated by relatively small, it to the deep portion of the molecular layer, the suprapyra-
tightly packed pyramidal cells. The parasubiculum is charac- midal granule cell layer, the hilus, and nally the infrapyrami-
terized by a wedge-shaped layer II with cells that resemble but dal granule cell and molecular layers.
44 The Hippocampus Book

3.2.3 Major Fiber Bundles of many of the elds of the rodent hippocampal formation are
the Hippocampal Formation grossly C-shaped and vertically oriented, they tend to be much
more linear and horizontally oriented in primates.
Three major ber systems are associated with the hippocam-
pal formation (Fig. 34, see color insert). The rst is the angu- 3.3.1 Rat Hippocampal Formation
lar bundle, which carries bers between the entorhinal cortex
and the other elds of the hippocampal formation. The The rat hippocampal formation is an elongated, banana-
second is the mbria-fornix pathway through which the hip- shaped structure with its long axis extending in a C-shaped
pocampal formation is interconnected with the basal fore- manner from the midline of the brain near the septal nuclei
brain, hypothalamic, and brain stem regions. The third (rostrodorsally) over and behind the thalamus into the incipi-
comprises the dorsal and ventral commissures, through which ent temporal lobe (caudoventrally). The long axis of the hip-
the hippocampal formation of one hemisphere is connected pocampal formation is referred to as the septotemporal axis
with the hippocampal formation of the contralateral hemi- and the orthogonal axis as the transverse axis (Fig. 38). What
sphere. We deal with these in greater detail in Section 3.3.2. is not obvious from a surface view of the hippocampal forma-
tion is that different regions make up the structure at different
septotemporal levels. At extreme septal levels, for example,
only the dentate gyrus and the CA3CA1 elds of the hip-
3.3 Three-dimensional Organization pocampus are present. About a third of the way along the sep-
and Major Fiber Systems of the totemporal axis the subiculum first appears, and the
Hippocampal Formation presubiculum and parasubiculum are seen at progressively
more temporal levels. The entorhinal cortex is located even
The hippocampal formation is positioned quite differently in farther caudally and ventrally. The dorsolateral limit of the
the rodent and primate brains (Figs. 35 through 37, see entorhinal cortex occurs approximately at the rhinal sulcus,
color insert). This is due in part to the more developed cere- which forms a prominent, rostrocaudally/horizontally ori-
bral cortex in primates, which tends to force the dentate ented indentation on the ventrolateral surface of the rat brain.
gyrus and hippocampus into the temporal lobe. Whereas This sulcus nominally separates the entorhinal cortex ventrally

Figure 34. Major ber systems of the

rat hippocampal formation: angular bun-
dle; mbria-fornix; dorsal and central
Hippocampal Neuroanatomy 45

Figure 35. Magnetic resonance imaging

of the rat brain shows the position of the
hippocampus (dentate gyrus  hip-
pocampus proper  subiculum) in red
and the entorhinal cortex in green. A.
Oblique view. B. Dorsal view. C. Frontal
view. D. Lateral view. (Source: Courtesy of
Dr. G. Allan Johnson, Center for In Vivo
Microscopy, Duke University.)

Figure 36. Magnetic resonance imaging

of the monkey brain shows the position
of the hippocampus (dentate gyrus 
hippocampus proper  subiculum) in
red and the entorhinal cortex in green. A.
Oblique view. B. Dorsal view. C. Frontal
view. D. Lateral view.
46 The Hippocampus Book

Figure 37. Magnetic resonance imaging

of the human brain shows the position of
the hippocampus (dentate gyrus  hip-
pocampus proper  subiculum) in red
and the entorhinal cortex in green. A.
Oblique view. B. Dorsal view. C. Frontal
view. D. Lateral view.

Figure 38. Line drawing of the rat brain

shows the septotemporal and transverse
axes of the hippocampal formation.
Hippocampal Neuroanatomy 47

from the perirhinal and postrhinal cortices dorsally. At rostral mbria-fornix to emphasize the continuity of bers in these
levels, however, the perirhinal cortex extends somewhat ven- bundles.
tral to the rhinal sulcus, and at caudal levels the entorhinal The ventricular, or deep, surface of the hippocampus is
cortex extends just slightly dorsal to the rhinal sulcus. covered by a thin sheet of myelinated bers called the alveus.
These bers form a white sheet overlying the hippocampus
3.3.2 Major Fiber Systems of the Rat that can be clearly seen when the overlying neocortex is
Hippocampal Formation aspirated. The alveus is a complex ber system with both
extrinsic afferent and efferent bers, and bers forming part
Angular Bundle of the intrahippocampal network travel within it (i.e., the
entorhinalCA1 alvear pathway, the CA1subiculum projec-
A major ber pathway associated with the hippocampal for- tion, and the CA1entorhinal projection). Some alvear bers
mation is the angular bundle, a ber bundle interposed originate from the pyramidal cells of the hippocampus and
between the entorhinal cortex and the presubiculum and subiculum and are en route to subcortical termination sites
parasubiculum. The angular bundle is the main route taken by (Meibach and Siegel, 1975). At temporal levels of the hip-
bers originating from the ventrally situated entorhinal cortex pocampal formation, the subcortically directed output bers
as they travel to all septotemporal levels of the other hip- extend obliquely in the alveus, from medial to lateral, over the
pocampal elds, particularly the dentate gyrus, hippocampus, surface of the hippocampus and collect in a bundle called the
and subiculum. In addition, the angular bundle contains com- mbria (from the Latin word for fringe), which becomes pro-
missural bers of entorhinal and presubicular origin and gressively thicker as it progresses from the temporal to the
bers to and from a variety of cortical and subcortical struc- septal level (i.e., as axons from more septally located pyrami-
tures that are interconnected with the entorhinal cortex. dal cells are added to the bundle). The rat mbria has a at-
The perforant path comprises the efferent entorhinal pro- tened appearance, contains approximately 900,000 axons, and
jections that traverse, or perforate, the subiculum on their way is situated along the lateral and rostral aspects of the hip-
to the dentate gyrus and the hippocampus. This is the main pocampus. The bers of the mbria are not randomly distrib-
route by which neocortical inputs reach the dentate gyrus and uted but are organized in a topographic fashion (Wyss et al.,
the hippocampus. We present a more detailed description of 1980). Axons located medially in the mbria (i.e., those clos-
the perforant path in the discussion of the inputs to the den- est to the hippocampus) tend to arise from more septal levels,
tate gyrus and the connectivity of the entorhinal cortex. whereas those located laterally arise from more temporal lev-
Entorhinal bers also reach the hippocampus via the els. Fibers from the subiculum are situated deeper to those
alveus, the temporoammonic alvear pathway rst described from the hippocampus.
by Cajal. At temporal levels, most of the entorhinal bers The fornix is the continuation of this bundle of hippocam-
reach the CA1 eld of the hippocampus after perforating the pal output bers to the subcortical target structures; it forms
subiculum (via the classic perforant pathway). At more septal a attened bundle located just below the corpus callosum very
levels, the number of entorhinal bers that take the alvear close to the midline (fornix is the Latin word for arch, which
pathway increases (Deller et al., 1996). In fact, in the septal is the shape of this tract over the diencephalon). Both mbria
portion of the hippocampal formation, most of the entorhinal and fornix carry bers from the hippocampus and subiculum;
bers to CA1 reach this subeld via the alveus. These bers the fornix, however, carries bers primarily from the septal
make sharp right-angle turns in the alveus, perforate the third of these structures. As the bers of the mbria leave
pyramidal cell layer, and nally terminate in the stratum the hippocampus and descend into the forebrain, they are
lacunosum-moleculare. The alveus is therefore also a major referred to as the columns of the fornix. The fornix splits
route by which entorhinal bers reach their targets in CA1. around the anterior commissure to form a rostrally directed
precommissural fornix, which innervates the septal nuclei and
Fimbria-Fornix Pathway nucleus accumbens, and a caudally directed postcommissural
fornix, which extends toward the diencephalon. As the post-
The mbria-fornix ber system provides the major conduit commissural fornix begins its course into the diencephalon
for subcortical afferent and efferent connections (Daitz and (ultimately reaching the mammillary nuclei of the posterior
Powell, 1954; Powell et al., 1957). It is perhaps easiest to hypothalamus), two smaller bundles split off. One, the medial
understand the mbria-fornix system by analogy with the corticohypothalamic tract, innervates a number of anterior
corticospinal ber system. The corticospinal bers are given hypothalamic areas. The other, called the subiculothalamic
different names at different points on their journey from the tract, carries bers to the anterior thalamic nuclei (Swanson
motor cortex to the spinal cord. Similarly, the subcortical and Cowan 1975; Canteras and Swanson, 1992).
afferent and efferent bers of the hippocampal formation are The mbria and fornix also carry bers that are traveling
given different names at different points in their trajectory to the hippocampal formation. Many of the subcortical inputs
from or toward the forebrain or brain stem. Because of this, to the hippocampal formation, including those from the sep-
and because the exact transition between mbria and fornix is tal nuclei (to the septal portion of the hippocampal forma-
difficult to dene, some hippocampal researchers use the term tion), the locus coeruleus, and the raphe nuclei enter via the
48 The Hippocampus Book

mbria-fornix pathway. Some subcortical structures have pro- is located caudally. Because the term septotemporal is not
jections that follow other pathways into the hippocampal for- appropriate for the monkey hippocampal formation (no por-
mation. Fibers from the anterior thalamus, for example, travel tion of it approaches the septal area), it is more common to
through the thalamic radiations and supracallosal stria to refer to the long axis in the primate as the rostrocaudal axis.
innervate the presubiculum. Still other subcortical projec- The orthogonal axis, however, is still referred to as the trans-
tions, particularly those from the amygdala, travel to the hip- verse axis.
pocampal formation via the external capsule. Two additional points should be made about the position
of the monkey hippocampal formation. First, at the rostral
Dorsal and Ventral Hippocampal Commissures limit of the lateral ventricle, some elds of the monkey hip-
pocampal formation ex medially and then caudally. This is
A third major ber system associated with the hippocampal the monkey homologue of the pes hippocampi that is so
formation is the commissural system (Blackstad, 1956; prominent in the human brain. At the rostral levels where this
Raisman et al., 1965; Laatsch and Cowan, 1967; Laurberg, exure occurs, there are two representations of the hippocam-
1979). In the rat, there are both dorsal and ventral commis- pal formation in standard coronal views of the brain. It is
sures. Some 350,000 bers cross the midline in the ventral difficult in this exed region of the monkey hippocampal for-
hippocampal commissure, which is located just caudal to the mation, when viewed in standard coronal, Nissl-stained sec-
septal area and dorsocaudal to the anterior commissure. Many tions, to specify the identity and borders of the subdivisions
of these bers are true commissural bers and are directed to of the hippocampal formation. The most medial and caudal
both homotopic and heterotopic elds in the contralateral portion of the hippocampal formation (i.e., the part that is
hippocampal formation. A much smaller number of bers are bent backward) is the actual rostral pole of the monkey hip-
directed into the contralateral descending column of the pocampal formation, even though it is physically located
fornix and ultimately innervate the same structures on the somewhat caudal to the rostral extreme of the hippocampal
contralateral side of the brain that receive the ipsilateral formation. To make reference to different portions of the
pre- and postcommissural fornix. The dorsal hippocampal monkey hippocampal formation, we call the medial portion of
commissure crosses the midline just rostral to the splenium the hippocampal formation the uncal region (because it forms
(posterior part) of the corpus callosum and carries bers much of the medially situated bulge that in the human would
mainly originating from or projecting to the presubiculum be called the uncus). The exed, rostrally located portion that
and entorhinal cortex. The dorsal hippocampal commissure is runs mediolaterally is called the genu; and the laterally situ-
the route by which the presubiculum contributes a major pro- ated, main portion of the hippocampus is the body.
jection to the contralateral entorhinal cortex. The second point to note is that the monkey entorhinal cor-
tex is physically associated with only the rostral portion of the
3.3.3 Monkey Hippocampal Formation other hippocampal elds. The entorhinal cortex extends cau-
dally just to the level of the lateral geniculate nucleus, whereas
The rst point we address in this section is how the appear- the dentate gyrus, hippocampus, and subiculum extend well
ance and position of the nonhuman primate hippocampal caudal to this level. It is equally important to point out that the
formation differs from that of the rat. Much of the work on rostral half of the entorhinal cortex extends beyond the rostral
the monkey hippocampal formation has been carried out in limit of the other hippocampal elds, where it is located ven-
Old World monkeys, primarily the macaque monkey. The tromedial to the amygdaloid complex. Throughout virtually
description we provide here is for the hippocampal formation all of its rostrocaudal extent, the lateral border of the entorhi-
of typical research monkeys, such as Macaca fascicularis and nal cortex is at the rhinal sulcus, as in the rat.
Macaca mulatta (cynomolgus and rhesus macaque monkeys, The ber bundles of the monkey hippocampal formation
respectively). The hippocampal formation in the macaque are fundamentally similar to those in the rat, although there
monkey is not nearly as C-shaped in its long axis as it is in the are a number of minor differences. First, the mbria is located
rat. It lies almost horizontally in the temporal lobe; and, as in dorsomedially in the monkey rather than ventrolaterally as in
the human, it makes up the major portion of the oor of the the rat (Fig. 32). Second, the mbria leaves the substance
temporal horn of the fourth ventricle. The major determinant of the hippocampal formation at a point near the splenium of
of the change in position of the primate hippocampal forma- the corpus callosum. Thus, the compact bundle of bers that
tion is the massive development of the associational cortices is called the body of the fornix travels rostrally as a pendulous,
of the frontal and temporal lobes. As a result of the caudal and attened cable hanging under the corpus callosum. Upon
ventral transposition of the temporal lobes that takes place reaching the level of the anterior commissure, the bundle
developmentally to accommodate the larger cortical surface, descends as the columns of the fornix and follows the same
the primate hippocampal formation comes to lie almost trajectories outlined for the rat. Third, the ventral and dorsal
entirely within the medial temporal lobe. Because of the ven- hippocampal commissures are relatively less prominent in the
trorostral rotation of the monkey hippocampal formation, the monkey than in the rat, reecting the more restricted com-
homologue of the temporal pole of the rat hippocampal for- missural connections observed in the monkey brain (Amaral
mation is located rostrally in the monkey brain, and the et al., 1984; Demeter et al., 1985). The more prominent of the
equivalent of the septal pole of the rat hippocampal formation two commissures in the monkey is the dorsal hippocampal
Hippocampal Neuroanatomy 49

commissure. It carries bers from the presubiculum and encephalography, there is almost no commissural interaction
entorhinal cortex to the contralateral side; it also carries bers between the hippocampal formations located on each side of
originating in the parahippocampal cortex. the human brain (Wilson et al., 1987).
The subdivision of the hippocampal formation into vari- The ventral surface of the human temporal lobe is demar-
ous regions is fundamentally the same in rat and monkey cated into mediolateral strips by two prominent rostrocau-
brains. There are, however, substantial species differences in dally oriented sulci (Figs. 39 and 310). The more lateral of
certain subregions, especially CA1 and the entorhinal cortex, the two is the occipitotemporal sulcus, which is often broken
which are addressed in greater detail later in the chapter. by small, transverse gyri. The more medial of the sulci, and the
one that is more closely associated with the hippocampal for-
mation, is the collateral sulcus. The collateral sulcus is often
3.3.4 Human Hippocampal Formation continuous with the rostrally situated rhinal sulcus. Unlike the
situation in the rat and the monkey, however, the rhinal sulcus
The three-dimensional position of the human hippocampal is relatively insignicant in the human brain and is associated
formation is similar to that in the macaque monkey brain only with the most rostral portion of the entorhinal cortex. It
(Figs. 36 and 37, respectively). However, owing to the larger is thus not a useful border for the lateral boundary of the
development of the temporal association cortex, in particular entorhinal cortex.
that of the entorhinal and perirhinal cortices, the structure of The collateral sulcus forms most of the lateral border of
the ventromedial surface of the brain, including the gyral pat- what has classically been termed the parahippocampal gyrus.
terns, is substantially different in the human and monkey The parahippocampal gyrus is a complex region that contains
brains. After a brief summary of the gross anatomical attrib- a number of distinct cytoarchitectonic elds and has been
utes of the main or intraventricular portions of the human dened in different ways by different authors. In recent years,
hippocampal formation, we review some of the differences of the parahippocampal gyrus has often been broken up into an
the associated cortical regions. anterior part, which comprises mainly the entorhinal cortex
The classic gross anatomical image of the human hip- and associated perirhinal cortex, and a posterior part, which
pocampal formation is of a prominent bulge in the oor of includes the areas TF and TH of Von Economo (1929).
the temporal horn of the lateral ventricle (Fig. 33). As in the Unlike the situation in the monkey, where the rhinal sulcus
monkey, this portion of the hippocampal formation is widest forms a reasonably reliable lateral border for the entorhinal
at its rostral extent where the structure bends toward the cortex, the collateral sulcus does not provide a discrete lateral
medial surface of the brain. In this area, two to ve subtle gyri, boundary for the human entorhinal cortex. The entorhinal
or digitationes hippocampi, form the pes hippocampi (Gertz cortex actually ends approximately midway along the medial
et al., 1972). The substance of the pes hippocampi is formed bank of the collateral sulcus. The two elds of the perirhinal
by several of the hippocampal elds, and the constituents dif- cortex, areas 35 and 36 of Brodmann (1909), form the remain-
fer at different rostrocaudal levels (see below). Continuing der of the medial bank, fundus, and a portion of the lateral
caudally from the pes hippocampi, the main body of the hip- bank of the collateral sulcus. The perirhinal cortex is massively
pocampus gets progressively thinner as it bends dorsally enlarged in the human brain and may account, in part, for the
toward the splenium of the corpus callosum. prominence of the collateral sulcus. The border zone between
The mbria is situated on the medial surface of the human the entorhinal cortex and the perirhinal cortex was called the
hippocampus, as in the monkey (Fig. 32). At rostral levels, trans-entorhinal zone by Braak (1972, 1980). In this region,
the mbria is thin and at but becomes progressively thicker neuroanatomical markers that label layer II of the entorhinal
caudally as bers are continually added to it. As the mbria cortex demonstrate an oblique band of labeled cells, which
leaves the caudal extent of the hippocampus, it fuses with the makes it appear that layer II is diving underneath the layers of
ventral surface of the corpus callosum and travels rostrally in the perirhinal cortex. Because the perirhinal cortex terminates
the lateral ventricle. The portion of these mbrial bers at a variable point along the lateral bank of the collateral sul-
located between the caudal limit of the hippocampal forma- cus, it is extremely difficult to dene the borders of the
tion and the fusion with the corpus callosum is called the crus entorhinal and perirhinal cortices using imaging modalities
of the fornix, whereas the major portion of the rostrally such as magnetic resonance imaging. The only way it is cur-
directed ber bundle is, as in the monkey, called the body rently possible to dene the border of these elds accurately is
of the fornix. At the end of its rostral trajectory, the body of through histological analysis.
the fornix descends as the columns of the fornix. At about the Interestingly, much of the areal extent of the human
point where the mbria fuses with the posterior portion of entorhinal cortex (at least the rostral portions) can be visually
the corpus callosum, bers extend across the midline to identied on the surface of the brain by the conspicuous
form the hippocampal commissure. A variety of gross bumps, named verrucae (latin for warts), that mark its pial
anatomical terms have been applied to this commissure, but surface. These bumps mark the islands of cells that constitute
the term psalterium (alluding to a harp-like stringed instru- layer II of the entorhinal cortex. The dorsomedial aspect of
ment) is most common. As noted previously, the primate hip- the entorhinal cortex is marked by a conspicuous mound, or
pocampal commissural connections are much more limited secondary gyrus, referred to as the gyrus ambiens (Fig. 310).
than in the rodent; and as suggested by stereotaxic depth The dorsomedial limit of the entorhinal cortex with the amyg-
50 The Hippocampus Book

Figure 39. Ventromedial view of the human tem-

poral lobe. Bar  1 cm.

daloid complex is marked by the shallow sulcus semiannu- (or hook) in gross anatomical descriptions (Fig. 39). The
laris, which is located dorsal to the gyrus ambiens. most rostral of the uncal bulges is often separately labeled
There is a series of prominent bulges on the medial surface the gyrus uncinatus. Histological sections through the gyrus
of the hemisphere just caudal to the dorsomedial portion of uncinatus indicate that it is made up of the amygdalo-
the entorhinal cortex; this is generally labeled the uncus hippocampal region (of the amygdaloid complex) and the

Figure 310. Ventral surface of the human brain

after the brain stem and cerebellum have been
removed. Bar  1 cm.
Hippocampal Neuroanatomy 51

transition zone between the amygdala and hippocampus. The

middle bulge, called the band of Giacomini, is generally com-
posed of the most medial part of the dentate gyrus. The most
caudal of the uncal bulges, called the intralimbic gyrus, is
composed mainly of a portion of the CA3 eld of the hip-
As the parahippocampal gyrus meets the retrosplenial
region caudally, a group of small bumps known as the gyri
Andreae Retzii are seen on the medial surface of the brain
(Fig. 39). This irregular region of human cortex marks
the caudal limit of the hippocampal formation and is com-
posed principally of CA1 and the subiculum. Two obliquely
oriented small gyri located deep to the gyri Andreae Retzii are
called the fasciola cinerea (which corresponds to the caudal-
most part of the dentate gyrus) and the gyrus fasciolaris
(which corresponds to the caudal pole of the CA3 eld). A
thin remnant of the hippocampal formation surrounds the
splenium of the corpus callosum and ascends dorsally over the
corpus callosum to form the induseum griseum (or supra-
commissural hippocampal formation), which comprises rem-
nants of the subiculum and other elds of the hippocampal

3.4 Neuroanatomy of the Rat

Hippocampal Formation

We have now concluded our overview of the gross or three-

dimensional organization of the rat, monkey, and human hip-
pocampal formations. In the sections that follow, we review
the neuroanatomical organization of each of the elds of the
hippocampal formation of the rat, beginning with the dentate
gyrus and ending with the entorhinal cortex. In each section,
we describe the cytoarchitectonic organization followed by a
summary of the resident neurons and nally a survey of the
pattern of intrinsic and extrinsic connections. The amount of
information available for each of the cytoarchitectonic elds
of the hippocampal formation varies substantially. The most
Figure 311. Horizontal section through the rat hippocampal for-
detailed anatomical information is available for the dentate
mation. A. Nissl-stained section. B. Line drawing shows the various
gyrus, perhaps because it is the simplest of the hippocampal regions, layers, and ber pathways. C. Timms sulde silver-stained
elds. For this reason and because the dentate gyrus is the rst section. Note the three bands of the molecular layer of the dentate
structure to receive direct entorhinal input and does not proj- gyrus in C. The outer band corresponds to the terminal zone of the
ect outside the hippocampal formation, we begin our descrip- lateral perforant pathway; the middle unstained region corresponds
tion here. We then progress through the hippocampus and to the terminal zone of the medial peforant pathway; the inner band
other hippocampal elds. corresponds to the zone of termination of the associational and
We have prepared a series of illustrations to demonstrate commissural pathways of the dentate gyrus. ab, angular bundle; al,
the cytoarchitectonic organization of the rat hippocampal alveus; CA1, CA1 eld of the hippocampus; CA2, CA2 eld of the
formation (Figs. 311 through 314). The rst gure (Fig. hippocampus; CA3, CA3 eld of the hippocampus; DG, dentate
gyrus; EC, entorhinal cortex; , mbria; gcl, granule cell layer of the
311) is a photomicrograph of a section cut in the horizontal
dentate gyrus; hf, hippocampal ssure; ml, molecular layer of the
plane. This plane has the advantage of illustrating all compo-
dentate gyrus; Para, parasubiculum; pcl, pyramidal cell layer of the
nents of the rat hippocampal formation. A Nissl-stained sec- hippocampus; pl, polymorphic layer of the dentate gyrus; Pre, pre-
tion is shown at the top, and a Timms sulde silver section is subiculum; sl, stratum lucidum of CA3; sr, stratum radiatum of the
shown at the bottom. The latter has been used extensively for hippocampus; sl-m, stratum lacunosum moleculare of the hip-
highlighting various ber and terminal systems in the hip- pocampus. Roman numerals: cortical layers. Bar in A  500 m
pocampal formation. After Figure 311, we provide images of and applies to all panels.
52 The Hippocampus Book

Figure 312. Coronal sections at three rostrocaudal levels through the rat brain show the rela-
tive position of the hippocampal formation. The three panels on the left are Nissl-stained sec-
tions; the three panels on the right are adjacent sections stained with Timms sulde silver
stain; the line drawings (middle column) highlight the regions of the rat hippocampal forma-
tion seen in the stained sections. For abbreviations see Figure 311. Bar  1 mm.
Hippocampal Neuroanatomy 53

Figure 313. Horizontal sections at three dorsoventral levels through the rat brain show the
relative position of the hippocampal formation. The three panels on the left are Nissl-stained
sections; the three panels on the right are adjacent sections stained with Timms sulde silver
stain; the line drawings (middle column) highlight the regions of the rat hippocampal forma-
tion seen in the stained sections. For abbreviations see Figure 311. Bar  1 mm.
54 The Hippocampus Book

Figure 314. Sagittal sections at three mediolateral levels through the rat brain show the
relative position of the hippocampal formation. The three panels on the left are Nissl-stained
sections; the three panels on the right are adjacent sections stained with Timms sulde silver
stain; the line drawings (middle column) highlight the regions of the rat hippocampal forma-
tion seen in the stained sections. For abbreviations see Figure 311. Bar  1 mm.
Hippocampal Neuroanatomy 55

similar stains of coronal (Figure 312), horizontal (Fig. 313), the branches directed toward the supercial portion of the
and sagittal (Fig. 314) sections through different levels of the molecular layer; most of the distal tips of the dendritic tree
hippocampal formation. Finally, we must point out that to end just at the hippocampal ssure or at the ventricular sur-
keep this chapter readable we have used citations sparingly. face. The dendritic trees of granule cells located in the
Detailed references for the information provided in this chap- suprapyramidal blade tend, on average, to be larger than those
ter can be found in Witter and Amaral (2004). of cells located in the infrapyramidal blade (3500 m vs. 2800
m). Desmond and colleagues (Desmond and Levy, 1985)
3.4.1 Dentate Gyrus provided estimates for the number of dendritic spines on the
granule cell dendrites. They found that cells in the suprapyra-
Cytoarchitectonic Organization midal blade have 1.6 spines/m, whereas cells in the
infrapyramidal blade have 1.3 spines/m. With these num-
The dentate gyrus is comprised of three layers. Supercially, bers and the mean dendritic lengths given above, an estimate
closest to the hippocampal ssure is a relatively cell-free layer of the number of spines on the average suprapyramidal gran-
called the molecular layer. In the rat, this layer has an average ule cell would be 5600 and for an infrapyramidal cell 3640.
thickness of approximately 250 m. The principal cell layer The total number of granule cells in one dentate gyrus
(granule cell layer) lies deep to the molecular layer and is of the rat is about 1.2106 (West et al., 1991; Rapp and
made up of a densely packed layer that is four to eight granule Gallagher, 1996). Although cell proliferation and neurogenesis
cells thick. The granule cell and molecular layers (which in the dentate gyrus persist into adulthood and appear to be
together are sometimes referred to as the fascia dentata) form under environmental control, modern stereological studies
a V- or U-shaped structure (depending on the septotemporal have shown that the total number of granule cells does not
position) that encloses a cellular region, the polymorphic cell vary in adult animals (Rapp and Gallagher, 1996). Only infant
layer, which constitutes the third layer of the dentate gyrus and juvenile mice exposed to running wheels or socially com-
(Fig. 311). plex, enriched environments demonstrate a larger dentate
gyrus and a greater number of granule cells that persist into
Neuron Types adulthood than animals raised in standard laboratory cages
(Kempermann et al., 1997). Similar manipulations performed
Dentate Granule Cell in adult animals can affect cell proliferation and/or survival of
The principal cell type of the dentate gyrus is the granule cell newly generated neurons, but they have no signicant impact
(Figs. 315 and 316) The dentate granule cell has an ellipti- on the volume of the dentate gyrus or the total number of
cal cell body with a width of approximately 10 m and a granule cells (Kempermann et al., 1998).
height of 18 m (Claiborne et al., 1990). Each cell is closely Before leaving the topic of granule cell number, we should
apposed to other granule cells, and in most cases there is no note that the packing density and thickness of the granule cell
glial sheath intervening between the cells. The granule cell has layers varies somewhat along the septotemporal axis of the
a characteristic cone-shaped tree of spiny dendrites with all dentate gyrus (Gaarskjaer, 1978a). The packing density of

Figure 315. Dendritic arborization of

the principal cells in the rat dentate gyrus
(granule cells) and hippocampus proper
(pyramidal cells). See the text for details.
56 The Hippocampus Book

Figure 316. Dendritic organization

of dentate granule cells. Computer-
generated reconstructions of horserad-
ish peroxidase (HRP)-lled granule
cells from the suprapyramidal blade.
Bar  100 m. (Source: Adapted from
Claiborne et al., 1990.)

granule cells is higher septally than temporally. Because the cal dendrite directed into the molecular layer (where it divides
packing density of CA3 pyramidal cells follows an inverse gra- into several aspiny branches) and several principal basal den-
dient, the net result is that at septal levels of the hippocampal drites that ramify and extend into the polymorphic cell layer.
formation the ratio of granule cells to CA3 pyramidal cells is Most of these cells contain biochemical markers for the
something on the order of 12:1, whereas at the temporal pole inhibitory transmitter -aminobutyric acid (GABA) and are
the ratio drops to 2:3. Because the CA3 pyramidal cells are the thus presumably inhibitory (Ribak et al., 1978; Ribak and
major recipients of granule cell innervation and the number Seress, 1983). The number of basket cells is not constant
of mossy ber synapses is roughly the same along the sep- throughout the transverse or septotemporal extents of the
totemporal axis, contact probability is much lower septally dentate gyrus (Seress and Pokorny, 1981). At septal levels, the
than temporally. ratio of basket cells to granule cells is 1:100 in the suprapyra-
The granule cell is the only principal cell of the dentate midal blade and 1:180 in the infrapyramidal blade. At tempo-
gyrus; that is, it is the only cell type that gives rise to axons that ral levels, the number is 1:150 for the suprapyramidal blade
leave the dentate gyrus to innervate another hippocampal and 1:300 for the infrapyramidal blade. These data raise a
eld (i.e., CA3). There is one type of neuron, called the mossy theme that is repeated throughout the chapter: that despite
cell, whose axon leaves the dentate gyrus of one side of the the apparent cytoarchitectonic homogeneity of the hip-
brain only to innervate the dentate gyrus of the other side. pocampal elds, there are several differences (especially
There are numerous other types of neuron, most of which are regarding neurochemical innervation) at different septotem-
inhibitory interneurons. We describe them in turn. poral levels of the hippocampal formation.

Pyramidal Basket Cell Other Interneurons

The most intensively studied interneuron is the pyramidal There has been an explosion in the number of interneurons
basket cell (Fig. 317). These cells are generally located along identied in the rat hippocampal formation. A detailed
the deep surface of the granule cell layer. They have pyramid- overview of the characteristics of the various hippocampal
shaped cell bodies 25 to 35 m in diameter and are wedged interneurons has been published by Freund and Buzsaki
slightly into the granule cell layer. The basket portion of the (1996). Hippocampal interneurons, which form a heteroge-
name refers to the fact that the axon of these cells forms peri- neous population, are designed to carry out different func-
cellular plexuses that surround and form synapses with the tions. Many of the cell types can be distinguished on the basis
cell bodies of granule cells. Ramon y Cajal rst described the of the distribution of their axonal plexus. Some have axons
pyramidal basket cells as having a single, principal aspiny api- that terminate on cell bodies, whereas others have axons that
Hippocampal Neuroanatomy 57

Figure 317. Morphological classication of the interneurons of ing an indication of which domain is innervated by which inter-
the rat dentate gyrus. Filled circles indicate the location of the cell neuron groups. The laminar distribution of various inputs, often
bodies, and thick lines indicate the predominant orientation and showing correspondence with the interneuron type or axon distri-
laminar distribution of the dendritic tree. The dentate granule bution, is also indicated. (Source: Adapted from Freund and
cells (principal neurons) are illustrated in the background, provid- Buzsaki, 1996.)

terminate exclusively on the initial segments of other axons. active for GABA. They form symmetrical synaptic contacts
Interneurons have also been distinguished on the basis of with the cell bodies, proximal dendrites, and occasionally
their inputs. Some are preferentially innervated (e.g., by the axon initial segments of granule cells and therefore function
serotonergic fibers originating from the raphe nuclei). as inhibitory interneurons. These cells are not neurochemi-
Interneurons can also be differentiated from principal cells on cally homogeneous, however, as subsets appear to colocalize
the basis of their electrophysiological characteristics. At least distinct categories of other neuroactive substances.
some interneurons have high rates of spontaneous activity
and re in relation to the theta rhythm. For this reason, Neurons of the Molecular Layer
interneurons are often called theta cells (see Chapter 11). The molecular layer is occupied primarily by dendrites of the
A major challenge is to determine if different classes of granule, basket, and polymorphic cells as well as axons and
interneurons demonstrate distinct electrophysiological terminal axonal arbors from the entorhinal cortex and other
response proles. sources. At least two neuron types are also present in the
Within the same subgranular region occupied by the cell molecular layer. The rst is located deep in the molecular
bodies of the pyramidal basket cells are several other cell types layer, has a multipolar or triangular cell body, and gives rise to
with distinctly different somal shapes, as well as different den- an axon that produces a substantial terminal plexus largely
dritic and axonal congurations (Amaral, 1978). Some of limited to the outer two-thirds of the molecular layer. This
these cells are multipolar with several aspiny dendrites enter- neuron, which has aspiny dendrites that remain mainly within
ing the molecular and polymorphic layers, whereas others the molecular layer, has been called the MOPP cell (molecular
tend to be more fusiform-shaped with a similar dendritic dis- layer perforant path-associated cell). This terminology was
tribution. As Ribak and colleagues pointed out, many of these proposed by Han et al. (1993) to bring some order to naming
cells share ne structural characteristics such as infolded interneurons in the hippocampal formation. The lettering sys-
nuclei, extensive perikaryal cytoplasm with large Nissl bodies, tem refers to the location of the cell body and to the region
and intranuclear rods. Moreover, it appears that all of these where the axon is distributed.
cells give rise to axons that contribute to the basket plexus in Frotscher and colleagues described a second type of neu-
the granule cell layer. Many of these neurons are immunore- ron in the molecular layer that resembles the so-called chan-
58 The Hippocampus Book

delier, or axo-axonic, cell originally found in the neocortex most of the daughter dendritic branches remain within the
(Soriano and Frotscher, 1989). These cells are generally polymorphic layer, an occasional dendrite pierces the granule
located immediately adjacent to or even within the supercial cell layer and enters the molecular layer. The mossy cell den-
portion of the granule cell layer. The axo-axonic cell is named drites virtually never enter the adjacent CA3 eld.
for the fact that its axon descends from the molecular layer The most distinctive feature of the mossy cell is that all of
into the granule cell layer, collateralizes profusely, and then its proximal dendrites are covered by large, complex spines
terminates, with symmetrical synaptic contacts, exclusively on evocatively called thorny excrescences. These spines are the
the axon initial segments of granule cells. Thus, their shape distinctive sites of termination of the mossy ber axons (i.e.,
resembles that of a chandelier. Each axo-axonic cell may axons of the dentate granule cells). Although thorny excres-
innervate the axon initial segments of as many as 1000 gran- cences are also observed on the proximal dendrites of pyram-
ule cells. Because these cells are immunoreactive for markers idal cells in CA3, they are never as dense as the ones on the
of GABAergic neurons and make symmetrical synapses, it is mossy cells. The distal dendrites of the mossy cell have typical
likely that they provide a second means of inhibitory control pedunculate spines that appear to be less densely distributed
of granule cell output. The inputs to the axo-axonic cells are than those on the distal dendrites of the pyramidal cells in the
currently unknown although their dendrites remain mainly in hippocampus. The mossy cells are immunoreactive for gluta-
the molecular layer, where they are likely to receive perforant mate and give rise to axons that project to the inner third of
path input from the entorhinal cortex. the molecular layer of the ipsilateral and contralateral dentate
Other neurons with cell bodies located in the molecular gyrus, making asymmetrical terminations on the dendrites of
layer are members of the IS (interneuron-specic) class of granule cells. The mossy cells thus appear to be the major
interneuron, which are specialized for termination on other source of the excitatory associational/commissural projection
interneurons. These IS neurons are demonstrated using to the dentate gyrus. For this reason, the mossy cell does not
immunohistochemistry for vasoactive intestinal peptide t the classic description of an interneuron.
(VIP), and their axons overlap with the dendrites of the O-LM There are also a number of fusiform cells in the polymor-
and HIPP cells (these cell types are dened shortly). phic layer. The main difference between the fusiform cell types
is whether they have spines. One type, the long-spined multi-
Neurons of the Polymorphic Cell Layer polar cell rst described by Amaral (1978), has recently been
The polymorphic layer harbors a variety of neuron types, but called the HIPP cell (hilar perforant path-associated cell). This
little is known about many of them (Amaral, 1978). The most cell type has two or three principal dendrites that originate
common type, and certainly the most impressive, is the mossy from the poles of the cell, run mainly parallel to the granule
cell (Fig. 318). This cell type is probably what Ramon y Cajal cell layer, and can extend for nearly the entire transverse
referred to as the stellate or triangular cells located in his length of one blade of the granule cell layer. The conspicuous
subzone of fusiform cells; and it is undoubtedly what Lorente feature of this cell is the distribution of copious, long, often
de N referred to as modied pyramids. The cell bodies of branched spines over its cell body and dendrites. Intracellular
the mossy cells are large (2535 m) and are often triangular staining techniques demonstrate that these cells have axons
or multipolar in shape. Three or more thick dendrites origi- that ascend into the outer two-thirds of the molecular layer
nate from the cell body and extend for long distances in the (i.e., the perforant path zone) and terminate with symmetri-
polymorphic layer. Each principal dendrite bifurcates once or cal and presumably inhibitory synapses on the dendrites of
twice and generally gives rise to a few side branches. Although granule cells. An amazing feature of these neurons is that their

Figure 318. Mossy cell of the poly-

morphic layer of the rat dentate gyrus.
A. Camera lucida drawing of the soma
and dendritic arbor. Note that the den-
drites extend widely in the polymor-
phic layer but do not penetrate the
granule cell layer. B. Camera lucida
drawing of the soma and axonal plexus.
The axon ramies throughout the poly-
morphic layer and enters the dentate
granule cell layer at considerable dis-
tances from the soma. A single collat-
eral projects through the CA3 cell body
layer and toward the mbria. Bar 
100 m. (Source: Adapted from
Buckmaster et al., 1992.)
Hippocampal Neuroanatomy 59

axonal plexus can extend for as much as 3.5 mm along the the perforant path bers originating from the medial entorhi-
septotemporal axis of the dentate gyrus (the entire length of nal area terminate in the middle third of the molecular layer.
the dentate gyrus in the rat is only about 10 mm) and may These terminal zones are readily distinguished by the classic
generate as many as 100,000 synaptic terminals. Because Timms staining method for visualizing heavy metals; this
inhibitory interneurons typically have aspiny dendrites and method demonstrates dense staining in the outer third of the
relatively local axonal plexuses, this long-spined multipolar molecular layer, a near absence of staining in the middle third,
HIPP cell is a highly atypical interneuron. At least some of and dark staining in the inner third that is associated with
these HIPP cells appear to correspond to the somatostatin/ the commissural/associational connection (Fig. 311). Projec-
GABA cells that give rise to the somatostatin innervation of tions from both areas of the entorhinal cortex innervate the
the outer portion of the molecular layer. entire transverse extent of the molecular layer. The thin axon
There are also multipolar or triangular cells in the poly- branches (0.1 m) in the molecular layer of the dentate gyrus
morphic layer with thin, aspiny dendrites that extend within show periodic varicosities with a thickness of 0.5 to 1.0 m.
both the hilus and the molecular layer. The axons of these Most of entorhinal cortex layer II spiny stellate cells project up
HICAP cells (hilar commissural-associational pathway- to 2 mm in the septotemporal direction, forming a sheet-like
related cells) extend through the granule cell layer and branch axon arbor in the molecular layer (Fig. 319) (Tamamaki
profusely in the inner third of the molecular layer. A variety of and Nojyo, 1993). There is surprisingly little quantitative
other neuron types exist in the polymorphic layer of the den- information about the termination of the perforant path pro-
tate gyrus whose axonal plexus has not yet been well jections.
described. Because the perforant path bers terminate exclusively in
Before leaving the typing of interneurons in the dentate the molecular layer, certain neurons of the dentate gyrus are
gyrus, it is perhaps worth noting that the eld has come a long not innervated by this input. Cells with dendrites conned to
way from believing that interneurons merely damp down the polymorphic layer do not receive entorhinal input. The
neuronal activity. Rather, most current hippocampologists mossy cells, for example, are thus likely to receive little or no
highlight the heterogeneity of interneurons and see them as direct perforant path input.
integral components of normal information processing in the It has often been assumed that the perforant path bers
hippocampal formation. from the entorhinal cortex are the only hippocampal input
reaching the dentate gyrus, but it is now clear that at least a
Extrinsic Connections minor projection also arises in the presubiculum and para-
subiculum (Kohler, 1985). These bers enter the molecular
Entorhinal Cortex Projection to the Dentate Gyrus layer of the dentate gyrus and ramify in a zone that is inter-
The dentate gyrus receives its major input from the entorhinal spersed between the lateral and medial perforant path projec-
cortex via the so-called perforant pathway (Ramn y Cajal, tions. The presubicular axons tend to be thicker than those
1893). The projection to the dentate gyrus arises mainly from from the entorhinal cortex and give rise to collaterals that take
cells located in layer II of the entorhinal cortex, although a
minor component of the projection also comes from layers V
and VI (Steward and Scoville, 1976). In the molecular layer of
the dentate gyrus, the entorhinal terminals are strictly con-
ned to the supercial (outer) two-thirds, where they form
asymmetrical synapses that account for nearly 85% of the
total axospinous terminations (Nafstad, 1967; Hjorth-
Simonsen and Jeune, 1972). These contacts occur primarily
on the dendritic spines of granule cells, although a small num-
ber of perforant path bers also form asymmetrical synapses
on the shafts of GABA-positive interneurons. The organiza-
tion of the perforant path projection in the mouse is similar
to that in the rat, although there are some detectable species
differences, such as a paucity of commissural connections in
the mouse (van Groen et al., 2002, 2003).
The perforant pathway can be divided into two parts based
on the region of origin, pattern of termination, and appear-
Figure 319. Perforant path projections. Distribution of labeled
ance in histochemical and immunohistochemical prepara- axon branches of a layer II spiny stellate neuron in the molecular
tions. In the rat, the two divisions have been called the lateral layer of the dentate gyrus and the stratum lacunosum-moleculare
and medial perforant paths because they originate from the of the CA2-CA3 elds of the hippocampus observed in a parasagit-
lateral and medial entorhinal areas, respectively. Perforant tal section. Bar  500 m. (Source: Tamamaki and Nojyo, 1993.
path bers originating in the lateral entorhinal area terminate With permission of Wiley-Liss, a subsidiary of John Wiley &
in the most supercial third of the molecular layer, whereas Sons.
60 The Hippocampus Book

a radial course in the molecular layer, in contrast to the pre- pocampal formation are cholinergic. Although it had long
dominantly transverse orientation of the entorhinal perforant been assumed that the septal projection to the hippocampal
pathway bers. Virtually nothing is currently known about formation was entirely cholinergic, it is now clear that many
which cells these bers innervate or what type of transmitter of the septal cells that project to the dentate gyrus are actually
they use. Because the presubiculum receives the only direct GABAergic. The most interesting facet of this heterogeneous
input from the anterior thalamic nucleus, these bers provide septal projection is that the cholinergic and GABAergic com-
a potential link by which thalamic information could reach ponents target different cell types. Fibers of the septal
the dentate gyrus. GABAergic projections terminate preferentially on other
GABAergic nonpyramidal cells, such as the basket pyramidal
Basal Forebrain Inputs: Projections from the Septal Nuclei cells of the dentate gyrus, and form symmetrical, presumably
The dentate gyrus receives relatively few inputs from subcor- inhibitory, contacts. The heaviest GABAergic septal termina-
tical structures. Certainly the most robust and longest studied tion is on interneurons located in the polymorphic layer. The
is the projection from the septal nuclei (Mosko et al., 1973; cholinergic septal projection to the dentate gyrus, in contrast,
Swanson, 1978a; Baisden et al., 1984; Amaral and Kurz, 1985; terminates mainly on granule cells, making asymmetrical,
Wainer et al., 1985; Nyakas et al., 1987). The septal projection presumably excitatory, contacts on dendritic spines, chiey in
arises from cells of the medial septal nucleus and the nucleus the inner third of the molecular layer; only 5% to 10% of the
of the diagonal band of Broca, and it travels to the hippocam- cholinergic synapses are on interneurons.
pal formation via four routes: the mbria, dorsal fornix, The septal projection to the dentate gyrus and to the
supracallosal stria, and a ventral route through and around remainder of the hippocampal formation is topographically
the amygdaloid complex. Septal bers heavily innervate cells organized. Cells located medially in the medial septal nucleus
of the polymorphic layer, particularly in a narrow region just tend to project preferentially to septal or dorsal levels of the
subjacent to the granule cell layer. The large mossy cells are dentate gyrus, whereas cells located laterally in the medial sep-
innervated by cholinergic bers (Lubke et al., 1997). Septal tal nucleus tend to project to temporal levels. Because the
bers are lightly distributed throughout the molecular layer medially situated neurons in the medial septal nucleus tend to
(Fig. 320a). be GABAergic rather than cholinergic, septal levels of the den-
A large portion of the bers of the septal projection to the tate gyrus receive most of their cholinergic input from the
dentate gyrus are cholinergic. Altogether, 30% to 50% of the nucleus of the diagonal band. In contrast, temporal levels of
cells in the medial septal nucleus and 50% to 75% of the cells the dentate gyrus receive their cholinergic innervation prima-
in the nucleus of the diagonal band that project to the hip- rily from the medial septal nucleus. The ventral septal pathway

Figure 320. Septal and hypothalamic

inputs to the rat hippocampal formation.
A. Distribution of PHA-L-labeled bers
in the hippocampal formation resulting
from an injection focused in the medial
portion of the medial septal nucleus.
B. Distribution of PHA-L-labeled bers
resulting from an injection in the supra-
mammillary nucleus. Note the dense
plexus of bers located just supercial to
the granule cell layer and in the CA2
region of the hippocampus (between the
small arrows). (Sources: A. From
Gaykema et al., 1990. B. Adapted from
Haglund et al., 1984.)
Hippocampal Neuroanatomy 61

appears to arise mainly from the nucleus of the diagonal band that project to the dentate gyrus. Although taken together
of Broca. these cells constitute a sizable input to the hippocampal for-
mation, their diffuseness and lack of any distinguishing bio-
Supramammillary and Other Hypothalamic Inputs chemical marker has made it difficult to study their patterns
The major hypothalamic projection to the dentate gyrus arises of termination in the hippocampal formation.
from a population of large cells, the supramammillary area,
which caps and partially surrounds the medial mammillary Brain Stem Inputs
nuclei (Wyss et al., 1979; Dent et al., 1983; Vertes, 1993; The dentate gyrus receives a particularly prominent nora-
Magloczky et al., 1994). The supramammillary projection ter- drenergic input from the pontine nucleus locus coeruleus
minates heavily in a narrow zone of the molecular layer (Pickel et al., 1974; Swanson and Hartman, 1975; Loughlin et
located just supercial to the granule cell layer and only lightly al., 1986). The noradrenergic bers terminate mainly in the
in the polymorphic layer or the rest of the molecular layer polymorphic layer of the dentate gyrus and extend into
(Fig. 320b). Most of the supramammillary bers terminate the stratum lucidum of CA3, as if preferentially terminating
on the proximal dendrites of granule cells. There has been in the zones occupied by mossy bers (Fig. 3-21).
some controversy about the neurotransmitter of the supra- The dentate gyrus receives a minor, diffusely distributed
mammillary projection. There is substantial evidence that this dopaminergic projection that arises mainly from cells located
projection is excitatory and is likely using glutamate as a pri- in the ventral tegmental area. The dopaminergic bers termi-
mary neurotransmitter (Kiss et al., 2000). Most, but not all, of nate mainly in the polymorphic layer.
the glutamatergic suprammamillary neurons that project to The serotonergic projection that originates from median
the dentate gyrus also colocalize calretinin; some of these cells and dorsal divisions of the raphe nuclei also terminates most
also colocalize substance P (Borhegyi and Leranth, 1997). heavily in the polymorphic layer in an immediately subgran-
In addition to the supramammillary cells, there are cells ular portion of the layer (Conrad et al., 1974; Moore and
scattered in several hypothalamic nuclei (many of which are Halaris, 1975; Khler and Steinbusch, 1982; Vertes et al.,
in a perifornical position or in the lateral hypothalamic area) 1999). A number of GABAergic interneurons appear to be

Figure 321. Line drawing of horizontal sections through the rat hippocampal formation
shows the distribution of A. noradrenergic, B. serotonergic, and C. dopaminergic bers.
(Source: Adapted from Swanson et al., 1987.
62 The Hippocampus Book

preferentially innervated by the serotonergic bers. The tar- Given the small number of basket cells relative to granule
gets are often the pyramidal basket cells. Fusiform neurons cells, the question arises as to how widespread the inuence of
in the region, particularly those that stain for the calcium- a single basket cell is. Analysis of Golgi-stained axonal
binding protein calbindin, are also heavily innervated. As with plexuses from single basket cells indicates that they extend for
the cholinergic projection from the septum, many of the cells distances of more than 900 m in the transverse axis and
in the raphe nuclei that project to the hippocampal formation about 1.5 mm in the septotemporal axis. This widely distrib-
appear to be nonserotonergic, but their transmitter is not uted axonal plexus would allow a single basket cell to inu-
known. ence as many as 10,000 granule cells (about 1%). The other
inhibitory input to granule cells originates with the chandelier
(axo-axonic) cells located in the molecular layer. These cells
Intrinsic Connections form symmetrical contacts exclusively with the axons initial
segment of granule cells.
Basket Cell and Axo-axonic Cell Innervation
of the Granule Cell Layer Granule Cell Projection to the Polymorphic Layer
As already noted, a variety of basket cells are located just The granule cells give rise to distinctive unmyelinated axons
below the granule cell layer and appear to contribute to a that Ramon y Cajal called mossy bers. The mossy bers have
extremely dense terminal plexus that is conned to the gran- unusually large boutons that form en passant synapses with
ule cell layer (Struble et al., 1978; Sik et al., 1997). The termi- the CA3 pyramidal cells; we describe this projection in more
nals in this basket plexus are GABAergic and form detail shortly. What is not generally appreciated is that the
symmetrical, inhibitory contacts located primarily on the cell mossy ber axons form a distinctive set of collaterals that also
bodies and proximal dendritic shafts of apical dendrites of the innervate cells in the polymorphic layer of the dentate gyrus
granule cells. GABAergic neurons in the polymorphic layer (Fig. 322). Each principal mossy ber (which is on the order
are themselves innervated by other GABAergic terminals, of 0.20.5 m in diameter) gives rise to about seven thinner
some of which arise from extrinsic sources, such as the collaterals in the polymorphic layer before entering the CA3
GABAergic septal input. This polysynaptic cascade of eld of the hippocampus. As much as 2300 m of the collat-
inhibitory interconnections indicates that the hippocampal eral axonal plexus is generated by a single mossy ber in the
circuitry provides intricate inhibitory and disinhibitory con- polymorphic layer (Claiborne et al., 1986). In the polymor-
trol of cell excitability, an issue discussed in Chapters 5 and 6. phic layer, the mossy ber collaterals branch extensively, and

Figure 322. Granule cell projections to the

polymorphic cell layer. Axon arbors of two
adjacent granule cells (gray and black) and
their mGluR1a immunoreactive targets
reconstructed from three neighboring 60 m
thick sections. Of the 175 small terminals and
lopodiae (arrowheads) 52 innervated
mGluR1a immunoreactive targets, whereas
large mossy terminals contacted none. gcl,
granule cell layer; pl, polymorphic layer. Bar
 50 m. (Source: Adapted from Acsady et
al., 1998.)
Hippocampal Neuroanatomy 63

the daughter branches bear two types of synaptic varicosity. Most of the synaptic terminals of the associational/com-
Numerous small (approximately 2 m) spherical synaptic missural pathway form asymmetrical, presumably excitatory
varicosities are distributed unevenly along these collaterals. synaptic terminals on the spines of proximal dendrites of
There are about 160 to 200 of these varicosities distributed granule cells. Many, perhaps all, of the axons contributing to
throughout the axonal collateral plexus of a single granule these synaptic terminals originate from the mossy cells of the
cell, and they form contacts on dendrites located in the poly- polymorphic layer, and individual mossy cells contribute a
morphic layer. At the end of each of the collateral branches projection to both the ipsilateral associational and commis-
there are also larger (35 m diameter), irregularly shaped sural projections. The fact that the mossy cells are immunore-
varicosities that resemble, although are smaller than, the active for glutamate adds credence to the notion that the
mossy ber boutons found in CA3. The mossy ber terminals associational/commissural projection is excitatory.
in the polymorphic layer establish contacts with the proximal There are a number of interesting features of this feed-
dendrites of the mossy cells (Ribak et al., 1985), the basal den- back projection from the mossy cells to the granule cells.
drites of the pyramidal basket cells, and other, unidentied First, the projection from mossy cells located at any particular
cells. Acsady et al. (1998) made the surprising discovery that level of the dentate gyrus is distributed widely along the lon-
most of the granule cell collaterals in the polymorphic cell gitudinal axis, both septally and temporally from the point of
layer terminate on GABAergic interneurons. Because there are origin. Axons from any particular septotemporal point in the
160 to 200 such varicosities (compared with about 20 of the dentate gyrus may innervate as much as 75% of the long axis
larger thorny expansions), mossy ber axons synapse on a of the dentate gyrus (Amaral and Witter, 1989). Second, the
larger number of interneurons than do mossy cells or CA3 projection to the molecular layer at the septotemporal level of
pyramidal cells. Mossy ber collaterals may enter the granule origin is extremely weak but becomes increasingly stronger at
cell layer, but they virtually never enter the molecular layer. levels that are progressively more distant from the cells of ori-
The collaterals that enter the granule cell layer appear to ter- gin. Remembering that mossy cells are the recipients of mas-
minate preferentially on the apical dendritic shafts of pyram- sive innervation from the granule cells at their same level (via
idal basket cells. This lack of mossy ber innervation of the the mossy ber collaterals into the polymorphic layer), it
molecular layer is of some importance because in the presence appears that the mossy cells pass on the collective output of
of kindling and some pathological conditions, such as granule cells from one septotemporal level to granule cells
epilepsy, mossy bers can be induced to sprout into the located at distant levels of the dentate gyrus.
molecular layer. The full impact of this longitudinal organization of the
associational projection cannot be fully appreciated without
Mossy Cell Projection to the Associational/Commissural Zone one further piece of information: The associational bers con-
The inner third of the molecular layer receives a projection tact not only the spines of the granule cell dendrites but also
that originates exclusively from neurons in the polymorphic the dendritic shafts of the GABAergic basket cells, which in
layer (Laurberg and Sorensen, 1981; Buckmaster et al., 1992, turn innervate the granule cells. Thus, the associational pro-
1996; Frotscher, 1992). Because in the rat this projection orig- jection may function both as a feedforward excitatory path-
inates in both the ipsilateral and contralateral sides of the hip- way to distant granule cells and as a disynaptic feedforward
pocampus, it has been called the associational/commissural inhibitory pathway, with the pyramidal basket cell as the inter-
projection. This projection was initially thought to arise from mediary.
both the cells of the polymorphic layer and the CA3 pyrami-
dal cells located within the connes of the dentate gyrus. If GABA/Somatostatin Projection from the Polymorphic
this were true, it would mean that at least one of the intrinsic Layer to the Outer Molecular Layer
hippocampal connections would be bidirectional. It is now Although the neuroanatomy of the hippocampal formation
clear, however, that the associational/commissural projection has been analyzed for many years, new components of its
arises exclusively from cells in the polymorphic layer, and CA3 intrinsic circuitry continue to be discovered. This is often the
cells do not project to the molecular layer. Although this state- result of applying new neuroanatomical techniques. One
ment is generally true, it appears that at least some CA3 neu- example is the discovery of a second projection from the poly-
rons in the most temporal extreme of the hippocampal morphic layer to the molecular layer. Antibodies directed
formation might send collaterals into the molecular layer of against the peptide somatostatin have revealed that neurons
the dentate gyrus. Aside from the existence of this projection, scattered throughout the polymorphic layer are immunoreac-
almost nothing is known concerning its magnitude. Nor is it tive for this peptide and account for approximately 16% of the
known why this projection is observed only in the temporal GABAergic cells in the dentate gyrus (Morrison et al., 1982;
portion of the hippocampal formation. Thus, the principle of Bakst et al., 1986; Freund and Buzsaki, 1996; Sik et al., 1997;
unidirectionality, although generally true, may have excep- Boyett and Buckmaster, 2001). As noted earlier, the somato-
tions to the rule. For all practical purposes, the organization of statin immunoreactive cells may correspond, in part, to the
the commissural projection is similar to that of the ipsilateral HIPP cells. The somatostatin-positive cells all colocalize with
associational connection. GABA and are the source of the somatostatin immunoreactive
64 The Hippocampus Book

bers and terminals in the outer two-thirds of the molecular intra- and infrapyramidal bundles are largely eliminated, and
layer. This system of bers, which forms contacts on the distal virtually all mossy bers travel in the stratum lucidum.
dendrites of the granule cells, provides a third means of releas- Although our focus is primarily on human, monkey, and rat
ing inhibitory control of granule cell activity (in addition to hippocampus in this chapter, it is worth noting in passing that
the basket cell plexus and the axo-axonic terminals provided variations in the relative size of the infrapyramidal mossy
by the chandelier cells). Because electron microscopic studies bers have been noted across different strains of mice, includ-
have demonstrated that somatostatin cells are contacted by ing mouse strains used widely in transgenic experiments; this
mossy ber terminals, the projection to the outer molecular variation sometimes correlates with different behavioral pro-
layer thus constitutes a local feedback inhibitory circuit. les (Lipp et al., 1987; Hausheer-Zarmakupi et al., 1996).
Interestingly, unlike the mossy cell associational projec- Granule cells at all transverse positions in the granule
tion, which terminates more heavily at distant levels of the cell layer generate mossy bers that extend for the full prox-
dentate gyrus, the GABA/somatostatin projection terminates imo-distal distance of CA3 (Fig. 323). Cells located in the
most heavily at the level of the cells of origin; moreover, ter- infrapyramidal blade of the granule cell layer have axons that
mination rapidly decreases within approximately 1.5 mm sep- tend to enter CA3 in the infrapyramidal bundle but ultimately
tally and temporally to the cells of origin. Thus, the mossy cell cross the pyramidal cell layer to enter the deep portion of the
projection and the somatostatin/GABA cell projection have stratum lucidum. The axons of granule cells located in the
terminal elds that are spatially complementary in both radial crest of the dentate gyrus tend to enter CA3 in the intrapyra-
and horizontal axes; the distribution suggests that the two cell midal bundle and also ultimately ascend into the stratum
types mediate distal excitation and local inhibition, respec- lucidum. Cells located in the suprapyramidal blade of the
tively. dentate gyrus give rise to axons that enter CA3 in the stratum
lucidum and continue within the most supercial portion of
Dentate Gyrus Efferent Projection: Mossy Fibers the stratum lucidum (Claiborne et al., 1986).
The mossy bers give rise to unique, complex en passant
The dentate gyrus does not project to any brain region other presynaptic terminals called mossy ber expansions (Amaral
than the CA3 eld of the hippocampus. The axons that and Dent, 1981). Part of the uniqueness of these terminals is
project to CA3, the mossy bers, arise exclusively from the their size; they can be as large as 8 m in diameter but more
granule cells and terminate in a relatively narrow zone mainly typically range from 3 to 5 m in greatest dimension. Their
located just above the CA3 pyramidal cell layer (Blackstad et large size has attracted the attention of physiologists interested
al., 1970; Gaarskjaer, 1978b; Swanson et al., 1978; Claiborne et in using them for patch-clamp studies of transmitter release
al., 1986). In the proximal portion of CA3, mossy bers are (Henze et al., 2000). The mossy ber expansions form highly
also located below and within the pyramidal cell layer. The irregular, complex, interdigitated attachments with the intri-
layer of mossy ber termination located just above the pyram- cately branched spines called thorny excrescences that are
idal cell layer is called the stratum lucidum because the lack of located on the proximal dendrites of the CA3 pyramidal cells.
myelin on the mossy bers gives the layer a relatively clear The thorny excresences are so distinctive they clearly mark the
appearance in fresh tissue (as one might visualize in a hip- location of mossy ber synaptic termination. In the proximal
pocampal slice experiment). There is no indication that den- portion of CA3, for example, thorny excrescences are located
tate neurons other than the granule cells project to CA3; in on both the basal and apical proximal dendrites of pyramidal
particular, cells in the polymorphic layer do not project to the cells, which are therefore in contact with both the infra- and
hippocampus, at least in the rodent. The dentate projection to suprapyramidal mossy ber bundles. In the mid and distal
CA3 stops precisely at the border of CA3 with CA2, and the portions of CA3, however, thorny excresences are almost
lack of granule cell input is one of the main features that dis- entirely restricted to the apical dendritic processes that
tinguishes CA3 from CA2 pyramidal cells. traverse the stratum lucidum (Fig. 324). The CA2 pyramidal
All dentate granule cells project to CA3, and the axon tra- cells, which do not receive any mossy ber input, are devoid of
jectory is partially correlated with the position of the parent thorny excrescences (Fig. 325).
cell body. Before describing features of the mossy ber projec- Another distinctive feature of the mossy ber expansion is
tion, it is worth making a few points concerning the organiza- the number of active synaptic zones they demonstrate. A sin-
tion of the terminal regions in CA3. In the proximal portion gle mossy ber expansion can make as many as 37 synaptic
of CA3 (close to the dentate gyrus), mossy bers are distrib- contacts with a single CA3 pyramidal cell dendrite. Three-
uted below, within, and above the pyramidal cell layer. The dimensional analysis of serial sections through these synapses
bers located below the layer (i.e., those that are in the area indicates that although a mossy ber expansion may be in
occupied primarily by basal dendrites) are generally called the synaptic contact with more than one complex spine originat-
infrapyramidal bundle. The bers located in the pyramidal ing from the same parent dendrite it does not typically contact
cell layer are called the intrapyramidal bundle, and those spines on two different dendrites. Thus, one mossy ber
located above the pyramidal cell layer (i.e., in the area occu- expansion does not typically contact two pyramidal cells.
pied mainly by proximal apical dendrites) are called the The large mossy ber expansions occur approximately
suprapyramidal bundle; the suprapyramidal bundle occupies every 135 m along the parent axon (Fig. 323), and each
the stratum lucidum. At mid and distal portions of CA3, the mossy ber axon forms about 15 of these complex boutons.
Hippocampal Neuroanatomy 65

Figure 323. Topography of the mossy bers in the CA3 region. A. fourth granule cell located posteriorly. The original coronal images
Camera lucida drawing of mossy bers of three adjacent granule are rotated to emphasize the spatial characteristics of the bers.
cells (truncated). Note the numerous lopodial extensions of the C. Wire diagram of the three mossy bers shown in A, depicting
large mossy terminals in A (arrowheads) and thin stalks of large the distribution of mossy terminals. Note that a shorter interbouton
mossy terminals (arrows). Boxed area in the inset in A shows the distance prevails in the proximal portion of CA3. sl, stratum
position of the bers in CA3. B. Neurolucida reconstruction of the lucidum; pcl, pyramidal cell layer. Bars: A, 50 m; B, 400 m; C,
same three axons shown in A and an additional mossy ber of a 200 m. (Source: Adapted from Acsady et al., 1998.)
66 The Hippocampus Book

Early Golgi anatomists indicated that the mossy ber axons

were mainly oriented perpendicular to the long axis of the
hippocampus. Theodor Blackstad and colleagues conrmed
the largely transverse trajectory of the mossy bers using
degeneration track tracing methods, and Andersen and col-
leagues demonstrated this electrophysiologically. Modern
studies, such as those carried out by Claiborne and colleagues,
also indicate that mossy bers do not collateralize once they
enter the hippocampus; they stay largely within the same sep-
totemporal level as their cells of origin. There is one peculiar-
ity of the mossy ber projection, however, that has eluded
explanation. Lorente de N (1934) and McLardy (McLardy
and Kilmer, 1970) indicated that the mossy bers actually do
change their course and take a longitudinal direction but not
until they reach the distal portion of CA3. The extent of this
distal longitudinal projection was further claried by Swanson
and colleagues using the autoradiographic method of neural
tract tracing (Swanson et al., 1978). They showed that granule
cells located at septal levels give rise to mossy bers that travel
throughout most of the transverse extent of CA3 at the same
septotemporal level. Just at the CA3/CA2 border, however,
they abruptly change course and travel toward the temporal
pole for nearly 2 mm. The extent of this longitudinal compo-
nent appears to depend on the septotemporal location of the
cells of origin. Granule cells in the mid to temporal portions
of the dentate gyrus have mossy bers that exhibit only a
slight temporal inclination at their distal extremity; mossy
bers that originate at the extreme temporal pole of the den-
tate gyrus barely extend to the CA3/CA2 border and have lit-
tle or no longitudinal component. Thus, in the septal half of
the hippocampus, there appears to be some overlap of mossy
Figure 324. Camera lucida drawing of a CA3 pyramidal neuron bers originating from different septotemporal levels, but it
located in the midportion of the eld. As this neuron lies outside occurs only at a restricted distal portion of the CA3 eld. On
the zone of the infrapyramidal mossy ber bundle, most of the
the face of it, this indicates that some CA3 pyramidal cells
thorny excrescences are located on the proximal apical dendrites.
located extremely close to the border with CA2 may be con-
The axon of this neuron is indicated by an arrowhead. pcl, pyrami-
dal cell layer; sl, stratum lucidum; so, stratum oriens; sr, stratum
tacted by mossy ber axons from granule cells spread out over
radiatum. Bar  100 m. (Source: Adapted from Ishizuka et al., a much broader septotemporal extent of the dentate gyrus.
1995.) They might therefore form a special class of integrator CA3
cells. Elegant neuroanatomical verication of this has come
from the work of Acsady and colleagues (1998), who stained
Thus, each granule cell communicates with only 15 CA3 dentate granule cells using in vivo intracellular techniques.
pyramidal cells. It is important to point out, however, that Not only do mossy bers travel 2 to 3 mm in the longitudial
these 15 pyramidal cells are distributed throughout the full direction, they demonstrate typical mossy ber expansions
proximo-distal length (transverse axis) of CA3. Because there along this portion of their trajectory; these terminals do, in
are approximately 2.5  105 CA3 pyramidal cells and approx- fact, make contact with thorny excrescences.
imately 1.2  106 granule cells, each pyramidal cell is expected Before leaving this description of the mossy bers, it
to receive input from about 72 granule cells. That is, an indi- should be noted that in addition to the mossy ber expansions
vidual dentate granule cell has the potential to activate about there are infrequent thin collaterals that emanate from the
15 CA3 pyramidal cells, whereas an individual pyramidal cell parent axon or from the mossy ber expansions. These collat-
receives input from about 72 granule cells. This pattern, not erals appear to terminate preferentially on cells other than the
seen elsewhere in the brain, has attracted considerable interest pyramidal cells, such as the GABAergic pyramidal basket cells
among computational neuroscientists. The fact that each of in CA3.
the mossy ber expansions has many release sites might There is substantial evidence indicating that the granule
ensure highly efficient depolarization of the innervated cells use glutamate as their primary transmitter, and the asym-
pyramidal cells. This is, however, currently controversial (see metrical contacts made between the mossy ber expansions
Chapters 5 and 6). and the thorny excrescences tend to conrm this notion.
Hippocampal Neuroanatomy 67

Figure 325. Camera lucida drawing

of a CA2 pyramidal neuron. Note that
although the size and general characteris-
tics of this neuron are similar to those in
CA3, there are no thorny excrescences on
the proximal apical dendrites. Two basal
dendrites cut at the surface of the slice
are indicated by asterisks. The axon of
this neuron is indicated by an arrowhead.
CA1, CA1 eld of the hippocampus;
CA3, CA3 eld of the hippocampus; hf,
hippocampal ssure; pcl, pyramidal cell
layer; sl, stratum lucidum; sl-m, stratum
lacunosum-moleculare; so, stratum
oriens; sr, stratum radiatum. Bar  100
m. (Source: Adapted from Ishizuka et
al., 1995.)

Nonetheless, the mossy bers are also immunoreactive for narrow, relatively cell-free layer located deep to the pyramidal
several other neuroactive substances. At least some of the cell layer is called the stratum oriens. This layer contains the
mossy bers demonstrate immunoreactivity for the opioid basal dendrites of the pyramidal cells and several classes of
peptide dynorphin and are also immunoreactive for GABA interneurons. Stratum oriens can be dened as the infrapyra-
(Walker et al., 2002). Interestingly, although the granule cells midal region in which some of the CA3 to CA3 associational
do not normally demonstrate mRNA for the synthetic connections and the CA3 to CA1 Schaffer collateral connec-
enzymes GAD65 or GAD67, prolonged stimulation of the tions are located. Deep to the stratum oriens is the thin, ber-
perforant path can induce GAD messenger expression in rat containing alveus. In the CA3 eld, but not in CA2 or CA1, a
granule cells. The possible role of GABA or opioid peptides in narrow acellular zone, the stratum lucidum, is located just
the synaptic economy of the mossy bers is discussed in rela- above the pyramidal cell layer and is occupied by the mossy
tion to synaptic plasticity at these synapses in Chapter 10. bers. There is a slight thickening of the stratum lucidum at
its distal end, where, at least at septal levels of the hippocam-
3.4.2 Hippocampus pus, the mossy bers bend temporally and travel longitudi-
nally. This zone, called the end bulb, is more prominent in
Cytoarchitectonic Organization species such as the guinea pig than in the rat; it marks the
CA3/CA2 border. The stratum radiatum is located supercial
An overview of the major elds of the hippocampus was given to the stratum lucidum in CA3 and immediately above the
earlier in the chapter. We now go into more detail about its pyramidal cell layer in CA2 and CA1. The stratum radiatum
laminar organization, which is generally similar for all the can be dened as the suprapyramidal region in which the
elds of the hippocampus. The principal cellular layer is called CA3 to CA3 associational connections and the CA3 to
the pyramidal cell layer. The pyramidal cell layer is tightly CA1 Schaffer collateral connections are located. The most
packed in CA1 and more loosely packed in CA2 and CA3. The supercial layer of the hippocampus is called the stratum
68 The Hippocampus Book

lacunosum-moleculare. It is in this layer that bers from the

entorhinal cortex terminate. Afferents from other regions,
such as the nucleus reuniens of the midline thalamus, also
terminate in the stratum lacunosum-moleculare. Both the
stratum radiatum and the stratum lacunosum-moleculare
contain a variety of interneurons.
In the following sections, we sometimes deal with the
organization of the CA3 and CA2 elds of the hippocampus
rst and then move on to the CA1 eld. For other topics, such
as the interneurons of the hippocampus, there is not enough
known to warrant discussing CA3, CA2, and CA1 separately.
For these topics, we discuss data for the entire hippocampus.
Moving on from the dentate gyrus, we begin our discussion
with the CA3 eld.

Pyramidal Cells of CA3 and CA2

The principal neuronal cell type of the hippocampus is the

pyramidal cell, which makes up most of the neurons in the
pyramidal cell layer. Pyramidal cells have a basal dendritic tree
that extends into the stratum oriens and an apical dendritic
tree that extends to the hippocampal ssure.
Ishizuka and colleagues (Ishizuka et al., 1995) demon-
strated that the dendritic length and organization of CA3
pyramidal cells are quite variable (Figs. 324 and 327). The
smallest cells (with a soma size of about 300 m2 or 20 m in
diameter) are located in the limbs of the dentate gyrus and
have a total dendritic length of 8 to 10 mm. The largest cells
(with a soma size of about 700 m2 or 30 m in diameter),
which are located distally in the eld, have total dendritic
Figure 326. Camera lucida drawing of a CA1 pyramidal neuron
lengths of 16 to 18 mm. The distribution of the dendritic trees from the midportion of the eld. Note that side branches originate
of CA3 pyramidal cells also varies depending on where the cell from the primary dendrites throughout the full extent of the stra-
body is located. Cells located in the limbs of the dentate gyrus, tum radiatum. Note also the curved and irregular trajectories of
for example, have few or none of their dendrites extending dendritic branches in the stratum lacunosum-moleculare. The axon
into the stratum lacunosum-moleculare, and thus these cells of this neuron is indicated by an arrowhead. pcl, pyramidal cell
receive little or no direct input from the entorhinal cortex. The layer; sl-m, stratum lacunosum-moleculare; so, stratum oriens; sr,
cells, however, receive a larger number of mossy ber termi- stratum radiatum. Bar  100 m. (Source: Adapted from Ishizuka
nals on their apical and basal dendritic trees and are thus et al., 1995.)
under greater inuence of the granule cells than distally
located CA3 cells, which receive only apical mossy ber input. dendrite, and others have two. Cells with two apical dendrites
tend to have slightly greater total dendritic length in the apical
Pyramidal Cells of CA1 direction. Neurons with a single apical dendrite, however, tend
to have slightly larger basal dendritic trees; thus, overall, the
In contrast to the substantial heterogeneity of dendritic dendritic tree in all CA1 neurons have about the same total
organization characteristic of CA3 pyramidal cells, investiga- length. This anatomical homogeneity, however, cannot reect
tors such as Norio Ishizuka and Dennis Turner and colleagues functional homogeneity because, as we shall see shortly, there
(Pyapali et al., 1998) demonstrated that the CA1 pyramidal are differences in the entorhinal cortex inputs received at dif-
cells show remarkable homogeneity of their dendritic trees ferent locations along the transverse axis of CA1.
(Figs. 326 and 327). As well as being more homogeneous,
they are also, on average, smaller than CA3 cells. The total Interneurons
dendritic length averages approximately 13.5 mm, and the
average size of CA1 cell somata is about 193 m2 or 15 m in Pyramidal neurons are by far the most numerous neurons in
diameter. Regardless of where a pyramidal cell is located in the hippocampus. As in the dentate gyrus, however, there is a
CA1, it has about the same total dendritic length and the same fairly heterogeneous group of interneurons that are scattered
dendritic conguration. Some pyramidal cells have one apical through all layers (Fig. 328) (Freund and Buzsaki, 1996).
Hippocampal Neuroanatomy 69

Figure 327. Summary of the organization

of hippocampal pyramidal cells produced as
a composite of computer-generated line
drawings of neurons from CA3, CA2, and
CA1. Dashed line in CA3 marks the region
occupied by the infrapyramidal and
suprapyramidal mossy bers. Source:
(Adapted from Ishizuka et al., 1995.)

Figure 328. Morphological classication of the interneurons in ron groups. The laminar distribution of the inputs, which often
the hippocampus proper. Filled circles indicate the location of the show correspondence with the interneuron type or axon distribu-
cell bodies, and thick lines indicate the predominant orientation tion, is also indicated. pcl, pyramidal cell layer; sl-m, stratum
and laminar distribution of the dendritic tree. The pyramidal cells lacunosum-moleculare; so, stratum oriens; sr, stratum radiatum.
(principal neurons) are illustrated in the background, providing an (Source: Adapted from Freund and Buzsaki, 1996.)
indication of which domain is innervated by the various interneu-
70 The Hippocampus Book

Most types of interneuron are found in all the hippocampal located in the zones occupied by recurrent pyramidal cell col-
subelds. The pyramidal basket cell resides in or close to laterals. In CA3 this includes all strata except the stratum
the pyramidal cell layer, and its dendrites extend into the lacunosum-moleculare, but in CA1 it includes only the stra-
stratum oriens, stratum radiatum, and stratum lacunosum- tum oriens. The axons of the O-LM cell leave the stratum
moleculare. The dendrites are beaded and aspiny, and they oriens (or in whichever layer the cell body is located) and rise
receive both asymmetrical and symmetrical synapses. Most of directly to the stratum lacunosum-moleculare, ramifying
the excitatory inputs are known to arise from hippocampal there to form a dense plexus. These axons form symmetrical
pyramidal cells. In fact, the dendritic tree of a pyramidal bas- synapses with the distal apical dendrites of pyramidal neu-
ket cell receives at least 2000 excitatory inputs. Because each rons. Because most of the excitatory input to the O-LM cells
pyramidal cell contributes only a single synapse to a particu- appears to arise from recurrent collaterals of the pyramidal
lar basket cell, the degree of pyramidal cell convergence on an cells, this class of interneuron inhibits activity in the distal
individual basket cell is enormous. Neurons with basket cell- dendrites of pyramidal cells in a disynaptic, feedback manner.
like axons have a variety of morphologies. There are fusiform- Another class of hippocampal interneuron, the bistratied
shaped basket cells in the stratum oriens and stellate-shaped cell, also has its cell body located close to the pyramidal cell
basket cells in the stratum radiatum. In all cases, the axons of layer. The dendritic trees of these neurons are multipolar but
these cell types innervate the soma and proximal dendrites of do not reach the stratum lacunosum-moleculare. The axon of
the pyramidal cells. The transverse extent of the basket cell the bistratied cell sends collaterals into the stratum oriens
axonal plexus (in a 400-m in vitro slice preparation) is and the deep portion of the stratum radiatum, where a dense
between 900 and 1300 m. Within this plexus, there are as terminal plexus is produced. These neurons generate an enor-
many as 10,000 synaptic varicosities; and because a basket cell mous axonal plexus, on the order of 80 mm in total length and
makes only 2 to 10 synapses on each pyramidal cell, a typical generating up to 16,000 synaptic varicosities. The bistratied
basket cell innervates as many as 1000-plus pyramidal cells. A cells have axons that terminate on both the dendritic shafts
single basket cell could thus have substantial inhibitory inu- and the dendritic spines of pyramidal cells. Although the
ence over a large population of pyramidal cells. inputs to these cells have not been thoroughly investigated,
A second type of hippocampal interneuron is the chande- their dendrites reside in the zone of associational connections
lier, or axo-axonic, cell. The axo-axonic cells found in the hip- in CA3 and the Schaffer collateral bers in CA1. It is likely,
pocampus are similar to the ones described in the dentate therefore, that they are driven in both a feedforward and feed-
gyrus. Their cell bodies, like those of the basket cells, are back manner.
located in or adjacent to the pyramidal cell layer, and their There are other interneurons located in the stratum radia-
dendrites span all the hippocampal strata. The axons of the tum, and their stellate or multipolar dendritic plexus is con-
chandelier cells have a transverse spread of approximately 1 ned to the layer. The axons of these cells tend to ramify
mm. They travel just supercial to the pyramidal cell layer and locally in the stratum radiatum and terminate primarily on
periodically give rise to collaterals that enter the pyramidal the dendrites of pyramidal cells.
cell layer and terminate on the proximal axons of the pyrami- A fairly sizable population of interneurons is located in the
dal neurons. Each axo-axonic cell terminates on approxi- stratum lacunosum-moleculare or at the border between the
mately 1200 pyramidal cell axon initial segments, and each stratum lacunosum-moleculare and the stratum radiatum
initial segment is innervated by 4 to 10 axo-axonic cells. (Lacaille and Schwartzkroin, 1988). These LM neurons (stra-
The early Golgi studies of Ramon y Cajal and Lorente de tum lacunosum-moleculare interneurons) have dendrites that
N made it abundantly clear that there are a variety of non- are oriented horizontally (i.e., within the layer) but occasion-
pyramidal cell types in the stratum oriens, stratum radiatum, ally have branches that extend into the pyramidal cell layer.
and stratum lacunosum-moleculare of the hippocampus. The axon also takes a predominantly horizontal orientation
Ribak and colleagues (1978) were the rst to discover that and ramies mainly in the stratum lacunosum-moleculare or
most of these neurons are immunoreactive for GABAergic the supercial portion of the stratum radiatum. Although the
markers, and most are thought to be interneurons. Freund connectivity of this class of neurons is not well worked out,
and Buzsaki (1996) described the location, dendritic organi- their axons do form symmetrical synapses on the distal den-
zation, and axonal distribution of these cells based on a classi- drites of pyramidal cells.
cation system that is dependent on the region of innervation. An additional type of interneuron in the hippocampus is
One class of cells (Lacaille et al., 1987) has been called the the IS neuron (interneuron-selective). The IS neurons cell
O-LM cell (oriens lacunosum-moleclare(associated cell) and bodies are located in all layers and can be identied by stain-
has as its dening feature a dense axonal arbor that is conned ing with antibodies to the calcium-binding protein calretinin.
to the stratum lacunosum-moleclare (also known as cells The IS cells have a number of notable features, including the
terminating in conjunction with entorhinal afferents). The propensity for their dendrites to form bundles with dendrites
location of the cell body of this class of interneuron varies of other IS neurons. The major unifying feature, however, is
depending on which hippocampal eld it inhabits. The prin- that their axons terminate exclusively on other interneurons.
ciple seems to be that the cell body and dendritic tree are Little is yet known concerning their input/output characteris-
Hippocampal Neuroanatomy 71

tics. Their dendrites, however, are disposed so they could be originate from cells in layer II, and collaterals of the same layer
innervated by all major afferent and intrinsic connections of II cells reach both the dentate gyrus and CA3/CA2, implying
the hippocampus, and their axons could potentially innervate that similar information reaches these structures.
all of the interneurons described above.
All of the interneurons we have described are immunopos- Entorhinal Cortex Projection to CA1
itive for markers of GABA. As in other cortical regions, the Although the entorhinal cortex also projects to the CA1 eld
interneurons of the hippocampus colocalize a number of of the hippocampus, the organization of this projection is
other neuroactive substances. Thus, many of these neurons fundamentally different from the projection to CA3/CA2.
can be visualized with antibodies to peptides such as somato- First, the cells of origin are in layer III rather than in layer II.
statin, VIP, cholecystokinin, neuropeptide Y (NPY), and cal- Second, the pattern of terminal distribution is organized
cium-binding proteins such as parvalbumin, calbindin, and in a topographical fashion rather than in a laminar fashion
calretinin. It remains something of a mystery what these (Fig. 329). As in CA3/CA2, the entorhinal bers terminate
neuroactive substances are doing in GABAergic neurons, throughout the full width of the stratum lacunosum-molecu-
but at the very least they provide useful markers for establish- lare of CA1. However, bers originating in the lateral entorhi-
ing subcategories of the large population of GABAergic nal area terminate in the distal portion of CA1 (close to the
interneurons. subiculum), whereas bers originating in the medial entorhi-
nal area terminate in the proximal portion of CA1 (close to
Extrinsic Connections CA2). Thus, depending on where a CA1 pyramidal cell is
located in the transverse axis of the hippocampus, it receives
One of the distinguishing features of the connectivity of the inputs from a different portion of the entorhinal cortex. We
hippocampus is that most of its synaptic input arises from noted earlier that CA1 cells are remarkably homogeneous
within its own boundaries. CA3 and CA2 are heavily inner- in their dendritic length; the inputs they receive, however,
vated by collaterals of their own axons (i.e., associational con- differ as a function of their location along the proximo-
nections) and from axons of the contralateral CA3 and CA2 distal axis.
(i.e., commissural connections). CA1, in turn, receives its
heaviest input from CA3. A relatively lighter projection arises Hippocampal Projections to the Entorhinal Cortex
from the entorhinal cortex and terminates on the most distal Early studies using autoradiographic tract-tracing techniques
dendrites of the pyramidal cells as well as on interneurons suggested that all elds of the hippocampus send a return pro-
with dendrites in the stratum lacunosum-moleculare. There jection to the entorhinal cortex, but it is now clear that only
are relatively few other extrinsic inputs to the hippocampus, cells located in CA1 give rise to this projection (Naber et al.,
and the ones that do exist generally account for a relatively 2001). The projecting cells appear to send their axons to
small number of synapses. roughly the same region of the entorhinal cortex from which
they receive their input (Fig. 329). Thus, proximal CA1 cells
Entorhinal Cortex Projection to CA3/CA2 project to the medial entorhinal cortex, whereas distal CA1
Although the entorhinal innervation of CA3 is mentioned in cells project to the lateral entorhinal area. We return to a more
most studies of the perforant path projection, the organization detailed description of these projections in Section 3.4.5.
of this component of the projection is generally not dealt with
in much detail. Indeed, there has been little in-depth research Hippocampal Connections with the Neocortex
on the entorhinal projections to the hippocampus despite the and Amygdaloid Complex
fact that its prominence suggests that it is extremely important. Although this book focuses on the hippocampal formation,
The perforant path takes its name from the observation that it we repeatedly emphasize that no brain structure can be seen
perforates the subiculum and hippocampal ssure en route to in isolation. The hippocampus sends projections to and
the dentate gyrus, but it is now quite clear that collaterals of receives projections from numerous other brain regions, and
bers that project to the dentate gyrus also project to the hip- these interconnections are vital to understanding its function.
pocampus. In fact, the origin and laminar terminal distribu- Having said that, and notwithstanding the extremely impor-
tion of the perforant path projection to CA3 are similar to tant functional relationship between hippocampus and neo-
those to the dentate gyrus (Witter, 1993). Entorhinal terminals cortex, it turns out that only selected parts of the hippocampal
are distributed throughout the width of the stratum lacuno- formation have discrete, monosynaptic connections with the
sum-moleculare. As with the projection to the molecular layer neocortex. The CA3 and CA2 elds of the hippocampus, for
of the dentate gyrus, projections from the lateral entorhinal example, have no known connections with the neocortex.
area terminate supercially in the stratum lacunosum-molec- Sensitive tracing methods have recently shown that CA3, in
ulare, and those from the medial entorhinal area terminate in particular its temporal parts, receives input from the amyg-
the deep half of the layer. The laminar origin, types, and num- daloid complex, which was previously thought to send projec-
bers of synaptic contacts of this projection are similar to those tions only to CA1 and the subiculum (Pikkarainen et al., 1999;
to the dentate gyrus. The entorhinal projections to CA3/CA2 Pitkanen et al., 2000). These inputs originate mainly from the
72 The Hippocampus Book

Figure 329. Reciprocal entorhinalhippocampal connections. A. injections into the lateral entorhinal cortex and the distal portion of
Distribution of labeled bers in the hippocampus and entorhinal the CA1 eld of the hippocampus. The terminal bers originating
cortex after combined anterograde tracer injections into the medial in the entorhinal cortex overlap with the injection site in CA1, and
portion of the entorhinal cortex and the proximal portion of the the terminal bers originating in CA1 overlap with the injection site
CA1 eld of the hippocampus. The terminal bers originating in in the entorhinal cortex. The subdivisions of the entorhinal cortex
the entorhinal cortex overlap with the injection site in CA1, and the follow the nomenclature of Insausti et al. (1997). Areas CE and ME
terminal bers originating in CA1 overlap with the injection site in constitute the medial entorhinal cortex, whereas DLE, DIE, VIE, and
the entorhinal cortex. B. Distribution of labeled bers in the hip- AE constitute the lateral entorhinal cortex. (Source: Adapted from
pocampus and entorhinal cortex after combined anterograde tracer Naber et al., 2001.)

caudomedial portion of the parvicellular division of the basal stratum lacunosum-moleculare and overlaps the projection
nucleus and terminate heavily in the stratum oriens and stra- arising from the lateral entorhinal area. The same CA1 cells
tum radiatum. The best-documented cortical connection, give rise to a return projection to the perirhinal cortex. CA1
other than with the entorhinal cortex, which we have dened cells located in the septal portion of the hippocampus have
as part of the hippocampal formation, is with the perirhinal also been reported to project to the retrosplenial cortex, and
(areas 35 and 36) and postrhinal cortices. Cells in the perirhi- those located at mid-septotemporal levels provide a fairly sub-
nal cortex give rise to a relatively selective projection to the stantial projection to the medial frontal lobe.
most distal CA1 pyramidal cells (i.e., those located at the bor- The temporal two-thirds of the distal portion of CA1 is
der with the subiculum). This projection terminates in the reciprocally connected with the amygdaloid complex.
Hippocampal Neuroanatomy 73

Projections from the basal nucleus of the amygdala terminate The suparamammillary region projects weakly, if at all, to
in the stratum oriens and stratum radiatum of the CA1/ CA3 and CA1.
subiculum border region. In addition, the accessory basal and
cortical nuclei projects to the stratum lacunosum-moleculare. Thalamic Connections: Nucleus Reuniens
CA1 cells in the same region give rise to a return projection to and Other Midline Nuclei
the basal nucleus of the amygdala. The thalamic inputs to the hippocampal formation have
received relatively little attention. It has been known for some
Basal Forebrain Connections time that the anterior thalamic complex is intimately inter-
The septum provides the major subcortical input to CA3. As connected with the presubiculum. However, Herkenham and
with the dentate gyrus, the septal projection originates mainly others demonstrated fairly prominent projections from mid-
in the medial septal nucleus and the nucleus of the diagonal line (nonspecic) regions of the thalamus to several elds of
band of Broca. The projection terminates most heavily in the the hippocampal formation (Herkenham, 1978; Wouterlood et
stratum oriens and to a lesser extent in the stratum radiatum al., 1990; Dolleman-Van der Weel and Witter, 2000). The
(Fig. 320). The CA1 eld receives a substantially lighter sep- nucleus reuniens, located on the midline, gives rise to a promi-
tal projection than CA3, but the bers are also most densely nent projection to the stratum lacunosum-moleculare of CA1,
distributed in the stratum oriens. where it overlaps with bers from the entorhinal cortex. The
Until the mid-1970s, it was commonly assumed that the nucleus reuniens projection travels to CA1 via the internal
hippocampal elds gave rise to both the precommissural and capsule and cingulum bundle rather than through the fornix
postcommissural projections to the basal forebrain and dien- and mbria. This projection innervates all septotemporal lev-
cephalon. Indeed, Ramon y Cajals classic diagram of the els of CA1, with a preference for the mid-septotemporal levels.
hippocampus shows bers from CA1 coursing toward the The nucleus reuniens bers terminate with asymmetrical
mbria. Swanson and Cowan demonstrated, however, that synapses on spines and thin dendritic shafts in the stratum
most of these projections originate from the subiculum lacunosum-moleculare on both principal neurons and
(Swanson and Cowan, 1975). It is now quite clear that the only GABAergic interneurons.
sizable subcortical projection from CA3 is to the lateral septal
nucleus. The CA3 projection to the lateral septal nucleus trav- Brain Stem Inputs
els via the mbria and precommissural fornix. The CA3 The hippocampus, like the dentate gyrus, receives noradren-
projection to the septal complex is bilateral; and some CA3 ergic and serotonergic inputs from brain stem nuclei (Fig.
bers cross in the ventral hippocampal commissure to inner- 321). Noradrenergic bers and terminals arising from the
vate the homologous region of the contralateral lateral septal locus coeruleus are most densely distributed in the stratum
nucleus. This pathway is topographically organized such that lucidum and the most supercial portion of stratum lacuno-
septal portions of CA3 project dorsally in the lateral septal sum-moleculare. A much thinner plexus of axons is distrib-
nucleus, and progressively more temporal portions of CA3 uted throughout the other layers of CA3. Serotonergic bers
project more ventrally; proximal CA3 cells tend to project are distributed more diffusely and sparsely in CA3 than in the
medially in the lateral septal nucleus, and distally situated CA3 noradrenergic bers. Despite the rather low number of bers,
cells terminate more laterally. Interestingly, virtually all CA3 the serotonergic innervation of the hippocampus demon-
cells give rise to projections to both CA1 and the lateral septal strates several interesting features. First, there are two calibers
nucleus. of axon, thick and thin, arising from the raphe nuclei that
Essentially all CA1 pyramidal cells also project to the lateral innervate the hippocampus. Most of the serotonergic vari-
septal nucleus. However, the CA1 projection is strictly ipsilat- cosities, which are located on the thin bers, do not appear to
eral, and some of the bers travel to the septal nuclei via the have standard synaptic junctions and may release transmitter
dorsal fornix rather than through the mbria. into the extracellular space. The varicosities on the thicker
bers, in contrast, form standard asymmetrical synapses that
Hypothalamic Connections preferentially terminate on GABAergic inhibitory neurons,
There has been little work dealing specically with the extrin- specically on the classes of interneuron that project to the
sic inputs and outputs of CA2. In general, CA2 appears to dendrites of hippocampal neurons. Thus, even the relatively
have the same connections as CA3. However, the CA2 eld few serotonergic bers that innervate the hippocampus
receives particularly prominent innervation from the poste- may have a profound action by enhancing the GABAergic
rior hypothalamus, in particular from the supramammillary inhibitory activity of the hippocampal interneurons. There
area (Fig. 320B) and the tuberomammillary nucleus. These are few, if any, dopaminergic bers in CA3. In general, CA1
projections terminate mainly in and around the pyramidal receives much lighter monoaminergic innervation than CA3.
cell layer and mainly on principal cells (Magloczky et al., The functional implications of monoaminergic inputs to
1994). There is no evidence that CA2 returns the projection to the hippocampus, particularly in relation to long-term poten-
the supramammillary region. In fact, none of the hippocam- tiation (LTP), is considered in more detail in Chapters 5, 6
pal elds appears to project into the postcommissural fornix. and 7.
74 The Hippocampus Book

Intrinsic Connections: CA3 Associational Connections as extending only through the stratum radiatum, it should be
and Schaffer Collaterals emphasized that both the stratum radiatum and stratum
oriens of CA1 are heavily innervated by CA3 axons. Thus, the
As mentioned earlier, the major source of input to the hip- Schaffer collaterals are as highly associated with the apical
pocampus is the hippocampus itself. The CA3 to CA3 associ- dendrites of CA1 cells in the stratum radiatum as they are
ational connections and the CA3 to CA1 Schaffer collateral with the basal dendrites in the stratum oriens. Moreover,
connections are unique in many respects. Perhaps the major CA3 cells located at any particular septotemporal level dis-
distinguishing feature of these projections, however, is their tribute some of their collaterals to much of the full septotem-
extensive spatial distribution. Through these connections a poral extent of CA1. This projection is not chaotic, however,
particular pyramidal cell in CA3 can, in theory, interact with and its topographical organization develops a network in
other hippocampal neurons distributed throughout much of which certain CA3 cells are more likely to contact certain CA1
the ipsilateral and contralateral hippocampus. The massive cells.
potential for association in the hippocampus is undoubtedly The major organizational features of this projection are as
linked to its function. The hippocampal connections, follows: CA3 cells located close to the dentate gyrus (proximal
although widely distributed, are nonetheless systematically CA3), although projecting both septally and temporally, proj-
organized. These projections have been studied by Ishizuka ect more heavily to levels of CA1 located septal to their loca-
and colleagues (Ishizuka et al., 1990) and Buzsaki and col- tion. CA3 cells located closer to CA1, in contrast, project more
leagues (Li et al., 1994). heavily to the levels of CA1 located temporally (Fig. 330). At
All CA3 and CA2 pyramidal cells give rise to highly diver- or close to the septotemporal level of the cells of origin, those
gent projections to all portions of the hippocampus. CA3 cells located proximally in CA3 give rise to collaterals that
pyramidal cells give rise to highly collateralized axons that dis- tend to terminate supercially in the stratum radiatum.
tribute bers both within the ipsilateral hippocampus (to Conversely, cells located more distally in CA3 give rise to pro-
CA3, CA2, and CA1), to the same elds in the contralateral jections that terminate deeper in the stratum radiatum and
hippocampus (the commissural projections), and subcorti- stratum oriens. At or close to the septotemporal level of ori-
cally to the lateral septal nucleus. Some CA3 (especially those gin, CA3 pyramidal cells located near the dentate gyrus tend
located proximally) and CA2 cells contribute a small number to project somewhat more heavily to distal portions of CA1
of collaterals that innervate the polymorphic layer of the den- (near the subiculum), whereas CA3 projections arising from
tate gyrus. Although claims of other hippocampal connec- cells located distally in CA3 terminate more heavily in por-
tions are found in the literature, it is now quite clear that CA3 tions of CA1 located closer to CA2. The truly thick Schaffer
does not project to the subiculum, presubiculum, parasubicu- collaterals (those that Schaffer originally described) originate
lum, or entorhinal cortex. only from the proximal CA3 cells. These cells give rise to a
The CA3 projections to CA3 and CA2 are typically called thick axon that ascends from the stratum oriens into the most
the associational connections, and the CA3 projections to the supercial portion of the stratum radiatum and travels to
CA1 eld are typically called the Schaffer collaterals. This ter- the distal part of CA1, where it contributes many collaterals.
minology, however, may be misleading, as one should remem-
ber that these two projections are true collaterals and thus Figure 330. Organization of the projections from the CA3 eld to
potentially carry the same information. Both the CA3 to CA3 the CA1 eld of the hippocampusthe Schaffer collaterals. The
associational projections and the CA3 to CA1 Schaffer collat- location of the cells of origin is indicated by small triangles in the
erals demonstrate a systematic gradient-like projection pat- middle coronal section. Terminals from these cells are indicated by
tern. Although it is somewhat out of sequence to discuss the different shades of gray similar to those in the triangles.
CA3 to CA1 projections rst, we do so because they have been
worked out in somewhat better detail and provide a clear
model for understanding CA3 projections. Moreover, the
organization of the CA3 to CA1 projection shares many orga-
nizational similarities with the CA3 to CA3 projection.

CA3 to CA1 Connections: Schaffer Collaterals

All portions of CA3 and CA2 project to CA1. The pattern of
terminal distribution, however, depends on the location of the
CA3/CA2 cells of origin. The older notion that a typical CA3
pyramidal cell sends a single axon to CA1, traveling linearly
through the eld with equal contact probability at all regions
in CA1 is clearly incorrect. In fact, each CA3 pyramidal cell
gives rise to highly collateralized axons that follow both trans-
verse and oblique orientations through CA1 (Ishizuka et al.,
1990). Although Schaffer collaterals are typically illustrated
Hippocampal Neuroanatomy 75

The axons of distal CA3 cells tend to be much thinner and to To summarize, the CA3 to CA1 projection is the major
project directly to CA1, either within the stratum oriens or input to CA1 pyramidal cells. The projection terminates on
through the deep portion of the stratum radiatum. the basal dendrites in the stratum oriens and the apical den-
The position of the terminal eld in CA1 varies in a sys- drites in the stratum radiatum. Individual CA3 axons distrib-
tematic fashion relative to the distance from the cells of origin. ute extensively and may innervate neurons throughout as
Regardless of the septotemporal or transverse origin of a pro- much as two-thirds of the entire septotemporal extent of the
jection, the highest density of terminal and ber labeling in hippocampus. The probability that a particular CA1 cell is
CA1 shifts to deeper parts of the stratum radiatum and stra- contacted by a particular CA3 cell depends, in part, on the
tum oriens at levels septal to the cells of origin and shifts away transverse positions of the two cell bodies and their septotem-
from the stratum oriens and into supercial parts of the poral distance. At one particular septotemporal level, a distal
stratum radiatum at levels temporal to the cells of origin. CA3 cell is more likely to interact with a proximal CA1 cell,
Moreover, the highest density of ber and terminal labeling in whereas a proximal CA3 cells is more likely to interact with a
CA1 shifts proximally (toward CA3) at levels septal to the ori- distal CA1 cell. The proximal CA3 cells are the only ones with
gin and distally (toward the subiculum) at levels temporal to classic thick Schaffer collaterals, and the thickness of these ini-
the origin. tial axons is likely to reect the longer distance the axon must
The entire axonal plexus of several CA3 pyramidal cells travel to innervate distal CA1 cells.
have been labeled by intracellular staining techniques devised
by Buzsaki and colleagues (Li et al., 1994). These studies CA3 to CA3 Associational Connections
provide convincing evidence that the axons of individual The associational projections from CA3 to CA3 are also
CA3 cells can distribute to as much as two-thirds of the organized in a highly systematic fashion. One somewhat idio-
septotemporal extent of the ipsilateral and contralateral CA1 syncratic facet of this projection is that cells located proxi-
elds. The plexus from a single CA3 neuron comprises as mally in CA3 communicate only with other proximally
much as 150 to 300 mm of total axonal length, on which located CA3 cells. Associational projections arising from mid
30,000 to 60,000 synaptic varicosities are formed. Although and distal portions of CA3, however, project throughout
single neurons have not been evaluated at all septotemporal much of the transverse extent of CA3 and also project much
levels, it appears that the extent of the CA3 connections is more extensively along the septotemporal axis. The density of
more restricted at temporal levels. Here neurons may give rise CA3 associational projections also shifts along the septotem-
to axonal plexuses that innervate only the temporal third of poral axis. The radial gradient of termination (supercial to
the CA1 eld. deep in the stratum radiatum and stratum oriens) is similar to
A number of pieces of fundamental information concern- that described for the CA3 to CA1 projection. The transverse
ing the CA3 projection to CA1 are still unknown. For exam- gradient, however, is the reverse; CA3 projections shift proxi-
ple, it is not clear how many synapses a single CA3 cell makes mally in CA3 at levels located temporal to the cells of origin
on a typical CA1 cell. To answer this question using neu- and shift distally in CA3 at more septal levels.
roanatomical procedures would be a monumental task. First,
a CA3 and CA1 neuron would have to be impaled with dye- Commissural Connections of the Hippocampus
bearing pipettes, and the two cells would have to be tested for In the rat, the CA3 pyramidal cells give rise to commis-
connectivity by electrophysiological methods. The two cells sural projections to the CA3, CA2, and CA1 elds of the con-
would then have to be lled and processed histologically for tralateral hippocampal formation (Blackstad, 1956; Fricke
visualization. That is the easy part. The neurons would then and Cowan, 1978). In fact, the same CA3 cells give rise to
have to be sectioned serially for electron microscopy and all both the ipsilateral and commissural projections. Although
putative synapses evaluated. Only in this way would one be the commissural projections follow roughly the same topo-
able to determine how many synapses are made by all of the graphical organization as the ipsilateral projections and
collaterals of a particular CA3 neuron on an identied CA1 generally terminate in homologous regions on both sides,
pyramidal cell. A number of laboratories have estimated the there are minor differences in the distribution of terminals.
extent of connectivity between CA3 and CA1 cells, and the If a projection is heavier to the stratum oriens on the ipsilat-
numbers are always quite low. Work from Harris and col- eral side, for example, it may be heavier in the stratum radia-
leagues (Sorra and Harris, 1993), which has been replicated by tum on the contralateral side. The detailed topography of
Trommaid et al. (1996), indicates that there are perhaps as few the commissural connections has not been as thoroughly
as two to four contacts between a single axon in the stratum investigated as the ipsilateral connections. As with the com-
radiatum and a particular dendritic tree and certainly no missural projections from the dentate gyrus, CA3 bers to the
more than 10 synapses between typical CA3 and CA1 neu- contralateral hippocampus form asymmetrical synapses
rons. The surprising state of affairs, however, is that we simply on the spines of pyramidal cells in CA3 and CA1 but also
do not know with certainty how many contacts a single CA3 terminate on the smooth dendrites of interneurons. As noted
neuron makes with a single CA1 cell. These data are critical earlier, hippocampal commissural connections are much less
for interpreting some of the quantal analysis studies described abundant in the monkey and are likely to be absent in
in Chapter 10. humans.
76 The Hippocampus Book

CA1 Lacks Associational Connections

Unlike CA3, pyramidal cells in CA1 give rise to a much more
limited associational projection. As the CA1 axons travel in
the alveus or the stratum oriens toward the subiculum (see
below), occasional collaterals are generated that enter the stra-
tum oriens and the pyramidal cell layer, most likely contacting
interneurons such as basket cells in the stratum oriens and in
turn inhibiting CA1 pyramidal cells. It is also conceivable that
these collaterals might contact the basal dendrites of other
CA1 cells. What is clear, however, is that the massive associa-
tional network, which is so apparent in CA3, is largely missing
in CA1. Similarly, although a weak commissural projection to
the contralateral CA1 appears to be present, there is no exten-
sive commissural projection originating in CA1, as is the case
in CA3. Figure 331. Organization of the projections from the CA1 eld of
the hippocampus to the subiculum. A. Coronal section shows the
cells of origin (triangles) and their respective terminal elds, indi-
CA1 Projection to the Subiculum
cated by similar shades of gray. B. Two-dimensional unfolded map
of CA1 and the subiculum. The septal portion of these elds is at
The CA1 eld gives rise to two intrahippocampal projections.
the top of the gure, and the temporal pole is at the bottom; the
The rst is a topographically organized projection to the adja-
border between CA2 and CA1 is on the left.
cent subiculum. The second is to the deep layers of the
entorhinal cortex. The latter projection is discussed in Section
3.4.5. CA1 (dened as the region that receives CA3 projections) also
Axons of CA1 pyramidal cells descend into the stratum ends at this border and is replaced with the molecular layer of
oriens or the alveus and bend sharply toward the subiculum the subiculum. This layer can be subdivided into a deep por-
(Amaral et al., 1991). The bers reenter the pyramidal cell tion that is continuous with the stratum radiatum of CA1 and
layer of the subiculum and ramify profusely in the pyramidal a supercial portion that is continuous with the stratum
cell layer and the deep portion of the molecular layer. Unlike lacunosum-moleculare of CA1 and the molecular layer of the
the CA3 to CA1 projection, which distributes throughout CA1 presubiculum. The deep portion receives bers from CA1,
in a gradient fashion, the CA1 projection ends in a topo- whereas the supercial portion receives bers from the
graphical fashion in the subiculum. Proximally located CA1 entorhinal cortex. The stratum oriens of CA1 is no longer
cells project to the distal portion of the subiculum, whereas present in the subiculum.
distally located CA1 cells project just across the border into
the proximal portion of the subiculum; the midportion of Neuron Types
CA1 projects to the midportion of the subiculum (Fig. 331).
Nobuaki Tamamaki (Tamamaki et al., 1987) injected horse- Research on the cytology and connectivity of the subiculum
radish peroxidase into single CA1 pyramidal cells and demon- has lagged behind the research on other areas of the hip-
strated that individual axonal plexuses distribute to about pocampal formation; and until recently (Funahashi and
one-third of the transverse extent of the subicular pyramidal Stewart, 1997a,b; Harris et al., 2001) much of the information
cell layer and for approximately one-third of the septotempo- on subicular cell types came from the classic Golgi studies of
ral length. It appears, therefore, that the CA1 to subiculum Ramon y Cajal and Lorente de N in young mice. The princi-
projection segments these structures roughly into thirds. The pal cell layer of the subiculum is populated by large pyramidal
CA1 to subiculum projection, like the CA3 to CA1 projection, neurons (Fig. 332). This layer starts just underneath the dis-
is organized in a divergent fashion, as even a small portion of tal end of CA1 and continues in a position deep to layer II of
CA1 projects to about a third of the septotemporal extent of the presubiculum. These cells are relatively uniform in shape
the subiculum. and size, and they extend their apical dendrites into the
molecular layer and their basal dendrites into deeper portions
3.4.3 Subiculum of the pyramidal cell layer. A subdivision into at least two cell
types has been proposed based on their ring characteristics:
Cytoarchitectonic Organization regular spiking cells and intrinsically bursting cells (Greene
and Totterdell, 1997). Although these two cell types do not
The cytoarchitectonic characteristics of the rat subiculum exhibit distinct morphological characteristics (but see below),
have been studied to a limited extent (OMara et al., 2001). they show a differential distribution in the pyramidal cell
The CA1/subiculum border is marked by abrupt widening of layer. Bursting cells are more numerous deep in the pyramidal
the pyramidal cell layer and an increased staining intensity cell layer, whereas regular spiking cells are more common
with the Timm stain (Fig. 311). The stratum radiatum of supercially. The two populations can also be distinguished by
Hippocampal Neuroanatomy 77

Figure 332. Neurolucida reconstructions of supercial and deep subicular cells. Somata and
dendrites are shown in black. Axons are shown in gray. Local collaterals tend to be longer in the
cell layer for supercially located cells; deep cells tend to have axon collaterals that ascend close
to the primary dendrite. CA1 is at the left, and the presubiculum is at the right. Bar  500 m.
(Source: Adapted from Harris et al., 2001.)

preferential staining for either NADPH-diaphorase/nitric density of this local intrinsic connection, as estimated by the
oxide synthetase (regular spiking neurons) or somatostatin number of varicosities on locally distributed axon collaterals,
(bursting cells). Both cell types are projection neurons, but is much higher than in CA1. In addition, the two types of
they might differ with respect to their connectivity because subicular pyramidal neurons differ with respect to their local
there is evidence suggesting that only bursting cells project to connectivity. Intracellular labeling of electrophysiologically
the entorhinal cortex. Intermingled among the pyramidal cells identied bursting cells generally show an axonal distribution
are many smaller neurons, presumably representing the that remains in the region circumscribed by their apical den-
interneurons of the subiculum. Little is known, however, drites (i.e., a columnar organization), whereas the regular
about whether the interneurons seen in the subiculum are spiking cells generally give rise to an axon that shows more
similar to those observed in the hippocampus. Subpopula- widespread distribution along the transverse axis (Fig. 333).
tions of these neurons appear to have characteristics similar to It is not known whether differences exist with respect to pos-
those described for CA1; among these subpopulations are sible septotemporal spread. Although much work remains to
GABAergic cells that stain for the calcium-binding protein be done, available data indicate that the organization of the
parvalbumin. intrinsic connectivity of the subiculum is different from that
of CA3 and CA1. There is both crude columnar and laminar
Intrinsic Connections organization, such that the bursting cells form a set of
columns and the regular spiking neurons integrate columnar
The subiculum gives rise to a longitudinal associational pro- activity along the transverse axis.
jection that extends from the level of the cells of origin to
much of the subiculum lying temporally (or ventrally). Extrinsic Connections: Subiculum,
Interestingly, this projection seems to be largely unidirec- a Major Output Structure
tional, as few if any associational projections course septally or
dorsally from their point of origin (Harris et al., 2001). The The subiculum is a major source of efferent projections from
associational bers terminate diffusely in all layers of the the hippocampal formation. Following the discovery by
subiculum. Recent studies have consistently found that the rat Swanson and Cowan that the subiculum, rather than the
subiculum neither gives rise to nor receives commissural con- hippocampus, is the origin of the major subcortical connec-
nections. tions to the diencephalon and brain stem (via the postcom-
Subicular pyramidal cells provide a strong local input in missural fornix), evidence has mounted that the subiculum is
the pyramidal cell layer and just supercial to it, targeting one of the two primary output structures of the hippocampal
proximal portions of the apical dendrites. Interestingly, the formation (Swanson and Cowan, 1975; Swanson et al., 1981;
78 The Hippocampus Book

collective hunch was that bers simply passed through the

subiculum and probably did not terminate in it. Earlier
anterograde tracer studies indicated that perforant path bers
are directed toward the molecular layer of the subiculum, but
proof that these bers formed a terminal plexus among the
subicular pyramidal cells was lacking. Witter and colleagues,
however, have provided evidence, at both light and electron
microscopic levels, that the subiculum receives a strong pro-
jection from the entorhinal cortex. The topography of the
projection is similar to that described for CA1. Fibers are
directed toward restricted transverse portions of the subicu-
lum and terminate in the outer two-thirds of the molecular
layer. The lateral component of the perforant pathway prefer-
entially projects to the proximal part of the subiculum (i.e.,
the part of the subiculum that borders CA1), and the medial
Figure 333. Model of intrinsic organization of the subiculum. component distributes to more distal portions of the subicu-
Deep cells have ascending axon collaterals that remain in close prox- lum (i.e., closer to the presubiculum). The projection origi-
imity to their apical dendrites, giving rise to a columnar pattern for nates mainly from layer III, although some of the axons of
the deep cells. Supercial cells have axon collaterals that can run layer II cells that cross the subiculum on their way to the den-
long distances in the cell layer perhaps serving to integrate activity
tate gyrus and CA3 may also give off collaterals that terminate
between columns. (Source: Adapted from Harris et al., 2001.)
in the subiculum. Entorhinal bers target dendritic spines of
presumed principal neurons with asymmetrical synapses
(80%), whereas 5% to 10% of the asymmetrical synapses ter-
Donovan and Wyss, 1983; Groenewegen et al., 1987; Witter
minate on dendritic shafts, most likely belonging to interneu-
and Groenewegen, 1990; Witter et al., 1990; Canteras and
rons. Subicular interneurons may also receive a minor
Swanson, 1992; Naber and Witter, 1998; Ishizuka, 2001;
inhibitory perforant path input in view of the symmetrical
Kloosterman et al., 2003).
synapses onto dendritic shafts.
The subiculum reciprocates the entorhinal input (Fig.
Subiculum Projects to the Presubiculum and Parasubiculum 334). Projections from the subiculum reach all parts of the
The subiculum projects to the presubiculum, but this projec- entorhinal cortex and terminate in the deep layers; termina-
tion is not of the magnitude of other intrinsic hippocampal tion is particularly dense in layer V. A minor component of the
formation projections. It is probably better to think of the subicular projection also extends supercial to the lamina dis-
subiculum projection to the presubiculum as one of a series of secans, predominantly to layer III. Subicular bers generally
pathways that distributes information processed in the den- form asymmetrical synapses with spines and dendrites located
tate gyrus, hippocampus, and subiculum to a series of cortical in these layers. A small number of these bers form symmet-
and subcortical structures. The subiculum can be viewed, rical synapses, suggesting small inhibitory input from the
therefore, as the last step in the large loop of processing subiculum to layer V. Therefore, the overall organization of
through the hippocampal formation. the subiculo-entorhinal projection mimics that of the CA1-
Subicular bers terminate mainly in layer I of the pre- entorhinal projection. The presence of asymmetrical synapses
subiculum. The projection to the dorsal part of the pre- at the termination of this pathway is consistent with its
subiculum, however, terminates deep to the prominent layer reported excitatory inuences on the entorhinal cortex.
II, with weaker projections to layers I and II. The projections
from the subiculum to the parasubiculum mainly terminate in Subicular Connections with the Neocortex
layer I and the supercial portion of layer II. These projections and Amygdaloid Complex
are topographically organized. Septal (or dorsal) portions of The subiculum gives rise to prominent projections to the
the subiculum project to dorsal and caudal aspects of both the medial and ventral orbitofrontal cortices and to the prelimbic
presubiculum and the parasubiculum, and temporal (or ven- and infralimbic cortices (Verwer et al., 1997). In the orbito-
tral) portions of the subiculum project to ventral and rostral frontal and prelimbic cortices subicular bers mainly inner-
aspects of the pre- and parasubiculum. vate the deep layers, whereas in the infralimbic cortex the
projection extends into the supercial layers. Subicular pro-
Subiculum Reciprocally Connected with the Entorhinal Cortex jections also reach medial portions of the anterior olfactory
Because the perforant path bers traverse the subiculum on nucleus.
their way to the dentate gyrus and hippocampus, the question The subiculum provides a meager projection to the ante-
of whether some of these bers might terminate in the subicu- rior cingulate cortex, whereas the projection to the retrosple-
lum has long been a matter of controversy. Until recently, the nial cortex is substantial (Wyss and Van Groen, 1992). The
Hippocampal Neuroanatomy 79

Figure 334. Topographical organization of the subicular projec- tion of the subiculum projects to medial portions of the lateral and
tions to the parahippocampal region (entorhinal, perirhinal, and medial entorhinal cortex. B. Relation between the proximo-distal
postrhinal cortices). A. Relation between the septotemporal origin origin in the subiculum with a rostrocaudal termination in the
in the subiculum with a lateral-to-medial termination in the parahippocampal region. The proximal subiculum projects to the
parahippocampal region. The septal portion of the subiculum proj- perirhinal and lateral entorhinal cortex; the distal subiculum proj-
ects to the perirhinal and postrhinal cortices and the lateral portion ects to the postrhinal and medial entorhinal cortex. (Source:
of both the lateral and medial entorhinal cortex. The temporal por- Adapted from Kloosterman et al., 2003.)

subiculo-retrosplenial projection terminates predominantly subiculum, but this projection terminates only in the proxi-
in layers II and III. The projections to the retrosplenial cortex mal third of the eld (i.e., at the border region with CA1).
originate predominantly from the septal two-thirds of the The proximal portion of the subiculum also receives an
subiculum. The perirhinal cortex receives a strong input from input from the parvicellular portion of the basal nucleus of
the subiculum, which terminates in both supercial and deep the amygdaloid complex, the posterior cortical nucleus, and
layers. the adjacent amygdalohippocampal area. These amygdaloid
In the rat, there is a paucity of detailed information regard- inputs terminate mainly at the CA1/subiculum border region,
ing direct cortical inputs to the subiculum. Many of the corti- where they preferentially innervate the molecular layer of the
cal regions that project fairly heavily to the entorhinal cortex subiculum and stratum lacunosum-moleculare of CA1
(see below) do not project to the subiculum. No inputs, for (Pitkanen et al., 2000).
example, have been reported from the pre- and infralimbic The temporal one-third of the subiculum gives rise to
cortices. Similarly, no portion of the retrosplenial cortex proj- return projections to the amygdaloid complex. The major
ects to the subiculum. Reports of projections from the cingu- component of this projection terminates in the accessory
late cortex have been somewhat contradictory. In a combined basal nucleus, with more moderate projections reaching sev-
electrophysiological and neuroanatomical study, White et al. eral other nuclei but not the lateral nucleus. The ventral
(1990) reported a projection from the anterior cingulate cor- subiculum also projects heavily to the bed nucleus of the stria
tex to the subiculum. However, this projection has not been terminalis and moderately to the ventral part of the claustrum
observed consistently. The perirhinal cortex projects to the or endopiriform nucleus.
80 The Hippocampus Book

Basal Forebrain Connections: Septal Nucleus Interestingly, the projections to the subiculum and to CA1
and Nucleus Accumbens originate from different but intermingled populations of
The most prominent subcortical subicular projections are neurons in the nucleus reuniens. Although earlier studies
those to the septal complex, the adjacent nucleus accumbens, had indicated that the subiculum was interconnected with
and the mammillary nuclei. The projection to the septal area the anterior nuclear complex, it is now clear that the ante-
terminates predominantly in the lateral septal nuclei. Closely rior nucleus projects almost exclusively to the pre- and para-
associated with the septal projection is the equally robust subiculum.
projection to the nucleus accumbens and adjacent portions Some of the thalamic regions that project to the subiculum
of the olfactory tubercle. Subicular bers terminate through- receive a return projection from the subiculum. Subicular
out the nucleus accumbens, with the projection to its caudo- bers terminate bilaterally in the nucleus reuniens, the
medial part being most dense. As with other striatal nucleus interanteromedialis, the paraventricular nucleus, and
structures, the subicular projection to the nucleus accumbens the nucleus gelatinosus. Although subicular projections to
is unidirectional. Whereas the subicular projections to the parts of the anterior thalamic complex have been described in
lateral septal nucleus are almost entirely conned to the ipsi- the literature, more recent retrograde tracing studies have
lateral side, those to the nucleus accumbens show a weak con- shown that these projections arise from the presubiculum.
tralateral component. The subiculum receives a relatively
weak cholinergic projection from the septal complex; bers Brain Stem Inputs
originating from the medial septal nucleus and the nucleus of Monoaminergic ascending pathways from the noradrenergic
the diagonal band terminate in the pyramidal cell and molec- locus coeruleus, the dopaminergic ventral tegmental area, and
ular layers. the serotonergic median and dorsal raphe nuclei reach the
subiculum, but they do not show preferential innervation of
Hypothalamic Connections: Mammillary Nuclei this region. Few details are available concerning the regional
The subiculum provides the major input to the mammillary localization of these pathways in the subiculum.
nuclei. The projection is heavy and is distributed bilaterally in
nearly equal density. The subiculomammillary bers originate Topography of Subicular Efferent Projections
mainly from the septal two-thirds of the subiculum. Although As with the projections from CA3 to CA1 and from CA1 to the
the temporal one-third of the subiculum also contributes to subiculum, the subicular efferent projections are topographi-
the mammillary projection, the major hypothalamic target of cally organized. In large part, the subicular projections pre-
this portion of the subiculum is the ventromedial nucleus of serve the transverse topography established by the CA1 to
the hypothalamus. The subicular projections to the mammil- subiculum projection. It is clear that different projections
lary nuclei reach all portions of the medial nucleus but are originate from at least the proximal and distal halves of the
topographically organized; the lateral mammillary nucleus is subiculum. The subiculum also demonstrates marked sep-
only sparsely innervated by the subiculum. Subicular bers totemporal topography, such that the projections that arise
also project to the lateral hypothalamic region located adja- from the septal or dorsal two-thirds of the subiculum are dif-
cent to the lateral mammillary nucleus. ferent from those that arise from the temporal or ventral third
Whereas the subiculum does not receive a return projec- (Witter and Amaral, 2004).
tion from the medial or lateral mammillary nuclei, the supra- Turning rst to the septotemporal topography, it appears
mammillary region projects heavily to the subiculum, that the projections to the entorhinal cortex, the lateral septal
particularly to its temporal levels. This portion of the subicu- complex, the nucleus accumbens, and the medial mammillary
lum also receives an input from the premammillary nucleus. nucleus originate from the entire septotemporal extent of the
It is not clear whether there are any local connections between subiculum (Witter and Groenewegen, 1990; Ishizuka, 2001).
the medial mammillary nucleus and the supramammillary Different septotemporal levels of the subiculum, however,
area or the premammillary nucleus that might complete the project to different portions of these elds. In the entorhinal
subiculohypothalamic loop. cortex, for example, the septal-to-temporal origin in the
subiculum is related to a lateral-to-medial termination in the
Thalamic Connections: Nucleus Reuniens entorhinal cortex. Septal levels of the subiculum project pref-
and Other Midline Nuclei erentially to lateral and caudal parts of the entorhinal cortex
The thalamic inputs to the subiculum are similar to those to (i.e., the parts that lie adjacent to the rhinal sulcus).
CA1. Thalamic inputs originate mainly in the nucleus Progressively more temporal levels of the subiculum project to
reuniens, the paraventricular nucleus, and the parataenial more medially located parts of the entorhinal cortex.
nucleus. The septal and temporal extremes of the subiculum Although addressed in more detail below, it is important to
appear to be devoid of input from the nucleus reuniens. The point out that this topography is completely in register (i.e.,
midline thalamic projections terminate mainly in the molec- they are point-to-point reciprocal) with the projections from
ular layer of the subiculum, whereas in CA1 they are coexten- the entorhinal cortex to the subiculum. Thus, cells in the
sive with the projections from the entorhinal cortex. subiculum that receive input from a subregion in the entorhi-
Hippocampal Neuroanatomy 81

nal cortex give rise to a return projection to the same region ization. Thus, the proximal portions of the subiculum project
in the entorhinal cortex. In some respect this breaks the rule to rostral medial mammillary nuclei, and distal portions of
that all structures in the hippocampal formation have unidi- the subiculum project more caudally. A similar situation exists
rectional connectivity, but other aspects of the neuroanatomy for the subicular projections to the entorhinal cortex. The
of these areas completely justify their inclusion in the func- proximal half of the subiculum projects to the lateral entorhi-
tionally dened hippocampal formation. nal area, and the distal half of the subiculum projects to the
In the nucleus accumbens, the septotemporal axis of origin medial entorhinal area.
in the subiculum determines a caudomedial to rostrolateral Because the proximal third of the subiculum gives rise to
axis of termination. Dorsomedial portions of the lateral septal projections to at least several cortical and subcortical regions,
complex receive inputs from septal levels of the subiculum, the question arises as to whether it is the same or different
and ventral portions of the lateral septal complex are inner- populations of subicular cells that innervate each structure.
vated by bers originating in more temporal parts of the The answer initially appeared to be that projections to the sep-
subiculum. tal complex, entorhinal cortex and mammillary complex arise,
Similar septotemporal topography has also been described at least in part, as collaterals from single subicular neurons;
for the subicular projections to the presubiculum and the but it now appears that largely independent populations of
medial mammillary nuclei. The latter projection arises mainly intermixed neurons in the subiculum project to each of its ter-
from the septal two-thirds of the subiculum, whereas the ven- minal regions (Naber and Witter, 1998). Given our earlier
tral one-third gives rise to projections to other hypothalamic assertion that the subiculum is the last staging post of hip-
regions such as the ventromedial nucleus. pocampal processing, this state of affairs seems to create the
This dichotomy between the septal two-thirds and the possibility that the outputs destined for different target struc-
temporal one-third of the subiculum is reected in the organ- tures can be carrying distinctly different information.
ization of other projections. Projections to the amygdala and
the bed nucleus of the stria terminalis, for example, originate 3.4.4 Presubiculum and Parasubiculum
exclusively from the temporal one-third of the subiculum,
whereas projections to the retrosplenial and perirhinal cor- Cytoarchitectonic Organization and Neuron Types
tices originate predominantly from the septal two-thirds. The
subicular projections to the midline thalamus demonstrate The presubiculum, Brodmanns area 27, is relatively easily
even greater septotemporal topography. The most septal part differentiated from the subiculum in standard Nissl-stained
of the subiculum projects preferentially to the interanterome- material. It has a distinct, densely packed external cell layer
dial nucleus; mid-septotemporal levels of the subiculum pref- that consists mainly of darkly stained, small pyramidal cells.
erentially project to the nucleus reuniens; and the temporal The most supercial cells are the most densely packed (layer
third of the subiculum projects most heavily to the paraven- II), whereas the deeper cells have a somewhat looser arrange-
tricular nucleus. ment (layer III). The differentiation between layers II and III
Whereas the septotemporal topography appears to be is more clear-cut at dorsal levels of the presubiculum.
organized in a gradient, or gradual, fashion, the transverse The dorsal presubiculum (sometimes called the post-
organization of subicular efferents is remarkably discrete. subiculum) has clearly distinguishable supercial and deep
Along the transverse axis of the subiculum, two essentially cell layers. In the ventral portion of the presubiculum, how-
nonoverlapping populations of cells can be differentiated that ever, the deep layers are difficult to distinguish from the deep
give rise to projections to specic sets of brain structures. This layers of the entorhinal cortex or from the principal cell layer
transverse organization of the outputs of the subiculum is of the subiculum. Deep to the lamina dissecans there are one
consistently observed along its entire septotemporal axis, or two layers of large, darkly stained pyramidal cells; and deep
although it is clearer septally than temporally. to these cells is a rather heterogeneous collection of pyramidal
Neurons in the proximal half of the subiculum (closest to and polymorphic cells. The latter cells have not been studied
CA1) project to the infralimbic and prelimbic cortices, the anatomically in great detail (Fig. 335), although electrophys-
perirhinal cortex, the nucleus accumbens, the lateral septum, iological discoveries about their receptive eld characteristics
the amygdaloid complex, and the core of the ventromedial are important to one prominent theory of hippocampal func-
nucleus of the hypothalamus. Cells in the distal half of the tion (see Chapter 8).
subiculum project mainly to the retrosplenial cortex and the The parasubiculum (Brodmanns area 49) lies adjacent to
presubiculum. Cells projecting to the midline thalamic nuclei the presubiculum. Layers II and III of the parasubiculum con-
are mainly located in the midportion of the subiculum. sist of rather densely packed, lightly stained, large pyramidal
The subicular projections to the entorhinal cortex and to cells. This and other characteristics, such as the distinctive
the medial mammillary nucleus do not follow a strictly trans- staining for heavy metals observable with the Timms stain
verse organization, as cells in all proximo-distal portions of method, are the major features that differentiate the para-
the subiculum project to these areas. However, the topography subiculum from the presubiculum (Fig. 311). There is no
of these projections indicates a more subtle transverse organ- clear differentiation between layers II and III; and as with the
82 The Hippocampus Book

Commissural projections from the dorsal portion of the pre-

subiculum are relatively sparse. The parasubiculum also gives
rise to a commissural projection that terminates most densely
in layers I and III.

Extrinsic Connections

Presubiculum and Parasubiculum Reciprocally

Connected with Anterior Thalamic Nuclei
The presubiculum and parasubiculum receive a number of
subcortical inputs, but the one that is unique to these portions
of the hippocampal formation is their interconnection
with the anterior thalamic nuclear complex, primarily the
anteroventral and anterodorsal nuclei and the closely related
laterodorsal nucleus (Kaitz and Robertson, 1981; Robertson
and Kaitz, 1981). Although earlier studies had suggested that
the presubiculum also receives inputs from the anteromedial
Figure 335. Cell types of the presubiculum and parasubiculum. nucleus, the principal projections of this nucleus appear to be
Camera lucida drawings of neurobiotin-lled cells in a horizontal directed to the anterior parts of the cingulate cortex rather
section through the hippocampal formation. A. Pyramidal cell in than to the presubiculum or parasubiculum.
layer II of the parasubiculum. B. Stellate cell in layer II of the pre- There are topographical differences in the thalamic con-
subiculum. C. Pyramidal cell in layer III of the presubiculum. nections of the ventral and dorsal portions of the presubicu-
D. Stellate cell in layer V of the presubiculum. E. Pyramidal cell
lum. The ventral portion receives most of its input from the
in layer V of the presubiculum. F. Pyramidal cell in layer V of the
laterodorsal and anteroventral nuclei, whereas the dorsal
parasubiculum. G. Stellate cell in layer V of the parasubiculum.
Bar  100 m. (Source: Adapted from Funahashi and Stewart,
part receives projections mainly from the laterodorsal and
1997a.) anterodorsal nuclei. The thalamic projections mainly termi-
nate in layers I, III, and IV. The nucleus reuniens also projects
to layer I of the presubiculum, but this is a much lighter pro-
presubiculum, the deep layers are continuous with those of jection.
the entorhinal cortex. Like most other thalamocortical connections, the pre-
subiculum sends a return projection back to the thalamus.
Intrinsic Connections The presubiculum projects massively and bilaterally to the
anterior nuclear complex. The presubicular projections arise
There are well developed associational connections in the pre- mainly from cells located deep to the lamina dissecans (layer
subiculum that interconnect all dorsoventral levels in a highly VI). The fact that these deep cells are related to the anterior
directional manner. Layer II cells in ventral portions of the nuclei of the thalamus lends some credibility to the idea that
presubiculum project to more dorsal levels of the presubicu- they are, in fact, deep layers of the pre- and parasubiculum
lum. Projections in the opposite direction (dorsal to ventral), rather than of the entorhinal cortex.
however, arise preferentially from cells in the deep layers.
The parasubiculum also gives rise to associational connec- Presubiculum Projection to Layer III of the Entorhinal Cortex
tions that distribute both dorsally and ventrally from the cells The most prominent intrahippocampal projection from the
of origin. The dorsally directed projections extend for short presubiculum is to the entorhinal cortex (Shipley, 1975;
distances and are quite weak. The ventrally directed projec- Caballero-Bleda and Witter, 1993). This projection has a
tions are substantially denser and extend for long distances number of interesting features. First, the presubicular projec-
along the long axis of the parasubiculum. The parasubiculum tion is directed only to the medial entorhinal area. Second, the
gives rise to a particularly dense projection to the dorsally projection terminates almost exclusively in layer III and to a
located area 29e. This small wedge-shaped region was initially much lesser extent in the deep part of layer I. Third, the
described by Blackstad (1956) and later characterized in more crossed homotopic projection from the presubiculum to the
detail by Haug (1976). Although the neuroanatomy of area contralateral entorhinal cortex is every bit as dense as the ipsi-
29e remains sketchy, it might be part of the parasubiculum. lateral projection. The projection to the entorhinal cortex is
topographically organized. The location of the presubicular
Commissural Connections terminal eld in the entorhinal cortex is determined by both
the proximo-distal and dorsoventral location of the presubic-
The presubiculum gives rise to strong commissural projec- ular cells of origin.
tions to layers I and III of the homotopic part of the con- The parasubiculum selectively innervates layer II of the
tralateral presubiculum (Van Groen and Wyss, 1990). entorhinal cortex. In contrast to the presubiculum, the para-
Hippocampal Neuroanatomy 83

subiculum projects to both the medial and lateral entorhinal a radial manner, which is quite distinct from the predomi-
areas, although the projection to the lateral entorhinal area is nantly transverse orientation of the entorhinal perforant
less robust. The parasubiculum projects to the contralateral pathway bers.
entorhinal cortex, but these projections are much weaker than A fact that has not been generally appreciated is that the
the ipsilateral ones. The topographical organization of the parasubiculum gives rise to a fairly substantial projection to
parasubicular projection to the entorhinal cortex is compara- the molecular layer of the dentate gyrus (Kohler, 1985). Like
ble to that of the presubiculo-entorhinal projection. the lighter projection from the presubiculum, this projection
occupies the supercial two-thirds of the molecular layer
Presubiculum and Parasubiculum Connected (with a preference for the midportion of the molecular layer),
with Some Neocortical Regions and the bers have a predominantly radial orientation.
The presubiculum receives relatively few extrahippocampal Because the parasubiculum receives a projection from the
cortical inputs. The most prominent one originates in the ret- anterior thalamic nuclei, its projection to the molecular layer
rosplenial cortex (Van Groen and Wyss, 1990; Wyss and Van provides a route by which thalamic input might inuence the
Groen, 1992). Cells located in layer V of the retrosplenial cor- very early stages of hippocampal information processing. The
tex give rise to projections that terminate in layers I and III/V parasubiculum projects weakly to the stratum lacunosum-
of the presubiculum. A second cortical input originates from moleculare of the hippocampus and to the molecular layer of
layer V of the visual area 18b. This projection mainly distrib- the subiculum. The parasubiculum also projects bilaterally to
utes to the dorsal half of the presubiculum and terminates in layers I and III of the presubiculum.
layers I and III. Minor cortical inputs originate in the prelim-
bic cortex and in a dorsal portion of the medial prefrontal Basal Forebrain Connections
cortex. The presubiculum and parasubiculum receive heavy choliner-
Extrahippocampal projections from layer V of the pre- gic input. The medial septal nucleus and the vertical limb of
subiculum reach the granular retrosplenial cortex, where they the diagonal band of Broca mainly innervate layer II of the
terminate preferentially in layers I and II. These projections presubiculum.
are topographically organized, such that the ventral pre-
subiculum projects mainly to the ventral part of the granular Hypothalamic Connections: Mammillary Nuclei
retrosplenial cortex, and the dorsal part of the presubiculum The deep layers of the presubiculum project bilaterally to the
projects more dorsally. These projections also exhibit a rostro- medial and lateral mammillary nuclei. The projections to the
caudal organization, such that rostral portions of the pre- medial mammillary nuclei are topographically organized in a
subiculum project to rostral parts of the retrosplenial cortex, manner similar to those that originate in the subiculum
and caudal portions of the presubiculum project to caudal (Thompson and Robertson, 1987; Allen and Hopkins, 1989;
parts of the retrosplenial cortex. It has been suggested that the Van Groen and Wyss, 1990).
dorsal presubiculum projects to the deep layers of the most The presubiculum receives input from the area surround-
caudal portion of the perirhinal cortex, although an alterna- ing the mammillary nuclei. Fibers from the supramammillary
tive interpretation is that this projection is directed to the nucleus terminate preferentially in the deeper cell layers of the
most caudodorsal portion of the medial entorhinal cortex, presubiculum, although those characterized as being positive
which is difficult to differentiate from the caudal perirhinal for -melanocyte-stimulating hormone (an opiate peptide
cortex. The work of Burwell et al. (1998a,b) supports the expressed in neurons whose somata are located in the lateral
latter idea. hypothalamic area) terminate in the molecular layer.
With the exception of the relatively light projections from
the retrosplenial cortex and the occipital visual cortex, there Brain Stem Inputs
are no other known extrahippocampal cortical inputs to the The presubiculum receives input from various nuclei in the
parasubiculum. The laminar distribution of these inputs is brain stem. A particularly dense innervation arises from the
similar to that described for the presubiculum. dorsal and ventral raphe nuclei; at least a component of this
projection is serotonergic and innervates layer I. The nora-
Other Intrahippocampal Connections drenergic locus coeruleus innervates the plexiform layer.
The presubiculum projects to layers I and II of the para-
subiculum bilaterally. Anterograde tracing studies with the 3.4.5 Entorhinal Cortex
lectin tracer PHA-L have demonstrated that the presubiculum
and perhaps to a greater extent the parasubiculum contribute The entorhinal cortex plays an extraordinarily important role
projections, albeit modest ones, to many of the other regions in the ow of information through the hippocampal forma-
of the hippocampal formation. For example, there is a modest tion. It is not only the main entry point for much of the sen-
bilateral projection from the presubiculum to the subiculum. sory information processed by the hippocampal formation, it
There is also a weak projection to all elds of the hippocam- provides the main conduit for processed information to be
pus and to the molecular layer of the dentate gyrus. The pre- relayed back to the neocortex. As portrayed in this chapter, the
subicular bers to the dentate molecular layer are arranged in entorhinal cortex is the beginning and the end point of an
84 The Hippocampus Book

extensive loop of information processing that takes place in layer that can be subdivided into bands, and layer Va, which
the hippocampal formation. Although neuroanatomical forms a band of large, darkly stained pyramidal neurons and
investigation of the entorhinal cortex has historically lagged is most conspicuous in the central parts of the entorhinal cor-
behind the work conducted in other elds of the hippocampal tex. At other levels, the packing density of cells is not high, and
formation, many new ndings have been forthcoming in the smaller cells of the deeper part of this layer (Vb) inter-
recent years. This progress has been spurred on, in part, by the mingle with it. Finally there is layer VI, containing a highly
appreciation, initially by van Hoesen et al. (1991), that the heterogeneous population of cell sizes and shapes. This cell
entorhinal cortex is a site of early, devastating pathology in density decreases toward the border with the white matter.
degenerative diseases such as Alzheimers disease. It has also The cells of layer VI appear to blend gradually into the subja-
been fostered by the reemergence of the notion rst put forth cent subcortical white matter and the overlying layer V.
by Ramon y Cajal that (to paraphrase) whatever the rest of
the hippocampal formation is doing depends on what the Regional Organization
entorhinal cortex has done. We begin our description of There have been several attempts to subdivide the rat entorhi-
the entorhinal cortex with a discussion of some lingering con- nal cortex, and unfortunately there have been almost an equal
troversies concerning its laminar and regional organization. number of differing opinions concerning the number and ter-
minology of the subelds. The subject has been discussed and
Cytoarchitectonic Organization reviewed by Menno Witter, Ricardo Insausti, and their col-
leagues (Witter, 1993; Insausti et al., 1997; Witter et al., 2000;
Laminar Organization Burwell and Witter, 2002). Nevertheless, it is now generally
There are currently two schemes of cortical lamination accepted that the entorhinal cortex can be subdivided into two
applied to the entorhinal cortex. As one might expect, this general areas: the lateral entorhinal area (LEA) and the medial
causes substantial confusion, especially to the neophyte entorhinal area (MEA) (Fig. 336). Layer II is more clearly
hippocampologist. One nomenclature, which divided the demarcated in the LEA than in the MEA, and the cells are
entorhinal cortex into seven layers, was rst suggested by extremely densely packed and tend to be clustered in islands.
Ramn y Cajal and later modied to more closely resemble The cells in layer II of the MEA are somewhat larger and do
the standard six-layer scheme applied to the isocortex. not show a distinct clustering into islands; the border between
According to this scheme, there are four cellular layers (II, III, layers II and III is not as sharp as in the LEA. In both entorhi-
V, VI) and two acellular or plexiform layers (I, IV). The acel- nal areas, however, the overall differences in cell size between
lular layer IV is also called lamina dissecans. Ramon y Cajals layers II and III facilitate the delineation of the two layers. The
scheme with slight modication has been employed by other cell layers, particularly layers IV to VI, can be better dif-
Amaral and colleagues in several primate studies. The other ferentiated from each other in the MEA than in the LEA, and
commonly used scheme was proposed by Lorente de N, who cells in the MEA generally show a more radial or columnar
also differentiated six layers. Five of Lorente de Ns layers arrangement. The lamina dissecans of the MEA is sharply
were cellular (II, III, IV, V, VI) with a cell-free lamina dissecans delineated but is less clear in the LEA.
(layer IIIb) between layers III and IV. This scheme was used in It should also be stressed that the terms lateral and medial
most of the older studies of the entorhinal cortex. It is still entorhinal areas do not relate in a simple manner to the car-
used in rodent studies and in at least some studies of the dinal transverse plane of the rat brain (Fig. 337). Both the
human entorhinal cortex, particularly those of Van Hoesen LEA and the MEA have a more or less triangular shape. The
and colleagues. LEA occupies the rostrolateral part of the entorhinal cortex;
Primarily to emphasize the lack of an internal granular cell its base is oriented rostrally and its tip caudolaterally, next to
layer in the entorhinal cortex, we have decided to adopt the rhinal ssure. The MEA occupies the remaining triangular
Ramon y Cajals nomenclature and have labeled the cell-poor area, which has its base caudally and its tip rostromedially
layer IV lamina dissecans. Starting from the pial surface, the such that the tip lies medial to the LEA. A different nomencla-
layers include layer I, the most supercial plexiform or molec- ture has recently been proposed by Insausti and colleagues to
ular layer, which is cell-poor but rich in transversely oriented accommodate the oblique orientation of the rat entorhinal
bers; layer II, containing mainly medium-sized to large stel- cortex and to address the need for subdemarcation of the LEA
late cells and a population of small pyramidal cells that tend and MEA (Insausti et al., 1997).
to be grouped in clusters (cell islands) particularly in the lat-
eral entorhinal area; layer III, containing cells of various sizes Neuron Types
and shapes but predominantly pyramidal cells; layer IV (or
lamina dissecans), a cell-free layer located between layers III Our current knowledge of the cytology of the entorhinal cor-
and V that is most apparent in those portions of the entorhi- tex is based largely on the classic Golgi studies of Ramon y
nal cortex that lie close to the rhinal ssure, particularly at the Cajal and Lorente de N in young mice. A few intracellular
caudal levels of the entorhinal cortex. In the remainder of the labeling studies have also been conducted in rats, and they
entorhinal cortex, groups of cells invade this layer so it has an have contributed important new information concerning
incomplete or patchy appearance. Next are layer V, a cellular entorhinal cell types (Hamam et al., 2000, 2002). It is proba-
Hippocampal Neuroanatomy 85

Figure 336. Cytoarchitectonic characteristics of the rat entorhinal crographs of LEA (D) and MEA (E) taken from portions of the
cortex. AC. Photomicrographs of Nissl-stained coronal sections entorhinal cortex enclosed by boxes in B and C, respectively.
through three selected rostrocaudal levels of the entorhinal cortex, The layers of the entorhinal cortex are indicated with roman
arranged from rostral (A) to caudal (C), showing the lateral (LEA) numerals. PcoA, posterior cortical nucleus of the amygdala. Bars:
and medial (MEA) entorhinal area subdivisions. Arrowheads indi- AC, 1 mm; D and E, 250 m (Source: Dolorfo and Amaral,
cate LEA and MEA boundaries. D, E. High magnication photomi- 1998b.)
86 The Hippocampus Book

bly safe to predict that a number of new entorhinal cell types

will be discovered as these techniques are applied more inten-
sively to analysis of the entorhinal cortex. Other recent studies
have focused on determining which particular neuron types
project to the dentate gyrus, hippocampus, and subiculum.
These studies have employed retrograde tracing techniques
either alone or in conjunction with intracellular lling. The
following description provides a short overview of the various
cell types of each of the entorhinal layers and a brief comment
on the major characteristics of their dendritic and axonal
Layer I is populated by a small number of widely dispersed
neurons that have been further differentiated using various
criteria. There are stellate and horizontal GABAergic neurons
that have been shown electrophysiologically to terminate on
the dendrites of layer II cells that project to the dentate gyrus.
Layer II is populated by stellate and pyramidal cells (Fig.
338), both of which project to the dentate gyrus and CA3.
These cells give rise to extensive associational projections to
layer II of other regions of the entorhinal cortex, and they
contribute a few collaterals to deeper layers of the entorhinal
cortex. Although there are some exceptions, most layer II cells
have a dendritic tree that is conned to supercial layers I and
II. A class of axo-axonic cells, similar to the cortical chandelier
Figure 337. Relative positions of the lateral (LEA) and medial
cell, is also located in layer II; although some of these cells are
(MEA) entorhinal cortex and the perirhinal and postrhinal cortices. present in layer III, they have not been observed in the deep
A. Lateral view of the rat brain. B. Unfolded map of the entorhinal, layers.
perirhinal, and postrhinal cortices. Bar  1 mm. (Source: Adapted In layer III, the most numerous neurons are the pyramidal
from Burwell and Amaral, 1998b.) cells (Fig. 339), which give rise to the perforant and alvear

Figure 338. Morphological characteristics of entorhinal cortex III. B. Camera lucida drawing of a typical layer II stellate cell. Note
layer II neurons. A. Camera lucida drawing of a typical layer II the multiple thick primary dendrites and the widely diverging
pyramidal cell. Note the thick apical dendrite branching above the upper and lower dendritic trees, with supercially directed den-
border with layer I, the thin basal dendrites arising radially from the drites reaching the topmost portion of layer I. The axon (truncated)
soma, and the limited extent of the upper dendritic tree. The axon emerges from the base of the soma. Bar  100 m. (Source:
(arrow) emerges from the base of the soma and branches in layer Adapted from Klink and Alonso, 1997.)
Hippocampal Neuroanatomy 87

Layer IV, the lamina dissecans, although generally referred

to as a cell-free layer, does contain scattered cell bodies
described as fusiform or pyramidal cells that have apical den-
drites reaching up to layer I and an axon reaching the deep
white matter, a characteristic of projection neurons. Some of
the cells in the lamina dissecans stain positively with antibod-
ies against GABA, NPY, calretinin, calbindin, or somatostatin.
Layer V contains three main classes of neuronpyramidal
cells, small spherical cells, and fusiform neuronsaccording
to the Golgi-based description of Lorente de N. Layer Va is
characterized by its large, darkly staining pyramidal cells.
These cells have large apical dendrites that ascend toward the
supercial portion of layer II and into layer I. The axons of
these cells run into the deep white matter and the angular
bundle, and additional collaterals innervate the supercial
layers of the entorhinal cortex. According to Lorente de N,
the collaterals of such cells form a column-like plexus situated
close to the location of the parent cell body and span the
entire thickness of the entorhinal cortex. These collaterals
mainly distribute to the deep cell layers (V/VI), although
Figure 339. Morphological characteristics of entorhinal cortex occasional collaterals reach layer II. Layer Va also contains var-
layer III neurons. Camera lucida of a typical layer III pyramidal cell. ious small neurons that have been characterized as horizontal,
The axon is indicated by an arrow. Bar  100 m. (Source: Adapted stellate, and multipolar cells. Golgi studies indicate that the
from Gloveli et al., 1997.) axonal plexus of the latter group either remains in layer Va or
extends to the supercial layers I to III. Angel Alonso and col-
path projections to CA1 and the subiculum. Their axons give leagues have used intracellular staining techniques to study
rise to collaterals that distribute mainly to layers I to III. The the morphology of layer V neurons, particularly in the deeper
apical dendrites of layer III pyramidal cells ascend, give off portion of layer V, or layer Vb (Fig. 340). They also found
branches to layer II, and ultimately form a terminal tuft in that there are three main types of neuron. Large and small
layer I. Layer III also contains multipolar, stellate, fusiform, pyramidal cells, neurons with dendrites oriented horizontally
horizontal, and bipolar cells, all of which appear to contribute and mainly conned to layer V, and an atypical form of mul-
to the perforant pathway. This large, heterogeneous group tipolar neuron with long wavy dendrites. A surprising aspect
of cells undoubtedly coincides with what Lorente de N of the latter neurons is that some of their dendrites meander
described as atypical layer III neurons. from the entorhinal cortex, through the angular bundle, and

Figure 340. Morphological characteristics of entorhinal cortex layer V neurons. Computer-

aided reconstructions of pyramidal, horizontal, and polymorphic cells. Dashed lines represent
the borders of the indicated layers. Bar  100 m. (Source: Adapted from Hamam et al., 2002.)
88 The Hippocampus Book

into the subiculum. Some of these cells may thus contribute to tial way to the portions of the entorhinal cortex that project to
the projections to the dentate gyrus and the hippocampus. other levels of the dentate gyrus. Thus, the associational con-
The overall picture emerging from these recent studies is that nections seem to be organized to integrate all of the informa-
all three types of layer V neuron should be considered projec- tion that comes into a particular portion of the entorhinal
tion neurons in that they send an axon to the white matter. cortex and are relayed to a particular septotemporal level of
They also may function as local circuit neurons, connecting the dentate gyrus.
the deep layers to the supercial layers. Not much is known about the detailed microcircuitry of
Layer VI contains a wide variety of neuron types. Based on the entorhinal associational connections. Presumably, the
the predominant distribution of their axonal plexus, these layer II cells that project to other portions of layer II terminate
cells can be grouped into three categories: cells that mainly on the same stellate and pyramidal cells that give rise to the
inuence other cells in layer VI or Vb; cells that by means perforant path projection. It is not known, however, if associ-
of their highly collateralized axons can inuence a vertical ational connections also terminate on interneurons in layer II.
column of cells in layers I to III; and cells whose axons Of even greater functional signicance is the issue of whether
are directed toward the deep white matter and are there- the deep layer cells project to the cells of layers II and III that
fore likely to be projection neurons. Some of these cells con- give rise to the perforant pathway. This is a critical missing
tribute to the projections to the dentate gyrus and the hip- piece of information. Because the deep layer neurons receive
pocampus. feedback projections from CA1 and the subiculum, an associ-
On the basis of their restricted axonal distribution, several ational connection from deep cells to supercial cells would
of the smaller cell types in the entorhinal cortex have been provide the link for completing the loop through the hip-
classied as interneurons. Most of these interneurons are pocampal formation. Another issue is whether these associa-
likely GABAergic. GABAergic neurons are found in all layers tional connections are excitatory or inhibitory. If a substantial
of the entorhinal cortex, although they are most abundant portion of the deep to supercial pathway originates from
in the supercial layers. GABAergic interneurons can be sub- GABAergic neurons, or if the pathway is excitatory but termi-
categorized on the basis of their colocalization with various nates preferentially on inhibitory cells located in layer II, the
neuroactive substances (e.g., peptides) or their expression of output of the hippocampal formation from CA1 and the
one or more of the various calcium-binding proteins. The subiculum could inhibit layer II or III neurons that project to
supercial layers contain a large number of cells that are the dentate gyrus, hippocampus, and subiculum.
immunoreactive for parvalbumin, whereas the number of
calbindin-D28-positive neurons is much lower. Although Commissural Connections
most of the GABAergic neurons are undoubtedly interneu-
rons, at least some of the GABAergic neurons in layers II and Relatively strong commissural connections, arising from all
III project to the dentate gyrus. portions of the entorhinal cortex, terminate predominantly in
layers I and II of the homotopic area of the entorhinal cortex
Intrinsic/Associational Connections (Goldowitz et al., 1975; Hjorth-Simonsen and Zimmer, 1975;
Deller, 1998). The entorhinal cortex also gives rise to a com-
The entorhinal cortex contains a substantial system of associ- missural projection to other components of the contralateral
ational connections. Intraentorhinal bers are organized in hippocampal formation. The largest component of this pro-
three rostrocaudally oriented bands, and connections that link jection is directed toward the dentate gyrus, but elds CA3
different transverse (or mediolateral) regions of the entorhi- and CA1 of the hippocampus and the subiculum also receive
nal cortex are rather restricted (Dolorfo and Amaral, 1998b). a contralateral input. The crossed entorhinal projection is
Associational connections originate in both supercial and heaviest to septal portions of the hippocampal subelds and
deep layers. Projections originating from layers II and III tend rapidly diminishes in strength at more temporal levels. This
to terminate mainly in the supercial layers, whereas projec- crossed projection apparently arises exclusively from layer
tions originating from the deep layers terminate in both the III cells.
deep and supercial layers.
The global organization of the associational connections in Organization of the Perforant
the entorhinal cortex can be best understood in relation to the and Alvear Pathway Projections
topography of the entorhinal projections to other elds of the
hippocampal formation (Fig. 341). As we describe shortly, As a reminder of our earlier description of the perforant path
different parts of the entorhinal cortex project to different input to the dentate gyrus, hippocampus, and subiculum,
septotemporal levels of the dentate gyrus. The portions of the both the lateral and medial entorhinal areas project to all three
lateral and medial entorhinal areas that project to the septal areas. The lateral and medial components of the perforant
half of the dentate gyrus, for example, are located laterally and path terminate along the supercial-to-deep gradient in the
caudally in the entorhinal cortex, close to the rhinal sulcus. molecular layer of the dentate gyrus and the stratum lacuno-
Cells located in this region give rise to associational connec- sum-moleculare of CA3 and along the transverse axis of CA1
tions to other cells in the same region but not in any substan- and the subiculum (Fig. 342). Layer II cells give rise to the
Hippocampal Neuroanatomy 89

Figure 341. Summary of the intrinsic connections of the entorhi- projection zones has substantial associational connections (large
nal cortex and the association of the entorhinal cortex with the hip- arrows) that remain largely in the zone of origin. Projections
pocampal formation and neocortex. A. Organization of entorhinal between each band are less prominent (small arrows). C. Entorhinal
projections to the dentate gyrus. A band of layer II cells located in intrinsic projections and entorhinal afferent and efferent projec-
the lateral and caudomedial portion of the entorhinal cortex (light tions. Entorhinal associational connections arise from both the deep
gray) projects to the septal half of the dentate gyrus; a band in the and supercial layers of the entorhinal cortex (arrows in the boxes)
mid-mediolateral entorhinal cortex (medium gray) projects to the and terminate mainly in the supercial layers. A less prominent
third quarter of the dentate gyrus; and a band located in the most associational connection in the deep layers is apparent primarily in
rostromedial entorhinal cortex (dark gray) projects to the temporal lateral portions of the entorhinal cortex. (Source: Adapted from
pole of the dentate gyrus. B. Organization of entorhinal intrinsic Dolorfo and Amaral, 1998b.)
projections: projections within and between bands. Each of these
90 The Hippocampus Book

Figure 342. Laminar and topographical organization of the the subiculum, layer III entorhinal cortex projections are organized
entorhinal projection to the dentate gyrus, the hippocampus, and topographically; the LEA projects to the distal CA1 and proximal
the subiculum. The surface of the entorhinal cortex is represented subiculum (i.e., at the CA1/subiculum border), and the MEA proj-
on the left (the rhinal sulcus is to the left). Layer II entorhinal cor- ects to the proximal CA1 and distal subiculum. Laterally situated
tex projections to the dentate gyrus and the CA3 and CA2 elds of portions of the entorhinal cortex project to septal levels of the hip-
the hippocampus terminate in a laminar fashion; the LEA projects pocampal formation, whereas progressively more medial portions
supercially in the molecular layer and the stratum lacunosum- of the entorhinal cortex project to more temporal levels of the hip-
moleculare, and the MEA projects deeper. In contrast in CA1 and pocampal formation.

projection to the dentate gyrus and CA3, and layer III cells and medial entorhinal areas) projects to the septal half of
project to CA1 and the subiculum. Although current usage the dentate gyrus. Most of the cells in this domain project to
applies the term perforant path to all entorhino-hippocampal all portions of the septal half of the dentate gyrus; thus, the
projections, entorhinal bers also reach CA1 via the alveus projection is both divergent and convergent. The next domain
(i.e., the alvear pathway originally described by Ramon y is more medially situated and projects to the third quarter of
Cajal). In the temporal portion of the hippocampus, most of the dentate gyrus. The last domain is medially and rostrally
the entorhinal bers reach CA1 after perforating the subicu- situated and projects to the temporal quarter of the dentate
lum (classic perforant pathway). At more septal levels, how- gyrus.
ever, the number of entorhinal bers that take the alvear There are a number of functional implications of the
pathway increases; and in the septal portion of the hippo- organization of these projections. First and foremost, because
campus, most of the entorhinal bers reach CA1 via the alvear the associational connections of the entorhinal cortex seem to
pathway. These bers make a sharp turn in the alveus, perfo- respect this tripartite organization, it is reasonable to think of
rate the pyramidal cell layer, and terminate in the stratum three functional, parallel systems encompassed within the
lacunosum-moleculare. Both pathways demonstrate the same entorhino-hippocampal system. Second, when conducting
septotemporal organization of their projections. Thus, certain stimulation or lesion experiments designed to evaluate the
portions of the entorhinal cortex project to certain septotem- contributions of the medial and lateral perforant paths to hip-
poral levels of the other hippocampal elds. Following this pocampal function, one must be careful to bear in mind that
brief overview, we now delve more deeply into the topograph- the cells giving rise to the medial perforant path are located
ical organization of the perforant path projection. caudal (not medial) to the cells that give rise to the lateral per-
Dolorfo and Amaral (1998a) have shown that cells located forant path.
laterally and caudally in the entorhinal cortex project to septal One nal comment must be made on the medial and lat-
levels of the hippocampal elds, whereas cells located progres- eral perforant path projections. Although the cellular charac-
sively more medially and rostrally (in the medial entorhinal teristics of layers II and III of the medial and lateral entorhinal
cortex) project to more temporal levels of the hippocampal areas are similar, the lateral and medial perforant path pro-
subelds (Fig. 341). It now appears that there may be three jections demonstrate a number of differential features. Both
largely nonoverlapping domains of the entorhinal cortex that components of the perforant path use glutamate as their pri-
project to three distinct levels of the dentate gyrus. The later- mary transmitter, but they exhibit differential distributions of
ally situated domain (encompassing cells of both the lateral glutamate receptors. The medial perforant path bers, for
Hippocampal Neuroanatomy 91

example, are immunoreactive for the metabotropic glutamate the raw material the hippocampal formation uses to accom-
receptor mGLUR 2/3, whereas the lateral bers are not. In plish its purported function(s). This is an area in which there
contrast, the lateral perforant path demonstrates dynorphin are substantial species differences. Keeping to the format that
immunoreactivity, whereas the medial perforant path does we have followed so far, we give an overview of the inputs to
not. One would have thought that some of these differences the rat entorhinal cortex. However, doing so does a disservice
would show up more clearly in the cells of origin, but thus far to the signicance of the connections between the entorhinal
there is no distinctive marker at the level of the entorhinal cor- cortex and the neocortex, as these connections are so much
tex for cells that give rise to the lateral and medial perforant more prominent in the monkey brain. Thus, when we provide
path projections. our comparison of the organization of the hippocampal for-
mation in the rat, monkey, and human in a later portion of the
Feedback Projections from the Hippocampus chapter, we return to a detailed overview of the cortical inputs
and Subiculum of the monkey entorhinal cortex.
Burwell has carried out a thorough quantitative analysis of
We already mentioned that the dentate gyrus and the CA3 the organization of cortical inputs to the rat entorhinal cortex
eld of the hippocampus do not project back to the entorhi- and compared the results with inputs to the perirhinal and
nal cortex. Thus, the recipients of the layer II projection do postrhinal cortices (Fig. 343) (Burwell and Amaral, 1998a).
not have any direct inuence over the activities of the entorhi- She found many similarities in the complement of cortical
nal cortex. It is only after the layer II and layer III projection inputs to the lateral and medial entorhinal areas. Each receive
systems are combined in CA1 and the subiculum that return about one-third of its total input from the piriform cortex
projections to the entorhinal cortex are generated. The return (LEA 34%, MEA 31%). These areas also receive roughly equal
projections mainly terminate in the deep layers (V and VI), proportions from temporal (LEA 26%, MEA 21%) and frontal
although some bers ascend into layer I (Witter et al., 1988; (LEA 11%, MEA 10%) regions. Some differences are observed
Naber et al., 2001; Kloosterman et al., 2003). It is not known in the proportions of insular, cingular, parietal, and occipital
which entorhinal neurons are the recipients of these return inputs. The lateral entorhinal area receives more input from
projections. What is clear, however, is that the projections insular cortex (LEA 21%, MEA 6%). In contrast, the medial
from CA1 and the subiculum to the entorhinal cortex are also entorhinal area receives more input from cingulate (LEA 3%,
topographically organized (Figs. 329 and 334). Septal por- MEA 11%), parietal (LEA 3%, MEA 9%), and occipital (LEA
tions of CA1 and the subiculum project chiey to lateral parts 2%, MEA 12%) regions. Not all portions of the rat entorhinal
of the entorhinal cortex, and more temporal parts of CA1 and cortex receive substantial cortical input. In fact, it is only the
the subiculum project to more medial parts of the entorhinal lateral and caudal parts of the entorhinal cortex (those pro-
cortex. Moreover, the transverse location of the cells of origin jecting to septal levels of the dentate gyrus) that are heavily
in CA1 and the subiculum also determines whether these pro- innervated by the neocortex. This neuroanatomical organiza-
jections terminate in the medial or lateral entorhinal cortex. tion has strong functional implications and likely underlies
The projections from the proximal part of CA1 and the distal the behavioral dissociations observed with dorsal (septal) ver-
part of the subiculum distribute exclusively to the medial sus ventral (temporal) hippocampal lesions.
entorhinal cortex, whereas cells located in the distal part of The neocortical inputs to the entorhinal cortex of the rat
CA1 and the proximal part of the subiculum project mainly to comprise two groups: those that terminate in the supercial
the lateral entorhinal cortex. layers (IIII) and those that terminate in the deep layers
The important point about these return projections is (IVVI). The rst category delivers information to the super-
that they are exactly in register (i.e., they are point-to-point cially located entorhinal neurons, which are the source of the
reciprocal) with the entorhinal inputs to these areas. Thus, projections to the dentate gyrus, hippocampus, and subicu-
at the global level, all of the circuitry is available for rever- lum. The second group has greater inuence on the deeply
beratory circuits to be established through the loop, starting located cells of the entorhinal cortex, which receive processed
and ending at the entorhinal cortex. This remarkable topogra- information from the other hippocampal elds and give rise
phy conrms the critical role of the entorhinal cortex with to feedback projections to certain cortical regions. In general,
respect to the input to and output from the hippocampal for- the cortical afferents that reach the deep layers terminate
mation. rather diffusely, whereas those that terminate supercially
have a more restricted mediolateral and/or rostrocaudal dis-
Extrinsic Connections tribution. Although speculative, the rst group may constitute
a set of information-bearing inputs, whereas the second group
Interconnections of the Entorhinal Cortex may have more of a modulatory inuence on the output of
with Neocortical Regions the hippocampal formation.
If the entorhinal cortex is viewed as the rst step of processing A substantial input to the supercial layers of the entorhi-
in the hippocampal formation, it is reasonable to wonder nal cortex originates from olfactory structures such as the
what types of information it receives. In other words, what is olfactory bulb, the anterior olfactory nucleus, and the piri-
92 The Hippocampus Book

Figure 343. Pattern and strength of cortical and intrinsic connec- AId/v/p, dorsal, ventral, and posterior agranular insular cortices;
tivity of the rat parahippocampal region (entorhinal, perirhinal, and AUD, primary auditory cortex; MOs, secondary motor area; Pir, pir-
parahippocampal cortices). The thickness of the solid lines repre- iform cortex; PTLp, posterior parietal cortex; RSPd and RSPv, retro-
sents the relative strength of the connections based on the densities splenial cortex, dorsal and ventral; SSp and SSs, primary and
of retrogradely labeled neurons. Open lines represent reported con- supplementary somatosensory areas; Tev, ventral temporal area;
nections for which no comparable quantitative data are available. VISC, visceral granular insular cortex; VIS1 and VISm, lateral and
The weakest projections ( 250 labeled cells/mm3) are not shown medial visual association cortex; VISp, primary visual cortex.
here. ACAd and ACAv, dorsal and ventral anterior cingulate cortex; (Source: Adapted from Burwell and Amaral, 1998b.)

form cortex. These olfactory projections terminate through- sively in our discussion of the monkey entorhinal cortex, but
out most of the rostrocaudal extent of the entorhinal cortex, it is important to give an overview of this area because it pro-
mainly in layer I and the supercial portion of layer II. Only vides major input to the rat entorhinal cortex. The perirhinal
the most caudal portion of the rat medial entorhinal area does cortex in the rat is made up of two areas, 35 and 36, which
not receive any olfactory input. In addition to terminating on appear to receive slightly different complements of neocorti-
the principal cells of these layers, olfactory bers also termi- cal inputs (Fig. 343) (Burwell et al., 1995; Burwell, 2001).
nate on layer I GABAergic neurons, which presumably inter- Area 36 of the perirhinal cortex receives more, higher-level
act with the principal cells in layers II and III. cortical input than does area 35. Most of this input comes
A second prominent cortical input to the supercial layers from the ventral temporal associational area (Tev), which is
of the entorhinal cortex arises from the laterally adjacent located dorsally adjacent to area 36. Other major inputs are
perirhinal and postrhinal cortices. The perirhinal and postrhi- from the postrhinal cortex and the entorhinal cortex. The
nal cortices are polysensory convergence areas that receive predominant inputs to area 35 arise from the piriform,
inputs from a variety of unimodal and polymodal sensory entorhinal, and insular cortices. Area 35 receives more than
cortices. The perirhinal cortex terminates mainly in the lateral one-fourth of its input from the piriform cortex and only
entorhinal area, and the postrhinal cortex terminates heavily, slightly less from the lateral entorhinal area. About one-fth of
though not exclusively, in the medial entorhinal area. In both the total input arises from insular cortices. Interestingly, the
cases, the projections terminate preferentially in layers I and perirhinal projections to the entorhinal cortex arise preferen-
III of the entorhinal cortex. tially from area 35, and the intrinsic projections of the perirhi-
We address the issue of the perirhinal cortex more exten- nal cortex seem to be organized to funnel information into
Hippocampal Neuroanatomy 93

area 35, which in turn gives rise to the main perirhinal projec- additional sources of information. The substantial input from
tions to the entorhinal cortex. the amygdaloid complex, which originates mainly from the
Cortical afferents to the deep layers of the entorhinal cor- lateral and basal nuclei, is presumably conveying information
tex arise from a variety of cortical areas. These areas include about the emotional state of the organism (Pikkarainen et al.,
projections from the agranular insular cortex, the medial pre- 1999a). The densest termination of the amygdaloid bers is in
frontal region (particularly the infralimbic, prelimbic, and the ventrolateral part of the entorhinal cortex (Fig. 344).
anterior cingulate cortices and the retrosplenial cortex Efferents from the lateral amygdaloid nucleus distribute most
(Insausti et al., 1997). intensely to the deep portion of layer III but end also between
Efferents of the entorhinal cortex return projections to the cell islands of layer II and in layer I. The bers from the
many of the cortical areas that provide input to the entorhinal basal nucleus terminate diffusely in layers III to V, whereas
cortex. There are projections to olfactory areas, originating those from the cortical nuclei and the periamygdaloid cortex
predominantly from layers II, III, and Va. An important issue preferentially project to layers I and II. The entorhinal cortex
is whether the entorhinal cortex of the rat, like that of the sends feedback projections to the amygdala that terminate
monkey (see below), gives rise to prominent, widespread pro- mainly in the basal nucleus. These projections originate from
jections to multimodal association cortices. Initial studies in cells in layer V, although a few cells in more supercial layers
the rat indicated that the entorhinal cortex projects mainly to may also contribute to the projection. Dense inputs also orig-
adjacent, limited portions of the temporal cortex. Swanson inate from the ventral part of the claustrum or endopiriform
and Khler (1986), however, suggested that the rat entorhinal nucleus, but the topographical and laminar organization of
cortex gives rise to projections that reach a much larger these inputs is not well understood.
domain of the cortical surface. Whether one believes that the The entorhinal cortex projects bilaterally to the striatum,
entorhinal cortex has widespread neocortical connections particularly to the nucleus accumbens and adjacent parts of
seems to hinge on the demarcation of the entorhinal cortex the olfactory tubercle. These projections originate mainly
from the adjacent perirhinal cortex. Insausti and colleagues from layer V and are topographically organized; medial parts
conrmed the report by Sarter and Markowitsch (1985) that of the entorhinal cortex project to the caudomedial portion of
the only cells that give rise to the extensive connections the nucleus accumbens, and more lateral portions of the
reported by Swanson and Khler are located in the most dor- entorhinal cortex project to more lateral parts of the nucleus.
solateral region of the entorhinal cortex (i.e., on the border of
the entorhinal and perirhinal cortices). Whether these cells Basal Forebrain and Hypothalamic Connections
indeed belong to the deep layers of the entorhinal cortex or The entorhinal cortex receives its cholinergic innervation
form a population of perirhinal cells is not yet clearly estab- mainly from the septum. This projection is topographically
lished. Based on the distribution of immunoreactivity for par- organized such that cells in the horizontal limb of the nucleus
valbumin or for certain glutamate receptor subunits, the of the diagonal band preferentially project to the most lateral
border between the entorhinal and perirhinal cortex appears part of the entorhinal cortex, whereas the medial septal
to be oblique and perhaps intermixed in this critical region. nucleus and the vertical limb of the nucleus of the diagonal
Thus, it remains unresolved at this point whether it is the band project to more medial parts of the entorhinal cortex.
entorhinal cortex or the perirhinal cortex that projects widely Septal projections terminate densely in the cell-sparse lamina
to the neocortex. One can safely conclude, however, that the dissecans and less densely in layer II.
major portion of the entorhinal cortex does not contribute Like the hippocampus and the subiculum but unlike the
projections to the unimodal areas of the neocortex, and that pre- and parasubiculum, the entorhinal cortex projects back
the bulk of neocortically directed projections are to higher- to the septal region. The projection originates from cells in
order associational and polysensory cortices and not to sen- layer Va, although some layer II cells also contribute to these
sory or motor regions. In this respect, the situation in the rat projections. The bers from the entorhinal cortex preferen-
closely resembles the connectivity observed in the monkey. tially terminate in the lateral septal complex.
Among the cortical areas that do receive entorhinal input are The entorhinal cortex receives diffuse inputs from various
the infralimbic, prelimbic, orbitofrontal, agranular insular, structures in the hypothalamus. They include inputs from the
perirhinal, and postrhinal cortices; these projections originate supramammillary nucleus that terminate rather diffusely with
mainly from cells in layer Va. Only weak projections have been some preference for layers III to VI, from the tuberomammil-
reported to the retrosplenial cortex. lary nucleus distributing diffusely throughout the entorhinal
cortex, and from the lateral hypothalamic area reaching the
Other Telencephalic Connections: deep layers of the entorhinal cortex.
Amygdala, Claustrum, Striatum
In addition to the cortical inputs just described, the entorhi- Thalamic Connections: Nucleus Reuniens
nal cortex receives a number of subcortical inputs. Whereas and Other Midline Nuclei
some of them, such as the monoaminergic and cholinergic The major thalamic inputs to the entorhinal cortex originate
inputs, may be viewed as largely modulatory, others, such as in the nucleus reuniens and the nucleus centralis medialis
the input from the amygdaloid complex, might also provide (Van der Werf et al., 2002). The rhomboid, paraventricular,
94 The Hippocampus Book

Figure 344. Summary of the reciprocal connections between the of the medial entorhinal cortex do not receive amygdala inputs.
amygdala and the hippocampal formation and the perirhinal and Areas DLE, DIE, VIE, and AE of the lateral entorhinal cortex receive
parahippocampal cortices. The subdivisions of the entorhinal cortex inputs from the various amygdala nuclei. (Source: Pitknen et al.,
follow the nomenclature of Insausti et al. (1997). Areas CE and ME 2000.)

and parataenial nuclei contribute minor projections. The plies the entorhinal cortex with a diffusely organized nora-
nucleus reuniens bers densely innervate the deep portion of drenergic input that exhibits slightly more dense termination
layers I and III and give rise to a few collaterals that extend in layer I.
into layer II. Separate populations of nucleus reuniens cells
project to the entorhinal cortex, CA1, and the subiculum.
There is no evidence that the entorhinal cortex projects back
to the thalamus. 3.5 Chemical Neuroanatomy

Brain Stem Inputs 3.5.1 Transmitters and Receptors

The entorhinal cortex receives dopaminergic input from
cells located in the ventral tegmental area. The projection Detailed, comprehensive treatment of the chemical neu-
preferentially terminates in a restricted rostrolateral part of roanatomy of the hippocampal formation would demand
the lateral entorhinal area, where bers are arranged in dense, a lengthy chapter of its own. We have already described the
columnar patches in layers I to III. The serotonergic innerva- distribution of systems dened by their neurotransmitter con-
tion arises from the central and dorsal raphe nuclei and tentnoradrenaline (norepinephrine), serotonin, acetyl-
terminates diffusely in all layers but with a preference for choline, GABAand a variety of reviews are available on the
the supercial layers. The noradrenergic locus coeruleus sup- chemical neuroanatomy of the hippocampus (Swanson et al.,
Hippocampal Neuroanatomy 95

1987; Kobayashi and Amaral, 1998). We thus restrict ourselves eralcorticoid receptors. Receptors for both steroids were
here to providing an overview of the diversity of neurochem- highly expressed in cells of the dentate gyrus, CA3, and CA2.
ical substances in the hippocampal formation. Lower levels of expression were observed in CA1. This is the
A variety of peptides and other chemical markers have opposite of the distribution observed in the rat and the tree
been shown to subdivide the population of GABAergic shrew.
interneurons. Among the peptides that colocalize with partic-
ular populations of GABAergic interneurons are VIP, somato-

statin, NPY, corticotropin-releasing factor (CRF), substance P,
3.6 Comparative Neuroanatomy of the Rat,
cholecystokinin, galanin, and the opioid peptides dynorphin
Monkey, and Human Hippocampal Formation
and enkephalin. It is worth commenting further on the dis-
tribution of the opioid peptides because, in addition to being
When one views Nissl-stained sections of the hippocampus
localized to certain populations of interneurons, they are
from the rat, monkey, and human, it is immediately apparent
observed in intrinsic excitatory pathways of the hippocampal
that one is looking at the same brain region (Figs. 32 and
formation. The bers of the lateral perforant path, for exam-
345). The densely packed granule cell layer is obvious in all
ple, are immunoreactive for Leu-enkephalin, whereas the
three species, as is the progressively more complex lamination
mossy bers of the dentate gyrus are positive for dynorphin.
when one progresses from the dentate gyrus to the entorhinal
Another class of substances that appear to mark certain
cortex. On closer inspection, however, a number of differences
subsets of GABAergic neurons selectively is the family of
make it clear that the hippocampal formation of the monkey
calcium-binding proteins, including parvalbumin, calbindin,
or human is not simply a scaled-up version of the rat hip-
and calretinin. Whereas parvalbumin immunoreactivity
pocampal formation. Some of the hippocampal elds, such as
appears to be exclusively conned to a subset of GABAergic
CA1 and the entorhinal cortex, are disproportionately larger
interneurons, the other calcium-binding proteins can be
in the primate. The entorhinal cortex has many more subdivi-
found in both interneurons and principal neurons. Although
sions in the monkey and human than in the rat; and the lam-
the precise function of the various calcium-binding proteins
inar organization is much more distinct in the primate brain.
has not been well established, their existence has provided
In the following sections, we review some of the similarities
a useful anatomical tool. Although standard immunohisto-
and differences in the organization and connections of the
chemistry of GABAergic neurons with antibodies to glutamic
acid decarboxylase (GAD) or to GABA does not label den-
drites very well, parvalbumin-immunoreactive neurons are
fully labeled in a Golgi-like fashion. Thus, even though par-
valbumin does not label all GABAergic neurons, those that are
labeled can be subjected to precise analyses of their inputs.

3.5.2 Steroids

Neurons in the hippocampal formation have been shown to

concentrate glucocorticoids and mineralcorticoids (McEwen
and Wallach, 1973). Studies using uptake of 3H-corticosterone
show that cells in CA2 and CA1 exhibit the greatest uptake.
There is slightly less uptake in CA3 cells and cells of the poly-
morphic layer of the dentate gyrus; dentate granule cells also
demonstrate some 3H-corticosterone uptake. The amount of
uptake in other elds of the hippocampal formation has not
yet been described. What is perhaps most surprising is the fact
that, in the rat, the hippocampus demonstrates the highest
level of uptake of any brain region. The distribution of gluco-
corticoid receptor mRNA has also been evaluated in the tree
shrew. Here the highest density of mRNA was observed in the
pyramidal cells of the subiculum and in the granule cells of
the dentate gyrus. In the hippocampus, higher densities of
mRNA were observed in CA1 than in CA3. Mineralcorticoid
receptor mRNA is also expressed in the tree shrew hippocam-
pus. It is expressed most strongly in CA1 and less strongly in
CA3; this contrasts with the pattern in the rat, where mineral- Figure 345. Nissl-stained coronal sections through the hippocam-
corticoid expression is higher in CA3. pal formation of the rat (A), monkey (B), and human (C) presented
In situ hybridization has also been used in postmortem at the same magnication. Bar in C  2 mm and applies to all
humans to study the distribution of glucocorticoid and min- panels.
96 The Hippocampus Book

hippocampal formation in these three species. Of course, Cytoarchitectonic Organization

although we can address certain issues concerning cell num-
ber and distribution in the human brain, we are unable to say There are several cytoarchitectonic differences between the rat
anything about patterns of connectivity in the human hip- and monkey hippocampal formation. The polymorphic layer
pocampal formation. of the dentate gyrus is relatively smaller in the monkey, and
much of the territory located enclosed within the blades of the
3.6.1 Neuron Numbers granule cell layers is occupied by the CA3 eld. This hilar
portion of the CA3 region is so expansive that some
The number of neurons present in the various subdivisions of researchers have confused it for an enlarged polymorphic
the hippocampal formation have been counted by several layer. Cells in this region bear the gold standard of inclusion
investigators, but rarely has the same author analyzed the rat, in CA3, however, as they give rise to Schaffer collaterals to
monkey, and human hippocampal formations. Moreover, CA1.
there is a lot of variability in the estimates published in the lit- The other obvious difference between the rat and monkey
erature, and few investigators have used modern stereological hippocampal formation is the much thicker pyramidal cell
techniques to carry out these assessments. We have compiled layer in the monkey CA1 (Fig. 349). In the rat, this layer is
a summary of current estimates of neuron number obtained tightly packed and is typically about 5 cells thick. In the mon-
with modern stereological techniques. The data presented key, the cell layer is much more diffusely organized and is 10
here are derived from published and unpublished work from to 15 cells thick. As a result, not only is the boundary between
three laboratories: Mark West at Aarhus University, Peter the pyramidal cell layer and the stratum radiatum less clear, so
Rapp at New York University, and our own at the University of is the boundary between CA1 and the subiculum. In the mon-
California Davis (Table 3-1). key, at least some of the Schaffer collaterals from CA3 termi-
In our initial comparison of the hippocampal formation in nate in the pyramidal cell layer, presumably on the apical
the rat, monkey, and human, the volume of the hippocampal dendrites of cells located deep in the layer or on the basal den-
complex (dentate gyrus  hippocampus) was said to be about drites of neurons located supercially in the layer.
10 times larger in monkeys than in rats and 100 times larger The entorhinal cortex exhibits major cytoarchitectonic dif-
in humans than in rats (rat 32 mm3, monkey 340 mm3; ferences between the rat and the monkey. The laminar organ-
human 3300 mm3). If, however, we compare the total number ization of the monkey entorhinal cortex is much clearer than
of neurons in the various hippocampal areas, we observe that that in the rat. Throughout much of the entorhinal cortex, for
certain regions are comparatively more developed than others example, there is a clear distinction between layers V and VI in
in the monkey and human brain. In the dentate gyrus, there the monkey, whereas these layers tend to blur together in the
are approximately 10 times more granule cells in the monkey rat. The monkey entorhinal cortex is also much more differ-
than in the rat, a ratio that parallels the overall volume differ- entiated than in the rat. As noted previously, Amaral and col-
ences. However, there are only 15 times more dentate granule leagues typically divided the rat entorhinal cortex into only
cells in humans compared to rats, whereas the volume of the lateral and medial areas, whereas they divided the monkey
dentate gyrus plus the hippocampus is about 100 times larger entorhinal cortex into seven distinct subdivisions.
in humans than in rats. In CA1, however, there are only three
times more pyramidal cells in the monkey than in the rat, Neuron Types
whereas there are 35 times more cells in humans than in rats.
We must add, however, that some of these estimates are still Until recently, there has been little direct comparison of the
provisional and need to be replicated before drawing solid morphology of similar neurons in the rat and monkey hip-
inferences on the relative development of the various hip- pocampus. A few studies, using either the Golgi technique or
pocampal regions in rats, monkeys, and humans. intracellular staining techniques, have compared easily identi-
ed cell types.
3.6.2 Comparison of Rat and Monkey The granule cells of the monkey dentate gyrus have been
Hippocampal Formation examined with both the Golgi technique and intracellular ll-
ing techniques in the in vitro slice preparation (Duffy and
In monkeys, the hippocampal formation lies entirely within Rakic, 1983; Seress and Mrzljak, 1987, 1992). Depending on
the temporal lobe (Figs. 346 through 348) and lacks the which aspects of these studies one wishes to emphasize, it
pronounced C shape along its septotemporal axis that is so could be concluded that granule cells are basically similar in
characteristic in the rat. The region equivalent to the septal rats and monkeys or that there have been substantial modi-
pole of the rat hippocampus is located caudally in the mon- cations of at least some granule cells. In general, dentate gran-
key, and the equivalent of the temporal pole is located ros- ule cells have the same unipolar apical dendritic tree in the
trally (Figs. 35 and 36). Much of the entorhinal cortex is monkey as in the rat. The total dendritic length of each gran-
located rostral to the remainder of the hippocampal forma- ule cell is also relatively similar in the rat and the monkey.
tion. In fact, the rostral half of the entorhinal cortex lies ven- Seress and colleagues were the rst to point out that at least
tral to the amygdaloid complex rather than the hippocampus. some monkey granule cells have basal dendrites (Fig. 350),
Hippocampal Neuroanatomy 97

Figure 346. Coronal sections at three rostrocaudal levels through the monkey brain show the
relative position of the hippocampal formation. The three panels on the left are Nissl-stained
sections; the three panels on the right are adjacent sections stained with Timms sulde silver
stain. The line drawings (middle column) highlight the regions of the monkey hippocampal
formation seen in the stained sections. Bar  2 mm.

and a similar observation has been made by Scheibel and col- the dentate gyrus (Buckmaster et al., 1992; Buckmaster and
leagues. This feature has never been reported for normal rat Amaral, 2001). In the rat, these cells give rise to the associa-
granule cells. Thus, it appears that there are some species dif- tional-commissural connections to the inner portion of the
ferences in dentate granule cell morphology, but the func- molecular layer, and its dendrites are generally conned to the
tional signicance of these differences is yet unclear. polymorphic area (i.e., the dendrites extend neither into the
Another example of differences in an ostensibly similar molecular layer nor into the adjacent CA3 eld). The mossy
cell type in the rat and the monkey has emerged from intra- cell in the monkey is quite different. First, there appear to be
cellular staining of mossy cells of the polymorphic region of at least two forms. One is very much like the rat mossy cell,
98 The Hippocampus Book

Figure 347. Horizontal sections at three dorsoventral levels through the monkey brain show
the relative position of the hippocampal formation. The three panels on the left are Nissl-
stained sections; the three panels on the right are adjacent sections stained with Timms sulde
silver stain. The line drawings (middle column) highlight the regions of the monkey hippocam-
pal formation seen in the stained sections. Bar  2 mm.

with dendrites conned to the polymorphic cell layer and tion in the molecular layer of the dentate gyrus, whereas stan-
axons directed to the molecular layer. There is a second type, dard mossy cells are not (because the perforant path does not
however, that extends much of its dendritic arbor into the enter the polymorphic layer). Moreover, in rats the granule
molecular layer (Fig. 351). Moreover, many of these cells give cells are the only input to CA3, whereas in the monkey the
rise to projections into the adjacent CA3 region. The implica- mossy cells appear to contribute an additional projection.
tion of this altered morphology in the monkey is that these Although these structural alterations must be conrmed by
mossy cells are capable of receiving perforant path innerva- functional studies, they suggest that there are fundamental
Hippocampal Neuroanatomy 99

Figure 348. Sagittal sections at three mediolateral levels through the monkey brain show the
relative position of the hippocampal formation. The three panels on the left are Nissl-stained
sections; the three panels on the right are adjacent sections stained with Timms sulde silver
stain. The line drawings (middle column) highlight the regions of the monkey hippocampal
formation seen in the stained sections. Bar  2 mm.
100 The Hippocampus Book

Figure 349. High magnication photomicrographs of Nissl- dal cell layer is approximately 15 cells thick and shows some sub-
stained coronal sections through the CA1 eld of the hippocampus lamination, with the top half of the layer having a slightly higher
in the rat (A), monkey (B), and human (C). Note that the CA1 density of neurons. The pyramidal cell layer in the human CA1 is
pyramidal cell layer in the rat (which is the darkly stained layer at even thicker and has a more laminated structure. Bar in A  200
the bottom of panel A) is about 5 cells thick. The monkey pyrami- m and applies to all panels.

differences regarding the circuit characteristics of the dentate little work has been carried out in the monkey, however, to
gyrus between the rat and the monkey. know if the topographical organization of intrinsic circuits is
the same in the two species. So far, minor differences have been
Connections observed. In the monkey, for example, perforant path bers
arising from the rostral entorhinal area (the equivalent of the
Basic Organization of the Intrinsic Hippocampal Circuitry rat lateral entorhinal area) terminate, as in the rat, mainly in
To the extent that it has been examined, the basic principles the outer third of the molecular layer of the dentate gyrus.
of organization of the intrinsic circuitry of the monkey hip- Some terminations, however, also continue in a decreasing
pocampal formation resemble those observed in the rat. Too gradient fashion into the middle third of the molecular layer.

Table 31.
Number of principal neurons in the subdivisions of the rat, monkey,
and human hippocampal formation (in millions)

Cells Rat Monkey Human

Gcl DG 1.20a.b 12c 18d

Hilus 0.05b 0.23e 1.72d
CA3 0.25a,b 1.27e 2.83d
CA1 0.39a,b 1.30e 14d
Subiculum 0.29b 0.75e 5.95d
Pre/parasubiculum 0.70f 2.08e
EC II 0.11f 0.26e 0.66g
EC III 0.25f 3.66g
EC V  VI 0.33f 3.78g
EC total 0.69f 2.16e 8.1g

Sources: aRapp and Gallagher (1996); bWest et al. (1991); cLavenex et al. unpublished data; dWest et al. (1994); eRapp et al.
unpublished data; fMulders et al. (1997); gWest and Slomianka (1998).
Hippocampal Neuroanatomy 101

Figure 350. Camera lucida drawings of

primate granule cells. A. Monkey granule
cell in the upper part of the granule cell
layer (gcl). The dendrites are cone-shaped
in the molecular layer (ml); the axon (a)
originates from the base of the soma and
extends into the polymorphic layer (pl).
B. Monkey granule cell in the deep part
of the granule cell layer with a basal den-
drite. Both apical and basal dendrites are
fully covered with spines. The apical den-
drites extend through the molecular
layer, and the basal dendrites extend into
the polymorphic layer. The axon origi-
nates from the base of the soma and
extends into the polymorphic layer. C.
Three types of human granule cell in the
human dentate gyrus. The neuron on the
right has dendrites in the molecular layer
only, whereas the other two neurons have
both apical dendrites extending into the
molecular layer and basal dendrites
extending into the polymorphic layer.
Bar in C  20 m and applies to all
panels. (Source: Adapted from Seress
and Mrzljak, 1987.)

Projections from the caudal entorhinal cortex (the region from the polymorphic layer of the dentate gyrus to the con-
equivalent to the rat medial entorhinal area) terminate in a tralateral molecular layer of the dentate gyrus and from the
similar fashion: heaviest in the middle third and gradually CA3 eld of the hippocampus to the contralateral CA3 and
decreasing in the outer third of the molecular layer. Thus, CA1 elds. In the monkey, these connections are virtually
the border between the lateral and medial entorhinal termina- absent. Only the most rostral part of the dentate gyrus and
tions is much less distinct in the monkey than in the rat. hippocampus (corresponding to the most temporal portion
in the rat) demonstrates any commissural connections, and
Lack of Commissural Connections in they are limited to the homotopic regions on the contralateral
the Monkey Hippocampal Formation side. Interestingly, whereas the commissural connections of
One of the most striking connectional differences between the the dentate gyrus and hippocampus are largely absent, the
rat and the monkey relates to the organization of the com- connection originating in the presubiculum and terminating
missural connections (Amaral et al., 1984; Demeter et al., in layer III of the contralateral entorhinal cortex appears to be
1985). In the rat, there are extensive commissural projections as robust in the monkey as in the rat.
102 The Hippocampus Book

Figure 351. Mossy cell of the polymorphic layer of the monkey dentate gyrus. A. Camera
lucida drawing of the soma and dendritic arbor. Note that the dendrites extend widely in the
polymorphic layer and extend into the molecular layer, in contrast to what is observed in the
rat. B. Camera lucida drawing of the soma and axonal plexus. Bar  100 m. (Source: Adapted
from Buckmaster and Amaral, 2001.)

Increased Cortical Interconnectivity of communication with the neocortex (Insausti et al., 1987). For
the Monkey Hippocampal Formation essentially all cortical regions, the return projections from the
The big difference between the rat and monkey brain is the entorhinal cortex originate in layers V and VI and terminate
much greater amount of neocortex in the primate. Much of both deeply and supercially in a manner typical of other cor-
this neocortex is dedicated to visual processing, but there is tical feedback projections. The laminar organization of the
also a substantial increase in the amount of association cortex, inputs to the entorhinal cortex varies according to the cortical
particularly in the frontal and temporal lobes. A substantial area of origin. Projections from some regions, such as the
portion of this association cortex is polysensory, and most of perirhinal and parahippocampal cortices, terminate preferen-
the polysensory cortical regions are interconnected with the tially in layers I to III, which give rise to the perforant path
hippocampal formation via connections with the entorhinal projections. Projections from other cortical areas, such as
cortex. The more robust neocortical connectivity has given rise the orbitofrontal and insular cortices, project to the deep lay-
to the suggestion that processing in the primate hippocampal ers of the entorhinal cortex, which generate the major output
formation is more highly dependent on and integrated with pathways to the neocortex and other regions. These latter con-
processing of sensory information in the neocortex. nections, therefore, might preferentially be involved in modu-
In the monkey temporal lobe, there is massive expansion of lating the output of the hippocampal formation. However,
the perirhinal and parahippocampal regions that border the they may also contribute to the perforant pathway via some of
hippocampal formation. Rostrally, this region includes the the few layer V and VI neurons that contribute to this pathway
perirhinal cortex, which comprises areas 35 and 36 of or through intrinsic deep to supercial connections in the
Brodmann. The perirhinal cortex starts at a level through the entorhinal cortex.
rostral hippocampal formation and extends rostrally to form Whereas the perirhinal and parahippocampal cortices pro-
the medial portion of the temporal polar cortex. The caudal vide the major input to the hippocampal formation, other
continuation of the perirhinal cortex is the parahippocampal substantial inputs originate in the retrosplenial cortex (area 29
cortex, which includes areas TF and TH of von Bonin and in the caudal cingulate gyrus), the polysensory region of the
Bailey, both of which are bordered laterally by the unimodal superior temporal gyrus (along the dorsal bank of the supe-
visual areas TE or TEO. The perirhinal and parahippocampal rior temporal gyrus), and the orbitofrontal cortex (mainly
cortices are important players in the economy of the neocor- caudal area 13). The general rule is that the entorhinal cortex
tical innervation of the hippocampal formation, as they pro- receives input only from cortical regions with demonstrated
vide nearly two-thirds of the neocortical inputs to the polysensory convergence. Thus, only olfactory input (which
entorhinal cortex. originates from all levels of the olfactory system including the
The major inputs to the monkey entorhinal cortex are olfactory bulb in the primate) has direct access to the hip-
summarized in Figures 352 and 353. In both rats and mon- pocampal formation. The olfactory connection, however, pro-
keys, the entorhinal cortex provides the major interface for vides another example of a difference between the rat and
Hippocampal Neuroanatomy 103

Figure 352. Summary of the major entorhinal cortical afferents in the macaque monkey.
(Source: Adapted from Insausti et al., 1987.)

monkey hippocampal formations. Whereas in the rat almost ing. From the unimodal association cortices, information
the entire entorhinal cortex receives direct input from the converges on a few polysensory cortical regions, which in turn
olfactory bulb, in the monkey this is restricted to about 10% project in a convergent fashion on the entorhinal cortex
of the surface area of the entorhinal cortex. (Lavenex and Amaral, 2000). The entorhinal cortex relays this
Given this pattern of neocortical inputs to the monkey information to the other hippocampal elds via the perforant
entorhinal cortex, the hippocampal formation can be viewed path projection. Once the information processed by the hip-
as the nal stage in a cascade of neocortical sensory process- pocampal formation is returned to the deep layers of the
104 The Hippocampus Book

Figure 353. Summary of the organization and strength of cortical inputs to the monkey
entorhinal, perirhinal (areas 35 and 36), and parahippocampal (areas TF and TH) cortices.
(Source: Adapted from Suzuki and Amaral, 1994.)

entorhinal cortex, they return this highly processed informa- human brain do need to be developed. Perhaps the com-
tion to many of the polysensory cortical regions that provide bination of transcranial electromagnetic stimulation during
feedback projections to the unimodal sensory regions. functional magnetic resonance imaging (fMRI) can provide
There are a number of implications of this cascade of pro- some information on pathways in the human brain, and one
jections into the entorhinal cortex and the organization of the can only hope that the wizardry of molecular biology will
entorhinal projections to the other hippocampal elds. One is someday provide new pathway-selective markers that allow
that the information the hippocampal formation is using has comparative studies of the rat, monkey, and human hip-
been highly preprocessed and integrated. Except for the olfac- pocampal formation. That being said, we next provide a short
tory input, little elemental or unimodal sensory information overview of the neuroanatomy of the human hippocampal
is directed to the hippocampal formation. A second, related formation.
implication is that it is unlikely that there is appreciable segre-
gation of sensory information processing in the hippocampal Cytoarchitectonic Organization
formation (i.e., visual information is not processed separately
from auditory information). In fact, the pattern of neu- The general appearance of the human hippocampal forma-
roanatomical connectivity predicts that the hippocampal for- tion is reminiscent of the monkey hippocampal formation
mation carries out whatever functions it accomplishes with (Figs. 32, 345, 354). Yet, differences are also apparent. For
high level, multimodal representations of sensory experiences. example, the CA1 eld of the hippocampus is even thicker in
humans than in monkeys (Fig. 349); in some regions, it is
3.6.3 Comparison of Monkey and as much as 30 cells thick. In addition to being thicker, the
Human Hippocampal Formation CA1 pyramidal cell layer takes on a distinctly multilaminate
appearance, with cells of different size and shape predominat-
At the risk of making points that may seem obvious, our ing at different depths of the layer. Unfortunately, there is lit-
understanding of the human hippocampus is necessarily tle information concerning the exact morphological or
primitive compared to that of the rat or the monkey. Because neurochemical characteristics of these neurons; there is not
standard experimental tract-tracing studies cannot be even a comprehensive Golgi analysis of the neurons of the
done, novel approaches for mapping connections in the human dentate gyrus and hippocampus.
Hippocampal Neuroanatomy 105

Figure 354. Nissl-stained coronal sections through the human the hippocampus; ML, molecular layer of the dentate gyrus; o,
hippocampal formation, arranged from rostral (A) to caudal (F). A, stratum oriens of the hippocampus; ot, optic tract; p, pyramidal
amygdaloid complex; a, alveus; ab, angular bundle; ac, anterior cell layer of the hippocampus; PaS, parasubiculum; PHG, parahip-
commissure; AHA, amygdalo-hippocampal area; CA1, CA1 eld of pocampal gyrus (areas TF and TH); pl, polymorphic layer of the
the hippocampus; CA2, CA2 eld of the hippocampus; CA3, CA3 dentate gyrus; PRC, perirhinal cortex (areas 35 and 36); PrS,
eld of the hippocampus; cas, calcarine sulcus; cf, choroidal ssure; presubiculum; r, stratum radiatum of the hippocampus; RSP,
cos, collateral sulcus; DG, dentate gyrus; EC, entorhinal cortex; f, retrosplenial cortex; S, subiculum; ssa, sulcus semiannularis; V,
mbria; GL, granule cell layer of the dentate gyrus; hc, hippocampal temporal horn of the lateral ventricle. The numbers in B indicate
commissure; hf, hippocampal ssure; irs, intrarhinal sulcus; LGN, the layers of the entorhinal cortex. Bar in A  2 mm and applies
lateral geniculate nucleus: l-m: stratum lacunosum-moleculare of to all panels.
106 The Hippocampus Book

Another obvious difference is the cellular composition of 1997a,b). They found that both the mossy ber projection and
the region enclosed within the limbs of the granule cell layer. the Schaffer collateral system (from CA3 to CA1) can be
There are obviously far more cells in this region, and most labeled over short distances by DiI and that these projections
appear to be cells of the CA3 eld. The polymorphic layer of resemble those observed in the monkey. As important as these
the dentate gyrus (the hilus) forms a narrow band that lies just limited observations are, they highlight the fact that we do not
subjacent to the granule cell layer. Little is known about the currently know whether the neuroanatomical connections
morphology of cells in the human dentate gyrus. As indicated that have been described for the rat and the monkey are appli-
in the comparison between the rat and the monkey, however, cable to the human brain.
there are clear differences in the morphology of certain cell
types. As many as 30% of granule cells in the human dentate Magnetic Resonance Imaging
gyrus have basal dendrites (Fig. 350); this number may be of the Human Hippocampal Formation
even larger than in the monkey.
Even less is known about the human subiculum, pre- With the emergence of MRI studies of the human brain, there
subiculum, and parasubiculum. Although the distribution of have been numerous structural (Jack et al., 2003) and func-
various neurochemicals or receptors has been incidentally tional (Maguire et al., 2003) studies of the human hippocam-
illustrated during the course of large survey studies, almost pal formation. Studies have used both manual tracing
nothing has been written on the cellular morphology of these techniques as well as algorithms for automated segmentation.
regions. The human entorhinal cortex has attracted consider- Of course, without the benet of cytoarchitectonic guidance,
ably more attention because of its vulnerability in Alzheimers it has generally proven difficult if not impossible to differenti-
disease. Although the same general regions identied in the ate the component parts of the hippocampal formation. It has
monkey can be identied in the human, there is evidence even been complicated to dene accurately the boundaries of
for additional elds in the human entorhinal cortex. Classic regions, such as the entorhinal cortex (Insausti et al., 1998).
cytoarchitectonicists, for example, have dened as many as 27 Figure 355 presents a coronal section stained by the Nissl
divisions. method. To get a sense of the dimensions of this region and
what can be distinguished during a standard functional imag-
Connections ing study, we have placed a grid of 4-mm square boxes over
the medial temporal lobe. With resolution at this level (which
There is little information concerning the organization of might be typical for a standard functional imaging study), one
connections in the human hippocampal formation owing to might expect to have one voxel focused over the subiculum or
the fact that most neuroanatomical tracing techniques require perhaps two voxels over the entorhinal cortex. Other voxels,
injection of tracers into the living brain. Some investigators however, would clearly overlap adjacent elds, such as the
have relied instead on techniques such as local application of dentate gyrus and CA3. Positron emission tomography stud-
the lipophilic dye DiI to map short pathways in human post- ies have even poorer spatial resolution. Thus, given current
mortem material. Clifford Saper and colleagues have used functional imaging technologies, it is difficult to dene spe-
this technique to study the connections of the granule cells cic activations for dened subelds of the hippocampal
and the projections of the enclosed portion of CA3 (Lim et al., formation.

Figure 355. Nissl-stained coronal sections through

the human hippocampal formation at the caudal
aspect of the uncus. The grid overlay consists of
squares that are 4 mm on edge. For abbreviations, see
Figure 311. (Source: Adapted from Amaral, 1999.)
Hippocampal Neuroanatomy 107

Critical Comparison of Neuroanatomy of Rat, Monkey, and formation is both massively divergent and convergent. Cells
Human Hippocampal Formation Must Await Comparable located in a focused point in the entorhinal cortex, for exam-
Quantitative Data for Each Species ple, can inuence neurons in as much as 50% of the entire
This chapter has dealt primarily with the rat hippocampal for- septotemporal axis of the dentate gyrus. Cells in the polymor-
mation because, as we pointed out at the beginning, much of phic layer of the dentate gyrus, in turn, can interconnect dif-
the neuroanatomical information available has been gained ferent levels of the dentate gyrus, providing further spread of
from studies of the rat. Furthermore, much of the electro- information to as much as 75% of the septotemporal axis.
physiological and behavioral literature discussed throughout Similarly, divergent projections have been described for the
the rest of the book is based on studies in the rat. With the CA3 to CA1 connections and for the CA1 to subiculum
dramatic increase in the use of noninvasive imaging of the connections. Thus, a perspective most consistent with the
human hippocampal formation, particularly in studies of known neuroanatomy is that the hippocampal formation
memory, the need has grown for more detailed information contains a series of three-dimensional networks of connec-
about the neuroanatomical organization of the human hip- tions. Depending on the synaptic interactions of these con-
pocampal formation. This is also important as increasingly nections (excitatory versus inhibitory), as well as the setting of
sophisticated models of hippocampus-related pathology are the myriad physiological parameters described in Chapter 5
introduced based primarily on rodent studies. A variety of (e.g., density and efficiency of synapses, transmitter release,
models for temporal lobe epilepsy, for example, have been synaptic plasticity), they might focus information processing
advanced based on cell degeneration and ber sprouting in to a specic level, recruit neurons at distant levels into the
the rat hippocampus. However, if the fundamental circuit dia- ongoing processing, or have no effect at all. Presumably, the
gram of the rat hippocampal formation is different from that role of this distributed network will become better under-
of humans, models based on the rat may be misleading. stood as increasingly massive parallel electrophysiological
Unfortunately, the extent of the similarities and differences recording procedures are employed.
in the hippocampal formation of the rat, monkey, and human
cannot yet be accurately gauged. Based on the few examples of 3.7.2 Functional Implications of the
established species differences described previously, it would Septotemporal Topography of Connections
not be surprising if substantial differences exist in the cellular
morphology, connectivity, and chemical neuroanatomy of the Given the septotemporal topography observed throughout
hippocampal formation across species. The relevance of these the hippocampal formation, there may still be some segre-
differences to the normal function and the pathology of the gation of functional divisions in this system. Although
hippocampal function will be an interesting area of compara- the massively divergent/convergent nature of hippocampal
tive neuroanatomy in the future. connections is inconsistent with a high degree of spatially
restricted information processing, the possibility remains that
there are a few septotemporally positioned domains of rela-
tively restricted information processing, at least at early stages,
3.7 Principles of Hippocampal Connectivity through the hippocampal circuitry. It appears that there are at
and Implications for Information Processing least three separable domains of neurons in the entorhinal
cortex: a lateral region, a mid region and a medial region. The
Neuroanatomy has the unfortunate characteristic of being most laterally situated domain of neurons innervates the
detail-oriented, and it is easy to lose sight of the forest for the septal half of the dentate gyrus and other innervated hip-
trees when absorbing and recounting information on the cells pocampal elds; the mid region innervates the next quarter;
and connections of the hippocampal formation. In the and the medial domain innervates the temporal quarter of the
following sections, we take a somewhat more integrative dentate gyrus (Fig. 341). Because the lateral portion of the
approach and attempt to draw implications from what we rat entorhinal cortex receives the bulk of the input from other
have described so far. These implications are generally quite neocortical areas, it is reasonable to assume that septal levels
speculative at this point, but they concern the important issue of the hippocampal formation are more highly involved with
of the functional correlates of the known hippocampal neu- processing exteroceptive sensory information. Because the
roanatomy. medial portions of the rat entorhinal cortex are preferentially
innervated by structures such as the amygdaloid complex, the
3.7.1 Highly Distributed Three-Dimensional temporal portion of the hippocampal formation may prefer-
Network of Intrinsic Connections entially deal with interoceptive or emotional information.
Because virtually all of the in vivo electrophysiological analy-
What is the hallmark of hippocampal connectivity? Beyond sis of the rat hippocampal formation has been carried out in
the dening aspect of unidirectional connectivity between the septal portion (and most of the in vitro work was carried
individual components, perhaps the facet that is most striking out to give slices from the rostal and mid-septotemporal
is that each of the intrinsic connections of the hippocampal levels), it is of interest to determine whether fundamentally
108 The Hippocampus Book

different types of response patterns are observed when tem- jections to these elds. The lateral entorhinal cortex projects
porally situated neurons are probed. Several lines of behav- to the border region of CA1 and the subiculum, whereas the
ioral data from lesioned animals are already consistent with medial entorhinal cortex projects either more proximally in
this idea (see Chapter 14). CA1 or more distally in the subiculum. Layer III neurons of
the medial entorhinal cortex are heavily inuenced by pre-
3.7.3 Functional Implications of the Transverse subicular input and thus by the thalamus, whereas layer III
Topography of Connections neurons in the lateral entorhinal area are heavily innervated
by the amygdaloid complex.
Despite the fact that intrinsic hippocampal connections have Taken together, this organization suggests that CA1 cells
an extensive septotemporal distribution, they are not uni- located close to the CA3 eld receive information from a sub-
formly distributed along the transverse axis. This organization set of CA3 cells located close to CA2 and a subset of layer III
raises the possibility of partially independent channels of cells located in the medial entorhinal cortex. CA1 cells located
information ow through and out of the hippocampal forma- closer to the subiculum, in contrast, receive preferential input
tion (Fig. 356). As we mentioned previously, it is not the case from CA3 cells located close to the dentate gyrus and from
that every CA1 cell has equal probability of input from every layer III cells in the lateral entorhinal cortex. A prediction
CA3 cell. The intrinsic circuitry of the hippocampal formation would be, therefore, that the response properties of CA1 cells
appears to be organized such that cells located at a particular at different transverse positions, at a particular septotemporal
transverse position in a eld are much more likely to be con- level, should be different.
nected with cells located at a particular transverse position of The notion of transverse topography of hippocampal con-
the innervated eld. This allows the possibility, therefore, that nections is made all the more compelling when it is appreci-
there is some channeling of information processing through ated that the subicular output is also organized in a columnar
the various hippocampal elds. As described fully above, cells fashion, with projections to different brain regions, or differ-
located proximally in CA3 tend to project to the most distal ent parts of the same brain region, originating from the prox-
CA1 cells. Projections from the distal portion of CA3, in con- imal, middle, and distal thirds of the subiculum. Neurons in
trast, terminate mainly in the proximal portion of CA1. The the proximal third of the subiculum project to the infralimbic
midportion of CA3 lls in the spaces between these two pro- and prelimbic cortices, the nucleus accumbens, and the lateral
jections. The CA1 projection to the subiculum demonstrates septal region. Projections from this portion of the ventral part
even more striking transverse topography. The proximal por- of the subiculum also project to the ventromedial nucleus of
tion of CA1 projects discretely to the distal third of the subicu- the hypothalamus and to the amygdala. The midtransverse
lum; the distal portion of CA1 projects just across the border portion of the subiculum projects mainly to the midline thal-
into the proximal third of the subiculum; and the middle por- amic nuclei, and neurons in the distal portion of the subicu-
tion of CA1 projects to the midregion of the subiculum. lum project to the retrosplenial portion of the cingulate cortex
The tripartite transverse organization in CA1 and the and the presubiculum. Although all portions of the subiculum
subiculum also appears to be respected by the entorhinal pro- project to the entorhinal cortex, the pattern of projections

Figure 356. Summary of the transverse

organization of the connections through
the hippocampal formation. This gure
highlights the possibility that informa-
tion is segregated, or channeled,
through the hippocampal formation and
ultimately reaches different recipients of
hippocampal output.
Hippocampal Neuroanatomy 109

reciprocates the topography of the perforant path projections

to the subiculum. Thus, the proximal portion of the subicu-
lum projects to the lateral entorhinal area, and more distal
portions of the subiculum project to the medial entorhinal
area. Thus, from CA3 on through the entorhinal cortex, there
exists the potential for information that arrives at the earliest
point in this chain of connections to maintain some uniquess
from information that arrives at different portions of CA3.
The functional signicance of this neuroanatomical pathway
segregation is unknown at present.
To summarize, although the intrinsic connectivity of the
hippocampal formation is highly divergent, the network that
is formed is not without some topography. The septotempo-
ral and transverse components of this topography provide the
increased probability of communication between certain neu-
rons in each eld and thus the possibility of some form of
channeling or segregation of information processing in the
hippocampal formation.

3.7.4 Serial and Parallel Processing Figure 357. Summary of the serial and parallel pathways through
in the Hippocampal Formation the hippocampal formation. Although the intrinsic hippocampal
circuitry is largely unidirectional, it contains both serial and parallel
A unique feature of the intrinsic hippocampal circuitry is the projections. See text for details.
largely unidirectional organization of the projections that
interconnect the various hippocampal regions. A popular
notion is that these unidirectional projections also imply an
exclusively serial or sequential ow of informationrst
from the entorhinal cortex to the dentate gyrus, then to the 3.8 Conclusions
CA3 eld of the hippocampus, and so on. However, the data
summarized in the body of this chapter clearly indicate that Although this chapter has presented a substantial amount of
the intrinsic hippocampal circuitry has both serial and paral- neuroanatomical information, it has only begun to scratch
lel projections (Fig. 357). the surface of what is available in the primary literature. How-
The entorhinal cortex, in particular, contributes many ever, beause we have focused on general features and princi-
parallel projections to several elds of the hippocampal for- ples of organization, it should be more than adequate for
mation. The same layer II entorhinal cells probably give rise delving into the molecular and cellular aspects of hippocam-
to projections that terminate in both the dentate gyrus and pal organization (see Chapters 5 and 7) as well as the function
the CA3 eld of the hippocampus. Thus, whatever the of the hippocampal formation in the living organism. The
information conveyed by layer II entorhinal cells, it arrives neuroanatomy of the hippocampal formation is unique and
both monosynaptically and disynaptically (through mossy predicts that the behavioral functions it subserves are
ber intermediaries) at CA3. It is not known, of course, undoubtedly unique as well.
whether information from a single entorhinal cell reaches
a particular CA3 cell both monosynaptically and disynapti-
cally. This is an important question to resolve in future ACKNOWLEDGMENTS
The existence of prominent associational connections in The authors are indebted to the many colleagues who have worked in
the dentate gyrus, hippocampus, and entorhinal cortex also our laboratory over the years. Much of the knowledge they have
provides the substrate for polysynaptic activation in hip- uncovered has gone into the pages of this chapter. We are grateful
pocampal circuits. The functional implication of this more particularly to Dr. Menno Witter and Dr. Ricardo Insausti, who not
complex circuitry is that each hippocampal region is not only co-authored earlier comprehensive summaries on the neu-
roanatomy of the rat and human hippocampal formation, but also
entirely dependent on the preceding region for input, which
reviewed the current chapter to help ensure its accuracy. We also
raises the prospect that each region may be acting semi-inde-
thank Dr. Lazlo Seress for additional comments on the manuscript.
pendently from, as well as in concert with, other hippocampal The original research summarized in this chapter has been supported
elds. Hippocampal neuroanatomy is thus consistent with the by the National Institutes of Health and the Human Frontier Science
electrophysiological ndingsfor example, that CA1 place Program. Finally, we are grateful to our families who understand that
elds are apparently normal even after pharmacological inac- science requires some sacrice of time with them. Their support has
tivation of the dentate gyrus. made this chapter possible.
110 The Hippocampus Book

REFERENCES Brodal A (1947) The hippocampus and the sense of smell: a review.
Brain 70:179222.
Acsady L, Kamondi A, Sik A, Freund T, Buzsaki G (1998) GABAergic Brodmann K (1909) Vergleichende Lokalisationslehre der Grosshirn-
cells are the major postsynaptic targets of mossy bers in the rat rinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues.
hippocampus. J Neurosci 18:33863403. Leipzig.
Allen GV, Hopkins DA (1989) Mamillary body in the rat: topography Buckmaster PS, Amaral DG (2001) Intracellular recording and label-
and synaptology of projections from the subicular complex, ing of mossy cells and proximal CA3 pyramidal cells in macaque
prefrontal cortex, and midbrain tegmentum. J Comp Neurol monkeys. J Comp Neurol 430:264281.
286:311336. Buckmaster PS, Strowbridge BW, Kunkel DD, Schmiege DL,
Amaral DG (1978) A Golgi study of cell types in the hilar region of Schwartzkroin PA (1992) Mossy cell axonal projections to the
the hippocampus in the rat. J Comp Neurol 182:851914. dentate gyrus molecular layer in the rat hippocampal slice.
Amaral DG (1999) Introduction: what is where in the medial tempo- Hippocampus 2:349362.
ral lobe? Hippocampus 9:16. Buckmaster PS, Wenzel HJ, Kunkel DD, Schwartzkroin PA (1996)
Amaral DG, Dent JA (1981) Development of the mossy bers of Axon arbors and synaptic connections of hippocampal mossy
the dentate gyrus. I. A light and electron microscopic study of cells in the rat in vivo. J Comp Neurol 366:271292.
the mossy fibers and their expansions. J Comp Neurol Burwell RD (2001) Borders and cytoarchitecture of the perirhinal
195:5186. and postrhinal cortices in the rat. J Comp Neurol 437:1741.
Amaral DG, Kurz J (1985) An analysis of the origins of the choliner- Burwell RD, Amaral DG (1998a) Perirhinal and postrhinal cortices of
gic and noncholinergic septal projections to the hippocampal the rat: interconnectivity and connections with the entorhinal
formation of the rat. J Comp Neurol 240:3759. cortex. J Comp Neurol 391:293321.
Amaral DG, Witter MP (1989) The three dimensional organization Burwell RD, Amaral DG (1998b) Cortical afferents of the perirhinal,
of the hippocampal formation: a review of anatomical data. postrhinal, and entorhinal cortices of the rat. J Comp Neurol
Neuroscience 31:571591. 398:179205.
Amaral DG, Insausti R, Cowan WM (1984) The commissural con- Burwell RD, Witter MP (2002) Basic anatomy of the parahippocampal
nections of the monkey hippocampal formation. J Comp Neurol region in monkeys and rats. Oxford: Oxford University Press.
224:307336. Burwell RD, Witter MP, Amaral DG (1995) Perirhinal and postrhinal
Amaral DG, Dolorfo C, Alvarez-Royo P (1991) Organization of CA1 cortices of the rat: a review of the neuroanatomical literature
projections to the subiculum: a PHA-L analysis in the rat. Hip- and comparison with findings from the monkey brain.
pocampus 1:415436. Hippocampus 5:390408.
Andersen P, Bliss TVP, Skrede KK (1971) Lamellar organization of Caballero-Bleda M, Witter MP (1993) Regional and laminar organi-
hippocampal excitatory pathways. Exp Brain Res 13:222238. zation of projections from the presubiculum and parasubicu-
Arantius G (1587) De humano foetu . . . Ejusdem anatomicorum obser- lum to the entorhinal cortex: an anterograde tracing study in the
vationum liber. Venice. rat. J Comp Neurol 328:115129.
Baisden RH, Woodruff ML, Hoover DB (1984) Cholinergic and non- Canteras N, Swanson L (1992) Projections of the ventral subicu-
cholinergic septo-hippocampal projections: a double-label lum to the amygdala, septum, and hypothalamus: a PHAL
horseradish peroxidase-acetylcholinesterase study in the rabbit. anterograde tract-tracing study in the rat. J Comp Neurol
Brain Res 290:146151. 324:180194.
Bakst I, Avendao C, Morrison JH, Amaral DG (1986) An experi- Claiborne BJ, Amaral DG, Cowan WM (1986) A light and electron
mental analysis of the origins of the somatostatin immunoreac- microscopic analysis of the mossy bers of the rat dentate gyrus.
tive bers in the dentate gyrus of the rat. J Neurosci 6:14521462. J Comp Neurol 246:435458.
Blackstad TW (1956) Commissural connections of the hippocampal Claiborne BJ, Amaral DG, Cowan WM (1990) A quantitative three-
region in the rat, with special reference to their mode of termi- dimensional analysis of granule cell dendrites in the rat dentate
nation. J Comp Neurol 105:417537. gyrus. J Comp Neurol 302:206219.
Blackstad TW, Brink K, Hem J, Jeune B (1970) Distribution of hip- Conrad LCA, Leonard CM, Pfaff DW (1974) Connections of the
pocampal mossy bers in the rat: an experimental study with median and dorsal raphe nuclei in the rat: an autoradiographic
silver impregnation methods. J Comp Neurol 138:433450. and degeneration study. J Comp Neurol 156:179206.
Borhegyi Z, Leranth C (1997) Distinct substance P- and calretinin- Daitz HM, Powell TPS (1954) Studies of the connexions of the fornix
containing projections from the supramammillary area to the system. J Neurol Neurosurg Psychiatry 17:7582.
hippocampus in rats: a species difference between rats and De Garengeot RJC (1742) Splanchnologie ou lanatomie des visceres,
monkeys. Exp Brain Res 115:369374. 2nd ed, pp 250251. Paris: C. Osmond.
Boyett JM, Buckmaster PS (2001) Somatostatin-immunoreactive Deller T (1998) The anatomical organization of the rat fascia dentata:
interneurons contribute to lateral inhibitory circuits in the den- new aspects of laminar organization as revealed by anterograde
tate gyrus of control and epileptic rats. Hippocampus tracing with Phaseolus vulgaris-Leucoagglutinin (PHAL). Anat
11:418422. Embryol (Berl) 197:89103.
Braak H (1972) Zur Pigmentarchitektonic der Grosshirnrinde des Deller T, Adelmann G, Nitsch R, Frotscher M (1996) The alvear path-
Menschen. I. Regio entorhinalis. Z Zellforsch Mikrosk Anat way of the rat hippocampus. Cell Tissue Res 286:293303.
127:407438. Demeter S, Rosene DL, Van Hoesen GW (1985) Interhemispheric
Braak H (1980) Architectonics of the human telencephalic cortex. New pathways of the hippocampal formation, presubiculum and
York: Springer. entorhinal and posterior parahippocampal cortices in the rhe-
Broca P (1878) Anatomie compare des circonvolutions crbrales: le sus monkey: the structure and organization of the hippocampal
grand lobe limbique. Rev. Anthropol 1:385498. commissures. J Comp Neurol 233:3047.
Hippocampal Neuroanatomy 111

Dent JA, Galvin NJ, Staneld BB, Cowan WM (1983) The mode of pocampus and dentate gyrus of the rat. J Comp Neurol
temination of the hypothalamic projection to the dentate gyrus: 149:393420.
an EM autoradiographic study. Brain Res 258:110. Greene JRT, Totterdell S (1997) Morphology and distribution of elec-
Desmond NL, Levy WB (1985) Granule cell dendritic spine density trophysiologically dened classes of pyramidal and nonpyrami-
in the rat hippocampus varies with spine shape and location. dal neurons in rat ventral subiculum in vitro. J Comp Neurol
Neurosci Lett 54:219224. 380:395408.
Dolleman-Van der Weel MJ, Witter MP (2000) Nucleus reuniens Groenewegen HJ, Vermeulen-Van Der Zee E, Te Kortschot A, Witter
thalami innervates gamma aminobutyric acid positive cells in M (1987) Organization of the projections from the subiculum
hippocampal eld CA1 of the rat. Neurosci Lett 278:145148. to the ventral striatum in the rat: a study using anterograde
Dolorfo CL, Amaral DG (1998a) Entorhinal cortex of the rat: topo- transport of Phaseolus vulgaris leucoagglutinin. Neuroscience
graphic organization of the cells of origin of the perforant path 23:103120.
projection to the dentate gyrus. J Comp Neurol 398:2548. Haglund L, Swanson LW, Khler C (1984) The projection of the
Dolorfo CL, Amaral DG (1998b) Entorhinal cortex of the rat: organ- supramammillary nucleus to the hippocampal formation: an
ization of intrinsic connerctions. J Comp Neurol 398:4982. immunohistochemical and anterograde transport study with
Donovan MK, Wyss JM (1983) Evidence for some collateralization the lectin PHA-L in the rat. J Comp Neurol 229:171185.
between cortical and diencephalic efferent axons of the rat Hamam BN, Amaral DG, Alonso AA (2002) Morphological and elec-
subicular cortex. Brain Res 259:181192. trophysiological characteristics of layer V neurons of the rat lat-
Duffy CJ, Rakic P (1983) Differentiation of granule cell dendrites in eral entorhinal cortex. J Comp Neurol 451:4561.
the dentate gyrus of the rhesus monkey: a quantitative Golgi Hamam BN, Kennedy TE, Alonso A, Amaral DG (2000)
study. J Comp Neurol 214:224237. Morphological and electrophysiological characteristics of layer
Felleman DJ, Van Essen DC (1991) Distributed hierarchical process- V neurons of the rat medial entorhinal cortex. J Comp Neurol
ing in the primate cerebral cortex. Cereb Cortex 1:147. 418:457472.
Freund TF, Buzsaki G (1996) Interneurons of the hippocampus. Han ZS, Buhl EH, Lorinczi Z, Somogyi P (1993) A high degree of
Hippocampus 6:347470. spatial selectivity in the axonal and dendritic domains of physi-
Fricke R, Cowan WM (1978) An autoradiographic study of the com- ologically identied local-circuit neurons in the dentate gyrus of
missural and ipsilateral hippocampo-dentate projections in the the rat hippocampus. Eur J Neurosci 5:395410.
adult rat. J Comp Neurol 181:253269. Harris E, Witter MP, Weinstein G, Stewart M (2001) Intrinsic con-
Frotscher M (1992) Application of the Golgi/electron microscopy nectivity of the rat subiculum. I. Dendritic morphology and
technique for cell identication in immunocytochemical, retro- patterns of axonal arborization by pyramidal neurons. J Comp
grade labeling, and developmental studies of hippocampal neu- Neurol 435:490505.
rons. Microsc Res Tech 23:306323. Haug FM (1976) Sulphide silver pattern and cytoarchitectonics of
Funahashi M, Stewart M (1997a) Presubicular and parasubicular cor- parahippocampal areas in the rat. Special reference to the subdi-
tical neurons of the rat: electrophysiological and morphologi- vision of area entorhinalis (area 28) and its demarcation from
cal properties. Hippocampus 7:117129. the pyriform cortex. Adv Anat Embryol Cell Biol 52:373.
Funahashi M, Stewart M (1997b) Presubicular and parasubicular Hausheer-Zarmakupi Z, Wolfer DP, Leisinger-Trigona MC, Lipp HP
cortical neurons of the rat: functional separation of deep and (1996) Selective breeding for extremes in open-eld activity of
supercial neurons in vitro. J Physiol (Lond) 501:387403. mice entails a differentiation of hippocampal mossy bers.
Gaarskjaer FB (1978a) Organization of the mossy ber system of the Behav Genet 26:167176.
rat studied in extended hippocampi. I. Terminal area related to Henze DA, Urban NN, Barrioneuvo G (2000) The multifarious hip-
number of granule and pyramidal cells. J Comp Neurol pocampal mossy eer pathway: a review. Neuroscience 98:
178:4972. 407427.
Gaarskjaer FB (1978b) Organization of the mossy ber system of the Herkenham M (1978) The connections of the nucleus reuniens thal-
rat studied in extended hippocampi. II. Experimental analysis of ami: evidence for a direct thalamo-hippocampal pathway in the
ber distribution with silver impregnation methods. J Comp rat. J Comp Neurol 177:589610.
Neurol 178:7388. Hjorth-Simonsen A, Jeune B (1972) Origin and termination of the
Gaykema RPA, Luiten PGM, Nyakas C, Traber J (1990) Cortical pro- hippocampal perforant path in the rat studied by silver impreg-
jection patterns of the medial septum-diagonal band complex. nation. J Comp Neurol 144:215232.
J Comp Neurol 293:103124. Hjorth-Simonsen A, Zimmer J (1975) Crossed pathways from the
Gertz SD, Lindenberg R, Piavis GW (1972) Structural variations entorhinal area to the fascia dentata. J Comp Neurol 161:
in the rostral human hippocampus. Johns Hopkins Med J 130: 5770.
367376. Insausti R, Amaral DG, Cowan WM (1987) The entorhinal cortex of
Gloveli T, Schmitz D, Empson RM, Dugladze T, Heinemann U (1997) the monkey. II. Cortical afferents. J Comp Neurol 264:356395.
Morphological and electrophysiological characterization of Insausti R, Herrero MT, Witter MP (1997) Entorhinal cortex of the
layer III cells of the medial entorhinal cortex of the rat. rat: cytoarchitectonic subdivisions and the origin and distribu-
Neuroscience 77:629648. tion of cortical efferents. Hippocampus 7:146183.
Goldowitz D, White WF, Steward O, Lynch G, Cotman C (1975) Insausti R, Juottonen K, Soininen H, Insausti AM, Partanen K, Vainio
Anatomical evidence for a projection from the entorhinal cortex P, Laakso MP, Pitkanen A (1998) MR volumetric analysis of the
to the contralateral dentate gyrus of the rat. Exp Neurol human entorhinal, perirhinal, and temporopolar cortices. AJNR
47:433441. Am J Neuroradiol 19:659671.
Gottlieb DI, Cowan WM (1973) Autoradiographic studies of the Ishizuka N (2001) Laminar organization of the pyramidal cell layer
commissural and ipsilateral association connections of the hip- of the subiculum in the rat. J Comp Neurol 435:89110.
112 The Hippocampus Book

Ishizuka N, Weber J, Amaral DG (1990) Organization of intrahip- hippocampal formation in humans. I. The mossy ber pathway.
pocampal projections originating from CA3 pyramidal cells in J Comp Neurol 385:325351.
the rat. J Comp Neurol 295:580623. Lim C, Mufson EJ, Kordower JH, Blume HW, Madsen JR, Saper
Ishizuka N, Cowan WM, Amaral DG (1995) A quantitative analysis CB (1997b) Connections of the hippocampal formation in
of the dendritic organization of pyramidal cells in the rat hip- humans. II. The endfolial ber pathway. J Comp Neurol 385:
pocampus. J Comp Neurol 362:1745. 352371.
Jack CR Jr, Slomkowski M, Gracon S, Hoover TM, Felmlee JP, Stewart Lipp HP, Schwegler H, Heimrich B, Cerbone A, Sadile AG (1987)
K, Xu Y, Shiung M, OBrien PC, Cha R, Knopman D, Petersen Strain-specic correlations between hippocampal structural
RC (2003) MRI as a biomarker of disease progression in a ther- traits and habituation in a spatial novelty situation. Behav Brain
apeutic trial of milameline for AD. Neurology 60:253260. Res 24:111123.
Kaitz SS, Robertson RT (1981) Thalamic connections with limbic cor- Lorente de N R (1933) Studies on the structure of the cerebral cor-
tex. II. Corticothalamic projections. J Comp Neurol 195:527545. tex. I. The area entorhinalis. J Psychol Neurol 45:381438.
Kempermann G, Kuhn G, Gage FH (1997) More hippocampal neu- Lorente de N R (1934) Studies on the structure of the cerebral
rons in adult mice living in an enriched environment. Nature cortex. II. Continuation of the study of the ammonic system.
386:493495. J Psychol Neurol 46:113177.
Kempermann G, Kuhn HG, Gage FH (1998) Experienced-induced Loughlin SE, Foote SL, Grzanna R (1986) Efferent projections of
neurogenesis in the senescent dentate gyrus. J Neurosci nucleus locus coeruleus: morphologic subpopulations have dif-
18:32063212. ferent efferent targets. Neuroscience 18:307319.
Kiss J, Csaki A, Bokor H, Shanabrough M, Leranth C (2000) The Lubke J, Deller T, Frotscher M (1997) Septal innervation of mossy
supramamillo-hippocampal and supramammillo-septal gluta- cells in the hilus of the rat dentate gyrus: an anterograde tracing
matergic/aspartatergic projections in the rat: a combined and intracellular labeling study. Exp Brain Res 114:423432.
[3H]D-aspartate autoradiographic and immunohistochemical Magloczky Z, Acsady L, Freund TF (1994) Principal cells are the post-
study. Neuroscience 97:657669. synaptic targets of supramammillary afferents in the hippocam-
Klink R, Alonso A (1997) Morphological characteristics of layer II pus of the rat. Hippocampus 4:322334.
projection neurons in the rat medial entorhinal cortex. Maguire EA, Spiers HJ, Good CD, Hartley T, Frackowiak RS, Burgess
Hippocampus 7:571583. N (2003) Navigation expertise and the human hippocampus: a
Kloosterman F, Witter Menno P, van Haeften T (2003) Topo- structural brain imaging analysis. Hippocampus 13:250259.
graphical and laminar organization of subicular projections McEwen BS, Wallach G (1973) Corticosterone binding to hippocam-
to the parahippocampal region of the rat. J Comp Neurol 455: pus: nuclear and cytosol binding in vitro. Brain Res 57:373386.
156171. McLardy T, Kilmer WL (1970) Hippocampal circuitry. Am Psychol
Kohler C (1985) Intrinsic projections of the retrohippocampal region 25:563566.
in the rat brain. I. The subicular complex. J Comp Neurol Meibach RC, Siegel A (1975) The origin of fornix bers which proj-
236:504522. ect to the mammillary bodies in the rat: a horseradish peroxi-
Khler C, Steinbusch H (1982) Identication of serotonin and non- dase study. Brain Res 88:508512.
serotonin-containing neurons of the mid-brain raphe project- Moore RY, Halaris AE (1975) Hippocampal innervation by serotonin
ing to the entorhinal area and the hippocampal formation: a neurons of the midbrain raphe in the rat. J Comp Neurol
combined immunohistochemical and uorescent retrograde 164:171184.
tracing study in the rat brain. Neuroscience 7:951975. Morrison JH, Benoit, R., Magistretti PJ, Ling N, Bloom FE (1982)
Laatsch RH, Cowan WM (1967) Electron microscopic studies of the Immunohistochemical distribution of prosomatostatin-related
dentate gyrus of the rat. II. Degeneration of commissural affer- peptides in hippocampus. Neurosci Lett 34:137142.
ents. J Comp Neurol 130:241262. Mosko S, Lynch G, Cotman CW (1973) The distribution of septal
Lacaille JC, Schwartzkroin PA (1988) Stratum lacunosum-molecu- projections to the hippocampus of the rat. J Comp Neurol
lare interneurons of hippocampal CA1 region. I. Intracellular 152:163174.
response characteristics, synaptic responses, and morphology. Naber PA, Witter MP (1998) Subicular efferents are organized mostly
J Neurosci 8:14001410. as parallel projections: a double-labeling, retrograde-tracing
Lacaille JC, Mueller AL, Kunkel DD, Schwartzkroin PA (1987) Local study in the rat. J Comp Neurol 393:284297.
circuit interactions between oriens/alveus interneurons and Naber PA, da Silva FHL, Witter MP (2001) Reciprocal connections
CA1 pyramidal cells in hippocampal slices: electrophysiology between the entorhinal cortex and hippocampal elds CA1 and
and morphology. J Neurosci 7:19791993. the subiculum are in register with the projections from CA1 to
Laurberg S (1979) Commissural and intrinsic connections of the rat the subiculum. Hippocampus 11:99104.
hippocampus. J Comp Neurol 184:685708. Nafstad PHJ (1967) An electron microscope study on the termina-
Laurberg S, Sorensen KE (1981) Associational and commissural col- tion of the perforant path bres in the hippocampus and the
laterals of neurons in the hippocampal formation (hilus fasciae fascia dentata. Z Zellforsch Und Mikrosk Anat 76:532542.
dentatae and subeld CA3). Brain Res 212:287300. Nyakas C, Luiten PGM, Spencer DG, Traber J (1987) Detailed pro-
Lavenex P, Amaral DG (2000) Hippocampal-neocortical interaction: jection patterns of septal and diagonal band efferents to the hip-
a hierarchy of associativity. Hippocampus 10:420430. pocampus in the rat with emphasis on innervation of CA1 and
Li X-G, Somogyi P, Ylinen A, Buzsaki G (1994) The hippocampal dentate gyrus. Brain Res Bull 18:533545.
CA3 network: an in vivo intracellular labeling study. J Comp OMara SM, Commins S, Anderson M, Gigg J (2001) The subiculum:
Neurol 339:181208. a review of form, physiology and function. Prog Neurobiol
Lim C, Blume HW, Madsen JR, Saper CB (1997a) Connections of the 64:129155.
Hippocampal Neuroanatomy 113

Papez JW (1937) A proposed mechanism of emotion. Arch Neurol Soriano E, Frotscher M (1989) A GABAergic axo-axonic cell in the
Psychiatry 38:725743. fascia dentata controls the main excitatory hippocampal path-
Pickel VM, Segal M, Bloom FE (1974) A radioautographic study of way. Brain Res 503:170174.
the efferent pathways of the nucleus locus coeruleus. J Comp Sorra KE, Harris KM (1993) Occurrence and three-dimensional
Neurol 155:1542. structure of multiple synapses between individual radiatum
Pikkarainen M, Ronkko S, Savander V, Insausti R, Pitkanen A (1999) axons and their target pyramidal cells in hippocampal area CA1.
Projections from the lateral, basal, and accessory basal nuclei of J Neurosci 13:37363748.
the amygdala to the hippocampal formation in rat. J Comp Steward O, Scoville SA (1976) Cells of origin of entorhinal cortical
Neurol 403:229260. afferents to the hippocampus and fascia dentata of the rat. J
Pitkanen A, Pikkarainen M, Nurminen N, Ylinen A (2000) Reciprocal Comp Neurol 169:347370.
connections between the amygdala and the hippocampal for- Struble RG, Desmond NL, Levy WB (1978) Anatomical evidence
mation, perirhinal cortex and postrhinal cortex in the rat. Ann for interlamellar inhibition in the fascia dentata. Brain Res 152:-
NY Acad Sci 911:369391. 580585.
Powell TPS, Guillery RW, Cowan WM (1957) A quantitative study of Suzuki WA, Amaral DG (1994) Perirhinal and parahippocampal cor-
the fornix-mamillo-thalamic system. J Anat 91:419437. tices of the macaque monkey: cortical afferents. J Comp Neurol
Pyapali GK, Sik A, Penttonen M, Buzsaki G, Turner DA (1998) 350:497533.
Dendritic properties of hippocampal CA1 pyramidal neurons in Swanson LW (1978a) The anatomical organization of septo-hip-
the rat: intracellular staining in vivo and in vitro. J Comp Neurol pocampal projections. In: Functions of the septo-hippocampal
391:335352. system, pp 2548. Amsterdam: Elsevier North Holland.
Raisman G, Cowan WM, Powell TPS (1965) The extrinsic afferent, Swanson LW, Cowan, WM (1975) Hippocampo-hypothalamic con-
commissural and association bres of the hippocampus. Brain nections: origin in subicular cortex, not Ammons horn. Science
88:963997. 189:303304.
Ramn y Cajal S (1893) Estructura del asta de Ammon y fascia den- Swanson LW, Hartman BK (1975) The central adrenergic system:
tata. Ann Soc Esp Hist Nat 22. an immunouorescence study of the location of cell bodies
Rapp PR, Gallagher M (1996) Preserved neuron number in the hip- and their efferent connections in the rat utilizing dopamine-
pocampus of aged rats with spatial learning decits. Proc Natl -hydroxylase as a marker. J Comp Neurol 163:467506.
Acad Sci USA 93:99269930. Swanson LW, Khler C (1986) Anatomical evidence for direct projec-
Ribak CE, Seress L (1983) Five types of basket cell in the hippocam- tions from the entorhinal area to the entire cortical mantle in
pal dentate gyrus: a combined Golgi and electron microscopic the rat. J Neurosci 6:30103023.
study. J Neurocytol 12:577597. Swanson LW, Wyss JM, Cowan WM (1978) An autoradiographic
Ribak CE, Vaughn JE, Saito K (1978) Immunocytochemical localiza- study of the organization of intrahippocampal association path-
tion of glutamic acid decarboxylase in neuronal somata follow- ways in the rat. J Comp Neurol 181:681716.
ing colchicine inhibition of axonal transport. Brain Res Swanson LW, Sawchenko PE, Cowan WM (1981) Evidence for collat-
140:315332. eral projections by neurons in ammons horn, the dentate gyrus,
Ribak CE, Seress L, Amaral DG (1985) The development, ultrastruc- and the subiculum: a multiple retrograde labeling study in the
ture and synaptic connections of the mossy cells of the dentate rat. J Neurosci 1:548559.
gyrus. J Neurocytol 14:835857. Swanson LW, Khler C, Bjorklund A (1987) The limbic region. I. The
Robertson RT, Kaitz SS (1981) Thalamic connections with limbic septohippocampal system. In: Handbook of chemical neu-
cortex. I. Thalamocortical projections. J Comp Neurol roanatomy (Bjorklund A, Hkfelt T, Swanson LW, eds), pp
195:501525. 125227. Amsterdam: Elsevier.
Sarter M, Markowitsch HJ (1985) Convergence of intra- and inter- Tamamaki N, Nojyo Y (1993) Projection of the entorhinal layer II
hemispheric cortical afferents: lack of collateralization and evi- neurons in the rat as revealed by intracellular pressure-injection
dence for subrhinal cell group projecting heterotopically. J Comp of neurobiotin. Hippocampus 3:471480.
Neurol 236:283296. Tamamaki N, Abe K, Nojyo Y (1987) Columnar organization in the
Seress L (1992) Morphological variability and developmental subiculum formed by axon branches originating from single
aspects of monkey and human granule cells: differences CA1 pyramidal neurons in the rat hippocampus. Brain Res
between the rodent and primate dentate gyrus. Epilepsy Res 412:156160.
Suppl 7:328. Thompson SM, Robertson RT (1987) Organization of subcortical
Seress L, Mrzljak L (1987) Basal dendrites of granule cells are normal pathways for sensory projections to the limbic cortex. I.
features of the fetal and adult dentate gyrus of both monkey and Subcortical projections to the medial limbic cortex in the rat.
human hippocampal formations. Brain Res 405:169174. J Comp Neurol 265:175188.
Seress L, Pokorny J (1981) Structure of the granular layer of the rat Trommald M, Hulleberg G, Andersen P (1996) Long-term potentia-
dentate gyrus: a light microscopic and Golgi study. J Anat tion is associated with new excitatory spine synapses on rat den-
133:181195. tata granule cells. Learn Mem 3:218228.
Shipley MT (1975) The topographical and laminar organization of Van der Werf YD, Witter MP, Groenewegen HJ (2002) The intralam-
the presubiculums projection to the ipsi- and contralateral inar and midline nuclei of the thalamus: anatomical and func-
entorhinal cortex in the guinea pig. J Comp Neurol 160:127146. tional evidence for participation in processes of arousal and
Sik A, Penttonen M, Buzsaki G (1997) Interneurons in the hip- awareness. Brain Res Brain Res Rev 39:107140.
pocampal dentate gyrus: an in vivo intracellular study. Eur J Van Groen T, Wyss JM (1990) The connections of presubiculum and
Neurosci 9:573588. parasubiculum in the rat. Brain Res 518:227243.
114 The Hippocampus Book

Van Groen T, Kadish I, Wyss JM (2002) Species differences in the pro- Witter MP (1993) Organization of the entorhinal-hippocampal
jections from the entorhinal cortex to the hippocampus. Brain system: a review of current anatomical data. Hippocampus 3:
Res Bull 57:553-556. 3344.
Van Hoesen GW, Hyman BT, Damasio AR (1991) Entorhinal cortex Witter MP, Amaral DG (2004) The hippocampal region. In: The rat
pathology in Alzheimers disease. Hippocampus 1:18. brain (Paxinos G, ed, 3rd ed), pp 637670. San Diego, CA,
Van Groen T, Miettinen P, Kadish I (2003) The entorhinal cortex of London: Elsevier Academic Press.
the mouse: organization of the projection to the hippocampal Witter MP, Groenewegen HJ (1990) The subiculum: cytoarchitec-
formation. Hippocampus 13:133149. tonically a simple structure, but hodologically complex. In:
Vertes RP (1993) PHA-L analysis of projections from the supramam- Understanding the brain through the hippocampus (Storm-
millary nucleus in the rat. J Comp Neurol 326:595622. Mathisen J, Zimmer J, Ottersen OP, eds). Amsterdam:
Vertes RP, Fortin WJ, Crane AM (1999) Projections of the median Elsevier.
raphe nucleus in the rat. J Comp Neurol 407:555582. Witter MP, Griffioen AW, Jorritsma-Byham B, Krijnen JLM (1988)
Verwer RWH, Meijer RJ, Vanuum HFM, Witter MP (1997) Collateral Entorhinal projections to the hippocampal CA1 region in the
projections from the rat hippocampal formation to the lateral rat: an underestimated pathway. Neurosci Lett 85:193198.
and medial prefrontal cortex. Hippocampus 7:397402. Witter MP, Ostendorf RH, Groenewegen HJ (1990) Heterogeneity in
Von Economo C (1929) The cytoarchitectonics of the human cerebral the dorsal subiculum of the rat: distinct neuronal zones project
cortex. New York: Oxford University Press. to different cortical and subcortical targets. Eur J Neurosci
Wainer BH, Levey AI, Rye DB, Mesulam MM, Mufson EJ (1985) 2:718725.
Cholinergic and non-cholinergic septohippocampal pathways. Witter MP, Wouterlood FG, Naber PA, van Haeften T (2000)
Neurosci Lett 54:4552. Structural organization of the parahippocampal-hippocampal
Walker MC, Ruiz A, Kullmann DM (2002) Do mossy bers release network. Ann NY Acad Sci 911:125.
GABA? Epilepsia 43:196202. Wouterlood FG, Saldana E, Witter MP (1990) Projection from
West MJ, Slomianka L, Gundersen HJG (1991) Unbiased stereologi- the nucleus reuniens thalami to the hippocampal region: light
cal estimation of the total number of neurons in the subdivi- and electron microscopic tracing study in the rat with the
sions of the rat hippocampus using the optical fractionator. anterograde tracer Phaseolus vulgaris-leucoagglutinin. J Comp
Anat Rec 231:482497. Neurol 296:179203.
White TD, Tan AM, Finch DM (1990) Functional reciprocal connec- Wyss JM, Van Groen T (1992) Connections between the retrosplenial
tions of the rat entorhinal cortex and subicular complex with cortex and the hippocampal formation in the rat: a review.
the medial frontal cortex: an in vivo intracellular study. Brain Hippocampus 2:112.
Res 533:95106. Wyss JM, Swanson LW, Cowan WM (1979) Evidence for an input to
Wilson CO, Isokawa-Akesson M, Babb TL, Engel J Jr, Cahan L D, the molecular layer and the stratum granulosum of the dentate
Crandall PH (1987) Comparative view of local and interhemi- gyrus from the supramammillary region of the hypothalamus.
spheric limbic pathways in humans: an evoked potential analy- Anat Embryol (Berl) 156:165176.
sis. In: Fundamental mechanisms of human brain function (Engel Wyss JM, Swanson LW, Cowan WM (1980) The organization of the
J Jr, Ojemann GA, Lders HO, Williamson PD, eds), pp 2338. mbria, dorsal fornix and ventral hippocampal commissure in
New York: Raven Press. the rat. Anat Embryol (Berl) 158:303316.
4 Michael Frotscher and Lszl Seress

Morphological Development of the Hippocampus

neurons are formed, what molecules and mechanisms deter-

4.1 Overview mine the fate of newly generated cells? For example, postmi-
totic neurons migrate to their nal destination. How does this
A fundamental question in neurobiology is how the various happen? Upon arrival in their appropriate structure or layer,
brain structures and the enormous number of specic the postmigratory neuroblasts start to develop their processes.
interneuronal connections that serve the complex functions of Why does only one axon originate from the cell body, whereas
the mature CNS are formed during development. Work on the many dendrites emerge from it? Moreover, as these processes
hippocampal formation has shed light on the principles emerge, they do so in an organized manner. For example, in
underlying this process. With the advent of modern genetic the various hippocampal elds, principal neurons display a
approaches, including mutant mice lacking or overexpressing rather uniform structure. What are the molecules and mecha-
genes important for developmental processes, new insights nisms controlling this uniform differentiation? Why do the
into the development of the hippocampal formation are in the nonprincipal cells that invade the hippocampus from the
process of being obtained. ganglionic eminence (Pleasure et al., 2000) form such a het-
Not all aspects of hippocampal development are covered in erogeneous population of neurons? Finally, what are the
this chapter. Traditionally, developmental processes are stud- determinants of synapse formation? The hippocampus is a
ied in a descriptive manner by monitoring the formation of laminated structure in which all bers originating from a par-
mature structures from simpler, undifferentiated stages. ticular afferent source terminate at identical dendritic seg-
Major contributions have also come from studies of the phy- ments. How is this layer-specic termination of hippocampal
logenetic development of the hippocampal formation and afferents to be explained? To what extent is neuronal activity
have demonstrated that it is a form of phylogenetically old involved in the differentiation of dendrites, dendritic spines,
cortex, the archicortex, that develops in the medial wall of the and synaptic contacts?
telencephalic vesicle. Owing to the expansion of the cerebral An attempt is rst made to summarize our present knowl-
cortex in higher vertebrates, the hippocampal formation is edge on neuronal generation and migration in the hippocam-
translocated medially and ventrally into the inferior portion pus. We then address the development of major hippocampal
of the lateral ventricle. In rodents, the primary focus of this connections by describing determinants of pathnding, target
chapter, much of the hippocampus is located dorsally, the layer recognition, and synapse formation of hippocampal
curved structure of the hippocampus reecting this rotation afferents.
of the telencephalon (see Chapter 3). Phylogenetic develop-
ment is not dealt with in detail here, although we do present
some information on species other than rodents where appro-
priate. For a fuller treatment of phylogenetic differences, the 4.2 Neurogenesis and Cell Migration
reader is referred to the exhaustive coverage by Stephan
(Stephan, 1975). Most of our knowledge about cell formation in the rodent
What are the fundamental issues regarding the develop- hippocampal formation dates back to the seminal studies by
ment of the hippocampus? As in other organs, it is important Angevine (1965) and Altman and coworkers (Altman, 1966;
to understand the signals governing cell proliferation. Once Altman and Das, 1966; Altman and Bayer, 1975; Bayer, 1980).

116 The Hippocampus Book

These important papers provided the rst descriptions of the gin might explain their characteristic morphological differ-
origin, the time course of the generation, and the laminar dis- ences. On E 14.5 and E 15.5, the newly formed pyramidal cells
tribution of principal neurons, pyramidal neurons, and gran- already show eld-specic genetic markers characteristic for
ule cells. the CA1 or CA3 area of the mouse hippocampus (Tole et al.,
4.2.1 Pyramidal Neurons At birth, the pyramidal layer in the rat is thick and com-
posed of 6 to 10 rows of neuronal somata. As the hippocam-
Stem cells of both pyramidal neurons and granule cells origi- pus gets larger postnatally, the pyramidal layer becomes
nate from the ventricular germinal layers (or ventricular neu- thinner; and in adult rats it comprises two or three cell layers.
roepithelial layers) that are located below the ventricular wall This process of pyramidal cell rearrangement is unlikely to be
along the CA1 area. In contrast to the neocortical ventricular a passive one due, for example, to the volumetric increase of
germinal layers, there is no subventricular zone at the hip- the hippocampal formation. Postnatally generated glial cells
pocampal germinal area. Therefore the multiplying neurons might contribute to the reorganization. In fact, neonatal X-ray
directly migrate from the ventricular zone to their nal target irradiation, which has its greatest impact on glial cells, pre-
region (Altman and Bayer, 1990b). Pyramidal neurons of the vents formation of the mature pyramidal cell layer (Czurk et
rodent hippocampus are generated between embryonic days al., 1997). However, early-generated reelin-synthesizing Cajal-
(E) 10 and E 18 in the mouse (Angevine, 1965) and between Retzius cells also appear to play a role in the formation of the
E 16 and E 21 in the rat brain (Bayer, 1980). Pregnancy lasts pyramidal cell layer (see below).
19 days in mice and 21 days in rats. CA3 pyramidal cells are It is worth noting that the pyramidal cell layer of the hip-
generated with a peak of generation on E 17, whereas the peak pocampus also forms quite early in the human brain. The
of cell generation of CA1 pyramidal cells is on E 18 and E 19 pyramidal layer is formed during the rst half of pregnancy
(Bayer, 1980). (Arnold and Trojanowsky, 1996). In addition, the CA1 to C3
Except for the CA3 pyramidal cells, the route of migration subregions of the hippocampus can be differentiated as early
is short because the hippocampus closely follows the curve of as during the 16th embryonic week. Regional differentiation
the ventricle (Fig. 41). Future CA1 neurons form cell rows of the hippocampal formation thus substantially precedes that
oriented perpendicular to the ventricular area that end in of the neocortex (Kostovic et al., 1989).
the incipient CA1 pyramidal layer. From the ventricular ger-
minal layer, the early-generated CA3 pyramidal cells migrate 4.2.2 Granule Cells
3 to 4 days longer (than the CA1 cells) along the ventricular
wall in the only prenatally existing intermediate zone (Altman The formation of the granule cell layer of the dentate gyrus
and Bayer, 1990b). As a consequence, the later-generated differs in many respects from that of the pyramidal cell layer
CA1 pyramidal cells establish a distinct layer earlier than of the hippocampus. The rst granule cells of the mouse den-
the CA3 cells. It remains an open question whether the tate gyrus appear at approximately the same time as the rst
pyramidal neurons in CA1 and CA3 originate from the same pyramidal cells (i.e., on E 10) (Angevine, 1965). There is a
segment of the hippocampal neuroepithelium. A separate ori- suprapyramidal to infrapyramidal, or medial to lateral, gradi-
ent in the formation of the two blades of the granule cell layer;
that is, granule cells at the tip of the suprapyramidal blade are
Figure 41. Route of migration (arrows) of pyramidal cells from the rst to be generated. The generation of granule cells lasts
the ventricular germinative layer to the subiculum and to areas a much longer time than that of pyramidal neurons. There is
CA1 to CA3. DG, dentate gyrus; H, hilus. substantial evidence that it continues long into the postnatal
period and, at a reduced level, into adulthood. In the rat den-
tate gyrus, rst granule cells are generated 1 day later than the
rst pyramidal neurons (i.e., on E 17) (Bayer, 1980). In rats,
15% of the granule cells are generated before birth (Altman
and Bayer, 1975). Therefore, the time span of granule cell
generation is approximately three times longer than that of
pyramidal neurons and is likely to continue for the rest of
the animals life (for adult neurogenesis, see Chapter 9).
Interestingly, adult granule cell formation does not follow the
suprapyramidal to infrapyramidal sequence observed during
the embryonic and early postnatal period. Granule cells are
generated with no obvious pattern in both blades (Kemper-
mann et al., 1997a,b).
The prolonged postnatal formation of granule cells is not a
unique feature of rodents. In humans, granule cell formation
lasts more than 30 weeks, beginning at approximately the 13th
Morphological Development of the Hippocampus 117

period of granule cell neurogenesis, and persisting hilar stem

cells may also form new neurons during adulthood (Altman
and Bayer, 1975, 1990a; Seress, 1978). In rodents, the total
number of hilar cells rapidly decreases between postnatal days
10 and 25, with a parallel increase in the number of granule
cells in the granule cell layer, suggesting that newly formed
cells from the hilus move to the granule cell layer (Seress,
1977). During the entire period of postnatal development,
labeled granule cells are always observed inside the granule cell
layer as early as 1 hour following the injection of bromod-
eoxyuridine (BrdU) or thymidine (Fig. 43, see color insert). It
is likely that stem cells from the hippocampal ventricular zone
move to the hilar region and form a secondary germinal zone
(Fig. 42). This possibility is supported by the fact that a sin-
gle neonatal X-ray irradiation prevents further cell formation
in both the ventricular zone and hilus without affecting the
Figure 42. Route of migration of dentate granule cells. There are already formed neuronal subpopulations (Czurko et al., 1997).
two possible ways for granule cells to reach the suprapyramidal To the extent that it has been studied, the developmental
or infrapyramidal blade of the dentate gyrus. 1, Granule cells characteristics of the rodent hilar region are also observed in
formed in the germinative layer move along the pyramidal layer
humans. In human fetuses, a large number of mitotic cells
of Ammons horn directly to the granule cell layer (arrows); 2, after
(indicated by the presence of Ki-67) are present in the hilar
migration from the ventricular germinative layer, stem cells form
a secondary germinative layer (black spot) in the hilus (H) and, region from the 13th week onward (Fig. 44A, see color
following further proliferation, move to the granule cell layer insert). These proliferating cells account for a large percentage
(arrows) of the dentate gyrus (DG). During the embryonic period of the total hilar cell population in 16- to 18-week-old fetuses
both routes are used, whereas the latter route prevails during the (Seress, unpublished observations). Proliferating cells can also
postnatal period. be found in the hilus during the late gestational period and
after birth (Seress et al., 2001) (Fig. 44B, see color insert).
Persisting stem cells have also been found in the hilus of the
embryonic week (Humphrey, 1967; Seress et al., 2001). There adult human hippocampal formation (Eriksson et al., 1998).
is also evidence for adult neurogenesis in the human dentate However, proliferative activity drops off considerably during
gyrus (Eriksson et al., 1998). the postnatal period (Seress et al., 2001).
The route of migration of the postmitotic granule cells is There is little morphological variability among granule
much longer than that of pyramidal neurons. Postmitotic cells cells even though they may be located at different places in the
must migrate from the ventricular germinal layer along the granule cell layer and may be born at different time points.
already formed hippocampus through the narrow intermedi- This morphological uniformity leads one to assume that all
ate zone between the alveus and future stratum oriens toward granule cells originate from the same stem cell population.
the hippocampus, where they form the cup-shaped dentate However, granule cells may originate from different germinal
gyrus that surrounds the tip of CA3 (Fig. 42). sources (Altman and Bayer, 1990a), and precursors of the
The two blades of dentate granule cells and the tip of the perinatally and postnatally forming granule cells can be dis-
CA3 area border the hilus of the dentate gyrus (see Chapter 3). tinguished (Altman and Bayer, 1990c). A newly discovered
During postnatal ontogenesis, the hilus (Fig. 43A, see color gene (Lef1) has been found to distinguish granule cell precur-
insert) contains large numbers of mitotic cells (Bayer, 1980). sors. This gene is expressed in the proliferating precursors of
Large hilar neurons, such as the mossy cells, are generated granule cells in the ventricular zone as well as in those of the
between E 15 and E 21 (Bayer, 1980). In addition, virtually all secondary germinal zone. Lef1-decient mice fail to form den-
hilar interneurons, representing approximately half of the tate granule cells (Galceran et al., 2000).
hilar cells (Seress and Ribak, 1983), are of subcortical origin Most of the granule cells display similar features in the
and invade the hilus from the lateral and medial ganglionic rodent and primate dentate gyrus. However, a subpopulation
eminence (Pleasure et al., 2000). Thus, the enduring cell pro- of granule cells in the primate dentate gyrus is characterized
liferation in the hilar region (Fig. 42) mainly represents local by the formation of basal dendrites (Seress and Mrzljak,
generation of granule cells (Altman and Bayer, 1990c). In fact, 1987). It is likely that local environmental factors, such as
the rate of postnatal cell formation in the hilus is similar to available space for dendritic expansion, may play an impor-
that in the periventricular germinal layer of the telencephalon tant role in the formation of basal dendrites on granule cells
(Seress, 1978). It is therefore reasonable to assume that the (Seress and Frotscher, 1990). Basal dendrites have also been
hilar region is a displaced secondary germinal layer of the regarded as a pathological phenomenon. For example,
dentate gyrus (Altman and Bayer, 1975; Seress, 1977). increased activity of the granule cells may evoke the formation
Proliferating cells persist in the hilar region during the entire of basal dendrites in epileptic rats (Spigelman et al., 1998).
118 The Hippocampus Book

Figure 43. A. Photomicrograph

showing 3H-thymidine-labeled cells
(black) in the hilus (H) and granule
cell layer (GCL) of the dentate gyrus
of a 5-day-old rat that received a single
injection of thymidine 1 hour before
sacrice. At this age, the infrapyrami-
dal blade is still very short and hardly
reaches the tip of the CA3 pyramidal
layer. The large number of labeled
cells above the granule cell layer sug-
gests that glial cells are also formed at
this age because the molecular layer
contains only a few interneurons,
which are born prenatally. Note the
lack of labeling in the CA3 area of
Ammons horn where pyramidal cells
are generated prenatally. n I set . Labeled
cells in the granule cell layer of this
animal at higher magnication.
Section is 10 m thick and stained
with toluidine blue. B. BrdU-labeled
cells (brown) in the suprapyramidal
blade of the dentate gyrus of a 3-
month-old rat that received daily
injections of BrdU from P4 to P6.
Bars: A, B, 50 m; inset, 10 m.

Epileptic seizures also appear to lead to an increased number uniform granule cells and pyramidal neurons, local circuit
of granule cells with basal dendrites in the human dentate neurons of the hippocampal formation form a heterogeneous
gyrus (von Campe et al., 1997). population (Freund and Buzski, 1996) (see Chapters 3 and
In summary, dentate granule cells in different mammalian 8). Even the classic basket cell can be subdivided into ve types
species form a largely homogeneous neuronal population. based on differences in location and dendritic arborization
Their dendritic morphology, however, may be modied to (Ribak and Seress, 1983). Neurochemical studies have con-
some extent depending on the local environmental or patho- rmed the existence of at least two basket cell types, which dis-
logical factors. play different receptors and co-transmitters (Hjos et al., 1998;
Katona et al., 1999). In addition to dendritic morphology, dif-
4.2.3 Local Circuit Neurons and Hilar Neurons ferences in axonal projections also distinguish subpopulations
of local circuit neurons in both the dentate gyrus (Halasy and
Local circuit neurons of the hippocampal formation differ Somogyi, 1993a,b) and the hippocampus proper (Soriano and
from the principal cells in both their morphological and Frotscher, 1989, 1993; Halasy et al., 1996; Spruston et al., 1997;
developmental features. In contrast to the morphologically Vida et al., 1998; Vida and Frotscher, 2000).
Morphological Development of the Hippocampus 119

Figure 44. A. MIB-1 (Ki-67)-labeled

cells (brown) in the dentate gyrus
of a 16-week-old human fetus. A large
number of labeled cells are seen in
the hilus (H) underneath the granule
cell layer (GCL). Note the absence
of labeled cells in the CA3 area of
Ammons horn. In contrast, numerous
labeled cells, migrating from the ven-
tricular germinative layer along the
pyramidal layer toward the hilus, are
seen in the intermediate zone. B. MIB-
1 (Ki-67)-labeled cells are still frequent
in the hilus (H) of a 28-week-old fetus.
The granule cell layer and molecular
layer (ML) contain only a few labeled
cells. Sections are 10 m thick and are
stained with cresyl violet. Bars: A, 50
m; B, 20 m.

In contrast to the heterogeneity of the local circuit neuron cedes formation of the rst pyramidal cells by 1 day (Bayer,
phenotype, their neurogenetic prole seems much more 1980). Cell formation of the cells destined to reside in the den-
homogeneous; that is, there are no developmental events that dritic layers ceases relatively early. It is reasonable to suggest
would allow one to predict the large diversity among - that this exquisite timing may play some role in pyramidal and
aminobutyric acid (GABA)ergic cells. GABAergic interneu- granule cell morphogenesis, although there is no evidence for
rons are generated early in both the dentate gyrus and the this at this time.
hippocampus (Bayer, 1980). This conclusion was based on the Double-labeling studies have conrmed that GABAergic
observation that neurons in the molecular layer and hilus of neurons are generated earlier than the principal cells of the
the dentate gyrus as well as in the strata oriens, radiatum, and hippocampal formation (Amaral and Kurz, 1985; Lbbers
lacunosum-moleculare appear earlier than the neurons in the et al., 1985; Soriano et al., 1989a,b; Dupuy and Houser,
principal cell layers (Bayer, 1980). The peak cell formation in 1997, 1999). The site of origin of hippocampal local circuit
both the hilus and the molecular layer precedes the onset of neurons is in the ventricular/subventricular zone of the
granule cell generation. Similarly, in the CA1 to C3 areas, the medial and lateral ganglionic eminence. Available data do not
peak cell generation in the strata oriens and radiatum pre- allow one to correlate the site of generation with the diverse
120 The Hippocampus Book

morphological features of the local circuit neurons (Pleasure Although this scenario is certainly an oversimplication, it
et al., 2000). can be employed to explain the unusual lamination in the
Although GABAergic neurons are generated and posi- reeler hippocampus and dentate gyrus. In the hippocampus, a
tioned before their target cells, some of them change their double pyramidal cell layer is formed in reeler mice. As in the
position later on in development (Dupuy and Houser, 1999). neocortex, the marginal zone is located directly underneath
Moreover, the differentiation of axonal collaterals and the the pial surface, which in the hippocampus proper corre-
arborization of dendrites takes place relatively late in postna- sponds to the stratum lacunosum-moleculare. The stratum
tal life (Seress et al., 1989; Lang and Frotscher, 1990; Seress and lacunosum-moleculare contains Cajal-Retzius cells, and its
Ribak, 1990). This raises the possibility that the maturation of extracellular matrix is enriched in reelin. The absence of the
local circuit neurons is inuenced by interactions with their stop signal reelin in reeler mutants may allow some pyramidal
synaptic partners. cells to migrate deep into the stratum radiatum, forming there
a second pyramidal layer (Staneld and Cowan, 1979; Deller
4.2.4 Determinants of Neuronal et al., 1999b).
Migration in the Hippocampus The marginal zone of the dentate gyrus corresponds to the
outer portion of the molecular layer. Assuming that reelin acts
In the neocortex, cell layers are formed in an inside-out man- as a stop signal, one would expect that many granule cells
ner; that is, early-generated neurons form deep layers, whereas would invade the molecular layer in reeler mice because they
supercial layers are established by cells born late in ontoge- would not be stopped earlier. However, in the reeler dentate
netic development. As in the neocortex, the hippocampus fol- gyrus, the granule cells do not invade the molecular layer,
lows a similar inside-out migration. The dentate gyrus, in whereas many granule cells are found in the hilar region with
contrast, has an outside-in migration pattern; that is, early- their dendrites oriented in all directions (Staneld and
generated granule cells populate supercial portions of the Cowan, 1979; Drakew et al., 2002). What might be the expla-
granule cell layer and later-formed granule cells are located nation of the hilar location of the granule cells? Frster et al.
progressively deeper. (2002) found that the radial glial bers required for neuronal
What are the signals controlling neuronal migration that migration are dramatically altered in the dentate gyrus of
lead to such well dened cell layers of principal neurons? It reeler mice. Thus, it has been proposed that the granule cell
has been known for some time that these migrational migration defect seen in the reeler mutant is due to a malfor-
processes are under fairly strict genetic control. In 1951, a mation of the radial glial scaffold required for proper neu-
mouse mutant, the reeler mouse, was described (Falconer, ronal migration (Frster et al., 2002; Frotscher et al., 2003;
1951) in which neocortical lamination was reversed. Recently, Zhao et al., 2004).
the gene missing in reeler mice was discovered (dArcangelo et
al., 1995; Hirotsune et al., 1995). The gene product, reelin, was
found to be synthesized by a well known group of neurons in
the marginal zone, originally described by Cajal and Retzius 4.3 Development of Hippocampal
more than 100 years ago (Retzius, 1893, 1894; Ramn y Cajal, Connections
1911). These Cajal-Retzius cells are remarkable neurons. They
extend long dendrites horizontally in the marginal zone, par- When describing the development of neuronal connections, it
allel to the pial surface. Cajal-Retzius cells synthesize and is useful to distinguish at least three processes. First, there is
secrete reelin, so it forms a component of the extracellular the way an axon navigates to the target region, called axonal
matrix in the marginal zone (dArcangelo et al., 1997). pathnding. During this process, both soluble and membrane-
Available data are compatible with the proposal that reelin bound molecules and attractant and repellent signals are
acts as a stop signal for migrating neurons. Early-generated involved (see Tessier-Lavigne and Goodman, 1996, for review).
neurons are stopped as they approach the marginal zone, Next, the axon must recognize the appropriate target region
which has a high concentration of reelin. For later neurons to and target cell, called target recognition. As far as the hip-
reach this same stop signal, they must migrate past the early- pocampus is concerned, the segregated ber systems must rec-
generated neurons and approach the reelin-containing mar- ognize their appropriate layers. Finally, there is the process of
ginal zone. In this way the characteristic inside-out lamination synapse formation.
of the cortex is formed (Frotscher, 1997, 1998). Because the There is substantial and increasing evidence that a variety
marginal zone ultimately becomes layer I of the cortex, it of molecules are involved in these three processes. In princi-
remains an almost cell-free layer in normally developing mice. ple, membrane-bound molecules that attract or repel the
In reeler mice, in contrast, layer I (which does not have the growing tip of an axon, diffusible factors building up a gradi-
stop signal) is densely lled with neurons. Early-generated ent, and components of the extracellular matrix are likely to
neurons migrate until they reach the pial surface, and later- play a role. The involvement of various ligand/receptor fami-
generated neurons accumulate underneath. The normal lies in the formation of hippocampal connections has been
inside-out lamination is reversed (Fig. 45). summarized by Skutella and Nitsch (2001). Thus, the various
Morphological Development of the Hippocampus 121

Figure 45. Hypothetical function of reelin in cortical lamination. tic inside-out pattern of cortical lamination is formed. Apical
A. In wild-type mice, the preplate splits into the marginal zone, dendrites, branching in the reelin-containing marginal zone, are
largely containing Cajal-Retzius cells (black cells), and the subplate extended by the increase in cortical thickness. D. In reeler mice
(white cells). Cajal-Retzius cells synthesize and secrete reelin (stip- lacking reelin, the separation of the subplate from the marginal
pled zone). B. The cortical plate develops between the marginal zone does not take place. E. In the absence of reelin, migrating cells
zone and the subplate. An early-generated cell (A) migrates from (A) are not stopped and reach the pial surface. F. Later-generated
the ventricular zone toward the marginal zone, where its migration neurons (B, C) accumulate underneath the early-generated neurons
is stopped by reelin. C. With increasing cortical thickness, the reelin- (A). As a result, the normal inside-out pattern of cortical lamination
containing marginal zone is moved further outward, allowing later- is reversed, and apical dendrites are not anchored in the marginal
generated neurons (B, C) to migrate farther than their predecessors zone. (Source: Frotscher, 1997.)
before their migration is stopped by reelin. This way the characteris-

semaphorins and their receptors neuropilin 1 and 2 and the gyrus. From a developmental point of view, one would expect
plexins netrin 1 and its receptor DCC (deleted in colon can- that the perforant path would utilize robust guidance signals
cer), the ephrin (Eph) family of tyrosine kinases and their lig- for axonal pathnding and target layer recognition. The
ands, the ephrins, and Slit and Robo were found to contribute entorhinal bers terminating in the outer two-thirds of the
to establishment of the hippocampal circuitry. dentate molecular layer and in stratum lacunosum-moleculare
In the following paragraphs, we deal with the develop- of the hippocampus proper form a sharp boundary toward the
ment of some of the main hippocampal afferent connections: adjacent commissural/associational bers in the inner molec-
those from the entorhinal cortex, the septum, and the con- ular layer of the dentate gyrus and in the stratum radiatum of
tralateral hippocampus. Whereas the bers from both the the hippocampus (see Chapter 3).
entorhinal cortex and the contralateral hippocampus are Recent studies have shed some light on the mechanisms
examples of a strictly laminated termination of hippocampal underlying the pathnding of entorhinal axons (del Rio et al.,
afferents, bers from the septum have a more diffuse dis- 1997; Chedotal et al., 1998; Frster et al., 1998; Frotscher,
tribution. 1998; Ceranik et al., 1999; Deller et al., 1999a,b; Savaskan et
al., 1999; Skutella et al., 1999; Steup et al., 1999). Investigators
4.3.1 Entorhinal Connections have focused on repulsive molecules expressed in the hip-
pocampus that would prevent the entorhinal afferents from
Neuroanatomists have been intrigued by the unique course of terminating in proximal layers close to the cell bodies of prin-
the entorhino-hippocampal projection, which perforates the cipal neurons. Similarly, molecules specically expressed in
subiculum and hippocampal ssure en route to the dentate the entorhinal cortex were studied for their capacity to repel
122 The Hippocampus Book

entorhinal axons, thereby directing them toward the hip-

pocampus. In particular, repulsive effects of semaphorins
(Sema3A) on entorhinal bers have been observed (Steup et
al., 1999), an effect that can be blocked by antibodies against
the semaphorin receptor neuropilin 1 (Chedotal et al., 1998).
As Sema3A is expressed by granule cells, it has been speculated
that this expression may be involved in the termination of
entorhinal bers on distal granule cell dendrites. Neuropilin 2,
the receptor for Sema3F, is strongly expressed in the hip-
pocampus, whereas the ligand is expressed in the developing
entorhinal cortex. As Sema3F is repulsive for growing axons
bearing its receptor, this particular pattern may be responsible
for the lack of ingrowth of hippocampal and dentate axons
into the entorhinal cortex (Chedotal et al., 1998; Steup et al.,
1999; Skutella and Nitsch, 2001). Similarly, Slit 1 and Slit 2,
being expressed in the entorhinal cortex, can repel axons of
hippocampal neurons expressing their receptors Robo 1 and
Robo 2 (Nguyen-Ba-Charvet et al., 1999). Also ephrins were
found to be involved in repulsive effects on entorhinal axons,
restricting their termination to distal dendritic portions of
hippocampal neurons (Stein et al., 1999). In addition to these
repulsive effects, neurotrophic factors with an attractive neu-
rotropic effect, components of the extracellular matrix, and a
guiding scaffold provided by pioneer neurons have been
found to navigate the axons of entorhinal neurons to their Figure 46. AC. Early-generated hippocampal Cajal-Retzius (CR)

hippocampal target structures. cells form a template for outgrowing axons of projection neurons in
the entorhinal cortex (EC). D. Later on in development when many
What could be the sequence of events underlying this com-
CR cells have disappeared, the axons of projection cells in the
plex process? Probably as a rst step, a guiding scaffold sup- entorhinal cortex establish denitive connections with distal den-
porting the directed growth of entorhinal axons toward the drites of dentate granule cells (GC).
hippocampus is formed. In fact, Ceranik et al. (1999) have
shown that early-generated Cajal-Retzius cells, located in the
marginal zones of the dentate gyrus (outer molecular layer) fact, Cajal-Retzius cells and their projections to the entorhinal
and hippocampus proper (stratum lacunosum-moleculare), cortex are also present in these mutants and may thus serve
give rise to a heavy projection to the entorhinal cortex that is their normal function as a template for entorhinal axons. In
established before the entorhino-hippocampal projection. normal mice, reelin affects the branching pattern of entorhi-
Thus, Cajal-Retzius cells in the outer molecular layer of the rat nal terminals. In the absence of reelin, entorhinal axons give
dentate gyrus are retrogradely labeled from the entorhinal rise to fewer collaterals and synapses (del Rio et al., 1997;
cortex as early as on E 17, whereas the rst entorhino- Borrell et al., 1999). The growth of entorhinal bers along
hippocampal bers were found to arrive in the dentate outer Cajal-Retzius cell axons is likely to be controlled by a variety
molecular layer only thereafter. Moreover, with combined of repulsive and attractive molecules (see above).
injections of two tracers, DiI and DiO, into the entorhinal cor- How do the entorhinal bers recognize their target layer?
tex and hippocampus, respectively, entorhinal axons were Current evidence strongly suggests that components of the
found to grow along axons of Cajal-Retzius cells. These nd- extracellular matrix play an important part in the segregation
ings suggest that the early-formed projection of hippocampal of hippocampal afferents. Frster et al. (1998), for example,
Cajal-Retzius cells to the entorhinal cortex provides a tem- demonstrated that uorescent beads coated with entorhinal
plate for outgrowing entorhinal axons Fig. 46A(C). A role of membranes precisely adhered to the correct entorhinal ter-
Cajal-Retzius cells in the pathnding of entorhinal axons is mination zone of the hippocampus. They then used this pat-
strongly supported by experiments in slice cultures, in which tern as an assay to study the role of extracellular matrix
Cajal-Retzius cells had been caused to degenerate by excito- components in this layer-specic adhesion. When slice cultures
toxic lesions (del Rio et al., 1997). In the absence of Cajal- of hippocampus were treated with hyaluronidase, the mem-
Retzius cells, entorhinal axons were unable to nd their way to brane-coated beads no longer adhered with layer specicity,
the hippocampus, whereas commissural bers invaded their indicating that hyaluronic acid and proteoglycans bound to it
termination elds as normal. may contribute to the layer-specic adhesion. Such adhesion
Because the entorhino-hippocampal projection develops seems also to be an important factor for target layer recogni-
almost normally in reeler mice, it does not appear that reelin tion of entorhinal bers. When hippocampal slices were
synthesized by Cajal-Retzius cells is essential to this process. In treated with hyaluronidase and then cocultured with entorhi-
Morphological Development of the Hippocampus 123

nal cortex, entorhinal bers invaded the molecular layer but ber of long, immature spines, indicating that neuronal activ-
were no longer restricted to its outer two-thirds (Frster et al., ity may be required for the maturation of spines.
2001). It appears that hyaluronic acid and proteoglycans are Many questions concerning the layer-specic termination
essential for the formation of boundaries between afferent ter- of entorhinal bers remain open at present. As an example,
minal zones in the hippocampus. Interestingly enough, no the development of the ber lamination in the infrapyramidal
similar role of extracellular matrix components was found for blade of the dentate gyrus is different from that in the
the laminar specicity of commissural/associational bers suprapyramidal blade. Whereas the entorhinal bers arrive
(Zhao et al., 2003). In conclusion, secreted molecules such as before the commissural axons in the suprapyramidal blade,
the semaphorins, membrane-bound receptors, extracellular this sequence is reversed in the infrapyramidal blade
matrix components, and a template formed by Cajal-Retzius (Tamamaki, 1999). Do the bers to these different blades orig-
cell axons are likely to be involved in the directed growth and inate from the same cells of origin in the entorhinal cortex? If
layer-specic termination of entorhinal bers. so, what causes the delay in the ingrowth into the infrapyra-
What is the role of the postsynaptic target cell in the layer- midal blade?
specic termination of entorhinal afferents? In the rodent
brain, entorhinal axons innervate the hippocampus proper at 4.3.2 Commissural Connections
E 15 and the dentate gyrus at E 18/E 19 (Super and Soriano,
1994; Ceranik et al., 1999). The entorhinal bers arrive much Commissural bers, originating from CA3 pyramidal neurons
earlier than the commissural bers from the contralateral hip- and hilar mossy cells, project via the hippocampal commis-
pocampus (see below). At this early stage, pyramidal neurons sure to the contralateral hippocampus and dentate gyrus. In
and granule cells have not yet grown their distal dendritic tips the mouse hippocampus, the rst commissural axons arrive in
to the stratum lacunosum-moleculare and the outer molecu- the contralateral hippocampus at E 18 and in the dentate
lar layer of the dentate gyrus. Most of the granule cells have gyrus at P 2. This is considerably later than the arrival of the
not even been generated yet. Interestingly, the entorhinal entorhinal axons. The sequential generation of the entorhinal
bers recognize their appropriate target layers from the very neurons and commissurally projecting hippocampal cells and
beginning despite the absence of their appropriate target den- their sequential ingrowth into their target elds has led to a
drites at the stage of ber ingrowth. Evidently, the target cells hypothesis concerning their laminated termination in both
are unlikely to be involved in pathnding of entorhinal axons the hippocampus proper and the dentate gyrus: The later the
and target layer recognition. The early-generated Cajal- bers invade their target zones, the more proximally do they
Retzius cells, located in the termination zones of entorhinal terminate on their target cell dendrites (Bayer and Altman,
bers, may serve as primary targets. In fact, entorhinal bers 1987). Entorhinal axons, arriving early in development, con-
were found to establish synaptic contacts with cell bodies and tact distal dendrites of pyramidal cells and granule cells,
dendrites of Cajal-Retzius cells (del Rio et al., 1997; Frotscher whereas later-arriving commissural bers establish synapses
et al., 2001). A role of Cajal-Retzius cells as primary target in the stratum radiatum of the hippocampus proper and in
neurons of entorhinal bers is supported by X-irradiation the inner molecular layer of the dentate gyrus. One of the last
experiments. Irradiation of newborn rats, which eliminates projections to be formed, the ipsilateral mossy ber projec-
most of the postnatally generated granule cells but not the tion, accordingly terminates on the most proximal dendritic
early-born Cajal-Retzius cells, does not alter the layer-specic segments of CA3 pyramidal neurons.
termination of entorhinal bers (Laurberg and Hjorth- This hypothesis is not easy to test in vivo. However, it can
Simonsen, 1977; Frotscher et al., 2001). Conversely, ablation of be addressed in slice culture experiments in which the time of
Cajal-Retzius cells prevented layer-specic ingrowth (del Rio arrival of growing axons can be brought under precise exper-
et al., 1997). imental control. Frotscher and Heimrich (1993) cocultured a
Many Cajal-Retzius cells degenerate later during postnatal hippocampal target culture rst with an entorhinal culture
development (i.e., after the entorhino-hippocampal projec- and then with a second hippocampal culture, thus imitating
tion has been formed and pyramidal cells and granule cells the normal developmental sequence in this sequential culture
have grown their distal dendrites into the entorhinal termina- system. They then traced the entorhinal and commissural
tion elds). At that time, the contacts of entorhinal bers are connections to the hippocampal target culture and observed,
reorganized, and denitive synapses with distal dendritic por- as one would expect, that the entorhinal bers terminated on
tions of pyramidal cells and granule cells are made (Fig. the distal dendritic segments of their target neurons, as nor-
46D). Although pathnding and target recognition are not mal. Next, they reversed this sequence: First, the two hip-
affected by the absence of neuronal activity, there is an effect pocampal cultures were cocultivated; then, with a delay of a
on the maturation of synaptic structures. Drakew and col- couple of days, an entorhinal slice culture was added. Despite
leagues (Drakew et al., 1999; Frotscher et al., 2000b) observed the reversal of the sequence of ber ingrowth, the commis-
normal formation of the entorhino-hippocampal projection sural and entorhinal bers still terminated in their appropri-
in cocultures of entorhinal cortex and hippocampus incu- ate layers, indicating that the sequence of ber arrival in the
bated in the presence of tetrodotoxin (TTX), which blocks fast target region is not responsible for the laminated, proximo-
sodium channels. However, they observed an increased num- distal termination of these two afferents.
124 The Hippocampus Book

As the time of arrival is not the factor governing lamina- insert). This is in striking contrast to the entorhino-dentate
tion, the question arises as to potential cellular and molecular projection, which forms a compact termination zone with
factors determining the layer specicity of commissural bers. sharp borders in the outer molecular layer of the reeler mutant
Soriano and colleagues (Super et al., 1998) have suggested that (Deller et al., 1999a); moreover, this characteristic layer-
there are pioneer neurons, early-generated GABAergic cells, specic projection is nicely preserved in slice cultures from
that serve as primary targets for the commissural bers. Their wild-type animals and reeler mutants (Zhao et al., 2003) (Fig.
role is comparable to that of the Cajal-Retzius cells as primary 48, see color insert). Together, these ndings support the
targets of the entorhinal afferents. This idea became testable hypothesis that entorhinal bers, because of their early arrival
with the possibility of depleting the hippocampus of in the dentate gyrus at a time when their denitive target cells,
GABAergic neurons during development (Pleasure et al., the granule cells, have not yet been generated, do require pio-
2000). Contrary to Sorianos idea, commissural bers still neer target neurons. These neurons are likely to be the Cajal-
innervated their proper target areas in the absence of Retzius cells, which are, as normal, located in the outer
GABAergic neurons. molecular layer of reeler mice (see above). One is tempted to
As the commissural bers arrive relatively late in the hip- speculate that whereas the entorhinal bers require pioneer
pocampus and dentate gyrus, an alternative proposal was that neurons the commissural afferents establish contacts directly
pioneer neurons are not required. This is because the target with principal cell dendrites, which is reected in reeler mice
cells, the principal neurons, are already present at that time and by the irregular, loose distribution of both commissural bers
have already grown a dendritic arbor (Deller et al., 1999a). and granule cells (Fig. 49). In fact, when commissural bers
Evidence relevant to this idea has come from studies in reeler from a reeler culture are traced to a wild-type culture, they are
mice. In this mutant, the granule cells are loosely distributed found to form a normal, compact projection to the inner
throughout the hilar region (Staneld and Cowan, 1979; Deller molecular layer, precluding a cell-autonomous effect of the
et al., 1999a; Drakew et al., 2002) and the commissural bers reeler mutation on commissural neurons (Zhao et al., 2003).
do not form a compact layer but are, instead, distributed all Little is known about the molecules involved in pathnd-
over the hilar region (Deller et al., 1999a). This target-depend- ing and target recognition of commissural bers. As with
ent termination of commissural bers is nicely seen when other commissures, Netrin 1 and its receptor DCC are impor-
commissural axons from a wild-type culture are traced to both tant for formation of the hippocampal commissure. Thus, a
a second wild-type culture and a reeler culture: Whereas the hippocampal commissure does not form in Netrin 1-decient
commissural axons terminating in the wild-type culture form animals. Along this line, neurites of hippocampal neurons are
their normal compact projection to the inner molecular layer attracted by Netrin 1 in the stripe choice assay (Steup et al.,
of the dentate, the commissural bers that innervate the reeler 2000). The molecules governing the laminar specicity of the
culture terminate diffusely, corresponding to the scattered dis- commissural bers in their target zones remain to be deter-
tribution of their target granule cells (Fig. 47, see color mined.

Figure 47. Triplet cocultures of two wild-type

slices of hippocampus together with a reeler
hippocampal slice. The biocytin injection site
into the dentate gyrus of one of the wild-type
cultures (DG1) is marked with an asterisk.
Labeled commissural bers invade the two
other cultures, but only in the dentate gyrus of
the second wild-type culture (DG2) do they
show their characteristic compact termination
in the inner molecular layer (arrow). In the
reeler hippocampal culture, the commissural
bers arising from the same cells of origin have
lost their laminar specicity and terminate
throughout the hilar region of the dentate gyrus
(DG3). Arrowhead points to the mossy ber
projection in the wild-type culture that was also
labeled by the tracer injection into the hilus. p,
pyramidal layer; p1, p2 two pyramidal layers in
the reeler culture. Bar  100 m. (Source: Zhao
et al., 2003, 2003 by the Society for
Morphological Development of the Hippocampus 125

Figure 48. A. Entorhino-hippocampal projec-

tion in a complex culture of entorhinal cortex
(EC) and hippocampus. Biocytin injection sites
are labeled by asterisks. Note the layer-specic
termination of the entorhinal bers in the outer
molecular layer (OML) of the dentate gyrus. B.
In a slice culture of reeler hippocampus and
entorhinal cortex, entorhinal bers labeled by
biocytin injection into the entorhinal cortex
(asterisks) form a sharply segregated projection
to the outer portion of the molecular layer of the
dentate gyrus (DG) despite the malpositioned
granule cells, demonstrating that the trajectory
of entorhinal bers is not inuenced by the posi-
tion of their target neurons. p1, p2, two pyrami-
dal layers in the reeler hippocampus. Bar  100
m. (Source: Zhao et al., 2003; 2003 by the
Society for Neuroscience.)

4.3.3 Septal Connections pocampal bers can nd their way to the hippocampus. It has
in fact been shown that growth cones of septohippocampal
The septohippocampal projection consists of two parts: a bers grow along hippocampo-septal axons (Linke and
cholinergic one and a GABAergic one. The cells of origin for Frotscher, 1993). The molecular mechanisms underlying the
both parts are located in the medial septal nucleus/diagonal change of hippocampo-septal bers from the medial to the
band complex (MSDB) (see Chapter 3). The development of lateral septum remain to be elucidated.
the septohippocampal projection can be understood only by Unlike the entorhino-hippocampal and commissural pro-
taking into account the hippocampo-septal projection. The jections to the hippocampus, septohippocampal cholinergic
latter develops relatively early and surprisingly terminates in bers do not terminate in clearly demarcated layers. They are
the medial septum during development as early as on E 15. In found in all layers of the hippocampal formation but are cer-
adults, hippocampo-septal bers project to the lateral septum tainly more concentrated in the cell body layers. In contrast,
(Linke and Frotscher, 1993; Super and Soriano, 1994; Linke et the septohippocampal GABAergic projection cells display a
al., 1995). This early termination in the medial septal nucleus high target cell specicity terminating almost exclusively on
has led to the hypothesis that the early formed hippocampo- GABAergic interneurons in the hippocampus. Inhibitory neu-
septal projection serves as a template by which septohip- rons afferent to inhibitory neurons serve a disinhibitory role
126 The Hippocampus Book

Figure 49. Different signals control the laminar specicity of outgrowing granule cell dendrites. B. Findings in the reeler mutant,
entorhinal bers (black) and commissural bers (gray) to the den- where the granule cells are not aligned but scattered over the den-
tate gyrus. A. In wild-type mice, entorhinal bers, guided by tate gyrus, conrm that the entorhinal axons are not guided by their
Cajal-Retzius cell axons, reach the outer molecular layer (OML) target granule cells. As for the wild-type bers, entorhinal bers ter-
of the dentate gyrus before the granule cells in the granule cell minate with laminar specicity in the OML. In contrast, commis-
layer (GCL) have grown their distal dendritic tips to the OML. sural bers are loosely distributed over the dentate gyrus, thus
Molecules of the extracellular matrix control the laminar speci- following their scattered target cells. (Source: Adapted from
city of entorhinal axons. In contrast, later-arriving commissural Frotscher, 1998).
bers, terminating in the inner molecular layer (IML), meet the

with respect to the principal neurons (Freund and Antal, for axonal orientation (see above). A neurotropic effect of
1988). The septal cholinergic bers establish contacts with NGF was demonstrated some time ago (Gundersen and
both principal neurons and interneurons, and their contacts Barrett, 1979; see Tessier-Lavigne and Goodman, 1996, for
are on cell bodies, dendritic shafts, and spines. Two types of review), but it needs to be conrmed that the ingrowth of
contact are established: symmetrical and asymmetrical septohippocampal bers is controlled by neurotrophins.
(Frotscher and Leranth, 1985, 1986). It remains to be estab- One of the semaphorins, Sema3C, was found to repel septal
lished whether one of these types derives from the few bers (Steup et al., 2000); and neuropilin 2, its receptor, is
intrahippocampal cholinergic cells (Frotscher et al., 1986, present along the pathway of septal bers (Chedotal et al.,
2000a) that are found in the rat hippocampus but not in the 1998; Steup et al., 2000), suggesting involvement of this lig-
monkey hippocampus. andreceptor system in the formation of the septohippocam-
The rst septohippocampal bers arrive in the hippocam- pal projection.
pus at E 17 (i.e., 2 days after the hippocampo-septal projec-
tion is established) (Linke and Frotscher, 1993; Super and 4.3.4 General Principles Underlying
Soriano, 1994). Little is known about the molecules involved the Formation of Synaptic
in pathnding and target recognition of septohippocampal Connections in the Hippocampus
bers. Cholinergic bers are known to bear the p75 neu-
rotrophin receptor (p75NTR), suggesting that neurotrophins We are now beginning to understand at least some of the prin-
play a role. Nerve growth factor (NGF), one of the ligands for ciples underlying pathway formation in the hippocampus.
the p75 receptor, is synthesized by GABAergic hippocampal Different cellular and molecular factors, probably being effec-
neurons, which are distributed over all hippocampal layers, tive only during certain developmental time windows, are
compatible with the diffuse termination of septohippocampal involved in the formation of the different projections to the
bers. Moreover, NGF-synthesizing hippocampal nonprinci- hippocampus. When describing the formation of a pathway,
pal neurons project to the septum (Acsady et al., 2000), we found it useful to distinguish between axonal path nding ,
thus providing both a neurotrophic source and a template target recognition, and synapse formation. For pathnding and
Morphological Development of the Hippocampus 127

target recognition, pioneer neurons provide a template by an 4.4.1 Neurogenesis

early-formed projection. Membrane-bound and soluble mol-
ecules and components of the extracellular matrix also seem The duration of embryonic development is 19 to 21 days in
to play a role at different times of development. A variety of mice and rats (life-span 1.53.0 years), 165 days in rhesus
ligandreceptor interactions has been elucidated that, by their monkeys (life span 2530 years), and 280 days in humans. As
repulsive or attractive effects, may guide the growth cone of a we have already noted, cell formation is a relatively late event
growing axon to the target region. Undoubtedly, new mole- in rodents: Pyramidal cells are generated during the second
cules or families of molecules of this kind will be discovered in half of the embryonic period, and granule cell formation
the future. Extracellular matrix molecules, proteoglycans such starts only a few days before birth. Furthermore, 85% of the
as neurocan, may be essential for the segregation of ber sys- dentate granule cells are formed postnatally. In contrast, neu-
tems in the hippocampus (Frster et al., 2001; Zhao et al., rogenesis in the primate hippocampal formation takes place
2003). relatively early. In rhesus monkeys, the rst neurons appear
Synapse formation is dealt with only briey here, and almost simultaneously in the various subregions of the hip-
the reader is referred to quantitative electron microscopic pocampal formation, from the entorhinal cortex to the den-
analyses and Golgi studies on the development of synapses tate gyrus, between E 36 and E 38 (Rakic and Nowakowski,
and dendritic spines (e.g., Crain et al., 1973; Minkwitz 1976; 1981). Except for the dentate gyrus, cell formation ceases dur-
Frotscher et al., 1977). This is a rapidly developing eld, ing the rst half of pregnancy, between E 62 and E 65. Granule
and more recent studies have provided evidence that the cell formation lasts until the end of the rst postnatal month,
mechanisms leading to new contacts or changes in the shape with only 15% of the granule cells forming postnatally (Rakic
of synapses or spines may not only be effective during and Nowakowski, 1981). Granule cell formation has also been
development but may also underlie plastic processes in the found in the dentate gyrus of adult monkeys (Kornack and
adult organism. A striking example is Engert and Bonhoeffers Rakic, 1999).
(1999) in vitro observation of new spines formed during In humans, the time of pyramidal cell neurogenesis is
the development of long-term potentiation (LTP). Similarly, not known and must be extrapolated from the time of the
Woolley and McEwen (1993) have shown that estrogen appearance of the cytoarchitectonic layers. Subelds of the
application to ovariectomized adult rats induces new spines hippocampal formation, such as the entorhinal cortex, subicu-
on CA1 pyramidal cell dendrites, an effect likely caused lum, and hippocampus, are discernible between E 70 and E 80,
by estrogen acting primarily on CA3 pyramidal cells and whereas the dentate granular layer appears around E 90
inducing the sprouting of their Schaffer collaterals to CA1 (Humphrey, 1967). Between E 160 and E 170, the cytoarchitec-
pyramidal neurons (Rune et al., 2002). Interestingly, recent tonic characteristics of all of the divisions of the hippocampal
studies have shown that estrogens synthesized in the formation are similar to what is observed in adults
hippocampus are important for hippocampal synaptic (Humphrey, 1967; Arnold and Trojanowski, 1996). With the
plasticity (Kretz et al., 2004). Future studies will help us estab- aid of the mitotic marker MIB-1 (Ki-67), we have directly
lish whether these plastic changes at spine synapses are shown that the ventricular zone and hilus contain a large num-
specic for hippocampal neurons or are a more general phe- ber of proliferating cells at E 100 (Fig. 44). The number of
nomenon occurring at many synapses in the central nervous labeled cells decreases rapidly from E 170 onward (24th week),
system. but MIB-1 (Ki-67)-positive cells are still observed in the hilus
and granule cell layer during the rst 6 months postnatally
(Seress et al., 2001).

4.4 Development of the Primate 4.4.2 Neuronal Differentiation

Hippocampal Formation
Differentiation of pyramidal neurons and granule cells con-
As in the previous chapter (see Chapter 3), one may raise the tinues for a relatively long period in rodents and primates.
question of how hippocampal development varies across phy- Spine density increases until the time of sexual maturation,
logenetically distant species such as rats, monkeys, and and total dendritic length is not reached before P 90 in the rat
humans. What is similar and what is different? Do the steps of CA1 area (Pokorny and Yamamoto, 1981). In the monkey, the
hippocampal development change proportionally with an spine density of CA3 and CA1 pyramidal cells does not reach
extended life-span, or are there some fundamental differences adult levels before sexual maturation (i.e., during the fourth
in the mechanisms and the timing of cell formation, cell year in rhesus monkeys) (Seress and Ribak, 1995b, unpub-
migration, dendritic development, and formation of neuronal lished observation). In humans, spine density and total den-
connectivity? We provide a short comparative analysis of hip- dritic length of CA1 pyramidal neurons increase until at least
pocampal development in primates. For details the reader the third postnatal year (Simic et al., 2001). In contrast to the
should consult a review on monkey and human hippocampal long-term differentiation of principal cells, local circuit neu-
development (Seress, 2001). rons in the primate hippocampal formation mature fast, as
128 The Hippocampus Book

indicated by the early fetal appearance of neurochemical Altman J (1966) Autoradiographic and histological studies of post-
markers of these neurons (Del Fiacco and Quartu, 1989; natal neurogenesis. II. A longitudinal investigation of kinetics,
Berger et al., 1993, 1999). migration and transformation of cells incorporating tritiated
Arrival of afferents at the rat hippocampal formation coin- thymidine in infant rats, with special reference to postnatal neu-
rogenesis of some brain regions. J Comp Neurol 128:431474.
cides with the development of their target cells. Thus, it takes
Altman J, Bayer SA (1975) Postnatal development of the hippocam-
place during the perinatal and early postnatal period, as pre-
pal dentate gyrus under normal and experimental conditions.
viously noted. In both the monkey and human, however, In: The hippocampus, vol 2 (Isaacson RL, Pribram KH, eds), pp.
afferent projections from the hypothalamus as well as from 95122. New York: Plenum Press.
the entorhinal cortex arrive early during the fetal period Altman J, Bayer SA (1990a) Mosaic organization of the hippocampal
(Kostovic et al., 1989; Hevner and Kinney, 1996; Berger et al., neuroepithelium and the multiple germinal sources of dentate
2001). In contrast to extrinsic afferents, intrinsic associational granule cells. J Comp Neurol 301:325342.
connections of the rodent hippocampus develop relatively Altman J, Bayer SA (1990b) Prolonged sojourn of developing pyram-
late, partly owing to the late formation of granule cells. idal cells in the intermediate zone of the hippocampus and their
Connections between dentate granule cells and CA3 pyrami- settling in the stratum pyramidale. J Comp Neurol 301:343364.
dal neurons as well as between dentate granule cells and hilar Altman J, Bayer SA (1990c) Migration and distribution of two popu-
lations of hippocampal granule cell precursors during the peri-
mossy cells develop postnatally (Ribak et al., 1985). This late
natal and postnatal periods. J Comp Neurol 301:365381.
development coincides with the prolonged postnatal develop-
Altman J, Das DG (1966) Autoradiographic and histological studies
ment of hippocampal functions (Nadel and Willner, 1989). In of postnatal neurogenesis. I. A longitudinal investigation of the
contrast, the development of associational connections is rel- kinetics, migration and transformation of cells incorporating
atively fast in rhesus monkeys, and the mainly prenatally tritiated thymidine in neonate rats, with special reference to
formed granule cells establish a well developed synaptic net- postnatal neurogenesis in some brain regions. J Comp Neurol
work by the fth to seventh postnatal month (Seress and 126:337390.
Ribak, 1995a,b). Surprisingly, this is not the case in humans, Amaral DG, Kurz J (1985) The time of origin of cells demonstrating
where the prenatally formed granule cells appear to form the glutamic acid-like immunoreactivity in the hippocampal for-
rst connections with hilar mossy cells and CA3 pyramidal mation of the rat. Neurosci Lett 59:3339.
neurons as late as during the 33rd gestational week, as indi- Angevine JB (1965) Time of neuron origin in the hippocampal
region: an autoradiographic study in the mouse. Exp Neurol
cated by the late appearance of excrescences on hilar mossy
Suppl 2:170.
cells and CA3 pyramidal neurons (Purpura, 1975). In fact,
Arnold SE, Trojanowski JQ (1996) Human fetal hippocampal devel-
there are almost no light microscopic signs of such connec- opment. I. Cytoarchitecture, myeloarchitecture and neuronal
tions at birth (Seress and Mrzljak, 1992; Seress, 2001). morphologic features. J Comp Neurol 367:274292.
Taken together, developmental events of the hippocampal Bayer SA (1980) Development of the hippocampal region in the rat.
formation are different in rodents and primates. However, I. Neurogenesis examined with 3H-thymidine autoradiography.
there are also common traits. In both rodents and humans the J Comp Neurol 190:87114.
late development of intrahippocampal associational connec- Bayer SA, Altman J (1987) Directions in neurogenetic gradients and
tions may inuence the maturation of the entire hippocampal patterns of anatomical connections in the telencephalon. Prog
network. The factors contributing to this late development are Neurobiol 29:57106.
likely to be different in rodents and primates. In rodents, the Berger B, Alvarez C, Goldman-Rakic PS (1993) Neurochemical devel-
opment of the hippocampal region in the fetal rhesus monkey.
late formation of granule cells may play a role. It remains to be
I. Early appearance of peptides, calcium binding proteins,
elucidated why the early-formed granule cells in humans do
DARPP-32 and the monoamine innervation in the entorhinal
not immediately form synapses with their target cells as is the cortex during the rst half of gestation (E47 to E90). Hippocam-
case in nonhuman primates. pus 3:279305.
Berger B, DeGrissac N, Alvarez C (1999) Precocious development of
parvalbumin-like immunoreactive interneurons in the hip-
ACKNOWLEDGMENTS pocampal formation and entorhinal cortex of the fetal
cynamolgus monkey. J Comp Neurol 403:309331.
This work was supported by Deutscher Akademischer Austausch- Berger B, Esclapez M, Alvarez C, Meyer G, Catala M (2001) Human
dienst (DAAD) with grants 323/2000 and 324/jo-H/2002, the Hun- and monkey fetal brain development of the supramammillary-
garian Ministry of Education (FKFP) with grant 0515/2000, hippocampal projections: a system involved in the regulation of
and the Deutsche Forschungsgemeinschaft (SFB 505 and Transregio theta activity. J Comp Neurol 429:515529.
SFB TR-3). Borrell V, Del Rio JA, Alcantara S, Derer M, Martinez A, DArcangelo
G, Nakajima K, Mikoshiba K, Derer P, Curran T, Soriano E
(1999) Reelin regulates the development and synaptogenesis of
REFERENCES the layer-specic entorhino-hippocampal connections. J Neu-
rosci 19:13451358.
Acsady L, Pascual M, Rocamora N, Soriano E, Freund TF (2000) Ceranik K, Deng J, Heimrich B, Lbke J, Zhao S, Frster E, Frotscher
Nerve growth factor but not neurotrophin-3 is synthesized by M (1999) Hippocampal Cajal-Retzius cells project to the
hippocampal GABAergic neurons that project to the medial entorhinal cortex: retrograde tracing and intracellular labelling
septum. Neuroscience 98:2331. studies. Eur J Neurosci 11:42784290.
Morphological Development of the Hippocampus 129

Chedotal A, del Rio JA, Ruiz M He Z, Borrell V, de Castro F, Ezan F, Frster E, Tielsch A, Saum B, Weiss KH, Johanssen C, Graus-Porta
Goodman CS, Tessier-Lavigne M, Sotelo C, Soriano E (1998) D, Mller U, Frotscher M (2002) Reelin, disabled 1 and beta1-
Semaphorins III and IV repel hippocampal axons via two dis- integrins are required for the formation of the radial glial
tinct receptors. Development 125:43134323. scaffold in the hippocampus. Proc Natl Acad Sci USA 99:
Crain B, Cotman C, Taylor D, Lynch G (1973) A quantitative electron 1317813183.
microscopic study of synaptogenesis in the dentate gyrus of the Freund TF, Antal M (1988) GABA-containing neurons in the septum
rat. Brain Res 63:195204. control inhibitory interneurons in the hippocampus. Nature
Czurk A, Czeh B, Seress L, Nadel L, Bures J (1997) Severe spatial 336:170173.
navigation decit in the Morris watermaze after single high dose Freund TF, Buzski G (1996) Interneurons of the hippocampus.
of neonatal X-ray irradiation in the rat. Proc Natl Acad Sci USA Hippocampus 6:347470.
94:27662771. Frotscher M (1997) Dual role of Cajal-Retzius cells and reelin in cor-
DArcangelo G, Miao GG, Chen S-C, Soares HD, Morgan JI, tical development. Cell Tissue Res 290:315322.
Curran T (1995) A protein related to extracellular matrix Frotscher M (1998) Cajal-Retzius cells, reelin, and the formation of
proteins deleted in the mouse mutant reeler. Nature 374: layers. Curr pOin Neurobiol 8:570575.
719723. Frotscher M, Heimrich B (1993) Formation of layer-specic ber
DArcangelo G, Nakajima K, Miyata T, Ogawa M, Mikoshiba K, projections to the hippocampus in vitro. Proc Natl Acad Sci USA
Curran T (1997) Reelin is a secreted glycoprotein recognized by 90:1040010403.
the CR-50 monoclonal antibody. J Neurosci 17:2331. Frotscher M, Leranth C (1985) Cholinergic innervation of the rat
Del Fiacco M, Quartu M (1989) Ontogenetic variance of substance P hippocampus as revealed by choline acetyltransferase immuno-
in the human hippocampus. In: The hippocampus: newvistas cytochemistry: a combined light and electron microscopic
(Chan Palay V, Kohler C, eds), pp 131144. New York: Alan R. study. J Comp Neurol 239:237246.
Liss. Frotscher M, Leranth C (1986) The cholinergic innervation of the rat
Deller T, Drakew A, Frotscher M (1999a) Different primary target fascia dentata: identication of target structures on granule cells
cells are important for ber lamination in the fascia dentata: a by combining choline acetyltransferase immunocytochemistry
lesson from reeler mutant mice. Exp Neurol 156:239253. and Golgi impregnation. J Comp Neurol 243:181190.
Deller T, Drakew A, Heimrich B, Frster E, Tielsch A, Frotscher M Frotscher M, Hmori J, Wenzel J (1977) Transneuronal effects of
(1999b) The hippocampus of the reeler mutant mouse: ber entorhinal lesions in the early postnatal period on synapto-
segregation in area CA1 depends on the position of the postsy- genesis in the hippocampus of the rat. Exp Brain Res 30:
naptic target cells. Exp Neurol 156:254267. 549560.
Del Rio JA, Heimrich B, Borrell V, Frster E, Drakew A, Alcntara S, Frotscher M, Schlander M, Leranth C (1986) Cholinergic neurons in
Nakajima K, Miyata T, Ogawa M, Mikoshiba K, Derer P, the hippocampus: a combined light and electron microscopic
Frotscher M, Soriano E (1997) A role for Cajal-Retzius cells and immunocytochemical study in the rat. Cell Tissue Res 246:
reelin in the development of hippocampal connections. Nature 293301.
385:7074. Frotscher M, Vida I, Bender R (2000a) Evidence for the existence of
Drakew A, Frotscher M, Heimrich B (1999) Blockade of neuronal non-GABAergic, cholinergic interneurons in the rodent hip-
activity alters spine maturation of dentate granule cells but not pocampus. Neuroscience 96:2731.
their dendritic arborization. Neuroscience 94:767774. Frotscher M, Drakew A, Heimrich B (2000b) Role of afferent inner-
Drakew A, Deller T, Heimrich B, Gebhardt C, Del Turco D, Tielsch A, vation and neuronal activity in dendritic development and
Frster E, Herz J, Frotscher M (2002) Dentate granule cells in spine maturation of fascia dentata granule cells. Cereb Cortex
reeler mutants and VLDLR and ApoER2 knockout mice. Exp 10:946951.
Neurol 176:1224. Frotscher M, Seress L, Abraham H, Heimrich B (2001) Early gener-
Dupuy ST, Houser CR (1997) Developmental changes in GABAergic ated Cajal-Retzius cells have different functions in cortical
neurons of the rat dentate gyrus: an in situ hybridization and development. In: Brain stem cells (Miyan J, Thorndyke M,
birth dating study. J Comp Neurol 389:402418. Beesley PW, Bannister C, eds), pp. 4349. Oxford: BIOS
Dupuy ST, Houser CR (1999) Evidence for changing positions of Scientic Publishers.
GABA neurons in the developing rat dentate gyrus. Frotscher M, Haas CA, Frster E (2003) Reelin controls granule cell
Hippocampus 9:186199. migration in the dentate gyrus by acting on the radial glial scaf-
Engert F, Bonhoeffer T (1999) Dendritic spine changes associated fold. Cereb Cortex 13:634640.
with hippocampal long-term synaptic plasticity. Nature Galceran J, Miyashita-Lin EM, Devaney E, Rubenstein JLR,
399:6670. Grosschedl R (2000) Hippocampus development and genera-
Eriksson PS, Perlieva E, Bjrk-Eriksson T, Alborn A, Nordborg C, tion of dentate gyrus granule cells is regulated by LEF1.
Peterson DA, Gage FH (1998) Neurogenesis in the adult human Development 127:469482.
hippocampus. Nat Med 4:13131317. Gundersen RW, Barrett JN (1979) Neuronal chemotaxis: chick dorsal-
Falconer DS (1951) Two new mutants, trembler and reeler, with root axons turn toward high concentrations of nerve growth
neurological actions in the house mouse (Mus musculus L.). J factor. Science 206:10791080.
Genet P( aris) 50:192201. Hjos N, Papp E, Acsdy L, Levey AI, Freund TF (1998) Distinct
Frster E, Kaltschmidt C, Deng J, Cremer H, Deller T, Frotscher M interneuron types express m2 muscarinic receptor immunore-
(1998) Lamina-specic cell adhesion on living slices of hip- activity on their dendrites or axon terminals in the hippocam-
pocampus. Development 125:33993410. pus. Neuroscience 82:355376.
Frster E, Zhao S, Frotscher M (2001) Hyaluronan-associated adhe- Halasy K, Somogyi P (1993a) Distribution of GABAergic synapses
sive cues control fiber segregation in the hippocampus. and their targets in the dentate gyrus of rat: a quantitative
Development 128:30293039. immunoelectron microscopic analysis. J Hirnforsch 34:299308.
130 The Hippocampus Book

Halasy K, Somogyi P (1993b) Subdivisions in the multiple Nadel L, Willner J (1989) Some implications of postnatal maturation
GABAergic innervation of granule cells in the dentate gyrus of of the hippocampus. In: The hippocampus: newvistas (Chan-
the rat hippocampus. Eur J Neurosci 5:411429. Palay V, Kohler C, eds), pp. 1731. New York: Alan R Liss.
Halasy K, Buhl EH, Lorinczi Z, Tams G, Somogyi P (1996) Synaptic Nguyen-Ba-Charvet KT, Brose K, Marillat V, Kidd T, Goodman CS,
target selectivity and input of GABAergic basket and bistratied Tessier-Lavigne M, Sotelo C, Chedotal A (1999) Slit2-mediated
interneurons in the CA1 area of the rat hippocampus. chemorepulsion and collapse of developing forebrain axons.
Hippocampus 6:306329. Neuron 22:463473.
Hevner RF, Kinney HC (1996) Reciprocal entorhinal-hippocampal Pleasure SJ, Anderson S, Hevner R, Bagri A, Lowenstein DH,
connections established by human fetal midgestation. J Comp Rubenstein JLR (2000) Cell migration from the ganglionic emi-
Neurol 372:384394. nences is required for the development of hippocampal
Hirotsune S, Takahare T, Sasaki N, Hirose K, Yoshiki A, Ohashi T, GABAergic interneurons. Neuron 26:727740.
Kusakabe M, Murakami Y, Muramatsu M, Watanabe S, Nakao Pokorny J, Yamamoto Y (1981) Postnatal ontogenesis of hippocam-
K, Katsuki M, Hayashizaki Y (1995) The reeler gene encodes a pal CA1 area in rats. I. Development of dendritic arborization in
protein with an EGF-like motif expressed by pioneer neurons. pyramidal neurons. Brain Res Bull 7:113120.
Nat Genet 10:7783. Purpura DS (1975) Normal and aberrant neuronal development in
Humphrey T (1967) The development of the human hippocampal the cerebral cortex of human fetus and young infant. In: Brain
ssure. J Anat 101:655676. mechanisms in mental retardation: UCLA forum in medical sci-
Katona I, Sperlgh B, Sk A, Kofalvi A, Vizi ES, Freund TF (1999) ences pp. 141169. San Diego: Academic Press.
Presynaptically located CB1 cannabinoid receptors regulate Rakic P, Nowakowski RS (1981) The time of origin of neurons in the
GABA release from axon terminals of specic hippocampal hippocampal region of the rhesus monkey. J Comp Neurol
interneurons. J Neurosci 19:45444558. 196:99128.
Kempermann G, Kuhn HG, Gage FH (1997a) More hippocampal Ramn y Cajal S (1911) Histologie du systme nerveux de lhomme et
neurons in adult mice living in an enriched environment. des vertbrs , vol 2. Paris: Maloine.
Nature 386:493495. Retzius G (1893) Die Cajalschen Zellen der Grosshirnrinde beim
Kempermann G, Kuhn HG, Gage FH (1997b) Genetic inuence on Menschen und bei Sugetieren. Biol Unters 5:19.
neurogenesis in the dentate gyrus of adult mice. Proc Natl Acad Retzius G (1894) Weitere Beitrge zur Kenntnis der Cajalschen Zellen
Sci USA 94:1040910414. der Grosshirnrinde des Menschen. Biol Unters 6:2934.
Kornack DR, Rakic P (1999) Continuation of neurogenesis in the Ribak CE, Seress L (1983) Five types of basket cell in the hippocam-
hippocampus of the adult macaque monkey. Proc Natl Acad Sci pal dentate gyrus. J Neurocytol 12:577597.
USA 96:57685773. Ribak CE, Seress L, Amaral DG (1985) The development, ultrastruc-
Kostovic I, Seress L, Mrzljak L, Judas M (1989) Early onset of synapse ture and synaptic connections of the mossy cells of the dentate
formation in the human hippocampus: a correlation with Nissl- gyrus. J Neurocytol 14:717730.
Golgi architectonics in 15 and 16.5-week-old fetuses. Rune GM, Wehrenberg U, Prange-Kiel J, Zhou L, Adelmann G,
Neuroscience 30:105116. Frotscher M (2002) Estrogen up-regulates estrogen receptor
Kretz O, Fester L, Wehrenberg U, Zhou L, Brauckmann S, Zhao S, alpha and synaptophysin in slice cultures of rat hippocampus.
Prange-Kiel J, Naumann T, Jarry H, Frotscher M, Rune GM Neuroscience 113:167175.
(2004) Hippocampal synapses depend on hippocampal estro- Savaskan NE, Plaschke M, Ninnemann O, Spillmann AA, Schwab
gen synthesis. J Neurosci 24:59135921. ME, Nitsch R, Skutella T (1999) Myelin does not inuence the
Lang U, Frotscher M (1990) Postnatal development of nonpyramidal choice behaviour of entorhinal axons but strongly inhibits their
neurons in the rat hippocampus (areas CA1 and CA3): a com- outgrowth length in vitro. Eur J Neurosci 11:316326.
bined Golgi/electron microscope study. Anat Embryol 181: Seress L (1977) The postnatal development of rat dentate gyrus and
533545. the effect of early thyroid hormone treatment. Anat Embryol
Laurberg S, Hjorth-Simonsen A (1977) Growing central axons 151:335339.
deprived of normal target neurones by neonatal X-ray irradia- Seress L (1978) Divergent responses to thyroid hormone treatment of
tion still terminate in a precisely laminated fashion. Nature the different secondary germinal layers in the postnatal rat
269:158160. brain. J Hirnforsch 19:395403.
Linke R, Frotscher M (1993) Development of the rat septohip- Seress L (2001) Morphological changes in the human hippocampal
pocampal projection: tracing with DiI and electron microscopy formation from midgestation to early childhood. In: Handbook
of identied growth cones. J Comp Neurol 332:6988. of developmental cognitive neuroscience (Nelson CA, Luciana M,
Linke R, Pabst T, Frotscher M (1995) Development of the hippocam- eds), pp. 4558. Cambridge, MA: MIT Press.
poseptal projection in the rat. J Comp Neurol 351:602616. Seress L, Frotscher M (1990) Morphological variability is a charac-
Lbbers K, Wolff JR, Frotscher M (1985) Neurogenesis of GABAergic teristic feature of granule cells in the primate fascia dentata: a
neurons in the rat dentate gyrus: a combined autoradiographic combined Golgi-electron microscopic study. J Comp Neurol
and immunocytochemical study. Neurosci Lett 62:317322. 293:253267.
Minkwitz H-G (1976) Zur Entwicklung der Neuronenstruktur des Seress L, Mrzljak L (1987) Basal dendrites of granule cells are normal
Hippocampus whrend der pr- und postnatalen Ontogenese features of the fetal and adult dentate gyrus of both monkey and
der Albinoratte. III. Mitteilung: Morphometrische Erfassung human hippocampal formations. Brain Res 405:169174.
der ontogenetischen Vernderungen in Dendritenstruktur und Seress L, Mrzljak L (1992) Postnatal development of mossy cells in
Spinebesatz an Pyramidenneuronen (CA1) des Hippocampus. J the human dentate gyrus: a light microscopic Golgi study.
Hirnforsch 17:255275. Hippocampus 2:127142.
Morphological Development of the Hippocampus 131

Seress L, Ribak CE (1983) GABAergic neurons in the dentate gyrus Staneld BB, Cowan WM (1979) The morphology of the hippocam-
appear to be local circuit and projection neurons. Exp Brain Res pus and dentate gyrus in normal and reeler mice. J Comp Neurol
50:173182. 185:393422.
Seress L, Ribak CE (1990) Postnatal development of the light and Stein E, Savaskan NE, Ninnemann O, Nitsch R, Zhou R, Skutella
electron microscopic features of basket cells in the hippocampal T (1999) A role for the Eph ligand ephrin-A3 in entorhino-
dentate gyrus of the rat. Anat Embryol 181:547566. hippocampal axon targeting. J Neurosci 15:88858893.
Seress L, Ribak CE (1995a) Postnatal development and synaptic con- Stephan, H (1975) Allocortex: Handbuch der mikroskopischen
nections of hilar mossy cells in the hippocampal dentate gyrus Anatomie des Menschen, vol IV/9. Berlin: Springer.
of rhesus monkeys. J Comp Neurol 355:93110. Steup A, Ninnemann O, Savaskan NE, Nitsch R, Pschel AW, Skutella
Seress L, Ribak CE (1995b) The postnatal development of CA3 T (1999) Semaphorin D acts as a repulsive factor for entorhinal
pyramidal neurons and their afferents in the Ammons horn of and hippocampal neurons. Eur J Neurosci 11:729734.
rhesus monkeys. Hippocampus 5:217231. Steup A, Lohrum M, Hamscho N, Savaskan NE, Ninnemann O,
Seress L, Frotscher M, Ribak CE (1989) Local circuit neurons in both Nitsch R, Fujisawa H, Pschel AW, Skutella T (2000) Sema3C
the dentate gyrus and Ammons horn establish synaptic connec- and netrin-1 differentially affect axon growth in the hippocam-
tions with principal neurons in ve day-old rats: a morpholog- pal formation. Mol Cell Neurosci 15:141155.
ical basis for inhibition in early development. Exp Brain Res Super H, Soriano E (1994) The organization of the embryonic and
78:19. early postnatal murine hippocampus. II. Development of
Seress L, brahm H, Tornczky T, Kosztolnyi GY (2001) Cell for- entorhinal, commissural and septal connections studied with
mation in the human hippocampal formation from mid- the lipophilic tracer DiI. J Comp Neurol 344:101120.
gestation to the late postnatal period. Neuroscience 105:3951. Super H, Martinez A, del Rio JA, Soriano E (1998) Involvement of
Simic G, Bjelos M, Rasin MR, Seress L, Kostovic I (2001) Postnatal distinct pioneer neurons in the formation of layer-specic con-
development of human CA1 pyramidal neurons. Soc Neurosci nections in the hippocampus. J Neurosci 18:46164626.
Abstr 27:Program 250.2. Tamamaki N (1999) Development of afferent ber lamination in the
Skutella T, Nitsch R (2001) New molecules for hippocampal develop- infrapyramidal blade of the rat dentate gyrus. J Comp Neurol
ment. Trends Neurosci 24:107113. 411:257266.
Skutella T, Savaskan NE, Ninnemann O, Nitsch R (1999) Target- and Tessier-Lavigne M, Goodman CS (1996) The molecular biology of
maturation-specic membrane-associated molecules determine axon guidance. Science 274:11231132.
the ingrowth of entorhinal bers into the hippocampus. Dev Tole S, Christian C, Grove EA (1997) Early specication and
Biol 211:277292. autonomous development of cortical elds in the mouse hip-
Soriano E, Frotscher M (1989) A GABAergic axo-axonic cell in the pocampus. Development 124:49594670.
fascia dentata controls the main excitatory hippocampal path- Vida I, Frotscher M (2000) A hippocampal interneuron associated
way. Brain Res 503:170174. with the mossy fiber system. Proc Natl Acad Sci USA
Soriano E, Frotscher M (1993) GABAergic innervation of the rat fas- 97:12751280.
cia dentate: a novel type of interneuron in the granule cell layer Vida I, Halasy K, Szinyei C, Somogyi P, Buhl EH (1998) Unitary IPSPs
with extensive axonal arborization in the molecular layer. J evoked by interneurons at the stratum radiatum/lacunosum-
Comp Neurol 334:385396. moleculare border in the CA1 area of the rat hippocampus in
Soriano E, Cobas A, Fairn A (1989a) Neurogenesis of glutamic acid vitro. J Physiol (Lond) 503:755773.
decarboxylase-immunoreactive cells in the hippocampus of the Von Campe G, Spencer DD, de Lanerolle NC (1997) Morphology of
mouse. I. Regio superior and regio inferior. J Comp Neurol dentate granule cells in the human epileptogenic hippocampus.
281:586602. Hippocampus 7:472488.
Soriano E, Cobas A, Fairen A (1989b) Neurogenesis of glutamic acid Woolley CS, McEwen BS (1993) Roles of estradiol and progesterone
decarboxylase-immunoreactive cells in the hippocampus of the in regulation of hippocampal spine density during the estrous
mouse. II. Area dentata. J Comp Neurol 281:603611. cycle in the rat. J Comp Neurol 336:293306.
Spigelman I, Yan XX, Obenaus A, Lee EY-S, Wasterlain CG, Ribak CE Zhao S, Frster E, Chai X, Frotscher M (2003) Different signals con-
(1998) Dentate granule cells form novel basal dendrites in a rat trol laminar specicity of commissural and entorhinal bers to
model of temporal lobe epilepsy. Neuroscience 86:109120. the dentate gyrus. J Neurosci 23:73517357.
Spruston N, Lbke J, Frotscher M (1997) Interneurons in the stratum Zhao S, Chai X, Frster E, Frotscher M (2004) Reelin is a positional
lucidum of the rat hippocampus: an anatomical and electro- signal for the lamination of dentate granule cells. Development
physiological characterization. J Comp Neurol 385:427440. 131:51175125.
This page intentionally left blank
5 Nelson Spruston and Chris McBain

Structural and Functional Properties

of Hippocampal Neurons

models that can reproduce or explain virtually any aspect of

5.1 Overview neuronal function. The purpose of such models, of course, is
to offer insight into the neurons function within a network
One of the most striking qualities of Cajals drawings is the and provide experimentally testable predictions.
great diversity of form exhibited by neurons in the central In this chapter, our aim is to summarize the wealth of
nervous system (CNS). Looking at these drawings, it is diffi- structural, histological, and physiological information that is
cult to escape the conclusion that neurons with different mor- available for hippocampal neurons. Most of the available data
phologies are likely to function differently. Indeed a plethora on this topic come from studies of the rat hippocampal for-
of anatomical, histological, and physiological studies support mation. Accordingly, we mention species primarily when
the view that diversity in form is mirrored by diversity of referring to data from species other than rat. We begin with
function. The hippocampus presents, in many ways, a micro- the CA1 pyramidal neuron because it is arguably the most
cosm of the cellular diversity exhibited on a larger scale studied class of neuron in the brain and probably better
throughout the CNS. An important step on the path to under- understood from both structural and functional points of
standing the function of the hippocampus is to reveal the view than any other type of neuron in the hippocampus. From
unique morphological and physiological properties of the there we proceed by comparing the function of the other
great variety of neurons it comprises. pyramidal neurons in the hippocampus: those in CA3 and the
Much can be learned by examining neuronal structure. The subiculum. We next cover two nonpyramidal excitatory neu-
dendritic tree denes which inputs can be received, and the rons in the hippocampus: granule cells of the dentate gyrus
axonal arborization indicates the targets of the output and mossy cells of the hilus. Finally, we conclude by describ-
(although dendro-dendritic and axo-axonal communication ing the properties of a variety of neurons in other regions
should not be forgotten). Electron microscopy provides comprising the hippocampal formation: the entorhinal cor-
details about the nature of each neurons inputs and output. It tex, presubiculum, and parasubiculum. In the nal section, we
allows us to infer which synapses are inhibitory (symmetrical) consider the many types of interneurons found in the hip-
and which are excitatory (asymmetrical) and where modula- pocampus.
tory machinery resides (e.g., endoplasmic reticulum for cal-
cium release, ribosomes for protein synthesis, vesicles for
neurotransmitter release or insertion of membrane proteins).
Histological studies (antibody or cDNA binding, histochemi- 5.2 CA1 Pyramidal Neurons
cal assays) provide further information, ranging from which
neurotransmitters are used to which voltage-gated channels The vastness of the literature on CA1 pyramidal neurons is
are present and where. Physiological studies, using electrodes largely attributable to a combination of structural considera-
and imaging both in vitro and in vivo, reveal the dynamic tions, cell viability, and historical accidents. Although long-
function of each neuron. term potentiation was rst described in the dentate gyrus,
A detailed picture of the function of each neuron can only most of the studies in the decades that followed its original
emerge by bringing together information from this variety of description have focused on the CA1 region. This is because of
techniques. Ultimately, complete understanding of the func- the relative ease of obtaining eld potential recordings and
tion of each type of neuron may be encapsulated by detailed intracellular recordings in this region. Also, the Schaffer col-

134 The Hippocampus Book

lateral axons from CA3 form a homogeneous pathway that is The number of these oblique branches varies from 9 to 30,
easily activated to study synaptic transmission and plasticity. with a mean of 17 (Bannister and Larkman, 1995a; Pyapali et
Studies of CA1 are more numerous than adjacent CA3 al., 1998) (Fig. 51C). Oblique dendrites branch no more than
because it is generally easier to keep cells in this region alive a few times, with a typical branch bifurcating just once at a
and healthy in slice preparations. Recently, CA1 pyramidal location close to its origin from the apical trunk. Despite their
neurons have been the focus of several studies of dendritic limited branching, however, oblique dendrites constitute most
integration because of the large primary apical dendrite, from of the dendritic length in the stratum radiatum (Bannister
which dendritic patch-clamp recordings can be obtained rou- and Larkman, 1995a; Megias et al., 2001). After the primary
tinely. These factors have contributed to tremendous advances apical trunk enters the stratum lacunosum-moleculare the
in understanding synaptic transmission, integration, and plas- apical dendrites continue to branch, forming a structure
ticity in the CNS. referred to as the apical tuft, which has an average of about 15
terminal branches (Bannister and Larkman, 1995a; Trommald
5.2.1 Dendritic Morphology et al., 1995).
Emerging from the base of the pyramidal soma are two to
Two elaborately branching dendritic trees emerge from the eight dendrites (a mean of ve). Most of these dendrites
pyramid-shaped soma of CA1 neurons (Fig. 51A). The basal branch several times (maximum 15 branch points), forming a
dendrites occupy the stratum oriens, and the apical dendrites basal dendritic tree with about 40 terminal segments
occupy the stratum radiatum (proximal apical) and stratum (Bannister and Larkman, 1995; Pyapali et al., 1998). Most
lacunosum-moleculare (distal apical). [Note: in this context branches in the basal dendrites occur rather close to the soma,
the terms proximal and distal refer to the relative distance so the terminal segments are quite long, constituting about
from the soma and should not be confused with the proximo- 80% of the total dendritic length (Bannister and Larkman,
distal axis of the hippocampal circuit (dentate-CA3-CA1- 1995a; Trommald et al., 1995). CA1 neurons differ consider-
subiculum-entorhinal cortex) introduced in Chapter 3.] Both ably in the position of the cell body, with some cells located in
the apical and basal dendritic trees occupy a roughly conical the stratum oriens as far as 100 m from the border between
(sometimes ovoid) volume (Pyapali et al., 1998). The size of the stratum pyramidale and the stratum radiatum. Cells with
the CA1 dendritic tree depends on species and age, but the best somata farther from this border tend to have more terminal
available data are from adult rats. The distance from the stra- dendrites in the stratum oriens (basal dendrites) and fewer
tum pyramidale to the hippocampal ssure is about 600 m, terminal dendrites in the stratum radiatum (apical oblique
and the distance from the stratum pyramidale to the alveus is dendrites) (Bannister and Larkman, 1995a).
about 300 m, yielding a distance of just under 1 mm from Many of the organelles found in the cell body extend into
the tips of the basal dendrites to the tips of the apical den- the proximal apical dendrites of CA1 neurons (Harris and
drites. The combined length of all CA1 dendritic branches is Stevens, 1989; Spacek and Harris, 1997). These include struc-
12.0 to 13.5 mm: basal dendrites contribute about 36% of the tures such as the smooth endoplasmic reticulum (SER) and
total length, apical dendrites in the stratum radiatum con- the Golgi apparatus. At greater distances from the soma, how-
tribute about 40%, and apical dendrites in the stratum lacuno- ever, a more limited set of organelles is found in dendrites.
sum-moleculare contribute the remaining 24% (Bannister Microtubules, neurolaments, and actin are prominent and
and Larkman, 1995a; Ishizuka et al., 1995; Trommald et al., serve transport and motility functions, and they are likely to
1995; Megias et al., 2001). CA1 dendites are studded with be important in synaptic plasticity. A network of SER is also
spines (Fig. 51B), the sites of most excitatory synaptic con- present in dendrites, where it is likely to serve important func-
tacts throughout the dendritic tree (Ramon y Cajal, 1904; tions in calcium buffering and release. The SER forms a con-
Lorente de No, 1934; Bannister and Larkman, 1995b; Megias et tinuous reticulum, which can extend into dendritic spines
al., 2001). (Spacek and Harris, 1997). Mitochondria are also numerous
Studies of dendritic morphology in adult rats suggest that in dendrites and are often associated with the SER, but glyco-
CA1 pyramidal neurons can be broadly classied into two gen granules are largely absent, except in sliced tissue, where
groups on the basis of dendritic morphology. In one group of they probably represent a response to stress (Fiala et al., 2003).
neurons, the primary apical dendrite extends all the way Dendritic mitochondria likely contribute to calcium handling
through the stratum radiatum before bifurcating in the stra- in addition to serving as the primary energy source. In apical
tum lacunosum-moleculare. The other group of CA1 neurons dendrites of CA1 pyramidal neurons, mitochondria ll about
has apical dendrites that bifurcate in the stratum radiatum 2% of the area of the main apical dendrite but 13% of smaller-
(Bannister and Larkman, 1995a). Most other features of these diameter branches (Nafstad and Blackstad, 1966). Sorting
two classes of CA1 neurons are similar, so most structural endosomes and multivesicular bodies form the endosomal
studies consider CA1 pyramidal neurons as a single morpho- pathway, which is responsible for sorting and recycling pro-
logical class (Fig. 51C). teins in dendrites as well as packaging them for transport to
Along the length of the primary apical dendrite, several the soma for degradation. This endosomal compartment
dendritic branches emerge obliquely in the stratum radiatum. forms a network that spans as many as 20 spines in CA1 den-
Hippocampal Neurons: Structure and Function 135

EC layer III

Schaffer s.r.
retrosplenial cortex
perirhinal cortex
s.p. pt lateral septal nucleus
diagonal band of Broca
taenia tecta
s.o. mi medial frontal cortex
collaterals anterior olfactory nucleus
olfactory bulb
axon o ral nucleus accumbens
subiculum mp basal nucleus of amygdala
EC layer V alveus anterior/dorsomedial hypothalamus
parasubiculum fimbria


C CA3 s.l.m.

Figure 51. CA1 dendritic morphology, spines, and synaptic inputs from Bannister and Larkman, 1995a.) B. Three views of a three-
and outputs. A. Camera lucida drawing of a CA1 pyramidal neuron dimensional reconstruction of a stretch of dendrite from the s.r.,
from an adult rat, showing the cell body in the stratum pyramidale illustrating the density of dendritic spines and their diverse struc-
(s.p.), basal dendrites in the stratum oriens (s.o.), and apical den- ture on CA1 pyramidal neuron dendrites. Bar  1 m. (Source:
drites in the stratum radiatum (s.r.) and stratum lacunosum-molec- Adapted from Synapse Web, by Kristen Harris (http://synapses.mcg.
ulare (s.l.m.). The major excitatory inputs in each layer and the edu/anatomy/ca1pyrmd/radiatum/K18/K18.stm). C. Line drawings
major outputs are also indicated. For the mbrial projection, the of several CA1 pyramidal neurons illustrating their diversity of den-
septo-temporal positions noted indicate the source of CA1 cells dritic structure. Dashed line represents the hippocampal ssure
projecting to different target regions. For the alveus projection, the (h.f.). Bar  500 m. (Source: Adapted from Pyapali et al., 1998.)
subiculum is the major target. Bar  100 m. (Source: Adapted
136 The Hippocampus Book

drites (Cooney et al., 2002). Clathrin-coated pits and vesicles receive roughly twice as many inhibitory synapses as excita-
are also present in CA1 dendrites, often near the ends of tubu- tory synapses. The remaining 24% of inhibitory synapses
lar endosomal compartments. Ribosomes are present in CA1 contact the distal basal dendrites. These dendrites are there-
dendrites, where they are usually clustered in the form of fore similar to the apical obliques in terms of diameter,
polyribosomes (Steward et al., 1996; Steward and Worley, spine density, and numbers of excitatory versus inhibitory
2002). synapses.
The dendritic spines studding the surface of CA1 dendrites
5.2.2 Dendritic Spines and Synapses exhibit a broad range of size and morphological complexity
(Fig. 51B). Although quantitative analysis of spine shapes
CA1 pyramidal neurons are covered with about 30,000 den- does not indicate clear groups of spines (Trommald and
dritic spines. Electron microscopic, immunocytochemical, Hulleberg, 1997), several investigators have used a variety of
and physiological analyses have all converged on the conclu- names to describe spines of different shapes (Laatsch and
sion that most dendritic spines receive excitatory synaptic Cowan, 1966; Peters and Kaiserman-Abramof, 1970; Harris
inputs, indicating that spine density can be used as a reason- and Kater, 1994; Sorra and Harris, 2000). Thin spines are long,
able measure of excitatory synapse density (Gray, 1959; narrow protrusions terminating in a small, bulbous head.
Andersen et al., 1966; Megias et al., 2001). The distribution of Sessile spines are long, narrow protrusions that do not termi-
spines on different dendritic domains has been carefully nate in a head. Stubby spines are small protrusions lacking a
quantied (Andersen et al., 1980; Bannister and Larkman, clearly distinguishable neck and a head. Mushroom spines
1995b; Megias et al., 2001). The density of dendritic spines have a narrow neck and a large, bulbous head. Branched
and synapses on CA1 pyramidal neurons is highest in the stra- spines consist of a neck that branches and terminates in two
tum radiatum and stratum oriens but lower in the stratum bulbous heads, each of which receives synaptic input from dif-
lacunosum-moleculare. The soma and the rst 100 m of the ferent axons. These ve types of spines are not specically
main apical dendrite (1.82.5 m diameter) are almost com- localized to any particular region of the CA1 dendritic tree but
pletely devoid of spines. The next 150 m of the apical den- can be found in close apposition on virtually any dendritic
drite (1.62.2 m diameter) has a very low spine density, but branch.
the nal 150 m of the main apical dendrite (1.01.5 m Spine structure is not static but may change in response to
diameter) has a high spine density (about seven spines per lin- neurotransmitter receptor activation or environmental and
ear micrometer). Spine density is lower in the oblique apical hormonal signals (Hering and Sheng 2001; Bonhoeffer and
branches (0.50.6 m diameter; about three spines per Yuste, 2002; Nikonenko et al., 2002; Nimchinsky et al., 2002).
micrometer); but because these dendrites constitute a large Growth of new spines and changes in the structure of existing
fraction of the total dendritic length, about 47% of all CA1 spines are possible substrates of synaptic plasticity in the hip-
spines are located on these branches. Together with the spines pocampus (Geinisman, 2000; Popov et al., 2004). Although
on the main apical dendrite, about 54% of all excitatory the plasticity of spine structure in vivo is a matter of some
synapses contact spines in the apical dendrites in the stratum controversy (Grutzendler et al., 2002; Trachtenberg et al.,
radiatum. Spine density is considerably lower in the apical tuft 2002), imaging studies of hippocampal neurons in vitro have
(0.21.2 m diameter). Although these dendrites contribute revealed a rather dynamic picture of spine morphology
about 20% of the total dendritic length, they contain only (Hosokawa et al., 1995; Dailey and Smith, 1996; Engert and
about 6% of the dendritic spines. Asymmetrical synapses, Bonhoeffer, 1999; Maletic-Savatic et al., 1999; Matsuzaki et al.,
however, are also found on dendritic shafts in this region. If 2004). One particularly noticeable feature of time-lapse
these synapses are excitatory, as is usually presumed, about movies of hippocampal dendrites is the continuous extension
10% of all excitatory synapses are located in the apical tuft. and retraction of lopodia. Occasionally, these lopodia
The rst 30 to 50 m of the basal dendrites (0.50.9 m extend without retracting fully, leading to the hypothesis that
diameter) have a low spine density, whereas the distal basal they form the precursors for mature, stable spines, which
dendrites (0.30.5 m diameter) have a spine density compa- eventually establish functional synaptic connections with
rable to that of the apical oblique dendrites. Accordingly, presynaptic boutons (Dailey and Smith, 1996; Fiala et. al.
about 36% of all excitatory synapses on CA1 neurons contact 1998; Parnass et al., 2000). The motility of dendritic lopodia
spines in the basal dendrites. and spines, which notably lack microtubules and neurola-
The distribution of symmetrical, GABA-positive synapses ments, is facilitated by a network of lamentous actin (Sorra
has also been quantied in CA1 neurons (Megias et al., and Harris, 2000).
2001). About 24% of these synapses contact the soma and CA1 neurons have been the target of numerous studies of
spine-free proximal dendrites (apical and basal). Somewhat the mechanisms of calcium entry, buffering, and extrusion in
surprisingly, the sparsely spiny and densely spiny regions dendritic spines (Yuste and Denk, 1995; Yuste et al., 1999;
of the main apical dendrite (in the stratum radiatum) receive Majewska et al., 2000). Such studies have contributed to
similar numbers of inhibitory synapses, at about 3% of the important advances in our understanding not only of calcium
total each. About 26% of the inhibitory synapses contact handling in spines but the mechanisms of synaptic transmis-
oblique apical branches. About 20% of all inhibitory synapses sion at single synapses and a variety of forms of morphologi-
contact dendrites of the apical tuft. These dendrites therefore cal and functional plasticity (Emptage et al., 1999; Matsuzaki
Hippocampal Neurons: Structure and Function 137

et al., 2001; Nimchinsky et al., 2002; Oertner et al., 2002; meability (Verdoorn et al., 1991; Burnashev et al., 1992;
Sabatini et al., 2002; Yasuda et al., 2003; Matsuzaki et al., 2004; Vissavajjhala et al., 1996; Wenthold et al., 1996). mGluRs are
Nimchinsky et al., 2004). located predominantly on the periphery of the PSD (Lujan et
Spines also contain numerous organelles, including al., 1996). These receptors are coupled directly to the SER via
smooth endoplasmic reticulum (SER). SER is found in about a molecule called Homer, which may be coupled to mecha-
half of the spines in CA1 but is present in most of the largest, nisms for releasing calcium from the SER (Xiao et al., 2000).
morphologically complex spines (Spacek and Harris, 1997; Individual excitatory synapses on CA1 neurons vary con-
Cooney et. al. 2002). SER is occasionally associated with a siderably in their expression of AMPA and NMDA receptors,
spine apparatus, which consists of stacks of SER associated even within particular dendritic domains. Indeed, silent
with other electron-dense material including polyribosomes. synapses, which are thought to contain NMDA receptors
CA1 spines also contain free polyribosomes and mRNA, lead- but few or no functional AMPA receptors, were rst discov-
ing to the hypothesis that proteins can be synthesized on ered and most extensively studied at SC synapses on CA1
demand in individual dendritic spines (Steward et al., 1996). neurons (Isaac et al., 1995; Liao et al., 1995; Isaac, 2003).
Endosomal organelles and coated vesicles are found in about Immunocytochemical analysis suggests that whereas the
one-third of CA1 spines (Cooney et al., 2002). Interestingly, number of NMDA receptors is relatively invariant, a tremen-
mitochondria are absent from most spines (Sorra and Harris, dous range exists in the number of AMPA receptors at indi-
2000). The presence of a large number of molecules and vidual synapses (Nusser et al., 1998; Racca et al., 2000).
organelles in spines, together with the separation that the Quantitative immunogold data suggest a correlation between
spine neck provides from the dendritic shaft and other spines, synapse size and AMPA receptor number (Nusser et al., 1998;
has led to the hypothesis that spines function as isolated Takumi et al., 1999; Ganeshina et al., 2004a,b) and recent
molecular compartments (Wickens, 1988; Koch and Zador, physiological data indicate that there are correlations between
1993; Harris and Kater, 1994; Sorra and Harris, 2000) and that spine morphology and glutamate receptor distribution, with
such compartmentalization is necessary for synapse-specic mushroom-shaped spines containing the most AMPA recep-
changes in synaptic strength (Wickens, 1988; Harris and tors and thinner spines and lopodia containing only NMDA
Kater, 1994). receptors (Matsuzaki et al., 2001).
A prominent feature of almost all CA1 spines (and spines
throughout the nervous system) is a postsynaptic density 5.2.3 Excitatory and Inhibitory Synaptic Inputs
(PSD), an electron-dense thickening of the postsynaptic
membrane. The PSD is located adjacent to the presynaptic Like most neurons in the CNS, CA1 neurons receive input
bouton(s) associated with the spine. Structurally, the size and from both excitatory and inhibitory presynaptic neurons. The
shape of the PSD denes two classes of axo-spinous synapses: principle excitatory inputs arrive from the entorhinal cortex
perforated synapses, which have large PSDs with complex (EC) and CA3 pyramidal neurons. Direct inputs from layer III
shapes, and nonperforated synapses, which have smaller, disk- pyramidal neurons in the EC project to CA1 neurons via the
like PSDs (Harris et al., 1992). Functionally, the PSD is a bio- perforant path (PP), so named because the bers leaving the
chemical specialization that allows numerous molecules (e.g., angular bundle perforate the subiculum (Cajal, 1911). The PP
receptors, kinases, cytoskeletal elements) to be associated in a input from the EC to CA1 (also referred to as the temporo-
structured array at the synapse (Sheng, 2001). Perforated ammonic path) selectively innervates the distal apical den-
PSDs on CA1 pyramidal neurons have more -amino- drites in the stratum lacunosum-moleculare (Blackstad,
3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA) and N- 1958). Some of the PP bers forming synapses on CA1
methyl-D-aspartate (NMDA) receptors than their nonperfo- pyramidal neurons reach their targets in the stratum lacuno-
rated counterparts, suggesting that they may constitute a sum-moleculare via the temporo-alvear pathway (Deller et al.,
population of relatively powerful synapses (Ganeshina et al., 1996a). Additional inputs from the nucleus reuniens of the
2004a,b). thalamus and the basolateral nucleus of the amygdala also
A picture is now emerging concerning the organization of innervate CA1 neurons via synapses on the distal apical den-
the glutamate receptors in the PSD found in excitatory drites (Krettek and Price, 1977; Amaral and Witter, 1995;
synapses on CA1 spines (see Chapters 6 and 7). These synapses Dolleman-Van der Weel and Witter, 1996, 1997; Kemppainen
contain three types of glutamate receptors: NMDA, AMPA, et al., 2002). Inputs from CA3 pyramidal neurons on both
and metabotropic glutamate (mGluR). NMDA receptors, sides of the brain form the Schaffer collateral/commissural
which mediate a slow synaptic current blocked in a voltage- system (SC), which form synapses on the apical dendrites in
dependent manner by Mg2 (see Chapter 6), occupy a disk- the stratum radiatum and on the basal dendrites in the stra-
like space near the center of the PSD. AMPA receptors, which tum oriens (Schaffer, 1892; Blackstad, 1956; Storm-Mathisen
mediate a fast synaptic current, are distributed more evenly and Fonnum, 1972; Hjort-Simonsen, 1973).
throughout the PSD (Lujan et al., 1996; Racca et al., 2000). The A few notable differences between PP and SC synaptic
presence of the edited form of the GluR2 subunit, which is inputs have been identied. First, the PP synapses are located
expressed abundantly in CA1 neurons under normal condi- farther from the soma. Without a mechanism for compensat-
tions, renders most of these AMPA receptors impermeable to ing for this dendritic disadvantage, these synapses would be
Ca2, whereas NMDA receptors have a considerable Ca2 per- expected to have a weaker inuence on action-potential initi-
138 The Hippocampus Book

ation in the axon than the SC synapses (see Sections 5.2.8 and synapses (Otmakhova and Lisman, 1999). Similarly, the nora-
5.2.9). A second difference is that, relative to the stratum drenergic bers from the locus coeruleus project preferentially
radiatum, a greater proportion of synapses in the stratum to the stratum lacunosum-moleculare region of CA1
lacunosum-moleculare are formed on dendritic shafts, rather (Pasquier and Reinoso-Suarez, 1978).
than spines (Megias et al., 2001). Thus, any functions nor-
mally attributed to spines (e.g., biochemical compartmental- 5.2.4 Axon Morphology and Synaptic Targets
ization) must be absent at many of these synapses (but see
Goldberg et al., 2003a). The functional signicance of the A single axon emanates from the pyramidal soma of CA1
shaft synapses in the stratum lacunosum-moleculare is not pyramidal neurons and projects through the stratum oriens
known, but approximately equal numbers of these synapses and into the alveus. CA1 axons branch extensively, forming
are formed by the perforant path and thalamic nucleus collaterals with several targets, both within and beyond the
reuniens projections (Wouterlood et al., 1990). Another pos- hippocampus. In the CA1 subeld, CA1 axons form a limited
sible difference between PP and SC synapses is that there arbor restricted primarily to the stratum oriens; very few col-
appears to be more NMDA receptor activation at stratum laterals enter the stratum radiatum (Amaral et al., 1991).
lacunosum-moleculare synapses (Otmakhova and Lisman, Anatomical and physiological analyses indicate that CA1 neu-
1999), although immunogold EM studies do not reveal rons show a remarkably low connection probability of about
increased NMDA receptor density in the stratum lacunosum- 1% (Knowles and Schwartzkroin, 1981; Deuchars and
moleculare (Nicholson et al., 2006). Finally, PP synapses are Thomson, 1996). Thus, unlike CA3 pyramidal neurons, CA1
inhibited by dopamine, serotonin, and noradrenaline (norep- cells do not make many connections among themselves,
inephrine) to a much greater degree than SC synapses except in the developing hippocampus (Tamamaki et al.,
(Otmakhova and Lisman, 2000). 1987; Amaral et al., 1991; Aniksztejn et al., 2001). By contrast,
Numerous inhibitory neurons also target CA1 pyramidal CA1-interneuron connectivity is much higher, and the
neurons. Some of these interneurons target the soma and strength of excitatory postsynaptic potentials (EPSPs) on
axon, and others target the dendrites. This selective targeting interneurons is powerful (Gulys et al., 1993a; Ali et al., 1998;
suggests that the PP and SC synapses are under differential Csicsvari et al., 1998; Marshall et al., 2002). Because the CA1
control of different populations of interneurons. For example, axon does not enter the stratum radiatum, local synaptic con-
the oriens-alveus/lacunosum-moleculare (O-LM) interneu- nections onto other CA1 neurons occur on basal dendrites.
rons project a single axon to the stratum lacunosum-molecu- Interneurons outside the stratum oriens are not signicantly
lare region of CA1, where the axon collateralizes extensively contacted by CA1 axons.
and forms synapses on the same dendritic branches as the PP The most signicant intrahippocampal connection of CA1
synapses from entorhinal cortex. (For more details on neurons is to pyramidal neurons in the subiculum (Tamamaki
interneuron targeting, see Section 5.9 and Chapter 8.) Despite et al., 1987; Tamamaki and Nojyo, 1990; Amaral et al., 1991).
the differential targeting of the various interneurons in CA1, a This connection is likely to be especially important because
common feature is that they all use -aminobutyric acid the subiculum forms a powerful output of the hippocampus
(GABA) as the principal neurotransmitter. GABA activates (see Section 5.4). CA1 axons collateralize extensively in the
GABAA (ionotropic Cl
channels) and GABAB receptors (G subiculum but form a topographical projection. CA1 neurons
protein receptors coupled to K channel activation). GABAA closest to the subiculum contact their nearest neighbors in the
receptor function, however, is inhomogeneous, with more proximal subiculum, whereas CA1 neurons farther from the
rapid kinetics associated with proximal GABAA receptors and subiculum project to the most distal regions of the subiculum
slower kinetics associated with more distal GABAA receptors (see Chapter 3). Collaterals from individual CA1 axons form a
(Pearce, 1993). column extending the full height of the subiculum (500 m in
CA1 pyramidal neurons also receive neuromodulatory the rat) occupying about one-third the width of the subicu-
inputs from a number of subcortical nuclei. A large input lum (250300 m) and extending about 2 mm along the
arriving from the septum contains cholinergic afferents septo-temporal length of the hippocampus (Tamamaki et al.,
(Swanson et al., 1987). Projections from the locus coeruleus 1987; Tamamaki and Nojyo, 1990).
contain noradrenergic inputs; the raphe nuclei projections The extrahippocampal projections of CA1 neurons via the
contain serotonergic inputs; and the ventral tegmental area mbria depend on their position along the septo-temporal
sends dopaminergic afferents to the hippocampus (Storm- axis (Fig. 51A). Septal CA1 neurons project to retrosplenial
Mathisen, 1977). CA1 pyramidal neurons express numerous and perirhinal cortex as well as to the lateral septal nucleus and
receptor subtypes for each of these neuromodulators, but the diagonal band of Broca. Midsepto-temporal CA1 neurons
their distribution is not always uniform throughout various project to the taenia tecta and medial frontal cortex. Temporal
subcellular domains. For example, dopamine receptors are CA1 neurons project to the taenia tecta, medial frontal cortex,
localized preferentially in the stratum lacunosum-moleculare anterior olfactory nucleus, olfactory bulb, nucleus accumbens,
(Swanson et al., 1987; Goldsmith and Joyce, 1994), suggesting basal nucleus of the amygdala, and anterior and dorsomedial
that neuromodulation via this pathway is selectively posi- hypothalamus (Jay et al., 1989; van Groen and Wyss, 1990;
tioned to inuence PP synapses to a greater extent than SC Amaral and Witter, 1995). Importantly, however, not all CA1
Hippocampal Neurons: Structure and Function 139

neurons project to each of these regions. For example, only a contributes to the fast AHP, whereas a slow, Ca2-activated K
subset of CA1 pyramidal neurons, along with a population of conductance contributes to the slow AHP (IAHP, mediated by
giant, nonpyramidal principal neurons, project to the olfac- SK channels). A K conductance reduced by muscarinic
tory bulb (van Groen and Wyss, 1990; Gulys et al., 1998). receptor activation (IM) and IC contribute to the medium AHP
No studies have explicitly quantied the total axonal length (Storm, 1990) (Fig. 52B). The ADP is mediated in part by a
or number of synapses formed by CA1 axon collaterals. More Ca2 tail current mediated by R-type Ca2 channels (Metz et
extensive data are available for the CA3 axon, which is dis- al., 2005) and may also have a contribution from persistent
cussed in Section 5.3.4. Light microscopic studies indicate, Na current (Yue et al., 2005).
however, that the CA1 axon, which has a diameter of less than Action potentials in CA1 pyramidal neurons typically have
1 m, has numerous en passant and terminal synaptic spe- a half-width of about 1 ms (Staff et al., 2000). In mature
cializations along its length (Tamamaki and Nojyo, 1990). The rats, repolarization is slowed by application of either
total number of synaptic connections is likely on the order of tetraethylammonium (TEA) or 4-aminopyridine (4-AP) at
several thousand. concentrations that block the delayed rectier K current, the
inactivating A-type and D-type K currents, and fast voltage-
5.2.5 Resting Potential and Action and Ca2-gated K currents (Storm, 1987; Golding et al.,
Potential Firing Properties 1999). Such results suggest that a variety of voltage-gated and
Ca2-sensitive K currents contribute to action-potential
CA1 pyramidal neurons have resting potentials, recorded in repolarization as well as the various phases of the AHP
slice preparations, in the range of
60 to
70 mV. Similar (Storm, 1990) (Fig. 52B).
values have been recorded using either sharp microelectrodes Longer depolarizing current injections via somatic record-
or patch pipettes (e.g., Storm, 1987; Spruston and Johnston, ing electrodes typically elicit continuous action potential r-
1992; Staff et al., 2000). At birth, resting potentials are 5 to 10 ing at a rate that is roughly proportional to the amount of
mV depolarized from this value but gradually hyperpolarize current injected. These spike trains usually exhibit spike-
to their adult value by 2 to 3 weeks of age (Spiegelman et al., frequency accommodation: a high frequency of action poten-
1992). In response to depolarizing current injections, action tials at the beginning of a step current injection followed by a
potentials typically have a threshold in the range of
40 to gradual reduction in spike frequency later in the current injec-

50 mV (Spruston and Johnston, 1992; Staff et al., 2000; but tion (Madison and Nicoll, 1984) (Fig. 52C1). Spike-fre-
see Fricker et al., 1999). Thus, CA1 neurons must be depolar- quency accommodation is caused by gradual activation of K
ized by about 20 mV before action potentials are triggered conductances such as IM and IAHP (Lancaster and Adams,
in vitro. The resulting action potentials have amplitudes of 1986; Lancaster and Nicoll, 1987). These K currents increase
about 100 mV. cumulatively during the action potential train because they do
The data discussed above are from recordings made in hip- not inactivate and also deactivate slowly, resulting in a larger
pocampal slices. Such measurements provide good estimates AHP and longer times to repolarize back to threshold for
of the intrinsic properties of CA1 neurons, but the values may the next spike. Trains of action potentials are followed by an
be slightly different in an active network. In vivo the ongoing after-hyperpolarization (AHP) lasting more than a second
synaptic activity causes action potentials at a rate of 1 to 10 (Fig. 52C2); and the amplitude of each of these AHPs
Hz, and the average membrane potential between spikes is increases with the number of action potentials in the preced-
sometimes as much as 10 mV depolarized from the in vitro ing train (Hotson and Prince, 1980; Madison and Nicoll,
measurements (Henze and Buzski, 2001). During locomo- 1984) owing to increased Ca2 entry and activation of
tion through a place eld, action potential ring transiently SK-type Ca2-activated K channels (Marrion and Tavalin,
increases to rates of around 8 Hz (theta frequency), with occa- 1998; Bowden et al., 2001).
sional bursts of high-frequency action potential ring ( 100 K currents that are activated below threshold for action
Hz) (Frank et. al. 2001) (see Chapter 11). potentials can also affect spike-ring patterns. The best exam-
Action potential ring patterns in vivo are determined in ple of this is the delay to ring of the rst action potential
part by the timing of synaptic inputs and in part by the intrin- observed with current injections just above threshold. K
sic ring properties of CA1 neurons. The intrinsic ring prop- conductances, such as A-type and D-type K currents, can be
erties of CA1 neurons have been studied extensively in the activated by subthreshold depolarizations, thus keeping the
slice preparation, in which it is relatively easy to obtain record- membrane potential from reaching the action potential
ings and to isolate the effects of intrinsic properties on spike threshold. As these conductances inactivate, however, their
ring due to the low rate of spontaneous background synap- hyperpolarizing inuence is removed, allowing the membrane
tic input. Under these conditions, brief depolarizing current to reach threshold. The delay to rst spike is partly deter-
injections typically elicit a single action potential followed by mined, therefore, by the activation and inactivation rates of
four after-potentials (Storm, 1987, 1990): (1) a fast after- subthreshold K currents (Storm, 1990).
hyperpolarization (AHP); (2) an after-depolarization (ADP); A common feature of action-potential ring patterns in
(3) a medium AHP; and (4) a slow AHP (Fig. 52A). A fast K vivo is bursting (Kandel and Spencer, 1961; Ranck, 1973; Fox
conductance gated by voltage and Ca2 (named IC or BK) and Ranck, 1975; Suzuki and Smith, 1985; Frank et al., 2001),
140 The Hippocampus Book


C1 C2

Figure 52. CA1 spike frequency adaptation and slow after- delayed rectier K current (IK); M current (IM); fast Ca2-depend-
hyperpolarization (AHP). A. Four after-potentials following single ent K current (IC); transient Ca2-dependent K current (ICT);
spikes (S) in CA1 pyramidal neurons: (1) fast AHP (); (2) after- slow Ca2-dependent K current (IAHP); voltage-dependent Cl

depolarization (ADP) (); (3) medium AHP (); and (4) slow AHP current (ICl(V)); hyperpolarization-activated mixed cation current
(). The lower trace is on a slower time scale. Bar  5 mV, 50 ms (IQ  IH). (Source: Adapted from Storm, 1990.) C. Spike-frequency
(top) or 5 mV, 500 ms (bottom). Spikes are truncated. (Source: adaptation and the slow AHP in CA1 pyramidal neurons. C1 shows
Adapted from Storm, 1987.) B. Overview of a variety of voltage- a train of adapting action potentials in response to a step current
and Ca2-dependent currents in hippocampal pyramidal neurons: injection. Bar  40 mV and 2 nA vertical and 40 ms horizontal.
transient Na current (INaT); persistent Na current (INaP); T-type C2 shows the slow AHP corresponding to 1, 4, 5, and 7 action
Ca2 current (ICaT); N-type Ca2 current (ICaN); L-type Ca2 cur- potentials. Bar  5 mV, 1 second. (Source: Adapted from Madison
rent (ICaL); fast transient K current (IA); delay K current (ID); and Nicoll, 1984.

dened broadly as a brief period of high-frequency spiking increase the action potential ring rate. The intrinsic proper-
( 100 Hz) followed by a longer period of inactivity. The rel- ties of CA1 neurons may, however, provide an additional con-
ative contributions of synaptic drive and intrinsic membrane tribution to bursting in vivo.
properties to the burst ring of CA1 neurons in vivo are In vitro studies have identied two types of intrinsic burst-
unclear. Although CA1 cells exhibit some intrinsic bursting ing in CA1 neurons: Low-threshold bursting occurs in
under normal in vitro conditions (Wong and Prince, 1981; response to somatic current injections just above action
Masukawa et al., 1982), it is modest compared to other neu- potential threshold; they are generated when the ADP that fol-
rons, such as pyramidal neurons of subiculum, which burst lows an action potential is large enough to trigger additional
much more robustly (see Section 5.4.4). In the absence of a action potentials (Jensen et al., 1994; Metz et al., 2005; Yue et
strong intrinsic burst mechanism, bursting may occur as a al., 2005). High-threshold bursting occurs in response to
consequence of large synaptic inputs, which could transiently strong dendritic current injection or synaptic activation and is
Hippocampal Neurons: Structure and Function 141

caused by the generation of Ca2 spikes in CA1 dendrites and inhibition of the AHP (Andrade and Nicoll, 1987; Ropert,
(Wong and Prince, 1978; Golding et al., 1999; Magee and 1988). Numerous other modulatory effects have been
Carruth, 1999) (see Section 5.2.11). reported in CA1 neurons, indicating that the resting and
A variety of factors inuence the spike-ring mode of CA1 active properties of these neurons in vivo are likely to depend
pyramidal neurons. Increasing extracellular K, decreasing on behavioral states.
extracellular Ca2, and decreasing osmolarity all enhance
bursting through different mechanisms (Jensen et al., 1994; 5.2.6 Resting Membrane Properties
Azouz et al., 1997; Su et al., 2002). Bursting is also likely to be
a target of modulation by neurotransmitters. For example, One key to understanding neurons at the cellular level is to be
cholinergic activation facilitates the induction of plateau able to predict action potential output in response to a given
potentials, which may be associated with some kinds of burst spatiotemporal pattern of synaptic inputs. In addition to the
ring in vivo (Fraser and MacVicar, 1996). Because so many properties of the synapses themselves, a neurons response to
factors inuence bursting, it is important to determine the a synaptic input depends on three things: its dendritic geom-
mechanism of bursting under conditions mimicking in vivo etry and the location of the synapse(s), its passive membrane
states as closely as possible. Intrinsic burst-ring mechanisms properties, and its active membrane properties. Passive mem-
have also been implicated as targets of epileptogenesis in the brane properties are those that do not depend on membrane
hippocampus (Wong et al., 1986), so understanding the patho- potential (i.e., they are linear). Active membrane properties,
physiology of bursting may lead to new treatments of epilepsy. by contrast, do depend on membrane potential (e.g., the
A number of aspects of pyramidal cell physiology are also voltage-gated Na and K channels that mediate the action
under modulatory control. Activation of muscarinic receptors potential, which are highly nonlinear). These properties are
causes depolarization and a reduction in spike-frequency very much interdependent, but physiologists attempt to treat
accommodation due to inhibition of IM, IAHP, and voltage- them separately, as we do in our discussion here. One justi-
insensitive K conductances (Benardo and Prince, 1982a,b,c; cation for doing this is that detailed neuronal simulations
Madison et al., 1987; Benson et al., 1988). Norepinephrine require a description of the passive membrane properties,
also reduces spike-frequency accommodation due to inhibi- which constitute a foundation on which active membrane
tion of IAHP (Madison and Nicoll, 1982; Pedarzani and Storm, properties are superimposed.
1996). Dopamine causes hyperpolarization and elevated CA1 pyramidal neurons, like most neurons in the brain,
action-potential threshold owing in part to activation of have long and extensively branching dendrites. A quantitative
Ca2-sensitive K conductances; it also raises the action- understanding of how membrane potential changes (such as
potential threshold in CA1 neurons (Benardo and Prince, EPSPs) spread through these structures requires a theoretical
1982d,e; Stanzione et al., 1984). Serotonin has biphasic effects treatment of dendrites combined with experimental measure-
on CA1 neurons, initially causing hyperpolarization owing to ments. Wilfrid Rall provided the theoretical framework with
activation of a K current, but later causing depolarization his seminal work on the cable theory (Box 51). Hippocam-

Box 51
Cable Theory

During the 1950s and 60s, Wilfrid Rall developed a theory for treating the ow of current in
passive dendrites. The seminal articles comprising this theory have been compiled and repub-
lished in book form (Segev et al., 1995). Because the theory was based on a theory similar to
that for trans-Atlantic telegraph cables, it is referred to as the cable theory. With this mathe-
matical treatment, neurons are considered to be long, leaky cables immersed in a conductive
medium (cerebrospinal uid). A synaptic current propagating along a passive dendrite is inu-
enced by three factors: membrane resistance, membrane capacitance, and axial (or intracellu-
lar) resistance. These properties depend in part on geometry (narrower dendrites have higher
membrane resistance, smaller capacitance, and higher axial resistance) and in part on the
composition of the membrane and cytoplasm. In his theory, Rall dened three geometry-
independent variables: membrane resistivity (Rm), specic membrane capacitance (Cm), and
intracellular resistivity (Ri). He also used two useful measures. The rst is called the space
constant (), which is equal to the length, in an innitely long cable, over which the mem-
brane potential decays to 1/e of its original value. The second is the electrotonic length
(L  l/), which is the ratio of the physical length to the space constant, a measure of the
total electrical length of a nite cable.
142 The Hippocampus Book

Box 51
Cable Theory (Continued)

Rall also showed that a class of branching dendritic trees with certain characteristics could be
collapsed to an equivalent cylinder with a characteristic L. He also described methods for esti-
mating L experimentally. Because most neurons deviate from the assumptions necessary to col-
lapse them to a single cable, however, the lasting impact of Ralls cable theory is that it provides
the foundation for modern computational analysis of neurons. Properties such as dendritic
structure and Rm (and possibly Ri, but less so Cm) differ across different types of neurons and
inuence their integrative properties, which can be predicted using computer models.

Box Fig. 51. A. Current owing along a dendrite may ow longitudinally (e.g., toward the
soma), or it may leak out across the dendritic membrane. B. Voltage attenuates with distance
along a dendritic cable in a way that depends on Rm, Ri, and dendritic geometry. Curve a
shows the decay of voltage in a semi-innite cylinder. Curves bd show attenuation in nite,
open-ended cylinders of decreasing length. Curves eg show attenuation in nite, closed-end
cylinders of decreasing length. All curves are for dendritic cylinders with Rm, Ri and diameter
yielding a space constant of 1. (Source: Adapted from Rall, 1959.) C. Dendritic trees with
particular geometries can be collapsed to an equivalent cylinder representation. Two criteria
must be met: At each branch point, the diameter of the parent dendrite, raised to the 3/2
power, must equal the sum of the daughter-dendrite diameters, each raised to the 3/2 power.
In addition, each dendritic branch must terminate at the same electrotonic length.
(Source: Adapted from Rall, 1964.)
Hippocampal Neurons: Structure and Function 143

pal neurons were among the rst to have their cable proper- Another complication when determining Rm is the pres-
ties determined experimentally. Initially, microelectrode ence of voltage-gated channels that are open at the resting
recordings were used to estimate the electrotonic length (L) of membrane potential. Investigators typically seek to minimize
CA1 pyramidal neurons (Brown et al., 1981; Johnston, 1981). the contribution from activation or deactivation of these
These estimates proved to be of limited value, however, channels by making the evoked voltage change small (e.g., 15
because CA1 dendrites clearly violate the conditions required mV). In CA1 pyramidal neurons, however, this problem can-
to collapse them to a single cylinder (Mainen et al., 1996). For not be avoided entirely because of the presence of a powerful
example, basal, apical, and oblique dendritic branches termi- hyperpolarization-activated, nonspecic-cation current (Ih)
nate at different electrotonic distances (Turner, 1984; Pyapali that is active at rest (Halliwell and Adams, 1982; Spruston and
et al., 1998). Furthermore, there are indications that the mem- Johnston, 1992). This conductance introduces a noticeable
brane properties that contribute to L are not uniform over the sag in the voltage response. Hyperpolarizing voltage
surface of the dendritic tree (Golding et al., 2005). For these responses activate Ih, resulting in a gradual return of Vm
reasons, a more accurate picture of the passive behavior of the toward the resting potential during the current injection
CA1 dendritic tree is best achieved by computer simulation (hyperpolarizing sag). Depolarizing voltage responses deac-
methods, which facilitate the characterization based on the tivate Ih, similarly resulting in a gradual return of Vm toward
measured geometry and membrane properties of CA1 pyram- the resting potential (depolarizing sag). The presence of Ih-
idal neurons. mediated sag makes it difficult or impossible to determine m.
The most reliable measures of the parameters affecting To solve this problem, m is usually measured with Ih blocked
cable propertiesRm, Cm, Ricome from patch-clamp by bath application of 2 to 5 mM CsCl or 50 to 100 M
recordings made in hippocampal slices. Earlier estimates of ZD7288. Under these conditions, CA1 pyramidal neurons
these parameters using microelectrodes were affected by the from adult rats and guinea pigs have an RN of about 120 to 150
leak that is introduced by microelectrode penetration M and m of about 35 to 40 ms (Spruston and Johnston,
(Spruston and Johnston, 1992; Staley et al., 1992). Patch- 1992; Staff et al., 2000; Golding et al., 2005) (Fig. 53).
clamp recordings are not without their own problems, namely Assuming a Cm of 1 F/cm2, this corresponds to an Rm of
the possible complications introduced by dialysis of the cyto- about 40,000 cm2 (to understand the units of Rm, it may
plasm. There is good evidence, however, that this problem help to understand that the units of its inverse are conduc-
does not affect passive membrane properties substantially tance per unit area). It is important to realize, however, that
(Spruston and Johnston, 1992; Staff et al., 2000). More serious these values are strongly affected by blocking Ih. In the absence
problems are posed by the voltage-gated channels that are of its blockers, RN is 38% to 54% lower, and the membrane
active near the resting potential, as well as differences between potential changes decay accordingly faster (Spruston and
in vitro and in vivo conditions (see below). Johnston, 1992; Staff et al., 2000; Golding et al., 2005).
Passive membrane properties are typically estimated by In guinea pig CA1 neurons, hyperpolarizations or depolar-
measuring the response of a neuron to a step current injec- izations of about 5 mV from rest result in about a 20%
tion. Two features dene the response: the steady-state ampli- decrease and increase in RN, respectively (Hotson et al., 1979;
tude and the time course of the voltage change. Theoretically, Spruston and Johnston, 1992). Even after block of Ih, estimates
this time course can be described by the sum of several expo- of Rm and RN are sensitive to small changes in the resting
nentials, with the slowest component having a time constant membrane potential, but in the opposite direction of the volt-
equal to the membrane time constant (m), given by the prod- age dependence caused by Ih. These ndings indicate that the
uct RmCm. The steady-state voltage response ( V) is deter- experimental estimates of these parameters are inuenced by
mined by the neurons input resistance (RN) in a way that multiple voltage-gated conductances; therefore, these param-
depends on Rm and geometry (large neurons with many den- eters may never be truly passive. Thus, it may be more appro-
drites and low Rm have the lowest RN). Input resistance can be priate to refer to the resting membrane properties of CA1
measured directly using Ohms law (RN  V/ I), but deter- neurons, which should be regarded as approximations of the
mination of Rm (from m) requires a value of Cm. Because Cm theoretical passive membrane properties.
is largely dependent on the lipid composition of the mem- Despite the voltage dependence of these parameters, esti-
brane, its value has long been assumed to be nearly constant mates of their values near rest are essential because they pro-
at 1 F/cm2, an assumption that was validated by experimen- vide important parameters for computer models of neurons.
tal measurements of capacitance in cells with simple geome- One of the purposes of constructing such models, using
try (Gentet et al., 2000). Estimates of Cm from neurons with methods originally developed by Rall (Segev et al., 1995) and
more complex geometry are often close to this value; in currently incorporated into user-friendly programs such as
instances where substantial deviations are reported, however, NEURON ( and GENESIS (http://www.
it is difficult to know if these reect real variability in Cm or, is to estimate how synaptic
the extreme difficulty of accurately reconstructing the surface potentials attenuate between the dendrites and the soma. Such
area of neurons with branching dendrites studded with estimates, however, are highly dependent on the intracellular
spines. Most estimates of Rm have therefore been derived by resistivity, Ri. This parameter is best estimated using computer
measuring m and assuming a value for Cm of 1 F/cm2. models of carefully reconstructed neurons along with a meas-
144 The Hippocampus Book

A soma and distal dendrites is limited. One of the interesting

consequences of this is that it makes it extremely difficult to
study synaptic or voltage-gated conductances in dendrites
using electrodes placed at the soma. Space-clamp errors are
10 mV likely to be enormous for distal conductances and substantial
100 ms even for relatively proximal conductances (Spruston et al.,
1993, 1994; Major et al., 1994; Mainen et al., 1996). These
errors are greatest for transient conductance changes; and
even when measures of steady-state voltage control (such as
15 reversal potential measurements) indicate reasonable voltage
V(mV) control, errors can be very large for the transient case. The
10 presence of voltage-gated conductances in dendrites, though
presumably advantageous for the normal function of the neu-
I(pA) ron, makes it still more difficult to control dendritic mem-

brane potential using somatic electrodes. Thus, other methods
100 200 are needed to complement somatic recordings to determine

the functional properties of dendrites. Fortunately, several
Control such methods are now available (see Sections 5.2.11 through

10 CsCl 5.2.13).

15 5.2.8 Attenuation of Synaptic
Potentials in CA1 Dendrites
3 mV
100 ms The functional signicance of the dendritic tree, with these
nonuniform membrane properties and high Ri, is perhaps
best understood by considering the extent to which synaptic
potentials attenuate as they propagate from the dendrite to the
soma. Compartmental models constrained by simultaneous
Figure 53. CA1 passive properties and sag. A. Voltage responses to