HANDBOOK OF DEVELOPMENTAL

COGNITIVE NEUROSCIENCE
Developmental Cognitive Neuroscience
Mark Johnson and Bruce Pennington, editors

Neurodevelopmental Disorders, Helen Tager-Flusberg, editor, 1999

Handbook of Developmental Cognitive Neuroscience, Charles A. Nelson

and Monica Luciana, editors, 2001

HANDBOOK OF DEVELOPMENTAL
COGNITIVE NEUROSCIENCE

Edited by

Charles A. Nelson
Monica Luciana

A BRADFORD BOOK
THE MIT PRESS
CAMBRIDGE, MASSACHUSETTS
LONDON, ENGLAND
© 2001 Massachusetts Institute of Technology

All rights reserved. No part of this book may be reproduced

in any form by any electronic or mechanical means (including
photocopying, recording, or information storage and re-
trieval) without permission in writing from the publisher.

This book was set in Baskerville by Impressions Book and Jour-

nal Services, Inc., Madison, Wisconsin, and was printed and
bound in the United States of America.

Library of Congress Cataloging-in-Publication Data

Handbook of developmental cognitive neuroscience / edited

by Charles A. Nelson and Monica Luciana.
p. cm.
"A Bradford book"
Includes bibliographical references and index.
ISBN 0-262-14073-X (alk. paper)
1. Developmental neurobiology—Handbooks, manuals,
etc. 2. Cognitive neuroscience—Handbooks, manuals, etc.
I. Nelson, Charles A. (Charles Alexander) II. Luciana,
Monica. [DNLM: 1. Nervous System—growth & develop-
ment. 2. Nervous System Physiology. 3. Cognition—physi-
ology. 4. Human Development. WL 102 H23535 2001]

QP363.5.H365 2001
612.8'2—dc21 00-046567
CONTENTS

Foreword ix
JOHN T. BRUER
Introduction xi

I. FUNDAMENTALS OF DEVELOPMENTAL
NEUROBIOLOGY 1

1. Neocortical Neuronogenesis: Regulation, Control Points, and a Strategy

of Structural Variation 3
T. TAKAHASHI, R. S. NOWAKOWSKI, AND V. S. CAVINESS, JR.

2. Synaptogenesis in the Neocortex of the Newborn: The Ultimate Frontier

for Individuation? 23
J.-P. BOURGEOIS

3. Myelination in the Developing Human Brain 35

RICARDO C. SAMPAIO AND CHARLES L. TRUWIT

4. Morphological Changes of the Human Hippocampal Formation

from Midgestation to Early Childhood 45
LASZLO SERESS

5. Effects of Sex Hormones on Brain Development 59

JUDY L. CAMERON

6. The Development of Prefrontal Cortex: The Maturation of Neurotransmitter

Systems and Their Interactions 79
FRANCINE M. BENES

7. Adult Neurogenesis in the Hippocampal Formation 93

PATIMA TANAPAT, NICHOLAS B. HASTINGS, AND ELIZABETH GOULD

CONTENTS V
II. METHODOLOGICAL PARADIGMS 107

8. Inferences about the Functional Development of Neural Systems in Children

via the Application of Animal Tests of Cognition 109
WILLIAM H. OVERMAN AND JOCELYNE BACHEVALIER

9. The Use of Event-Related Potentials in the Study of Cognitive

Development 125
CHARLES A. NELSON AND CHRISTOPHER S. MONK

10. Applications of Magnetic Resonance Imaging to the Study of

Development 137
B. J. CASEY, KATHLEEN M. THOMAS, AND BRUCE MCCANDLISS

11. Genetic Methods 149

BRUCE F. PENNINGTON

12. Neural Network Models of Cognitive Development 159

YUKO MUNAKATA AND JENNIFER MfiRVA STEDRON

III. NEURAL PLASTICITY OF DEVELOPMENT 173

13. Early Brain Injury, Plasticity, and Behavior 175

BRYAN KOLB AND ROBBIN GIBB

14. Neural Plasticity and Development 191

THOMAS ELBERT, SABINE HEIM, AND BRIGITTE ROCKSTROH

IV. SENSORY AND SENSORIMOTOR SYSTEMS 203

15. Development, Plasticity, and Learning in the Auditory System 205

RICHARD N. ASLIN AND RUSKIN H. HUNT

16. Brain-Behavior Relationships in Early Visual Development 221

JAMES L. DANNEMILLER

17. Visual Acuity and Spatial Contrast Sensitivity: Normal Development

and Underlying Mechanisms 237
DAPHNE MAURER AND TERRI L. LEWIS

18. Stability and Flexibility in the Acquisition of Skilled Movement 253

MELISSA W. CLEARFIELD AND ESTHER THELEN

VI CONTENTS
V. LANGUAGE 267

19. Speech and Language Processing in Infancy: A Neurocognitive

Approach 269
JANET F. WERKER AND ATHENA VOULOUMANOS

20. Language Development in Children with Unilateral Brain Injury 281

ELIZABETH BATES AND KATHERINE ROE

21. Experience-Dependent Plasticity and the Treatment of Children with Specific

Language Impairment or Dyslexia 309
GAIL C. BEDI

VI. COGNITION 319

22. Attention in Young Infants: A Developmental Psychophysiological

Perspective 321
JOHN E. RICHARDS

23. The Functional Development and Integration of the Dorsal and Ventral Visual
Pathways: A Neurocomputational Approach 339
MARK H.JOHNSON, DENIS MARESCHAL, AND GERGELY CSIBRA

24. Mechanism and Variation in the Development of Attentional

Networks 353
MARY K. ROTHBART AND MICHAEL I. POSNER

25. Neural Bases of Memory Development: Insights from Neuropsychological

Studies in Primates 365
JOCELYNE BACHEVALIER

26. The Neuropsychology of Face Processing during Infancy

and Childhood 381
MICHELLE DE HAAN

27. Spatial Cognitive Development 399

JOAN STILES

28. Bridging the Gap between Cognition and Developmental Neuroscience:

The Example of Number Representation 415
SUSAN CAREY

29. A Model System for Studying the Role of Dopamine in the Prefrontal Cortex
during Early Development in Humans: Early and Continuously Treated
Phenylketonuria 433
ADELE DIAMOND

30. Age-Related Changes in Working Memory and Frontal Lobe Function:

A Review 473
MONICA FABIANI AND EMILY WEE

CONTENTS Vll
VII. NEURODEVELOPMENTAL ASPECTS OF CLINICAL
DISORDERS 489

31. The Role of Nutrition in Cognitive Development 491

MICHAEL K. GEORGIEFF AND RAGHAVENDRA RAO

32. Fetal Alcohol Syndrome and Other Effects of Prenatal Alcohol:

Developmental Cognitive Neuroscience Implications 505
ANN P. STREISSGUTH AND PAUL D. CONNOR

33. The Effects of Cocaine on the Developing Nervous System 519

GREGG D. STAN WOOD AND PAT LEVITT

34. Advances in the Cognitive Neuroscience of Autism 537

SALLY OZONOFF

35. Tics: When Habit-Forming Neural Systems Form Habits of Their Own 549
JAMES F. LECKMAN, BRADLEY S. PETERSON, ROBERT T. SCHULTZ,
AND DONALD J. COHEN

36. Developmental Disorders of Attention 561

CANAN KARATEKIN

37. The Neuropsychology of Schizophrenia and Its Relationship

to the Neurodevelopmental Model 577
BRITA ELVEVAG AND DANIEL R. WEINBERGER

VIII. EMOTION AND COGNITION INTERACTIONS 597

38. Toward a Neurobiology of Attachment 599

MYRON A. HOFER AND REGINA M. SULLIVAN

39. Effects of Early Deprivation: Findings from Orphanage-Reared Infants

and Children 617
MEGAN R. GUNNAR

40. The Biology of Temperament: An Integrative Approach 631

NATHAN A. Fox, HEATHER A. HENDERSON, AND PETER J. MARSHALL

41. Dopamine-Opiate Modulations of Reward-Seeking Behavior: Implications

for the Functional Assessment of Prefrontal Development 647
MONICA LUCIANA

Contributors 663
Index 665

Vlll CONTENTS
FOREWORD

The first Handbook of Cognitive Neuroscience appeared in 1984. According to that

volume's preface, this was only 8 years after George Miller and Michael Gazzaniga
coined the term "cognitive neuroscience" to describe a new research front on the
border between biological and behavioral science. At the time, Miller, Gazzaniga,
and a few like-minded colleagues, saw the need for a new research program that
would use the techniques of both brain science and cognitive science to explore
the biological foundations of human cognition. That first handbook, edited by
Gazzaniga, contained only one chapter on development. For the most part, that
chapter described the methods of cognitive science, what they had revealed about
the human mind, and what, in combination with brain science, they might reveal
about the development of the mind-brain.
Now, Charles Nelson and Monica Luciana have organized and edited the first
Handbook of Developmental Cognitive Neuroscience. Its publication testifies to the rapid
progress developmental cognitive neuroscience has made over the past 16 years.
The handbook more than fulfills the vision that Miller, Gazzaniga, and other early
cognitive neuroscientists had for their new research program. As one can see from
the handbook's sections on sensory and sensorimotor systems, language, executive
function, and cognition, the theories and methods of cognitive science have a cen-
tral role in developmental cognitive neuroscience. Furthermore, cognitive neuro-
science has progressed to a point where scientists can begin asking and answering
fundamental questions about interactions between emotion and cognition. In these
areas, methods of cognitive science allow researchers to conduct careful analyses
and to develop detailed models of the mental processes that guide and regulate
our behavior. Using these models, scientists can ask how the components of those
models might map onto the neural structures and circuits that provide the bio-
logical substrate for cognition.
Brain imaging and recording technologies are among the central methodological
paradigms of cognitive neuroscience. Imaging and recording technologies allow
cognitive neuroscientists to study neural structure/mental function correlations in
normal human brains. Initially recording and imaging techniques, such as PET,
that required administration of radio isotopes allowed scientists to study structure-
function relations only in adult brains. Developmentalists could use adult studies
to frame hypotheses about what might be happening within a child's developing
brain. Recent technological advances, such as functional magnetic resonance im-
aging and multi-electrode brain recording, have yielded methods that can be safely
used to study structure-function relations and their development in children's
brains. These new techniques plus more refined cognitive models account for the
progress and heightened interest in developmental cognitive neuroscience.

FOREWORD IX
 This handbook also illustrates how the brain sciences contribute to cognitive
neuroscience. Neuropsychology has a long history of the study of how brain trauma
affects behavior. The power of studying normal versus impaired brains, both mature
and developing, is evident in the handbook's section on neurodevelopmental as-
pects of clinical disorders.
Basic developmental neurobiology has elucidated developmental phenomena
such as corticogenesis, neurogenesis, and synaptogenesis, as well as the role of
critical and sensitive periods in development. There have also been attempts to
discern how such phenomena relate to human behavior. Combining basic neuro-
biology with systems neuroscience, imaging, and cognitive science will no doubt
expedite our understanding of how changes at the cellular and subcellular levels
affect and are affected by experience. Some of the most recent, and exciting, find-
ings in basic neurobiology establish that the human brain remains malleable and
plastic throughout our lifetimes. One of the great opportunities in developmental
cognitive neuroscience, as this handbook indicates, is to explore what this
new understanding of lifelong, neural plasticity implies for brain and child
development.
Recently, claims about what brain science might mean for parenting and early
childhood have captured the public's imagination. We should encourage the pub-
lic's interest in developmental brain science and applaud attempts to base early
childhood policy and practice on a scientific basis. However, in some instances
public enthusiasm far outstrips our scientific understanding. Too often the mes-
sages broadcast by advocates and the media do not accurately reflect what scientists
currently know about synapses, critical periods, neural plasticity, and how experi-
ence affects the brain. If we wish to base policy and practice on brain science or
cognitive neuroscience, it is imperative that scientists knowledgeable in the relevant
disciplines engage in careful discussion of what their disciplines, alone and collec-
tively, can contribute to better policy and practice and articulate their conclusions
to the interested public. This handbook provides an excellent starting point for
that discussion.
The power, promise, and importance of the attempt to understand the biological
foundations of human cognition, are amply illustrated throughout the handbook's
chapters.

John T. Bruer

X FOREWORD
INTRODUCTION

Hebb's seminal volume, The Organization of Behavior, was instrumental in helping to

organize the field of behavioral neuroscience. From the time that book was pub-
lished in 1949 through the early 1980s, there was tremendous interest in the neural
bases of behavior, particularly cognitive behavior. Admittedly, much of the work in
this field was dominated by animal studies, but that was a product of the methods
available at the time (a point repeatedly stressed to the first author of this volume
by his undergraduate advisors at McGill, when he expressed interest in studying the
relation between brain and behavior in the human child).
This began to change, however, in the mid 1980s as excitement grew over a
noninvasive new tool believed to be capable of peering into the human brain. This
tool, Positron Emission Tomography (PET), exploded into the scientific main-
stream with the seminal publication by Michael Posner and Marcus Raichle and
their colleagues at Washington University. In a paper this group published in 1988
(Posner, Petersen, Fox, and Raichle, 1988), it became clear that it was possible to
ascertain where processes underlying complex thought (in this case, attention)
were taking place in the brain.
Closely monitoring this change was the James S. McDonnell Foundation. Antici-
pating the partial merger of three distinctive fields—cognitive psychology, neuro-
science, and computational science—the Foundation recognized the need to
train scientists in the language of these three fields. To do so, they funded what
became known as the Summer Institute in Cognitive Neuroscience. In some small way
the impetus for this book began over a decade ago, in 1988, in Cambridge, Mas-
sachusetts, when the first author of this volume attended the inaugural Summer
Institute (the agenda for which is reproduced at the end of this Introduction). This
first meeting was brilliantly organized by Michael Posner, Gordon Shepherd, and
Stephen Kosslyn, and graciously hosted by Harvard University. For two full weeks
it brought to an audience of graduate students, postdocs, and junior faculty an
outstanding panel of scientists. Labs were interspersed with lectures, which were
interspersed with hallway conversation, collectively making this a very high caliber
intellectual endeavor.
One element of this meeting that was particularly memorable, however, was what
was not discussed, at least not formally: the study of development. Clearly the work
of Jay McClelland, David Hubel, and many others who presented at the Summer
Institute had implications for development, but development as development re-
ceived little formal attention. An exception was a small breakout session on the
development of memory that Stephen Kosslyn and Michael Posner (who themselves
have long been interested in this topic) asked the author and a number of other
attendees to organize. With this one exception, development received little discus-
sion at the meeting.

INTRODUCTION XI
 Through the late 1980s into the 1990s, the field of cognitive neuroscience took
hold. The Summer Institutes continued and flourished, the Cognitive Neuroscience
Society and the Journal of Cognitive Neuroscience were launched, and psychology de-
partments throughout the United States began to advertise for faculty positions in
this field. However, interest in the neural underpinnings of development occurred
very slowly. There are many reasons for this trend (for discussion, see Nelson and
Bloom, 1997), but suffice to say that change began to occur on three fronts. One
was that through the 1980s the field of developmental neuroscience began to yield
discoveries of profound importance (e.g., the pattern of synaptogenesis; the role
of experience in influencing brain development). A second was that a number of
prominent cognitive neuroscientists who had historically studied mature function
acknowledged the importance of studying development (e.g., in the mid 1990s a
meeting of the Cognitive Neuroscience Society devoted a symposium to develop-
mental cognitive neuroscience). Finally, a new generation of developmental psy-
chologists began to recognize the need for considering the neurobiological
underpinnings of behavioral development (again, see Nelson and Bloom, 1997 for
discussion). Collectively, then, the time was right for the field of developmental cog-
nitive neuroscience to be born (approximately 16 years after Michael Gazzaniga's
first Handbook of Cognitive Neuroscience was published).
This Handbook of Developmental Cognitive Neuroscience represents the distillation of
the best this new field has to offer; it also reflects a number of strong biases by the
editors. One such bias is that the field of developmental cognitive neuroscience
must be grounded in basic neurodevelopmental science, particularly developmen-
tal neurobiology. To this end, the first part of this volume (Fundamentals of De-
velopmental Neurobiology) is devoted to basic studies and principles of neural
development. Here the reader learns about pre- and postnatal neurogenesis, syn-
aptogenesis, and myelination; the effects of sex hormones on brain development;
and about development of the hippocampus and prefrontal cortex in particular
(given the importance of these regions for cognitive development). A second bias
is our emphasis on the importance of methodological advances. Thus, the second
part of the volume (Methodological Paradigms) is devoted to describing methods
that have proved so important in elucidating brain-behavior relations in the con-
text of cognitive development. These methods include behavioral "marker" tasks,
along with event-related potentials (ERPs), functional Magnetic Resonance Imag-
ing (fMRI), and genetic and computational (neural network) modeling.
Over the past several years the area of neural plasticity has received tremendous
attention by both neuroscientists and behavioral scientists. Indeed, the forces that
shape and mold the brain may well represent the "new" developmental psychology,
albeit a more mechanistic and reductionistic version than offered by previous gen-
erations of developmental psychologists and one that emphasizes development
within a lifespan context. That is, the forces that mold the brain's structure and
physiology are now recognized to operate well into adulthood (see Tanapat, Has-
tings, and Gould, chapter 7, this volume). To this end, the third part (Neural Plas-
ticity of Development) is devoted to a discussion of this area, emphasizing both
normative and atypical aspects of development.
Because the development of sensory and sensorimotor systems and language has
played such a prominent role in both neuroscience and in developmental psy-
chology, the fourth (Sensory and Sensorimotor Systems) and fifth (Language) parts
of this volume are devoted to this area. Here we are enlightened about the devel-
opment of the visual and auditory systems, the development of skilled motor move-
ments, and several aspects of language development. Embedded within these
chapters is yet another bias of the editors: the need to juxtapose the study of nor-

xii INTRODUCTION
mative development against that of atypical development, as each mutually informs
the other.
The sixth part of the book (Cognition) reflects the substance of the volume, as
befits a book on cognitive neuroscience. Here we provide chapters on the develop-
ment of attention, memory, face/object recognition, spatial cognition, number
comprehension, and executive functions. Naturally, the work captured by the dis-
tinguished authors of these chapters is framed in a neuroscience context.
The seventh part of this volume (Neurodevelopmental Aspects of Clinical Dis-
orders) returns the reader to the theme of how studies of atypical development can
inform the study of typical development. However, these chapters directly target
studies of atypical populations. We introduce the reader to the importance of nu-
trition on brain-behavioral development, on the risks of prenatal alcohol and co-
caine exposure, on the development of autism, Tourette's syndrome, and
schizophrenia, and on disorders of attention.
The eighth and final part of the book (Emotion and Cognition Interactions)
anticipates what may well represent the next cutting edge area, that of develop-
mental affective neuroscience and the manner in which affective experience may
act to shape cognitive processes. This part begins with a tutorial on the neurobiology
of attachment, and progresses to a discourse on the effects of deprivation on emo-
tional development, on the neurobiology of temperament, and on reward-seeking
behavior.
Overall, this volume possesses the breadth and depth necessary to do justice to
this exciting new scientific field. It does so by drawing on internationally known
experts who have graciously contributed their time and energy to bring the reader
the first complete Handbook of Developmental Cognitive Neuroscience. We hope that you
are as pleased with the result as we are.
Charles A. Nelson
Monica Luciana

REFERENCES
Hebb, D. O., 1949. The Organization of Behavior. New York: Wiley Press.
Nelson, C. A., and F. E. Bloom, 1997. Child development and neuroscience. Child Devel.
68:970-987.
Posner, M. I., S. E. Petersen, P. T. Fox, and M. E. Raichle, 1988. Localization of cognitive
operations in the human brain. Science 240(4859) :1627-1631.

JAMES S. MCDONNELL FOUNDATION SUMMER INSTITUTE IN

COGNITIVE NEUROSCIENCE

13-24 June 1988, Harvard University

Lectures will be in Room 1 of William James Hall, which is located at 33 Kirkland Street (at
the corner of Kirkland and Divinity Avenue) in Cambridge. Laboratories and discussion
groups will meet in smaller rooms (to be announced) within William James Hall. Unless
otherwise noted, lunch and dinner will be open; a booklet describing the many and varied
Harvard Square restaurants will be distributed at the Institute. Coffee and donuts will be
available beginning at 8:15 A.M. in Room 105 of William James Hall (off the lobby), and soft
drinks and cookies will be provided in the laboratory rooms in the afternoon.

Prior reading:

Cognition (2nd ed.). A. L. Glass and K. Holyoak (New York. Random House).
Chapters 1, 2, 4, and 9.

INTRODUCTION Xlll
 Parallel Distributed Processing. D. E. Rumelhart andj. McClelland
(Cambridge, MA: MIT Press).
Chapters 1-4; 16 and 17 optional.
Principles of Behavioral Neurology. M.-M. Mesulam (Philadelphia: F. A. Davis).
Chapters 3, 4, and 7.
Neurobiology (2nd ed.). G. M. Shepherd (New York: Oxford University Press).
Chapters 16, 29, and 30.

WEEK 1: MEMORY
Monday, 13 June

8:15 Registration and parking permits

(in Room 105 of William James Hall)
9:00 Overview
(Michael I. Posner, Stephen M. Kosslyn, and Gordon M. Shepherd)
10:00 A memory system in the monkey (Mortimer Mishkin)
11:30 Lab Toward a computational model of a memory system in the monkey
(Stephen M. Kosslyn and John D. E. Gabrieli)
Lunch
2:00 Hands-on use of computer model of a memory system
3:00 Demo Cellular and synaptic organization of cerebral cortex (Alan Peters)
4:00 Demo Interpreting scans, observing and testing amnesic patients
(William Hirst and Bruce T. Volpe)
5:00 General review and discussion

Tuesday, 14 June

9:00 The anatomy of declarative memory: Focus on hippocampus

(Larry R. Squire)
10:30 The anatomy of a procedural memory: Focus on cerebellum
(Richard F. Thompson)
Lunch
2:00 Demo Circuit organization of hippocampus and cerebellum
(Gordon M. Shepherd)
3:00 Lab Separating declarative and procedural memory in amnesics and
normals (John D. E. Gabrieli)
4:00 Hands-on demonstrations of experimental paradigms used to test
amnesics; small group discussion
5:00 General review and discussion
Dinner
(Evening sessions during the first week are optional)
8:00 Review of study panel on memory (Michael S. Gazzaniga)
Review of study panel on higher cognitive processes
(Terence W. Picton)

Wednesday, 15 June

9:00 Overview of neural computation (Tomaso Poggio)

10:30 Physiology of memory mechanisms (Gary S. Lynch)
Lunch
2:00 Lab Properties of connectionist systems (James L. McClelland)
3:00 Disc Making connections between real neural architectures and neural
networks (Gordon M. Shepherd)
4:00 Hands-on explorations of connectionist models
5:00 General review and discussion

xiv INTRODUCTION
 Dinner

8:00 Review of study panel on motor control (Emilio Bizzi)

Review of study panel on emotion (Stanley Schachter and
Jerome E. Singer)

Thursday, 16 June

9:00 Possible transmitter systems in memory: Psychopharmacology

(Trevor W. Robbins)
10:30 Disorders of memory and cognition in Alzheimer's disease
(Mary Jo Nissen)
12:00 Memory from a single-cell point of view: Discussant (Robert
Desimone)
Lunch

2:00 Demo Neurotransmitters and neuromodulators in cerebral cortex

(Gordon M. Shepherd)
3:00 Lab Connectionist models of memory disorders (Jay G. Rueckl and
John D. E. Gabrieli)
4:00 Hand-on use of connectionist models to explore the bases of memory
disorders
5:00 General review and discussion
Dinner

8:00 Review of study panel on attention (Michael I. Posner)

Friday, 17 June

9:00 Cognitive psychophysiology in the study of memory: Event-related

potentials and event-related magnetic fields as tools for cognitive
science (Emanuel Donchin)
10:30 PET and related approaches to studying memory function
(Marcus E. Raichle)
Lunch

2:00 Demo Lesion method approach to memory systems in humans

(Antonio R. Damasio)
3:00 Disc Summary of basic principles of cortical circuits
(Gordon M. Shepherd)
4:00 Small group discussion of memory
5:00 General review and discussion

Saturday Evening: Clam bake

Sunday Evening: Lobster buffet

WEEK 2: HIGH-LEVEL VISION

Monday, 20June
9:00 Anatomy of the visual system (David C. Van Essen)
10:30 Physiology of the visual system: Single cell recordings
(Robert H. Wurtz)
Lunch
2:00 Demo Functional organization of the visual system (David C. Van Essen)
3:00 Lab Models of the "two cortical visual systems" (Jay G. Rueckl,
Kyle R. Cave, and Stephen M. Kosslyn)
4:00 Analysis of "hidden units" in "two cortical visual systems"
connectionist model
5:00 General review and discussion

INTRODUCTION XV
Tuesday, 21 June
9:00 The anatomy and physiology of the posterior parietal lobe
(Richard A. Andersen)
10:30 The functions of the visual-spatial attention system
(Anne M. Treisman)
Lunch
2:00 Demo A connectionist/control architecture for attention and skill
acquisition: From potential physiology to behavior
(Walter Schneider)
3:00 Lab Studying visual attention: Pop-out, illusory conjunctions
(William Prinzmetal)
4:00 Hands-on demonstrations of visual attention phenomena; small
group discussion
5:00 General review and discussion
Dinner
8:00 Anatomical, physiological, and psychophysical evidence for separate
pathways for form, color, movement and stereo (David F. L. Hubel)

Wednesday, 22 June
9:00 Neurological patients as an approach to studying spatial attention
(M. Marsel Mesulam)
10:30 Cognitive studies of neurological deficits (Michael I. Posner)
Lunch
2:00 Demo Neuropsychological examination of patients (William P. Milberg)
3:00 Demo Anatomical methods for brain studies (M. Marsel Mesulam)
4:00 Lab Hands-on demonstrations: Cerebral laterality experiments (Stephen
M. Kosslyn and John D. E. Gabrieli)
5:00 General review and discussion

Thursday, 23 June
9:00 The visual agnosias (Elizabeth K. Warrington)
10:30 The anomias (Harold Goodglass)
Lunch
3:00 Demo Neural basis of cortical maps (Gordon M. Shepherd and John S.
Kauer)
4:00 Small group discussion
5:00 General review and discussion

Friday, 24 June
9:00 Visual object recognition (Irving Biederman)
10:30 Visual imagery and visual perception (Stephen M. Kosslyn)
Lunch
2:00 Lab A computer simulation of neurological disorders of high-level visual
recognition processes (Stephen M. Kosslyn and Gretchen Wang)
4:00 Hands-on experimentation with computer simulation of high-level
visual recognition syndromes
5:00 General review and discussion
8:00 Banquet
After dinner: Brief summation by Michael I. Posner, Stephen M. Kosslyn, and Gordon M.
Shepherd

xvi INTRODUCTION
I
FUNDAMENTALS OF
DEVELOPMENTAL
NEUROBIOLOGY
This page intentionally left blank
 1 Neocortical Neuronogenesis:
Regulation, Control Points, and
a Strategy of Structural Variation
T. TAKAHASHI, R. S. NOWAKOWSKL
AND V. S. CAVINESS, JR.

ABSTRACT Cognitive and adaptive behaviors both across and ture across and within species (Hahn, Jensen, and
within mammalian species vary in association with variation in Dudek, 1979). That is, the study of the evolutionary and
neocortical structure. Here we present an algorithm of neo- ontogenetic origins of these structural variations is a
cortical histogenetic operation, based upon experiments in
the mouse with an emphasis upon neuronogenesis, that de- cornerstone of cognitive neuroscience.
fines control points at which large-scale revisions may be as- The general structural features of the neocortex are
sociated with speciation and small-scale revisions with the greatly regular across mammalian species. In all species,
structural distinctions within species. We suggest that the mas- neurons with long axons, which are excitatory, group
ter control mechanism for the neuronogenetic sequence oc- broadly into five classes which are arrayed tangentially
curs at the Gl restriction point and that it is at this control
point that both large- and small-scale variations in numeration by class into regular laminar order (figure 1.1 A; table
are coordinated with those of specification of neuronal class 1.1; Cajal, 1952; Lorente de No, 1938; Peters and Jones,
and regional specification. Lower amplitude and regionally 1984). These general classes are, by layer, the small pyr-
diverse modulations, typical of within-species differences, will amids of layer II, the medium-sized pyramids of layer
also arise from modulation of cortical development down- III, the granule cells of layer TV, the large pyramids of
stream from neuronogenesis.
layer V, and the polymorphic neurons of layer VI.
Inhibitory interneurons, a minority population, distrib-
Introduction ute without laminar grouping more or less uniformly
throughout the height of the cortex (Lavdas, Mione,
The neocortex is the central and indispensable "organ" and Parnavelas, 1996; Peters and Jones, 1984). The gen-
of cognitive process. It is by far the largest structure eral laminar structure varies regionally in terms of these
of the brain and is bilaterally represented in the cere- cytologic details as well as the relative numbers and the
bral hemispheres of the mammalian forebrain. It is packing density of the separate neuronal classes. These
bounded at the medial and ventral margin of the hemi- variations have represented the principal criteria for the
sphere by the hippocampal formation and amygdaloid regional parcellation of the neocortex into architec-
nucleus and rostrally and ventrally by piriform cortex, tonic fields that form the basis for understanding the
the diagonal band, and septal structures. To the extent multimodal and hierarchic neocortical representations
that "cognition follows form," the range and quality of of neural and cognitive systems (Brodmann, 1909; Cav-
cognitive processes, as these vary across and within spe- iness, 1975; Zilles, 1990). Thus, the generic neocortical
cies, are correlates of the variations in neocortical struc- map places perceptual (visual, acoustic, somatosensory,
and gustatory representations) in constant positions
relative to each other in the posterior and lateral
T. TAKAHASHI Department of Neurology, Massachusetts
General Hospital, Harvard Medical School, Boston, Massachu-
regions of the hemisphere and motor representations
setts; Department of Pediatrics, Keio University School of anteriorly (figure LIB).
Medicine, Tokyo, Japan. The general similarity in organization of the neocor-
R. s. NOWAKOWSKI Department of Neuroscience and Cell Bi- tex of all mammals stands in marked contrast to the
ology, UMDNJ-Robert Wood Johnson Medical School, Piscat- variable features that differentiate the neocortex from
away, New Jersey.
v. s. CAVINESS, JR. Department of Neurology, Massachusetts species to species. Most notable among these variable
General Hospital, Harvard Medical School, Boston, features is neocortical volume or, more specifically,
Massachusetts. surface area (table 1.2; Haug, 1987; Holloway, 1974;

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL N E U R O N O G E N E S I S 3

 TABLE 1.1
General mammalian system of neocortical specification

Systems Topology by Neuronal Class by

Hierarchical Order Laminar Order
Primary representations1 Layers by class: Polymorphic
V-SS VI, large pyramidal V,
M granular IV, medium and
A-G small pyramidal III/II.
(Primary areas, e.g., VI are
more granular than other
areas.)
Unimodal associative areas
Multimodal associative and
limbic areas
]
The primary visual (V), somatosensory (SS), acoustic (A),
gustatory (G), and motor (M) representations are here posi-
tioned schematically to represent their relative caudal to ros-
tral (left to right) and medial to lateral (top to bottom) spatial
relationships in the cortex.
Adapted from Mesulam (1998).

tion (Krubitzer, 1995; Manger et al., 1998; Morgane,

Glezer, and Jacobs, 1988; Northcutt and Kaas, 1995; Van
Essen, Newsome, and Maunsell, 1984; Van Essen et al.,
1986; Woolsey and van der Loos, 1970) as well as a cor-
responding increase in the number of neurons, which
may be larger and more complex in their form and cir-
cuitry relationships (Northcutt and Kaas, 1995).
The scale of variation in neocortical structure is more
FIGURE 1.1 Generic neocortical format. (A) Laminar archi- muted among normal members of a species and tends
tecture from cutout in B. The mammalian neocortical format to be regional rather than global in its expression (table
comprehends a range of specific neuronal classes, ordered by
class into layers II-VT. (ML: molecular layer.) (B) Regional 1.2). In particular, variation in overall neocortical vol-
variations in cytologic detail and in the relative numbers and ume is of relatively low order, reflecting a general con-
packing density of neurons distinguish architectonic fields, stancy of neuron number and size, which are the
mapped from caudal to rostral as visual, somatosensory, acous- essential determinants of volume (Andrews, Halpern,
tic, and gustatory sensory and motor representations. The and Purves, 1997; Kennedy et al., 1998; Rademacher et
map is represented bilaterally in paired cerebral hemispheres.
al., 1993; Van Essen, Newsome, and Maunsell, 1984; Van
Essen et al., 1986). Whereas the total mean volume of
the neocortex of normal young adults appears to be
Jerison, 1973, 1982, 1987; Zilles et al., 1988). The scal- minimally variant (Kennedy et al., 1998), the surface
ing (or proportionality) of neocortical volume to that area of individual architectonic representations (Rade-
of other brain regions or somatic structures (Schmidt- macher et al., 1993; Stensaas, Eddington, and Dobelle,
Nielsen, 1984) is species-specific and highly variable 1974), the form of specific gyri (Ono, Jubik, and Ab-
with, for example, a 400-fold or more greater surface ernathy, 1990), and the mapping of architectonic rep-
areas between the brain of a human and that of a resentations to specific gyri (Rademacher et al., 1993)
mouse. Moreover, the surface contour is smooth in may be as idiosyncratic as, perhaps, the individual
smaller species but folded in species-specific gyral pat- fingerprint.
terns in species with larger and more complex brains.
Moreover, with the brains of larger, complex species or THE NEOCORTICAL HISTOGENETIC SEQUENCE The
in species with distinctive adaptive specializations, there structure of neocortex emerges in the course of a series
is an increase in the hierarchic complexity of the ar- of histogenetic events common to all mammals (figure
chitectonic map and corresponding systems organiza- 1.2; Sidman and Rakic, 1982). The histogenetic se-

4 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 TABLE 1.2
Variation and regulation of phenotype in the neocortex

Numeration Interspecific Individual

(numbers of cells) Characteristics Characteristics Regulation
Volume Volume Species-typical Regionally variable, Output domain and
(surface area) low order preneuronogenetic
parcellation
Surface Species-typical smooth vs. Regionally variable —
conformation gyriform conformation low-order variation
where general surface in gyral pattern,
pattern among gyren- contour, size
cephalic mammals is
species- typical
Scaling Species-typical scaling re- Regionally variable,
lationships respecting low-order variation
neocortex to other
CNS structures and re-
specting corresponding
neocortical representa-
tions
Architectonic Fields Interspecific Individual
(maps) Characteristics Characteristics Regulation
Cytoarchitectonic Change in both size and Low-order changes Time domain and size
fields number of fields in size but not of TNG
number
Neuronal Subtypes1
(neuron classes Interspecific Individual
and laminae) Characteristics Characteristics Regulation
Specification Laminar order by Species-typical by subclass Invariant
class and sublaminae
Neurons by class Species-typical in parallel Invariant or only Time domain and
size, and with other domains of low-order regional intracortical processes
complexity of size and complexity variation
form and circuitry
relationships
1
We suggest that neuronal form, laminar and sublaminar destinations, and final circuitry affiliations are ultimately directly or
indirectly attributable to properties inherent in specification.

quence is initiated with the generation of neurons, or 1990, 1992; Finlay and Pallas, 1989; Finlay and Slattery,
neuronogenesis, which occurs in a pseudostratified ven- 1983; Spreafico et al., 1995; Thomaidou et al., 1997;
tricular epithelium (PVE; for abbreviations, see table Verney et al., 1999). Surviving neurons form synapses
1.3) at the margins of the forebrain ventricle. Prior to with other neurons, grow, and further differentiate as
the formation of the earliest neurons, the epithelium they become integrated into neural circuitry (Bour-
proliferates exponentially (Rakic, 1988) and is already geois, Jastreboff, and Rakic, 1989; Bourgeois and Rakic,
partitioned regionally with reference to expression do- 1993; Rakic, 1995; Rakic, Bourgeois, and Goldman-
mains of a diversity of transcription factors (Puelles Rakic, 1994). The progression of the sequence, ex-
and Rubenstein, 1993; Shimamura et al., 1995). In all pressed in terms of the origin of successive neocortical
species, once neuronogenesis is initiated, postmitotic layers, is scaled identically to the overall time course of
neurons migrate across the width of the cerebral wall, neuronogenesis in each species. As indicated in figure
where they become sorted by class into layers (Rakic, 1.3, progression through this temporal interval is ex-
1972; Takahashi et al., 1999b). A substantial number are pressed as a percentile of the duration of the interval
then eliminated by histogenetic cell death (Ferrar et al., (Caviness, Takahashi, and Nowakowski, 1995).

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 5

FIGURE 1.2 The principal events of neocortical histogenesis. where they are assembled into neocortex more or less in
Histogenesis of the neocortex is initiated with cell prolifera- inside-to-outside order. Thus, the earliest formed are destined
tion in the pseudostratified ventricular epithelium (PVE). The for layer VT of the cortex while later formed cells will take up
founder population proliferates exponentially through an ex- positions in progressively more superficial layers V-II. Lines
tended series of cycles in the preneuronogenetic phase (pre- (for granular/supragranular layers) and broken lines (for in-
NI) during which no cells exit the cycle (Q = 0, P = 1.0). fragranular layers) schematically indicate cell cycle of neuron
The onset of the neuronogenetic interval (NI) corresponds to origin in approximate relation to laminar destination). At the
CC], the first cycle when Q becomes greater than 0 and post- conclusion of migration, a substantial proportion of young
mitotic cells exit the cycle as young neurons. The cells of the neurons are eliminated by cell death while surviving neurons
PVE then execute a series of integer cycles (CC1-CC11 in grow, differentiate, and become integrated into cortical
mouse). As young neurons of the Qfraction exit their respec- circuitry,
tive cycles, they migrate across the embryonic cerebral wall

These commonalities of sequence and scaling of se- tures that distinguish species. Yet, in order to assure
quence progression suggest that histogenesis is regu- characteristic constancy of structure among individuals
lated by similar mechanisms across species (Caviness, of a species, the execution of these molecular and cel-
Takahashi, and Nowakowski, 1995). Therefore, it can lular biological events must be greatly regular and stably
be inferred from comparative study that regulatory buffered in the face of modest variations in fetal geno-
mechanisms operate at three levels: (1) a general mam- type and developmental conditions that must exist be-
malian program, which assures the generic events of tween individuals (Kennedy et al., 1998). With respect
specification and histogenesis; (2) a species-specific to neocortical development, the nature of variations in
variation of the generic histogenetic program, which histogenetic process associated with species-specific dif-
assures specific structural adaptations; and (3) modu- ferences is not readily tackled directly by experiment,
latory mechanisms within species, which determine On the other hand, the workings of histogenetic pro-
structural variations distinctive to the individual. cess may be elucidated through experiment as it pro-
The execution of the histogenetic sequence must de- ceeds in a given species. Through experiment in a
pend upon a great number of interdependent molec- species, one may construct a specific program or algo-
ular and cellular biological events (Gerhart and rithm of the generic sequence in its detailed operations
Kirschner, 1997). These must vary in systematic ways and with respect to control points that regulate the
across species to provide for the distinct structural fea- whole sequence. In principle, such an algorithm and,

6 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 TABLE 1.3 complexity of cell behaviors and individual cell biologi-
Abbreviations cal steps, may be understood to be globally regulated
by molecular events coordinated at a master control
PVE Pseudostratified ventricular epithelium mechanism within the Gl phase of the cell cycle. For
TNG Transverse neurogenetic gradient
reasons to be described, this master control mechanism
Gl First gap phase of the cell cycle that occurs
between M and S phases could readily operate to determine structural modifi-
G2 Second gap phase of the cell cycle that cations in neocortex introduced with the origin of new
occurs between S and M phases species. Low-level modulations in the operation of this
M Mitosis phase of the cell cycle same control mechanism may also determine the lower
S DNA synthetic phase of the cell cycle order structural variations of the neocortex that distin-
Tc Length of the whole cell cycle
guish individuals of a species. However, other mecha-
Ts Length of the S phase
TGZ + M Length of the G2 and M phases combined nisms and cellular events occuring further downstream
Tc-ccn Length of the whole cell cycle of cell cycle n in the histogenetic sequence will be suggested as addi-
P Proportion of cells that continue to tional determinants of individual variation with species.
proliferate
<2 Proportion of cells that become The pseudostratified ventricular epithelium:
proliferatively quiescent (or that quit the
cell cycle) Architecture and elementary operation
Qn <2 for that cell cycle
k Dimensionless constant that relates Q to cell Neurons arise at locations that are remote from their
cycle number final destinations. They arise from the pseudostratified
OUTn Output of the PVE at any given cell cycle n ventricular epithelium (PVE), a specialized proliferative
OUTTOTAI Total neuron production from the PVE over epithelium that lines the ventricular cavities of the
the entire neuronogenetic interval
embryonic telencephalic pallium (figures 1.2 and 1.4;
PVE,, Size of the PVE at that cell cycle
CCn Cell cycle n Boulder Committee, 1970; His, 1889; Sauer, 1935, 1936).
LE Leading edge of a CC domain The PVE is pseudostratified, a feature that determines
TE Trailing edge of a CC domain two complex features of its operation that are common
to all proliferative pseudostratified epithelia. First, its
cells, attached at the ventricular margin, vary systemat-
ically in their form with progression through the cell
in particular, its control points comprise both a descrip- cycle (figures 1.2 and 1.5; Sauer, 1936, 1937; Takahashi,
tion and a theoretical framework for the developmental Nowakowski, and Caviness, 1995). M phase occurs with
origins of the cortical features common to all mam- the cell in rounded configuration at the ventricular
malian species and for mechanisms that regulate the margin. The cell elongates through Gl phase and en-
global and large-scale structural differences that distin- ters S phase in maximally elongate configuration with
guish neocortex of different species or, on a smaller its nucleus in the outer half of the epithelium. It short-
scale, those that distinguish neocortical structure ens again with progression during G2 (Hayes and
among individuals of a species. Nowakowski, 1999). Second, proliferative activity is
In the ensuing discussion, we construct an algorith- asynchronous. This quality has the critical implication,
mic framework for the neocortical histogenetic se- to which we return later, that proportionate represen-
quence in the mouse. The principal, emphasis is given tation of the sequential phases of the cell cycle will cor-
to neuronogenesis, the initial step in the sequence; but respond to the duration of that phase as a fraction of
this perspective is extended, in a limited fashion, to the the duration of the total cell cycle, Tc (Nowakowski,
events that both precede and follow neuronogenesis. Lewin, and Miller, 1989).
The cell cycle—proceeding regularly in proliferative There are two systematic operational properties of
cells through the first gap (Gl), the DNA synthetic (S), the neocortical PVE that are critical to the integrated
the second gap (G2), and finally the cell division (M) operation of this epithelium. First, neurons arise se-
phases—is the engine that drives neuronogenesis and quentially from any given region of the epithelium
is the cell biological function in which control mecha- more or less in an inside-out order with respect to their
nisms are vested (Caviness, Takahashi, and Nowakowski, ultimate position in the cortex (Bisconte and Marty,
1999a). The algorithmic construction of the operation 1975a,b; Caviness and Sidman, 1973; Fernandez and
of the neuronogenetic process is parameter-based and Bravo, 1974; Hicks and D'Amato, 1968; Luskin and
quantitative, which, we suggest, is essential to any mean- Shatz, 1985; Rakic, 1974; Takahashi et al., 1999b). That
ingful model of histogenetic regulation. The central is, the neurons of neocortical layer VI are the first to
thesis of this review is that neuronogenesis, in all its arise while neurons of successively overlying layers arise

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 7

FIGURE 1.3 Scaling of neuronogenetic process across species. cat, and monkey—as a function of the percentage of the neu-
The time of production of neurons for each layer of the neo- ronogenetic interval in each species that has elapsed at the
cortex is compared in four different species—mouse, rat, time that the cortical layer is produced.

more or less in order. We refer to this origin of class by

sequence as the neuronogenetic sequence (figure 1.2). Sec-
ond, the proliferative process does not proceed syn-
chronously across the epithelium, but rather is initiated
at the far rostrolateral margin of the PVE where the
neocortical PVE is continuous with that of the gangli-
onic eminence (figure 1.4; Bisconte and Marty, 1975b;
Caviness and Sidman, 1973; Fernandez and Bravo,
1974; Hicks and D'Amato, 1968; McSherry, 1984; Mc-
Sherry and Smart, 1986; Miyama et al., 1997; Rakic,
1974; Smart and McSherry, 1982; Smart and Smart,
1982). Neuronogenesis, once initiated, propagates
from rostrolateral to caudomedial PVE. As a conse-
quence, at any time during neuronogenesis, the pro-
gression of the neuronogenetic sequence will be
spatially graded across the PVE, an operational aspect
of the proliferative process in the PVE referred to as the
transverse neurogenetic gradient or TNG (Bayer and Alt-
FIGURE 1.4 Embryonic brain with the PVE drawn in per- man, 1991).
spective. The proliferative process, initiated at the far rostro- Once a young neuron exits the proliferative cycle, it
lateral margin of the PVE, propagates along the principal must migrate across the developing hemispheric wall to
(rostrolateral to caudomedial) axis of the TNG (curved wide the emerging cortex, a journey of thousands of microns
arrow). The neighborhood relationships of proliferative cells
in large mammals (Rakic, 1972; Sidman and Rakic,
within the PVE generally determine the neighborhood rela-
tionships of their postmigratory progeny within the cortex (ar- 1973). Upon arrival, each neuron is further sorted as
rows projecting from the PVE to the hemispheric surface). appropriate to its class into the emerging layers VI-II/

8 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

FIGURE 1.5 Time and output modes of operation of the cell of facilitatory (cdk4 and 6, cyclin D) and inhibitory (p27 and
cycle of the PVE. The shape of cells of the PVE correlates with p21) cell-internal molecules; this balance is modulated by cell-
phase of the cell cycle. Cells undergo mitosis (M) in a rounded external factors (mitogens, antimitogens). The pre-RP interval
configuration at the ventricular margin. Postmitotic cells in of Gl (horizontal bracket), which is fixed in duration, is
Gl phase progressively elongate as the nucleus ascends to ini- thought to be the only phase of the cell cycle where the be-
tiate S phase in the outer half of the PVE. As cells move havior of proliferative cells of the PVE may be modified by
through G2 phases, they shorten and the nucleus descends to such cell-external agents. The post-RP interval is variable, in-
undergo mitosis. The proliferative cell passes through two reg- creasing with successive cell cycles as the determinant of the
ulatory transitions in Gl phase, the restriction point (RP, filled increase in the length of Gl phase (T G1 ) over the NI. Q in-
arrowheads) and the Gl/S phase transition (open arrow- creases concurrently with TG1. The increases in the post-RP
heads) . The decision to leave the cycle as (Q fraction or to interval and Q with progression of neuronogenesis are rep-
remain as P fraction corresponds to whether the cell passes resented in left and right panels.
RP. Such a decision will be determined by the relative balance

Ill of the neocortex (Takahashi et al., 1996, 1999b). It opmentally primitive PVE in the course of the
is significant that the neighborhood relationships of the proliferative process (Miyama et al., 1997; Rakic, 1988).
proliferative population generally determine the neigh- In summary, the advance of the neuronogenetic
borhood relationships of the postmigratory population sequence within all regions of the PVE is ordered tem-
within the cortex (figure 1.4), and this appears to be a porally according to the TNG and spatially by the to-
particularly stringent rule for the cells destined for lay- pographic organization of the PVE. This integrated
ers V and VI (Kornack and Rakic, 1995; Misson et al., operation is the principle through which the two-
1991; Walsh and Cepko, 1992a,b, 1993). This observa- dimensional (rostrocaudal and mediolateral) structure
tion is only one component of the body of evidence of the PVE is "translated" upon what becomes the lam-
suggesting that a protomap of the neocortical architec- inate—i.e., three-dimensional—structure of the neo-
tonic map is in some way represented within the devel- cortex. Our investigations, conducted in the mouse,

T A K A H A S H I , N O W A K O W S K I , AND CAVINESS: NEOCORTICAL NEURONOGENESIS 9

have revealed that integration of these operational do- where TOTAL is the total number of cell cycles that
mains of the PVE is regulated by two parameters and constitute the neuronogenetic interval in a given spe-
that this regulation is tightly constrained. We will ap- cies (Caviness, Takahashi, and Nowakowski, 2000; Taka-
proach the evidence for these assertions from the van- hashi, Nowakowski, and Caviness, 1996). Q reaches 1.0
tage of a larger discussion of the proliferative operation at the completion of the final cycle; that is, the entire
of the epithelium. set of cells corresponding to 2PVE of the final cycle will
exit the VZ as the terminal output of the proliferative
Neuronogenesis: An algorithm based upon process (figure 1.2).
experiments in mouse
PARAMETERS The parameters that satisfy these equa-
EQUATIONS The cell cycle of histogenetic systems op- tions are the numbers of cell cycles that constitute the
erates in two modes. The first mode is the time mode neuronogenetic interval and Q and P for each cycle.
with reference to the duration of the cell cycle (T c ) and These parameters, derived in mouse through experi-
the durations of its successive Gl, S, G2, and M (TG1, ments based upon patterns of labeling with the S phase
Ts, TG2 + M) phases (figures 1.5 and 1.6). The second is marker bromodeoxyuridine (BUdR) or BUdR and tri-
the output mode with reference to the fraction (0 of tiated thymidine (Takahashi, Nowakowski, and Cavi-
postmitotic cells that exits, as opposed to the fraction ness, 1992, 1995), have been found to be identical in
(P) that returns to S phase: P + Q = 1.0 (figures 1.2 widely separated regions of the PVE, suggesting that the
and 1.5). neuronogenetic algorithm applies generally to the ep-
Neurons of the neocortex arise from a founder pop- ithelium. Independently of location in the PVE, the
ulation through a discrete neuronogenetic interval, neuronogenetic interval is 6 days in duration and Tc
corresponding to a sequence of integer cell cycles increases from just over 8 h to approximately 20 h (fig-
(Caviness, Takahashi, and Nowakowski, 1999a). That is, ure 1.6A) with variation no greater than ±5-7% in any
a neuron precursor cannot execute a partial cycle (fig- small area of the PVE at any time (Cai, Hayes, and No-
ure 1.2; Takahashi, Nowakowski, and Caviness, 1995). wakowski, 1997). From the pattern of advance of Tc
The equation for calculation of the size of the PVE de- with each day, we calculate that 11 cell cycles can be
rived from a PVE founder cell population with unit size executed during the 6-day neuronogenetic interval.
of 1 over the course of n cell cycles (PVEn) is There is no systematic change in TG2+ M or Ts (figure
1.6A). That is, the doubling of Tc reflects a near quad-
rupling of TG1. In other words, Gl phase is the only
phase of the cycle whose duration is regulated in the
where Pi is the P fraction at cell cycle i (2 < i< 11 in course of murine neocortical neuronogenesis. The in-
mouse) (Caviness, Takahashi, and Nowakowski, 2000; crement in Q per cycle is found to be low at first. The
Takahashi, Nowakowski, and Caviness, 1996). With each
"pivot point" of the system where the number of post-
passing cell cycle of the neuronogenetic interval, the mitotic cells leaving the cycle is the same as the number
size of the PVE (PVEn) increases when P > 0.5 and de- returning to S phase (P = Q = 0.5) is achieved only
creases when P < 0.5. during the eighth cycle (figure 1.6B). This corresponds
The number of neurons formed during each cycle to a point nearly 75% of the way through the full set of
will be equal to twice the size of the PVE at that cell 11 cycles. The continued advance of the two parame-
cycle (each cell will have two daughter cells) times Qor ters, Q to 1.0 and P to 0, is much more rapid over the
terminal three cycles.
where OUTn is the output of the PVE at any given cell
cycle n, PVEn is the size of the PVE at that cell cycle (as NEURON PRODUCTION As a correlate of the slow initial
given in equation 1.1) and Qn is Q for that cell cycle rise of Q, the initial rate of neuron production per cycle
(Caviness, Takahashi, and Nowakowski, 2000; Taka- is low and the PVE increases approximately 50-fold in
hashi, Nowakowski, and Caviness, 1996). volume over the initial 8 cycles (figure 1.6C) with only
Over the entire neuronogenetic interval the total a 2-fold increase in its height (Caviness, Takahashi, and
neuron production (OUTTOTAL) corresponds to the Nowakowski, 2000; Takahashi, Nowakowski, and Cavi-
sum of neurons produced with each cycle; or, as derived ness, 1996). Thus, the PVE must increase approximately
from equation 1.2: 25 times in area (i.e., approximately 5-fold in each of
its two tangential dimensions) between the beginning
of the neuronogenetic interval (i.e., CC1) and CC8.
With respect to a founder cell present at the outset of

10 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 FIGURE 1.6 Parameters for cell cycle and neuron production
in the PVE. (A) The advance in cell cycle length (ordinate)
as a function of cycle (abscissa) is attributable entirely to a
near quadrupling of the duration of the Gl phase. (B) The
progression of Q as a function of cell cycle is expressed as Q
= 0.009 X CC1'97 in mouse. Note that the "pivot point" where
the number of postmitotic cells leaving the cycle is the same
as the number returning to S phase (P = Q = 0.5, broken
line) is achieved only during the eighth cycle. (C) The volume
of the PVE increases approximately 50-fold over the initial
eight cycles, i.e., until the pivot point (Q = P — 0.5) is reached
(arrow). The rate of neuron production per cycle (Output/
Cycle) is initially low but rises steadily after the pivot point is
passed. The multiplier effect of the proliferative process in
mouse is such that the lineage arising from the average
founder cell, i.e., a single cell proliferating in the PVE at the
outset of the NI, will generate a cumulative progeny (Cumu-
lative Output) of approximately 140 cells.

methods, cell death is not entered as a parameter into

our equations. These estimates of neuron production
based upon parameters measured experimentally in the
murine neocortical PVE accord well with numbers that
have been based on direct counts (Caviness, Takahashi,
and Nowakowski, 2000; Rhee et al., 1998; Tan et al.,
1998; Vaccarino et al., 1999).

Two domains of neuronogenetic operation

In the foregoing sections we introduced the cellular
structure of the PVE and its modes of operation and
provided the parameters that determine its proliferative
behavior. Within this detail we discern two coordinate
domains of regulation, each with its separate and dis-
tinct regulatory principles (Caviness, Takahashi, and
Nowakowski, 1999b). Locally there is the output domain,
which encompasses the parameters that determine
neuronal output from a founder population as the neu-
ronogenetic sequence is enacted in a given region.
Globally there is the temporospatial domain, which en-
compasses those parameters governing the pace at
which the neuronogenetic sequence advances down the
TNG.

OUTPUT DOMAIN Neuronal output from a founder

population is the expression of an iterative sequence
the 11 cell cycle neuronogenetic period, the average through which the ascent of Q determines the output
total cumulative output (OUTTOTAL) is approximately for each respective cycle. Total output is the sum of the
140 cells (from equation 1.3). Of these, approximately output for the set of cycles. That is, only two parameters,
30% are formed through the first 7 cycles, but 70% are the number of cell cycles and the ascent of Qwith each
formed at CC8and afterwards, i.e., after Qhas become cycle, determine the output of the proliferative process
greater than 0.5 (figure 1.6C). Because the rate of cell from a founder population. Time, sensu strictu, is not
death in the PVE, as estimated by reliable methods directly a parameter governing output (Caviness, Taka-
(Thomaidou et al., 1997), is less than 1% per cell cycle hashi, and Nowakowski, 1995; Takahashi, Nowakowski,
and within the standard error of our experimental and Caviness, 1996) because output per cycle is deter-

T A K A H A S H I , N O W A K O W S K I , AND CAVINESS: NEOCORTICAL NEURONOGENESIS 11

mined by the number of mitoses that occur, taken to- (Sauer, 1936; Sauer and Walker, 1959; Takahashi,
gether with Q, and not the number of hours between Nowakowski, and Caviness, 1995), and the proportional
mitoses. representation of cells by phase will correspond to the
The proliferative sequence is initiated from a popu- duration of that phase as a fraction of Tc.
lation that is proliferating exponentially, i.e., Q = 0 and Consider the initial temporospatial progression of Q
P = 1 (figure 1.2) (Caviness, Takahashi, and Nowa- from 0 at the beginning of CCl (figure 1.7). CCl is ini-
kowski, 2000; Rakic, 1988). It terminates when Q tiated in Gl progeny of the founder cycle (designated
reaches 1 and P becomes 0, and hence no cells re-enter as CC0) at origin of the TNG in the rostrolateral PVE
the proliferative populations. In the course of the in- (Caviness, Takahashi, and Nowakowski, 2000). We will
tervening interval, the progeny of the founder popu- refer to the narrow region of the PVE where cells now
lation will execute a mean number of integer cycles operate in CCj as the CC: domain. The leading edge
corresponding to the number of cycles required for Q (LE) of this domain will advance spatially along the
to traverse its obligatory path from Q = 0 to Q = 1.0 principal axis of the TNG at an hourly rate corre-
(Takahashi, Nowakowski, and Caviness, 1996). We have sponding to the rate that cells along the LE exit CC0
seen that the ascent of Qwith each cycle is nonlinear (1/Tc-cco)- After a time equal to TC_CC1 has elapsed, a
such that the relationship between Q and cell cycle at second LE for CC2 will leave the rostrolateral edge of
any moment has the form the PVE, and CC2 cells will become intermixed with CC1
cells; thereafter, a new LE will be formed for each suc-
cessive CC. However, because the time between each LE
where k is a constant corresponding to the proportion-
lengthens as each Tc lengthens, the distance between
ate advance of Q with each cycle (Caviness, Takahashi,
successive LE will increase. The result of this process is
and Nowakowski, 2000; Takahashi, Nowakowski, and
a wave-like sequence of progression of LEs across the
Caviness, 1996, 1997, 1999). The constant, which is di-
PVE (figure 1.7). Between each LE are the cells of two
mensionless, has a value of approximately 0.009 in the
successive CCs; thus each CC is found between the LE
mouse.
of both the preceding and the succeeding LE. At any
Read abstractly, this formulation states a best-fit re-
given time during the neuronogenetic interval there is
lationship between the progression of Qand cell cycle.
an overlapping mosaic pattern of CC domains where
However, when we consider the biology represented by
the borders of each CC domain are continuously shift-
Q more closely, we realize that this formulation carries
ing. The number of CC domains active across the PVE
a substantially deeper significance. That is, the number
at any given time has not been determined experimen-
of cycles that constitute the neuronogenetic sequence
tally; but based on the lengthening of Tc, we have else-
is determined by those molecular biological mecha-
where estimated that this number is approximately 5 at
nisms that determine the probability at each successive
the time early in development when the LE of the CC]
cycle that postmitotic daughter cells will exit the prolif-
domain has traveled the full extent of the principal axis
erative process. In other words, the number of cycles is
and when Tc is short (Caviness, Takahashi, and Nowa-
set by the molecular mechanism that regulates the out-
kowski, 2000). At the termination of the neuronoge-
put mode of the proliferative process.
netic interval when Tc is maximal, we have estimated
TEMPOROSPATIAL DOMAIN Whereas the output do- that there are only about 1.2 CC domains present at the
main is governed by parameters that are without time time that trailing edge (TE) of CC11 leaves the rostro-
dimension, global coordination of the neuronogenetic caudal origin of the TNG.
sequence across the expanse of the PVE is strictly reg- We have elsewhere suggested that the wave-like prop-
ulated in terms of temporospatial order (figure 1.7; Cav- agation of the CC domains across the surface of the PVE
iness, Takahashi, and Nowakowski, 2000; Miyama et al., provides a plausible positional encoding mechanism by
1997). Specifically, the initiation of the neuronogenetic which PVE cells "know" where they are in the TNG
sequence, as it occurs with respect to the principal axis (Caviness, Takahashi, and Nowakowski, 1999a; Miyama
of the TNG (rostrolateral to caudomedial), is strictly et al., 1997). This positional coding could provide
regulated in time. Regulation is monotonic, we suggest, boundary imposition for cytoarchitectural landmarks,
under the control of the parameter Tc. Temporospatial provided that there is a synchronizing signal coupled to
constraints are imposed by the operating properties of the wave action. Specifically, mapping information ap-
the pseudostratified proliferative epithelium in which pears to reside in neurons of layers VI and V (De Carlos
cells proliferate asynchronously. Thus, in all regions of and O'Leary, 1992; Erzurumlu and Jhaveri, 1992; Mol-
the PVE there will be cells in all phases of the cell cycle nar and Blakemore, 1995); and, we suggest, mapping

12 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

FIGURE 1.7 Temporospatial domain of neuronogenetic schematically in a wedge cutaway on the left and with respect
operation. The cell cycle domains (CC1} CC2, and so on) prop- to the full PVE surface on the right. The leading edge of a
agate across the neocortical PVE along the principal axis of given domain corresponds to the trailing edge of the cycle
the transverse neurogenetic gradient (TNG, a thick arrow on that is two cycles ahead. Thus, in this diagram each domain
the "surface" of the PVE, upper panel) as neuronogenesis pro- has the form of a parallelogram, reflecting the idea that at any
ceeds (from left to right in the upper panel, horizontal arrows, point there will be cells representative of two cell cycles. The
continuing to the lower panel, a broken arrow). Approxi- size of a given domain is the distance between the leading
mately 40 h after the onset of neuronogenesis (the lower edge and trailing edge, and the height is proportional to the
panel), cycle domains are distributed in descending order, relative prevalence of the cells of that CC. The size of a given
CCs-CCj, along the principal axis of the TNG, represented domain increases with increasing cell cycle length.

information is encoded as they arise with cycles 1-7. We Alternatively, they might modulate the encoding
envision the following operating properties of such a process within the territory of overlap. The process
mechanism. At any given time there are multiple CC would be repeated as each TE leaves the origin so that
domains spanning the TNG in overlapping fashion (fig- successive waves of parcellation would lead to an in-
ure 1.7). The critical mechanism complementary to do- creasingly atomized mosaic of map units within the
main propagation, we suggest, would be an encoding PVE. We suggest that this mechanism of parcellation
signal that emanates from origin as each LE passes would extend the relatively coarse-grained parcellation
through the origin and then is propagated rapidly (i.e., implicit in the domains of distribution of transcription
traveling across the PVE in a time much less than a factors, already instantiated in the PVE by the outset of
single cell cycle) down the TNG. One possible mecha- the neuronogenetic interval (Puelles and Rubenstein,
nism for the propagation of synchronizing signal is via 1993; Shimamura et al., 1995). To the extent that such
gap junctions between cells of the PVE (Bittman et al., a mechanism might serve to "translate" a CC domain
1997). The borders of a "mapping unit" encoded by a into an architectonic area, there will be linkage between
given CC domain could correspond to its LE and trail- the temporospatial domain (the set of Tcs plus the
ing edge (TE) where they abut the TE and LE of do- TNG) and the cytoarchitectonic subdivisions of the ma-
mains of CC V _ 2 and CCN+2t respectively (figure 1.7). ture neocortex. Such specification of a unit of the

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 13

protomap would also be closely coordinated with the represented by the cyclin-dependent kinases cdk4 and
same molecular events governing the numbers, and per- 6 and their regulatory subunit cyclin D (Massague and
haps also the classes, of neurons that will populate the Polyak, 1995; Roberts et al., 1994; Sherr, 1993, 1994). A
laminae of the corresponding region of the neocortex. second set that works in opposition to the kinases,
blocking passage through the restriction point and
A master integrator thereby forcing the cell to leave the cycle, are the in-
hibitors represented by p27 and p21 (Massague and Po-
The two domains of proliferative process, i.e., the out- lyak, 1995; Sherr and Roberts, 1995). This balance at
put and temporospatial domains, are coordinated sub- the restriction checkpoint is modulated by a heteroge-
routines within the overall complexity of the neous array of cell external factors that are both positive
proliferative activity of the PVE. As we have discussed, (physiologic mitogens) and negative (physiologic an-
each is regulated in its dynamic progression by a single timitogens) with respect to restriction point passage. In
elementary proliferative parameter. In the case of the nonneural cell lines maintained in vitro, the actions of
output domain, the main parameter is k (hence Q), physiologic mitogens include the upregulation of the
which determines for a specified founder population expression and activity of the elements of Cdk4/6-
both the number of cells that become neurons and the cyclin D complexes and the downregulation of the ac-
expansion size of the cortex. In the case of the tempo- tions of the inhibitors (Koff et al., 1993; Massague and
rospatial domain, it is the kinetic parameter Tc that de- Polyak, 1995; Roberts et al., 1994; Sherr and Roberts,
termines the pace of advance of the CC domains across 1995). The physiologic antimitogens acting in Gl phase
the PVE. Tc, in turn, advances systematically with cycle are opposite in effect (Koff et al., 1993; Massague and
as TGl is regulated. In the following discussion, we pro- Polyak, 1995; Roberts et al., 1994; Sherr and Roberts,
pose that both the output and temporospatial domains 1995). A principal consequence of a cell passing the
are coordinated and integrated by a shared integrative restriction point is activation of master transcription
mechanism. factors, in particular E2F (Follette and O'Farrell, 1997;
The Gl restriction point (figure 1.5; Caviness, Taka- Lees and Harlow, 1995; Sherr, 1994), which apparently
hashi, and Nowakowski, 1999a; Pardee, 1989) is an ob- promotes the passage through the Gl/S transition. It
vious candidate for the "master" integrative mechanism remains to future investigations to consider whether
from its functions, which are well established in general other consequences of this transcriptional activity are
cell biology. "Restriction point" is a short-hand desig- those fundamental to histogenetic specification.
nation for a sequence of molecular actions involving Mechanisms that regulate TG1 are less completely ex-
multiple molecular agents (Coats et al., 1996; Pardee, plored. Nevertheless, the restriction point remains the
1989; Zetterberg, Larsson, and Wiman, 1995). These plausible control point in that experimentally induced
molecular actions determine whether a cell in Gl phase variations in Q are closely coordinate with variations in
will leave the cycle or advance to the Gl/S transition rG1. For example, overexpression of cyclin E in diploid
(figure 1.5; Roberts et al., 1994; Sherr, 1993; Sherr and fibroblasts not only drives Qto near 0 (and P toward 1),
Roberts, 1995). Thus, from the perspective of a large but also coordinately decreases TGl (Ohtsubo and Rob-
population of daughter cells entering Gl phase of a erts, 1993; Ohtsubo et al., 1995). We are not aware of
given cell cycle, P corresponds to the probability that complementary studies that have considered whether
the cell will pass the restriction point for that cycle, re- upward pressure on Q is associated with a correspond-
enter S, then divide again. Qfor that cycle corresponds, ing increase in TG1. Nor has it been clarified in terms
therefore, to the probability that the cell will not pass of molecular mechanisms how the Gl phase becomes
the restriction point but instead will leave the PVE and shortened in the models mentioned previously.
migrate to the cortical plate (figure 1.5). The restriction Thus, an extensive body of investigation in nonneural
point is of further interest to mechanisms regulating Q cells establishes the restriction point as an integrative
in that its operation is modulated by cell external influ- mechanism general to proliferative vertebrate cells.
ences that are known to modulate Q and also to mod- Limited observations now relate more specifically to the
ulate other domains of proliferative behavior. restriction point in the proliferative cells that form
Whether or not a cell passes the restriction point will the neocortical PVE. Thus, check points, which arrest
be determined by the relative balance of facilitatory and the cycle of the PVE, operate adaptively within both the
inhibitory actions acting at the restriction point (figure G2 and Gl phases according to patterns consistent with
1.5; Massague and Polyak, 1995; Roberts et al., 1994; other proliferative vertebrate cells (Takahashi et al.,
Sherr, 1993, 1994). One set of molecular operators that 1999a). The principal regulatory operators, including
acts to drive the cell through the restriction point is cdk2, its regulatory subunit cyclin E (indispensable to

14 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

the Gl/S transition), and the cycle progression inhibi- ordinated with these events of specification (Caviness,
tors p27 and p21 (known to operate at the restriction Takahashi, and Nowakowski, 2000).
point of vertebrate cells), also have established activity
in the PVE (Fero et al., 1996; Kiyokawa et al., 1996; Lee Structural variation and the
et al., 1996; Nakayama et al., 1996; Tsai et al., 1993). neuronogenetic algorithm
Moreover, the patterns of expression of these operators
across the neuronogenetic interval are closely in accord As described in our introduction, differences between
with what would be predicted from the progression of species are characterized by variations in the surface
the proliferative process itself (Delalle et al., 1999). Spe- area of the neocortex as well as in the hierarchic com-
cifically, in the course of the early cycles when Q is low plexities of systems organization reflected in the neo-
and TG1 is short, levels of cyclin E expression are max- cortical map and its systems of connections. Within
imum and those of p27 are minimum. Subsequently, as species, by contrast, there is constancy of overall neo-
the pivot point is approached where Q = P = 0.5, the cortical magnitude and in the topology of the systems
expression of cyclin E plummets to low asymptote while map, where interindividual variation is expressed re-
that of p27 surges to its peak levels of expression, de- gionally in terms of the size and configuration of indi-
clining only modestly thereafter as the epithelium in- vidual gyri and in the areal extent of individual
volutes and the neuronogenetic interval ends. In cytoarchitectonic fields. In the following sections, we
addition, when dissociated cells of the PVE are cultured suggest that the global structural variations attendant
with the mitogen FGF, not only is Q driven downward upon speciation reflect the consequences of parameter
toward 0, but also 7G1 is shortened to what is essentially changes driving the regulatory controls of the neuron-
its minimal duration observed in the earliest cycles in ogenetic algorithm. These may be large in scale and
vivo (Cavanagh et al., 1997). The coordinate nature of potentially saltatory in their appearance in support of
these progressions of cycle parameters, Q, P, and rG1, speciation. We suggest that the regionally limited, low-
and the molecular mechanisms of cycle regulation open amplitude structural variations that distinguish individ-
a window upon the fundamental workings of the neu- uals of a species may reflect modulatory influences
ronogenetic algorithm that has been formulated here. acting upon these regulatory controls or, alternatively,
Thus far, in attempting to formulate a neuronoge- influences that modulate the course of later-occurring
netic algorithm our emphasis has been upon neuronal intracortical events. It is our intent to propose a set of
output, how it is regulated within region and through- hypothetical strategic principles offered by this frame-
work; it should be evident that these principles do not
out all regions of the PVE. Another set of critical mech-
constitute a comprehensive system of origin of neocor-
anisms of histogenesis comprises those of specification,
tical structural variation.
specification of cell class, and regional specification
(Caviness, Takahashi, and Nowakowski, 2000). That is,
THE NEURONOGENETIC ALGORITHM AND STRUCTURAL
cells are produced by class in appropriate numbers;
DIFFERENCES ACROSS SPECIES In principle, species-
cells are assigned by class to the separate neocortical specific changes in cortical structure could be pro-
fields in appropriate numbers. In that a fuller consid- duced by changes at any of the control points of the
eration of proliferative mechanisms fundamental to his- neuronogenetic algorithm. Here we emphasize how
togenetic specification would strain the intended scope two important features of neocortical organization,
of this review, we will limit this discussion to the sug- magnitude, and cytoarchitectonic subdivisions, could
gestion that the restriction point may serve an even be affected by variation of controls governing output
more elaborate master control mechanism than we and temporospatial domains, respectively.
have thus far postulated. That is, those mechanisms de- Variations in overall neocortical magnitude are read-
terminant of cell number may at the restriction point ily programmed within the context of the output do-
become coordinate with those of class and regional main of the neuronogenetic algorithm (figure 1.8). The
specification. By "coordinate" we intend that regulatory focal point for such control of neocortical magnitude
mechanisms are interlocking. Their interlocking activ- is encapsulated by the constant k (equation 1.4) which,
ities may be scaled upward or downward (nonpropor- we have noted, is 0.009 in mouse. If k is made smaller
tionately) or uniformly (proportionately) scaled. The than 0.009, more cell cycles would be required to trav-
argument for coordinate regulation is in part circum- erse the pathway from Q = 0 to Q = 1. The conse-
stantial in that the events of specification of both class quence of this change would be a cortex larger than
and region are not only in effect within the proliferative that of mouse, with more neurons as specified by the
epithelium but also that cell numeration is precisely co- changes in Qand the number of cell cycles (using equa-

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 15

FIGURE 1.8 Neuronogenetic determinants of structural vari- is established pre-NI. Variation in the output mode arises from
ation of neocortex. Variations in the three cardinal features changes in the constant k, which determines the relationship
of neocortical structure—size, number of mapping units, and between Qand cell cycle number (equation 1.4). In mouse k
neuronal class—are specified by variations in three develop- is 0.009; neuron number will vary upward or downward from
mental determinants. These are the preneuronogenetic or that of mouse with lower or higher k (more or fewer cycles
pre-NI events (the starting size of the PVE at the outset of during the neuronogenetic interval). The time mode is mea-
neuronogenesis: "Size of founder population, "y-axis), the out- sured as variation in Tc but regulated by the post-restriction
put (k: x-axis), and time (length of Gl phase: TG1 z-axis) point length of Gl. Values for mouse Tc range from about 10
modes of operation of the cell cycle. The coordinate opera- to 20 h, and the size and number of mapping units will be
tion of these three determinants is integrated and illustrated dependent upon variation in Tc (represented in the time
here as the three orthogonal axes of a cube. Each axis is mode) interacting with the size of the founder population
orthogonal to the other two because, we suggest, the three (figures 1.7 and 1.9). An increasing range of specification of
determinants are regulated independently, but adult pheno- neuronal subclasses will correlate with increasing duration of
types are produced by an interaction of two of these three Gl phase, related, we suggest, to an increasing range of post-
determinants. For example, the number of neurons formed, restriction point transcriptional activity secondary to the re-
and hence the actual size of the neocortex, is influenced in- lease of master transcription factors including E2F. The force
teractively by the output mode and the pre-NI events. The of any of the three determinants will reflect a balance of cell
mouse neocortex occupies the center of the cube with a internal and cell external (e.g., mitogens, antimitogens) influ-
"moderate-sized" neocortex; the human neocortex is at one ences (see the text and the legend to figure 1.5).
extreme corner of the cube. Variation in founder cell number

tions 1.1, 1.2, and 1.3). An increase in k greater than internal organs, including the brain. It is pertinent to
0.009 would, via the same logic, produce a cortex the argument that the enlargement occurs as a conse-
smaller than that of mouse. These changes in size would quence of increase in cell number (Fero et al., 1996;
apply to the entire neocortex and do so uniformly. This Kiyokawa et al., 1996; Nakayama et al., 1996). As far as
is because the number of neurons produced from is known, p27 operates only within the cell cycle at the
equivalent founder populations will be identical as the restriction point, and the inference from the knockout
neuronogenetic sequence is enacted sequentially along is that the "computation" governed by p27 is to deter-
the TNG and under the regulation of Q. mine the gating of probability that cells will pass the Gl
Thus, a single change in k, induced at the Gl restric- restriction point. Importantly, the p27-governed com-
tion point, could produce the step changes in neocor- putation can be only partially rate-limiting; otherwise,
tical magnitude that are necessary for speciation. This p27 knockout animals would be infinitely large. Other
is illustrated by the consequences of knockout of the cycle progression inhibitors acting at the restriction
p27 gene in mice, which results in an animal that is point, for example p21, must complement the action of
some 30-40% larger than animals with the wild-type p27. These may include inhibitors that act globally, such
gene. The enlargement in this animal is a more-or-less as p27, but also others which are strictly regional in their
uniformly scaled increase in body size and in (most) jurisdictions. Note that the actions of either restriction

16 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

point promoters or inhibitors on the mechanisms that ber of founder cells, which sets the starting size of the
set k would, in principle, modify equation 1.4 and TNG. If the founder cell population is unchanged, then
thereby regulate the output domain. In this way differ- TNG is a constant size and a change in Tc would change
ential action mediated by a family of inhibitors might the number of mapping units without an effect upon
provide for uniform or nonuniform differential scaling neuronal number. But if, for example, founder cell
of body and brain size, or even of neocortical and non- number is increased, producing an increased TNG, and
neocortical brain structures such as those characteristic this change occurs coordinately with an increase Tc ,
of species diversity (Finlay and Darlington, 1995; Hol- then both neuron number and the numbers of cortical
loway, 1974, 1979; Jerison, 1979; Stephan, Bauchot, and areas would increase while the number of architectonic
Andy, 1970). fields would not increase. Such a mechanism might un-
Species differences also involve variation in the size derlie mapping differences that distinguish the neocor-
of corresponding architectonic representations that are tex across the range of large and small rodents, say,
components of the separate sensory and motor systems mouse to rat to beaver (Killackey, Rhoades, and Ben-
in the neocortex. Such variations would be expected to nett-Clarke, 1995; Krubitzer, 1995; Northcutt and Kaas,
arise from variations in the temporospatial pace of 1995; Welker, 1990).
movement of the CC domains across the principal axis Variations in map unit size, dependent upon Tc-
of the TNG. Thus, an upward shift in Tc would enlarge induced variations in the pace of progression of CC
while a downward shift would reduce the size of corre- domains, would complement species characteristic
sponding map representations (figure 1.9). These variations in the preneuronogenetic size of the neo-
changes in the time domain will interact with the num- cortical PVE. Thus, the number of CC domains to be

FIGURE 1.9 Schema correlating variation in the growth of the (Dl, D2), the PVE would be partitioned into a smaller number
PVE and progression of cell cycle domains. Depending upon of CC domains. With greater pre-NI expansion of the PVE
the number of cycles antedating the onset of neuronogenesis, (A2) giving rise to a larger PVE (B2), the same cell cycle length
the PVE might enlarge moderately (Al) or greatly (A2) prior as that in mouse would result in more CC domains (C2). With
to the onset of neuronogenesis (pre-NI expansion of PVE). greater pre-NI expansion but with longer cell cycle length the
Once neuronogenesis is initiated (Bl, B2) in the far rostro- PVE would be larger PVE but partitioned in the same number
lateral region of the PVE with cell cycle 1 ("CCj begins"), the of CC domains (D2). (The theoretical consequences of other
PVE will become partitioned into cell cycle domains (Cl, C2, possible patterns of variation in Tc, i.e., shorter cell cycle
Dl, D2). In mouse (Cl) it is estimated that five cycle domains length, and preneuronogenetic size of PVE are not illus-
will be established before the domain of CCj invades and re- trated.) Mechanisms postulated here might lead to an in-
places completely the last vestige of the preneuronogenetic creased number of mapping units or, alternatively, assure a
PVE at the caudomedial extremity of the PVE (see also figure common mapping pattern across the range of small to large
1.7). In the specific instances where cell cycle length is longer rodents, for example, from mouse to beaver.

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 17

instantiated simultaneously in the PVE by a series of cell 1995; Rakic, Bourgeois, and Goldman-Rakic, 1994). In
cycles of identical Tc will vary directly with the length the mouse, cell death also proceeds monotonically and
of the preneuronogenetic TNG. Moreover, it is con- uniformly in widely separated neocortical fields. The
ceivable that these two determinants, i.e., the progres- pattern of progression of cell death is fully dissociated
sion of Tc with cell cycle and the expanse of neocortical from both neuronogenetic sequence and the tempo-
PVE (the length of the principal axis of TNG), might rospatial pattern of progression of CC domains down
vary independently, providing for a great range of vari- the principal axis of the TNG (Verney et al., 1999). That
ation in numbers of map representations or in the sizes is, the pace of histogenetic process within the cortex
of corresponding representations across a range of spe- marches to the beat of its own drummer, not to the pace
cies such as rat, cat, monkey, and man (figure 1.9; Kil- set by neuronogenesis. Moreover, the amplitudes of the
lackey, Rhoades, and Bennett-Clarke, 1995; Krubitzer, intracortical histogenetic processes are not only idiosyn-
1995; Northcutt and Kaas, 1995). cratic with respect to cortical region, and even layer
within region, but also are of relatively low amplitude,
THE NEURONOGENETIC ALGORITHM AND STRUCTURAL modulating population numbers by cell death, for ex-
DIFFERENCES WITHIN SPECIES The regional diversity ample, by some 15-35% at most (Ferrar et al., 1990,
of neocortical structure that distinguishes individuals of 1992; Finlay and Pallas, 1989; Finlay and Slattery, 1983;
a species is expressed, for the most part, in terms of low- Spreafico et al., 1995; Thomaidou et al., 1997; Verney
amplitude variation in the size of architectonic repre- et al., 1999). Such patterns of modulation, of variable
sentations or gyral size and conformation projected expression in different species and in special systems,
upon a map of invariant topology. That is, each member are generally appropriate to the character and magni-
of the species will have the identical set of map repre- tude of variations that distinguish the structure of neo-
sentations, but corresponding map representations will cortex among individuals. It should be recognized,
vary in the size of their surface projections and to some further, that the post-proliferative regulatory events
extent will do so independently of variations in the pro- occur in an early time frame of life, which would
jection of other mapping units. Such variation can be favor exposure to adaptive influences imposed by
produced by modulation of the neuronogenetic algo- experience.
rithm via coordinate local variation of both the pace of
ascent of Q and Tc with cell cycle (figure 1.8). Thus, Recapitulation and synthesis
local expansion (contraction) of a "map unit," reflect-
ing a local decrease (or increase) in k, must be matched Our most general hypothesis, then, is that those mech-
with a corresponding increase (reduction) in Tc so that anisms acting to produce the saltatory changes associ-
the boundaries of mapping units are adjusted coordi- ated with speciation are not essentially different with
nately with the variation in their size. respect to the algorithmic control points that modulate
adaptive distinctions among individuals of species (see
Postproliferative regulation of structural variation also Nijhout, 1999). For the former, potentially large in
scale and expressed in terms of variations in both mag-
As neurons are produced in the PVE, they migrate to nitude and topology of mapping, we look to the globally
the cortical plate where they begin an additional set of coherent, uniformly regulated mechanisms that sup-
developmental events that culminates in their assembly port the neuronogenetic algorithm. Within this frame-
into functional systems. Chief among these post- work, we suggest, revisions of computations driven by
proliferative events are synaptogenesis and cell death, just two elementary proliferative parameters—one
which are the principal histogenetic cellular processes from the output domain and one from the temporo-
of the intracortical phase of neocortical histogenesis. It spatial domain—would readily scale the ranges of mag-
is to be anticipated that the workings of synaptogenesis nitudes and topologies presented by comparative study.
and cell death determine the contributions to struc- Such revisions could act over several orders of magni-
tural variation of this phase of neocortical histogenesis. tude to build the neocortical areas associated with small
In primates, synaptogenesis proceeds more or less at the (e.g., mouse) and large (e.g., human) species (Cavi-
same pace and through the same interval in all cerebral ness, Takahashi, and Nowakowski, 1995). Speculatively,
regions in a way that does not suggest a strictly regu- one might imagine that speciation could occur in these
lated and systematic global ordering of the sort seen processes via the introduction of major differences
with neuronogenesis (Bourgeois, Goldman-Rakic, and (e.g., timing, binding constants, tissue-specific expres-
Rakic, 1994; Bourgeois and Rakic, 1993; Granger et al., sion) for the expression of molecular actions crucial for

18 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

controlling events occurring at the "master integrator" BITTMAN, K, D. OWENS, A. KRIEGSTEIN, and J. LoTuRco,
associated with the Gl restriction checkpoint. 1997. Cell coupling and uncoupling in the ventricular zone
of developing neocortex./. Neurosci. 17:7037-7044.
For variations that distinguish the neocortex of indi- BOULDER COMMITTEE, 1970. Embryonic vertebrate nervous
viduals within a species, typically regionally discrete and system: Revised terminology. Anat. Rec. 166:257-262.
of relatively low amplitude, we also consider a low- BOURGEOIS, J.-P., P. S. GOLDMAN-RAKIC, and P. RAKIC, 1994.
amplitude modulation of the same control points of the Synaptogenesis in the prefrontal cortex of rhesus monkeys.
neuronal algorithm. Here, however, we anticipate that Cereb. Cortex 4:7'8-96.
BOURGEOIS, J.-P., P. J. JASTREBOFF, and P. RAKIC, 1989. Syn-
a second source of structural distinctions respecting the
aptogenesis in visual cortex of normal and preterm
neocortex of individuals within species are traceable to monkeys: Evidence for intrinsic regulation of synaptic over-
variations in the course of intracortical histogenetic production. Proc. Natl. Acad. Sci. (USA) 86:4297-4301.
events, synaptogenesis and cell death. Whereas no BOURGEOIS, J.-P., and P. RAKIC, 1993. Changes of synaptic den-
parameter-driven formulation yet characterizes these sity in the primary visual cortex of the macaque monkey
late intracortical histogenetic processes, it is clear that from fetal to adult stage./. Neurosci. 13:2801-2860.
BRODMANN, K., 1909. Vergleichende Lokalisationslehre der Gross-
they operate coordinately with the adaptive process of hirnrinde. Leipzig: Barth.
systems matching and coordinately with large-scale sys- CAI, L., N. HAYES, and R. S. NOWAKOWSKI, 1997. Local ho-
tems assembly as the organism optimizes its neural ap- mogeneity of cell cycle length in developing mouse cortex.
paratus to the specific offerings of situation. In keeping / Neurosci. 17:2079-2087.
with this perspective, it has been found that variance of CAJAL, S. RAMON Y., 1952. Histologie du Systeme Nerveux de
neocortical volumes is determined virtually on a gyrus- I'Homme e des Vertebres. Madrid: Consejo Superior de Inves-
tigaciones Cientificas.
by-gyrus basis, principally by a nonlinear interaction of CAVANAGH,J., M. MIONE, I. PAPPAS, andj. PARNAVELAS, 1997.
individual and gyrus (Kennedy et al., 1998). This gen- Basic fibroblast growth factor prolongs the proliferation of
eral insight is consistent with other investigations, in rat cortical progenitor cells in vitro without altering their
which it was established that the neocortical volumes of cell cycle parameters. Cereb. Cortex 7:293-302.
gyri serving specific operations are enlarged when these CAVINESS, V. S., JR., 1975. Architectonic map of neocortex of
functions are overlearned from early childhood the normal mouse./. Comp. Neurol. 164:247-263.
CAVINESS, V. S., JR., and R. L. SIDMAN, 1973. Time of origin of
(Amunts et al., 1997). By implication, the cortical phase corresponding cell classes in the cerebral cortex of normal
of neocortical histogenesis is geared to serve those and reeler mutant mice: An autoradiographic analysis. /.
mechanisms by which the specific experience of the or- Comp. Neurol. 148:141-152.
ganism "builds the brain" as an adaptive response to this CAVINESS, V. S., JR., T. TAKAHASHI, and R. S. NOWAKOWSKI,
experience as it is acquired (Van der Loos et al., 1978). 1995. Numbers, time and neocortical neuronogenesis: A
general developmental and evolutionary model. Trends Neu-
rosci. 18:379-383.
ACKNOWIJEDGMENTS Supported by NIH grants NS12005 CAVINESS, V. S., JR., T. TAKAHASHI, and R. S. NOWAKOWSKI,
and NS33433, NASA grant NAG2-750 and a grant from Phar-
1999a. The Gl restriction point as critical regulator of neo-
macia-Upjohn Fund for Growth and Development Research.
cortical neuronogenesis./. Neurochem. Res. 24:497-506.
T.T. was supported by a fellowship of The Medical Foundation,
CAVINESS, V. S., JR., T. TAKAHASHI, and R. S. NOWAKOWSKI,
Inc., Charles A. King Trust, Boston, Massachusetts.
1999b. Neocortical neuronogenesis: The Gl restriction
point as master integrative mechanism. In The Newborn
Brain—Scientific Basis and Clinical Application, H. Lager-
REFERENCES crantz and M. Hanson, eds. Cambridge: Cambridge Univer-
ANDREWS, T, S. HALPERN, and D. PURVES, 1997. Correlated sity Press (in press).
size variations in human visual cortex, lateral geniculate nu- CAVINESS, V. S., JR., T. TAKAHASHI, and R. S. NOWAKOWSKI,
cleus, and optic tract./. Neurosti. 17:2859-2868. 2000. Neuronogenesis and the early events of neocortical
AMUNTS, K, G. SCHLAUG, L. JAENCKE, H. STEINMETZ, A. histogenesis. In Development of the Neocortex, A. Goffinet, and
SCHLEICHER, A. DABRiNGHAUS, and K. ZILLES, 1997. Motor P. Rakic, eds. Berlin: Springer Verlag, pp. 107-143.
cortex and hand motor skills: Structural compliance in the COATS, S., W. M. FLANAGAN,J. NOURSE, andj. M. ROBERTS,
human brain. Hum. Brain Map 5:206-215. 1996. Requirement of p27Kipl for restriction point control
BAYER, S. A., andj. ALTMAN, 1991. Neocortical Development. New of fibroblast cell cycle. Science 272:877-880.
York: Raven Press. DE CARLOS, J. A., and D. M. O'LEARY, 1992. Growth and tar-
BISCONTE, J.-C., and R. MARTY, 1975a. Analyse chronoarchi- geting of subplate axons and establishment of major cortical
tectonique du cerveau de rat par radioautographie. I. His- pathways.J Neurosci. 12:1194-1211.
togenese du telencephale./. Hirnforsch. 16:55-74. DELALLE, I., T. TAKAHASHI, R. S. NOWAKOWSKI, L.-H. TSAI,
BISCONTE, J.-C., and R. MARTY, 1975b. Etude quantitative du and V. S. CAVINESS, JR., 1999. Cyclin E-p27 opposition and
marquage radioautographique dans le systeme nerveux du regulation of the Gl phase of the cell cycle in the murine
rat. II. Caracteristiques finales dans le cerveau de 1'animal neocortical PVE: A quantitative analysis of mRNA in situ
adulte. Exp. Brain Res. 22:37-56. hybridization. Cereb. Cortex 9:824-832.

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 19

 ERZURUMLU, R. S., and S. JHAVERI, 1992. Emergence of con- JERISON, H., 1979. The evolution of diversity in brain size. In
nectivity in the embryonic rat parietal cortex. Cereb. Cortex Development and Evolution of Brain Size, M. Hahn, C. Jensen,
2:336-352. and B. Dudek, eds. New York: Academic Press, pp. 30-57.
FERNANDEZ, V., and H. BRAVO, 1974. Autoradiographic study JERISON, H., 1982. Allometry, brain size, cortical surface, and
of the cerebral cortex in the rabbit. Brain Behav. Evol. 9:317- convolutedness. In Primate Brain Evolution: Methods and Con-
332. cepts, E. Armstrong and D. Falk, eds. New York: Plenum
FERO, M., M. RIVKIN, M. TASCH, P. PORTER, C. CAROW, E. Press, pp. 77-84.
FIRPO, K. POLYAK, L.-H. TSAI, V. BROUDY, R. PERLMUTTER, JERISON, H., 1987. Brain size. In Encyclopedia of Neuroscience, G.
K. KAUSHANSKY, and J. ROBERTS, 1996. A syndrome of multi- Adelman, ed. Boston: Birkhauser, pp. 168-170.
organ hyperplasia with features of gigantism, tumorigenesis, KENNEDY, D., N. LANGE, N. MAKRIS, J. BATES, and V. S. CAVI-
and female sterility in p27Kipl-deficient mice. Cell 85:733- NESS.JR., 1998. Gyri of the human neocortex: An MRI-based
744. analysis of volumes and variance. Cereb. Cortex 8:372-385.
FERRAR, I., E. BERNET, E. SORIANO, T. DEL Rio, and M. FON- KILLACKEY, H. P., R. W. RHOADES, and C. A. BENNETT-CLARKE,
SECA, 1990. Naturally occurring cell death in the cerebral 1995. The formation of a cortical somatotopic map. Trends
cortex of the rat and removal of dead cells of transitory Neurosci. 18:402-407.
phagocytes. Neuroscience 39:451-458. KlYOKAWA, H., R. KlNEMAN, K. MANOVA-ToDORAVA, V. SCARES,
FERRAR, I., E. SORIANO, J. A. DEL Rio, S. ALCANTARA, and C. E. HOFFMAN, M. ONO, D. KHANAM, A. HAYDAY, L. FROHMAN,
AULADELL, 1992. Cell death and removal in the cerebral and A. KOFF, 1996. Enhanced growth of mice lacking the
cortex during development. Prog. Neurobiol. 39:1-43. cyclin-dependent kinase inhibitor function of p27 Klpl . Cell
FINLAY, B. L., and R. B. DARLINGTON, 1995. Linked regulari- 85:721-732.
ties in the development and evolution of mammalian KOFF, A., M. OHTSUKI, K. POLYAK, J. M. ROBERTS, andj. MAS-
brains. Science 268:1578-1584. SAGUE, 1993. Negative regulation of Gl progression in
FINLAY, B. L., and S. L. PALLAS, 1989. Control of cell number mammalian cells; inhibition of cyclin E-dependent kinase
in the developing mammalian visual system. Prog. Neurobiol. by TGF-p. Science 260:536-539.
32:207-234. KORNACK, D. R., and P. RAKIC, 1995. Radial and horizontal
deployment of clonally related cells in the primate neocor-
FINLAY, B. L., and M. SLATTERY, 1983. Local differences in the
tex: Relationship to distinct mitotic lineages. Neuron 15:311 -
amount of early cell death in neocortex predict adult local
321.
specializations. Science 219:1349-1351.
KRUBITZER, L., 1995. The organization of neocortex in mam-
FOLLETTE, P., and P. O'FARRELL, 1997. Connecting cell behavior
mals: Are species differences really so different? Trends Neu-
to patterning: Lessons from the cell cycle. Cell 88:309-314.
rosci. 18:408-417.
GERHART, J., and M. KIRSCHNER, 1997. Cell, Embryos, and Evo-
LAVDAS, A., M. C. MIONE, and J. G. PARNAVELAS, 1996. Neu-
lution. London: Blackwell Science.
ronal clones in the cerebral cortex show morphological and
GRANGER, B., F. TEKAIA, A. LE SOURD, P. RAKIC, andJ.-P. BOUR-
neurotransmitter heterogeneity during development. Cereb.
GEOIS, 1995. Tempo of neurogenesis and synaptogenesis in Cortex 6:490-497.
the primate cingulate mesocortex: Comparison with the LEE, M.-H., M. NIKOLIC, C. BAPTISTA, E. LAI, L.-H. TSAI, and
neocortex./. Comp. Neurol. 360:363-376. J. MASS AGUE, 1996. The brain-specific activator p35 allows
HAHN, M., C.JENSEN, and B. DUDEK, 1979. Introduction: To- Cdk5 to escape inhibition by p27Kipl in neurons. Proc. Natl.
ward understanding the brain-behavior relationship. In De- Acad. Sci. (USA) 93:3259-3263.
velopment andEvolution of Brain Size, M. Hahn, C.Jensen, and LEES, E., and E. HARLOW, 1995. Cell cycle progression and cell
B. Dudek, eds. New York: Academic Press, pp. 2-7. growth in mammalian cells: Kinetic aspects of transition
HAUG, H., 1987. Brain sizes, surfaces, and neuronal sizes of events. In Cell Cycle Control, C. Hutchison and D. Glover, eds.
the cortex cerebri: A stereological investigation of man and Oxford: Oxford University Press, pp. 228-263.
his variability and a comparison with some mammals (pri- LORENTE DE No, R., 1938. Cerebral cortex: Architecture, in-
mates, whales, marsupials, insectivores, and one elephant). tracortical connections, motor projections. In Physiology of
Am.J. Anat. 180:126-142. the Nervous System ,J. F. Fulton, ed. London: Oxford Univer-
HAYES, N., and R. NOWAKOWSKI, 2000. Exploiting the dynam- sity Press, pp. 274-313.
ics of S-phase tracers in developing brain: Interkinetic nu- LUSKIN, M. B., and C.J. SHATZ, 1985. Neuro/ genesis of the cat's
clear migration for cells entering vs leaving the S-phase. Dev. primary visual cortex./. Comp. Neurol. 242:611-631.
Neurosci. 22:44-55. MANGER, P., M. SUM, M. SZYMANSKI, S. RIDGWAY, and L. KRU-
HICKS, S. P., and C. J. D'AMATO, 1968. Cell migration to the BITZER, 1998. Modular subdivisions of dolphin insular cor-
isocortex in the rat. Anat. Rec. 160:619-634. tex: Does evolutionary history repeat itself?J.Cogn. Neurosci.
His, W., 1889. Die Neuroblasten und deren Entstehung im 10:153-166.
embryonalen Mark. Abh. Math. Phys. CL, Kgl. Saechs. Ges. MASSAGUE, J., and K, POLYAK, 1995. Mammalian antiprolifer-
Wiss. 15:313-372. ative signals and their targets. Curr. Opin. Gen. Dev. 5:91-96.
HOLLOWAY, R., 1974. On the meaning of brain size. Science McSHERRY, G. M., 1984. Mapping of cortical histogenesis in
184:677-679. the ferret./ Embryol Exp. Morph. 81:239-252.
HOLLOWAY, R., 1979. Brain size, allometry, and reorganiza- McSHERRY, G. M., and I. H. M. SMART, 1986. Cell production
tion: Toward a synthesis. In Development andEvolution of Brain gradients in the developing ferret isocortex./ Anat. 1-14.
Size, M. Hahn, C. Jensen, and B. Dudek, eds. New York: MESULAM, M.-M., 1998. From sensation to cognition. Brain
Academic Press, pp. 61-88. 121:1013-1052.
JERISON, H., ed., 1973. TheEvolution of the Brain and Intelligence. MISSON, J.-P, C. AUSTIN, T. TAKAHASHI, C. CEPKO, and V. S.
New York: Academic Press. CAVINESS, JR., 1991. The alignment of migrating neural cells

20 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 in relation to the murine neopallial radial glial fiber system. van Pelt, M. A. Corner, H. B. M. Uylings, and F. H. Lopes
Cereb. Cortex 1:221-229. da Silva, eds. Amsterdam: Elsevier, pp. 227-243.
MIYAMA, S., T. TAKAHASHI, R. S. NOWAKOWSKI, and V. S. CAV- RHEE, J., R. RABALLO, M. SCHWARTZ, and F. VACCARINO, 1998.
INESS, JR., 1997. A gradient in the duration of the Gl phase Lineage and non-lineage specific effects of fibroblast growth
in the murine neocortical proliferative epithelium. Cereb. factor (FGF2) on progenitor cells in the developing cere-
Cortex 7:678-689. bral cortex. Soc. Neurosci. Abst. 24:281.
MOLNAR, Z., and C. BLAKEMORE, 1995. How do thalamic axons ROBERTS, J., A. KOFF, K. POLYAK, E. FIRPO, S. COLLINS, M.
find their way to the cortex? Trends Neurosci. 18:389-397. OHTSUBO, and J. MASSAGUE, 1994. Cyclins, cdks and cyclin
MORGANE, P., I. GLEZER, and M. JACOBS, 1988. Visual cortex kinase inhibitors. Cold Spring Harbor Symp. Quant. Biol.
of the dolphin: An image analysis study. J. Comp. Neurol. 59:31-38.
273:3-25. SAUER, F. C., 1935. Mitosis in the neural tube. J. Comp. Neurol.
MURRAY, A., and T. HUNT, 1993. The Cell Cycle. New York: 62:377-405.
W. H. Freeman. SAUER, F. C., 1936. The interkinetic migration of embryonic
NAKAYAMA, K., N. ISHIDA, M. SHIRANE, A. INOMATA, T. INOUE, epithelial nuclei. J Morphol. 60:1-11.
N. SHISHIDO, I. HORII, D. LOH, and K.-I. NAKAYAMA, 1996. SAUER, F. C., 1937. Some factors in the morphogenesis of ver-
Mice lacking p27 Klpl display increased body size, multiple tebrate embryonic epithelia. J. Morphol. 61:563-579.
organ hyperplasia, retinal dysplasia, and pituitary tumors. SAUER, M. E., and B. E. WALKER, 1959. Radioautographic study
Cell 85:707-720. of interkinetic nuclear migration in the neural tube. Proc.
NIJHOUT, H., 1999. When developmental pathways diverge. Soc. Ext. Biol. (NY) 101:557-560.
Proc. Natl. Acad. Sa. (USA) 96:5348-5350. SCHMIDT-NIELSEN, K., 1984. Scaling: Why Is Animal Size So Im-
NORTHCUTT, R. G., and J. H. KAAS, 1995. The emergence and portant? Cambridge: Cambridge University Press.
evolution of mammalian neocortex. Trends Neurosci. 18:373- SHERR, C.J., 1993. Mammalian Gl cyclins. Cell 73:1059-1065.
379. SHERR, C. J., 1994. Gl phase progression: Cycling on cue. Cell
NOWAKOWSKI, R. S., S. B. LEWIN, and M. W. MILLER, 1989. 79:551-555.
Bromodeoxyuridine immunohistochemical determination SHERR, C. J., andj. M. ROBERTS, 1995. Inhibitors of mamma-
of the lengths of the cell cycle and the DNA-synthetic phase lian Gl cyclin-dependent kinases. Genes Dev. 9:1149-1163.
for an anatomically defined population. J. Neurocytol. SHIMAMURA, K., D. HARTIGAN, S. MARTINEZ, L. PUELLES, and
18:311-318. J. RUBENSTEIN, 1995. Longitudinal organization of the an-
OHTSUBO, M., and J. M. ROBERTS, 1993. Cyclin-dependent terior neural plate and neural tube. Development 121:3923-
regulation of Gl in mammalian fibroblasts. Science 259: 3933.
1908-1912. SIDMAN, R. L., and P. RAKIC, 1973. Neuronal migration, with
OHTSUBO, M., A. M. THEODORAS, J. SCHUMACHER, J. M. ROB- special reference to developing human brain: A review.
ERTS, and M. PAGANO, 1995. Human cyclin E, a nuclear pro- Brain Res. 62:1-35.
tein essential to the Gl-to-S phase transition. Mol. Cell. Biol. SIDMAN, R. L., and P. RAKIC, 1982. Development of the human
15:2612-2624. central nervous system. In Histology and Histopathology of the
ONO, M., S. JUBIK, and C. D. ABERNATHY, 1990. Atlas of the Nervous System, W. Haymaker and R. D. Adams, eds. Spring-
Cerebral Sulci. New York: Thieme Verlag. field: Charles C Thomas, pp. 3-145.
PARDEE, A. B., 1989. Gl events and regulation of cell prolif- SMART, I. H. M., and G. M. MCSHERRY, 1982. Growth patterns
eration. Science 246:603-608. in the lateral wall of the mouse telencephalon. II. Histolog-
PETERS, A., and E. G. JONES, 1984. Classification of Cortical Neu- ical changes during and subsequent to the period of iso-
rons. New York: Plenum Press. cortical neuron production.J Anat. 131:415-442.
PUELLES, L., andj. RUBENSTEIN, 1993. Expression patterns of SMART, I. H. M., and M. SMART, 1982. Growth patterns in the
homeobox and other putative regulatory genes in the em- lateral wall of the mouse telencephalon. I. Autoradio-
bryonic mouse forebrain suggest a neuromeric organiza- graphic studies of the histogenesis of the iso-cortex and ad-
tion. Trends Neurosci. 16:472-479. jacent areas.J. Anat. 134:273-298.
RADEMACHER,J., V. S. CAVINESS,JR., H. STEINMETZ, and A. M. SPREAFICO, R., C. FRASSONI, P. ARCELLI, M. SELVAGGIO, and S.
GALABURDA, 1993. Topographical variation of the human DE BIASI, 1995. In situ labeling of apoptotic cell death in
primary cortices: Implications for neuroimaging, brain the cerebral cortex and thalamus of rats during develop-
mapping and neurobiology. Cereb.Cortex3:313-329. ment. J. Comp. Neurol. 363:281-295.
RAKIC, P., 1972. Mode of cell migration to the superficial layers STENSAAS, S. S., D. K. EDDINGTON, and W. H. DOBELLE, 1974.
of fetal monkey neocortex. J. Comp. Neurol. 145:61-84. The topography and variability of the primary visual cortex
RAKIC, P., 1974. Neurons in rhesus monkey visual cortex: Sys- in man.J. Neurosurg. 40:747-755.
tematic relation between time of origin and eventual dis- STEPHAN, H., R. BAUCHOT, and O. ANDY, 1970. Data on the
position. Science 183:425-427. size of the brain and its various brain parts in insectivores
RAKIC, P., 1988. Specification of cerebral cortical areas. Science and primates. In The Primate Brain, C. Noback and W. Mon-
241:170-176. tagna, eds. New York: Appleton-Century-Crofts, pp. 289-
RAKIC, P., 1995. Corticogenesis in human and nonhuman pri- 297.
mates. In The Cognitive Neurosciences, M. S. Gazzaniga, ed. TAKAHASHI, T., P. BHIDE, T. GOTO, S. MIYAMA, and V. S. CAV-
Cambridge, Mass.: MIT Press, pp. 127-145. INESS, JR., 1999a. Proliferative behavior of the murine ce-
RAKIC, P., J.-P. BOURGEOIS, and P. S. GOLDMAN-RAKIC, 1994. rebral wall in tissue culture: Cell cycle kinetics and
Synaptic development of the cerebral cortex: Implications checkpoints. Exp. Neurol. 156:407-417.
for learning, memory and mental illness. In The Self- TAKAHASHI, T., T. GOTO, S. MIYAMA, R. S. NOWAKOWSKI, and
Organizing Brain: From Growth Cones to Functional Networks, J. V. S. CAVINESS, JR., 1999b. Sequence of neuron origin and

TAKAHASHI, NOWAKOWSKI, AND CAVINESS: NEOCORTICAL NEURONOGENESIS 21

 neocortical laminar fate: Relation to cell cycle of origin in VAN DER Loos H., and J. DORFL, 1978. Does the skin tell the
the developing murine cerebral wall. J. Neurosci. 19:10357- somatosensory cortex how to construct a map of the pe-
10371. riphery? Neurosci. Lett. 7:23-30.
TAKAHASHI, T., T. GOTO, S. MIYAMA, R. S. NOWAKOWSKI, and VAN ESSEN, D. C., W. T. NEWSOME, and H. R. MAUNSELL, 1984.
V. S. CAVINESS, JR., 1996. Intracortical distribution of a co- The visual field representation in striate cortex of the ma-
hort of cells arising in the PVE. Soc. Neurosci. Abst. 22:284. caque monkey: Asymmetries, anisotropies, and individual
TAKAHASHI, T., R. S. NOWAKOWSKI, and V. S. CAVINESS, JR., variability. Vision Res. 24:429-448.
1992. BUdR as an S-phase marker for quantitative studies VAN ESSEN, D. C., W. T. NEWSOME, J. H. R. MAUNSELL, and
of cytokinetic behaviour in the murine cerebral ventricular J. L. BIXBY, 1986. The projections from striate cortex (V1)
zone. J. Neurocytol. 21:185-197. to areas V2 and V3 in the macaque monkey: Asymmetries,
TAKAHASHI, T, R. S. NOWAKOWSKI, and V. S. CAVINESS, JR., areal boundaries, and patchy condensations. J Comp. Neurol.
1995. The cell cycle of the pseudostratified ventricular epi- 244:451-480.
thelium of the murine cerebral wall. J. Neurosci. 15:6046- VERNEY, C., T. TAKAHASHI, P. G. BHIDE, R. S. NOWAKOWSKI,
6057. and V. S. CAVINESS, JR., 1999. Independent controls for neo-
TAKAHASHI, T, R. S. NOWAKOWSKI, and V. S. CAVINESS, JR., cortical neuron production and histogenetic cell death. Dev.
1996. The leaving or Q fraction of the murine cerebral pro- Neurosci. 22:125-138.
liferative epithelium: A general computational model of WALSH, C., and C. CEPKO, 1992a. Generation of widespread
neocortical neuronogenesis.J. Neurosci. 16:6183-6196. cerebral cortical clones. Soc. Neurosci. Abst. 18:925.
TAKAHASHI, T, R. S. NOWAKOWSKI, and V. S. CAVINESS, JR.,
WALSH, C., and C. CEPKO, 1992b. Widespread dispersion of
1997. The mathematics of neocortical neuronogenesis. Dev.
neuronal clones across functional regions of the cerebral
Neurosci. 19:17-22.
cortex. Science 255:434-440.
TAKAHASHI, T, R. S. NOWAKOWSKI, and V. S. CAVINESS, 1999.
WALSH, C., and C. CEPKO, 1993. Clonal dispersion in prolif-
Cell cycle as operational unit of neocortical neuronogene-
erative layers of developing cerebral cortex. Nature 362:632-
sis. Neuroscientist 5:155-163.
TAN, S.-S., M. KALLONIATIS, K. STURM, P. TAM, and B. REESE, 635.
1998. Separate progenitors for radial and tangential cell dis- WELKER, W., 1990. Why does cerebral cortex fissure and fold?
persion during development of the cerebral cortex. Neuron A review of determinants of gyri and sulci. In Cerebral Cortex:
21:295-304. Comparative Structure and Evolution of Cerebral Cortex, Part II,
THOMAIDOU, D., M. MIONE, J. CAVANAGH, and J. PARNAVELAS, E.Jones and A. Peters, eds. New York: Plenum Press, pp. 3-
1997. Apoptosis and its relation to the cell cycle in the de- 136.
veloping cerebral cortex. J. Neurosci. 17:1075-1085. WOOLSEY, T. A., and H. VAN DER Loos, 1970. The structural
TSAI, L.-H., T. TAKAHASHI, V. S. CAVINESS, JR., and E. HARLOW, organization of layer IV in the somatosensory region (SI)
1993. Activity and expression pattern of cyclin-dependent of mouse cerebral cortex. Brain Res. 17:205-242.
kinase 5 in the embryonic mouse nervous system. Develop- ZETTERBERG, A., O. LARSSON, and K. WIMAN, 1995. What is
ment 119:1029-1040. the restriction point? Curr. Opin. Cell Biol. 7:835-842.
VACCARINO, F., M. SCHWARTZ, R. RABALLO, J. NILSEN,J. RHEE, ZILLES, K, 1990. Cortex. In The Human Nervous System, G. Pax-
M. ZHOU, T. DOETSCHMAN, J. COFFIN, J. WYLAND, and E. Yu- inos, ed. New York: Academic Press, pp. 757-802.
TING, 1999. Changes in the size of the cerebral cortex are ZILLES, K, E. ARMSTRONG, A. SCHLEICHER, and H.-J.
governed by fibroblast growth factor during embyrogenesis. KRETSCHMANN, 1988. The human pattern of gyrification in
Nat. Neurosci. 2:246-253. the cerebral cortex. Anat. Embryol. 179:173-179.

22 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

2 Synaptogenesis in the Neocortex
of the Newborn: The Ultimate
Frontier for Individuation?
J.-P. BOURGEOIS

ABSTRACT We have identified five distinct phases of synap- The formation of synaptic contacts can be described
togenesis in the cerebral cortex of primates from conception at two distinct levels. On the one hand, one can describe
to death. The first two phases, with a low density of synapses, the assembly of the numerous molecular components
occur during early embryonic life. They are followed by a third
phase of very rapid accumulation of synapses around birth. building up the pre- and postsynaptic domains of the
During the fourth phase the density of synapses is maintained neuron. On the other hand, one can describe the de-
at a very high level from the early postnatal period through velopment of ensembles of synaptic contacts in the
puberty. This is followed by a true loss of synapses during pu- cortical neuropil. Putting these two aspects of synapto-
berty. The fifth phase begins after sexual maturity and extends
into old age. It is characterized by a slow decline in the density
genesis in parallel is not trivial because the synaptic con-
of synapses. Experimental observations indicate that the early tact constitutes a crucial point of articulation between
phases of synaptogenesis are robust neurodevelopmental pro- the cellular properties of the single cortical neuron and
cesses determined by mechanisms intrinsic and common to the neurophysiological functions associated with large
the whole cortical mantle. The various postnatal phases coin- ensembles of these cortical neurons, which are under
cide with final adjustments of many aspects of synaptoarchi-
tectony, depending on normal functional inputs. They overlap distinct categories of constraints:
with critical periods for diverse cortical functions until pu- 1. The constraints related to the multiple intracellu-
berty. Inside each of these phases, distinct classes of synapses
appear in successive waves of synaptogenesis, providing the lar mechanisms control the type, amount, and distri-
first examples of a long list of discrete synaptogenetic events bution of presynaptic active zones, as well as
we are just beginning to explore. postsynaptic densities differentiated on the cell surface
of a cortical neuron. Are these constraints all under
The development of the neocortex in mammals is a strict genetic control?
highly orchestrated cascade of histological events in- 2. The constraints related to intercellular mecha-
cluding, successively, the generation and differentiation nisms are linked to the neurophysiological functions of
of neurons (Rakic, 1995), the navigation and organi- large ensembles of these cortical neurons and control
zation of the axonal projections between ensembles of
the development of the topological distribution of the
neurons (Barone et al., 1995), then the formation and
synaptic contacts in the neuropil. Are these constraints
maturation of the synaptic contacts that constitute the
major final steps of corticogenesis (Bourgeois, Goldman- all under strict epigenetic control?
Rakic, and Rakic, 1994, 2000; Bourgeois and Rakic, The interactions between genetically defined factors
1993; de Felipe et al., 1997; Huttenlocher and Dabhol- and experientially modifying factors are crucial in the
kar, 1997). All these biological events ultimately lead to development and morphofunctional maturation of the
individuation, the process by which individual mam- neocortex, in which synaptogenesis is a key event.
mals become differentiated in their societies. Knowing the exact timing of synaptogenesis in the cas-
As reviewed by Zoli and colleagues (1998), intercel- cade of histological events will provide a neuroanatom-
lular communications in the neocortex can occur ei- ical basis for the experimental identification of these
ther via volume transmission or by wiring transmission. interactions through the time course of development
This chapter describes the development of the wiring and maturation. The description of formation of en-
transmission via synaptic contacts. sembles of synaptic contacts is currently less explored
J.-P. BOURGEOIS Laboratoire de Recepteurs et Cognition,
than the description of the mechanisms of assembling
Departement des Biotechnologies, Institut Pasteur, Paris, of synaptic macromolecules. We tried to re-equilibrate
France. this. The neocortex of the rhesus monkey provides an

BOURGEOIS: SYNAPTOGENESIS IN NEOCORTEX 23

excellent animal model for the analysis of the devel- formed by axons of subcortical origin penetrating hor-
oping synaptoarchitectony, plasticity, and physiology of izontally within the neuroepithelium (horizontal ar-
the human neocortex (Teller, 1997). Using quantitative rows in figure 2.1). This phase of synaptogenesis begins
electron microscopy, we first delineated the kinetics of about 40-60 days after conception (DAC) and coin-
synaptogenesis in several cortical areas of macaque cides with the onset of neurogenesis (Rakic, 1995).
monkeys from conception to death (Bourgeois, Gold- Phase 2 is an early synaptogenesis at low density of
man-Rakic, and Rakic, 1994, 2000; Bourgeois and Rakic, synapses, now within the cortical plate itself, occurring
1993). at midgestation between 70 and 100 days after concep-
tion, i.e., during the peak of neurogenesis and migra-
Kinetics of synaptogenesis tion of neuroblasts. These synapses appear first in the
infragranular layers, then progressively in the more su-
In the primary visual cortex of the macaque five differ- perficial cortical layers of the CP following the vertical
ent phases were identified (figure 2.1): penetration of axonal projections (vertical arrow in fig-
ure 2.1) from cortical and subcortical origins (Bour-
Phase 1 is a very early synaptogenesis at low density
geois, Goldman-Rakic, and Rakic, 1994). All these
of synapses, occurring first in the primordial plexiform
"early" synaptic contacts are formed on the dendritic
layer (Marin-Padilla, 1988), then in the marginal zone
shafts of the neurons (figure 2.2A,B).
(MZ, or prospective layer I) and in the subplate (SP),
Phase 3 is a phase of very rapid accumulation of syn-
but not in the cortical plate (CP). These synapses are
apses. This phase begins 2 months before birth ( 100
days after conception) and initially proceeds in the ab-
sence of patterned stimulation from the external world.
The maximal density of synapses is reached about 2
months after birth. We have shown that the onset of
phase 3 is not necessarily linked to the end of neuro-
genesis (Granger et al., 1995). The most rapid part of
phase 3 occurs around birth (in macaque, delivery oc-
curs 165 days after conception), when 40,000 new syn-
apses are formed every second in each striate cortex of
the macaque (Rakic, Bourgeois, and Goldman-Rakic,
1994), mainly on the dendritic spines (figure 2.2A,B).
This rapid accumulation of synapses coincides with the
growth of the dendritic and axonal arbors. However, the
onset of phase 3 of synaptogenesis may either precede,
as in the primary visual cortex (Bourgeois and Rakic,
1993), or follow, as in the prefrontal cortex (Bourgeois,
Goldman-Rakic, and Rakic, 1994), the segregation of
the cortical columns.
Phase 4 is a plateau phase during which the mean
density of synapses remains at a very high level, 600-
FIGURE 2.1 Changes in the relative density of synapses (dis- 900 million synapses per cubic millimeter of neuropil
continuous line in the upper frame) as a function of days after
conception expressed on a log scale on the abscissae (t) in the
(Bourgeois and Rakic, 1993), throughout infancy and
primary visual cortex of macaque monkey during normal adolescence until puberty (in macaque, puberty occurs
development. Five different phases of synaptogenesis are iden- 3 years after birth).
tified between conception and death. Each phase is superim- Phase 5 is a slow, steady decline in density of synapses
posed above the density distribution of synapses in the cortical from puberty throughout adulthood, resulting mainly
layers of the neocortex represented in vertical frames in the from the loss of synapses located on dendritic spines.
middle part of the figure. During phase 1 synapses appear first
in the marginal zone (MZ), subplate (SP), and intermediate This decline coincides with another crucial and lengthy
zone (IZ). During phase 2, synapses now appear also in the phase in the late maturation of cortical functions. Fi-
cortical plate (CP) with a gradient of density represented by nally, a drop in density of synapses is observed during
a vertical arrow. Roman numerals indicate cortical layers. The senescence before death (figure 2.1; Peters, Sethares,
proposed effects of experience on cortical development and and Moss, 1998).
maturation are represented in horizontal frames in the lower
part. (Reproduced from Bourgeois, 1997, with permission Our observation of these five distinct phases has been
from Scandinavian University Press.) confirmed by studies done on macaque using either

24 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 brain imaging (Jacobs et al., 1995) or immunocyto-
chemical and histological techniques (Anderson et al.,
1995). These phases have also been identified in mouse
neocortices (de Felipe et al., 1997) and human brains
(Chugani, 1999; Huttenlocher and Dabholkar, 1997).
These facts strongly suggest that the mechanisms of cor-
ticogenesis and synaptogenesis are highly conserved
during the extensive evolution of the neocortex. In hu-
man cortex it is easy neither to obtain precise quanti-
tative data on the time course of these phases nor to
establish time points equivalent to those described in
the macaque. However, from the observations of Ze-
cevic (1998) on the human occipital cortex, we tenta-
tively propose that phase 1 might begin around 6-8
weeks of gestation. Phase 2 would begin near 12-17
weeks of gestation. The onset of phase 3 of rapid syn-
aptogenesis would occur around midgestation (17-24
weeks). On the other hand, data from Huttenlocher
and Dabholkar (1997) indicate that, in the primary vi-
sual cortex, phase 3 ends somewhere between 8 and 12
months after birth. For phase 4, a plateau of a high
density of synaptic contacts, differences appear between
cortical areas. In the human primary visual cortex the
plateau phase has the same length as that in the ma-
caque (2-3 years; figure 2.3), while in the prefrontal
cortex it lasts for a decade, until puberty (figure 2.3;
Huttenlocher and Dabholkar, 1997). Both in macaque
and human neocortices, phase 5 begins after puberty
and proceeds throughout adulthood with almost no de-
FIGURE 2.2 Distinct classes of synapses appear in distinct cline in densities of synaptic contacts, and is followed
waves of synaptogenesis in the primary visual cortex of ma- by a significant loss of synapses during senescence (fig-
caque monkey during normal development. As examples,
changes in the relative densities of synaptic contacts located ures 2.1 and 2.3).
on dendritic shafts (black line) or on dendritic spines (dis-
continuous line) are represented here as a function of days Distinct waves of synaptogenesis
after conception expressed on a log scale on the abscissae (t)
in two cortical layers (layer III in A and layer IVC in B). The The curve presented in figure 2.1 is actually an enve-
same pattern of synaptogenesis is observed during the first
lope curve "covering" many distinct waves of synapto-
three phases of synaptogenesis in all cortical layers (A and B),
suggesting the existence of highly conserved mechanisms. genesis differing in location, timing, or tempo, as
During phase 2 a first wave of synaptogenesis occurs on den- revealed by more detailed descriptions (Bourgeois,
dritic shafts followed two weeks later by a second wave of syn- Goldman-Rakic, and Rakic, 1994; Bourgeois and Rakic,
aptogenesis now on dendritic spines of the same neurons. 1993). With respect to location, we observed a wave of
During phase 3, before birth, the proportions of these classes synaptogenesis outside the cortical plate during phase
of synapses are reversed. During the plateau phase 4 the ki- 1, and another one inside this plate later on, during
netics of these waves of synaptogenesis differ. A protracted
plateau of high density of synapses on dendritic spines is ob- phase 2 (figure 2.1). With respect to timing, we ob-
served in supragranular layer III (A), while in granular layer served sequential waves of synaptogenesis first on den-
IVC (B) we observed a very short wave of synaptogenesis on dritic shafts, and later on dendritic spines of the same
dendritic spines and a protracted wave of synaptogenesis on neurons (figure 2.2A,B). As for tempo, in the primary
dendritic shafts. After an early bilateral enucleation was per- visual cortex we observed a protracted plateau until
formed during phase 1 of synaptogenesis (arrow in C), the puberty in supragranular layer III (figure 2.2A), while
same pattern of synaptogenesis unfolds normally during the
first three phases in granular layer IVC (compare A, B, and
a short phase 4 of synaptogenesis on dendritic spines
C). However, during plateau phase 4, the second reversal of was observed postnatally in granular layer IVC (figure
densities of synapses on dendritic spines and shafts does not 2.2B). In layer III of the prefrontal cortex of the
take place (C). macaque, a protracted plateau of high density of

BOURGEOIS: SYNAPTOGENESIS IN NEOCORTEX 25

 geois, Goldman-Rakic, and Rakic, 1994, 2000; Bour-
geois and Rakic, 1993; Granger et al., 1995; Zecevic,
Bourgeois, and Rakic, 1989; Zecevic and Rakic, 1991).
The very same early developmental pattern of synap-
togenesis—i.e., the ensembles of percentages of differ-
ent classes of synapses at successive steps of
development and maturation—is also observed in all
cortical layers (see the example shown in figure
2.2A,B,C) and in all the cortical areas examined.
Unexpectedly, during the perinatal period, we also
observed that the rapid phase 3 of synaptogenesis oc-
curs more concurrently than sequentially in the sen-
sory, motor, and associational cortical areas (Bourgeois,
Goldman-Rakic, and Rakic, 2000; Rakic et al., 1986;
Rakic, Bourgeois, and Goldman-Rakic, 1994). Excit-
atory and inhibitory synapses also accumulate concur-
rently during phase 3 in the prefrontal cortex of the
FIGURE 2.3 Changes in the relative densities of synapses in
the primary visual cortex (discontinuous line) and prefrontal macaque monkey (Anderson et al., 1995). These obser-
cortex (continuous line) of human brain as a function of days vations are corroborated by brain imaging studies on
after conception expressed on a log scale on the abscissae. the cortical mantle of the macaque (Jacobs et al.,
Phases 3 and 4 are indicated with numbers 3 and 4. (This is a 1995).
schematic representation of data published in Huttenlocher We have defined a "time window" of 41 days encom-
and Dabholkar, 1997, with permission from the authors and
John Wiley and Sons, Inc.) passing the midpoints of the rapid phase 3 of synapto-
genesis in all the cortical areas thus far described in the
macaque (Granger et al., 1995). Using available data,
we proposed that this is also the case in the human
parvalbumin-positive axon cartridges from chandelier
neocortex with a time window of about 3-5 months (fig-
cells was observed until puberty, while a short peak of
ure 2.3; Huttenlocher and Dabholkar, 1997; Rakic,
density of dopaminergic varicosities was observed ex-
Bourgeois, and Goldman-Rakic, 1994). Although the
actly during puberty (Anderson et al., 1995). In the hu- duration of this time window increases through evolu-
man cerebral cortex, sequential developments of tion, it always remains short as compared to the long
distinct cortical circuits (Burkhalter, Bernardo, and duration of the succeeding plateau phase 4 (a decade
Charles, 1993) most probably correspond to distinct in the human cortex; see figure 2.3), which extends un-
waves of synaptogenesis. These are the first examples of til puberty.
a long list of discrete synaptogenetic events we are just Recent observations reveal that, as in the macaque,
beginning to explore. phases 2 and 3 of synaptogenesis appear to occur con-
All these preliminary studies show that phase 3 is a currently in the prenatal human cerebral cortex (Ze-
period of rapid production of all categories of synaptic cevic, 1998). Similarly, correspondence analysis of the
contacts, even if during the next phases different cate- developmental patterns in several cytoarchitectonically
gories of synapses may display different kinetics in distinct cortical areas also shows a concurrent morpho-
macaque (Anderson et al., 1995; Bourgeois, Goldman- logical maturation of the whole cortical mantle in the
Rakic, and Rakic, 1994; Bourgeois and Rakic, 1993; Ze- human postnatal brain (Shankle et al., 1998).
cevic, Bourgeois, and Rakic, 1989; Zecevic and Rakic,
1991), mouse (de Felipe et al., 1997), and human (fig- Modifiability of phase 3 of synaptogenesis
ure 2.3; Huttenlocher and Dabholkar, 1997) cerebral
cortices. The onset of synaptogenesis long before birth, its
tempo, and the concurrent phase 3 in all cortical layers
Phase 3 of synaptogenesis in diverse cortical areas and cortical areas of the macaque strongly suggested to
us that these events might be determined by mecha-
In diverse cortical fields characterized by different cy- nisms intrinsic and common to the whole cortical
toarchitectonies and subserving very different physio- mantle (Bourgeois, Goldman-Rakic, and Rakic, 2000;
logical functions, the same phases of synaptogenesis Goldman-Rakic, Bourgeois, and Rakic, 1997; Rakic et
were found at the same developmental stages (Bour- al., 1986; Rakic, Bourgeois, and Goldman-Rakic, 1994).

26 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

We have tested this hypothesis in the following Effects of environment on diverse phases
experiments: of synaptogenesis
1. Using a mild intervention with experimental pre-
The transition from intrinsic to extrinsic control of syn-
term monkeys (Bourgeois, Jastreboff, and Rakic, 1989),
aptogenesis in the cerebral cortex most likely involves
we found that a premature bilaterally equilibrated ex-
many steps. Experience, i.e., the presence of patterned
posure to visual environment does not accelerate or de-
evoked activities coming from the world external to the
lay the rate of synaptic accretion during the rapid phase
neocortex, has diverse effects on cortical maturation
3 of synaptogenesis in the primary visual cortex. This
during the different phases of synaptogenesis. As
phase of synaptogenesis proceeds in relation to the time
sketched in figure 2.1, we tentatively propose the fol-
of conception rather than time of delivery, i.e., the on-
lowing interpretations:
set of visual stimulation.
2. Using a more drastic intervention, we found that 1. Phases 1 and 2 of very early synaptogenesis: The
an early bilateral enucleation (Bourgeois and Rakic, early events, such as neurogenesis, neuronal migration,
1996) does not alter the final mean densities of synapses neuritic navigation, individualization of cortical layers,
reached at the end of rapid phase 3 and maintained neurochemical differentiation, and early synaptogene-
during the phase 4 plateau in the striate cortex of blind sis, are dominated by genetic and epigenetic mecha-
monkeys. Our study also indicates that a few weeks after nisms, all of which are intrinsic to the neocortex. These
birth the proportions of synaptic contacts situated on events can be severely disturbed by genetic mutations,
dendritic spines (75%) and shafts (25%) were similar viral infections, toxic agents, surgical interventions, and
in all cortical layers of normal and enucleated animals. so forth; however, they are not yet influenced by expe-
Four months after birth, the localization on dendritic rience coming from the world external to the neocor-
spines or shafts in the thalamorecipient granular layers tex. Using Greenough's terminology (Greenough and
fails to mature properly in the absence of normal func- Alcantara, 1993), these early events are "experience-
tional input from the periphery (figure 2.2C). These independent" (figure 2.1) and, possibly, are common to
proportions, which normally become reversed during all mammals. A spontaneous activity most likely already
infancy in sublayers IVAB and IVC, were not reversed circulates via the early formed synapses and participates
in the enucleates. The proportions of symmetric (20%) in the wiring of cortical neurons (Maffei and Galli-
versus asymmetric (80%) synapses located on dendritic Resta, 1990; Mooney et al., 1996).
spines were within the normal range of variability in 2. Rapid phase 3 of synaptogenesis: This phase
both groups of animals. As a result, the granular layers is dominated early on by "experience-expectant," and
in the striate cortex of early-blind macaques have more later by "experience-dependent" mechanisms (figure 2.1;
Greenough and Alcantara, 1993). The intrinsic mech-
excitatory axodendrospines than do normal animals.
anisms become epigenetically modulated by experience
This experimental model also provides an additional coming to the neocortex from the external world. Ex-
example of the fact that a perturbation at an early neu- perience-expectant means that the presence of some visual
rodevelopmental stage (before midgestation in the parameters in the external world (orientation, color,
present case) may have a late (late infancy in the pres- movement, disparity, etc.) are necessary for the proper
ent case) and long-lasting effect of disorganization final adjustment of the cortical circuits being estab-
despite an apparently normal intermediary period (la- lished for their specific processing. This phase 3 coin-
tency period for this aspect of synaptoarchitectony). cides with the beginning of critical periods. The
These experimental observations indicate that early mechanisms involved during this phase are assumed to
synaptogenesis is a very robust neurodevelopmental be common to all individuals of a given mammalian
process. They reinforce our hypothesis that the onset, species.
time course, magnitude, and rate of cortical synapto- Huttenlocher and Dabholkar (1997) claimed that the
genesis during phase 3 in the primate striate cortex pro- onset of cortical functions occurs only at the end of
ceed without stimulation from the periphery. They are phase 3 of synaptogenesis and at different ages in dif-
determined and coordinated by mechanisms intrinsic ferent cortical regions. This is in contradiction with ac-
and common to the whole cortical mantle. The nature cumulating evidence of very early maturation of
of these mechanisms and their interactions (metabolic, anatomical and physiological parameters in the neo-
trophic, hormonal, genetic, etc.) is under investigation. cortex of primates.
However, final adjustments of some aspects of the syn- In macaques, many sensory, motor, visceral, and cog-
aptoarchitectony depend on normal functional input. nitive functions subserved by the neocortex are present

BOURGEOIS: SYNAPTOGENESIS IN NEOCORTEX 27

very early after birth, when the main aspects of synap- (Daw et al., 1992), monkeys (Le Vay, Wiesel, and Hubel,
toarchitectony are still being laid down. The patterns of 1980), and humans (Johnson and Newport, 1989).
ocular dominance columns are already adult-like in the Critical periods for different visual functions subserved
primary visual cortex of newborn macaques (Horton by different cortical layers in the primary visual cortex
and Hocking, 1996). Complex receptive field proper- have different timings. Neuronal plasticity lasts longer
ties, such as face recognition, of inferotemporal neu- in the upper and lower cortical layers than in the gran-
rons are already adult-like, as early as investigators were ular layers. For example, the short postnatal peak of
able to test, i.e., only a few weeks after birth in macaques density observed in granular layer IVC in the striate cor-
(Rodman, 1994). The possibility of a cortical circuitry tex of the macaque (see the section Distinct Waves of
prewired even for highly integrated functions has been Synaptogenesis and figure 2.2B)fitswell with the short
invoked (Rodman, 1994). A cardinal cognitive function critical period for orientation selectivity subserved by
thought to be subserved by the dorsolateral prefrontal this layer, while the long-lasting plateau phase 4 ob-
association cortex [i.e., performance on Piaget's A not served in supragranular layer III (see the section Dis-
B task; see this volume, Stiles (chapter 27), Diamond tinct Waves of Synaptogenesis and figure 2.2A)
(chapter 29), Luciana (chapter 41)] is present long be- coincides with the protracted critical periods described
fore the end of the rapid phase 3 of synaptogenesis a for contrast sensitivity, binocular selectivity, and Vernier
few weeks after birth (Bourgeois, Goldman-Rakic, and hyperacuity, subserved by this layer (Blakemore, Vital-
Rakic, 1994; Diamond and Goldman-Rakic, 1989; Gold- Durand, and Garey, 1981; Harwerth et al., 1986).
man-Rakic, Bourgeois, and Rakic, 1997). Harlow and 4. Phase 5: This phase is dominated by experience-
Harlow (1962) showed that critical periods for social dependent mechanisms (figure 2.1; Greenough and
skills of the macaque infants could also begin as early Alcantara, 1993). It corresponds mainly to a very slow
as 2 months after birth. Evoked activity in response to decrease in the density of synapses through adulthood.
maternal voice seems to be present before birth in hu- Over several decades of aging the efficacy and the local
man fetuses (Schaal, Lecanuet, and Granier-Deferre, plastic reorganizations of synaptic contacts are now
1999), and newborn human infants most certainly see related only to the experience of each individual
(Teller, 1997). (Greenough and Alcantara, 1993), but the main aspects
Our working hypothesis is that the very rapid and of synaptoarchitectony in the cortical networks remain
concurrent synaptogenesis in the whole cortical mantle unchanged.
during the rapid phase 3 allows the early coordinated We observed a true loss of synapses in neocortex of
emergence of all these cortical functions (Rakic et al., macaque monkey around puberty (Bourgeois, Goldman-
1986; Rakic, Bourgeois, and Goldman-Rakic, 1994). Rakic, and Rakic, 1994; Bourgeois and Rakic, 1993). A
This is essential for competitive and selective (Chan- similar loss of synapses near puberty was also observed,
geux and Danchin, 1976) interactions among the very using quantitative synaptology, in the cortices of human
heterogeneous cortical inputs in each point of the cor- brain (Huttenlocher and Dabholkar, 1997) and mouse
tex. However, these cortical functions will also require brains (de Felipe et al., 1997). These observations are
up to 3 years to reach full maturity in the macaque confirmed by brain imaging in the macaque (Jacobs et
(Goldman-Rakic, Bourgeois, and Rakic, 1997) and al., 1995) and human cortices (Chugani, 1999).
more than a decade in human (Huttenlocher and Dab- Although we arbitrarily included this period of rapid
holkar, 1997; Rakic, Bourgeois, and Goldman-Rakic, loss of synapses during puberty in phase 5, we do not
1994), i.e., until puberty. rule out the possibility that it might be a singular phase
3. Phase 4, en plateau: This phase is also dominated distinct from the other phases of synaptogenesis, with
by experience-expectant and experience-dependent its own mechanisms and tempo.
mechanisms (figure 2.1; Greenough and Alcantara, One possibility is that the loss of synapses in the pre-
1993). This extended period of synaptic plasticity cor- frontal cortex during puberty might correspond to a
responds to a process of continuous reorganization of "hormonal sanction" on some aspects of neuronal plas-
the intracortical axonal arborizations (Callaway and ticity. The deep hormonal "reorganization" occurring
Katz, 1990; Lowel and Singer, 1992), allowing fine- in and near puberty would contribute to the definitive
tuning and maturation of the neuronal circuits during elimination of labile synapses not stabilized during the
the 3 years after birth until puberty. This topological preceding plateau phase 4. In humans, central visual
reorganization of synaptic contacts occurs with a con- defects due to cataract become less treatable as puberty
stant mean density of synapses. approaches (Mitchell and Timney, 1984). At age 12,
Phases 3 and 4 of synaptogenesis coincide with the people stop being able to learn nonmaternal languages
diverse critical periods extending until puberty in cats without effort and without accents (Johnson and New-

28 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

port, 1989). And, as frequently reiterated in textbooks, During the plateau phase of synaptogenesis, a selec-
the period around puberty corresponds to a crystalli- tive stabilization of synapses might occur during re-
zation of personality, marking the end of several basic organization of the tridimensional distribution of
learning capacities and skills. synapses in the neuropil, without any net loss of syn-
apses, because different waves for different classes of
Selectionists versus constructivists hypothesis synapses occur at different developmental stages (figure
for synaptogenesis 2.2A,B; Bourgeois and Rakic, 1993). All these models
include the existence of extensive production, remod-
Studies addressing phases 3 and 4 of synaptogenesis eling, and retraction of dendritic and axonal branches
represent foci of contention between selectionist and and synapses during development of cortex. For ex-
constructivist hypotheses. Both hypotheses relate the ample, synapses identified on proximal axon collaterals
modifications of the synaptoarchitectony during devel- at a juvenile stage relocate to distal axonal domains at
opment, maturation, and learning in the neocortex to the adult stage on local circuit neurons in cortex (Cal-
the evoked activity resulting from individual experi- laway and Katz, 1990).
ence. And both hypotheses recognize that there is an The quantitative observations from Greenough and
early and rapid accumulation of synapses, the most ac- Alcantara (1993) suggest that learning induces forma-
tive of which will ultimately be retained. But the hy- tion of new synapses. However, if such an increase in
potheses differ in the sequence of events invoked. density of synapses occurs everywhere and permanently
According to the selective stabilization hypothesis as primates learn, this density should increase con-
proposed by Changeux and Danchin (1976), connec- stantly across the life span. This does not fit with the
tivity progresses from a larger to a smaller number of unambiguous loss of diverse classes of synapses, repeti-
connections and synapses. Intrinsic mechanisms con- tively observed at different moments of phases 4 and 5,
trol the initial formation of large pools of synaptic con- in different cortical layers and cortical areas of the ma-
tacts, while the subsequent stabilization of the most caque (figures 2.1-2.4; Anderson et al., 1995; Bour-
active synapses, and the elimination of the less active geois, Goldman-Rakic, and Rakic, 1994; Bourgeois and
ones, is regulated by evoked activity. Rakic, 1993; Zecevic, Bourgeois, and Rakic, 1989; Ze-
In the constructivist hypothesis proposed by Katz and cevic and Rakic, 1991), mouse (de Felipe et al., 1997),
Shatz (1996), Purves, White, and Riddle (1996), and or human (Huttenlocher and Dabholkar, 1997) cortices.
Quartz and Sejnowski (1997), connectivity matures In addition, Geinisman and colleagues (2000) showed
from a smaller to a larger number of connections and that associative learning does not increase the number
synapses. According to these instructionist models, the of synapses in hippocampus. It is difficult to reconcile
spontaneous activity and the patterns of evoked activi- all these facts with a strict interpretation of the construc-
ties cause the formation of synapses, control the extent tivist models. To go further, one could even hypothesize
of their domains, and orient the elaboration of their that these transient bursts of synaptogenesis in the adult
synaptoarchitectonic organization. Losses of neurons, neocortex may in fact correspond to a transient phase
axons, and synapses, if any, are just epiphenomena. 4 within the phase 5, i.e., transient selectionist episodes
In the primary visual system of the macaque or cat, of synaptoarchitectonic reorganization.
the coincidence between the continuous increase in
axonodendritic branching, the segregation of ocular Evolution of phase 3 of synaptogenesis in neocortex
dominance columns (Katz and Shatz, 1996; Purves,
White, and Riddle, 1996; Quartz and Sejnowski, 1997), Through evolution of neocortex, there is a shift in the
and the rapid phase 3 of synaptogenesis (Bourgeois and onset time of neurogenesis in the primary visual cortex
Rakic, 1993) are taken as support for constructionist (figure 2.4); it occurs 14 days after conception (DAC)
hypothesis. However, the fact that the onset of phase 3 in the rat (Frantz et al., 1994), 30 DAC in the cat (Lus-
of rapid synaptogenesis slightly precedes the loss of a kin and Shatz, 1985), 40 DAC in the macaque, and 43
few LGN axonal branches from contralateral ocular DAC in the human neocortex (Rakic, 1995). With re-
dominance columns does not rule out a selective com- spect to synaptogenesis, the comparison of the tempo
petition between these few axons subserving the two of phase 3 in the primary visual cortices of these four
eyes. The massive growth and sprouting of axonal ar- mammalian species also reveals two significant modifi-
bors that follow might just as well hide local selectionist cations, as sketched in figure 2.4:
competitions between diverse axons now subserving the 1. The onset of the rapid phase 3 of synaptogenesis
same eye inside the same dominance column. occurs progressively later after conception: from 23

BOURGEOIS: SYNAPTOGENESIS IN NEOCORTEX 29

 period, a crucial parameter which, otherwise, could
make delivery problematic.
These observations become even more interesting
when one relates them to the evolution of synaptoar-
chitectony. The final mean density of synapses per unit
volume of cortical neuropil in the mature striate cortex
(column 5 in table 2.1) is not higher in human than in
rat, although it takes about 30 times as many days to
produce these synapses in each point of the cortex.
Even the mean number of synapses per neuron in the
striate cortex remains in the same range from rat to
man (column 6 in table 2.1).
As plausible mechanisms for this extension of phase
3 of synaptogenesis, two working hypotheses, which are
not mutually exclusive, are now considered:
1. A genetic hypothesis: The time course of phase 3
is controlled by genetic mechanisms, and the number
and/or the time needed by the developmental master
genes to control phase 3 of synaptogenesis might in-
crease during evolution. This hypothesis leads to one
FIGURE 2.4 Changes in the relative density of synapses (dis- falsifiable experimental prediction: The mutation of
continuous lines) in the primary visual cortices of four differ-
ent mammalian species—rat, cat, macaque, and man—as a these genes or the deregulation of their expression
function of days after conception expressed on a log scale on should result in a modification of the onset and/or time
the abscissae. Only phases 3, 4, and 5 of synaptogenesis are course of phase 3 of synaptogenesis.
sketched here. Phase 3 is indicated with the number 3. The 2. A morphofunctional hypothesis: Morphological
striped and the solid black horizontal bars represent the time and functional heterogeneities of the axonal inputs to
of neurogenesis in the lateral geniculate nucleus (LGN) and
primary visual cortex (V1). (Reproduced from Bourgeois, the primary visual cortex increase with the addition of
1997, with permission from Scandinavian University Press.) extrastriate cortical areas during mammalian evolution,
and the increased diversity of synaptic inputs causes the
extension of phase 3 of synaptogenesis. During evolu-
DAC in the rat (Blue and Parnavelas, 1983) to 110 DAC tion of mammals the neocortex grows more in surface
in the human neocortex (table 2.1 and figure 2.4; Hut- area than in the vertical dimension. From rat to human,
tenlocher and Dabholkar, 1997). Through phylogeny of the total surface area of the cortical mantle increases by
mammals, this shift is less pronounced than that of the 3 orders of magnitude. Cortical expansion in evolution
time of delivery (figure 2.4). As a result, the onset of is achieved largely by addition of radial units, and thus
phase 3 of synaptogenesis, a postnatal event in rodents by increases in the number, not the size, of constituent
and carnivores, progressively becomes a precocious pre- columns (Rakic, 1995). Although the two-dimensional
natal event in primates. shape of the primary visual cortex appears highly con-
2. More importantly, the duration of phase 3 in the served across mammalian species (Duffy, Murphy, and
primary visual cortex increases significantly from 14 Jones, 1998), its surface increases significantly. The total
days in the rat (Blue and Parnavelas, 1983) to 30 days number of cortical areas also increases significantly
in the cat (Cragg, 1975), 136 days in the macaque (column 7 in table 2.1). More relevant to our purpose,
(Bourgeois and Rakic, 1993), and about 400 days in the the total number of visual areas increases from 4 in the
human brain (table 2.1 and figure 2.4; Huttenlocher rat (Coogan and Burkhalter, 1993) to about 25 in the
and Dabholkar, 1997). Normalization of the duration macaque and probably many more in the human brain
of phase 3 of synaptogenesis to the onset time of pu- (column 8 in table 2.1; Van Essen, 1985). Most of these
berty (table 2.1) shows that it represents a constant pro- visual areas are directly or indirectly connected to the
portion of 10-17% of this time in the four species primary visual cortex (Coogan and Burkhalter, 1993;
studied, suggesting that phase 3 is a developmental pro- Salin and Bullier, 1995; Van Essen, 1985), and their in-
cess not apart from others. In an evolutionary perspec- puts are quite heterogeneous. This heterogeneity may
tive this protracted phase 3 postpones the increase in be histologically related, for example, to the high di-
volume of the primate brain well into the late postnatal versity of cell adhesion molecules (Alcantara, Pfennin-

30 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 TABLE 2.1
Parameters related to the development of the primary visual cortex of four distinct mammalian species

Total
Phase 3 Number of Total
Puberty Duration Density of Synapses per Cortical Number of
Gestation Onset (onset-end) synapses1 Neuron Areas Visual Areas
Rat 21 days 82 DAC (2 14 days 320-946 12,500-13,500 21 4-6
months (P2-P16)
after birth)
Cat 65 days 248 DAC (6 30 days 276-406 5800-9300 30-50? 12-17
months (P9-P39)
after birth)
Macaque 165 days 1260 DAC (3 136 days 276-620 2000-5600 72 25
years after (E90-P61)
birth)
Human 280 days 4660 DAC 470 days 350 6800-10,000 200? 50?
(12 years (E120-P310)
after birth)
1
Density of synapses is expressed in millions per cubic millimeter of cortical tissue.
DAC = days after conception; E = embryonic day; P = postnatal day.
Reproduced from Bourgeois (1997) with permission from Scandinavian University Press.

ger, and Greenough, 1992) and diffusible or signaling natal developmental events, while complete maturation
molecules (Edlund and Jessel, 1999). This heteroge- of cortical functions is protracted until puberty. Al-
neity may also be functional: In different cortical areas though kinetics is not the whole story of synaptogenesis,
neurons fire with different temporal patterns (Ferster our identification of distinct phases of synaptogenesis
and Spruston, 1995) and different conduction times provides a new developmental frame for the description
(Salin and Bullier, 1995). In the morphofunctional hy- of the establishment and successive adjustments of syn-
pothesis all these heterogeneities increase the number aptoarchitectony in neocortex of primates. The density
and the complexity of cellular interactions between of synapses is the same at birth and after puberty; how-
nerve endings during the formation of synapses, their ever, the developmental constraints on synaptogenesis
selective stabilization, their elimination, or all these are expected to be totally different. We observed that
steps. This, along with an increased number of waves of the onset and time course of phase 3 of synaptogenesis
synaptogenesis, might increase the duration of phases are very robust developmental events, while influences
3 and 4 of synaptogenesis. from environment increase thereafter. Among multiple
The morphofunctional hypothesis leads to another mechanisms, this plasticity is related to the high vari-
falsifiable experimental prediction: The anatomical or ability in the density, localization, assembly, and nature
functional suppression of a specified number of pre- of many types of molecules in the pre- and postsynaptic
striate cortical areas should shorten the duration of domains of neurons. To cite only a few examples, one
phase 3 of synaptogenesis observed in the primary vi- should refer to the large diversity and plasticity of the
sual cortex. Conversely, the addition of cortical areas subunits of classical pharmacological receptors (Chan-
should prolong phase 3. geux et al., 1997; Craig 1998; Vannier and Triller, 1997)
These two hypothesis are not mutually exclusive: The as well as to new families of molecules such as class I
two classes of mechanisms might be at work at different MHC (Corriveau, Huh, and Shatz, 1998) and synaptic
moments of phase 3 or/and for distinct ensembles of cadherins (Hagler and Goda, 1998; Tanaka et al., 2000),
synapses. These mechanisms are expected to be iden- which have been observed during formation of synap-
tical in the whole cortical mantle. tic contacts. These numerous structural and signal-
transducing molecules assemble and deassemble in
New perspectives on genetics of synaptogenesis, highly dynamic pre- and postsynaptic domains of mor-
evolution of epigenesis, and individuation phologically permanent synaptic contacts.
We hypothesize that, upstream of all these molecular
In neocortices of nonhuman and human primates, neu- events, new families of genes control the onset and
rogenesis and synaptogenesis are very precocious pre- duration of the different phases of synaptogenesis.

BOURGEOIS: SYNAPTOGENESIS IN NEOCORTEX 31

Different groups of these genes may be activated, some psychosocial competencies, reaching its maximum in
transiently, some permanently, during distinct waves of the human brain.
synaptogenesis. The identification of the functional in- All these descriptions, including that of the devel-
teractions between different groups of genes during dis- opment of the synaptoarchitectony in the cerebral cor-
tinct critical periods of synaptogenesis will constitute an tex, participate in the identification of the mechanisms
entirely new field of research on corticogenesis. The involved in the process of individuation.
regulatory genes controlling the onset of phase 3 of
synaptogenesis will have to be identified first. The ACKNOWLEDGMENTS I thank Professors Jean-Pierre Chan-
sequencing of the human genome, to be completed geux, Patricia Goldman-Rakic, and Pasko Rakic for their con-
within the early years of this century, will allow transla- stant support and numerous discussions about epigenesis.
tion of the cascade of histological events observed dur- This work was supported at its inception by a Fogarty Inter-
national Fellowship and constantly by the Centre National de
ing corticogenesis into a cascade of genetic events.
la Recherche Scientifique.
The aim of future projects is to explore the mecha-
nisms of the transitions from strictly intrinsic control
(genes) to epigenetic control (environment) of the for-
mation of synaptoarchitectony during cortical devel- REFERENCES
opment of the normal or pathological individual. ALCANTARA, A. A., K. H. PFENNINGER, and W. T. GREENOUGH,
Whether these transitions result from a changing over- 1992. 5B4-CAM expression parallels neurite outgrowth and
synaptogenesis in the developing rat brain. J. Comp. Neurol.
lap of independent mechanisms subserved by distinct
319:337-348.
ensembles of synaptic circuits, or whether it is a pro- ANDERSON, S. A., J. D. CLASSEY, F. CONDE, J. S. LUND, and D. A.
gressive epigenetic transformation from one mecha- LEWIS, 1995. Synchronous development of pyramidal neu-
nism into the next in the very same ensembles of ron dendritic spines and parvalbumin-immunoreactive
synapses, is not yet known. chandelier neuron axon terminals in layer III of monkey
The recognition of distinct and successive waves of prefrontal cortex. Neurosdence 67(1):7-22.
BARONE, P., C. DEHAY, M. BERLAND, J. BULLIER, and H. KEN-
synaptogenesis, along with these studies of genetic con- NEDY, 1995. Developmental remodeling of primate visual
trols of synaptogenesis, will also provide new biological cortical pathways. Cereb. Cortex 5:22-38.
approaches to testing diverse neurodevelopmental hy- BLAKEMORE, C., F. VITAL-DURAND, and L. J. GAREY, 1981. Re-
potheses for psychiatric disorders (Bloom, 1993; Fein- covery from monocular deprivation in the monkey. I. Re-
berg, 1983, 1990; Stefan and Murray, 1997). According versal of physiological effects in the visual cortex. Proc. R.
Soc. Land. (B) 213:399-423.
to these hypotheses, early perturbations in the devel- BLOOM, R. E., 1993. Advancing a neurodevelopmental origin
opment of neuronal circuits remain hidden until some of schizophrenia. Arch. Gen. Psychiatry 50:224-227.
aspects of their late normal maturation reveal the BLUE, M. E., and J. G. PARNAVELAS, 1983. The formation and
defect. maturation of synapses in the visual cortex of the rat. II.
Synaptogenesis is a crucial part of the phylogenetic Quantitative analysis. J. Neurocytol. 12:697-712.
BOURGEOIS, J.-P, 1997. Synaptogenesis, heterochrony and epi-
and ontogenetic history of the neocortex. However, we genesis in the mammalian neocortex. Acta Paediatrica
do not yet know the sequence of causality, if any, be- 422(suppl.):27-33.
tween the extension of phases 3 and 4 of synaptogene- BOURGEOIS, J.-P., P. S. GOLDMAN-RAKIC, and P. RAKIC, 1994.
sis, the appearance of many new local neuronal circuits, Synaptogenesis in the prefrontal cortex of rhesus monkey.
the functional refinement of the multiple receptive Cereb. Cortex 4:78-96.
BOURGEOIS, J.-P., P. S. GOLDMAN-RAKIC, and P. RAKIC, 2000.
fields of the striate cortex, and the increased numbers
Formation, elimination and stabilization of synapses in the
and plasticity of the neuronal networks, all observed primate cerebral cortex. In The New Cognitive Neurosciences,
through phylogeny. M. Gazzaniga, ed. Cambridge, Mass.: MIT Press, pp. 45-53.
According to the heterochronic epigenesis hypothe- BOURGEOIS, J.-P., P. J. JASTREBOFF, and P. RAKIC, 1989. Syn-
sis (Bourgeois, 1997), the extension of phase 3 of syn- aptogenesis in visual cortex of normal and preterm
monkeys: Evidence for intrinsic regulation of synaptic over-
aptogenesis significantly increases (1) the scale and
production. Proc. Natl. Acad. Sci. (USA) 86:4297-4301.
number of possible epigenetic combinations, (2) the BOURGEOIS, J.-P., and P. RAKIC, 1993. Changes of synaptic den-
amplitude of the "overshoot" of synapses, (3) the exten- sity in the primary visual cortex of the macaque monkey
sion of their plasticity well into adulthood, and (4) the from fetal to adult stage. J Neuwsci. 13:2801-2820.
morphofunctional interindividual variabilities observed BOURGEOIS, J.-P., and P. RAKIC, 1996. Synaptogenesis in the
through mammalian phylogeny. This hypothesis also occipital cortex of macaque monkey devoid of retinal input
from early embryonic stages. Eur. J. Neurosci. 8:942-950.
proposes that the extension of phases 3 and 4 of syn- BURKHALTER A., K. L. BERNARDO, and V. CHARLES, 1993. De-
aptogenesis during evolution of neocortex significantly velopment of local circuits in human visual cortex. J. Neu-
increases the process of maturation of cognitive and rosci. 13:1916-1931.

32 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

CALLAWAY, E. M., and L. KATZ, 1990. Emergence and refine- GRANGER, B., F. TEKAIA, A. M. LESOURD, P. RAKIC, and J.-P.
ment of clustered horizontal connections in cat striate cor- BOURGEOIS, 1995. Tempo of neurogenesis and synapto-
tex. J. Neurosci. 10:1134-1153. genesis in the primate cingulate mesocortex: Comparison
CHANGEUX, J. P., A. BESSIS, J.-P. BOURGEOIS, P. J. CORRINGER, with the neocortex. J. Comp. Neurol. 360:363-376.
A. DEVILLERS-THIERY, J. L. EISELE, M. KERSZBERG, C. LENA, GREENOUGH, W. T, and A. A. ALCANTARA, 1993. The roles of
N. LENOVERE, M. PICCIOTO, and M. ZOLI, 1997. Nicotinic experience in different developmental information stage
receptors and brain plasticity. In Cold Spring Harbor Sympo- processes. In Developmental Neurocognition, B. de Boysson-
sium on Function and Dysfunction in the Nervous System, No. Bardies, S. de Schonen, P. Jusczyk, P. McNeilage, and J. Mor-
61, pp. 343-362. ton, eds. Dordrecht: Kluwer Academic Publishers, pp. 3-16.
CHANGEUX, J. P., and A. DANCHIN, 1976. Selective stabilization HAGLER, D. J., and Y. GODA, 1998. Synaptic adhesion: The
of developing synapses as a mechanism for the specification building blocks of memory? Neuron 20:1059-1062.
of neural network. Nature 264:705-712. HARLOW, H. F, and M. K. HARLOW, 1962. Social deprivation
CHUGANI, H. T., 1999. Metabolic imaging: A window on brain in monkeys. Scientific American 207:136-146.
development and plasticity. Neurosdentist 5:29-40. HARWERTH, R. S., E. L. SMITH III, G. C. DUNCAN, M. L. J.
COOGAN, T. A., and A. BURKHALTER, 1993. Hierarchical or- CRAWFORD, and G. K. VON NOORDEN, 1986. Multiple sensi-
ganization of areas in rat visual cortex. J. Neurosci. tive periods in the development of the primate visual system.
13(9):3749-3772. Science 232:235-238.
CORRIVEAU, R. A., G. S. HUH, and C. J. SHATZ, 1998. Regula- HORTON, J. C., and D. R. HOCKING, 1996. An adult-like pattern
tion of class I MHC gene expression in the developing and of ocular dominance columns in striate cortex of newborn
mature CNS by neural activity. Neuron 21:505-520. monkeys prior to visual experience. J. Neurosci.l6(5):l79l-
CRAGG, B. G., 1975. The development of synapses in the visual 1807.
system of the cat. J. Comp. Neurol. 160:147-166. HUTTENLOCHER, P. R., and A. S. DABHOLKAR, 1997. Regional
CRAIG, A. M., 1998. Activity and synaptic receptor targeting: differences in synaptogenesis in human cerebral cortex. J.
The long view. Neuron 21:459-462. Comp. Neurol. 387:167-178.
JACOBS, B., H. T. CHUGANI, V. ALLADA, S. CHEN, M. E. PHELPS,
DAW, N. W., K. Fox, H. SATO, and D. CZEPITA, 1992. Critical
D. B. POLLACK, and M. J. RALEIGH, 1995. Developmental
period for monocular deprivation in the cat visual cortex.
changes in brain metabolism in sedated rhesus macaques
J. Neurophysiol 67:197-202.
and vervet monkeys revealed by positron emission tomog-
DIAMOND, A., and P. S. GOLDMAN-RAKIC, 1989. Comparison
raphy. Cereb. Cortex 3:222-233.
of human infants and rhesus monkeys on Piaget's AB task:
JOHNSON, J. S., and E. L. NEWPORT, 1989. Critical period ef-
Evidence for dependence on dorsolateral prefrontal cortex.
fects in second language learning: The influence of matu-
Exp. Brain Res. 74:24-40.
rational state on the acquisition of English as a second
DUFFY, K. R., K. M. MURPHY, and D. R.JONES, 1998. Analysis
language. Cogn. Psychol. 21:60-99.
of the postnatal growth of visual cortex. Vis. Neurosci. KATZ, L. C., and C. J. SHATZ, 1996. Synaptic activity and the
15:831-839. construction of cortical circuits. Science 274:1133-1138.
EDLUND, T, and T. M. JESSEL, 1999. Progression from extrinsic LE VAY, S., T. N. WIESEL, and D. H. HUBEL, 1980. The devel-
to intrinsic signalling in cell fate specification: A view from opment of ocular dominance columns in normal and visu-
the nervous system. Cell 96:211-224. ally deprived monkeys. J Comp. Neurol. 191:1—51.
FEINBERG, I., 1983. Schizophrenia: Caused by a fault in pro- LOWEL, S., and W. SINGER, 1992. Selection of intrinsic hori-
grammed synaptic elimination during adolescence? J. Psy- zontal connections in the visual cortex by correlated neu-
chiat. Res. 17:319-334. ronal activity. Science 255:209-212.
FEINBERG, I., 1990. Cortical pruning and the development of LUSKIN, M. B., and C. J. SHATZ, 1985. Neurogenesis of the cat's
schizophrenia. Schizophrenia Bull. 16:567-568. primary visual cortex. J Comp. Neurol. 242:611-631.
DE FELIPE, J., P. MARCO, A. FAIREN, and E. G. JONES, 1997. MAFFEI, L., and L. GALLI-RESTA, 1990. Correlation in the dis-
Inhibitory synaptogenesis in mouse somatosensory cortex. charges of neighboring rat retinal ganglion cells during pre-
Cereb. Cortex 7:619-634. natal life. Proc. Natl. Acad. Sci. (USA) 87:2861-2864.
FERSTER, D., and N. SPRUSTON, 1995. Cracking the neuronal MARIN-PADILLA, M., 1988. Early ontogenesis of the human ce-
code. Science 270:756-757. rebral cortex. In Cerebral Cortex: Development and Maturation
FRANTZ, G. D., A. P. BOHNER, R. M. AKERS, and S. K. Mc- of Cerebral Cortex, Vol. 7, A. Peters and E. G.Jones, eds. New
CONNELL, 1994. Regulation of the POU domain gene SCIP York: Plenum Press, pp. 1-34.
during cerebral cortical development. J. Neurosci. 14:472- MITCHELL, D. E., and B. TIMNEY, 1984. Postnatal development
485. of function in the mammalian visual system. In Handbook of
GEINISMAN, Y., J. F. DISTERHOFT, H. J. G. GUNDERSEN, M. D. Physiology: Sec. I. The Nervous System: Vol. 3. Sensory Processes
McEcHRON, I. S. PERSINA, J. M. POWER, E. A. VAN DER ZEE, (part 1), I. Darian-Smith, ed. Bethesda, Md.: American Phys-
and M. J. WEST, 2000. Remodeling of hippocampal synapses iological Society, pp. 507-555.
after hippocampus-dependent associative learning. J. Comp. MOONEY, R., A. A. PENN, R. GALLEGO, and C. J. SHATZ, 1996.
Neurol. 417:49-59. Thalamic relay of spontaneous retinal activity prior to vi-
GOLDMAN-RAKIC, P. S., J.-P. BOURGEOIS, and P. RAKIC, 1997. sion. Neuron 17:863-874.
Synaptic development of the prefrontal cortex and the PETERS, A., C. SETHARES, and M. B. Moss, 1998. The effects
emergence of cognitive function. In Development of the Pre- of aging on layer 1 in area 46 of prefrontal cortex in the
frontal Cortex: Evolution, Neurobiology and Behavior, N. A. Kras- rhesus monkey. Cereb. Cortex 8:671 -684.
negor, G. R. Lyon, and P. S. Goldman-Rakic, eds. Baltimore: PURVES, D., L. E. WHITE, and D. R. RIDDLE, 1996. Is neural
Paul Brookes, pp. 27-47. development Darwinian? Trends Neurosci. 19:460-464.

BOURGEOIS: SYNAPTOGENESIS IN NEOCORTEX 33

QUARTZ, S. R., and T. J. SEJNOWSKI, 1997. The neural basis of human cerebral cortex from birth to 6 years examined by
cognitive development: A constructivist manifesto. Behav. correspondence analysis. Proc. Natl. Acad. Sci. (USA) 95:
Brain Sci. 20(4): 1-60. 4023-4028.
RAKIC, P., 1995. A small step for the cell, a giant leap for man- STEFAN, M. D., and R. M. MURRAY, 1997. Schizophrenia: De-
kind: A hypothesis of neocortical expansion during evolu- velopmental disturbance of brain and mind? ActaPaediatrica
tion. Trends Neurosci. 18:383-388. 422(suppl.):112-116.
RAKIC, P., J.-P. BOURGEOIS, M. F. ECKENHOFF, N. ZECEVIC, and TANAKA H., W. SHAN, G. R. PHILLIPS, K. ARNDT, O. BOZDAGI,
P. S. GOLDMAN-RAKIC, 1986. Concurrent overproduction of L. SHAPIRO, G. W. HUNTLEY, D. L. BENSON, and D. R. COL-
synapses in diverse regions of the primate cerebral cortex. MAN, 2000. Molecular modification of N-cadherin in re-
Science 232:232-235. sponse to synaptic activity. Neuron 25:93-107.
RAKIC, P., J.-P. BOURGEOIS, and P. S. GOLDMAN-RAKIC, 1994. TELLER, D. Y., 1997. First glances: The vision of infants. Invest.
Ophthalmol Vis. Sci. 38(11):2183-2203.
Synaptic development of the cerebral cortex: Implications
VAN ESSEN, D., 1985. Functional organization of primate visual
for learning, memory, and mental illness. Prog-. Brain Res.
cortex. In Cerebral Cortex, Vol. 3, A. Peters and E. G. Jones,
102:227-243.
eds. New York: Plenum Press, pp. 259-329.
RODMAN, H. R., 1994. Development of inferior temporal cor-
VANNIER, C., and A. TRILLER, 1997. Biology of the postsynaptic
tex in the monkey. Cereb. Cortex 5:484-498. glycin receptor. Int. Rev. Cytol. 176:201-244.
SALIN, P. A., and J. BULLIER, 1995. Corticocortical connections ZECEVIC, N., 1998. Synaptogenesis in layer I of the human
in the visual system: Structure and function. Phys. Rev. cerebral cortex in the first half of gestation. Cereb. Cortex
75:107-154. 8:245-252.
SCHAAL B., J. P. LECANUET, and C. GRANIER-DEFERRE, 1999. ZECEVIC, N., J.-P. BOURGEOIS, and P. RAKIC, 1989. Synaptic
Sensory and integrative development in the human fetus density in motor cortex of rhesus monkey during fetal and
and perinate: The usefulness of animal models. In Animal postnatal life. Dev. Brain Res. 50:11-32.
Models of Human Emotion and Cognition, M. Hang and R. E. ZECEVIC, N., and P. RAKIC, 1991. Synaptogenesis in monkey
Whalen, eds. Washington D.C.: American Psychological As- somatosensory cortex. Cereb. Cortex 1:510-523.
sociation, pp. 119-142. ZOLI, M., C. TORRI, R. FERRARI, A. JANSSON, I. ZINI, K. FUXE,
SHANKLE, W. R., A. K. ROMNEY, B. H. LANDING, and J. HARA, and L. AGNATI, 1998. The emergence of the volume trans-
1998. Developmental patterns in the cytoarchitecture of the mission concept. Brain Res. Rev. 26:136-147.

34 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

3 Myelination in the Developing
Human Brain
RICARDO C. SAMPAIO
AND CHARLES L. TRUWIT

ABSTRACT Myelination in the human central nervous system weeks of gestation in the spinal cord and continues well
(CNS) is a complex but orderly process, occurring in predict- into the third and fourth decades of life in the intra-
able topographical and chronological sequences. CNS myeli- cortical fibers of the cerebral cortex; but the most
nation begins as early as 12-14 weeks of gestation in the spinal
cord and continues well into the fourth decade of life. The important and dramatic changes occur between mid-
most significant period of CNS myelination, however, occurs gestation and the end of the second postnatal year, with
between midgestation and the second postnatal year. Myeli- myelination accounting in large part for the large gain
nation can be studied by various methods, including myelin- in brain weight, which more than triples during this
stained histopathological brain sections and, more recently, period. The process slows markedly after 2 years of age
magnetic resonance (MR) imaging. The latter constitutes the
focus of this chapter. (Yakovlev and Lecours, 1967).

Myelin is a wrapping of surface membrane of oligo- General rules of myelination

dendrocytes (Schwann cells in the peripheral nervous
system) around the axons, with little or no cytoplasm in According to Yakovlev and Lecours (1967), myelination
between (Everett, 1971). The process of myelin forma- is a synchronized and orderly process of maturation of
tion occurs in two, partially overlapping stages. Initially, functionally allied systems of fibers. Several general
oligodendrocytes proliferate and differentiate. Subse- rules governing the chronological and topographical
quently, myelin is synthesized. Although the precise sequences of the myelination process have been de-
chemical structure has been only partially elucidated scribed. No single rule acts alone; rather, the interplay
(Kinney et al., 1994), myelin, like other membranes, is is complex, and one rule may supersede another to dic-
composed of a lipid bilayer with several large proteins. tate the sequence of myelination within or across a par-
Proteolipid protein (PLP) and myelin basic protein ticular axonal system.
(MBP) are two such proteins, both of which are neces- As summarized by Kinney and colleagues (1994),
sary for membrane compaction and span the bilayer. these general rules are as follows: (1) Proximal path-
The outer layer of the membrane is composed mainly ways myelinate earlier and faster than do distal path-
of cholesterol and glycolipids while the inner portion ways; (2) sensory pathways myelinate before motor
of the lipid bilayer is composed mainly of phospholipids pathways; (3) projection pathways myelinate before as-
(Braun, 1984). sociative pathways; (4) central telencephalic sites my-
Myelination in the human central nervous system elinate before poles; and (5) occipital poles myelinate
(CNS) is a complex but orderly process, occurring in before frontotemporal poles.
predictable topographical and chronological se- The more proximal the components of the fiber sys-
quences. The sequence of myelination in the human tem, the shorter its myelination interval, as exemplified
brain has been carefully defined by histochemical (Kin- in (a) the visual system, in which the optic tract myeli-
ney et al., 1994; Yakovlev and Lecours, 1967) as well as nates faster than the optic radiations which myelinate
imaging (Barkovich et al., 1988; Bird et al., 1989; Martin faster than the subcortical association fibers of the vi-
et al., 1991; Nakagawa et al., 1998; Staudt et al., 1993; sual cortex; (b) the auditory system, in which the prox-
van der Knaap and Valk, 1990) methods. Histochemi- imal auditory radiation myelinates faster than Heschl's
cally, the CNS myelination begins as early as 12-14 gyrus (thalamic projections into the auditory cortex);
and (c) the pyramidal system, in which the posterior
limb of the internal capsule myelinates faster than the
RICARDO C. SAMPAIO AND CHARLES L. TRUWIT Department of bulbar pyramid which myelinates faster than the lateral
Radiology, University of Minnesota, Minneapolis, Minnesota. corticospinal tracts.

SAMPAIO AND TRUWIT: MYELINATION IN THE DEVELOPING HUMAN BRAIN 35

 Axonal myelination also proceeds from proximal to posterior limb of the internal capsule myelinates earlier
distal. Considering trans-synaptic systems, myelination and faster than the anterior limb. The body and sple-
proceeds from the neuron along the direction of im- nium of the corpus callosum, interconnecting the
pulse conduction. Thus, the proximal components of a posterior frontal and parieto-occipital hemispheres,
fiber system myelinate earlier than do distal compo- myelinate earlier and faster than the genu and rostrum,
nents. Moreover, these proximal components myelinate interconnecting the frontal poles.
faster. Corollary to this, however, are the principles of There are exceptions to these rules, however. In the
central region before pole and occipital before tem- spinal and cranial nerves, motor roots myelinate earliest
poral. For example, in the telencephalon, Heschl's gy- and rapidly, and anticipate myelination of the sensory
rus, the distal auditory thalamic projection to the roots. Maturation of myelin occurs in the optic tract
temporal lobe, myelinates after the distal optic radia- before it does in the optic chiasm. Similarly, myelina-
tion to the occipital pole. Proximal motor components tion is completed in the auditory radiation before it is
also myelinate faster than do distal sensory compo- in the brachium of the inferior colliculus. In addition,
nents. Thus, the posterior limb of the internal capsule, myelination of the callosal rostrum, genu, body, and
which contains the proximal portion of the pyramidal splenium occurs before the respective subcortical and
system, myelinates earlier and faster than Heschl's gy- central white matter origins (Kinney et al., 1994).
ms, the distal portion of the auditory system (Kinney et
al., 1994).
MR of postnatal brain development
Myelination of fiber systems mediating sensory input
to the thalamus and cerebral cortex precedes myelina- Myelination of the human brain has been studied in
tion of those fiber systems carrying output relating to vivo by means of magnetic resonance (MR) imaging
movement, and these latter fibers myelinate before the (Barkovich et al., 1988; Bird et al., 1989; Martin et al.,
association fibers. Thus, the fiber systems mediating ves- 1991; Nakagawa et al., 1998; Staudt et al., 1993; van der
tibular and acoustic input myelinate early and rapidly Knaap and Valk, 1990). The fundamental work in this
before birth, the optic radiation and the pre- and post- domain was done by Barkovich and colleagues (1988),
central cortical thalamic projections myelinate rapidly
later complemented by others (Bird et al., 1989; Huppi
during the first year after birth, and these sensory sys-
et al., 1998; Koenig et al., 1990; Martin et al., 1991;
tems anticipate the myelination of the pyramidal sys-
Nakagawa et al., 1998; Staudt et al., 1993; Takeda et al.,
tems. Postnatally, the visual and auditory systems tend
1997; van der Knaap and Valk, 1990). To better under-
to have shorter myelination intervals than do pyramidal
stand the following description, a brief review of MR
systems (Kinney et al., 1994; Yakovlev and Lecours,
1967). terminology may be helpful. Most MR images used to
Primary projection systems begin to myelinate earlier assess myelination are based on the concept of Tl or
and have shorter myelination intervals than do associ- T2 weighting (table 3.1), which reflect differences in
ative systems. The thalamocorticopyramidal fiber sys- tissue water. Tl-weighted images are typically more "an-
tems have a relatively short cycle of myelination that is atomic," whereas T2-weighted images typically show
completed during the first postnatal year. The nonspe- subtle abnormalities reflected in changes in water, or
cific thalamocorticopontine fiber systems show a long edema. More importantly, images can be Tl- or T2-
myelination cycle, prolonging into early childhood. In weighted to optimize tissue characteristics.
general, the association fibers in the supratentorial Tl-weighted images look like "cut brain." Cerebro-
brain and the reticular formation in the brainstem con- spinal fluid spaces, such as the ventricles and sulci, are
tinue to myelinate to at least the third decade (Kinney dark on Tl-weighted images. Fatty tissues are bright.
et al., 1994; Yakovlev and Lecours, 1967). Myelin, containing phospholipids, is also bright relative
Overall, myelination progresses from caudal to ceph- to other intracranial structures. Thus, by the process of
alad and from dorsal to ventral. Within any particular myelination, brain areas that are myelinated appear
region of the brain, the posterior region tends to my- bright or hyperintense relative to other areas on the
elinate first. Myelination in the central white matter of image. Moreover, areas of the brain with very tightly
the cerebral hemispheres proceeds from the region packed fiber bundles, such as the corpus callosum, ex-
around the central sulcus toward the poles, with the trude any free water from their myelin fibers and thus
occipital pole myelinating before the frontal pole, appear even brighter on the image.
which in turn myelinates before the temporal pole. In T2-weighted images are essentially inverted in their
addition, the posterior poles have shorter myelination signal characteristics. Hence, myelinated brain, with an
intervals than the anterior frontotemporal regions. The abundance of phospholipid, appears hypointense.

36 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 TABLE 3.1
MR milestones of CNS myelination

Age at Which Myelination Appears

Anatomic Region Tl-Weighted Images Tl-Weighted Images
Inferior cerebellar peduncle Present at birth Present at birth
Middle cerebellar peduncle 1 month 1-2 months
Superior cerebellar peduncle Present at birth Present at birth
Posterior cerebellar peduncle
Anterior portion Present at birth 4-7 months
Posterior portion Present at birth Birth to 2 months
Anterior limb internal capsule 2-3 months 7-11 months
Genu corpus callosum 4-6 months 5-8 months
Splenium corpus callosum 3-4 months 4-6 months
Occipital white matter
Central 3-5 months 9-14 months
Peripheral 4-7 months 11-15 months
Frontal white matter
Central 3-6 months 11-16 months
Peripheral 7-11 months 14-18 months
Centrum semiovale 2-6 months 7-11 months
Modified from Barkovich et al., 1988.

Fluid, on the other hand, is ideally imaged on T2- myelination provides an excellent parameter of brain
weighted images: The ventricles are bright. maturation.
On MR imaging, brain maturation during the first 2 The changes in signal intensity seen on Tl- and T2-
years of life consists primarily of signal changes second- weighted images closely parallel the known pattern of
ary to the process of myelination and occurs at different myelination as determined by histochemical studies. At
rates and at different times on different types of images. birth, the dorsal pons, portions of the superior and in-
Changes of brain signal due to myelination will ap- ferior cerebellar peduncles, the decussation of the su-
pear first on Tl-weighted images as increased signal perior cerebellar peduncles, the posterior limb of the
relative to gray matter. Eventually, a corresponding internal capsule, the central corona radiata, the optic
drop of T2 signal will be seen. In general, changes in chiasm, the optic tract, the ansa lenticularis, and the
white matter maturation are best seen on Tl-weighted fibers originating in the external segment and travers-
images during the first 6-8 months of life and on the ing the internal segment of the globus pallidus are par-
T2-weighted images between the ages of 6 and 18 tially myelinated. These are seen as areas of increased
months (Barkovich et al., 1988). Most recently, a new signal intensity on the Tl-weighted images and as de-
MR imaging technique, diffusion-weighted imaging, creased signal intensity on the T2-weighted images of
has been used for assessment of brain maturation in newborns (figures 3.1, 3.2, and 3.3).
humans (Huppi et al., 1998; Takeda et al., 1997). On Tl-weighted images, unmyelinated white matter
Diffusion-weighted imaging is sensitive to brownian mo- is hypointense relative to gray matter. With myelination,
tion of water. Acute injury to the brain, such as edema phospholipids occupy the space otherwise occupied by
and/or cytotoxicity, results in restricted brownian mo- interstitial water and the Tl signal increases, from hy-
tion, and hence altered diffusion, of extracellular water. pointense to hyperintense, relative to gray matter. On
Similarly, on diffusion-weighted images, it is possible to T2-weighted images, the reverse picture can be seen,
map brain maturation, as unbound water tends to fol- with unmyelinated white matter being hyperintense
low pathways of least resistance. With myelination, water relative to gray matter. As the white matter matures, its
is more likely to follow an axonal bundle than to cross T2 signal decreases from hyperintense to hypointense
it. This so-called diffusion anisotropy becomes apparent relative to gray matter, passing through a phase of iso-
within white matter tracts. In fact, the signal changes intensity to gray matter.
on diffusion-weighted images may precede correspond- The causes of the Tl shortening (increase in Tl sig-
ing changes on conventional Tl-weighted images. Ac- nal intensity) associated with myelination are related to
tive myelination from the perinatal period onward can but, at least in part, independent from the causes of T2
be identified by imaging, and thus the progression of shortening (decrease in T2 signal intensity). As already

SAMPAIO AND TRUWIT: MYELINATION IN THE DEVELOPING HUMAN BRAIN 37

FIGURE 3.1 Sagittal Tl-weighted image (A) of an 8-week-old the T2-weighted image (C). Axial T2-weighted images (D-F)
boy shows thin, and not yet myelinated corpus callosum. Hy- show subtle myelination of internal capsule and basal ganglia,
perintense pituitary is normal at this age. Coronal inversion but the white matter is still hyperintense overall. Note early
recovery images (B, C) show myelination of corticospinal myelination (mottled hypointensity) of corona radiata (ar-
tracts (arrows), but the remainder of white matter is still hy- rows, F).
pointense on the Tl-weighted image (B) and hyperintense on

mentioned, brain maturation occurs at different rates myelination with even a temporary decline in brain
and times on Tl-weighted and T2-weighted images. The weight. It is, however, probable that the changes in Tl
discrepancy is mild in certain areas (2 months in the and T2 values are due not only to a decreasing water
corpus callosum) but marked in others (6-10 months content of the brain, but also to a change in biochem-
in the centrum semiovale) (Barkovich et al., 1988). ical composition caused by the arrival of precursors of
From estimations of Tl and T2 relaxation times, it is myelin constituents and materials necessary in the pro-
evident that a significant and global decrease in Tl and cess of myelin formation (van der Knaap and Valk,
T2 relaxation times occurs during the first few months 1990). When myelin is deposited, another important
of life, preceding the myelination of large parts of the decrease in Tl and T2 values occurs. This decrease is
brain (van der Knaap and Valk, 1990). This observation presumed to be related to the degree of myelination.
is in accord with animal experimental work, which The further Tl shortening correlates temporally with
shows that an impressive decrease in water content of the increase in cholesterol and glycolipids that accom-
the brain occurs just prior to the onset of a wave of pany the formation of myelin from oligodendrocytes

38 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

FIGURE 3.2 A sagittal Tl-weighted image (A) of the same myelination by virtue of relatively hypointense white matter.
child at 8 months shows normal myelination of corpus cal- Axial T2-weighted images (C, D) show that much of the sub-
losum (pituitary is normal on adjacent slice, not shown). A cortical white matter is not yet myelinated.
coronal inversion recovery image (B) shows considerable

and probably reflects shortening of the Tl relaxation nation, whereas T2-weighted images seem better suited
time of the water in the white matter resulting from a for evaluating the associated changes in water content
very large interaction with myelin lipid (Koenig et al., (Barkovich et al., 1988). Diffusion-weighted images are
1990; Nakagawa et al., 1998; Poduslo and Jang, 1984). able to detect the process of myelination owing to their
The further decrease in T2 signal intensity of myelinat- sensitivity to changes of water diffusion induced by the
ing white matter correlates temporally with the tight- myelin membranes (Huppi et al., 1998; Takeda et al.,
ening of the spiral of myelin around the axon, which is 1997). They may be more sensitive than Tl-weighted
accompanied by loss of white matter water content. As images to white matter myelination, but their use has
myelin matures, it becomes increasingly hydrophobic been limited essentially to experimental work.
because of the development of the inner layer of phos- Neonatal posterior fossa structures that exhibit high
pholipids. This hydrophobic layer results in the pres- Tl and low T2 signal intensity at birth include the dor-
ence of fewer aqueous protons and a diminution in sal brainstem and the inferior and superior cerebellar
signal intensity on the T2-weighted images secondary to peduncles (Barkovich et al., 1988). An increase in Tl
a combination of shortened T2 relaxation time and a signal intensity denoting early myelination of the deep
decrease in water content. Tl-weighted images are a cerebellar white matter appears near the end of the first
better way of detecting the primary process of myeli- month of life and steadily increases, with high signal

SAMPAIO AND TRUWIT: MYELINATION IN THE DEVELOPING HUMAN BRAIN 39

FIGURE 3.3 A sagittal Tl-weighted image (A) of another corpus callosum genu and splenium, deep white matter, and
child at 13 months shows normal corpus callosum. Coronal internal capsules. Note the still incompletely myelinated sub-
inversion recovery image (B) shows more advanced myelina- cortical white matter and persistent terminal zones of myeli-
tion of white matter, even at the subcortical level. Axial T2- nation at peritrigonal level (arrows, D).
weighted images (C, D, E) show expected myelination of

intensity developing in the subcortical white matter of In the supratentorial brain, the decussation of the
the cerebellar folia by the third month. At 3 months of superior cerebellar peduncles, the ventral lateral region
age the cerebellum has a Tl appearance similar to that of the thalamus, the dorsal putamen, and the posterior
seen in the adult. On T2-weighted images, the middle limb of the internal capsule exhibit high Tl signal in-
cerebellar peduncle begin to decrease in signal inten- tensity at birth. To a lesser extent, these structures also
sity during the second month of life. Low T2 signal in- demonstrate areas of low T2 signal intensity at birth.
tensity in the subcortical white matter of the cerebellar The development of Tl and T2 signal changes proceeds
folia (arborization) begins to develop at approximately rostrally from the pons along the corticospinal tracts
the eighth month (5 months later than on Tl-weighted into the cerebral peduncles, the posterior limb of the
images), and the cerebellum reaches an adult appear- internal capsule, and the central portion of the cen-
ance at approximately 18 months (15 months later than trum semiovale. The white matter of the pre- and post-
on Tl-weighted images). Signal intensity in the ventral central gyri is of high Tl and low T2 signal intensity,
pons increases less rapidly, occurring during the third respectively, compared to the surrounding cortex by
through the sixth months. about 1 month of age. The change to high Tl signal

40 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

intensity in the subcortical motor tracts is essentially weighted images after 8 months, consisting of increas-
complete by 3 months of age. By age 2 months, patches ing signal intensity in the most peripheral regions of
of low T2 signal intensity are seen in the central cen- the frontal, temporal, and parietal white matter. On T2-
trum semiovale. weighted images, peripheral extension of the low signal
In infants less than 1 month old, high Tl signal in- intensity into the subcortical white matter begins at
tensity is present in the optic nerve, optic tracts, and about 1 year and is essentially complete by 22-24
optic radiations. By age 3 months, the occipital white months.
matter surrounding the calcarine fissure is of high Tl On Tl-weighted images, a brain pattern essentially
signal intensity. Low T2 signal intensity is seen in the identical to that seen in adults is seen by approximately
optic tracts at age 1 month, the decrease in signal in- 8 months of age. With the exception of the so-called
tensity extends posteriorly along the optic radiations terminal zones (persistent areas of high T2 signal in-
during the subsequent 2 months, and by 4 months of tensity in the white matter dorsal and superior to the
age the calcarine fissure shows some low T2 signal in- trigones of the lateral ventricles), white matter matu-
tensity. Diffusional anisotropy denoting myelination is ration, as assessed by MR imaging, is complete by the
present in the optic radiations in infants less than 1 end of the second year of life. The terminal zones prob-
month old. ably represent the areas of known delayed myelination
The posterior limb of the internal capsule is of high of fiber tracts involving the association areas of the pos-
Tl signal intensity at birth; high signal intensity does terior and inferior parietal and posterior temporal cor-
not develop in the anterior limb of the internal capsule tex, as described by Yakovlev and Lecours (1967). These
until 2-3 months of age. The more anterior portion of areas of high T2 signal intensity are seen throughout
the posterior limb of the internal capsule contains a the first decade and, in some subjects, into the second
thin strip of T2 hypointensity by approximately 7 decade of life.
months; progressive thickening of the hypointense area It is difficult to distinguish fiber bundles (e.g., medial
continues up to 10 months of age. The anterior limb of lemniscus or optic radiation) from surrounding tissue
the internal capsule is completely T2-hypointense by 11 by MR in adults. This is because the white matter sur-
months in all subjects; T2 hypointensity can be detected rounding each fiber bundle has the same signal inten-
as early as 7 months in some subjects but is always pre- sity as the fiber bundle. Early in development, fiber
ceded by detectability of low signal intensity in the pos- bundles can be distinguished from surrounding white
terior limb. matter because myelination of fiber bundles precedes
The splenium of the corpus callosum shows high Tl that of surrounding white matter in the neonatal pe-
signal intensity in all infants by 4 months. The increase riod; later, however, they become indistinguishable
in signal intensity proceeds rostrally; the genu is always owing to the progression of myelination in the sur-
of high signal intensity by age 6 months. Likewise, on T2- rounding white matter. The time of blurring (when the
weighted images, myelination proceeds in a posterior- fiber bundle becomes indistinguishable from surround-
to-anterior fashion. The splenium of the corpus ing white matter) may be a useful parameter of brain
callosum shows low T2 signal intensity by age 6 months maturation in infancy. Eighteen fiber bundles were an-
and the genu by age 8 months. The basal ganglia begin alyzed by Nakagawa and colleagues (1998). These au-
to diminish in T2 signal intensity relative to the subcor- thors found that not only do different bundles start to
tical white matter at 7 months of age. The basal ganglia myelinate at different times and myelinate at different
appear essentially isointense with the subcortical white rates, but also they blur at different times, with some of
matter by the age of approximately 10 months. the bundles showing no blurring (e.g., posterior limb
Maturation of the subcortical white matter, other of internal capsule).
than the visual and motor regions, begins at 3 months
on Tl-weighted images and at 9-12 months on T2- Sequences of CNS myelination
weighted images. The deep white matter matures in a
posterior-to-anterior direction, with the deep occipital Historically, in elucidating the factors that determine
white matter maturing first and the frontal and tem- the topographical and chronological sequences of CNS
poral white matter last. On Tl-weighted images, pe- myelination, the process of myelination has been con-
ripheral extension and increasing complexity of sidered relative to the development of human behavior.
arborization of the subcortical white matter continue This approach resulted in the analysis of the sequence
approximately until age 7 months in the occipital white of myelination in terms of relationships among ana-
matter and 8-11 months in the frontal and temporal tomic systems associated with different functions.
white matter. Only minimal changes are seen on Tl- Yakovlev and Lecours (1967) held that the myelination

SAMPAIO AND TRUWIT: MYELINATION IN THE DEVELOPING HUMAN BRAIN 41

proceeds along a hierarchy of increasingly complex vestibulocochlear system. In sharp contrast, the outer
CNS functions ending in the maturation of association division, containing spino-, olivo- and reticulocerebellar
and intracortical fibers critical to the highest intellec- fibers, shows no myelinated fibers until the eighth fetal
tual functions, and that myelination represents the month and thereafter continues to myelinate until
anatomic correlate of neurophysiological maturation. about the third postnatal month; its cycle is long and
Functionally related systems myelinate together, and comparable to that of the medial lemniscus. Myelina-
the timing of myelination reflects the position of the tion of the superior cerebellar peduncles begins in the
fiber system in the hierarchy of the functional organi- eighth fetal month, approximately 6 weeks later than
zation of the developing nervous system. They referred the inner division of the inferior peduncle but before
to the temporal aspect of myelination in a fiber system, myelinated fibers are present in the outer division of
from its onset to completion, as its myelogenetic cycle and this peduncle. In comparison with the superior and in-
observed that this cycle may have an early or late onset ferior peduncles, the middle cerebellar peduncles demon-
with a rapid or slow course. Different sites not only be- strate the most protracted cycle of myelination.
gin myelination at different times, but also progress to Myelinated fibers in the transverse bundles of the pons
maturation over different time intervals (Kinney et al., appear only during the first postnatal month, and my-
1994; Yakovlev and Lecours, 1967). The early onset of elination extends into the middle cerebellar peduncles
myelination does not predict early myelin maturation during the second postnatal month. The pontocerebel-
(Kinney et al., 1988; Yakovlev and Lecours, 1967). lar fibers continue to gain slowly in intensity of myeli-
Myelin sheaths appear in the motor root fibers of the nation at least until the fourth postnatal year.
spinal nerves at the end of the fourth fetal month, while In man the reticular formation of the brainstem exhib-
the sensory fibers begin to myelinate at the end of the its a very long cycle of myelination and, in this respect,
fifth month. The motor nerve roots reach their adult matches the protracted cycle of myelination in the
pattern of myelination at about term while the sensory commissural and association fiber systems of the
nerve rootlets continue to myelinate for several months telencephalon.
after birth. Among the cranial nerves, the roots of the In the supratentorial brain, no myelinated fibers are
eighth pair are the first to show myelinated fibers. At encountered before the seventh fetal month. Early in
the end of the fifth fetal month, the roots of both divi- the seventh month the first myelinated fibers appear in
sions of the eighth nerve are myelinated. The oculo- the fasciculus habenulointerpeduncularis, which my-
motor nerves (III, IV, VI) and the motor division of the elinates rapidly and completes its cycle of myelination
trigeminal nerve myelinate next, at about the same before term. In the last trimester intense myelination
time. As in the spinal roots, the cranial motor roots occurs in the subthalamic region, thalamus, and
seem to myelinate at a faster pace than the sensory pallidum.
roots. The cycle of myelination of the cranial nerve The brachia of the inferior colliculi and the medial
roots appears to be completed early in the first postnatal geniculate nuclei are among the earliest extrinsic tha-
year. lamic afferents to myelinate. They begin to myelinate in
Except for the dorsal root fibers in the posterior col- the seventh fetal month, well in advance of the terminal
umns of the spinal cord, there are no myelinated fibers fibers of the medial lemniscus and optic tract, and com-
in the CNS before the end of the fifth fetal month. In plete the cycle at about the fourth postnatal month. The
the brain, the vestibulocochlear system of fibers is the first contrast between early myelination of the acoustic and
to demonstrate myelination, having not only an early late myelination of the optic colliculi is noteworthy. My-
beginning but also a very short cycle of myelination. elination of the optic nerves and tracts, of the superior
The fibers of this system in the brainstem complete the colliculi and their brachia, and of the lateral geniculate
cycle of myelination at about the middle of the ninth nuclei begins late in the ninth fetal month. However,
fetal month, and from the sixth fetal month to term the optic system, as a whole, completes the cycle rapidly
they dominate the myelin-staining preparations of the near the third postnatal month. By the middle of the
brainstem. eighth fetal month, the myelinated fibers of the medial
The cycles of myelination of the inferior, middle, and lemniscus reach their thalamic nuclei. Their myelina-
superior cerebellar peduncles exhibit remarkable dif- tion is complete by the eighth postnatal month. The
ferences. The inner, vestibulocerebellar division of the reticulum and the intrinsic fibers of the thalamus my-
inferior cerebellar peduncle begins to myelinate early in the elinate at a slower pace.
sixth fetal month and attains an adult pattern of my- Myelination of thalamocortical radiations is related to
elination by the end of eighth fetal month. The cycle the thalamic nucleus involved and has been described
of the inner division is about the same as that of the in terms of so-called "specific" and "nonspecific" tha-

42 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

lamic nuclei. Projections from the specific nuclei of the the corticopontine tracts myelinate latest and longest. The
basal complex (or "relay" nuclei) myelinate earlier and cycle of myelination of the corticopontine tracts ap-
exhibit a shorter cycle than do projections from the pears to be synchronized with the cycle of myelination
nonspecific nuclei of the anteromedial and dorsal com- of the nonspecific thalamocortical projections from the
plex. Among the specific thalamic projections, the optic medial dorsal and pulvinar nuclei.
radiations to the primary visual cortex of the occipital In the white matter of the telencephalon, myelination
lobe exhibit the shortest cycle of myelination, begin- of the long association and commissural fiber systems
ning late in the first postnatal month and reaching ma- is protracted, lasting until at least the end of the first
turity about the fourth postnatal month. Myelination of decade of life. The commissural fibers in the splenium
the projections from the ventrobasal complex of the of the corpus callosum and forceps major begin to my-
thalamic nuclei to the postcentral cortex appears in elinate at about the fourth postnatal month. Myelina-
the lenticulothalamic region of the posterior limb of tion spreads gradually and slowly from the splenium
the internal capsule at the end of the ninth month and toward the genu, the rostrum, and forceps minor of the
during the tenth month rapidly extends into the core corpus callosum. From pathologic studies, myelination
of the postcentral gyrus; but, in contrast to the optic of commissural fibers in the corpus callosum seems to
radiations, these projections seem to complete the cycle continue slowly even after the first decade. The intra-
only at about the first postnatal year. Myelination of the cortical neuropil of the anterolateral convexity of the
acoustic radiations from the medial geniculate to frontal lobes, of the inferior parietal lobes, and of the
Heshl's gyrus in the temporal lobe and from the latero- basolateral convexity shows even longer myelination
ventral complex to the precentral cortex of the frontal cycles.
lobe exhibits a still longer cycle, which is completed at
about 4 years of age. It is also noteworthy that myeli- Summary
nation of the cortical end of the acoustic system in the
temporal lobe is protracted beyond the first postnatal The unique ability of MR imaging to reveal not only
year, in sharp contrast to the visual system, which my- gross structural organization of the human brain but
elinates rapidly soon after birth in one short spurt from also macroscopic changes in intra- and extracellular wa-
retina to its end about the calcarine fissure. ter (diffusion-weighted MR imaging) enables us to see
The first myelinated fibers in the internal capsule ap- the in vivo human brain as never before. In this chapter,
pear in the ninth fetal month in the lenticulothalamic the process and sequence of brain myelination have
sector of the posterior limb. From the ninth fetal month been reviewed, as has the correlate appearance on MR
to the fourth postnatal month, the myelination spreads images. In tandem with the structural changes that oc-
fanwise posteriorly into the retrolenticular sector of the cur with myelination are functional changes of the hu-
posterior limb and, much more slowly, anteriorly into man central nervous system. Over the past five years,
the anterior limb. The thalamocortical radiations reach and certainly over the next five years, cognitive neuro-
the postcentral cortex (somesthetic area) during the scientists and their neuroradiological counterparts have
first postnatal month, and later appear in the precentral had and will have incredible opportunities to "see"
gyrus (propriokinesthetic area). During the first post- brain function with the relatively new field of functional
natal month there is also a spurt of myelination of the MR imaging. While fMRI of the pediatric brain remains
corticofugal fibers from the postcentral and the precen- in its infancy (see Casey, Thomas, and McCandliss,
tral cortices. Myelination of the fibers in the anterior chapter 10), it is anticipated that the field of develop-
limb of the internal capsule lags behind that of the pos- mental fMRI will emerge and offer the opportunity to
terior limb but accelerates from the fourth postnatal marry anatomic structure with histologic maturation
month, and by the eighth month is similar to that of and cognitive development.
the posterior limb.
Myelination of the pyramidal tract is synchronized
closely with that of specific thalamocortical projections
REFERENCES
from the ventrobasal complex of thalamic nuclei. My-
elination of the pyramidal tract fibers appears first at BARKOVICH, A., B. KJos, D.JACKSON, and D. NORMAN, 1988.
the pontine level and spreads cephalad into the cere- Normal maturation of the neonatal and infant brain: MR
imaging at 1.5 T. Radiology 166:173-180.
bral peduncles and internal capsule and caudad into
BIRD, R., M. HEDBERG, B. DRAYER, P. KELLER, R. FLOM, and
the bulbar pyramids. Below the pons, fibers of the bul- J. HODAK, 1989. MR assessment of myelination in infants
bar pyramids complete their myelination cycle at about and children: Usefulness of marker sites. Am. J. Neuroradiol.
1 year. Of all the long descending transcapsular tracts, 10:731-740.

SAMPAIO AND TRUWIT: MYELINATION IN THE DEVELOPING HUMAN BRAIN 43

BRAUN, P. E., 1984. Molecular organization of myelin. In My- MR imaging of the brainstem: Normal postnatal develop-
elin, 2d ed., P. Morell, ed. New York: Plenum Press, pp. 97- ment. Neuroradiology 33:391-395.
116. NAKAGAWA, H., S. IAWSAKI, K. KICHKAWA, et al., 1998. Normal
EVERETT, N. B., 1971. Histology and cytology of neurons and myelination of anatomic nerve fiber bundles: MR analysis.
neuroglia. In Functional Neuroanatomy, N. B. Everett, ed. Am. J. Neuroradiol 19:1129-1136.
Philadelphia: Lea & Febiger, pp. 16-28. PODUSLO, S., and Y.JANG, 1984. Myelin development in infant
HUPPI, P., S. MAIER, S. PELED, G. P. ZIENTARA, P. D. BARNES, brain. Neurochem. Res. 9:1615-1626.
F. A. JOLESZ, and J. J. VOLPE, 1998. Microstructural devel- STAUDT, M., C. SCHROPP, F. STAUDT, N. OBLETTER, K. BISE,
opment of human newborn cerebral white matter assessed and A. BREIT, 1993. Myelination of the brain in MRI: A
in vivo by diffusion tensor magnetic resonance imaging. Pe- staging system. Pediatr. Radiol. 1993:169-176.
diatr. Res. 44:584-590. TAKEDA, K., Y. NOMURA, H. SAKUMA, T. TAGAMI, T. OKUDA,
KINNEY, H., B. BRODY, A. KLOMAN, and F. GILLES, 1988. Se- and T. NAKAGAWA, 1997. MR assessment of normal brain
quence of central nervous system myelination in human in- development in neonates and infants: Comparative study of
fancy: Patterns of myelination in autopsied infants. J. Tl- and diffusion-weighted images. J. Comput. Assist. Tomogr.
Neuropathol. Exp. Neurol. 47:217-234. 21:1-7.
KINNEY, H., J. KARTHIGASAN, N. BORENSHTEYN, J. FLAX, and VAN DER KNAAP, M., and J. VALK, 1990. MR imaging of the
D. KIRSCHNER, 1994. Myelination in the developing human various stages of normal myelination during the first year
brain: Biochemical correlates. Neurochem. Res. 19:983-996. of life. Neuroradiology 31:459-470.
KOENIG, S., R. BROWN, M. SPILLER, and N. LUNDBOM, 1990. YAKOVLEV, P. I., and A. R. LECOURS, 1967. The myelogenetic
Relaxometry of brain: Why white matter appears bright on cycles of regional maturation of the brain. In Regional De-
MRI. Magn. Reson. Med. 14:482-495. velopment of the Brain in Early Life, A. Mankowski, ed. Phila-
MARTIN, E., S. KRASSNITZER, P. KAELIN, and C. BOESCH, 1991. delphia: Davis, pp. 3-69.

44 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

4 Morphological Changes of the Human
Hippocampal Formation from
Midgestation to Early Childhood
LASZLO SERESS

ABSTRACT Cytoarchitectonic layers of the human hippo- prematurely born infants survive. The aim of this chap-
campal formation are formed by the 24-25th fetal weeks. Con- ter is to provide age-specific morphological data against
sequently, cell formation in the hippocampal ventricular zone which behavioral scientists might correlate their rele-
is occasional after the 24th week. Cell formation in the hilar
region is continuous perinatally, but neuronal cell formation vant behavioral observations. The results of animal ex-
and cell death occur at a very low rate. Cell migration from periments are only briefly reviewed.
the proliferative zones lasts until the 32-34th fetal weeks in
Ammon's horn. Immature cells accumulate in the hilus of the Neuronal cell formation
dentate gyrus throughout the first year, when the subgranular
hilar zone appears cell-free, suggesting that immature cells from in the hippocampal formation
the hilus migrate to the granule cell layer and differentiate into
granule cells. Dendritic development and synapse formation In animal experiments, multiplying neurons can be
last for several years. Light microscopic changes of dendrites marked by isotope-labeled thymidine or the uridine an-
of the hippocampal neurons have been reported until the alogue bromodeoxyuridine (BrdU). It has been shown,
fifth postnatal year. Although the hippocampal formation of
newborn infants has the necessary synaptic connections for both in rodents and primates, that a majority of neurons
memory formation, the number of postnatal morphological in the entorhinal cortex, subicular complex, and Am-
changes suggests a significant modification of hippocampal mon's horn are formed before birth (Angevine, 1975;
circuits between the neonatal period and adulthood. Rakic and Nowakowski, 1981). In rodents, a few hip-
pocampal neurons continue to form in the last days of
The hippocampal formation plays an important role pregnancy, but no neurons are formed postnatally (An-
in the mechanisms of memory and learning. Experi- gevine, 1975). In primates, all neurons are formed in
mental data in rodents and primates as well as clinical the first half of pregnancy (Rakic and Nowakowski,
data in humans suggest that an individual with an im- 1981). There is only one region of the hippocampal
mature or lesioned hippocampus is unable to perform formation where postnatal neuronal formation occurs,
memory and learning tasks as well as a normal adult. and that is the dentate gyrus. The major difference be-
Brain development does not stop at birth, and recent tween rodents and primates in the formation of granule
data suggest that development may last until adulthood. cells is that in rodents approximately 85% of these neu-
In harmony with that notion, accumulating data indi- rons are formed postnatally, whereas similar numbers
cate a prolonged development of the human hippo-
of cells are formed prenatally in primates (Bayer, 1980;
campal formation that includes cell formation, cell
Rakic and Nowakowski, 1981). In rodents, the process
migration, and synapse formation. This chapter sum-
of granule cell formation lasts for 3 weeks, whereas in
marizes the morphological changes that occur from the
monkeys newly formed granule cells have been visual-
24th week of gestation to the period when further
ized through the first postnatal month.
changes in the hippocampal formation are not detect-
In humans, data about neuronal formation are based
able under the light microscope and the structure ap-
on the first appearance of stained neurons at their final
pears to be adult-like. The 24th week has been chosen,
cytoarchitectonic position; this is because direct label-
as that week is the approximate time after which most
ing of dividing cells is not possible. In the human hip-
pocampus, pyramidal neurons already form distinct cell
layers by the 15th gestational week. During the 23-25th
LASZLO SERESS Central Electron Microscopic Laboratory, gestational weeks the pyramidal cell layer of Ammon's
Faculty of Medicine, University of Pecs, Pecs, Szigeti, Hungary. horn and the cytoarchitectonics of the subiculum and

SERESS: HIPPOCAMPAL FORMATION 45

entorhinal cortex are close to what has been observed zone and move toward the developing dentate gyrus
in adults (Arnold and Trojanowski, 1996a; Humphrey, (figure 4.2C). There are labeled cells in all layers of
1967). The only exception to this general pattern con- the subicular complex and entorhinal cortex (figure
cerns the granule cell layer of the dentate gyrus, which 4.3A,B; see color plate 2). Often, the type of labeled
first appears during the 13-14th gestational weeks and cells, based on the morphology of cell nuclei, is unclear;
continues to grow after birth (Humphrey, 1967; Seress, but in several cases, glial and endothelial cells are pos-
1992). However, formation of cell layers includes mi- itively identified among the labeled elements (figure
gration of neurons from the place of final cell division. 4.3A). While there are no multiplying cells in the hip-
Therefore, in humans, there are no direct data about pocampal ventricular zone, a large mass of cells persists
the exact time of final cell multiplication in the hip- in the subventricular zone underlying the parahippo-
pocampal formation. campal formation and the adjacent neocortex (figure
With recent developments in immunocytochemistry, 4.3C; see color plate 2). In that subventricular cell mass
it is possible to detect cell multiplication in paraffin- at least half of the cells are immunolabeled by Ki-67,
embedded postmortem tissue. Ki-67 is a commercially suggesting very active cell formation (figure 4.3C).
available mouse monoclonal antibody that reacts with Large numbers of cells are labeled in the temporal cor-
an antigen, mainly present in the nucleoli (Verheijen tex, especially in the upper three layers. There are pro-
et al., 1989a,b). This monoclonal antibody is widely portionately fewer labeled cells in the hippocampal
used in histopathology, and comparison with other cell formation in 28-32-week and 34-week-old infants. In
proliferation markers concluded that Ki-67 is the best 36-40-week-old newborns practically all labeled cells
proliferation marker in conventional histological prep- are glial or endothelial cells in the subicular complex
arations (Brown and Gatter, 1990; Rose, Maddox, and and entorhinal cortex. Similarly, only a few glial ele-
Brown, 1994). In our studies we have used the Ki-67 ments are labeled in Ammon's horn, but no neurons
antibody (MIB-1; Immunotech) to detect cell formation (figure 4.4C; see color plate 3). The dentate gyrus con-
in postmortem brains. The human hippocampi have sists of several labeled cells in the hilar region, both
been obtained from autopsy, 6-12 hours after death. In inside the granule cell layer and in the molecular layer
the current study, only those patients who had no his- (figure 4.4A; see color plate 3). Based on their nuclear
tory of brain-related disorders and in whom autopsy morphology, these cells can be neurons, although no
confirmed the cause of death were included. Our ma- cells were immunoreactive for both Ki-67 and neuron-
terial consists of brains from premature infants (24-35 specific enolase (NCL-NSE2; Novocastra) in children
weeks), newborns (36-40 weeks), infants (1-11.5 born at term who died in a few hours after birth. How-
months), young children (1-10 years), and adults (ta- ever, in the very early stage of cellular development,
ble 4.1). The gestational age of premature infants and when cell nuclei are labeled by Ki-67, the cytoplasm may
newborns is based on postovulation time and on so- not be sufficiently mature to show enolase activity. Most
matic measurements at birth, and considered to be labeled cells are glial cells in the hilus and the molec-
accurate by ± 1 week. In this chapter, the term "hip- ular layer of the dentate gyrus of newborn infants (fig-
pocampal formation" refers to the dentate gyrus, Am- ure 4.4B,C; see color plate 3). In the postnatal period,
mon's horn, the subicular complex, and the entorhinal labeled cells are less and less frequent in the dentate
cortex. The terminology suggested by Amaral (1990) gyrus. There are a few immunopositive cells in every
has been applied to the subdivisions of each structure. section of the hilus in 3- or 3.5-month-old children (fig-
There is a significant increase in the volume of the ure 4.5B; see color plate 4), but most labeled cells are
hippocampal formation from fetal to postnatal devel- in the molecular layer (figure 4.5A; see color plate 4).
opment, but the borders of subdivisions do not change In most cases, cellular morphology of labeled cells
and are distinguishable from the earliest age group indicates that those labeled are glial cells (figure
demonstrated in this study (figure 4.1). In 24-week-old 4.5A,B,C; see color plate 4). In a few cases the labeled
infants, there are several labeled cells in all parts of the nuclei indicate neurons both in the dentate gyrus (fig-
hippocampal formation. Many cells are labeled in the ure 4.5F; see color plate 4) and the entorhinal cortex
ventricular zone underlying the hippocampal forma- (figure 4.5G; see color plate 4). Double labeling with
tion (figure 4.2D; see color plate 1) and in the hilar neuron-specific enolase is unlikely; therefore, our aim
region (figure 4.2A,B; see color plate 1). There are only is to identify the Ki-67 labeled cell nuclei in the electron
a few labeled cells inside the pyramidal layer of Am- microscope, since the nuclear ultrastructures of glial
mon's horn (figure 4.2C; see color plate 1). Most la- cells and young neurons are distinguishable from each
beled cells in the CA1-3 areas are in the subplate zone other (Eckenhoff and Rakic, 1988). It is interesting that
where immature neurons pass from the ventricular the number of labeled glial cells is much higher in the

46 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 TABLE 4.1
Summary of personal data, histological stainings, and clinical diagnoses verified by autopsy

No. Case Age Diagnosis Stain

1 Z. K. 24 gestational weeks IRDS CV, MIB-1
2 H.I. 28 weeks IRDS CV, MIB-1
3 K. L. 30 weeks IRDS CV
4 T. R. 32 weeks Sepsis, pneumonia, CHD CV
5 O.R. 34 weeks Pneumonia, IRDS CV, MIB-1
6 K. L. 36 weeks IRDS, CHD CV
7 J.P- 36 weeks Respiratory distress, asphyxia CV, MIB-1
8 T. V. 38 weeks CHD, IRDS CV, MIB-1
9 R. M. 38 weeks Esophageal atresia, pneumonia CV
10 B. B. 38 weeks Spina bifida CV, MIB-1
11 B. B. 39 weeks Respiratory distress, asphyxia CV, MIB-1
12 SZ. T. 39 weeks Pneumonia, asphyxia CV
13 K. M. 40 weeks Respiratory distress, asphyxia CV
14 A. R. 40 weeks CHD CV
15 L. D. 1 postnatal week BPD CV, MIB-1
16 T. T. 1 week CHD CV, MIB-1
17 X.Y. 1 week SIDS Rapid-Golgi
18 K. I. 1 month Pneumonia CV
19 G. U. 1 month SIDS Rapid-Golgi
20 B.J. 2 months Sepsis CV
21 S. P. 2.5 months CHD Rapid-Golgi
22 T. M. 3 months Pneumonia, muscular dystrophy CV
23 N. L. 3 months CHD CV, MIB-1
24 K. F. 3 months CHD CV, MIB-1
25 H.G. 5 months Respiratory distress, asphyxia CV, MIB-1
26 J. M. 7 months SIDS Rapid-Golgi
27 B. K. 8.5 months Pneumonia CV
28 F.E. 11.5 months Muscular atrophy CV, MIB-1
29 S. K. 11 months Ileus CV
30 T. P. 15 months Pneumonia Rapid-Golgi
31 M. P. 3 years Intestinal neoplasm Rapid-Golgi
32 L. H. 5 years Drowning Rapid-Golgi
33 N. D. 9 years CO intoxication Rapid-Golgi
34 A. N. 10 years Traffic accident Rapid-Golgi
35 K. I. 47 years Heart attack CV, Rapid-Golgi
CV — cresyl-violet staining; MIB-1 = immunostaining with Ki-67 antibody; IRDS = infant respiratory distress syndrome;
CHD = congenital heart disease; BPD = bronchopulmonary dysplasia; SIDS = sudden infant death syndrome.

granule cell and molecular layers of the dentate gyms the labeling index is above 35%, while the hippocampal
in the 11.5-month-old child (figure 4.5D,E; see color formation shows only a few labeled cells (Abraham et
plate 4) than in either the 3- or the 5-month-old chil- al., 1999). There were only a few cases when material
dren. The large number of small, round, darkly stained could be obtained from children older than 1 year, and
labeled cells in the upper molecular layer suggests an those specimens were not suitable for immunostaining.
active cell multiplication of oligodendroglia cells that Therefore, we have no direct information about cell for-
might be related to active myelin formation during this mation in children older than 1 year.
age span. Considering the total number of neurons of
the granule cell layer of a newborn child, the labeling Evidence of cell death
index per section is in the range of 1:1000. In later age
groups this index is lower. The cerebellum of the same Cell formation is always accompanied by cell death.
children has always been processed for immunostaining Therefore, we have examined the frequency of occur-
together with the hippocampus. Sparsity of immuno- rence of pycnotic nuclei, which indicate that cell death
reactive cells in the hippocampal formation is not a has occurred in the hippocampal formation. In those
methodological failure: In the external germinal layer cases where hypoxic damage was not significant we have
of the cerebellum of the 3- and 5-month-old children, found only a few pycnotic nuclei inside the granule cell

SERESS: HIPPOCAMPAL FORMATION 47

 FIGURE 4.2 Photomicrographs of Ki-67 labeled cells in cresyl
violet-stained coronal sections of the hippocampal formation
of the 24-week-old child. (A) Large numbers of labeled cells
are in the hilus (H). (B) Higher magnification reveals differ-
ences in the nuclear structure. Some of the labeled cells ap-
FIGURE 4.1 Camera lucida drawings of the hippocampal for- pear to be neuronal precursors (arrows), whereas others are
mation and the adjacent parahippocampal gyrus in a 24-week- glial cells (open arrows). (C) There are a few labeled cells
old (A), newborn (B), and 1-year-old child (C). CA1-CA3 = inside the pyramidal cell layer (str. pyr. = stratum pyramidale
subfields of Ammon's horn; DG = dentate gyrus; GL = gran- of Ammon's horn) of the CA1 area (arrows). Most labeled
ule cell layer; H = hilus of the dentate gyrus; Ent. c. = en- cells are in the intermediate and subplate zones, showing a
torhinal cortex; ML = molecular layer; Sub. = subiculum; band of probably migrating immature cells (arrowheads). (D)
SVZ = subventricular zone; Temp. c. = temporal cortex. Cal- There are a few labeled cells in the ventricular zone (VZ) at
ibration bar = 1 mm. the hippocampus. Calibration bar = 20mm.

layer and the hilus of the dentate gyrus. Those are the 4.6A; see color plate 5). The young cells have no cyto-
areas where the frequency of labeled cells is the highest plasm and their nuclei are darkly stained (figure 4.6A).
in the hippocampal formation. In order to avoid mis- Pycnosis has been found rarely in the dentate granular
calculation caused by postmortem autolysis or hypoxic layer and the pycnotic cells are not exclusively at the
damage, we have examined the frequency of pycnotic hilar border, where newly generated cells are supposed
nuclei in the hippocampi of newborn perfused mon- to be located, but are distributed in the entire width of
keys. Monkeys have been used to study the postnatal the granule cell layer (figure 4.6C; see color plate 5).
development of hippocampal connections and the In a few cases, apoptosis may be assumed based on the
maturation of the ultrastructure (Seress and Ribak, nuclear changes of the neurons that were probably
1995a,b). It is known that granule cell formation occurs granule cells (figure 4.6D; see color plate 5). Such apop-
in the monkey hippocampus until the 32nd postnatal totic cells were evident both at the molecular layer (fig-
day (Rakic and Nowakowski, 1981). In the perinatal pe- ure 4.6.D) and hilar borders (figure 4.6B; see color
riod a large number of young granule cells locate them- plate 5) of the granule cell layer, suggesting that cell
selves at the hilar border of the granule cell layer. death does not necessarily occur among the newly
Therefore, we assume that in newborn and 4-day-old generated cells. The frequency rate of cell death was
monkeys cell death may accompany cell formation. In extremely low in the monkey dentate gyrus, approxi-
newborn monkeys, cresyl violet-stained semithin sec- mately 2-3 cells per 1000 granule cells.
tions reveal the developed granule cells owing to their In conclusion, the overwhelming majority of neurons
purple-stained cytoplasm which sharply contrasts with of the human hippocampal formation are formed in
the thin, pale cytoplasmic rim of young cells (figure the first half of pregnancy, before the 24th week. There

48 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 FIGURE 4.4 Photomicrographs of Ki-67-labeled cells in cre-
syl violet-stained coronal sections of the dentate gyrus (A, B,
and D) and Ammon'shorn (C) of a newborn child. (A) There
are a few labeled cells (arrows) in the hilus (H) and in the
FIGURE 4.3 Photomicrographs of Ki-67-labeled cells in cre- granule cells layer (GL). (B) Labeled glial cell (arrowhead)
syl violet-stained coronal sections of the entorhinal cortex at the border of the granule cell layer (GL) and another that
(A), subiculum (B), and the subventricular zone under the could be a nonaligned granule cell (arrow) in the molecular
temporal cortex (C). (A and B) Some of the labeled cells ap- layer (ML). (C) Labeled glial cells (arrows) in the stratum
pear to be neurons (arrow on A), whereas the others are glial oriens (str. or.) of the CA1 area of Ammon's horn. (D) Some
cells. (C) Large numbers of labeled cells are in the ventricular of the labeled cells in the hilar region (H) of the dentate gyrus
zone (VZ) as well as in the subventricular zone (SVZ) at the may be neuronal precursors (arrow), whereas others are glial
temporal neocortex. Calibration bar = 20 (am. cells (open arrows). Calibration bar = 20 mm.

is limited cell formation within the dentate gyrus in the torhinal cortex, bypass previously generated neurons
perinatal period, but the number of newly formed cells on their way to the superficial limits of the developing
is very low relative to the total number of cells. Similarly, cortical plate (Nowakowski and Rakic, 1981). There-
there is no significant cell death in the perinatal period, fore, the inside-out migration pattern is similar to that
neither in the human nor in the monkey hippocampal in the neocortex. The exception is again the dentate
formation. This suggests that formation of neuronal gyrus, where the granule cell layer is formed in an
connections is not accompanied by cell death in the outside-in pattern (Bayer, 1980; Nowakowski and Rakic,
primate hippocampus and that most of the newly gen- 1981). The dentate gyrus receives neurons both from
erated neurons survive. the ventricular zone and from the hilar proliferative
zone, which may be a hippocampal equivalent of the
Cell migration in the hippocampal formation subventricular zone, because the dynamics of cell for-
mation appear to be similar in the subventricular zone
Experimental studies in rodents and primates have and in the hilus (Bayer, 1980; Seress, 1977). Cell migra-
shown that the ventricular zone underlying the hippo- tion to Ammon's horn, the subiculum, and the ento-
campal formation is the source of neurons (Bayer, 1980; rhinal cortex terminates early in pregnancy, whereas
Nowakowski and Rakic, 1981). Migrating neurons in migration of granule cells continues for a long period
Ammon's horn, in the subiculum as well as in the en- of time postnatally. In rodents, granule cell migration

SERESS: HIPPOCAMPAL FORMATION 49

 FIGURE 4.6 Photomicrographs of cresyl violet-stained semi-
thin sections of the dentate gyrus of a newborn monkey.
(A) Inside the granule cell layer (GL), matured granule cells
(g) display large pale cell nuclei with a purple cytoplasmic rim,
FIGURE 4.5 Photomicrographs of Ki-67-labeled cells in dif- whereas at the hilar (H) border many small dark immature
ferent parts of the hippocampal formation of a 3-month-old cells (i) accumulate. Some of the large hilar neurons display
(A, B, C, F, G) and an 11.5-month-old child (D and E). (A) deep nuclear infoldings (arrows) characteristic of matured
Labeled cells appear to be glial cells in the molecular layer GABAergic neurons of the hippocampus. (B) An apoptotic
(ML) and granule cell layer (GL). (B) Labeled cells (arrows) cell (arrow) at the hilar border of the granule cell layer.
in the granule cell layer (GL) and hilus (H). (C) A labeled (C) A pycnotic cell nucleus (arrow) at the molecular layer
astroglia in satellite position to a pyramidal cell (p) in the CA1 border of the granule cell layer. (D) An apoptotic granule cell
area of Ammon's horn. (D) Many small, darkly stained labeled (arrow) at the molecular layer border (ML) of the granule
cells (arrows) in the different layers of dentate gyrus of the cell layer. Note the shrunken soma and dense cell nucleus as
11.5-month-old child. (E) Higher magnification reveals that well as the basophilic dendrite of the apoptotic cell. Calibra-
the Ki-67-labeled cells (arrows) are all glial elements. (F) An tion bar =10 mm.
immunolabeled cell in the hilus of a 3-month-old child. Nu-
clear morphology of labeled cell suggests that, when com-
pared with an unlabeled neuron (arrow), it may be a neuron.
entiated, immature-looking cells are positive for glial
(G) Labeled cells in the entorhinal cortex of the 3-month-old fibrillary acidic protein in the 6-month- and 1.5-year-old
child. Calibration bars = 40 mm for D, 20 mm for A, B, C, E, monkeys, suggesting that these immature cells are neu-
F, G. rons and may differentiate into granule cells later in
development (Eckenhoff and Rakic, 1988). Indeed, re-
cent observations suggest that neurogenesis and cell mi-
lasts only a few days after final cell division occurs, a gration occur in the dentate gyrus of adult monkeys
finding that may be explained by the smaller size of the (Gould et al., 1999; Kornack and Rakic, 1999). Similar
rodent dentate gyrus. In primates, however, groups of results have previously been described in rats (Bayer,
small cells with thin cytoplasm and dark cell nucleus 1982). In cresyl violet-stained preparations the imma-
persist in the subgranular zone of the dentate gyrus ture neurons display dark, ovoid cell nuclei and very
throughout the first postnatal year (Eckenhoff and thin cytoplasmic rims that lack organelles (Seress,
Rakic, 1988). Since no neurons were labeled with thy- 1992). The morphology of these immature neurons is
midine after the first month, these cells had to be very similar to that of those migrating cells described
formed in the perinatal period. Not all of the undiffer- for the developing cerebellum (Rakic, 1971).

50 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 The period of time during which immature, migrat-
ing cells are detectable in Ammon's horn and the den-
tate gyrus has been examined. In 24-30-week-old
infants, a large number of migrating cells are visible in
the subgranular proliferative zone of the hilus as well
as in the subplate and intermediate zone between the
alveus and pyramidal cell layer (figure 4.2A,C). In the
32-week-old and older infants, Ammon's horn lacks
these immature cells (Arnold and Trojanowski, 1996a).
However, in the dentate gyrus large numbers of imma-
ture cells still persist in the subgranular zone, not only
in 32-36-week-olds but also in neonates (figure 4.7A,D;
see color plate 6). Cells with small, darkly stained,
mostly elongated-ovoid nuclei are dispersed in the hilar
region including the subgranular zone (figure 4.7D).
In a 5-month-old child, the number of immature cells
is much lower (figure 4.7B; see color plate 6); however,
clusters of dark, immature cells still occur in the
subgranular zone of the 8.5-month-old child (figure
4.7C; see color plate 6). In 1-year-old children the hilar
border of the granule cell layer and the deep hilus are
free of immature cells (figure 4.7E; see color plate 6)
and the general cytoarchitectonic features of the den-
tate gyrus appear adult-like. It is evident that the cell
density of the hilar region is higher in newborn infants FIGURE 4.7 Photomicrographs of cresyl violet-stained coro-
than in older children and that the number of granule nal sections of the human dentate gyrus. (A) In the newborn
cells inside the granule cell layer increases after birth infant, the subgranular zone and the deep hilus contain large
(Seress, 1992). It is also known that glial cells are rare numbers of darkly stained small cell nuclei. (D) Higher mag-
inside the granule cell layer. Therefore, it can be as- nification reveals the elongated cell nuclei (arrows) of im-
mature neurons that are probably migrating granule cells. In
sumed that the immature cells of the hilar region mi- 5-month-old (B) and 8.5-month-old (C) children, migrating
grate to their final position—the granule cell layer— cells are still seen in the hilus (H) and at the hilar border of
and develop into granule cells. It is likely that another the granule cell layer (GL). (E) In the dentate gyrus of a
population of cells remains in the hilus, becoming glial 1-year-old child, the subgranular zone contains only glial cells
cells, because the large neurons of the hilar region are (compare D with E). Calibration bar = 50 mm.
known to form at the same prenatal period when py-
ramidal cells of Ammon's horn are formed. rons. An unknown number of immature cells may per-
Our current results suggest that cell migrations in the sist until adulthood, and those cells may consist of the
human hippocampal formation terminate prenatally in stem cells for adult neurogenesis (Eriksson et al., 1998).
Ammon's horn, the subicular complex, and the ento- However, immature dividing cells in the adults must be
rhinal cortex. In the dentate gyrus, migration of gran- rare, since they are already infrequent in the early post-
ule cells lasts for approximately 1 year after birth. This natal period. The cell number in the granule cell layer
is a long period of time considering that active cell does not change significantly after the 30th postnatal
formation in the ventricular and intermediate zones day in rats, and no data demonstrate an increase in
terminates in the first half of pregnancy. In the second number of granule cells in the adult primate hippocam-
half of pregnancy, a few cells may still divide in the pus (Seress, 1977, 1987). In monkeys, cell migration
migratory zone described by Humphrey (1966) or in may last for 3 months; the last identified granule cells
the subplate and intermediate zone (Arnold and are formed 32 days after birth (Rakic and Nowakowski,
Trojanowski, 1996a; Kostovic et al., 1989), but most of 1981) and the migrating neurons disappear from the
the cells in those zones are migrating, not dividing, neu- subgranular zone by the 4th postnatal month. There-
rons. A large number of nondifferentiated granule cells fore, the prolonged migration of granule cells is a
appear to reside in the subgranular zone for a long pe- unique characteristic of the human dentate gyrus, al-
riod of time (1 year or more) before they move to their though one must consider that the lifespan of the mon-
final positions and begin to proliferate into mature neu- key is three times shorter than that of the human.

SERESS: HIPPOCAMPAL FORMATION 51

Therefore 3-4 months in monkeys may be equivalent opment of CA3 pyramidal cells; CA1 pyramidal cells
to 1 year in the human. Similarly, as in rodents and show extensive dendritic growth and an increasing syn-
subhuman primates, it is clear that cell migration in the aptic density up to the 90th postnatal day. This corre-
dentate gyrus lasts as long as the migration of granule lates with an increasing number of synapses in the
cells of the cerebellum, since the external granular stratum radiatum (Pokorny and Trojan, 1986; Pokorny
layer of the cerebellum disappears between 8.5 and 10 and Yamamoto, 1981).
months in human (Rakic, 1971; Abraham et al., 1999). The circuitries of the hippocampal formation have
There is, however, one important difference between been known since the time of Ramon y Cajal (1911).
the two regions: The cerebellar external granular layer Andersen and colleagues (1971) used the term "hip-
displays significant cell multiplication as long as the pocampal trisynaptic circuit" to refer to the unique uni-
layer persists (Abraham et al., 1999), whereas in the hi- directional progression of excitatory pathways that link
lar region, a similar rate of cell formation is a prenatal the subregions of the hippocampal formation (Ander-
event. sen, Bliss, and Skrede, 1971). Within this circuit the
granule cells are the first elements. They connect the
Synapse formation and the development dentate gyrus to the CA3 pyramidal neurons of Am-
of neuronal connections in the mon's horn, which in turn project to the CA1 pyramidal
hippoca mpal formation cells. CA1 pyramidal neurons are the sole cortical ef-
ferents of the hippocampus (for further discussion, see
Postnatal development of the hippocampal formation Johnston and Amaral, 1998).
and its connections has been described in detail in rats. Granule cells are the last to be formed among the
There are but few available data in primates, and data neurons of the trisynaptic circuit. As a consequence, the
pertaining to human development are particularly synaptic connections between the dentate gyrus and
sparse. The lack of empirical study can be partially ex- Ammon's horn are established late in ontogenesis. This
plained by the difficulties of tissue preservation due to highlights the functional importance of this first link in
the regulated postmortem delay between the time of the trisynaptic circuit, because its morphological mat-
death and autopsy. After a few hours' delay, the cellular uration correlates with hippocampus-dependent behav-
ultrastructure is obscured in human tissue. In rats, syn- ioral maturation (Nadel and Willner, 1989).
aptic connections and dendritic development of a few
cell types have been described, such as those of the Development of principal (pyramidal and granule
granule cells (Seress and Pokorny, 1981), mossy cells cells) and nonprincipal (GABAergic) neurons of
(Ribak, Seress, and Amaral, 1985), and the pyramidal the primate hippocampus
cells of the CA1 area of Ammon's horn (Pokorny and
Yamamoto, 1981). Light microscopic changes of the Granule cells, hilar mossy cells, and CA3 pyramidal cells
dendrites, such as spine density and spine morphology, of monkeys are in an advanced stage of development at
correlate with the increasing number of afferent ter- birth (Seress and Ribak, 1995a,b). Most granule cells
minals that establish synapses with dendrites of hippo- have a complete dendritic arbor, although both spine
campal neurons. Studies in rats suggest that in the density and the number of synapses in the molecular
dentate gyrus an increasing number of mossy fiber axon layer increase after birth (Seress, Baumgartner, and Ri-
terminals establish connections with the hilar mossy bak, 1995). Mossy cells and CA3 pyramidal cells display
cells and the pyramidal cells of the CA3 area (Amaral thorny excrescences, and terminals of mossy fibers (ax-
and Dent, 1981; Ribak, Seress, and Amaral, 1985). First ons of granule cells) establish multiple synapses with
connections are established with dendritic shafts at those excrescences (Seress and Ribak, 1995a,b). These
birth (Amaral and Dent, 1981). The more complex ter- features indicate that in monkeys developmental events
minals of mossy fibers induce the development of are mainly prenatal, but the chronological sequence of
thorny excrescences on the dendrites of 2-week-old rats synaptic development is similar to what occurs in the
(Amaral and Dent, 1981). Maturation of this connec- rat. At present we have no data about the development
tion appears to terminate 2-3 weeks after the last gran- of CA1 pyramidal cells in monkeys. Our preliminary
ule cells have been formed at the end of the third week. data indicate a change in spine density and myelin for-
The CA3 pyramidal cells send the Schaffer collaterals mation up to the postnatal seventh month.
to the CA1 pyramidal cells. It is not known when the In conclusion, the monkey hippocampal formation
Schaffer collaterals reach the CA1 cells, nor when this displays a fast and dynamic development that termi-
connection is fully mature. However, the CA1 pyramidal nates several months after birth. In harmony with the
cells show a delayed maturation relative to the devel- advanced stage of development of principal neurons,

52 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

most GABAergic cells of the hippocampal formation
mature very early in monkeys (Berger, Alvarez, and
Goldman-Rakic, 1993; Berger, deGrissac, and Alvarez,
1999). This is in contrast to the development of non-
principal cells in rats, where GABAergic neurons show
a very immature appearance at birth (Seress and Ribak,
1990). Berger, Alvarez, and Goldman-Rakic (1993) as-
sume that functional circuits are forming between the
hippocampus and the entorhinal cortex of primates
during the first half of gestation. Other data also suggest
an early maturation of the human entorhinal cortex
(Kostovic, Petanjek, and Judas, 1993; Ulfig, 1993). Light
and electron microscopic data are sparse with respect
to the developing human hippocampal formation.
Nonetheless, the cell types and cytoarchitectonics of the
adult hippocampus have been described in several clas-
sic reports (Koelliker, 1896; Lorente de No, 1934).
Purpura (1975) has included the hippocampus in his
description of the human cerebral cortex, but concen-
trated mainly on the early developmental stages. He ob-
served that a variety of growth processes observed in
hippocampal pyramidal neurons in the 20-29-week-old
fetus are not prominently displayed in granule cells. In
33-week-old premature infants, many granule cells dis-
play dendrites that appear to be mature, except for the FIGURE 4.8 Photomicrographs of Golgi-impregnated neu-
total number of spines, whereas other granule cells dis- rons in the human dentate gyrus. (A) Dendrites of the mossy
play immature dendrites. The time period from 20 to cell in the newborn infant display the conventional simple
28 weeks of gestation is a phase of maximum dendritic spines and a few small protrusions that can be thorny excres-
cences (arrows). (B) Dendrites of mossy cell of the 3-year-old
growth for hippocampal pyramidal neurons (Purpura, child display large complex spines, the so-called thorny ex-
1975). Pyramidal neurons of the CA3 area display only crescences, that are characteristic for mossy cells of the adult
a few, small spines (spicules) and filopodia on their den- brain (D). (C) Large numbers of granule cells display spiny
drites in the 22-week-old fetus. In the 33-week-old fetus, dendrites (open arrows) and a richly arborizing axon (arrows
the first thorn-like excrescences appear on the den- point to the axon and its collaterals) in the dentate gyrus of
the newborn child. Calibration bars = 40 mm for B and 20
drites of pyramidal-type neurons of the hilus and on mm for A, C, and D.
CA3 pyramidal cells, suggesting that the first connec-
tions between granule cells and their postsynaptic tar-
gets in the hilus and CA3 area are established by the growth cones (Seress, 1992). These latter granule cells
33rd gestational week. are still growing and may be those that were formed in
In children born at term, the hilus of the dentate the perinatal period. Therefore, the diversity in granule
gyrus contains large multipolar cells displaying a few cell development seen by Purpura (1975) in the 33-
complex thorn-like excrescences and a few small, con- week-old fetus is still observable at birth. It should be
ventional spines (figure 4.8A; see color plate 7). These noted that a few immature-looking granule cells are still
cells are probably young mossy cells that have already seen in the 15-month-old child (Seress, 1992), suggest-
established a few synapses with mossy fiber terminals of ing that granule cells exhibit a prolonged period of cell
granule cells (figure 4.9; see color plate 8). In agree- proliferation. The first real thorn-like excrescences ap-
ment with the previous observations of Purpura (1975), pear on mossy cells by the third postnatal month (Seress
we have demonstrated that well-developed granule cells and Mrzljak, 1992). At seven months thorny excres-
frequently appear in the granule cell layer (figure 4.8C; cences are frequent on mossy cells, but their number
see color plate 7). These granule cells have dendrites and size continue to increase up to the third year, when
that display a large number of spines and an axon that the first adult-like mossy cells appear (figure 4.8B; see
gives rise to several collaterals in the hilus (figure 4.8C). color plate 7). The complex spines or thorny excres-
In addition, several granule cells display varicose, cences of the adult human mossy cells (figure 4.8D; see
stubby, short, and spineless dendrites that terminate in color plate 7) are much larger than those in monkeys

SERESS: HIPPOCAMPAL FORMATION 53

 mossy cells (Seress and Ribak, 1995a,b). However, not
only the mossy cells but also the pyramidal cells of the
CA1 region of Ammon's horn are in an early stage of
proliferation at birth. If the CA1 pyramidal cells of
adults (figure 4.9D) are visually compared with the py-
ramidal cells of newborn infants (figure 4.9C), several
differences are obvious. The pyramidal cells of newborn
infants have few basal dendritic branches and poorly
developed side branches of the apical dendrites. A simi-
lar pattern has been found in the neocortex, where the
pyramidal cells of the newborn child display a few, var-
icose, short basal dendrites and a few poorly developed
side-branches of the apical dendrite (figure 4.9A).
There are only a few spines on the pyramidal dendrites
of the newborn infant (figure 4.9A), whereas the equiv-
alent pyramidal cell dendrites in the adult cortex are
fully packed with spines (figure 4.9B). The postnatal
human development of pyramidal cells in Ammon's
horn and the neocortex has not been described in de-
tail, but our observations suggest that the morphologi-
cal development of the principal neurons of those brain
areas lasts as long as that in the hilar region.

Synapse formation in the human hippocampus

Light microscopic studies of the development of den-

drites, and the early appearance of first spines, suggest
FIGURE 4.9 Photomicrographs of Golgi-impregnated pyram- that synapse formation starts early in the human hip-
idal cells in the cortex (A and B) and in Ammon's horn (C pocampal formation. It is extremely difficult to achieve
and D) of hippocampal formation in newborn (A and C) and
adult (B and D) brain. Pyramidal cells in the newborn child
adequate preservation of postmortem fetal or child
display immature, beaded, spine-free short dendrites, whereas brains for electron microscopy. Data relating to the syn-
in the adult, pyramidal cells have large dendrites fully covered aptic development of the human hippocampal forma-
with spines. Calibration bar = 20mm. tion are rare. In accord with expectations based on
studies of monkey brains (Berger, Alvarez, and Goldman-
Rakic, 1993), the first synapses have been observed in
or in rodents (Frotscher et al., 1991). In the 3-year-old the marginal zone and in the cortical subplate of Am-
child mossy cells still vary with respect to their dendritic mon's horn in a 15-week-old fetus (Kostovic et al.,
spine density and the size of their thorny excrescences. 1989). The axodendritic asymmetric synapses in the
Not until the fifth year do all impregnated mossy cells marginal zone suggest that entorhinal axons have al-
display similarly large thorny excrescences that are in- ready reached the hippocampus at that age, since those
distinguishable from those seen in adults (Seress and are the sole excitatory afferents in that zone. Recent
Mrzljak, 1992). The extended morphological develop- findings using anterograde and retrograde tracers in-
ment of granule, mossy, and pyramidal cells accords dicate that reciprocal entorhinal-hippocampal pro-
with the prolonged development of cytoskeletal pro- jections may be among the first corticocortical
teins, since an adult pattern of neuronal cytoskeletal connections to be established in the human brain (Hev-
protein expression in the hippocampus appears around ner and Kinney, 1996). The perforant pathway projec-
the second postnatal year (Arnold and Trojanowski, tion from the entorhinal cortex to the dentate gyrus
1996b). There is a similarly extended development of develops several weeks later than the connection to Am-
thorny excrescences for CA3 pyramidal neurons, al- mon's horn (Hevner and Kinney, 1996). In our electron
though it appears that CA3 pyramidal cells display more microscopic preparations of postmortem tissue from
advanced thorns at birth than the mossy cells. Similarly, neonates (36-40 weeks of gestation), small asymmetric
CA3 pyramidal cells of newborn monkeys appear to axodendritic and axospinous synapses are frequently
show more developed thorn-like excrescences than found in different layers of the dentate gyrus and Am-

54 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 in monkeys (Berger, Alvarez, and Goldman-Rakic,
1993), nonprincipal GABAergic neurons appear to de-
velop very early in the human hippocampal formation.
NADPH-d positive neurons appear at the 15th fetal
week and become frequent from the 28th to the 32nd
weeks (Yan and Ribak, 1997). In accord with light mi-
croscopic data, symmetric synapses were frequent in the
different layers of the dentate gyrus (figure 4.10A) and
Ammon's horn in newborn children.
In summary, the first afferent fibers to the hippocam-
pus proper arise from the entorhinal cortex and estab-
lish synapses with neurons of Ammon's horn early in
fetal development (Kostovic et al., 1989). This corre-
sponds to the suggestion of Berger, Alvarez, and Gold-
man-Rakic (1993) that the remarkably early maturation
of the entorhinal cortex during the first half of gesta-
tion, together with the early neurochemical develop-
ment of the hippocampal formation, indicates that
functional circuits in primates form during the first half
of gestation. Tracing studies also suggest that entor-
hinal-hippocampal connections are established by mid-
gestation (Hevner and Kinney, 1996). In addition, the
presence of the following indirect features suggests that
neuronal connections of the inner trisynaptic circuit of
the hippocampus, which connect the dentate gyrus with
FIGURE 4.10 Electron micrographs of the hilar region of the the entorhinal cortex, are also established prenatally:
newborn infant (A and B) and adult (C). (A) An axon ter- (1) mature-looking granule cells present in the dentate
minal (open star) that forms a symmetric synapse (arrows)
with a dendrite in the hilus of the dentate gyrus of a newborn
gyrus at birth give rise to richly arborizing axons;
child. (B) A small (star) and a large (mt) terminal in the hilus (2) small thorny excrescences on dendrites of mossy
of the newborn child. Both terminals contain a few vesicles and CA3 pyramidal cells in newborn children; (3) asym-
and form asymmetric synapses (arrows and curved arrow) metric synapses in the dentate gyrus and in the stratum
with dendrites. In the mossy terminal (mt) the mitochondria lucidum of Ammon's horn of newborn children; (4) a
locate away from the synaptic surface and the terminal con-
tains a few dense core vesicles (open arrows). (C) A large
large number of asymmetric synapses in the strata ra-
mossy fiber terminal (mt) in the hilus of the dentate gyrus of diatum and oriens of the CA1 area of Ammon's horn.
an adult human is fully packed with synaptic vesicles and a few Therefore, it is reasonable to assume that children born
of them contain a dense core (open arrows). Mitochondria at term have the synaptic connections necessary to es-
locate away from the asymmetric synaptic surface (arrow). Cal- tablish memory traces (for discussion, see Nelson,
ibration bar = 0.5 mm.
1995). However, the functional capability of those cir-
cuits has not been fully examined. One must be ex-
tremely cautious when correlating morphology with
mon's horn (figure 4.10B). However, the large complex function, either at the electrophysiological or behav-
mossy fibers terminals seen in the adult dentate gyrus ioral level. For instance, Purpura (1975) demonstrated
(figure 4.1OC) cannot be seen. It is highly probable that that immature photic-evoked potentials can be re-
those terminals are disrupted and the structure de- corded in 33-week-old premature infants whose pyram-
stroyed by the postmortem autolysis, but it is also pos- idal cells in the visual cortex have poorly developed
sible that mature mossy fiber terminals are rare in the basal dendrites and almost no spines. Despite the fact
newborn infant. There are a few large and pale termi- that the adult-like neuronal circuitry needed for infor-
nals that display a few vesicles close to the synaptic sur- mation processing was absent, some synaptic connec-
face and mitochondria that locate away from the tions were nevertheless capable of transmitting visual
synaptic surface (figure 4.1 OB). These are characteristic impulses. It is plausible that similar events occur in the
of mossy fiber terminals and, as in newborn monkeys, hippocampal formation, and thus the earliest forms of
the hilus contains both mature and immature mossy memory formation may not necessarily be related to the
fiber terminals (Seress and Ribak, 1995b). Similarly, as adult-like function of the hippocampal formation.

SERESS: HIPPOCAMPAL FORMATION 55

Conclusion and functional implications week to the third postnatal month, when cell formation
decreases to a minimum but cell migration continues
Three major milestones underlie the formation of neu- and the first thorny spines appear on the dendrites of
ronal circuits: (1) when the participating neurons are hilar mossy cells; (2) from the third month to the first
formed, (2) when they reach their final position, and year, when cell migration ceases and the first adult-like
(3) when they start to proliferate and establish their first large thorny excrescences appear on mossy and CA3
afferent and efferent connections. The neuronal cell pyramidal cells, suggesting the beginning of a period of
formation of the human hippocampus appears to occur excessive synapse formation; (3) from the first year to
within a relatively short period of time, with the excep- the third year, when the first mature-looking mossy cells
tion of the dentate granule cells. The last neurons for appear; (4) from the third year to the fifth year of life,
Ammon's horn, the subicular complex, and entorhinal when most neurons are adult-like both in the hilus and
cortex may form between the 20th and the 24th gesta- in Ammon's horn. Moreover, it is highly possible that
tional weeks, but major cell formation probably ceases synapse formation continues in the hippocampal for-
around the 15th gestational week. This finding indi- mation after the fifth year and results in volumetric and
cates that while pyramidal cells form in the first half of sexual differences (Caviness et al., 1996). However,
pregnancy, granule cell formation continues until birth, given currently available morphological methods, these
even if the major cell formation of granule cells hap- later events cannot be studied. In addition, the number
pens before the 34-36th week of gestation. Cell migra- of newly formed granule cells is small after birth; there-
tion is also rapid in Ammon's horn, the subiculum, and fore, it is difficult to evaluate the functional importance
entorhinal cortex; adult-like cytoarchitectonics of those of granule cell formation in the adult human (Ericsson
areas are observed at the 24th week (Arnold and Tro- etal., 1998).
janowski, 1996a), whereas the granule cell layer is still If Ammon's horn develops early and receives its ma-
growing at birth. At birth, the granule cell layer includes jor cortical afferent connection early in gestation, what
only about 70% of the adult numbers of granule cells might be the functional consequence of the late for-
and its volume is also about 25-30% less than that in mation of granule cells of the dentate gyrus (and with
adults (Seress, 1988; unpublished observations). Cell it the trisynaptic circuit) ? Can the dentate gyrus modify
migration lasts for eight months until the last immature- the function of the hippocampal formation? Animal ex-
looking cells disappear from the hilus. Such a long pe- periments demonstrate that it can. In rats there is an
riod of cell migration explains the presence of important postnatal morphological development of the
proliferating granule cells in the dentate gyrus of the dentate gyrus during normal development and a cor-
1.5-year-old child. This observation also suggests that at responding delay in the adult-like hippocampal func-
least 30% of granule cells start to proliferate and estab- tion of spatial navigation (Nadel and Willner, 1989).
lish their afferent and efferent connections postnatally, Moreover, if the granule cells of the dentate gyrus are
although only a very small percentage of those cells are removed by irradiation at birth, the normal function of
formed after the 34th week. hippocampal formation is never attained in the course
It is very likely that entorhinal afferents are among of later life (Czurko et al., 1997). A significant func-
the first to establish connections with pyramidal cells. tional reserve capacity of the dentate gyrus is suggested
There is no ultrastructural evidence concerning the by the fact that if 50% of granule cells remain intact in
time when entorhinal axons form synapses with granule the dentate gyrus, the effects of a functional lesion are
cells. Indeed, Golgi-impregnated granule cells in the 15- not detectable (Czeh et al., 2001). This finding might
week-old fetus display immature, stubby dendrites with- explain why not all children who suffer perinatal dam-
out spines. Spine formation on dendrites appears to age show signs of functional lesions. Previously, we have
begin after the 25th week (Seress and Mrzljak, 1987). hypothesized that the development of the hippocampal
Afferent circuits develop first in the Ammon's horn- trisynaptic circuit is under the influence of complex en-
entorhinal complex while the dentate gyrus receives vironmental stimuli (Seress and Mrzljak, 1992). After
afferents later (Hevner and Kinney, 1996). As a conse- birth, when sensory neocortical information processing
quence, the intrinsic connections (trisynaptic circuit) increases, it also likely increases the activity of the per-
of the hippocampal formation start to develop later. forant path from the entorhinal cortex to the hippo-
Synaptic elements of the trisynaptic connection are campus. As a consequence, higher activity within the
present at birth, but full maturation is not expected to perforant path would activate the granule cells, causing
occur before the 5th postnatal year in the human. Four them to form additional synapses with those postsyn-
periods can be distinguished by the morphological aptic neurons that are their already selected targets.
events during these five years: (1) from the 36th fetal The hippocampal formation is capable of memory for-

56 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

mation in the perinatal period when the intrahippo- BERGER, B., C. ALVAREZ, and P. S. GOLDMAN-RAKIC, 1993. Neu-
campal trisynaptic circuit is still forming. This finding rochemical development of the hippocampal region in the
fetal rhesus monkey. I. Early appearance of peptides, cal-
may explain why even very young infants show signs of cium-binding proteins, DARPP-32 and the monoamine in-
memory formation (Nelson, 1995). The later-forming nervation in the entorhinal cortex during the first half of
neuronal connections between granule cells of the den- gestation (E47 to E90). Hippocampus 3:279-305.
tate gyms and pyramidal neurons of Ammon's horn BERGER, B., N. DEGRISSAC, and C. ALVAREZ, 1999. Precocious
may alter the functional circuits of the hippocampus development of parvalbumin-like immunoreactive inter-
proper. It is therefore proposed that the trisynaptic cir- neurons in the hippocampal formation and entorhinal cor-
tex of the fetal Cynamogus monkey. J. Comp. Neurol.
cuit of the hippocampal formation is critical for the 403:309-331.
adult-like memory formation. If the morphological de- BROWN, D. C., and K. C. GATTER, 1990. Monoclonal antibody
velopment of the hippocampal formation correlates Ki-67: Its use in histopathology. Histopathology 17:489-503.
with functional capability, then adult-like memory for- CAJAL, S. RAMON Y, 1911. Histologie du Systeme Nerveux de
mation in humans may not be expected earlier than the l'Homme et des Vertebres, Vol. II. Paris: Maloine.
CAVINESS, V. S. JR., D. N. KENNEDY, C. RICHELME, J. RADE-
fifth postnatal year. It is not known how changes of
MACHER, and P. A., FILIPEK, 1996. The human brain age 7-
inner and outer environments affect human hippocam- 11 years: A volumetric analysis based on magnetic reso-
pal development, a question that brings into focus the nance images. Cereb. Cortex 6:726-736.
need to ensure the welfare of children through early CZEH, B, A. STUCHLIK, M. WESIERSKA, J. M. CIMADEVILLA, J.
childhood. POKORNY, L. SERESS, and J. BURES, 2001. Effect of dentate
gyrus lesion on allothetic and idiothetic spatial behavior.
ACKNOWLEDGMENTS The author is grateful to Maria Pal- Neurobiology of Learning and Memory (in press).
leszter, Maria Domjan, and Judit Mishley-Lorand for their ex- CZURKO, A., B. CZEH, L. SERESS, L. NADEL, and J. BURES, 1997.
cellent technical assistance, and thanks Dr. Hajnalka Abraham Severe spatial navigation deficit in the Morris water maze
for her participation in the experiments and critical com- after single high dose of neonatal X-ray irradiation in the
ments to this manuscript. This work was supported by the rat. Proc. Natl. Acad. Sti. (USA) 94:2766-2771.
Hungarian Science Foundation, grant No. T 029214 and by ECKENHOFF, M. F, and P. RAKIC, 1988. Nature and fate of pro-
ETT grant No. 80-4/1998. liferative cells in the hippocampal dentate gyrus during the
life span of the rhesus monkey.J. Neurosci. 8:2729-2747.
ERIKSSON, P. S., E. PERFILIEVA, T. BJORK-ERIKSSON, A. M. AL-
REFERENCES BORN, C. NORDBORG, D. E. PETERSON, and F. H. GAGE, 1998.
Neurogenesis in the adult human hippocampus. Nature
ABRAHAM, H., T. TORNOCZKY, GY. KOSZTOLANYI, and L. SER-
Med. 4:1313-1317.
ESS, 1999. Peri- and postnatal cell formation in the human
FROTSCHER, M., L. SERESS, W. K SCHWERDTFEGLER, and E.
brain. Soc. Neurosci. Abst. 25:2267.
BUHL, 1991. The mossy cells of the fascia dentata: A com-
AMARAL, D. G., 1990. Hippocampal formation. In The Human
parative study of their fine structure and synaptic connec-
Nervous System, G. Paxinos, ed. New York: Academic Press,
tions in rodents and primates.J. Comp. Neurol. 312:145-163.
pp. 711-755.
GOULD, E., A. J. REEVES, M. FALLAH, P. TAPANAT, C. G. GROSS,
AMARAL, D. G., and J. A. DENT, 1981. Development of mossy
fibers of the dentate gyrus: I. A light and electron micro- and E. FUCHS, 1999. Hippocampal neurogenesis in adult
scopic study of the mossy fibers and their expansions. J. Old World primates. Proc. Natl. Acad. Sci. (USA) 96:5263-
Comp. Neural. 195:51-86. 5267.
ANDERSEN, P., T. V. P. BLISS, and K. SKREDE, 1971. Lamellar HEVNER, R. F., and H. C. KINNEY, 1996. Reciprocal entorhinal-
organization of hippocampal excitatory pathways. Exp. hippocampal connections established by human fetal mid-
Brain Res. 13:222-238. gestation.J. Comp. Neurol. 372:384-394.
ANGEVINE, J. B., 1975. Development of the hippocampal for- HUMPHREY, T, 1966. The development of the human hippo-
mation. In The Hippocampus, R. L. Isaacson and K. H. Pri- campal formation correlated with some aspects of its phy-
bram, eds. New York: Plenum Press, pp. 61-90. logenetic history. In Evolution of Forebrain, R. Hassler and H.
ARNOLD, S. E., and J. Q. TROJANOWSKI, 1996a. Human fetal Stephan, eds. New York: Plenum Press, pp. 104-116.
hippocampal development: I. Cytoarchitecture, myeloar- HUMPHREY, T, 1967. The development of the human hippo-
chitecture, and neuronal morphologic features. J. Comp. campal fissure.J. Anat. 101:655—676.
Neurol. 367:274-292. JOHNSTON, D., and D. G. AMARAL, 1998. Hippocampus. In The
ARNOLD, S. E., and J. Q. TROJANOWSKI, 1996b. Human fetal Synaptic Organization of the Brain, G. M. Shepherd, ed. New
hippocampal development: II. The neuronal cytoskeleton. York: Oxford University Press, pp. 418-458.
J. Comp. Neurol. 367:293-307. KOELLIKER, A., 1896. Handbuch der Gewebekhre des Menschen.
BAYER, S. A., 1980. The development of the hippocampal re- Leipzig: Verlag von Wilhelm Engelmann.
gion in the rat. I. Neurogenesis examined with 3H-thymi- KORNACK, D. R., and P. RAKIC, 1999. Continuation of neuro-
dine autoradiography. J. Comp. Neurol. 190:87-114. genesis in the hippocampus of the adult macaque monkey.
BAYER, S. A., 1982. Changes in the total number of dentate Proc. Natl. Acad. Sci. (USA) 96:5768-5773.
granule cells in juvenile and adult rats: A correlated volu- KOSTOVIC, I., Z. PETANJEK, and M.JUDAS, 1993. The early areal
metric and 3H-thymidine autoradiographic study. Exp. Brain differentiation of the human cerebral cortex: Entorhinal
Res. 46:315-323. area. Hippocampus 3:447-458.

SERESS: HIPPOCAMPAL FORMATION 57

KOSTOVIC, I., L. SERESS, L. MRZLJAK, and M.JUDAS, 1989. Early SERESS, L., 1988. Interspecies comparison of the hippocampal
onset of synapse formation in the human hippocampus: A formation shows increased emphasis on the regio superior
correlation with Nissl-Golgi architectonics in 15- and 16.5- in the Ammon's horn of the human brain. J Hirnforsch.
week-old fetuses. Neuroscience 30:105-116. 29:335-340.
LORENTE DE No, R., 1934. Studies on the structure of the ce- SERESS, L., 1992. Morphological variability and developmental
rebral cortex. II. Continuation of the study of the Ammonic aspects of monkey and human granule cells: Differences
system.J. Psychol. Neurol. 214:113-177. between the rodent and primate dentate gyrus. Epilepsy Res.
NADEL, L., andj. WILLNER, 1989. Some implications of post- 7(suppl.):3-28.
natal maturation of the hippocampus. In The Hippocam- SERESS, L., B. J. BAUMGARTNER, and C. E. RIBAK, 1995. Post-
pus—New Vistas, V. Chan-Palay and C. Kohler eds. New York: natal development of granule cells and their afferent syn-
Alan R. Liss, pp. 17-31. apses in the dentate gyrus of Rhesus monkeys. Soc. Neurosci.
NELSON, C. A., 1995. The ontogeny of human memory: A cog- Abst. 21:929.
nitive neuroscience perspective. Dev. Psychol. 31:723-738. SERESS, L., and L. MRZLJAK, 1987. Basal dendrites of granule
NOWAKOWSKI, R. S., and P. RAKIC, 1981. The site of origin and cells are normal features of the fetal and adult dentate gyrus
route and rate of migration of neurons to the hippocampal of both monkey and human hippocampal formation. Brain
region of the rhesus monkey.J. Comp. Neurol. 196:129-154. Res. 405:169-174.
POKORNY, J., and S. TROJAN, 1986. The development of hip- SERESS, L., and L. MRZLJAK, 1992. Postnatal development of
pocampal structure and how it is influenced by hypoxia. mossy cells in the human dentate gyrus: A light microscopic
Ada. Univ. Carol. Med. Monograph. 113:5-79. Golgi study. Hippocampus 2:127-142.
POKORNY, J., and Y. YAMAMOTO, 1981. Postnatal ontogenesis of SERESS, L., and J. POKORNY, 1981. Structure of the granular
hippocampal CA1 are in rats. I. Development of dendritic layer of the rat dentate gyrus. A light microscopic and Golgi
arborization in pyramidal neurons. Brain Res. Bull. 7:113- study.J. Anat. 133:181-195.
120. SERESS, L., and C. E. RIBAK, 1990. Postnatal development of
the light and electron microscopic features of basket cells
PURPURA, D. S., 1975. Normal and aberrant neuronal devel-
in the hippocampal dentate gyrus of the rat. Anat. Embryol.
opment in the cerebral cortex of human fetus and young
181:547-565.
infant. In Brain Mechanisms in Mental Retardation, UCLA Fo-
SERESS, L., and C. E. RIBAK, 1995a. Postnatal development of
rum in Medical Sciences. New York: Academic Press, pp.141-
CA3 pyramidal neurons and their afferents in the Ammon's
169.
horn of Rhesus monkey. Hippocampus 5:217-231.
RAKIC, P., 1971. Neuron-glia relationship during granule cell
SERESS, L., and C. E. RIBAK, 1995b. Postnatal development and
migration in developing cerebellar cortex. A Golgi and elec-
synaptic connections of hilar mossy cells in the hippocampal
tron microscopic study in Macacus rhesus. J. Comp. Neurol. dentate gyrus of Rhesus monkeys. J Comp. Neurol. 355:93-
141:283-312. 110.
RAKIC, P., and R. S. NOWAKOWSKI, 1981. The time of origin of ULFIG, N., 1993. Ontogeny of the entorhinal cortex. Hippocam-
neurons in the hippocampal region of the rhesus monkey. pus 3:27-32.
J. Comp. Neurol. 196:99-128. VERHEIJEN, R., H. J. H. KUIJPERS, R. O. SCHLINGEMANN, A. L.
RIBAK, C. E., L. SERESS, and D. G. AMARAL, 1985. The devel- M. BOEHMER, R. VAN DRIEL, G. J. BRAKENHOFF, and F.C.S.
opment, ultrastructure and synaptic connections of the RAMAEKERS, 1989a. Ki-67 detects a nuclear matrix-associ-
mossy cells of the dentate gyrus.J. Neurocytol. 14:835-857. ated proliferation-related antigen. I. Intracellular localiza-
ROSE, D. S. C., P. H. MADDOX, and D. C. BROWN, 1994. Which tion during interphase.J Cell Sci. 92:123-130.
proliferation markers for routine immunohistology? A com- VERHEIJEN, R., H. J. H. KUIJPERS, R. VAN DRIEL, J. L. M. BECK,
parison of five antibodies.J. Clin. Pathol. 47:1010-1014. J. H. VAN DlERENDOCK, G. J. BRAKENHOFF, and F. C. S. RA-
SERESS, L., 1977. The postnatal development of rat dentate MAEKERS, 1989b. Ki-67 detects a nuclear matrix-associated
gyrus and the effect of early thyroid hormone treatment. proliferation-related antigen. II. Localization in mitotic cells
Anat. Embryol. 151:335-339. and association with chromosomes.J Cell Sci. 92:531-540.
SERESS, L., 1987. Strain differences in granule cell numbers YAN, X. X., and C. E. RIBAK, 1997. Prenatal development of
of the hippocampal dentate gyrus are not the result of adult nicotinamide adenine dinucleotide phosphate-disphorase
cell formation. In Ontogenesis of the Brain, S. Trojan and F. activity in the human hippocampal formation. Hippocampus
Stastny, eds. Universitas Carolina-Pragensis, pp. 29-32. 7:215-231.

58 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

5 Effects of Sex Hormones
on Brain Development
JUDY L. CAMERON

ABSTRACT Sex steroid hormones play an important role in cies, however, it is now clear that there are sex
regulating plastic changes in neuronal structure and function differences in areas of the brain that are not associated
throughout development and adulthood. There is a diversity with the control of reproductive function, including the
of cellular mechanisms by which steroid hormones influence
neural function, including metabolism to more active forms, hippocampus, striatum, cerebellum, amygdala, and ce-
modulation of function through traditional nuclear receptors, rebral cortex, and/or are associated with nonreproduc-
binding to membrane receptors, and activation of a variety of tive behaviors, such as spatial problem solving abilities,
second messenger pathways. A wide variety of neural processes verbal abilities, aggression, defensive behaviors, motor
are influenced by sex steroid hormones, including neuro- activity, and various aspects of learning and memory.
genesis, cell migration, growth of the neuronal cell body, den-
dritic growth, differentiation and synapse formation, synapse
elimination, neuronal atrophy and apoptosis, neuropeptide Sex steroid receptors within the brain
expression, the expression of neurotransmitter receptors, and
neuronal excitability. In many cases steroid hormones must ESTROGEN RECEPTORS Early studies of the effects of
be present both developmentally and in adulthood to elicit sex steroid hormones on the brain showed that hypo-
maximal effects on behavior.
thalamic regions, including the preoptic area and ven-
When they have been looked for, differences tromedial hypothalamus, played a central role in
throughout the brains of males and females have been mediating the action of estrogen on reproductive be-
clearly demonstrated in nearly all species, both with re- haviors and control of the hypothalamic-pituitary-
gard to the structural organization of many brain nuclei ovarian axis (Madeira and Lieberman, 1995; Pfaff,
and neural circuits, and in the functionality and re- 1980). Complementary studies identifying regions of
sponsiveness of neural systems to internal and external the brain with bound labeled estrogens showed that
stimuli. Moreover, many of the sex differences in brain these hypothalamic regions had the highest concentra-
structure and function result from differential exposure tions of estrogen receptors (ER) in the brain (Pfaff and
of males and females to the sex steroid hormones (tes- Reiner, 1973; Stumpf, Sar, and Keefer, 1975). Further
tosterone in males, estradiol and progesterone in fe- localization studies, identifying estrogen receptors by
males) . Sex steroid hormones play a role early in brain immunocytochemistry or in situ hybridization, con-
development in the "organization" of neural circuits, firmed the strong presence of estrogen receptors in the
and have critical roles throughout adulthood in "acti- hypothalamus and in brain areas with strong connec-
vating" specific behaviors. Although some generaliza- tions to the hypothalamus, including the amygdala, sep-
tions can be made, the specific neural circuits that are tal nuclei, the bed nucleus of the stria terminalis, the
influenced by sex steroid hormones and the time pe- medial part of the nucleus of the solitary tract, and the
riods of sensitivity to sex steroid hormone influence lateral portion of the parabrachial nucleus. However,
show dramatic species specificity. these studies showed relatively few estrogen receptors
The earliest studies of sex hormone influences on in the hippocampus, cerebellum, or the cerebral cortex
brain structure and function examined the effects of (Cintra et al., 1986; DonCarlos, Monroy, and Morrell,
testosterone and estradiol on sexual functions, includ- 1991; Simerly et al., 1990). No major differences were
ing reproductive capacity and behavior. In many spe- found in the distribution of estrogen receptors in males
versus females (Simerly et al., 1990).
In 1996 a new member of the steroid hormone re-
ceptor superfamily, one with a high sequence homology
JUDY L. CAMERON Departments of Psychiatry, Neuroscience,
and Cell Biology & Physiology, University of Pittsburgh, Pitts- to the classical estrogen receptor (now referred to as
burgh, Pennsylvania; Oregon Regional Primate Research Cen- ER-a), was isolated from rat prostate; this receptor
ter, Oregon Health Sciences University, Beaverton, Oregon. was named ER-b (Kuiper et al., 1996, 1998). This novel

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 59

estrogen receptor was shown to bind estradiol and to bution of androgen and estrogen receptors throughout
activate transcription by binding to estrogen response the brain (Michael, Clancy, and Zumpe, 1995; Resko
elements (Kuiper et al., 1998). In situ hybridization and Roselli, 1997; Simerly et al., 1990). The highest den-
studies examining the localization of ER-b mRNA have sity of androgen receptors is found in hypothalamic nu-
shown that these receptors are present throughout the clei known to participate in the control of reproduction
rostral-caudal extent of the brain, with a high level of and sexual behaviors, including the arcuate nucleus,
expression in the preoptic area, bed nucleus of the stria paraventricular nucleus, medial preoptic nucleus, and
terminalis, paraventricular and supraoptic nuclei, ventromedial nucleus, and in brain regions with strong
amygdala, and laminae II-VI of the cerebral cortex, connections to the hypothalamus, including the amyg-
and lower levels of expression in Ammon's horn and dala, nuclei of the septal region, the bed nucleus of the
the dentate gyrus of the hippocampus (Shughrue, stria terminalis, the nucleus of the solitary tract, and the
Lane, and Merchenthaler, 1997; Shughrue and Mer- lateral division of the parabrachial nucleus. Moderate
chenthaler, 2000). Demonstration of ER-b in regions of densities of androgen receptors have been shown in the
the brain previously thought to have relatively few es- hippocampus, cerebral cortex, and areas of the telen-
trogen receptors—specifically the cerebral cortex and cephalon involved with olfaction and the processing of
much of the hippocampus—has provided new insight vestibular and auditory information, as well as in motor
into how estrogen may modulate neural functions in nuclei of the brainstem and spinal cord. As with estro-
these brain regions. gen receptors, no major differences in androgen recep-
tor distribution have been found in males versus
PROGESTERONE RECEPTORS In females of a number of females. Androgen receptors are present early in fetal
species, progesterone can either facilitate or inhibit development and increase in number with gestation in
estrogen-induced sexual behaviors, depending on the both sexes (Resko and Roselli, 1997), a finding that ac-
timing of the increase in relation to rising estrogen lev- counts for the fact that developmental treatment of fe-
els during the ovarian cycle (Feder and Marrone, 1977). males with exogenous androgens can masculinize
Specific receptors for progesterone are induced by es- numerous behaviors.
trogen in hypothalamic regions of the brain, including
the preoptic area, the ventromedial and ventrolateral
Organizational versus activational effects
nuclei, and the infundibular/arcuate nucleus (Bethea
et al., 1992; Blaustein et al., 1988; DonCarlos, Greene, of sex steroid hormones
and Morrell, 1989). In the monkey, ovariectomy has The study of sex differences in the brain grew out of
been shown to decrease, but not eliminate, progeste- our understanding of how gonadal hormones influence
rone receptor levels in much of the hypothalamus, and sexual differentiation of the body with regard both to
in the supraoptic nucleus ovariectomy does not de- primary sexual differences between males and females
crease progesterone receptor levels, suggesting that
(the differentiation of the sexual organs) and to the
there are populations of neurons in which progeste-
development of secondary sexual differences (body fat
rone receptors are constitutively expressed (Bethea et
distribution, muscle development, breast development,
al., 1992). Progesterone has been shown to downregu-
differences in hair distribution). In the case of sexual
late its own receptors in many peripheral tissues of the
differentiation of the body, it is clear that exposure of
reproductive tract, but it does not appear to have this
males to various testicular secretory products, including
effect in many regions of the hypothalamus, although
testosterone, during early prenatal development leads
progesterone treatment does decrease progesterone re-
to sexual differentiation of the internal and external
ceptors in the ventromedial nucleus in the monkey (Be-
genitalia. Later activation of the reproductive axis at pu-
thea, Brown, and Kohama, 1996; Bethea et al., 1992;
berty, with a sustained increase in circulating testoste-
DonCarlos, Greene, and Morrell, 1989). In contrast,
rone, then leads to the development of secondary
there is evidence that progesterone can act to enhance
estrogen induction of some hypothalamic progesterone sexual characteristics. Thus, testosterone has both or-
receptors in the guinea pig (Brown et al., 1990; Don- ganizational and activational influences on the sexual
Carlos, Greene, and Morrell, 1989). Thus, there ap- differentiation of the body. Organizational effects of go-
pears to be cellular specificity to the mechanisms nadal hormones are conceptualized as resulting from
regulating progesterone receptor expression in various the early influence of gonadal hormones on structural
regions of the hypothalamus. development that do not require continued hormone
exposure to maintain sexual differentiation. Activational
ANDROGEN RECEPTORS Androgen receptor mapping effects are conceptualized as later stimulation of revers-
studies have shown considerable overlap in the distri- ible influences on sexual differentiation that require

60 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

continued exposure to gonadal hormones to maintain In contrast to the organizational effects of sex steroid
sex differences. hormones on the neural circuits that control behavior,
The concept that sex steroid hormones have impor- the more common actions of sex steroids on behavior
tant and permanent organizational effects on the de- in adulthood are often conceptualized as activational
veloping brain was originally postulated on the basis of effects. That is, sex steroid treatment causes a tempo-
experimental findings that treatment of developing rary and reversible change in neural function and be-
mice with testosterone produced permanent effects on havior. An example of an activational effect of a sex
reproductive capacity (Barraclough and Leathern, steroid on behavior is the rapid increase in display of
1954), with early treatment with testosterone blocking female sexual receptive behavior, lordosis, that occurs
later activation of ovulation by estradiol. A similar co- in female rats after a brief exposure to elevated
ordination of early and later influences of gonadal ste- estrogen.
roid hormones on reproductive behavior was first Despite the useful conceptual framework of early or-
reported by Phoenix and coworkers (1959). These in- ganizational effects and later activational effects of ste-
vestigators found that exposure of female guinea pigs roid hormones on neural function, there are clear
to testosterone in the prenatal period increased the examples in which effects of steroid hormones in adult-
likelihood of animals' displaying masculine sexual be- hood cause major long-term structural reorganization
haviors in adulthood, and simultaneously decreased of the brain. For example, castration of male hamsters
the likelihood of their displaying feminine sexual be- and rats in adulthood has been shown to lead to dra-
haviors. These observations formed the basis for the matic decreases in neural size and dendritic branching
hypothesis that, during early brain development, ex- within the medial amygdala, with testosterone treat-
posure to sex steroids can have long-lasting, organiza- ment in adulthood reversing these changes (Cooke et
tional effects that will influence CNS neuronal activity al., 1998; Gomez and Newman, 1991). Another example
and behavior throughout life. Although there are ex- of long-term organizational effects of steroid hormones
ceptions, it is now recognized that for a multitude of in adulthood is the neurotrophic effects of testosterone
diverse actions of sex steroid hormones on the brain on substance P-containing neurons of the medial nu-
there are specific "critical" periods in early brain devel- cleus of the amygdala and the bed nucleus of the stria
opment when sex steroid hormone exposure has per- terminalis in the rat (Malsbury and McKay, 1994).
manent organizational effects on the regions of the Male rats have almost twice the area of substance P-
brain mediating these actions (Cooke et al., 1998; containing neurons in these brain regions compared to
MacLusky and Naftolin, 1981). females, and this sexually dimorphic difference is
In general, exposure of males to testicular hormones dependent on the maintenance of adult levels of tes-
during prenatal and early postnatal periods leads both tosterone. Castrating males in adulthood leads to a dra-
to masculinization of some tissues and functions (mas- matic decrease in the area of substance P-containing
culine changes in genital structure, copulatory behav- neurons.
ior, and other behaviors characteristic of males) and to A great many actions of steroid hormones on the
defeminization of other tissues and functions (ovula- brain require both early organizational and later acti-
tory competence, feminine sexual behaviors such as lor- vational effects of the hormones, but this is not true in
dosis, and other behaviors characteristic of females). all cases. The ventromedial hypothalamus (VMH) is a
Because the various steroid-sensitive tissues of the body sexually dimorphic nucleus that plays key roles in the
and brain have differing critical periods for the orga- control of feminine reproductive behaviors, including
nizational effects of testosterone, in some congenital maternal behavior and lordosis in rats (Madeira and
syndromes and under some experimental conditions Lieberman, 1995; Pfaff, 1980). The VMH is significantly
certain tissues of the body and brain may become mas- larger in males than in females, but individual cell nu-
culinized while others do not. Thus, it is often impor- clei are larger in females (Dorner and Staudt, 1969).
tant to consider the degree of masculinization or These differences depend on organizational effects of
feminization of specific tissues and functions. steroid hormones in the first few days of postnatal life,
In rodents, the critical period for steroid hormone- such that castration of males on postnatal day 1, but not
mediated organization of brain regions and sexually di- postnatal day 7, reduces the volume of the VMH to that
morphic behaviors appears to be postnatal, with most of control females (Dorner and Staudt, 1969). In adult-
effects occurring during the first 10 days of life. In pri- hood the VMH is a critical site for the induction of lor-
mates, sexual differentiation of the brain occurs pre- dosis behavior in response to elevations in estradiol and
natally, over an extended period in midgestation progesterone (Pfaff, 1980). Estradiol treatment of fe-
(Phoenix, Goy, and Resko, 1968). male rats leads to a number of acute changes in the

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 61

VMH, including an increase in protein synthesis and an
increase in dendritic spines in VMH neurons, and in-
duction of progesterone receptors and receptors for an-
other reproductive hormone oxytocin (McEwen, 1991).
Subsequent progesterone exposure leads to changes in
the location of oxytocin receptors and makes the estra-
diol-primed female sensitive to oxytocin-induced lor-
dosis (McEwen, 1991). Some of these actions depend
on both the organizational and activational effects of
steroid hormones on VMH neurons. For example, es-
tradiol increases dendritic spines in VMH neurons in
the female, but the VMH neurons in the male are re-
fractory to this action of estradiol (Frankfurt et al.,
1990; McEwen, 1991). Also, estradiol treatment sensi-
tizes this nucleus to actions of progesterone in females,
but not in males (Rainbow, Parsons, and McEwen, FIGURE 5.1 Pathways for metabolism of testosterone in brain
1982). Other actions of estradiol in this nucleus depend tissue.
only on acute estradiol exposure. For example, estra-
diol induces oxytocin receptors in the VMH in adult-
hood in both female and male rats (Coirini, Johnson, of androgens to estrogens in or near brain target cells
and McEwen, 1989). These contrasts highlight the need accounts for actions of androgens on behavior and sex-
for specificity when determining the contributions or- ual differentiation—was proposed.
ganizational versus activational actions of a sex steroid Whereas a great deal of data indicate that many ac-
hormone, both in reference to the particular action of tions of androgens on behavior in rodents are mediated
a steroid hormone and the cell type upon which it is by local conversion to estrogen, a similar role for aro-
acting. matization in regulating the behavioral effects of an-
drogens on behavior in primate species is more
controversial. Early evidence that androgens may not
The role of steroid metabolism in mediating
need to be converted to estrogen to influence behavior
the actions of sex steroids on the brain came from studies in rhesus monkeys showing that pre-
AROMATIZATION In the early 1940s, Frank Beach natal treatment of female monkeys with dihydrotestos-
(1942) showed that castrated male rats receiving exog- terone was effective in masculinizing sexual behavior
enous estrogen recovered normal male mating behav- (Goy and Resko, 1972). Later studies showed that estra-
ior, providing the first evidence that metabolism of diol was not effective in restoring sexual activity in cas-
androgens to estrogen plays a key role in mediating an- trated macaques (Michael, Zumpe, and Bonsall, 1990;
drogen actions at the level of the central nervous sys- Phoenix and Chambers, 1982) and that the addition of
tem. Later, experiments in female rodents showed that estradiol to dihydrotestosterone therapy in castrated
perinatal estradiol treatment was as effective as testos- macaques does not further increase sexual behavior
terone treatment both in suppressing lordosis and in (Michael, Bonsall, and Zumpe, 1987). Such studies fur-
stimulating male-like mounting behavior (Barraclough ther supported the notion that, in the primate, aro-
and Gorski, 1962; Levine and Mullins, 1964; Whalen matization of androgens to estrogen may not be
and Nadler, 1963). A more definitive indication that important in mediating the behavioral effects of andro-
aromatization of androgen to estrogen (figure 5.1) is gens. However, aromatase is present in the primate
necessary for actions of androgens to stimulate sexual brain (Abdelgadir et al., 1997; Naftolin et al., 1996) and
behavior was provided by studies showing that treat- is regulated by androgen exposure (Roselli and Resko,
ment of female or castrated male rats with dihydrotes- 1989; Resko et al., 1993). Moreover, recent studies have
tosterone (DHT), a nonaromatizable androgen, was shown that aromatase inhibitors can decrease male sex-
ineffective in stimulating male sexual behavior (Brown- ual behavior in macaques (Zumpe, Bonsall, and Mi-
Grant etal., 1971; McDonald etal., 1970). Shortly there- chael, 1993). Thus, aromatization of androgens to
after, the first direct evidence that neural tissues estrogens does occur in the brains of primates and ap-
aromatize androgens to estrogens was provided by Naf- parently plays a role in regulating at least some behav-
tolin and colleagues (1971, 1977). Based on this work, ioral effects of androgens; however, aromatization of
the "aromatization hypothesis"—that local conversion androgens does not appear to be as critical a step in

62 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

 mediating primate androgen-induced behaviors as it is
in rodent species.
The aromatase enzyme complex that catalyzes andro-
gens to estrogens is composed of a specific aromatase,
cytochrome P450, P450XIXAI, the product of the
CYP19 gene, and NADPH-cytochrome P450 reductase,
with a cellular location in the endoplasmic reticulum
(Lephart, 1996; Nebert and Gonzales, 1987). Aroma-
tase activity has been localized in specific brain regions
by demonstration of enzyme activity and mRNA levels,
including many brain nuclei that show sexually dimor-
phic differences in structure (Abdelgadir et al., 1997;
Roselli and Resko, 1987). In the brain, aromatase activ-
ity is specific to neurons, glial cells being devoid of the
enzyme (Abe-Dohmae, Tanaka, and Harada, 1994).
There is an abundance of aromatase activity in neurons
of the hypothalamus, preoptic area, and limbic region;
but it is present at much lower levels or undetectable in
a number of other brain regions including many
regions of the cerebral cortex, hippocampus, and cer-
ebellum (Abdelgadir et al., 1997; Lephart, 1996).
In many, but not all, regions of the brain, aromatase
activity is regulated by androgen concentrations, as in-
dicated by studies showing that androgen treatment in-
creases aromatase activity while castration decreases FIGURE 5.2 Developmental time course of hypothalamic aro-
matase activity (top panel) and testosterone secretion (bottom
aromatase activity, and by the finding that males have panel) in the male rat. (Adapted from McEwen, B. S., 1992.
higher levels of aromatase activity than females in many Steroid hormones: Effect on brain development and function.
brain regions (Roselli and Resko, 1993). Further evi- Harm. Res. 37:1-10.)
dence for androgen regulation of aromatase is provided
by data collected in Tfm (testicular feminization mu-
tation) mice, which have a defective structural gene for 3a-hydroxysteroid dehydrogenase pathway (figure 5.1).
the androgen receptor. These mice have aromatase lev- Transformation of testosterone into dihydrotestoste-
els that are significantly decreased relative to control rone (DHT) by 5a-reductase plays a critical role in dif-
males (Rosenfeld et al., 1977). A notable exception is ferentiation of some androgen-dependent structures in
the aromatase in areas of the limbic system, including the periphery—in particular, the prostate. DHT binds
the bed nucleus of the stria terminalis and amygdaloid to androgen receptors, but with about four times the
regions where aromatase enzymatic activity appears to affinity of testosterone (Grino, Griffin, and Wilson,
be regulated by an androgen-independent mechanism 1990). The androgen receptor complex appears to be
(Jakab et al., 1993; Roselli and Resko, 1993). more stable when DHT is bound, compared to testos-
Brain aromatase activity shows a marked developmen- terone, and DHT exhibits a much slower dissociation
tal pattern of expression, first appearing in the hypo- rate from the androgen receptor (Grino, Griffin, and
thalamus of rats on gestational day 16, peaking at Wilson, 1990). Thus, metabolism of testosterone to
gestational day 19, and declining to low but detectable DHT within specific cells provides a more potent an-
levels in the perinatal period, then further declining at drogen signal to those cells. There are two forms of 5a-
puberty (figure 5.2; Lephart et al., 1992). However, the reductase, type 1 and type 2. In the prostate, type 2
factors that regulate the developmental changes in enzyme plays a critical role in differentiation.
brain aromatase activity are not well understood, and The brain has both type 1 and type 2 forms of 5a-
little is known about the developmental pattern of aro- reductase (Normington and Russell, 1992). There ap-
matase expression in specific brain nuclei (Lephart, pears to be a preponderance of the type 1 isoform in
1996). adult brain, while the type 2 isoform varies with age,
being maximal in the perinatal period (Negri-Cesi, Po-
5a-REDUCTASE A second metabolic pathway for andro- letti, and Celotti, 1996). Interestingly, the formation of
gens that is present in the brain is the 5a-reductase— DHT appears to be severalfold higher in white matter

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 63

structures than in rest of the brain (Celotti, Melcangi, tracellular receptors for these hormones (described
and Martini, 1992). In general, neither in rodents nor earlier), once bound by hormone, bind to hormone
primates does the 5a-reductase system appear to be reg- response elements and act together with other tran-
ulated by the presence or absence of sex steroid hor- scription factors to regulate gene expression (Truss and
mones, and no sexual dimorphism in its distribution Beato, 1993). Neurons that are targets for this form of
has been reported (Negri-Cesi, Poletti, and Celotti, steroid hormone modulation must have both receptors
1996; Sholl, Goy, and Kim, 1989). The role of DHT in for a particular steroid hormone and the appropriate
guiding sexual differentiation of neural systems in the hormone response element. It has been generally be-
brain is currently unknown. However, in the rat there lieved that actions of steroid hormones mediated via
appears to be a peak in 5a-reductase activity in the neo- intracellular receptors are relatively slow to occur and
natal period (Negri-Cesi, Poletti, and Celotti, 1996); this are long-lasting. However, more recent evidence indi-
peak is coincident with the neonatal secretory peak of cates that some effects mediated by intracellular steroid
testosterone that plays a critical role in sexual differ- hormone receptors can occur quite rapidly (Boyle et
entiation of a number of behaviors in the rodent (Weisz al., 1987). There is considerable evidence that intra-
and Ward, 1980), suggesting that DHT may be impor- cellular steroid hormone receptors play a role in
tant in this process. regulating synthesis of a variety of neuropeptides,
neuropeptide and neurotransmitter receptors, and reg-
NEUROSTEROIDS The term "neurosteroid" refers to ste- ulatory enzymes for neurotransmitter synthesis (Alonso
roids that are synthesized within the nervous system, and Lopez-Coviella, 1998).
either de novo from cholesterol or by metabolism of Steroid hormones can also act via nongenomic mech-
precursors originating from the circulation (Baulieu, anisms. Actions that are mediated this way generally oc-
1991). Enzymes for the production of progesterone and cur quite rapidly, on the order of milliseconds to
its steroid precursor pregnenolone have been demon- minutes, and include alterations in neuronal excita-
strated in the brain (Guennoun et al., 1995; Mellon, bility and neuropeptide release (Alonso and Lopez-
1994). The A5,3(3-hydroxysteroid dehydrogenase isom- Coviella, 1998; McEwen and Alves, 1999). Estrogen
erase enzyme (figure 5.3) that converts pregnenolone binding sites have been reported on cell membranes
to progesterone has been found to be widely distributed (McEwen and Alves, 1999), but specific membrane-
in the brain, both in neurons and glial cells, and the bound estrogen receptors have not yet been cloned.
production of tritiated progesterone from tritiated However, recent studies using transient transfection of
pregnenolone has been demonstrated in cultured glial ERa and ERb in Chinese hamster ovarian cells have
cells (Guennoun et al., 1995). Changes in the concen- demonstrated that both of these estrogen receptors can
tration of pregnenolone and its derivatives, which vary be expressed in the cell membrane as well as in the
in concert with behavioral changes, have been found to nuclear fraction (Razandi et al., 1999). A-ring-reduced
be correlated with behavioral performance, including metabolites of progesterone bind to GABAA receptors
performance on memory tests (Baulieu, 1997). En- on the cell membrane and increase GABA-dependent
zymes that can metabolize progesterone, including chloride currents within neurons (Barker et al., 1987;
5a-reductase and 3a- and 3b-hydroxysteroid oxidore- Gee, 1988; McEwen, 1991).
ductases, have been detected in the brain; and these Increasingly, there is evidence that steroid hormones
allow production of the 5a-reduced metabolites of pro- can modulate neuronal activity by acting via intracel-
gesterone that have been shown to interact with the lular second messenger systems. In the hypothalamus,
GABAA receptor (Baulieu, 1997; Majewska, 1992). Such estrogen-induced depolarization has been reported to
progesterone metabolites have been reported to have activate a cAMP second messenger system, coupled to
anesthetic, anticonvulsant, hypnotic, anxiolytic, and G proteins (Minami et al., 1990; Nabekura et al., 1986).
analgesic activities (Baulieu, 1997). Estrogen has also been reported to activate MAP ki-
nases, ERK-1 and ERK-2, in cortical neurons (Toran-
Mechanisms by which sex steroids modulate Allerand, Singh, and Setalo, 1999). Changes in cellular
neuronal activity calcium flux constitute a third intracellular messenger
system that is affected by estrogen. In rat neostriatal
Over the past few years it has become apparent that sex neurons it has been shown that l7b-estradiol rapidly
steroid hormones can modulate neuronal function by suppresses currents mediated by L-type calcium chan-
acting via a variety of different mechanisms (Alonso and nels (Mermelstein, Becker, and Surmeier, 1996)—an
Lopez-Coviella, 1998; McEwen and Alves, 1999; Toran- action that is probably mediated via a membrane recep-
Allerand, Singh, and Setalo, 1999). The well known in- tor. In the hippocampus, however, estradiol has been

64 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

FIGURE 5.3 Biosynthesis of neurosteroids. Enzymes in this hydroxysteroid dehydrogenase (3bHSD), 5a-reductase, and
pathway include side-chain cleavage cytochrome P450, 3b- 3a-hydroxysteroid oxidoreductase.

shown to increase both sustained and transient calcium Sexual dimorphisms in the brain
currents—actions that may well be mediated by ge-
nomic actions of this steroid hormone (Joels and Karst, PROCESSES AFFECTED BY SEX STEROIDS Sex steroid
1995). hormones lead to sexual differentiation of behavior by
There is also evidence that estrogen may act as a neu- affecting a wide variety of cellular mechanisms within
ral growth factor within the brain, influencing neuronal the nervous system (Cooke et al., 1998). Processes that
development, survival, plasticity, and regeneration have been shown to be modulated by sex steroid hor-
(Toran-Allerand, Singh, and Setalo, 1999). Neurons in mones include neurogenesis (Gurney and Konishi,
the forebrain coexpress estrogen and neurotrophin re- 1979; Jacobson and Gorski, 1981), cell migration (Sen-
ceptors, and these neurons are also sites of both estro- gelaub and Arnold, 1986), growth of the neuronal soma
gen and neurotrophin synthesis. Moreover, estrogen (Breedlove and Arnold, 1981), dendritic growth, differ-
and nerve growth factor can reciprocally regulate the entiation and synapse formation (Frankfurt et al., 1990;
expression of each other's receptors, and accumulating Goldman and Nottebohm, 1983; Gomez and Newman,
evidence suggests that there is cross-coupling of their
1991; Gould, Woolley, and McEwen, 1990; Juraska,
signaling pathways (Toran-Allerand, Singh, and Setalo,
1991; Raisman and Field, 1973; Woolley and McEwen,
1999).
1992), synapse elimination (Leedy, Beattie, and Bres-
Further evidence of complex interactions between
steroid hormones and neural signaling systems comes nahan, 1987), neuronal atrophy and apoptosis (Davis,
from studies showing that dopamine can modulate sex- Shryne, and Gorski, 1996; McEwen, 1996; Woolley,
ual behavior in rodents via a mechanism that requires Gould, and McEwen, 1990), neuropeptide expression
the presence of intracellular progesterone receptors (De Vries, 1990; Malsbury and McKay, 1994), the ex-
within hypothalamic neurons (Mani, Blaustein, and pression of neurotransmitter receptors (Levesque, Gag-
O'Malley, 1997). This action of dopamine involves a non, and Di Paolo, 1989; McEwen, 1991; Turner and
ligand-independent activation of progesterone recep- Weaver, 1985), and neuronal excitability (Teyler et al.,
tors (Mani et al., 1994). 1980; Wong and Moss, 1992).

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 65

EXAMPLES OF SEXUALLY DIMORPHIC BRAIN REGIONS estradiol receptors in the brain (Herbison and Theo-
AND BEHAVIORS In mammalian species numerous sex dosis, 1992), and there is evidence that there are more
differences in brain structure and function have now estradiol receptors in the female than in the male
been documented. Sexual dimorphic brain areas in- (Brown, Naftolin, and MacLusky, 1992). This area also
clude a number of regions of the hypothalamus, in- accumulates androgens to a greater degree in males
cluding the medial preoptic area and the ventromedial than in females (Jacobson, Arnold, and Gorski, 1987).
hypothalamus, the vomeronasal system, amygdala, bed The SDN-POA is an excellent example of a nuclear re-
nucleus of the stria terminalis, hippocampus, striatum, gion in which organizational influences of testosterone
cerebellum, and various regions of the cerebral cortex. play a critical role and aromatization of testosterone to
Behaviors showing documented sex differences include estradiol is an important intermediary step. Castration
behaviors associated with reproduction (mating and of males in the early neonatal period leads to a per-
maternal behaviors), aggression, activity, and various manent decrease in the size of the SDN-POA, whereas
cognitive functions including spatial cognition, verbal treatment of female rats with testosterone in this same
skills, and various aspects of learning and memory. Ex- time period leads to a permanent increase in size of the
amples of sexually dimorphic brain regions and behav- SDN-POA (Gorski et al., 1978, 1980). Neonatal treat-
iors in nonmammalian species have also proven to be ment with estrogen is even more effective than early
very important for learning about basic mechanisms androgen treatment in increasing the size of the nu-
governing the establishment and maintenance of sex- cleus, and neonatal treatment of males with an anties-
ual dimorphisms, including bird and frog song systems. trogen decreases the size of the SDN-POA (Dohler et
It is beyond the scope of this chapter to review all of al., 1984). In this nucleus there is evidence that either
these areas and behaviors. Here, we review areas and androgen or estrogen treatment increases neurogenesis
behaviors that demonstrate especially useful examples in the early neonatal period (Jacobson and Gorski,
of general principles regarding sexual dimorphisms, or 1981), and later prevents programmed cell death (Da-
are particularly relevant to the topic of developmental vis, Shryne, and Gorski, 1996). Despite the abundance
cognitive neuroscience. of studies on the SDN-POA of the rat, and the profound
sexual dimorphism of this nucleus, there is relatively
The medial preoptic area and sexual behaviors One of the little information about the role of the SDN-POA in
first reports of a sexual dimorphism in a brain structure mediating behavior. Lesions of the nucleus have been
was by Raisman and Field (1973), who examined the rat reported to have few behavioral effects (Arendash and
medial preoptic area (mPOA; a brain area that plays an Gorski, 1983). Nevertheless, the mPOA has been shown
important role in the regulation of sexual behaviors) by to have significant sexual dimorphism in a wide variety
electron microscopy and found that females had more of species (Ayoub, Greenough, and Juraska, 1983; Com-
dendritic spine synapses and fewer axosomatic synapses mins and Yahr, 1985; Hines et al., 1985; Tobet, Zahniser,
than males. Moreover, they showed that the synapse and Baum, 1986), including humans (Allen et al., 1989;
structure was modulated by the presence of androgen LeVay, 1991; Swaab and Fliers, 1985); thus, there re-
in the neonatal period. Previous reports had shown that mains a great deal of interest in the possible function
the mPOA plays a role in male copulatory behavior in of this brain area in mediating sexually dimorphic
the rat, that implantation of testosterone into the behaviors.
mPOA of castrated males would restore mating behav- Other key areas involved in mediating sexual behav-
ior, and that similar implants in females would induce ior in the rodent are the VMN, the vomeronasal system,
male-like mounting and copulatory behaviors (Madeira and the bed nucleus of the stria terminalis (Cooke et
and Lieberman, 1995). Thus, this was the first report of al., 1998). In many cases, these brain areas show sexual
a structural difference in the brain induced by a go- dimorphisms in structure. And in many cases gonadal
nadal steroid hormone, one that may underlie a known hormones have both organizational and activational
behavioral effect of that hormone. activities.
Later studies by Gorski and colleagues (1978, 1980)
identified a more striking sexual dimorphism in the The bird song system Three years after Raisman and
mPOA. They showed a region of densely packed, darkly Field's report of subtle structural sexual dimorphisms
staining cells in the mPOA that was 2.5-5 times larger in synapse structure in the mPOA of the rat, Nottebohm
in the male rat than in the female rat—a region they and Arnold (1976) published a landmark report show-
designated as the sexually dimorphic nucleus of the me- ing a much more robust sexual dimorphism in brain
dial preoptic area, or SDN-POA. In both male and fe- structure in songbirds. They found that the nuclei in
male rats this region has the highest concentration of the vocal control areas of male songbirds (canaries and

66 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

zebra finches; species in which the male sings but the activity responses when the striatal dopamine system is
female does not) were five to six times the size of these stimulated on the evening of proestrous, when circulat-
same areas in their female counterparts. Later studies ing estradiol levels are high, compared to the day of
showed that species in which both the male and female diestrous, when circulating estradiol levels are much
sing exhibit no sexual dimorphism in these brain areas lower (Becker and Cha, 1989). They also show increases
(Brenowitz, Arnold, and Levin, 1985). Since that time in extracellular dopamine concentration within the
a great deal of research on the bird song system has striatum, amphetamine-stimulated striatal dopamine re-
been performed and there is excellent evidence linking lease, dopamine metabolism, and increased dopamine
structural differences between sexes with sexual dimor- receptors on estrous compared to diestrous (Becker,
phism of singing behavior (Bottjer and Arnold, 1997). 1990, 1999; Di Paolo, Falardeau, and Morisette, 1988;
In zebra finches, both organizational and activational Levesque, Gagnon, and Di Paolo, 1989). With ovariec-
affects of testosterone are apparent, and testosterone tomy, stimulated activity is decreased (Camp, Becker, and
works through aromatization to estrogen. When treated Robinson, 1986), as well as stimulated strial dopamine
with estrogens early in life and then given testosterone release and dopamine receptor density (Becker and
or estrogen steroids as adults, female zebra finches will Ramirez, 1980; Levesque, Gagnon, and Di Paolo, 1989).
sing in adulthood (Gurney and Konishi, 1979). In the Estradiol replacement can reverse these changes, al-
neonatal time period, estrogen treatment increases though there are differences with acute versus more
neuronal number (Gurney and Konishi, 1979). Sur- chronic estradiol treatment (Becker, 1999).
prisingly, in zebra finches the converse experiment— In male rats, estrogen does not affect striatal dopa-
castrating males or treating them with anti-estrogens— mine release; in addition, castration has little effect on
has not been effective either in preventing structural this system (Becker, 1999). Comparison of male and fe-
differentiation of male-like nuclei in vocal control areas male rats shows that there is sexual dimorphism in stria-
or in preventing singing, leading to the proposal that tal dopamine release and dopamine receptors. In
male songbirds may also have a steroid-independent ge- castrated male rats, basal extracellular dopamine con-
netic mechanism regulating development of the nuclei centrations in the striatum are twice as high as in ovari-
in the vocal control areas (Arnold, 1996, 1997). ectomized female rats (Xiao and Becker, 1994), and
The neural systems underlying song production in amphetamine-induced dopamine release is also higher
the canary differ from those in the zebra finch in that in males (Becker and Ramirez, 1980). In addition, male
organizational effects of androgens remain apparent rats have more dopamine D1 receptors than do females
into adulthood. Female canaries treated with andro- (Hruska et al., 1982).
gens in adulthood show structural changes in the nuclei It seems likely that these sex differences in activity
of the vocal control areas, specifically increased den- and ease of activation of the striatal dopamine system
dritic growth, eventually resulting in song production play important roles in governing behavior in various
(DeVoogd and Nottebohm, 1981; Goldman and Not- circumstances. Becker has postulated, based on studies
tebohm, 1983). This is a good example of steroid hor- administering estrogen directly into the striatum, that
mones in adulthood causing long-term organizational this system plays a role in pacing the rate of sexual en-
effects in the brain. counters between male and female animals to optimize
fertility (Becker, 1999; Xiao and Becker, 1997). One
Midbrain dopaminergic systems and motor activity The practical implication of these sex differences is that sci-
striatum is part of the midbrain system that plays a criti- entists need to be aware that sexual differences in activ-
cal role in sensorimotor integration. The striatum re- ity may underlie sexual differences that are measured
ceives input fibers from the motor and associated in animal's performances on a variety of laboratory
cortical areas, as well as from the substantia nigra. The tests, including tests of learning and memory.
nigrostriatal projection is dopaminergic, containing
more than 90% of the dopamine neurons in the brain. The hippocampus and learning and memory Clinical stud-
When dopamine activity in this pathway is increased, ies showing that lesions of the hippocampus lead to
animals display increased motor activity to sensory stim- severe memory damage indicate that this area of the
uli, whereas when dopaminergic activity in this pathway brain can play a critical role in learning and memory,
is decreased, animals become hyporesponsive to sen- specifically with regard to deficits in the formation of
sory input (Becker, 1991). new memories (anterograde amnesia; Milner, Teuber,
In females, estrogen and progesterone modulate the and Corkin, 1968). The hippocampus is known to
activity of this dopaminergic pathway (Becker, 1999). be influenced both by gonadal steroid hormones
Female rats have been shown to have greater behavioral and by adrenal hormones, specifically glucocorticoids.

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 67

Microstructure within the hippocampus has also been There are also functional changes in hippocampal
shown to be influenced by early experience and rearing neurons associated with experimentally manipulated or
environment (Juraska, 1991). Thus, interactions be- naturally occurring fluctuations in estradiol and pro-
tween these two endocrine systems and experiences of gesterone in the female rat. In general, there appears
an individual during development combine to result in to be an increase in neuronal excitability in the hip-
long-term consequences for this brain area, both in pocampus with elevated estradiol levels. EPSP duration
structure and function. is lengthened following estradiol administration, and
The hippocampus is composed of three distinct population spike amplitude is elevated in response to
regions, connected in a circuit—the dentate gyrus, estradiol (Teyler et al., 1980; Wong and Moss, 1992).
which has granule neurons that send mossy fibers to the Seizure threshold in the dorsal hippocampus is also
pyramidal neurons of the CA3 region of Ammon's lower following estrogen administration to ovariecto-
horn, which in turn send collaterals to the pyramidal mized female rats or on the day of proestrous in the
neurons in the CA1 region. In both males and females, naturally occurring estrous cycle (Buterbaugh and
cells in the CA1 region of the hippocampus contain ER Hudson, 1991; Teresawa andTimiras, 1968). Long-term
(Loy, Gerlach, and McEwen, 1988; Simerly et al, 1990), potentiation (LTP) in response to standardized stimu-
while ER-b receptors are present in Ammon's horn and lus trains has also been shown to be greater in the hip-
the dentate gyrus (Shughrue, Lane, and Merchen- pocampus of proestrous rats compared to other times
thaler, 1997; Shughrue and Merchenthaler, 2000). Es- in the estrous cycle (Warren et al., 1995).
tradiol administration can alter the morphology of Although estradiol receptors do not show sexual di-
neurons in the CA1 region, increasing the number of morphism in their distribution within the hippocam-
spines and the amount of branching in apical dendrites pus, there is a sexual dimorphism in glucocorticoid
and increasing the density of synapses in the hippocam- receptors, with female rats having larger numbers of
pal CA1 stratum radiatum, whereas the opposite glucocorticoid receptors compared to males (Turner
changes occur with ovariectomy (Gould et al., 1990; Jur- and Weaver, 1985). There is also an interaction between
aska, 1991; Woolley and McEwen, 1992). Progesterone gonadal hormones and the glucocorticoid receptor sys-
administration initially potentiates the effects of estra- tem such that ovariectomy in females increases gluco-
diol on neurons in this brain region, but later triggers corticoid receptors while castration in males has no
down-regulation of spines on CA1 neurons (Gould et effects on glucocorticoid receptors (Turner and
al., 1990; Woolley and McEwen, 1993). Moreover, nat- Weaver, 1985).
urally occurring changes in estradiol and progesterone Glucocorticoids have significant effects on hippocam-
across the rat estrous cycle have been associated with pal neuronal structure and function. In both the devel-
changes in synaptic density and dendritic morphology oping and adult rat, adrenalectomy leads to a decrease
in hippocampal CA1 neurons (Woolley and McEwen, in size and dendritic arborization of neurons in the den-
1992, 1993; Woolley et al., 1990). Estrogen effects on tate gyrus, which can be prevented with adrenal steroid
CA1 neurons appear to be mediated by an NMDA supplementation (Gould, Woolley, and McEwen, 1990;
receptor-mediated mechanism, in that NMDA recep- Woolley et al., 1991). In contrast, in the CA3 region of
tor antagonists block estrogen induction of spines and Ammon's horn high levels of glucocorticoids cause at-
estrogen induces NMDA receptors in the CA1 region rophy of pyramidal neurons (Woolley, Gould, and
(Woolley and McEwen, 1994). In the adult rat, the ef- McEwen, 1990). Studies by Sapolsky (1990) show that
fects of estradiol on the hippocampus are sexually di- glucocorticoids inhibit glucose uptake by pyramidal
morphic, with male rats showing fewer effects on neurons, which precipitates atrophy and eventual cell
hippocampal synapse formation when treated with es- death. These effects of glucocorticoids appear to play a
trogen (Lewis, McEwen, and Frankfurt, 1995). This ap- role in the loss of CA3 neurons with aging, in that this
pears to result from organizational effects of androgen loss can be prevented by adrenalectomy (Sapolsky,
(converted to estrogen) on the hippocampus, as treat- 1990). Glucocorticoids also lead to atrophy of CA3 neu-
ment of neonatal male rats with an aromatase inhibitor rons in conditions of chronic stress (McEwen, 1996).
can eliminate this sex difference in adulthood (Lewis, In addition to plasticity induced by hormonal signals,
McEwen, and Frankfurt, 1995). Developmental changes environmental conditions during development also
in estrogen receptors (O'Keefe and Handa, 1990) and have a marked influence on morphology of neurons in
aromatase (MacLusky et al., 1987) have been reported the hippocampus, and there are clear sexual dimor-
in the hippocampus, and these developmental changes phisms in the hippocampal responses to environmental
may play a critical role in sexual differentiation of this stimuli (Juraska, 1991). For example, granule cells in
region of the brain. the dentate gyrus of male rats raised in single cages

68 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

show relatively few changes in their dendritic morphol- maze than do the same lesions in male rats (Therrien,
ogy as opposed to animals raised in a complex environ- 1982).
ment containing both toys and other rats. In contrast, In humans, there are also reports of sex differences
these same hippocampal granule cells in female rats in spatial problem solving abilities, with males showing
show a marked plasticity in response to rearing in a greater proficiency (Delgado and Prieto, 1996; Linn
complex environment, with an increase in length of a and Petersen, 1985; Witkin and Berry, 1975). But these
specific population of dendrites. These sex differences differences are small in magnitude, and apparent only
appear to be regulated by organizational influences of in terms of population statistics; there is considerable
testosterone in the neonatal period, in that neonatally overlap in ability among individual males and females.
castrated males show plasticity equivalent to females It is also important to remember that spatial problem
(Juraska, 1991). solving is a higher order cognitive task and that there
Numerous reports have indicated sex differences in is evidence that sexual dimorphism in this task reflects
the processes of learning and memory, particularly not only differences in hippocampal function but also
learning that involves the use of spatial abilities, a task functional differences within other regions of the cor-
believed to be strongly dependent on hippocampal tex (De Courten-Myers, 1999; Kimura, 1987; Wis-
function (O'Keefe, 1993). Using a variety of different niewski, 1998). There is a small amount of evidence that
learning paradigms, including passive avoidance learn- prenatal androgen exposure may affect spatial problem
ing, active avoidance learning, and complex maze solving abilities. Resnick and colleagues (1986) re-
learning (using the radial-arm maze, T-maze, and Mor- ported that females with congenital adrenal hyperplasia
ris water maze), differences have been shown in the per- (a syndrome involving increased exposure to adrenal
formance of male versus female rats (Beatty, 1979; androgens in prenatal development) had significantly
Becker, 1991). In complex maze learning, male rats enhanced performance on tests of spatial ability com-
generally have been found to learn more rapidly and pared to their unaffected female relatives. However, the
with fewer errors compared to female rats. However, the effects of sex steroids on this ability do not appear to
issue of whether better performance by males results follow the typical organizational and activational
from differences in hippocampal function or from dif- scheme. There is limited evidence that sex steroid hor-
ferences in activity, which appear to be governed by mone levels in adulthood may influence spatial prob-
other brain systems (as discussed in the previous sec- lem solving abilities, with elevated levels of estradiol in
tion; see Midbrain Dopaminergic Systems and Motor females and testosterone in males being associated with
Activity), is not clear for some of these tasks. Beatty lower, rather than higher, performance on tests of spa-
(1979) argued that females make more errors in many tial ability. Several studies report that women perform
mazes simply because they are more active, and more worse on tests of spatial problem solving during the
active motions translate into a greater occurrence of midpoint of the menstrual cycle, when estradiol levels
errors. In an attempt to distinguish sex differences in are elevated, compared to other times during the men-
learning from differences in activity, some investigators strual cycle (Broverman et al., 1981; Hampson, 1990a,b;
have focused on the rate of initial learning of a maze Hampson and Kimura, 1988; Komnenich et al., 1978;
compared to that of later maze performance. In one Wickham, 1958). Moreover, men with higher testoste-
such study, Einon (1980) showed that in a radial-arm rone levels have been reported to perform worse on
maze males were more likely than females to use a suc- tests of spatial ability than their counterparts with lower
cessful strategy of visiting adjacent arms. Further studies testosterone levels (Gouchie and Kimura, 1991; Shute
with this paradigm have shown that males or neonatally et al., 1983).
androgenized females show faster acquisition of radial-
arm maze skill compared to females or neonatally cas- Cerebral cortical areas and higher cognitive functions We
trated males, but that following skill acquisition the sex are still in the very early stages of understanding how
difference on this task disappears (Williams, Barnett, sex differences in most higher cognitive functions and
and Meek, 1990). It thus appears that there may be mi- complex behaviors are linked to sex differences in the
nor sex differences in the cognitive strategies that rats brain. In general, research in this area has suffered
utilize to solve spatial learning paradigms, although from inadequate delineation of the behaviors, the an-
both sexes show equal mastery of these tasks in the end. atomical brain regions, and the functional neural sys-
Further evidence that there are sex differences in hip- tems that have been studied. Another confound in this
pocampal function comes from studies showing that bi- area of investigation has been that many of these func-
lateral hippocampal lesions in female rats lead to far tions can be studied only in humans (e.g., studies
greater impairment of function in the Morris water of speech), such that invasive experiments are rarely

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 69

possible and control of factors such as developmental in women. Evidence for organizational effects of pre-
experiences is much more difficult. With the advent of natal sex steroid hormones on the degree of laterali-
various neural imaging technologies, however, more in- zation of verbal skills was provided by a study by Hines
formation is becoming available concerning sexual di- (1982), which found that women whose mothers took
morphisms in the brain, particularly in the cerebral diethylstilbesterol (a potent estrogenic compound) in
cortex. Moreover, functional magnetic resonance im- pregnancy to prevent miscarriage had greater laterali-
aging (MRI) is beginning to link differences in cogni- zation than their nonexposed sisters.
tive abilities with differences in specific anatomical There are also reports of sex differences in the size
regions and patterns of neural circuit activation. of cortical regions involved in speech (figure 5.4). Wada
Studies of sex differences in verbal abilities provide and colleagues (1975) reported that the planum tem-
examples of the considerable difficulty in delineating porale (a flat region of the temporal lobe in a language-
the neuroanatomical substrates underlying a sexually associated area) is asymmetric, being larger on the left
dimorphic higher order behavior. A number of reports than the right side of the brain, but that this asymmetry
suggest that women perform better on some verbal was less marked in women than in men. This finding
tasks than do men (Hines, 1991; Hyde and Linn, 1988). was confirmed in two recent studies (Kulynych et al.,
There is limited evidence that this sexual dimorphism 1994; Witelson and Kigar, 1992). More recently, Witel-
is related to sex steroid hormones. Several studies have son, Glezer, and Kigar (1995) have reported that
reported that women's performance on verbal tasks is women have a greater density of neurons in this cortical
better when estradiol, or estradiol and progesterone, region compared to men. In contrast, however, Rabi-
are elevated during the menstrual cycle than when cir- nowicz and colleagues (1999) have found that, com-
culating sex steroid hormone levels are very low pared to females, males have thicker cortex, higher
(Hampson, 1990a,b; Silverman and Zimmer, 1975; Sil- neuronal densities, and more neurons in the left tem-
verman, Zimmer, and Silverman, 1974; Wickham, poral sites involved with language function. In other
1958). It is important to note that such changes in per- studies, females have been reported to have larger pro-
formance over the menstrual cycle are not seen with all portional volumes of gray matter in the dorsolateral
verbal tasks. Rather, there appears to be specificity for prefrontal cortex and superior temporal gyrus com-
the type of task that may be influenced by sex steroid pared with males (Schlaepfer et al., 1995), as well as
hormones (Hampson, 1990a,b). cortex associated with the cingulate sulcus (Paus et al.,
In most humans, the left side of the brain is primarily 1996), and proportionately larger Wernicke and Broca
responsible for the control of speech (Springer and areas (Harasty et al., 1997). At the microscopic level,
Deutch, 1998). However, there is evidence that there
may be less hemispheric specialization in women than
in men, and it is possible that this sexual dimorphism
may underlie the differences between the sexes in ver-
bal abilities. Compared to men, for example, women
have been reported to have less perceptual asymmetry
in dichotic listening tests (McGlone, 1980) and less
asymmetry in the size of the cortex, which leads to nam-
ing errors in response to electrical stimulus tests per-
formed during surgeries to map out areas of the cortex
involved in speech (Mateer, Polen, and Ojemann,
1982). More recently, using functional MRI, men were
shown to have left lateralized inferior frontal gyrus ac-
tivation in phonological tasks, whereas women were
shown to have more bilateral activation in this cortical
region when they performed the same phonological FIGURE 5.4 Sex differences in the proportional cortical vol-
tasks (Shaywitz et al., 1995). Unilateral damage to the umes in language-related areas of the brain. Data are pre-
brain has also been reported to have less devastating sented as percentages of brain hemisphere volume occupied
consequences for performance on tests of verbal abili- by each region. Asterisks indicate a significant differences (P
ties in women compared to men (Inglis et al., 1982; < 0.05) between males and females in the regional volume.
(Adapted from Harasty, J., K. L. Double, G. M. Halliday, J. J.
McGlone, 1978). Similarly, Lansdell (1961) found that Kril, and D. A. McRitchie, 1997. Language-associated cortical
surgical lesions to the left temporal cortex disrupted regions are proportionally larger in the female brain. Arch.
performance on a verbal proverb test in men, but not Neural. 54:171-176.)

70 FUNDAMENTALS OF DEVELOPMENTAL N E U R O B I O L O G Y
 greater dendritic arborizations in Wernicke's area in fe- shown that neonatal exposure to androgens can mas-
males compared to males have been reported (Jacobs, culinize some sexually dimorphic neuroanatomical dif-
Schall, and Scheibel, 1993). ferences found in the cortex, such as cortical thickness
A third sexual dimorphism that has been proposed (Lewis and Diamond, 1995), little is known about the
to underlie the sexual dimorphism in verbal abilities is effects of sex steroid hormones on the morphology of
the size of the corpus callosum. The corpus callosum is brain areas specifically involved in language processing
a nerve fiber tract that connects the two cerebral hemi- or the size or shape of the corpus callosum. However,
spheres, allowing transfer of information between the one principle is clear from studies in other parts of the
hemispheres for complex cognitive processing, includ- brain, including the hypothalamus and the hippocam-
ing language (Hines et al., 1992). In 1982, de Lacoste pus: There is profound cellular and circuit specificity
and Holloway hypothesized that the decrease in func- for the actions of sex steroid hormones. Thus, in mak-
tional neural asymmetry in women was due to increased ing conclusions about the role of steroid hormones on
interhemispheric exchange via the corpus callosum. cortical functions, it will be important to be very specific
Since that time, sexually dimorphic structural asym- with regard to both function and neuroanatomical
metries of the corpus callosum have been reported for substrate.
its area, shape, and fiber composition (Wisniewski,
1998). However, the issue of whether there is a consis- Effects of sex steroid hormones on behavior
tent sexual dimorphism of this fiber tract has been over the lifespan
highly controversial. The splenium is the posterior por-
tion of the corpus callosum that contains fibers pro- Most of the data reviewed in this chapter concern ef-
jecting between the main auditory and visual cortical fects of sex steroid hormones on adult behavior, result-
areas of the two hemispheres. In humans, a number of ing either from organizational effects of early hormonal
postmortem studies and MRI investigations provide exposure during the fetal or neonatal periods or from
support for the conclusion that females have a more activational effects of sex steroid hormones in adult-
bulbous splenium, although it is important to control hood. However, sex steroid hormone levels change dra-
for the overall corpus callosum shape, chronological matically both in children as they mature and enter
age of the individuals, cerebral volume, and handed- puberty, and in women when the activity of the repro-
ness when making such comparisons (Allen et al., 1991; ductive axis declines at menopause; thus, one would
Burke and Yeo, 1994; Johnson et al., 1996; Wisniewski, expect to see accompanying changes in behaviors mod-
1998). The isthmus of the corpus callosum connects the ulated by sex steroid hormones at these times. One
parietal and temporal lobes, and has been more consis- would also predict that sexually dimorphic areas of the
tently reported to have a larger area in women com- brain that are dependent on adult levels of sex steroid
pared to men, although for such comparisons it is again hormones for maintenance of the dimorphism would
important to control for the same factors as in studies change in morphology over puberty and menopause.
of the splenium (Denenberg, Kertesz, and Cowell, 1991; Although we know a good deal about the develop-
Steinmetz et al., 1992; Witelson, 1989). Sexual dimor- ment of sex-related behaviors at puberty (Baum, 1979),
phism in the corpus callosum of several rodent species surprisingly few studies have examined pubertal or
has also been reported (Wisniewski, 1998), and an im- menopausal changes in other behaviors that appear to
portant finding in these studies is that environmental be influenced by sex steroid hormones. In part, this
rearing conditions can have modulatory effects on the paucity may reflect the fact that much of the work ex-
sexual dimorphism of corpus callosum (Juraska and amining effects of sex steroid hormones on such be-
Kopcik, 1988). Another interesting finding of these lat- haviors as learning and memory and other cognitive
ter studies is that there are sexual dimorphisms in the functions is relatively recent. It may also reflect the dif-
ultrastructure of the rat corpus callosum, with sex dif- ficulty in determining what portion of the developmen-
ferences in the number and diameter of axons project- tal and aging changes in these behaviors results from
ing across the corpus callosum (Juraska and Kopcik, changes in sex steroid hormones and what portion is
1988; Kopcik et al., 1992). attributable to hormone-independent developmental
Despite the great interest in and numerous studies and aging processes. Finally, methodological prob-
devoted to the neural underpinnings of sexual dimor- lems—poor characterization of the activity of the re-
phisms in verbal abilities, little information is available productive axis; the use of widely varying tests of
regarding the role of sex steroid hormones in mediat- learning, memory, and other cognitive processes; and
ing the development of these functional differences. Al- extreme variation in hormone treatment regimens in
though some investigators using rodent models have postmenopausal studies—also contribute to the lack of

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 71

clarity in our understanding of how changes in sex ste- steroid hormones can act through traditional nuclear
roid hormones across the lifespan influence behavior. receptors to exert their actions, but they can also bind
Recent studies have started to document develop- to membrane receptors and activate a variety of second
mental changes within specific neuroanatomical messenger pathways to influence neuronal activity. Al-
regions of the brain. In a large imaging study of nor- ternatively, some sex steroid hormones appear to mod-
mally developing children between the ages of 4 and 18 ulate function by acting as growth factors. Additionally,
years, Giedd and colleagues (1997) found a progressive novel interactions between steroid hormones and clas-
increase in the volume of the lateral ventricles in boys, sical neurotransmitters and neuropeptides have be-
with an abrupt increase in volume starting at age 11. come apparent, such as the binding of progesterone to
Other areas that showed developmental changes in vol- GABAA receptors and the activation of progesterone re-
ume included the amygdala in boys and the hippocam- ceptors by dopamine. Third, a wide variety of neural
pus in girls (Giedd et al., 1997). Developmental changes processes are influenced by sex steroid hormones: neu-
in the density of fiber tracts within the brain have also rogenesis, cell migration, growth of the neuronal soma,
been reported for the internal capsule and arcuate fas- dendritic growth, differentiation and synapse forma-
ciculus across this same age span, (Paus et al., 1999). tion, synapse elimination, neuronal atrophy and apop-
In the past few years, a number of studies have ex- tosis, neuropeptide expression, the expression of
amined how cognitive functions change in women as neurotransmitter receptors, and neuronal excitability.
they enter menopause and how steroid hormone re- Fourth, in many cases steroid hormones must be pres-
placement therapy affects cognitive function. Although ent both developmentally and in the adult to elicit max-
some protective effects of sex steroid hormones on imal effects on behavior. Fifth, there is a great deal of
counteracting age-related changes in cognitive func- plasticity in the developing and adult nervous system,
tions have been reported, the majority of recent studies and sex steroid hormones play an important role in reg-
show little or no effect of sex steroid hormones on age- ulating plastic changes in neuronal structure and
related changes in cognitive functions (Barrett-Connor function.
and Goodman-Gruen, 1999; Haskell, Richardson, and
Horwitz, 1997; Hogervorst et al., 1999; Matthews et al.,
REFERENCES
1999). However, a detailed examination of various tests
of cognitive function and different hormone replace- ABDELGADIR, S. E., C. E. ROSELLI, J. V. A. CHOATE, and J. A.
ment therapies suggests that there may be some subtle RESKO, 1997. Distribution of aromatase cytochrome P450
messenger ribonucleic acid in adult rhesus monkey brains.
effects of specific steroid hormone regimens on specific Biol. Reprod. 57:772-777.
tests of cognition and memory (Hogervorst et al., ABE-DOHMAE, S., R. TANAKA, and N. HARADA, 1994. Cell type-
1999). In evaluating these studies, one must keep in and region-specific expression of aromatase mRNA in cul-
mind that general effects of hormone replacement tured brain cells. Mol. Brain Res. 24:153-158.
therapy on feelings of well-being may significantly influ- ALLEN, L. S., M. HINES, J. E. SHRYNE, and R. A. GORSKI, 1989.
Two sexually dimorphic cell groups in the human brain. J
ence performance on these tests, making it very diffi- Neurosci. 9:497-506.
cult to ascertain effects of hormones on specific ALLEN, L. S., M. F. RICKEY, Y. M. CHAI, and R. A. GORSKI, 1991.
cognitive functions. Sex differences in the corpus callosum in the living human
being.J Neurosci. 11:933-942.
ALONSO, R, and I. LOPEZ-COVIELLA, 1998. Gonadal steroids
and neuronal function. Neurochem. Res. 23:675-688.
General principles of sex steroid actions ARENDASH, G. W., and R. A. GORSKI, 1983. Effects of discrete
in the developing brain lesions of the sexually dimorphic nucleus of the preoptic
area or other medial preoptic regions on the sexual behav-
As we learn more about how sex steroid hormones in- ior of male rats. Brain Res. Bull. 10:147.
fluence behavior, some general principles regarding sex ARNOLD, A. P, 1996. Genetically triggered sexual differentia-
tion of brain and behavior. Horm. Behav. 30:495-505.
steroid actions in the developing brain are emerging. ARNOLD, A. P, 1997. Sexual differentiation of the zebra finch
First, it is apparent that the actions of steroid hormones song system: Positive evidence, negative evidence, null hy-
on brain systems are widespread, with steroid hormones potheses and a paradigm shift.J Neurobiol. 33:572-584.
acting at multiple sites within the neural circuitry to AYOUB, D. M., W. T. GREENOUGH, and J. M. JURASKA, 1983.
lead to changes in any one behavior. Second, we now Sex differences in dendritic structure in the preoptic area
of the juvenile macaque monkey brain. Science2.l9:l97-l98.
recognize great diversity in the cellular mechanisms by BARKER, J. L., N. L. HARRISON, G. D. LANGE, and D. G. OWEN,
which steroid hormones influence neural function. In 1987. Potentiation of g-aminobutyric-acid-activated chloride
some cases, hormones are metabolized to more active conductance by a steroid anesthetic in cultured rat spinal
forms to exert their actions on neuronal function. Sex neurons.J Physiol. 386:485-501.

72 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

BARRACLOUGH, C. A., and R. A. GORSKI, 1962. Studies on mat- BRENOWITZ, E. A., A. P. ARNOLD, and R. N. LEVIN, 1985. Neu-
ing behaviour in the androgen-sterilized female rat in re- ral correlates of female song in tropical duetting birds. Brain
lation to the hypothalamic regulation of sexual behaviour. Res. 343:104-112.
/. Endocrinol. 25:175-182. BROVERMAN, D. M., W. VOGEL, E. L. KLAIBER, D. MAJCHER, D.
BARRACLOUGH, C., and J. LEATHEM, 1954. Infertility induced SHEA, and V. PAUL, 1981. Changes in cognitive task perfor-
in mice by a single injection of testosterone propionate. mance across the menstrual cycle. J. Comp. Physiol. Psychol.
Proc. Soc. Exp. Biol. Med. 85:673-674. 95:646-654.
BARRETT-CONNOR, E., and D. GOODMAN-GRUEN, 1999. Cog- BROWN, T. J., N. J. MACLUSKY, C. LERANTH, M. SHANABROUGH,
nitive function and endogenous sex hormones in older and F. NAFTOLIN, 1990. Progestin receptor-containing cells
women.J. Am. Geriatr. Soc. 47:1289-1293. in guinea pig hypothalamus: Afferent connections, mor-
BAULIEU, E. E., 1991. Neurosteroids: A new function in the phological characteristics, and neurotransmitter content.
brain. Biol. Cell 71:3-10. Mol. Cell Neurosci. 1:58-77.
BAULIEU, E. E., 1997. Neurosteroids: Of the nervous system, BROWN, T. J., F. NAFTOLIN, and M. J. MACLUSKY, 1992. Sex
by the nervous system, for the nervous system. Rec. Prog. differences in estrogen receptor binding in the rat hypo-
Harm. Res. 52:1-32. thalamus: Effects of subsaturating pulses of estradiol. Brain
BAUM, M. J., 1979. Differentiation of coital behavior in mam- Res. 578:129-134.
mals: A comparative analysis. Neurosci. Biobehav. Rev. 3:265- BROWN-GRANT, K., A. MUNCK, F. NAFTOLIN, and M. R. SHER-
284. WOOD, 1971. The effecls of the administration of testoste-
BEACH, F. A., 1942. Copulatory behavior in prepubertally cas- rone propionate alone or with phenobarbilone and of
trated male rats and its modification by estrogen adminis- testosterone metabolites to neonatal female rats. Horm. Be-
tration. Endocrinology 31:679-683. hav. 2:173-182.
BEATTY, W. W., 1979. Gonadal hormones and sex differences BURKE, H. L., and R. A. YEO, 1994. Systematic variations in
in nonreproductive behaviors in rodents: Organizational callosal morphology: The effects of age, gender, hand pref-
and activational influences. Horm. Behav. 12:112-163. erence, and anatomic asymmelry. Neuropsychol. 8:563-571.
BECKER, J. B., 1990. Direct effect of I7p-estradiol on striatum: BUTERBAUGH, G. G., and G. M. HUDSON, 1991. Eslradiol re-
Sex differences in dopamine release. Synapse 5:157-164. placement to female rals facilitates dorsal hippocampal but
BECKER, J. B., 1991. Hormonal influences on extrapyramidal not venlral hippocampal kindled seizure acquisition. Exp.
sensorimotor function and hippocampal plasticity. In Be- Neurol. 111:55-64.
havioral Endocrinology, J. B. Becker, M. Breedlove, and D. CAMP, D. M., J. B. BECKER, and T. E. ROBINSON, 1986. Sex
Crews, eds. Cambridge, Mass.: MIT Press, pp. 325-356. differences in the effecls of gonadeclomy on amphelamine-
BECKER, J. B., 1999. Gender differences in dopaminergic func- induced rolational behavior in rats. Behav. Neuroal. Biol.
tion in striatum and nucleus accumbens. Pharm. Biochem. 46:491-495.
Behav. 64:803-812. CELOTTI, F, R. MELCANGI, and L. MARTINI, 1992. The 5 alpha-
BECKER, J. B., and J. CHA, 1989. Estrous cycle-dependent vari- reductase in the brain: Molecular aspects and relation to
ation in amphetamine-induced behaviors and striatal do- brain function. Front. Neuroendocrinol. 13:163-215.
pamine release assessed with microdialysis. Behav. Brain Res. CINTRA, A., K. FUXE, A. HARFSTRAND, L. F. AGNATI, L. S.
35:117-125. MILLER, J. L. GREENE, and J.-A. GUSTAFSSON, 1986. On the
BECKER, J. B., and V. D. RAMIREZ, 1980. Sex differences in cellular localization and distribution of eslrogen receptors
amphetamine stimulated release of catecholamines from in the rat tel- and diencephalon using monoclonal antibod-
rat striatal tissue in vitro. Brain Res. 204:361-372. ies to human estrogen receptor. Neurochem. Int. 8:587-595.
BETHEA, C. L., N. A. BROWN, and S. G. KOHAMA, 1996. Steroid COIRINI, H., A. JOHNSON, and B. S. McEwEN, 1989. Estradiol
regulation of estrogen and progestin receptor messenger modulation of oxytocin binding in the ventromedial hypo-
ribonucleic acid in monkey hypothalamus and pituitary. En- thalamic nucleus of male and female rats. Neuroendocrinology
docrinology 137:4372-4383. 50:193-198.
BETHEA, C. L., W. H. FAHRENBACH, S. A. SPRANGERS, and COOKE, B., C. D. HEGSTROM, L. S. VILLENEUVE, and S. M.
F. FREESH, 1992. Immunocytochemical localization of pro- BREEDLOVE, 1998. Sexual differentiation of the vertebrate
gestin receptors in monkey hypothalamus: Effect of estro- brain: Principles and mechanisms. Front. Neuroendocrinol.
gen and progestin. Endocrinology 130:895-905. 19:323-362.
BLAUSTEIN, J. D., J. C. KING, D. O. TOFT, and J. TURCOTTE, COMMINS, D., and P. YAHR, 1985. Autoradiographic localiza-
1988. Immunocytochemical localization of estrogen-in- tion of estrogen and androgen receptors in the sexually di-
duced progestin receptors in guinea pig brain. Brain Res. morphic area and other regions of the gerbil brain.J. Comp.
474:1-15. Neurol 231:473-489.
BOTTJER, S. W., and A. P. ARNOLD, 1997. Developmental plas- DAVIS, E. C., J. E. SHRYNE, and R. A. GORSKI, 1996. Structural
ticity in neural circuits for a learned behavior. Ann. Rev. sexual dimorphisms in the anteroventral periventricular
Neurosci. 20:459-482. nucleus of the rat hypolhalamus are sensitive to gonadal
BOYLE, M., N. MACLUSKY, F. NAFTOLIN, and L. KACZMAREK, steroids perinatally, but develop peripubertally. Neuroendo-
1987. Hormonal regulation of K + -channel messenger RNA crinology 63:142-148.
in rat myometrium during oestrus cycle and in pregnancy. DE COURTEN-MYERS, G. M., 1999. The human cerebral corlex:
Nature 330:373-375. Gender differences in slruclure and function.J. Neuropath.
BREEDLOVE, S. M., and A. P. ARNOLD, 1981. Sexually dimor- 58:217-226.
phic motor nucleus in the rat lumbar spinal cord: Response DE LACOSTE, M. C., and R. L. HOLLOWAY, 1982. Sexual di-
to adult hormone manipulation, absence in androgen-in- morphism in the human corpus callosum. Science216:1431-
sensitive rats. Brain Res. 225:297-307. 1432.

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 73

DELGADO, A. R., and G. PRIETO, 1996. Sex differences in visuo- sexually dimorphic nucleus in the preoptic area of the rat.
spatial ability: Do performance factors play such an impor- J. Comp. Neurol. 193:529-539.
tant role? Memory Cog. 24:404-510. GOUCHIE, C. T., and D. KIMURA, 1991. The relationship be-
DENENBERG, V. H., A. KERTESZ, and P. E. COWELL, 1991. A tween testosterone levels and cognitive ability patterns. Psy-
factor analysis of the human's corpus callosum. Brain Res. choneuroendocrinology 16:323-334.
548:126-132. GOULD, E., E. WESTLIND-DANIELSSON, M. FRANKFURT, and
DEVOOGD, T. J., and F. NOTTEBOHM, 1981. Gonadal hormones B. S. McEwEN, 1990. Sex differences and thyroid hormone
induce dendritic growth in the adult avian brain. Science sensitivity of hippocampal pyramidal cells. J. Neurosci.
214:202-204. 10:996-1003.
DE VRIES, G. J., 1990. Sex differences in neurotransmitter sys- GOULD, E., C. WOOLLEY, and B. S. McEwEN, 1990. Short-term
tems.J. Neuroendocrinol 2:1-12. glucocorticoid manipulations affect neuronal morphology
Di PAOLO, T., P. FALARDEAU, and M. MORISETTE, 1988. Striatal and survival in the adult dentate gyrus. Neuroscience37:367-
D-2 dopamine agonist binding sites fluctuate during the rat 375.
estrous cycle. Life Sci. 43:665-672. GOY, R. W., and J. A. RESKO, 1972. Gonadal hormones and
DOHLER, K. D., S. S. SRIVASTAVA, J. E. SHRYNE, B. JARZAB, A. behavior of normal and pseudohermaphroditic nonhuman
SIPOS, and R. A. GORSKI, 1984. Differentiation of the sexu- female primates. Rec. Prog. Harm. Res. 28:707-733.
ally dimorphic nucleus in the preoptic area of the rat brain GRINO, P. B., J. E. GRIFFIN, and J. D. WILSON, 1990. Testoste-
is inhibited by postnatal treatment with an estrogen antag- rone at high concentration interacts with the human an-
onist. Neuroendocrinology 38:297-301. drogen receptor similarly to dihydrotestosterone.
DONCAROLOS, L. L., G. L. GREENE, and J. I. MORRELL, 1989. Endocrinology 126:1165-1172.
Estrogen plus progesterone increases progestin receptor GUENNOUN, R., R. J. FIDDES, M. GOUEZOU, M. LOMBES, and
immunoreactivity in the brain of ovariectomized guinea E. E. BAULIEU, 1995. A key enzyme in the biosynthesis of
pigs. Neuroendocrinology 50:613-623. neurosteroids, 3b-hydroxysteroid dehydrogenase/D5,D4-
DONCAROLOS, L. L., E. MONROY, and J. I. MORRELL, 1991. Dis- isomerase (3P-HSD) is expressed in rat brain. Mol. Brain Res.
tribution of estrogen receptor-immunoreactive cells in the 30:287-300.
forebrain of the female guinea pig.J. Comp. Neurol. 305:591- GURNEY, M. E., and M. KONISHI, 1979. Hormone induced sex-
612. ual differentiation of brain and behavior in zebra finches.
DORNER, G., and J. STAUDT, 1969. Structural changes in the Science 208:1380 -1382.
hypothalamic ventromedial nucleus of the male rat, follow- HAMPSON, E., 1990a. Estrogen-related variations in human
ing neonatal castration and androgen treatment. Neuroen- spatial and articulatory motor skills. Psychoneuroendocrinology
docrinology 4:278-281. 15:97-111.
EINON, D., 1980. Spatial memory and response strategies in HAMPSON, E. 1990b. Variations in sex-related cognitive abili-
rats: Age, sex and rearing differences in performance. ties across the menstrual cycle. Brain Cogn. 14:26-43.
Quart.J. Exp. Psychol. 32:473-489. HAMPSON, E., and D. KIMURA, 1988. Reciprocal effects of hor-
FEDER, H. H., and B. L. MARRONE, 1977. Progesterone: Its role monal fluctuations on human motor and perceptual-spatial
in the central nervous system as a facilitator and inhibitor skills. Behav. Neurosci. 102:456-459.
of sexual behavior and gonadotropin release. Ann. N.Y. HARASTY, J., K L. DOUBLE, G. M. HALLIDAY, J. J. KRIL, and
Acad. Sci. 286:331-354. D. A. McRiTCHiE, 1997. Language-associated cortical
FRANKFURT, M., C. GOULD, C. WOOLLEY, and B. S. McEwEN, regions are proportionally larger in the female brain. Arch.
1990. Gonadal steroids modify dendritic spine density in Neurol. 54:171-176.
HASKELL, S. G., E. D. RICHARDSON, and R. I. HORWITZ, 1997.
ventromedial hypothalamic neurons: A Golgi study in the
The effect of estrogen replacement therapy on cognitive
adult rat. Neuroendocrinology 51:530-535.
function in women: A critical review of the literature.J. Clin.
GEE, K. W., 1988. Steroid modulation of the GABA/benzodi-
Epidemiol. 11:1249-1264.
azepine receptor-linked chloride ionophore. Mol. Neuwbiol.
HERBISON, A. E., and D. T. THEODOSIS, 1992. Immunocyto-
2:291-317.
chemical identification of oestrogen receptors in preoptic
GIEDD, J. N., F. X. CASTELLANOS, J. C. RAJAPAKSE, A. C. VAI-
neurones containing calcitonin gene-related peptide in the
TUZIS, and J. L. RAPOPORT, 1997. Sexual dimorphism of the
male and female rat. Neuroendocrinology 56:761-764.
developing human brain. Prog. Neuropsychopharmacol. Biol. HINES, M., 1982. Prenatal gonadal hormones and sex differ-
Psychiat. 21:1185-1201. ences in human behavior. Psychol. Bull. 92:56-80.
GOLDMAN, S., and F. NOTTEBOHM, 1983. Neuronal produc- HINES, M., 1991. Gonadal hormones and human cognitive de-
tion, migration, and differentiation in a vocal control nu- velopment. In Hormones, Brain and Behavior in Vertebrates, J.
cleus of the adult female canary brain. Proc. Natl. Acad. Sci. Balthazart, ed. Basel: Karger.
(USA) 80:2390. HINES, M., F. C. DAVIS, A. COQUELIN, R. W. GOY, and R. A.
GOMEZ, D. M., and S. W. NEWMAN, 1991. Medial nucleus of GORSKI, 1985. Sexually dimorphic regions in the medial
the amygdala in the adult Syrian hamster. A quantitative preoptic area and the bed nucleus of the stria terminalis of
Golgi analysis of gonadal hormonal regulation of neuronal the guinea pig brain: A description and an investigation of
morphology. Anat. Rec. 231:498-509. their relationship to gonadal steroids in adulthood. J. Neu-
GORSKI, R. A., J. H. GORDON, J. E. SHRYNE, and A. M. Sou- rosci. 5:40-47.
THAM, 1978. Evidence for a morphological sex difference HINES, M., L. A. McADAMS, L. CHIU, and J. LIPCAMON, 1992.
within the medial preoptic area of the rat brain. Brain Res. Cognition and the corpus callosum: Verbal fluency, visuo-
148:333-346. spatial ability, and language lateralization related to midsa-
GORSKI, R. A., R. E. HARLAN, C. D. JACOBSON, J. E. SHRYNE, gittal surface areas of callosal subregions. Behav. Neurosci.
and A. M. SOUTHAM, 1980. Evidence for the existence of a 106:3-14.

74 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

HOGERVORST, E., M. BOSHUISEN, W. RlEDEL, C. WlLLEKEN, and phometry of Heschl's gyrus and the planum temporale.
J. JOLLES, 1999. The effect of hormone replacement therapy Cereb. Cortex 4:107-118.
on cognitive function in elderly women. Psychoneuroendocri- LANSDELL, H., 1961. The effect of neurosurgery on a test of
nology 24:43-68. proverbs. Am. Psychol 16:488.
HRUSKA, R. E., L. M. LUDMER, K. T. PITMAN, M. DE RYCK, and LEEDY, M. G., M. S. BEATTIE, and J. C. BRESNAHAN, 1987. Tes-
E. K. SILBERGELD, 1982. Effects of estrogen on striatal do- tosterone-induced plasticity of synaptic inputs to adult mam-
pamine receptor function in male and female rats. Phar- malian motoneurons. Brain Res. 424:386-390.
macol. Biochem. Behav. 16:285-291. LEPHART, E. D., 1996. A review of brain aromatase cytochrome
HYDE, J. S., and M. C. LINN, 1988. Gender differences in verbal P450. Brain Res. Rev. 22:1-26.
ability: A meta-analysis. Psychol. Bull. 104:53-69. LEPHART, E. D., E. R. SIMPSON, M. J. MCPHAUL, M. W. KIL-
INGLIS, J., M. RUCKMAN, J. S. LAWSON, A. W. MACLEAN, and GORE, J. D. WILSON, and S. R. OJEDA, 1992. Brain aromatase
T. N. MONGA, 1982. Sex differences in the cognitive effects cytochrome P-450 messenger RNA levels and enzyme activ-
of unilateral brain damage. Cortex 18:257-276. ity during prenatal and perinatal development in the rat.
JACOBS, B., M. SCHALL, and A. B. SCHEIBEL, 1993. A quanti- Mol. Brain Res. 16:187-192.
tative dendritic analysis of Wernicke's area in humans. II. LEVAY, S., 1991. A difference in hypothalamic structure
Gender, hemispheric and environmental factors. J. Comp. between heterosexual and homosexual men. Science 253:
Neurol 327:97-111. 1034-1037.
JACOBSON, C. D., A. P. ARNOLD, and R. A. GORSKI, 1987. Ste- LEVESQUE, D., S. GAGNON, and T. DI PAOLO, 1989. Striatal Dl
roid autoradiography of the sexually dimorphic nucleus of dopamine receptor density fluctuates during the rat estrous
the preoptic area. Brain Res. 414:349-356. cycle. Neurosci. Lett. 98:345-350.
TACOBSON, C. D., and R. A. GORSKI, 1981. Neurogenesis of the LEVINE, S., and R. MULLINS, JR., 1964. Estrogen administered
sexually dimorphic nucleus of the preoptic area in the rat. neonatally affects adult sexual behavior in male and female
J. Comp. Neurol. 196:519-529. rats. Science 144:185-187.
JAKAB, R. L., T. L. HORVATH, C. LERANTH, N. HARADA, and LEWIS, C., B. S. McEwEN, and M. FRANKFURT, 1995. Estrogen-
F. NAFTOLIN, 1993. Aromatase immunoreactivity in the rat induction of dendritic spines in ventromedial hypothala-
brain: Gonadectomy-sensitive hypothalamic neurons and an mus and hippocampus: Effects of neonatal aromatase
unresponsive 'limbic ring' of the lateral septum-bed blockade and adult castration. Dev. Brain Res. 87:91-95.
nucleus-amygdala complex. J. Steroid Biochem. Mol. Biol. LEWIS, D. W., and M. C. DIAMOND, 1995. The influence of
44:481-498. gonadal steroids on the asymmetry of the cerebral cortex.
JOELS, M., and H. KARST, 1995. Effects of estradiol and pro- In Brain Asymmetry, R. J. Davidson and K. Hugdahl, eds.,
gesterone on voltage-gated calcium and potassium conduc- Cambridge, Mass.: MIT Press, pp. 31-49.
tances in rat CA1 hippocampal neurons. J. Neurosci. LINN, M. C., and A. C. PETERSEN, 1985. Emergence and char-
15:4289-4297. acterization of sex differences in spatial ability: A meta-anal-
JOHNSON, S. C., J. B. PINKSTON, E. D. BIGLER, and D. D. BLAT- ysis. Child Devel 56:1479-1498.
TER, 1996. Corpus callosum morphology in normal controls LOY, R., J. L. GERLACH, and B. S. McEwEN, 1988. Autoradio-
and traumatic brain injury: Sex differences, mechanisms of graphic localization of estradiol-binding neurons in the rat
injury, and neuropsychological correlates. Neuropsychol. hippocampal formation and entorhinal cortex. Dev. Brain
10:408-415. Res. 39:245-251.
JURASKA, J. M., 1991. Sex differences in "cognitive" regions of MACLUSKY, N., A. S. CLARK, F. NAFTOLIN, and P. S. GOLDMAN-
the rat brain. Psychoneuroendocrinology 16:105-119. RAKIC, 1987. Oestrogen formation in mammalian brain:
JURASKA, J. M., and J. R. KOPCIK, 1988. Sex and environmental Possible role of aromatase in sexual differentiation of the
influences on the size and ultrastructure of the rat corpus hippocampus and neocortex. Steroids 50:459-474.
callosum. Brain Res. 450:1-8. MACLUSKY, N. J., and F. NAFTOLIN, 1981. Sexual differentia-
KIMURA, D., 1987. Are men's and women's brains really dif- tion of the central nervous system.Science211:1294-1303.
ferent? Can. Psychol. 28:133-147. MADEIRA, M. D., and A. R. LIEBERMAN, 1995. Sexual dimor-
KOMNENICH, P., D. M. LANE, R. P. DICKEY, and S. C. STONE, phisms in the mammalian limbic system. Prog. Neurobiol.
1978. Gonadal hormones and cognitive performance. Phys- 45:275-333.
iol. Psychol. 6:115-120. MAJEWSKA, M. D., 1992. Neurosteroids: Endogenous bimodal
KOPCIK, J. R., P. SYMOURE, S. K. SCHNEIDER, J. KIM-HONG, and modulators of the GABAA receptor. Mechanism of action
J. M. JURASKA, 1992. Do callosal projection neurons reflect and physiological significance. Prog. Neurobiol. 38:379-395.
sex differences in axon number? Brain Res. Bull. 29:493- MALSBURY, C. W., and K. McKAY, 1994. Neurotrophic effects
497. of testosterone on the medial nucleus of the amygdala in
KUIPER, G. G. J. M., E. ENMARK, M. PELTO-HUIKKO, S. NILS- adult male rats.J. Neuroendocrinol. 6:57-69.
SON, and J.-A. GUSTAFSSON, 1996. Cloning of a novel estro- MANI, S. K, J. M. C. ALLEN, J. H. CLARK, J. D. BAUSTEIN, and
gen receptor expressed in rat prostate. Proc. Natl. Acad. Sci. B. W. O'MALLEY, 1994. Convergent pathways for steroid-,
(USA) 93:5925-5930. hormone- and neurotransmitter-induced rat sexual behav-
KUIPER, G. G. J. M., P. J. SHUGHRUE, I. MERCHENTHALER, and ior. Science 265:1246-1249.
J.-A. GUSTAFSSON, 1998. The estrogen receptor (3 subtype: MANI, S. K., J. D. BLAUSTEIN, and B. W. O'MALLEY, 1997. Pro-
A novel mediator of estrogen action in neuroendocrine sys- gesterone receptor function from a behavioral perspective.
tems. Front. Neuroendocrinol. 19:253-286. Harm. Behav. 31:244-255.
KULYNYCH, J. J., K. VLADAR, D. W.JONES, and D. R. WEINBER- MATEER, C. A., S. B. POLEN, and G. A. OJEMANN, 1982. Sexual
GER, 1994. Gender differences in the normal lateralization variation in cortical localization of naming as determined
of the supratemporal cortex: MRI surface-rendering mor- by stimulation mapping. Behav. Brain Sci. 5:310-311.

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT 75

MATTHEWS, K., J. CAULEY, K. YAFFE, and J. M. ZMUDA, 1999. reductase and aromatase in brain differentiation and func-
Estrogen replacement therapy and cognitive decline in tions.J. Steroid Biochem. Molec. Biol. 58:455-466.
older community women.J. Am. Geriatr. Soc. 47:518-523. NORMINGTON, K., and D. W. RUSSELL, 1992. Tissue distribu-
MCDONALD, P. G., C. BEYER, F. NEWTON, B. BRIEN, R. BAKER, tion and kinetic characteristics of rat steroid 5a-reductase
H. S. TAN, C. SAMPSON, P. KITCHING, R. GREENHILL, and D. isozymes.J. Biol. Chem. 267:19548-19554.
PRITCHARD, 1970. Failure of 5a-dihydrotestosterone to ini- NOTTEBOHM, F., and A. ARNOLD, 1976. Sexual dimorphism in
tiate sexual behavior in the castrated male rat. Nature vocal control areas of the songbird brain. Science 194:211-
227:964-965. 213.
McEwEN, B. S., 1991. Steroids affect neural activity by acting O'KEEFE, J., 1993. Hippocampus, theta, and spatial memory.
on the membrane and the genome. Trends Pharmacol. Curr. Opin. Neurobiol. 3:917-924.
12:141-147. O'KEEFE, J. A., and R. J. HANDA, 1990. Transient elevation of
McEwEN, B. S., 1996. Gonadal and adrenal steroids regulate estrogen receptors in the neonatal rat hippocampus. Dev.
neurochemical and structural plasticity of the hippocampus Brain Res. 57:119-127.
via cellular mechanisms involving NMDA receptors. Cell. PAUS, T., N. OTAKY, Z. CARAMANOS, 1996. In vivo morphom-
Mol. Neurobiol. 16:103-116. etry of the intrasulcal gray matter in the human cingulate,
McEwEN, B. S., and S. E. ALVES, 1999. Estrogen actions in the paracingulate, and superior-rostral sulci: Hemispheric asym-
central nervous system. Endocrinol. Rev. 20:279-307. metries, gender differences and probability maps. J. Comp.
McGLONE, J., 1978. Sex differences in functional brain asym- Neurol 376:664-673.
metry. Cortex 14:122-128. PAUS, T., A. ZIJDENBOS, K. WORSLEY, D. L. COLLINS, J. BLU-
McGLONE, J., 1980. Sex differences in human brain asymme- MENTHAL, J. N. GIEDD, J. L. RAPOPORT, and A. C. EVANS,
try: A critical survey. Behav. Brain Sci. 3:215-263. 1999. Structural maturation of neural pathways in children
MELLON, S. H., 1994. Neurosteroids: Biochemistry, modes of and adolescents: In vivo study. Science 283:1908-1911.
action, and clinical relevance. J. Clin. Endocrinol. Metab. PFAFF, D., 1980. Estrogens and Brain Function. New York:
78:1003-1008. Springer-Verlag.
MERMELSTEIN, P. G., J. B. BECKER, and D. J. SURMEIER, 1996. PFAFF, D. W., and M. KEINER, 1973. Atlas of estradiol-concen-
Estradiol reduces calcium currents in rat neostriatal neu- trating cells in the central nervous system of the female rat.
rons via a membrane receptor.J. Neurosci. 16:595-604. J Comp. Neurol. 151:121-158.
MICHAEL, R. P, R. W. BONSALL, and D. ZUMPE, 1987. Testos- PHOENIX, C. H., and K. C. CHAMBERS, 1982. Sexual behavior
terone and its metabolites in male cynomolgus monkeys in adult gonadectomized female pseudohermaphrodite, fe-
(Macaca fascicularis): Behavior and biochemistry. Physiol. Be- male, and male rhesus macaques (Macaca mulatto) treated
hav. 40:527-537. with estradiol benzoate and testosterone propionate. J.
MICHAEL, R. P., A. N. CLANCY, and D. ZUMPE, 1995. Distribu- Comp. Physiol Psychol. 96:823-833.
tion of androgen receptor-like immunoreactivity in the PHOENIX, C. H., R. W. GOY, A. A. GERALL, and W. C. YOUNG,
brains of cynomolgus monkeys. J. Neuroendocrinol. 7:713- 1959. Organizing action of prenatally administered testos-
719. terone propionate on the tissues mediating mating behavior
MICHAEL, R. P., D. ZUMPE, and R. W. BONSALL, 1990. Estradiol in the female guinea pig. Endocrinology 65:369-382.
administration and the sexual activity of castrated male rhe- PHOENIX, C. H., R. W. GOY, and J. A. RESKO, 1968. Psycho-
sus monkeys (Macaca mulatto). Harm. Behav. 24:71-88. sexual differentiation as a function of androgenic stimula-
MILNER, B., H. L. TEUBER, and S. CORKIN, 1968. Further anal- tion. In Perspectives in Reproduction and Sexual Behavior, M.
ysis of the hippocampal amnesic syndrome: 14 year follow Diamond, ed. Bloomington, Ind.: Indiana University Press,
up study of H. M. Neuropsychologia 6:215-234. pp. 33-49.
MINAMI, T., Y. OOMURA, J. NABEKURA, and A. FUKUDA, 1990. RABINOWICZ, T, D. E. DEAN, J. M.-C. PETETOT, and G. M. DE
l7b-Estradiol depolarization of hypothalamic neurons is COURTEN-MYERS, 1999. Gender differences in the human
mediated by cyclic AMP. Brain Res. 519:301-307. cerebral cortex: More neurons in males; more processes in
NABEKURA, J., Y. OOMURA, T. MINAMI, Y. MIZUNO, and A. FU- females.J. Child Neurol. 14:98-107.
KUDA, 1986. Mechanism of the rapid effect of I7b-estradiol RAINBOW, T., B. PARSONS, and B. S. McEwEN, 1982. Sex dif-
on medial amygdala neurons. Science 233:226-228. ferences in rat brain oestrogen and progestin receptors. Na-
NAFTOLIN, E, T. L. HORVATH, R. L. JAKAB, C. LERANTH, ture 300:648-649.
N. HARADA, and J. BALTHAZART, 1996. Aromatase immu- RAISMAN, G., and P. M. FIELD, 1973. Sexual dimorphism in the
noreactivity in axon terminals of the vertebrate brain. Neu- neuropil of the preoptic area of the rat and its dependence
roendocrinology 63:149-155. on neonatal androgen. Brain Res. 54:1-29.
NAFTOLIN, F, K. J. RYAN, and Z. PETRO, 1971. Aromatization RAZANDI, M., A. PEDRAM, G. L. GREENE, and E. R. LEVIN, 1999.
of androstenedione by the diencephalon.J. Clin. Endocrinol. Cell membrane and nuclear estrogen receptors (ERs) origi-
Metab. 33:368-370. nate from a single transcript: Studies of ERa and ERp ex-
NAFTOLIN, E, K. J. RYAN, and Z. PETRO, 1977. Aromatization pressed in Chinese hamster ovary cells. Mol. Endocrinol.
of androstenedione by the anterior hypothalamus of adult 13:307-319.
male and female rats. Endocrinology 90:295-298. RESKO, J. A., P. B. CONNOLLY, C. E. ROSELLI, S. E. ABDELGADIR,
NEBERT, D. W., and F. J. GONZALES, 1987. P450 genes: Struc- and J. V. A. CHOATE, 1993. Selective activation of androgen
ture, evolution, and regulation. Ann. Rev. Biochem. 56:945- receptors in the subcortical brain of male cynomolgus ma-
953. caques by physiological hormone levels and its relationship
NEGRI-CESI, P., A. POLETTI, and F. CELOTTI, 1996. Metabolism to androgen-dependent aromatase activity. J. Clin. Endocri-
of steroids in the brain: A new insight into the role of 5a- nol. Metab. 76:1588-1593.

76 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

RESKO, J. A., and C. E. ROSELLI, 1997. Prenatal hormones or- ence in the isthmus of the corpus callosum. Neurology
ganize sex differences of the neuroendocrine reproductive 42:749-752.
system: Observations on guinea pigs and nonhuman pri- STUMPF, W. E., M. SAR, and D. A. KEEFER, 1975. Atlas of estro-
mates. Cell. Mol. Neurobiol. 17:627-648. gen target cells in the rat brain. In Anatomical Neuroendocri-
RESNICK, S. M., S. A. BERENBAUM, I. I. GOTTESMAN, and T. J. nology, W. E. Stumpf and L. D. Grant, eds. Basel: Karger, pp.
BOUCHARD, 1986. Early hormonal influences on cognitive 104-119.
functioning in congenital adrenal hyperplasia. Dev. Psychol. SWAAB, D. F, and E. FLIERS, 1985. A sexually dimorphic nu-
22:191-198. cleus in the human brain. Science 228:1112-1115.
ROSELLI, C. E., and J. A. RESKO, 1987. The distribution and TERESAWA, S., and P. S. TIMIRAS, 1968. Electrical activity dur-
regulation of aromatase activity in the central nervous sys- ing the estrous cycle of the rat: Cyclic changes in limbic
tem. Steroids 50:495-508. structures. Endocrinology 83:207-216.
ROSELLI, C. E., and J. A. RESKO, 1989. Testosterone regulates TEYLER, T. J., R. M. VARDARIS, D. LEWIS, and A. B. RAWITCH,
aromatase activity in discrete brain areas of male rhesus ma- 1980. Gonadal steroids: Effects on excitability of hippocam-
caques. Biol Reprod. 40:929-934. pal pyramidal cells. Science 209:1017-1019.
ROSELLI, C. E., and J. A. RESKO, 1993. Aromatase activity in THERRIEN, B., 1982. Sex Differences in the Effects of Hippocampal
the rat brain: Hormonal regulation and sex differences. /. Lesions on Place Navigation. Doctoral Dissertation. Ann Ar-
Steroid Biochem. Mol. Biol. 44:499-508. bor: Mich.: University of Michigan.
ROSENFELD, J. M., J. D. DALEY, S. OHON, and E. V. YOUNG LAI, TOBET, S. A., D. J. ZAHNISER, and M. J. BAUM, 1986. Sexual
1977. Central aromatization of testosterone in testicular dimorphism in the preoptic/anterior hypothalamic area of
feminized mice. Experientia 33:1392-1393. ferrets: Effects of adult exposure to sex steroids. Brain Res.
SAPOLSKY, R., 1990. Glucocorticoids, hippocampal damage 364:249-257.
and the glutamatergic synapse. Prog. Brain Res. 86:13-23. TORAN-ALLERAND, C. D., M. SINGH, and G. SETALO, 1999.
SCHLAEPFER, T. E., G. J. HARRIS, A. Y. TlEN, L. PENG, S. LEE,
Novel mechanisms of estrogen action in the brain: New
and G. D. PEARLSON, 1995. Structural differences in the cer- players in an old story. Front. Neuroendocrinol. 20:97-121.
TURNER, B. B., and D. A. WEAVER, 1985. Sexual dimorphism
ebral cortex of healthy female and male subjects: A mag-
of glucocorticoid binding in rat brain. Brain Res. 343:16-23.
netic resonance imaging study. Psychiatry Res. 61:129-135.
TRUSS, M., and M. BEATO, 1993. Steroid hormone receptors:
SENGELAUB, D. R., and A. P. ARNOLD, 1986. Development and
Interaction with deoxyribonucleic acid and transcription
loss of early projections in a sexually dimorphic rat spinal
factors. Endocrinol. Rev. 14:459-479.
nucleus.J. Neurosci. 6:1613-1620.
WADA, J. A., R. CLARKE, and A. HAMM, 1975. Cerebral hemi-
SHAYWITZ, B., S. E. SHAYWITZ, K. R. PUGH, R. T. CONSTABLE,
spheric asymmetry in humans. Arch. Neurol. 32:239-246.
P. L. SKUDLARSKI, R. K. FULBRIGHT, R. A. BRONEN, J. M.
WARREN, S. G., A. G. HUMPHREYS, J. M. JURASKA, and W. T.
FLETCHER, D. P. SHANKWEILER, and L. KATZ, 1995. Sex dif-
GREENOUGH, 1995. LTP varies across the estrous cycle: En-
ferences in the functional organization of the brain for lan- hanced synaptic plasticity in proestrus rats. Brain Res.
guage. Nature 373:607-609. 703:26-30.
SHOLL, S. A., R. W. GOY, and K. L. KIM, 1989. 5a-Reductase, WEISZ, J., and I. L. WARD, 1980. Plasma testosterone and pro-
aromatase, and androgen receptor levels in the monkey gesterone levels of pregnant rats, their male and female fe-
brain during fetal development. Endocrinology 124:627-634. tuses, and neonatal offspring. Endocrinology 106:306-316.
SHUGHRUE, P. J., M. V. LANE, and I. MERCHENTHALER, 1997. WHALEN, R. E., and R. D. NADLER, 1963. Suppression of the
The comparative distribution of estrogen receptor-a and b development of female mating behavior by estrogen admin-
mRNA in the rat central nervous system. J Comp. Neurol. istered in infancy. Science 141:273-275.
388:507-525. WICKHAM, M., 1958. The effects of the menstrual cycle on test
SHUGHRUE, P. J., and I. MERCHENTHALER, 2000. Estrogen is performance. Br.J. Psychol. 49:34-41.
more than just a "sex hormone": Novel sites for estrogen WILLIAMS, C. L., A. M. BARNETT, and W. H. MECK, 1990. Or-
action in the hippocampus and cerebral cortex. Front. Neu- ganizational effects of early gonadal secretions on sexual dif-
roendocrinol. 21:95-101. ferentiation in spatial memory. Behav. Neurosci. 104:84—97.
SHUTE, V. J., J. W. PELLEGRINO, L. HUBERT, and R. W. REY- WISNIEWSKI, A. B., 1998. Sexually-dimorphic patterns of cor-
NOLDS, 1983. The relationship between androgen levels and tical asymmetry, and the role for sex steroid hormones in
human spatial abilities. Bull. Psychonom. Soc. 21:465-468. determining cortical patterns of lateralization. Psychoneu-
SILVERMAN, E. M., and C. H. ZIMMER, 1975. Speech fluency roendocrinology 23:519-547.
fluctuations during the menstrual cycle. J Speech Hear. Res. WITELSON, S. F, 1989. Hand and sex differences in the isthmus
18:202-206. and genu of the human corpus callosum. Brain 112:799-
SILVERMAN, E. M., C. H. ZIMMER, and F. H. SILVERMAN, 1974. 835.
Variability of stutterers' speech disfluency during the men- WITELSON, S. F, I. I. GLEZER, and D. L. KIGAR, 1995. Women
strual cycle. Percept. Mot. Skills 38:1037-1038. have greater density of neurons in posterior temporal cor-
SIMERLY, R. B., C. CHANG, M. MURAMATSU, and L. W. SWAN- tex.J Neurosci. 15:3418-3428.
SON, 1990. Distribution of androgen and estrogen receptor WITELSON, S. E, and D. L. KIGAR, 1992. Sylvian fissure mor-
mRNA-containing cells in the rat brain: An in situ hybrid- phology and asymmetry in men and women: Bilateral dif-
ization study.J Comp. Neurol. 294:76-95. ferences in relation to handedness in men.J. Comp. Neurol.
SPRINGER, S. P., and G. DEUTCH, 1998. Left Brain, Right Brain, 323:326-340.
5th ed. New York: W. H. Freeman. WITKIN, H. A., and J. W. BERRY, 1975. Psychological differen-
STEINMETZ, H., L. JANCKE, A. KLEINSCHMIDT, G. SCHLAUG,J. tiation in cross-cultural perspective. J Cross-Cult. Psychol.
VOLKMANN, and Y. HUANG, 1992. Sex but no hand differ- 6:4-87.

CAMERON: EFFECTS OF SEX HORMONES ON BRAIN DEVELOPMENT //

WONG, M., and R. L. Moss, 1992. Long-term and short-term WOOLLEY, C., and B. S. McEwEN, 1994. Estradiol regulates
electrophysiological effects of estrogen on the synaptic hippocampal dendritic spine density via an NMDA receptor
properties of hippocampal CA1 neurons. J. Neurosci. 12: dependent mechanism.J. Neurosci. 19:379.
3217-3225. WOOLLEY, C. S., E. GOULD, M. FRANKFURT, and B. S. McEwEN,
WOOLLEY, C., E. GOULD, and B. S. MCEWEN, 1990. Exposure 1990. Naturally occurring fluctuation in dendritic spine
to excess glucocorticoids alters dendritic morphology of density on adult hippocampal pyramidal neurons. J Neu-
adult hippocampal pyramidal neurons. Brain Res. 531:225- rosci. 10:4035-4039.
231. XIAO, L., and J. B. BECKER, 1994. Quantitative microdialysis
WOOLLEY, C., E. GOULD, R. SAKAI, R. L. SPENCER, and B. S. determination of extracellular striatal dopamine concentra-
McEwEN, 1991. Effects of aldosterone or RU28362 treat- tions in male and female rats: Effects of estrous cycle and
ment on adrenalectomy-induced cell death in the dentate gonadectomy. Neurosci. Lett. 180:155-158.
gyrus of the adult rat. Brain Res. 554:312-315. XIAO, L., and J. B. BECKER, 1997. Hormonal activation of the
WOOLLEY, C. S., and B. S. McEwEN, 1992. Estradiol mediates striatum and the nucleus accumbens modulates paced mat-
fluctuation of hippocampal synapse density during the es- ing behavior in the female rat. Horm. Behav. 32:114-124.
trous cycle in the adrenalectomized rat.J. Neurosci. 12:2549- ZUMPE, D., R. W. BONSALL, and R. P. MICHAEL, 1993. Effects
2554. of the nonsteroidal aromatase inhibitor, Fadrozole, on the
WOOLLEY, C. S., and B. S. McEwEN, 1993. Roles of estradiol sexual behavior of male cynomolgus monkeys (Macacafas-
and progesterone in regulation of hippocampal dendritic cicularis). Horm. Behav. 27:200-215.
spine density during the estrous cycle in the rat. J. Comp.
Neural. 336:293-303.

78 FUNDAMENTALS OF DEVELOPMENTAL NEUROBIOLOGY

6 The Development of Prefrontal
Cortex: The Maturation
of Neurotransmitter Systems
and Their Interactions
FRANCINE M. BENES

ABSTRACT This chapter provides an overview of the devel- other components of this system with which it is inti-
opment of the prefrontal cortex that begins with the differ- mately linked. The discussion that follows provides an
entiation of its intrinsic projection and local circuit neurons overview of corticolimbic development at various stages
during embryogenesis, but extends well beyond. Despite the
early appearance of laminar patterns that reflect those seen of the life cycle, beginning with gyral development dur-
in the adult brain, the architecture of synaptic connections ing embryogenesis and extending to the postnatal pe-
and the maturation of intrinsic and extrinsic neurotransmitter riod. Consideration is given to the maturation of both
systems continue to change in the prefrontal cortex during intrinsic and extrinsic neurons and their respective neu-
the postnatal period. It is particularly noteworthy that a variety rotransmitter systems, and the chapter concludes with
of studies have demonstrated that the monoaminergic neu-
rons, especially those employing dopamine as a neuromodu- a discussion of how the latter interact with one another
lator, continue to infiltrate the cortical mantle until the early during postnatal development.
adult period. More recent evidence, however, even suggests
that dopaminergic and serotonergic fibers form appositions
with the same projection and local circuit neurons of the pre-
Gyral development
frontal cortex. Because of this convergence, a disturbance in
the development of serotonergic projections in this region is The development of the cerebral cortex occurs at dif-
associated with an increase of its dopaminergic innervation. ferent rates that, to some extent, reflect phylogeny. For
Taken together, these findings suggest that these two mono- example, at 16-19 weeks of embryogenesis, the cingu-
aminergic systems may compete with each other for the estab- late gyrus can be distinguished, but this occurs long be-
lishment of functional territory on intrinsic neurons of the
cortex, although it is not clear whether this interaction is me- fore the parahippocampal gyrus and hippocampal
diated within the cortex itself or possibly at midbrain levels formation have become discernible in the medial tem-
where the respective nuclei are located. Overall, the results poral lobe (Gilles, Shankle, and Dooling, 1983). Corti-
described in this chapter suggest that the ontogenesis of the cal regions undergo ontogenesis at varying rates that,
prefrontal cortex is a protracted process in which a marked to some degree, reflect phylogeny (table 6.1). For ex-
degree of plasticity, even in the early stages of adulthood, con-
tributes to the establishment of mature patterns of connectivity. ample, limbic cortical areas tend to differentiate early
in gestation. At 16-19 weeks in humans, the cingulate
In recent years, it has become apparent that the de- region can be distinguished as a gyrus, but this is well
velopment of the corticolimbic system continues well before the parahippocampal gyrus and the hippocam-
beyond the neonatal period, perhaps even as late as the pal formation have begun to take shape in the medial
fifth and sixth decades (Benes, 1995, 1997b; Benes et temporal area at weeks 20-23 (Gilles, Shankle, and
al., 1994). Of particular relevance to our understanding Dooling, 1983). The superior and medial frontal gyri
of the maturation of cognitive and emotional behaviors (prefrontal regions) do not take on a clear gyral con-
is the development of the prefrontal cortex and the figuration until 24-27 weeks of gestational age, and the
angular and supramarginal gyri (inferior parietal area)
are not distinguished until 28-31 weeks (Gilles, Shan-
FRANCINE M. BENES Laboratory for Structural Neuroscience,
McLean Hospital, Belmont, Massachusetts; Program in Neu- kle, and Dooling, 1983). The orbital frontal gyri, which
roscience and Department of Psychiatry, Harvard Medical (like the cingulate gyri) are well represented in earlier
School, Boston, Massachusetts. mammalian forms such as rodents, are among the last

BENES: THE DEVELOPMENT OF THE PREFRONTAL CORTEX 79

 TABLE 6.1 VI, these cells already have elaborate dendritic arbori-
Comparison of the embryonic development of human cortical gyri zations. During this same interval, incoming fibers from
other regions are present in virtually all layers of the
Gestational cortex. In the months immediately prior to birth, inter-
Age (weeks) Gyrus
neurons begin to appear in the deeper portions of layer
16-19 Gyrus rectus, cingulate III, while in layers II and outer portions of III they are
20-23 Parahippocampal, superior temporal first beginning to form. By birth, both the second and
24-27 Pre- and post-rolandic, middle temporal,
superior and middle frontal, occipital, third laminae contain pyramidal neurons; however, bas-
cuneus, lingual ket cells in layer II are largely absent (Marin-Padilla,
28-31 Inferior temporal, medial and lateral 1970b). By 2.5 months postnatally, pyramidal neurons
orbital, callosomarginal, angular, continue to mature, showing a dramatic increase in
supramarginal, transverse temporal overall size, the amount of dendritic branching, and
32-35 Paracentral
36-39 Anterior and posterior orbital numbers of dendritic spines (Michel and Garey, 1984).
Interestingly, layers VI and I (the marginal zone) are
Adapted from Gilles, Shankle, and Dooling (1983) the first laminae to appear and differentiate, while lay-
ers II and III are the last to form and the latest to mature
to appear. It seems likely that perturbations of brain (Marin-Padilla, 1970a). At birth, cortical layer II in
development at various stages of the process might be motor cortex of human brains is still quite immature
capable of inducing marked alterations in the forma- (Marin-Padilla, 1970b).
tion of normal gyral patterns of human brain. Bayer and Altaian (1991) have reported that a similar
inside-out progression is observed in the medial pre-
Cytoarchitectural maturation of the cortex frontal cortex of rodent brain (figure 6.1). Cell migra-
tion in this region occurs earlier than it does for the
The ontogenesis of all cortical regions follows a care- retrosplenial cortex and subicular subdivision of the
fully timed sequence of events (Poliakov, 1965; Sidman hippocampal formation (Bayer, 1990). This pattern re-
and Rakic, 1973). All cortical neurons are generated flects that described for gyral development, in which
along the ventricular surface in the so-called marginal the cingulate gyrus differentiates weeks in advance of
zone and migrate toward the surface after undergoing the parahippocampal gyrus and hippocampal forma-
several mitotic divisions. Poliakov (1965) described five tion; in this instance, however, it follows a dorsal-
stages of cortical development in human brain, begin- to-ventral course (see table 6.1). Presumably, the
ning with Stage I at approximately the seventh fetal analogous timing of events for the prefrontal cortex of
week when postmitotic cells begin to move upward. human brain is somewhat different, but its growth and
Stage V, the longest period, occurs between the six- maturation as a gyrus is likely complete by the second
teenth fetal week and the early postnatal period. During year of life. Postnatally, the medial prefrontal cortex of
Stage V, postmitotic neuronal cells continue migrating rat shows a progressive expansion of its matrix as the
and reach their final destination within the cortical neuropil surrounding neuronal cell bodies increases.
plate. As neurons enter the cortical plate, those des- As shown in figure 6.2, the increase of neuropil occurs
tined for more superficial layers arrive later than those between P10 and P20, an interval that occurs within the
that occupy deeper layers and show an "inside-out" pro- preweanling period that is roughly equivalent to pre-
gression (Rakic, 1974). Studies of motor cortex in hu- adolescence in humans. The increase of neuropil is
mans have shown that by 7 months of gestation, layers probably related to two different events: (1) the in-
V and VI have attained a more advanced degree of de- growth of fibers from other cortical and subcortical
velopment than layers II and III (Marin-Padilla, 1970a; regions, and (2) the collateralization of axons from
Zilles, Busching, and Schleicher, 1986). The mor