New Aspects of Axonal Structure and Function

New Aspects of Axonal Structure and Function

Dirk Feldmeyer · Joachim H.R. Lübke Editors

New Aspects of Axonal Structure and Function

123

Editors Dirk Feldmeyer Forschungszentrum Jülich Institute of Neuroscience and Medicine 52425 Jülich Germany [email protected]

Joachim H.R. Lübke Forschungszentrum Jülich Institute of Neuroscience and Medicine 52425 Jülich Germany [email protected]

ISBN 978-1-4419-1675-4 e-ISBN 978-1-4419-1676-1 DOI 10.1007/978-1-4419-1676-1 Springer New York Dordrecht Heidelberg London Library of Congress Control Number: 2010931723 © Springer Science+Business Media, LLC 2010 All rights reserved. This work may not be translated or copied in whole or in part without the written permission of the publisher (Springer Science+Business Media, LLC, 233 Spring Street, New York, NY 10013, USA), except for brief excerpts in connection with reviews or scholarly analysis. Use in connection with any form of information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed is forbidden. The use in this publication of trade names, trademarks, service marks, and similar terms, even if they are not identified as such, is not to be taken as an expression of opinion as to whether or not they are subject to proprietary rights. While the advice and information in this book are believed to be true and accurate at the date of going to press, neither the authors nor the editors nor the publisher can accept any legal responsibility for any errors or omissions that may be made. The publisher makes no warranty, express or implied, with respect to the material contained herein. Printed on acid-free paper Springer is part of Springer Science+Business Media (www.springer.com)

Foreword

This book represents a timely and up-to-date series of papers on axons and follows in the illustrious footsteps of the volume by Stephen Waxman and colleagues on the structure, function and pathophysiology of the axon, dating from 1995. It is relatively unusual to see the axon singled out as the subject for an edition of original papers and the editors are to be congratulated. They have divided the contributions into developmental, functional, connectional and pathophysiological aspects. However, as might be expected, no single contribution compartmentalises itself so strictly and in each chapter we see multiple facets of axonal structure and function. This volume is not meant to be comprehensive but presents some important new developments and technical approaches. The axon was for long seen as a “simple” cable transmitting information via an all-or-none action potential from a neuronal cell body to the axon’s terminal, the presynaptic bouton. However, things are much more complex than that, as we see in this book. Any neuron normally has a single axon, although it can form numerous branches and typically innervates up to several thousand postsynaptic cells. Neurons can have long or short axons, from a few hundred microns long for some local, intrinsic neurons, to a metre or more for certain motor neurons. Axons can also vary in diameter, from a millimetre or so for the squid giant axons, to a fraction of a micron from many more “conventional” axons. The early ontological development of axons is a subject which is making rapid progress, although many mysteries still persist.

Development To begin with, Vera Niederkofler and Esther Stoeckli take a familiar model and discuss molecular aspects of commissural axonal guidance. They emphasise that developing axons need to find their way through their neural environment to establish functional circuits. We still do not know exactly how axons do this. The authors examine the growth cone at the tip of the developing axon where receptor proteins bind to guidance cues and transduce a signal to the cell. A change in cytoskeletal dynamics in turn induces the growth cone to change direction and microtubules consolidate these changes. Guidance cues can be either attractive or repulsive and either v

vi

Foreword

long or short range. Developing axons typically do not go straight to their target, but rather encounter intermediate targets which provide them with directional information. A wrong choice at such “decisions points” can be corrected by apoptosis, but if not they can manifest themselves pathologically. Next, Zoltán Molnár, Wei Zhi Wang, Maria Carmen Piñon, Shinichi Kondo, Franziska Oeschger and Anna Hoerder-Saubedissen pursue the theme of the formation of early neuronal circuits by considering a highly topical subject, the foetal cerebral cortex and its subplate. Cortical development involves remodelling of transient circuits through interaction of developmental programmes and environment. The fascinating cortical subplate contains some of the first generated neurons in the brain. Developing cortical afferents and efferents must find their way through the subplate, so its formation and fate are vital for the structure and function of the mature neocortex. Most subplate neurons die and make way for permanent cortical circuits. The developing brain is highly vulnerable to aberrations in transient subplate circuits which may be associated with developmental disorders, including epilepsy, schizophrenia and autism. The authors review data emerging from the use of transgenic mice and specific markers of subplate neurons to examine their synaptic input and output characteristics during development.

Axonal Function Tony Kelly and Christine Rose discuss the significance of sodium signals in axonal function. We have long known that regulation of intracellular sodium concentration is critical for neuronal function, because sodium ions are major current carriers and because many cellular functions are directly dependent on the sodium gradient. However, dynamic measurement of sodium in discrete cellular compartments has only recently become possible due to advances in spatial and temporal resolution of cellular imaging. Microspectrofluorescent recordings in neuronal cell bodies, dendrites or glial processes have revealed significant changes in sodium concentration during modest physiological stimuli. However, there have been few measurements of sodium concentration in axons. This chapter discusses sodium signals and routes of sodium entry in axons and possible physiological and pathological implications of changes in axonal sodium concentration. In contrast to large cell bodies which are amenable to measurements of sodium concentration by ion-sensitive microelectrodes, recording from small cellular compartments is not feasible. An appropriate method in such compartments is the use of sodium-sensitive dyes, such as sodiumbinding benzofuran isophthalate (SBFI), which diffuses into small cellular processes and is thus suited for sodium concentration measurements from axons and small dendrites. Dominique Debanne and Sami Boudkkazi provide new insights into information processing in the axon. They emphasise the great variability in axons, in terms of length, diameter and complexity of branching. The axon is defined as a neuronal process transmitting information via action potentials from a cell body to a nerve

Foreword

vii

terminal, but it is not limited to this and new findings suggest that other complex operations can take place along the axon. Axonal geometry and biophysical properties determine the timing of propagation of efferent messages in different axonal branches. In addition, axons link central neurons through gap junctions that allow fast network synchrony. Moreover, local shaping of an action potential may determine synaptic efficacy during repetitive stimulation. The authors review different ways in which the properties of axons can control the transmission of electrical signals. They consider two novel operations achieved by the axonal membrane, signal amplification and integration and review how axonal geometry and channel distribution can affect propagation by introducing delays in conduction or causing propagation failures or reflection of the action potential. The theme of electrical coupling of axons is pursued by Gunnar Birke, Dietmar Schmitz and Andreas Draguhn. Traditionally, neuronal information processing is a function of a rather static axonal transmission, chemical synapses and dendritic integration. But there can be backwards dendritic propagation of action potentials, branching point failures of propagating action potentials, ectopic origin of spikes at downstream locations and propagation of subthreshold synaptic potentials along axons. The authors summarise evidence for autonomous activity in axonal networks through fast and efficient electrical coupling between axons, circumventing signal integration at chemical synapses. Electrical coupling can be by gap junctions or ephaptic interactions. Such an ephaptic interaction is considered rather non-specific and has been mostly discussed in pathophysiological states: it may contribute to epileptic discharges, but it may be present in healthy tissue. Quan Wen and Dmitri Chklovskii ask what determines whether an axon should be myelinated or not. Was Rushton’s concept, dating from 1951, correct that for the same conduction velocity, a thinner fibre would be chosen for its smaller volume cost? His argument led to the prediction of a critical axon diameter for myelination. The authors discuss this concept in the light or recent measurements in the corpus callosum. Although myelinated axons are typically thicker than non-myelinated ones, the diameter distributions of both overlap. The question remains open.

Axons in Circuits We return to an axonal perspective on cerebral cortex by Tom Binzegger, Rodney Douglas and Kevan Martin who emphasise that complex cortical circuits remain a fundamental problem of neuroscience. Anatomy has for long traced pathways involving axons and dendrites, and electrophysiology revealed connections between neurons. New optical methods provide further clues about functional architecture, but a basic difficulty remains: Which neurons connect to which in a network of thousands? What is the role of feed forward and recurrent signal flow in cortical function or the degree of specificity involved in wiring up neurons? The authors discuss the possibilities and limitations of interpreting cortical circuits from analysis of overlapping axonal and dendritic arbors, either using Golgi impregnations or,

viii

Foreword

better, injected dyes. They follow the history of establishing cortical circuitry, from Ramón y Cajal to today, and illustrate the exquisitely complex results that have been obtained, and the scope for functional correlations. Sensory cortices have been demonstrated to be organised in functional units, oriented perpendicular to the surface, the so-called cortical columns, such as described some decades ago by Mountcastle and Powell, Hubel and Wiesel and Szentágothai. Moritz Helmstaedter and Dirk Feldmeyer take up the theme of the cortical column and axonal connectivity within and between columns. They discuss classic studies of long-range connections, such as between different regions of the cerebral cortex, using tract tracing methods. Analysis of local circuits is more challenging due to the proximity of the origin of short axons and their postsynaptic neurons. Ultrastructural identification of synapses has been used for over 50 years and synapse-specific labelling for light and electron microscopy more recently. It has, however, not been possible to study in detail synaptic contacts in large local populations of pre- and postsynaptic neurons. Indirect techniques at intermediate resolution can be used to infer synaptic connectivity based on light microscopic images, on the assumption that the probability of finding a synaptic contact is proportional to the numerical density of presynaptic axons and postsynaptic dendrites in a given region. The authors discuss how this can be used to define types of cortical interneurons and their capacity to inhibit neurons located within the same and in neighbouring cortical columns. Joachim Lübke and Dirk Feldmeyer develop some of these themes to consider projection patterns and target specificity of excitatory axons in the neocortex. The neocortex contains two major classes of neurons, excitatory pyramidal and spiny stellate neurons, and a heterogeneous population of GABAergic interneurons with axons mainly projecting within a single layer or column. Long-range projection excitatory neurons tend to have two axonal domains, one within its cortical column, and another to other cortical areas, ipsi- or contralateral, or to subcortical brain regions. The authors use the example of the “barrel field” in the somatosensory cortex of rodents in which each whisker is represented topographically in layer IV. These barrels project to the other layers to form a particularly well-demarcated column. The authors compare the axonal morphology of barrel column neurons with those of other areas, in particular the visual cortex.

Degeneration and Regeneration of Axons Lucy Broom and Hugh Perry discuss mechanisms and consequences of axonal degeneration. They characterise the axon as a functionally distinct and partly autonomous compartment. They also deal with consequences of axonal degeneration on glial cells and macrophages. Research on neurodegenerative diseases has for long concentrated on the cell body, but axonal loss in traumatic and neurodegenerative conditions is of utmost importance. Axonal loss is an early event in disease and is basic to the functional deficits. While therapeutic intervention to

Foreword

ix

protect the cell body may be of clinical value, protection of the axon may prove an important consideration. This requires an understanding of the extent to which the axon can function as an independent cellular compartment under normal physiological conditions. We have generally considered protein synthesis as a function of the cell body, with subsequent transport to the axon, notably as fast transport of vesicles and membranous organelles, and slow transport of cytoskeletal and cytoplasmic proteins. However, it seems unlikely that proteins would only use such forms of transport in, for example, very long axons of spinal motor neurons, for they would not survive long enough to reach their targets before being degraded. Following axotomy the earliest regeneration at the axonal tip can occur within 24 h, suggesting a local axonal response. Supportive evidence is the presence of ribosomes, either free or on structures similar to the endoplasmic reticulum, tRNA, initiation factors and mRNA in axons, and the authors discuss this in detail. Axons also have the capacity to export proteins to the cell surface, which suggests the existence of a Golgi apparatus. Axonal lipid biosynthesis has been demonstrated and the axon should now be considered as a metabolically active neuronal compartment. Sven Hendrix and Robert Nitsch ask if the immune system helps regeneration after central nervous lesions. Injury to the central nervous system is followed by an inflammatory response, characterised by a short first acute phase and a longterm remodelling phase. Different functions of T lymphocytes have been suggested in these two phases. The authors propose that specific subtypes of T lymphocytes may be responsible for beneficial or detrimental effects in this biphasic response to trauma. Type 2 T helper cells may play a beneficial role after lesions by promoting axonal outgrowth and protecting the nervous system from self-reactive inflammation. However, as the authors discuss, a systematic analysis of these effects is still needed. This book achieves a happy balance between our anatomical and physiological views of axons over several decades and very recent, more cellular and molecular biological, as well as developmental, concepts. It goes much further than simply discussing axons, and puts them in their rightful context, as part not only of neurons but also of neuronal networks, and even as far as the “highest” part of ourselves, our cerebral cortex. Perroy, Switzerland February 2010

Laurence Garey

Preface

Neurons are interconnected via axons, a term originally introduced in 1896 by the Swiss physiologist and anatomist Rudolph Albert von Kölliker. The importance and structural variability of axons was beautifully described by the work of Ramón y Cajal. Axons are effectively the primary transmission lines of the nervous system. They conduct electrical signals termed action potentials, initiated at the axon hillock. The frequency of action potentials are the code with which neurons communicate with each other or directly with the effector organs such as muscles, heart and the gastrointestinal tract. Using the squid giant axon as an experimental system, Alan Hodgkin and Andrew Huxley obtained first a quantitative description of the ionic mechanisms underlying an action potential, the well-known Hodgkin–Huxley model. Ever since, the axon has remained a focus of neurobiological research and a wealth of literature has been published on axonal structure and function. The intention of this book was to highlight some of these aspects. It was never meant to be comprehensive but focuses mainly on recent studies concerning axonal connectivity and its development. We would like to thank all the contributing authors for their willingness to participate in this endeavour and their patience during the composition of this book. We would also acknowledge numerous colleagues that – over time – have influenced our work and thereby contributed to the creation of this book. Furthermore, we are particularly grateful to Laurence Garey for agreeing to write such a nice introductory preface and to deliver it at such short notice. Finally, we are very thankful to Ann Avouris and Melissa Higgs from Springer for the invaluable help during the production of this book on axons. Jülich and Aachen, Germany February 2010

Dirk Feldmeyer Joachim Lübke

xi

Contents

Part I

Axons in Development

1 Molecular Aspects of Commissural Axon Guidance . . . . . . . . Esther T. Stoeckli and Vera Niederkofler 2 Subplate and the Formation of the Earliest Cerebral Cortical Circuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . Zoltán Molnár, Wei Zhi Wang, Maria Carmen Piñon, Daniel Blakey, Shinichi Kondo, Franziska Oeschger, and Anna Hoerder-Suabedissen Part II

3

19

Axonal Function

3 Sodium Signals and Their Significance for Axonal Function . . . . . . . . . . . . . . . . . . . . . . . . . . Tony Kelly and Christine R. Rose

35

4 New Insights in Information Processing in the Axon . . . . . . . . Dominique Debanne and Sami Boudkkazi

55

5 Electrical Coupling of Axons . . . . . . . . . . . . . . . . . . . . . Gunnar Birke, Dietmar Schmitz, and Andreas Draguhn

85

6 To Myelinate or Not to Myelinate? . . . . . . . . . . . . . . . . . . Quan Wen and Dmitri B. Chklovskii

103

Part III

Axons and Neuronal Circuits

7 An Axonal Perspective on Cortical Circuits . . . . . . . . . . . . . Tom Binzegger, Rodney J. Douglas, and Kevan A.C. Martin 8 Axons Predict Neuronal Connectivity Within and Between Cortical Columns and Serve as Primary Classifiers of Interneurons in a Cortical Column . . . . . . . . . . . . . . . . . Moritz Helmstaedter and Dirk Feldmeyer

117

141

xiii

xiv

Contents

9 The Axon of Excitatory Neurons in the Neocortex: Projection Patterns and Target Specificity . . . . . . . . . . . . . Joachim H.R. Lübke and Dirk Feldmeyer Part IV

157

Axons and Degeneration/Regeneration

10 Axon Degeneration: Mechanisms and Consequences . . . . . . . Lucy J. Broom and V. Hugh Perry

181

11 Regeneration After CNS Lesion: Help from the Immune System? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sven Hendrix and Robert Nitsch

209

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

233

Contributors

Tom Binzegger University of Newcastle, Newcastle upon Tyne, UK; Institute of Neuroinformatics, University Zürich and ETH Zürich, Zürich, Switzerland, [email protected] Gunnar Birke Institut für Physiologie und Pathophysiologie, Universität Heidelberg, Heidelberg, Germany, [email protected] Daniel Blakey Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK Sami Boudkkazi INSERM U641, Marseille, France; Faculté de médecine secteur nord, Université de la Méditerranée, Marseille, France, [email protected] Lucy J. Broom CNS Inflammation group, University of Southampton, Southampton, UK, [email protected] Dmitri B. Chklovskii Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, USA, [email protected] Dominique Debanne INSERM U641, Marseille, France; Faculté de médecine secteur nord, Université de la Méditerranée, Marseille, France, [email protected] Rodney J. Douglas Institute of Neuroinformatics, University Zürich and ETH Zürich, Zürich, Switzerland, [email protected] Andreas Draguhn Institut für Physiologie und Pathophysiologie, Universität Heidelberg, Heidelberg, Germany, [email protected] Dirk Feldmeyer Research Centre Jülich GmbH, Institute of Neuroscience and Medicine, INM-2, Jülich, Germany; Department of Psychiatry and Psychotherapy, RWTH Aachen University, Aachen, Germany, [email protected] Laurence Garey Centre for Psychiatric Neuroscience, Hôpital de Céry, Lausanne, Switzerland, [email protected]

xv

xvi

Contributors

Moritz Helmstaedter Max Planck Institute for Medical Research, Heidelberg, Germany, [email protected] Sven Hendrix Department of Functional Morphology & BIOMED Institute, Hasselt University - Campus Diepenbeck, Agoralaan Gebouw C, BE 3590 DIEPENBEEK, Belgium, [email protected] Anna Hoerder-Suabedissen Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK, [email protected] Tony Kelly Department of Epileptology, Life & Brain Center, University of Bonn Medical Center, Sigmund-Freudstr. 25, Bonn, Germany, [email protected] Shinichi Kondo Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK, [email protected] Joachim H.R. Lübke Research Centre Jülich GmbH, Institute of Neuroscience and Medicine, INM-2, Jülich, Germany; Department of Psychiatry and Psychotherapy, RWTH Aachen University, Aachen, Germany, [email protected] Kevan A.C. Martin Institute of Neuroinformatics, University Zürich and ETH Zürich, Zürich, Switzerland, [email protected] Zoltán Molnár Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK, [email protected] Vera Niederkofler Institute of Molecular Life Sciences, University of Zürich, Zürich, Switzerland, [email protected] Robert Nitsch Institute of Microanatomy and Neurobiology, University Medical Center, Johannes-Gutenberg-University, Langenbeckstrasse 1, 55131 Mainz, Germany, [email protected] Franziska Oeschger Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK, [email protected] V. Hugh Perry CNS Inflammation group, University of Southampton, Southampton, UK, [email protected] Maria Carmen Piñon Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK, [email protected] Christine R. Rose Institute for Neurobiology, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany, [email protected] Dietmar Schmitz Neuroscience Research Center, Charité - Universitätsmedizin Berlin, Schumannstr. 21/22, 10117 Berlin, Germany, [email protected] Esther T. Stoeckli Institute of Molecular Life Sciences, University of Zürich, Zürich, Switzerland, [email protected]

Contributors

xvii

Wei Zhi Wang Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK, [email protected] Quan Wen Department of Physics and Astronomy, State University of New York at Stony Brook, Stony Brook, NY, USA; Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, USA, [email protected]

Part I

Axons in Development

Chapter 1

Molecular Aspects of Commissural Axon Guidance Esther T. Stoeckli and Vera Niederkofler

During development of the nervous system, growing axons must navigate through their environment to find their correct target. Accurate pathfinding of axons is essential for the establishment of functional neuronal circuits. How do axons know where to go? What provides them with the necessary information? Which part of the axon makes the directional decision? These are questions scientists have been trying to answer over decades of axon guidance research. Few of them have been solved; many of them remain a mystery.

1.1 Basic Principles of Axonal Navigation The growth cone is a specialized and sensitive motile structure located at the tip of the elongating neurite (Smith, 1988; Huber et al., 2003; Zhou and Cohan, 2004). It has the ability to steer the axon by sensing guidance cues provided by the environment. Receptor proteins expressed on the surface of the growth cone bind to these guidance cues and transduce a signal to the cell. Virtually all signaling cascades triggered by the binding of a guidance cue to its receptor converge on the cytoskeleton (Guan and Rao, 2003; Huber et al., 2003; Kalil and Dent, 2005). A change in actin cytoskeletal dynamics in turn induces a directional change of the growth cone, while microtubules consolidate changes in direction and position (Gordon-Weeks, 2004; Kalil and Dent, 2005). Guidance cues can be either attractive or repulsive (Tessier-Lavigne and Goodman, 1996; Dickson, 2002). Importantly, this separation is not only dependent on the guidance cues themselves but also on the complement of receptors expressed on the growth cone, the intracellular state of active signaling molecules such as cyclic nucleotides, and cross talk between intracellular signaling cascades (Yu and Bargmann, 2001; Huber et al., 2003; Wen et al., 2004). Guidance cues can be divided into long- and short-range cues, depending on whether they are secreted E.T. Stoeckli (B) Institute of Molecular Life Sciences, University of Zürich, Zürich, Switzerland e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_1, 

3

4

E.T. Stoeckli and V. Niederkofler

or remain stationary. In summary, guidance cues can act in four different ways on a growing axon: long-range attractive, long-range repulsive, short-range attractive, and short-range repulsive. Developing neurites typically do not project straight to their target. Instead axons encounter several intermediate targets along their trajectory, known as “choice points,” which provide them with directional information in form of guidance molecules. Wrong decisions at a choice point can be corrected during development by induction of apoptosis. If no adjustment takes place, they can manifest themselves as “wiring” defects and subsequent neurological diseases (Lesnick et al., 2007; Yaron and Zheng, 2007).

1.2 Commissural Neurons as a Model System to Study Axonal Pathfinding To date, the best-characterized intermediate targets are the midlines of the vertebrate and insect central nervous system (CNS). The midline serves as a choice point for commissural axons on their journey through the developing nervous system. The vertebrate spinal cord and the insect ventral nerve cord have been extremely valuable not only in discovering and characterizing new axon guidance molecules but also in revealing general principles of axon pathfinding. Commissural neurons are interneurons that connect the two body halves in bilaterally symmetric organisms, by projecting their axons to the contralateral side. This ensures the vital information transfer between the two body halves. Although many aspects of commissural axon guidance are controlled by mechanistically and evolutionarily conserved ligand–receptor systems, others are different between species. Here we compare and contrast commissural axon pathfinding in the vertebrate spinal cord and the insect ventral nerve cord.

1.2.1 Commissural Axon Guidance in the Vertebrate Spinal Cord Developing spinal commissural neurons represent a large, heterogeneous class of interneurons that are widely distributed along the dorsoventral axis of the spinal cord. They can be classified on the basis of morphology, cell body position, and gene expression (Kadison and Kaprielian, 2004). In both lower and higher vertebrates spinal commissural axons follow a stereotypical, circumferential trajectory to the floorplate. After crossing the floorplate, most commissural axons abruptly alter their direction of growth from the transverse to the longitudinal plane and elaborate a variety of contralateral projections, but never recross the ventral midline (Kadison and Kaprielian, 2004; Nissen et al., 2005). In rodent and chick embryos, commissural axons whose cell bodies are located in the dorsal spinal cord turn almost exclusively rostrally, toward the brain,

1

Molecular Aspects of Commissural Axon Guidance

5

after crossing the midline (Imondi and Kaprielian, 2001). This uniform behavior makes them a powerful model to study axon guidance. In principle, the trajectory of dorsal commissural neuron pathfinding can be divided into three phases: first, ventral growth toward the floorplate; second, crossing of the floorplate; and third, rostral turn and projection parallel to the floorplate. 1.2.1.1 Guidance Toward the Floorplate Early in vertebrate development dorsal commissural neurons send out their axons ventrally toward the floorplate. The molecules causing this initial directional growth have been identified as members of the BMP (bone morphogenetic protein) family. BMP7 and GDF7 secreted by the roofplate, a specialized group of glial cells at the dorsal midline, have been shown to act as heterodimeric long-range repellents for commissural neurites, driving elongation ventrally (Augsburger et al., 1999; Butler and Dodd, 2003). The receptor complex transmitting this signal has not yet been identified. In addition to being repelled by the roofplate, commissural axons are attracted by the floorplate (Kennedy et al., 1994). Commissural axons express Dcc (deleted in colorectal cancer) and are attracted by the ligand Netrin-1 emanating from the floorplate and the ventral neural tube (Fig. 1.1) (Serafini et al., 1996; Charron et al., 2003). Netrin-1 protein accumulates in an increasing dorsoventral gradient through the developing spinal cord (Kennedy et al., 2006). Interestingly, mice deficient in Netrin-1 or Dcc still exhibit a thin ventral commissure

Fig. 1.1 Switch from attraction to repulsion at the vertebrate midline. Netrin-1 attracts commissural axons expressing Dcc to the midline. Robo1 activity is kept low by Robo3. After midline crossing Robo3 is downregulated by an unknown mechanism. Subsequently, Robo1 is upregulated and binds Dcc preventing Netrin-1-mediated attraction. Notably, Dcc remains expressed on commissural axons after midline crossing. Double-headed arrows indicate interactions. Crosses indicate no response

6

E.T. Stoeckli and V. Niederkofler

indicating that there might be other molecules attracting commissural neurons toward the floorplate. Gli-2 mutant mice lacking floorplate cells exhibit a severe reduction of commissural neurons growing toward and crossing the midline (Ding et al., 1998; Matise et al., 1998; Charron et al., 2003). Residual axons do occasionally approach and cross, presumably due to Netrin-1 produced by the periventricular zone. Finally, Netrin-1/Gli2 double mutant mice exhibit a stronger phenotype than either single mutant alone, namely a complete lack of ingrowth of commissural neurons into the ventral spinal cord resulting in the absence of the ventral commissure. This indicates that there must be additional chemoattractive signals produced by the floorplate, besides Netrin-1. And indeed, based on this hypothesis, Sonic hedgehog (Shh) was identified as the additional attractive source, acting in concert with Netrin1. Using collagen gel coculture assays, Charron et al. showed that commissural axons reorient toward a source of Shh, establishing a role for Shh in this process (Charron et al., 2003). Addition of cyclopamine, a compound which blocks the Patched/Smoothened signaling pathway, diminishes the remaining attractive activity of Netrin-1 mutant floorplates. Beside Patched, Boc (biregional Cdon-binding protein) has recently been suggested as Shh receptor mediating commissural axons’ attraction to the floorplate (Okada et al., 2006). Together, the combination of repulsive BMPs from the roofplate and attractive Netrin-1 and Shh from the floorplate steers axon bundles toward the ventral midline. Within the nervous system, morphogens, such as BMPs and Shh, were long believed to play exclusive roles in regulating patterning processes by initiating transcriptional changes (Lee and Jessell, 1999; Jessell, 2000; Patapoutian and Reichardt, 2000; Altmann and Brivanlou, 2001). Involvement of these molecules in axon guidance assigned a novel function to them, which in addition to occurring later in development, operated locally at the growth cone (Salie et al., 2005). Another morphogen family, the Wnt family, has recently been implicated in commissural axon guidance (Section 1.2.1.3). 1.2.1.2 Guidance Across the Floorplate The ventral midline of the developing CNS represents a binary choice point for growing axons. Upon reaching the midline, each axon must decide whether or not to cross over to the other side. There is now an increasing body of evidence from both vertebrate and invertebrate systems that a variety of short- and long-range guidance mechanisms control the axons’ decision to cross the midline. Cell adhesion molecules of the immunoglobulin (Ig)-superfamily have been shown in vertebrates to establish vital contacts between the commissural neurons and the floorplate (Stoeckli and Landmesser, 1998). Studies in the chick showed that a direct interaction between two cell adhesion molecules of the Ig superfamily (or IgCAMs), Axonin-1/TAG-1 and NrCAM on commissural axons and floorplate cells, respectively, normally renders the floorplate permissive for commissural growth cone entry and passage (Stoeckli and Landmesser, 1995). Blocking their interaction leads to pathfinding errors. In particular, commissural axons fail to enter the floorplate and instead turn along the ipsilateral floorplate border. This finding suggests

1

Molecular Aspects of Commissural Axon Guidance

7

that in the absence of the positive signal provided by the Axonin-1/NrCAM interaction, the floorplate is inhibitory for commissural axons (Stoeckli et al., 1997). The negative signal associated with the floorplate is most likely provided by the Slit family, although additional cues have been identified in the mouse (Zou et al., 2000). Thus, midline crossing by commissural axons is regulated by a balance between positive (Axonin-1/NrCAM) and negative (Slits/Robo receptors) signals. Commissural axons cross the midline once and leave the floorplate on the contralateral side again. What makes commissural axons cross the midline but prevents ipsilaterally projecting neurites from doing so? Why are commissural axons not attracted back to the midline after crossing it? It is generally believed that a repellent force provided by Slits from the floorplate, acts on commissural axon growth cones expressing the Robo receptors (Fig. 1.1) (Brose et al., 1999; Kidd et al., 1999; Long et al., 2004). Robo1 and Robo2 are expressed on commissural axons only after they have reached the midline. Commissural axons can cross once because they do not initially express Robo protein on their surface, even though Robo1 and Robo2 mRNA is found in their cell bodies (Long et al., 2004; Mambetisaeva et al., 2005). Upon crossing, through a mechanism that is still poorly understood in vertebrates, but quite well characterized in Drosophila (Section 1.2.2.2), growth cones upregulate Robo protein on their surface and become responsive to Slits, which prevents them from recrossing. Axons projecting ipsilaterally, exhibit high surface levels of Robo protein and therefore cannot cross, due to repulsion by Slits. How is Robo protein upregulated once commissural axons have reached the midline? Conversely, what represses Robo expression before commissural axons reach the midline? Of the four Robo genes present in mammals, all but Robo4 are expressed in commissural neurons, in a partially overlapping pattern (Kidd et al., 1998; Yuan et al., 1999; Huminiecki et al., 2002; Sabatier et al., 2004). Robo1 and Robo2 start being expressed strongly once commissural axons have reached the midline (Long et al., 2004). Interestingly, based on studies using a TAG-1 antibody (which stains the majority of precrossing commissural axons), only mice lacking Robo1 but not Robo2 show midline crossing defects, suggesting a distinct role of these two proteins. Unfortunately, due to the proximity of Robo1 and Robo2 on the same chromosome, double mutant animals have proven difficult to generate. In addition to Robo1 and Robo2 a third member of the Robo family expressed by commissural neurons, Robo3/Rig1, has been identified (Sabatier et al., 2004). In contrast to Robo1 and Robo2, Robo3 protein is expressed strongly only prior to crossing the midline, but not after (Fig. 1.1). Strikingly, lack of Robo3 in mouse results in a complete failure of commissural axons to enter the ventral midline region, reflected in the lack of ventral commissures throughout the spinal cord. Genetic and in vitro analyses indicate that Robo3 functions to repress Slit responsiveness of commissural axons by suppressing Robo1 repulsive activity. This allows commissural axons to cross the floorplate. After crossing, Robo3 is downregulated allowing Robo1 to be active and mediate the repulsive Slit signal. So far, the mechanism of Robo3 interference with Robo1 function is unclear. Furthermore, it is unknown how Robo3 is downregulated upon midline crossing. Noticeably, it has been shown that Robo1 and Robo2 protein levels are unchanged in Robo3

8

E.T. Stoeckli and V. Niederkofler

mutant mice. This observation is distinct from what is seen in Drosophila commissureless mutants, in which Robo protein is abnormally expressed on precrossing commissural axons (Section 1.2.2.2). Interestingly, a human syndrome called horizontal gaze palsy and progressive scoliosis (HGPPS) was found to be associated with mutations in the human Robo3 gene (Jen et al., 2004). A striking feature of this syndrome is the aberrant ipsilateral projections of major ascending and descending axon pathways, evidently due to a failure of these axons to cross the midline in the hindbrain (Jen et al., 2004; Bosley et al., 2005). Notably, Dcc, the Netrin receptor, remains expressed on commissural neurons once they have reached the midline (Fig. 1.1) (Keino-Masu et al., 1996). Why do commissural axons leave the floorplate, if it is still attractive? For growth cones to progress from their intermediate target, the floorplate, in an efficient manner, they should not only upregulate their responsiveness to a midline repellent, but should also lose the attractive responses to the midline that got them there in the first place. And indeed, this is exactly what happens. Commissural axons lose responsiveness to Netrin-1 upon crossing the midline, despite continued expression of the Netrin receptor Dcc (Shirasaki et al., 1998). Even more strikingly, the gain of responsiveness to the repellent cue, Slit2, and the loss of responsiveness to the attractant cue, Netrin-1, have been shown in Xenopus spinal axons to be causally linked (Stein and Tessier-Lavigne, 2001). Activation of the Robo1 receptor by Slit2 silences the attractive effect of Netrin-1, through direct interaction between the cytoplasmic domains of Robo1 and Dcc. In other words, the activation of Robo by Slit at the midline does not only serve to repel the axons from the midline but also to switch off the attraction of commissural axons to the midline. Three mammalian Slit proteins have been identified, all of which are expressed by floorplate cells (Brose et al., 1999). Their functions as midline repellents seem to be at least partially redundant, since commissural axons of Slit1;Slit2;Slit3 triple mutant mice exhibit a much more severe midline crossing defect compared to any single or double Slit mutant (Long et al., 2004). In Slit1;Slit2;Slit3 triple mutant spinal cords, many commissural axons stall in the floorplate, while others recross the floorplate multiple times. Nevertheless, normal projecting axons are still seen in the triple mutants. This contrasts with what has previously been described in Drosophila, where the removal of slit leads to the collapse of both contra- and ipsilaterally projecting axons into the ventral midline. This result suggests that in vertebrates, other repulsive cues beside Slit–Robo interactions are involved in guiding commissural axons beyond the floorplate. Based on current knowledge, Semaphorins or EphrinBs could provide this additional repulsive force. Semaphorin 3B expressed by the floorplate has been shown to be repulsive for postcrossing commissural axons expressing Neuropilin-2 in vitro (Zou et al., 2000). In Neuropilin-2 mutant mice commissural axons often stall within the floorplate, supporting the involvement of Semaphorin 3B as midline repellent. The observed defects are only partially penetrant and seem to be corrected as the embryo matures, indicating the operation of redundant guidance mechanisms, such as Slits and Ephrins. Furthermore, Neuropilin-2 is not expressed by commissural neurons

1

Molecular Aspects of Commissural Axon Guidance

9

in chick indicating that Semaphorin 3B repulsion of commissural axons may be species specific (Mauti et al., 2006). Furthermore, EphB–EphrinB interactions have been suggested to elicit a repulsive response from commissural axons. Expression analysis revealed transmembrane EphrinBs to be present in the floorplate region and the dorsal spinal cord in mouse and chick (Imondi et al., 2000; Imondi and Kaprielian, 2001; Jevince et al., 2006). EphB proteins are only present on postcrossing commissural axons that project longitudinally between the Ephrin expression domains, namely the floorplate and the dorsal spinal cord. Furthermore, all three EphrinBs collapse commissural growth cones in vitro (Imondi et al., 2000). Their repulsive activity is corroborated by commissural neurons executing their final turn in the longitudinal axis at the ventral-most boundary of EphrinB expression in the dorsal spinal cord. In vitro perturbation of endogenous EphB–EphrinB interactions in intact spinal cord explants results in the inappropriate growth of decussated commissural axons into dorsal regions of EphrinB expression (Imondi and Kaprielian, 2001). Crossing defects at the floorplate were not observed, despite the presence of EphrinB. These data suggest that the trajectory of commissural neurons is restricted by the repulsive force of EphrinBs on EphB-positive axons in the dorsal spinal cord. The function of EphrinBs in the floorplate appears negligible. In contrast to in vitro experiments, in mice lacking Ephrin-B3 or multiple EphB receptors, a small, but significant number of axons (max. 15%) follows aberrant trajectories in the immediate vicinity of the ventral midline (Kadison et al., 2006). From the crossing defects observed, it is not apparent that EphrinB3 emanating from the floorplate can act as repulsive cue for commissural axons expressing EphB receptors. In addition, only a minor fraction of commissural axons seems to be affected by the lack of EphrinB3–Eph interaction. Also inconsistent with the model of EphrinB acting as a repellent is the fact that contralateral projections along the dorsal EphrinB boundary are not altered in EphrinB mutant mice. Although EphrinBs have been implicated as key players in several other midline decisions, they appear to play only a minor role in commissural axon guidance in the spinal cord (Nakagawa et al., 2000; Williams et al., 2003).

1.2.1.3 Rostral Turning and Longitudinal Projection After Crossing After crossing and exiting the floorplate successfully, dorsal commissural axons make a sharp rostral turn and grow close to the floorplate toward the brain. Although much less is known about commissural axon pathfinding along the anterior– posterior axis than along the dorsal–ventral axis, the floorplate seems once again to be the key player in conveying directional information (Stoeckli, 2006). The importance of this structure becomes immediately apparent when looking at commissural axon guidance in vertebrates lacking a floorplate. In this situation commissural axons no longer turn into the longitudinal axis (Bovolenta and Dodd, 1991; Clarke et al., 1991; van Straaten and Hekking, 1991; Greenspoon et al., 1995; Matise et al., 1999).

10

E.T. Stoeckli and V. Niederkofler

For a long time, F-Spondin, a protein produced and secreted by the floorplate, was the only known guidance cue involved in the longitudinal turn of commissural axons (Burstyn-Cohen et al., 1999). F-Spondin has been shown by in vitro and in vivo experiments in the chick to restrict the turning angle of postcrossing commissural axons to the longitudinal axis at the contralateral floorplate border without affecting the direction of the turn. F-Spondin is required to prevent the lateral drifting of commissural axons after having crossed the floorplate via a receptor of unknown identity. The molecules responsible for determining the direction of the turn have only recently been identified. The morphogen Wnt4 was the first molecule identified to be involved in the rostral turn of dorsal commissural axons (Lyuksyutova et al., 2003). In the mouse, Wnt4 is expressed in a gradient by the floorplate with high rostral to low caudal levels during the time window when dorsal commissural axons make the decision to grow toward the brain (Fig. 1.2). In vitro, it was shown that Wnt4 attracts commissural neurons that have already crossed the midline, but does not affect axons that have not yet crossed the midline. In vivo and in vitro experiments suggest Frizzled3 to be the receptor expressed by commissural neurons that mediates Wnt4 attraction.

Fig. 1.2 Guidance of commissural axons along the longitudinal axis. a In mouse, Wnt4 expressed in a high-rostral to low-caudal gradient attracts postcrossing commissural axons expressing the Frizzled3 receptor. b In chick, Shh expressed in a low rostral to high caudal gradient repels postcrossing commissural axons expressing the Hhip receptor. RP: roofplate, FP: floorplate

Notably, high anterior/rostral to low posterior/caudal Wnt gradients have also been shown to be crucial for the formation of the corticospinal tract in the murine dorsal funiculus. Wnt1 and Wnt5a act as chemorepellents driving Ryk-positive corticospinal axons caudally (Liu et al., 2005). Disruption of the Wnt–Ryk interactions results in premature stalling of corticospinal axons. The attractive effect on commissural neurons and the repulsive effect on corticospinal axons suggest a general role of Wnt proteins in anterior–posterior guidance of multiple classes of axons. Similarly to Wnt4 in the mouse, Sonic hedgehog (Shh) was identified in the developing chick to provide postcrossing commissural axons with the directional cue to turn rostrally (Fig. 1.2) (Bourikas et al., 2005). In contrast to murine Wnt4, Shh was shown to act as a repellent rather than an attractant on postcrossing commissural axons and accordingly it is expressed in a high-caudal to low-rostral gradient

1

Molecular Aspects of Commissural Axon Guidance

11

in the floorplate. In ovo RNAi experiments demonstrated that Hhip (hedgehoginteracting protein) acts as the receptor mediating Shh repulsion of postcrossing commissural axons (Bourikas et al., 2005). In summary, there seem to be two forces provided by the floorplate acting selectively on postcrossing commissural axons: the attractive force of Wnt4 in a high-rostral to low-caudal gradient in the mouse and the repulsive force of Shh in a high-caudal to low-rostral gradient in chick. Whether Wnt4 and Shh are expressed in comparable gradients in chick and mouse, respectively, and whether their effects on postcrossing axons are similar between species remain to be elucidated.

1.2.2 Commissural Axon Guidance in Drosophila melanogaster Many molecules and mechanisms described in the vertebrate section above are similar in insect ventral commissure formation. Studying axon guidance in both vertebrates and invertebrates has very frequently resulted in synergism. The fact that molecules and mechanisms are often conserved between vertebrates and invertebrates allows for general conclusions about axon navigation. However, there are differences between invertebrates and vertebrates with respect to axonal navigation at the midline.

1.2.2.1 Guidance Toward the Midline Currently, there is no evidence that Hedgehog or BMPs contribute to commissure formation in Drosophila, as these molecules do in vertebrates. Drosophila commissural axons, like their vertebrate counterparts require Netrins (NetrinA and NetrinB) and their receptor Frazzled for midline crossing (Fig. 1.3) (Kolodziej et al., 1996; Garbe and Bashaw, 2007). However, in Drosophila Netrins appear to act as shortrange cues to facilitate crossing itself rather than as long-range chemoattractants that orient and guide commissural axons toward the midline (Brankatschk and Dickson, 2006). Drosophila commissural axons still orient normally and reach the midline even in the complete absence of netrins (netrinA and netrinB). Some of them fail to cross the midline which subsequently leads to a disruption of the anterior and posterior commissures. Expression of a membrane-tethered version of NetrinB in an otherwise netrin-deficient Drosophila embryo restores normal commissure formation. Despite the fact that Netrins do not act as a long-range attractant in midline crossing, they can act as a long-range repellent (Keleman and Dickson, 2001). In summary, Netrins only seem to be required for the actual commissure formation, but not for commissural neurons to grow toward the midline. It is not clear yet, what actually causes Drosophila commissural neurons to grow toward the midline. It is possible that long-range signals are not required due to the small size of the embryo and the short distances between the position of the neuronal cell body and the intermediate target.

12

E.T. Stoeckli and V. Niederkofler

Fig. 1.3 Commissural axon guidance in Drosophila. Commissural axons cross the midline with the help of Netrins (green) provided by the midline. Drl-positive neurons (grey) can only pass through the anterior commissure since they are repelled by Wnt5 (orange) expressed at the posterior commissure. Slit (yellow) pushes commissural axons away from the midline. Depending on the combinatorial expression of Robo proteins commissural axons project along either the medial, intermediate, or lateral tract (different blue shades) longitudinally. Drl: Derailed, PC: posterior commissure , AC: anterior commissure

1.2.2.2 Guidance Across the Midline Axons crossing the midline in Drosophila have to choose between two alternative routes: the anterior or the posterior commissure. Initial details regarding the mechanism responsible for this choice were revealed when Derailed (Drl; an atypical receptor tyrosine kinase with homology to vertebrate Ryk) was discovered to be both necessary and sufficient to direct axons through the anterior commissure (Fig. 1.3) (Bonkowsky et al., 1999). Drl is selectively expressed on axons choosing the anterior commissure and ectopic expression of Drl in neurons normally choosing the posterior commissure forces them to take the anterior path. Yoshikawa et al. have found that this effect is mediated by the interaction of Drl with Wnt5, which is present in a region surrounding the posterior commissure (Yoshikawa et al., 2003). Wnt5 appears to repel commissural axons expressing Drl and drive axons away from the posterior and into the anterior commissure (Fig. 1.3). In vertebrates, the Wnt–Ryk signaling pathway has been implicated in various axon guidance decisions (Keeble and Cooper, 2006). With respect to the midline, Ryk was shown to be necessary for the appropriate exit of callosal cortical neurons to the contralateral side after they reach the midline. In this system Wnt5a acts as chemorepellent on callosal neurons expressing Ryk (Keeble et al., 2006). Furthermore, Ryk functions as an axon-guidance receptor in the formation of the cortical spinal tract and the establishment of the retinotectal map (Liu et al., 2005; Schmitt et al., 2006). In all systems analyzed regarding axonal navigation to date,

1

Molecular Aspects of Commissural Axon Guidance

13

Wnt–Ryk interactions result in chemorepulsion, whereas Wnt–Frizzled interactions cause chemoattraction (Lyuksyutova et al., 2003; Liu et al., 2005; Keeble et al., 2006; Schmitt et al., 2006). Once at/in the midline, the question arises again of how to leave it. Although robo mRNA is expressed uniformly in all neurons of the Drosophila nerve cord, Robo protein is highly enriched on longitudinal axon tracts, i.e., ipsilateral and distal segments of contralateral projections. Expression of Robo protein inhibits midline crossing or recrossing of these axon bundles (Kidd et al., 1998). Although gainof-function studies showed that elevated levels of any of the three Robos can be sufficient to prevent crossing, loss-of-function experiments revealed that midline crossing is primarily regulated by Robo alone and not Robo2 and Robo3 (Kidd et al., 1999; Rajagopalan et al., 2000b; Simpson et al., 2000a). How are Robo protein levels regulated in Drosophila? In particular what keeps Robo protein levels low in commissural axons to allow midline crossing? In contrast to vertebrates, the underlying posttranscriptional mechanism is quite well characterized, at least for Robo. The first clues leading to the clarification of this mechanism came from the discovery of the commissureless (comm) mutant (Seeger et al., 1993). As the name suggests, loss of comm prevents the formation of the commissures because no axons cross the midline (Tear et al., 1996). In contrast, ectopic expression of Comm allows axons which normally would not cross the midline, because they express high Robo levels, to do so. According to the current model, the transmembrane protein, Comm, prevents Robo surface expression by binding to Robo directly and targeting it to endosomes (Keleman et al., 2002, 2005). After crossing the midline, Comm is downregulated and Robo is able to initiate repulsive signaling that prevents axons from recrossing. Thus, axons cross the midline, because Comm regulates the surface levels of Robo and thereby prevents premature repulsion from this intermediate target. Only axons that express Comm can cross the midline, while all others already express Robo before reaching the midline and are sensitive to the midline repellent Slit.

1.2.2.3 Longitudinal Projection After Crossing Longitudinally projecting axons can be divided into three major connectives or axon bundles based on their expression of a certain combination of Robos: axons forming the medial connective express only Robo, axons of the intermediate connective express both Robo and Robo3, and axons of the lateral connective express Robo, Robo2, and Robo3 (Fig. 1.3) (Rajagopalan et al., 2000a; Simpson et al., 2000b). Loss- and gain-of-function experiments revealed a “Robo code” that determines the lateral positions of longitudinal axons (Rajagopalan et al., 2000a; Simpson et al., 2000b). For example, loss of robo3 function leads to a shift of axons from the intermediate to the medial connective whereas loss of robo2 shifts some axons from the lateral to the intermediate axon bundle. Conversely, forced expression of either Robo2 or Robo3 can shift medial axons into more lateral pathways. It is unclear how the “Robo code” functions. It is still debated whether it is a certain combination of Robo receptors or whether it is the total level of Robo receptors

14

E.T. Stoeckli and V. Niederkofler

that matters (Rajagopalan et al., 2000a; Simpson et al., 2000b). Although there is no direct evidence it is believed that the lateral position of the axon bundles is determined by a Slit gradient read by the Robo receptors (Fig. 1.3). Slit expression and function in this context depends on the presence of the heparan sulfate proteoglycan, Syndecan (Johnson et al., 2004; Steigemann et al., 2004). In contrast to Drosophila, there is direct evidence that vertebrate Slit proteins are involved in lateral positioning of commissural neurons in the murine spinal cord. Specifically, in Slit1;Slit2;Slit3 triple mutant embryos the lateral funiculus is reduced in size compared to wild-type and Robo1-positive fibers are confined to more medial positions (Long et al., 2004). Vertebrate Robo receptors have been shown to interact homophilically with the same Robo receptor and heterophilically with different Robo receptors (Hivert et al., 2002). Direct binding of Robo receptors amongst each other provides an alternative Slit-independent explanation for sorting of axons into distinct mediolateral tracts.

1.3 Open Questions Despite extensive studies and numerous discoveries, the molecular basis of axon guidance and midline crossing in particular is far from resolved. It is clear that axons change their behavior when they have reached the floorplate, the intermediate target. But what is triggering these changes? Growth cones are known to change their complement of surface receptors at this choice point to be ready for the next leg of their journey. How is this achieved? Is there a signal derived from floorplate contact? How are the diverse signals generated by ligand/receptor interactions translated into axonal responses? How are these signals conveyed into cytoskeletal changes? Why are some molecules perceived as growth promoting in some situations and as guidance molecules in others? How is the temporal pattern of axon guidance molecules and their receptors regulated? With all these open questions axon guidance will remain an active and attractive field in the future.

References Altmann CR, Brivanlou AH (2001) Neural patterning in the vertebrate embryo. Int Rev Cytol 203:447–482 Augsburger A, Schuchardt A, Hoskins S, Dodd J, Butler S (1999) BMPs as mediators of roof plate repulsion of commissural neurons. Neuron 24:127–141 Bonkowsky JL, Yoshikawa S, O’Keefe DD, Scully AL, Thomas JB (1999) Axon routing across the midline controlled by the Drosophila Derailed receptor. Nature 402:540–544 Bosley TM, Salih MA, Jen JC, Lin DD, Oystreck D, Abu-Amero KK, MacDonald DB, al Zayed Z, al Dhalaan H, Kansu T, Stigsby B, Baloh RW (2005) Neurologic features of horizontal gaze palsy and progressive scoliosis with mutations in ROBO3. Neurology 64: 1196–1203 Bourikas D, Pekarik V, Baeriswyl T, Grunditz A, Sadhu R, Nardo M, Stoeckli ET (2005) Sonic hedgehog guides commissural axons along the longitudinal axis of the spinal cord. Nat Neurosci 8:297–304

1

Molecular Aspects of Commissural Axon Guidance

15

Bovolenta P, Dodd J (1991) Perturbation of neuronal differentiation and axon guidance in the spinal cord of mouse embryos lacking a floor plate: analysis of Danforth’s short-tail mutation. Development 113:625–639 Brankatschk M, Dickson BJ (2006) Netrins guide Drosophila commissural axons at short range. Nat Neurosci 9:188–194 Brose K, Bland KS, Wang KH, Arnott D, Henzel W, Goodman CS, Tessier-Lavigne M, Kidd T (1999) Slit proteins bind Robo receptors and have an evolutionarily conserved role in repulsive axon guidance. Cell 96:795–806 Burstyn-Cohen T, Tzarfaty V, Frumkin A, Feinstein Y, Stoeckli E, Klar A (1999) F-Spondin is required for accurate pathfinding of commissural axons at the floor plate. Neuron 23: 233–246 Butler SJ, Dodd J (2003) A role for BMP heterodimers in roof plate-mediated repulsion of commissural axons. Neuron 38:389–401 Charron F, Stein E, Jeong J, McMahon AP, Tessier-Lavigne M (2003) The morphogen sonic hedgehog is an axonal chemoattractant that collaborates with netrin-1 in midline axon guidance. Cell 113:11–23 Clarke JD, Holder N, Soffe SR, Storm-Mathisen J (1991) Neuroanatomical and functional analysis of neural tube formation in notochordless Xenopus embryos; laterality of the ventral spinal cord is lost. Development 112:499–516 Dickson BJ (2002) Molecular mechanisms of axon guidance. Science 298:1959–1964 Ding Q, Motoyama J, Gasca S, Mo R, Sasaki H, Rossant J, Hui CC (1998) Diminished Sonic hedgehog signaling and lack of floor plate differentiation in Gli2 mutant mice. Development 125:2533–2543 Garbe DS, Bashaw GJ (2007) Independent functions of Slit-Robo repulsion and NetrinFrazzled attraction regulate axon crossing at the midline in Drosophila. J Neurosci 27: 3584–3592 Gordon-Weeks PR (2004) Microtubules and growth cone function. J Neurobiol 58:70–83 Greenspoon S, Patel CK, Hashmi S, Bernhardt RR, Kuwada JY (1995) The notochord and floor plate guide growth cones in the zebrafish spinal cord. J Neurosci 15:5956–5965 Guan KL, Rao Y (2003) Signalling mechanisms mediating neuronal responses to guidance cues. Nat Rev Neurosci 4:941–956 Hivert B, Liu Z, Chuang CY, Doherty P, Sundaresan V (2002) Robo1 and Robo2 are homophilic binding molecules that promote axonal growth. Mol Cell Neurosci 21:534–545 Huber AB, Kolodkin AL, Ginty DD, Cloutier JF (2003) Signaling at the growth cone: ligandreceptor complexes and the control of axon growth and guidance. Annu Rev Neurosci 26: 509–563 Huminiecki L, Gorn M, Suchting S, Poulsom R, Bicknell R (2002) Magic roundabout is a new member of the roundabout receptor family that is endothelial specific and expressed at sites of active angiogenesis. Genomics 79:547–552 Imondi R, Kaprielian Z (2001) Commissural axon pathfinding on the contralateral side of the floor plate: a role for B-class ephrins in specifying the dorsoventral position of longitudinally projecting commissural axons. Development 128:4859–4871 Imondi R, Wideman C, Kaprielian Z (2000) Complementary expression of transmembrane ephrins and their receptors in the mouse spinal cord: a possible role in constraining the orientation of longitudinally projecting axons. Development 127:1397–1410 Jen JC, Chan WM, Bosley TM, Wan J, Carr JR, Rub U, Shattuck D, Salamon G, Kudo LC, Ou J, Lin DD, Salih MA, Kansu T, Al Dhalaan H, Al Zayed Z, MacDonald DB, Stigsby B, Plaitakis A, Dretakis EK, Gottlob I, Pieh C, Traboulsi EI, Wang Q, Wang L, Andrews C, Yamada K, Demer JL, Karim S, Alger JR, Geschwind DH, Deller T, Sicotte NL, Nelson SF, Baloh RW, Engle EC (2004) Mutations in a human ROBO gene disrupt hindbrain axon pathway crossing and morphogenesis. Science 304:1509–1513 Jessell TM (2000) Neuronal specification in the spinal cord: inductive signals and transcriptional codes. Nat Rev Genet 1:20–29

16

E.T. Stoeckli and V. Niederkofler

Jevince AR, Kadison SR, Pittman AJ, Chien CB, Kaprielian Z (2006) Distribution of EphB receptors and ephrin-B1 in the developing vertebrate spinal cord. J Comp Neurol 497: 734–750 Johnson KG, Ghose A, Epstein E, Lincecum J, O’Connor MB, Van Vactor D (2004) Axonal heparan sulfate proteoglycans regulate the distribution and efficiency of the repellent slit during midline axon guidance. Curr Biol 14:499–504 Kadison SR, Kaprielian Z (2004) Diversity of contralateral commissural projections in the embryonic rodent spinal cord. J Comp Neurol 472:411–422 Kadison SR, Makinen T, Klein R, Henkemeyer M, Kaprielian Z (2006) EphB receptors and ephrin-B3 regulate axon guidance at the ventral midline of the embryonic mouse spinal cord. J Neurosci 26:8909–8914 Kalil K, Dent EW (2005) Touch and go: guidance cues signal to the growth cone cytoskeleton. Curr Opin Neurobiol 15:521–526 Keeble TR, Cooper HM (2006) Ryk: a novel Wnt receptor regulating axon pathfinding. Int J Biochem Cell Biol 38:2011–2017 Keeble TR, Halford MM, Seaman C, Kee N, Macheda M, Anderson RB, Stacker SA, Cooper HM (2006) The Wnt receptor Ryk is required for Wnt5a-mediated axon guidance on the contralateral side of the corpus callosum. J Neurosci 26:5840–5848 Keino-Masu K, Masu M, Hinck L, Leonardo ED, Chan SS, Culotti JG, Tessier-Lavigne M (1996) Deleted in Colorectal Cancer (DCC) encodes a netrin receptor. Cell 87:175–185 Keleman K, Dickson BJ (2001) Short- and long-range repulsion by the Drosophila Unc5 netrin receptor. Neuron 32:605–617 Keleman K, Rajagopalan S, Cleppien D, Teis D, Paiha K, Huber LA, Technau GM, Dickson BJ (2002) Comm sorts robo to control axon guidance at the Drosophila midline. Cell 110: 415–427 Keleman K, Ribeiro C, Dickson BJ (2005) Comm function in commissural axon guidance: cellautonomous sorting of Robo in vivo. Nat Neurosci 8:156–163 Kennedy TE, Serafini T, de la Torre JR, Tessier-Lavigne M (1994) Netrins are diffusible chemotropic factors for commissural axons in the embryonic spinal cord. Cell 78:425–435 Kennedy TE, Wang H, Marshall W, Tessier-Lavigne M (2006) Axon guidance by diffusible chemoattractants: a gradient of netrin protein in the developing spinal cord. J Neurosci 26:8866–8874 Kidd T, Bland KS, Goodman CS (1999) Slit is the midline repellent for the robo receptor in Drosophila. Cell 96:785–794 Kidd T, Brose K, Mitchell KJ, Fetter RD, Tessier-Lavigne M, Goodman CS, Tear G (1998) Roundabout controls axon crossing of the CNS midline and defines a novel subfamily of evolutionarily conserved guidance receptors. Cell 92:205–215 Kolodziej PA, Timpe LC, Mitchell KJ, Fried SR, Goodman CS, Jan LY, Jan YN (1996) Frazzled encodes a Drosophila member of the DCC immunoglobulin subfamily and is required for CNS and motor axon guidance. Cell 87:197–204 Lee KJ, Jessell TM (1999) The specification of dorsal cell fates in the vertebrate central nervous system. Annu Rev Neurosci 22:261–294 Lesnick TG, Papapetropoulos S, Mash DC, Ffrench-Mullen J, Shehadeh L, de Andrade M, Henley JR, Rocca WA, Ahlskog JE, Maraganore DM (2007) A genomic pathway approach to a complex disease: axon guidance and parkinson disease. PLoS Genet 3:e98 Liu Y, Shi J, Lu CC, Wang ZB, Lyuksyutova AI, Song XJ, Zou Y (2005) Ryk-mediated Wnt repulsion regulates posterior-directed growth of corticospinal tract. Nat Neurosci 8:1151–1159 Long H, Sabatier C, Ma L, Plump A, Yuan W, Ornitz DM, Tamada A, Murakami F, Goodman CS, Tessier-Lavigne M (2004) Conserved roles for Slit and Robo proteins in midline commissural axon guidance. Neuron 42:213–223 Lyuksyutova AI, Lu CC, Milanesio N, King LA, Guo N, Wang Y, Nathans J, Tessier-Lavigne M, Zou Y (2003) Anterior-posterior guidance of commissural axons by Wnt-frizzled signaling. Science 302:1984–1988

1

Molecular Aspects of Commissural Axon Guidance

17

Mambetisaeva ET, Andrews W, Camurri L, Annan A, Sundaresan V (2005) Robo family of proteins exhibit differential expression in mouse spinal cord and Robo-Slit interaction is required for midline crossing in vertebrate spinal cord. Dev Dyn 233:41–51 Matise MP, Epstein DJ, Park HL, Platt KA, Joyner AL (1998) Gli2 is required for induction of floor plate and adjacent cells, but not most ventral neurons in the mouse central nervous system. Development 125:2759–2770 Matise MP, Lustig M, Sakurai T, Grumet M, Joyner AL (1999) Ventral midline cells are required for the local control of commissural axon guidance in the mouse spinal cord. Development 126:3649–3659 Mauti O, Sadhu R, Gemayel J, Gesemann M, Stoeckli ET (2006) Expression patterns of plexins and neuropilins are consistent with cooperative and separate functions during neural development. BMC Dev Biol 6:32 Nakagawa S, Brennan C, Johnson KG, Shewan D, Harris WA, Holt CE (2000) Ephrin-B regulates the Ipsilateral routing of retinal axons at the optic chiasm. Neuron 25:599–610 Nissen UV, Mochida H, Glover JC (2005) Development of projection-specific interneurons and projection neurons in the embryonic mouse and rat spinal cord. J Comp Neurol 483:30–47 Okada A, Charron F, Morin S, Shin DS, Wong K, Fabre PJ, Tessier-Lavigne M, McConnell SK (2006) Boc is a receptor for sonic hedgehog in the guidance of commissural axons. Nature 444:369–373 Patapoutian A, Reichardt LF (2000) Roles of Wnt proteins in neural development and maintenance. Curr Opin Neurobiol 10:392–399 Rajagopalan S, Vivancos V, Nicolas E, Dickson BJ (2000a) Selecting a longitudinal pathway: Robo receptors specify the lateral position of axons in the Drosophila CNS. Cell 103:1033–1045 Rajagopalan S, Nicolas E, Vivancos V, Berger J, Dickson BJ (2000b) Crossing the midline: roles and regulation of Robo receptors. Neuron 28:767–777 Sabatier C, Plump AS, Le M, Brose K, Tamada A, Murakami F, Lee EY, Tessier-Lavigne M (2004) The divergent Robo family protein rig-1/Robo3 is a negative regulator of slit responsiveness required for midline crossing by commissural axons. Cell 117:157–169 Salie R, Niederkofler V, Arber S (2005) Patterning molecules; multitasking in the nervous system. Neuron 45:189–192 Schmitt AM, Shi J, Wolf AM, Lu CC, King LA, Zou Y (2006) Wnt-Ryk signalling mediates medial-lateral retinotectal topographic mapping. Nature 439:31–37 Seeger M, Tear G, Ferres-Marco D, Goodman CS (1993) Mutations affecting growth cone guidance in Drosophila: genes necessary for guidance toward or away from the midline. Neuron 10:409–426 Serafini T, Colamarino SA, Leonardo ED, Wang H, Beddington R, Skarnes WC, Tessier-Lavigne M (1996) Netrin-1 is required for commissural axon guidance in the developing vertebrate nervous system. Cell 87:1001–1014 Shirasaki R, Katsumata R, Murakami F (1998) Change in chemoattractant responsiveness of developing axons at an intermediate target. Science 279:105–107 Simpson JH, Kidd T, Bland KS, Goodman CS (2000a) Short-range and long-range guidance by slit and its Robo receptors. Robo and Robo2 play distinct roles in midline guidance. Neuron 28:753–766 Simpson JH, Bland KS, Fetter RD, Goodman CS (2000b) Short-range and long-range guidance by Slit and its Robo receptors: a combinatorial code of Robo receptors controls lateral position. Cell 103:1019–1032 Smith SJ (1988) Neuronal cytomechanics: the actin-based motility of growth cones. Science 242:708–715 Steigemann P, Molitor A, Fellert S, Jackle H, Vorbruggen G (2004) Heparan sulfate proteoglycan syndecan promotes axonal and myotube guidance by slit/robo signaling. Curr Biol 14:225–230 Stein E, Tessier-Lavigne M (2001) Hierarchical organization of guidance receptors: silencing of netrin attraction by slit through a Robo/DCC receptor complex. Science 291:1928–1938 Stoeckli ET (2006) Longitudinal axon guidance. Curr Opin Neurobiol 16:35–39

18

E.T. Stoeckli and V. Niederkofler

Stoeckli ET, Landmesser LT (1995) Axonin-1, Nr-CAM, and Ng-CAM play different roles in the in vivo guidance of chick commissural neurons. Neuron 14:1165–1179 Stoeckli ET, Landmesser LT (1998) Axon guidance at choice points. Curr Opin Neurobiol 8:73–79 Stoeckli ET, Sonderegger P, Pollerberg GE, Landmesser LT (1997) Interference with axonin-1 and NrCAM interactions unmasks a floor-plate activity inhibitory for commissural axons. Neuron 18:209–221 Tear G, Harris R, Sutaria S, Kilomanski K, Goodman CS, Seeger MA (1996) commissureless controls growth cone guidance across the CNS midline in Drosophila and encodes a novel membrane protein. Neuron 16:501–514 Tessier-Lavigne M, Goodman CS (1996) The molecular biology of axon guidance. Science 274:1123–1133 van Straaten HW, Hekking JW (1991) Development of floor plate, neurons and axonal outgrowth pattern in the early spinal cord of the notochord-deficient chick embryo. Anat Embryol (Berl) 184:55–63 Wen Z, Guirland C, Ming GL, Zheng JQ (2004) A CaMKII/calcineurin switch controls the direction of Ca(2+)-dependent growth cone guidance. Neuron 43:835–846 Williams SE, Mann F, Erskine L, Sakurai T, Wei S, Rossi DJ, Gale NW, Holt CE, Mason CA, Henkemeyer M (2003) Ephrin-B2 and EphB1 mediate retinal axon divergence at the optic chiasm. Neuron 39:919–935 Yaron A, Zheng B (2007) Navigating their way to the clinic: emerging roles for axon guidance molecules in neurological disorders and injury. Dev Neurobiol 67:1216–1231 Yoshikawa S, McKinnon RD, Kokel M, Thomas JB (2003) Wnt-mediated axon guidance via the Drosophila Derailed receptor. Nature 422:583–588 Yu TW, Bargmann CI (2001) Dynamic regulation of axon guidance. Nat Neurosci 4(Suppl): 1169–1176 Yuan SS, Cox LA, Dasika GK, Lee EY (1999) Cloning and functional studies of a novel gene aberrantly expressed in RB-deficient embryos. Dev Biol 207:62–75 Zhou FQ, Cohan CS (2004) How actin filaments and microtubules steer growth cones to their targets. J Neurobiol 58:84–91 Zou Y, Stoeckli E, Chen H, Tessier-Lavigne M (2000) Squeezing axons out of the gray matter: a role for slit and semaphorin proteins from midline and ventral spinal cord. Cell 102:363–375

Chapter 2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits Zoltán Molnár, Wei Zhi Wang, Maria Carmen Piñon, Daniel Blakey, Shinichi Kondo, Franziska Oeschger, and Anna Hoerder-Suabedissen

The billions of cells and trillions of connections of the human brain are generated from the complex interactions between a developmental genetic programme and the environment. The resulting embryo is the ultimate readout of our genome; a combination of genetic susceptibility and environmental perturbations can lead to several devastating neurological and psychiatric conditions. The cerebral cortex constitutes half the volume of the human brain and is presumed to be responsible for the neuronal computations underlying such complex phenomena as perception, thought, language, attention, episodic memory and voluntary movements. These functions rely on elaborate cortical circuits, which are assembled during embryonic and early postnatal development. Understanding cortical circuit formation is a fundamental research underpinning all aspects of complex mammalian behaviour. Building the brain is like erecting a huge complex tower or bridge. The construction requires a dynamic scaffold, which is built and modified along with the final permanent structure. After the completion of the construction the scaffold must be dismantled at precisely the right places and time. The nervous system is particularly vulnerable at these stages. This is indicated by the high prevalence of cerebral cortical developmental disorders in the general population [schizophrenia (1:100); autism (1:166); attention deficit hyperactivity disorder (1:30); dyslexia (1:10); childhood epilepsy (1:200)]. In spite of recent progress, we are only beginning to understand basic neural developmental mechanisms and their involvement in the pathomechanisms of several debilitating diseases. In order to gain a more comprehensive understanding of the brain as a final product, we must thoroughly characterise neurodevelopmental processes contributing to its formation. In doing so, we will ascend to a platform from which we can critically analyse and perhaps treat the numerous disorders affecting the developing and mature brain. Early cortical circuit formation in all regions of the mammalian cerebral cortex have a largely uniform structure with characteristic six layers. However, in the adult they show variations in cell numbers, the proportion of these layers and the cortical Z. Molnár (B) Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford, UK e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_2, 

19

20

Z. Molnár et al.

sheet can be divided into numerous well-defined areas (Brodmann, 1909). The slight differences in cytoarchitecture, connectivity and physiological properties reflect differences in the cortical circuits adopted for the performance of various computational functions. These differences emerge during embryonic and early postnatal life from circuits, which are drastically different from the adult. They involve transient neurons and they are substantially remodelled during perinatal stages. This remodelling is dependent on signals originating from the sensory periphery and from the immature brain itself (Katz and Shatz, 1996). An important aspect of developmental neurobiology is to understand the contributions from the environment on the unfolding genetic programme (see Rubenstein and Rakic, 1999; Shimogori and Grove, 2005; Price et al., 2006; O’Leary et al., 2007, Walsh 2007). Thalamic projections arrive at the developing cerebral cortex at the peak of cortical neurogenesis while cell migration is ongoing and before the cortical cells differentiate or form their connectivity (Rakic 1976; Shatz and Luskin 1986); therefore, they are in a position to influence the area-specific development of the cerebral cortex (Dehay and Kennedy, 2007). The largely transient subplate neurons (SP) play an important role in thalamocortical axon pathfinding at the level of the initial areal targeting (Catalano and Shatz, 1998; Ghosh et al., 1990; Molnár and Blakemore, 1995; López-Bendito and Molnár, 2003) as well as the eventual innervation of cortical layer 4 by thalamic afferents and establishment of optical orientation columns (Kanold et al., 2003). They are also necessary for the maturation of inhibition in cortical layer 4 in areas innervated by the thalamus (Kanold and Shatz, 2006), and drive oscillations in the gap junction coupled early cortical syncytium (Dupont et al., 2006). Subplate cells were first described in primates (Kostovic and Rakic, 1980) and carnivores (Luskin and Shatz, 1985) as a substantial transient layer of cells below layer 6. In rodents, subplate is a thin band of cells separating the white matter from layer 6. Murine subplate (SP) cells, born around embryonic day (E)11, are among the earliest mature cortical neurones (Price et al., 1997), and begin extending axons towards the thalamus by E13 (DeCarlos and O’Leary, 1992; Molnár et al., 1998a,b). During development, SP neurons are electrically active and capable of firing action potentials (Hanganu et al., 2001) while incorporating (at least transiently) into the cortical and subcortical circuitry (Kanold et al., 2003; Friauf and Shatz, 1991; Higashi et al., 2002, 2005; Hoerder et al., 2006). The example for the subplate neuron in Fig. 2.1 only possesses intracortical projection, however, it is well known that subplate also develops subcortical projections (Allendoerfer and Shatz, 1994). Both electrophysiological properties and cell morphology point to a high degree of underlying diversity among subplate neurons (Hanganu et al., 2001; Antonini and Shatz, 1990; Hoerder, 2007). The diversity of subplate cells is further underscored by the heterogeneity of molecular markers of glutamatergic or GABAergic cells expressed in cells of the subplate (Allendoerfer and Shatz, 1994; Hevner and Zecevic, 2006). It is not yet clear whether the same types of subplate neurons possess intracortical or extracortical projections and how subplate neurons with different somatodendritic morphology relate to the diverse neurochemical properties and physiological fingerprints (Fig. 2.2). The developing corticofugal circuit closer to the thalamus establishes an equally dynamic and puzzling transient circuit. It has been proposed that corticofugal

2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits

21

Fig. 2.1 Schematic summary of the early development of cortical lamination (upper row) with special attention to subplate, marginal zone and thalamocortical afferents (lower row) from E12 until P6 in the mouse. The blue, round cells at the bottom of the panel represent neural progenitors in ventricular and subventricular zones (VZ, SVZ) as well as radial glia along which immature neurons migrate (black bipolar cells). At E12 the earliest generated neurons form the preplate (PP). By E15 the cortical plate (CP) splits the preplate into subplate (SP) and marginal zone (MZ). By P0 the neurogenesis is completed, the cortex is comprised of infragranular layers (IGL, layers 5 and 6) and still migrating future cortical layers 2–4, the dense cortical plate (DCP). The lower, IGL cells are born at an earlier stage than the upper layers (4–2). By postnatal day 6 most neurons complete their migration and the cortical lamination is established. Thalamocortical axons (thick black lines in lower panels) arrive to the subplate around E15. Thalamocortical projections enter the cortex by birth and begin to establish characteristic periphery-related patterning in the barrel field of S1 by P3-6. Thalamocortical axons establish functional interactions in subplate as they arrive and accumulate in this layer before they enter the cortex and recognise their ultimate target cells in layer 4. Subplate neurons integrate into the overlying cortex and also develop various extracortical projections providing a stable platform for the establishment of cortical circuitry from early stages

projections accumulate outside the thalamus before they innervate their final target cells (Shatz and Rakic, 1981; Allendoerfer and Shatz, 1994; Molnár and Cordery, 1999, Price et al., 2006). The sequence of their development is not well understood in spite of their obvious links to cross-modal plasticity, early brain damage and some cognitive disorders. The better understanding of subplate molecular taxonomy and reporter gene expressing transgenic lines opened up new opportunities to readdress several of these issues (Table 2.1). Recently, our group identified several markers

22

Z. Molnár et al.

Fig. 2.2 Proposed transient minicircuit formed between thalamocortical axons and subplate and layer 4 neurons during cortical development. Thalamocortical axons (TCAs) navigate to the cortex and by E16 in mice project into the subplate (SP) before extending radial branches into the cortical plate from around E17 (left schematic panel of a coronal section). During this period TCAs form synaptic connections with SP cells (synapse #1) which they retain during their initial contact with the main target cells in layer 4 (L4, #2). The minicircuit is completed by SP axons also terminating on layer 4 neurons (#3). Some models of this interaction also include inhibitory interneurons (IN) that receive input from both TCAs and SP (#4 and# 5 respectively) and they themselves synapse with layer 4 excitatory neurons (#6). Activity in this circuit acts as a coincidence detecting mechanism and may act to guide and stabilise TCA – layer 4 connections. After Arber (2004)

for subplate (Hoerder-Suabedissen et al., 2009; Wang et al., 2009), which provided a unique opportunity to selectively monitor and modulate these cells using genetic approaches. The genetic models shall allow us to label a subgroup of these neurons with all their projections and follow their dynamic integration into local and long-range circuits, and to selectively control their synaptic input and output characteristics during development.

2.1 Clinical Importance of the Understanding of Early Cortical Circuits The subplate layer is the foundation of the developing brain, and disruption of these cells may be the source of many developmental flaws, and therefore a fundamental topic to study (Kostovic and Rakic, 1990; Allendoerfer and Shatz, 1994; Kanold, 2004). Cerebral cortical developmental disorders (schizophrenia, autism, attention deficit/hyperactivity disorder, dyslexia) and perinatal hypoxic injuries such

2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits

23

Table 2.1 Questions surrounding early cortical circuit formation When do the early subplate projections enter into the thalamus? 1. Do they wait outside the LGN (Shatz and Rakic, 1981)? 2. Do they wait in TRN (Molnár and Cordery, 1999)? 3. Do they wait in white matter or internal capsule (Clasca et al., 1995)? 4. Do subplate projections enter the dorsal thalamus (Allendoerfer and Shatz, 1994)? Which cortical cells develop projections to the thalamus first? 1. Subplate cells (McConnell et al., 1989)? 2. Layer 5 cells (Clasca et al., 1995)? 3. Layer 6 cells? What is the relationship between early corticofugal and thalamic projections? 1. Fasciculate with each other in IC and IZ (Molnar and Blakemore, 1995)? 2. Run in separate compartments (Miller et al., 1993, 1995)? 3. Interdigitate in a restricted portion (Bicknese et al., 1994)? What is the mode of integration of subplate neurites into the cortical plate prior, during and after thalamic innervation? 1. Ocular dominance formation (Ghosh and Shatz, 1990)? 2. Orientation column formation (Kanold et al., 2003)? 3. Barrel formation (Jethwa et al., 2007; Pinon et al., 2009)? 4. Area-specificity (McConnell et al., 1989; Catalano and Shatz, 1998; Molnár and Blakemore 1995; Shimogori and Grove, 2005)? 5. Axonal and/or dendritic remodelling associated with thalamocortical ingrowth and periphery related patterning (Hoerder, 2007; Jethwa et al., 2007; Pinon et al., 2009)

as periventricular leucomalacia (PVL) involve cells of the subplate (McQuillen et al., 2003; Volpe, 2001). Some of these pathologies can be revealed by in utero imaging in human following the advent of novel image processing algorithms (Rutherford, 2002). We believe that it is impossible to understand pathological circuit formation without comprehending the transient structures contributing to its development, just as it is not possible to understand how a complex building or a bridge was constructed without the detailed knowledge of the actual scaffolding utilised during the time of the building. Recent molecular markers provide handles to monitor and modulate very specific classes of neurons. Developmental neurobiologists have a long-standing interest in understanding cerebral cortical circuit formation and function, and investigate them using anatomical, genetic, physiological and molecular approaches (Thomson and Bannister, 2003; Nelson et al., 2006). Developments in cell separation and gene expression analysis enabled the field to identify molecular tags for the multitude of neuronal subtypes, which can be used as molecular handles to identify, modulate or eliminate very specific classes of neurons. Consequently, we are gaining greater insight into the regulation of the process of cortical neurogenesis and the classification of laminar specific sub-classes of cells (Markram, 2004; Nelson et al., 2006; Molyneaux et al., 2007; Molnár and Cheung 2006). These provide unparalleled opportunities to exploit genetics to monitor and modulate a selected group of neurons during development or adulthood (Luo et al., 2008; Miyoshi and Fishell, 2006).

24

Z. Molnár et al.

2.2 Recent Microarray Screen of Murine Subplate Neurons Most previous studies on the function of subplate cells have used carnivores and primates (Kostovic and Rakic, 1990; Allendoerfer and Shatz, 1994), which have a prominent subplate, but remain relatively unexplored as genetic models. Improving our understanding of the role of subplate cells in mice is thus particularly important, as they are increasingly used as a model organism for developmental neurobiology. We have recently identified novel subplate cell-specific gene expression by using a microarray-based approach (Hoerder, 2007; Hoerder-Suabedissen et al., 2009; Wang et al., 2009). By comparing gene expression between the subplate and the adjacent layer 6 we have identified moxd1, cplx3, ddc and tmem163 and confirmed ctgf and nurr1 (Heuer et al., 2003; Arimatsu et al., 2003) as highly expressed in the subplate but not in layer 6. MoxD1 and cplx3 gene expression is exclusive to the subplate in the P8 mouse cortex as assessed by in situ hybridisation, while tmem163 expression was also detected in layer 5 cells. DDC protein is localised to cells in the white matter, subplate and lower layer 6. The molecular data is currently being integrated with cellular and anatomical data. So far, co-localisation of Cplx3 and DDC with the known subplate marker Nurr1 as well as GABA to identify interneurons was assessed. Neither Cplx3 nor DDC expression co-localised with GABA. DDC did not co-localise with the known subplate marker Nurr1, but Cplx3 expression is present in approximately 50% of Nurr1+ glutamatergic projection neurons suggesting several subpopulations within the subplate. We have recently extended the identification of these markers by performing screens at additional developmental ages and adult. The identification of subplate-specific molecular markers will facilitate the characterisation of neurochemical, morphological and functional properties of these neurons.

2.3 Recent Studies of Subplate Neuron Integration into the Cortical and Extracortical Circuitry in Reporter Gene Expressing Mouse Models Due to the lack of sensitive purely anterograde tracers and due to our inability to selectively label subplate neurons and their neurites, the exact timing and pattern of subcortical subplate projections is still controversial, although numerous functional suggestions have been based on their intimate association with thalamocortical projections and integration into layer 4 cortical circuitry. Moreover, the nature and the pattern of innervation of thalamic nuclei by various cortical projection neurons during development have not been resolved (Table 2.1). Recently, two transgenic mouse models were created with GFP expression restricted to a subset of SP and layer 6 neurons [Golli-tau-eGFP (GTE) mouse; Jacobs et al., 2007; Tbr1-driven GFP; Kolk et al., 2005]. Although neither of these two models is completely subplate specific, they both already help to answer some questions surrounding the integration of the subplate neurons into the intra- and extra-cortical circuitry (Molnár et al., 2007). Our studies on the GTE mouse, in collaboration with Campagnoni and Jacobs

2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits

25

(UCLA), suggest delayed and differential innervation of different nuclei of the dorsal thalamus by early GFP-positive corticofugal projections (Jacobs et al., 2007; Piñon et al., 2005; Piñon et al., 2009) and intimate association of early corticofugal projections with thalamic afferents (Piñon et al., 2005). At birth, GFP-positive neurites extending from the lower cortical layers into the overlying cortex are evenly distributed throughout the putative barrel field (Fig. 2.3). At P4, eGFP+ neurites aggregate within intra-barrels, a pattern becoming more defined by P6. By P10 the intra-barrel aggregation inverts to an inter-barrel pattern and remains so until P14 (Fig. 2.3).

Fig. 2.3 Fluorescence micrographs of the barrel cortex in the coronal plane of the GTE mouse at ages E (embryonic day) 17 – P (postnatal day) 14. GFP neurites are shown in green and the bisbenzimide nuclear counterstain are shown in blue. Filled arrowheads indicate intra-barrel GFP densities; arrows indicate barrel septa; outlined arrowheads indicate inter-barrel GFP concentration. Abbreviations: mz marginal zone; cp, cortical plate; sp, subplate; wm, white matter; IV, layer IV; hp, hippocampus. Scale bar = 100 μm (a–c) and 200 μm (d–h) (Reproduced from Piñon et al., 2009)

2.4 Manipulation of the Sensory Periphery Alters Subplate Integration into the Barrel Field Optical recording and current source density (CSD) analysis both demonstrated immature synapses between subplate and thalamocortical projections in both mouse and rat (Molnár et al., 2003; Higashi et al., 2002; 2005) as it was originally described in carnivores (Friauf et al., 1990; Kanold, 2004). The nature of these synaptic contacts is not fully understood, but it is likely that the combination of glutamatergic (mostly mediated through NMDA receptors) and depolarizing

26

Z. Molnár et al.

GABAergic inputs plays a central role as in the development of other excitatory cortical synapses (Akerman and Cline, 2006; Iwasato et al., 2000). We examined whether changing the flow of sensory information from the whiskers can alter the transition from initial intra-barrel aggregation of GFP-positive neurites into inter-barrel pattern around P10. We found that the pattern inversion can be delayed until at least P10 by whisker removal at P0, suggesting that sensory input is modulating this rearrangement (Jethwa et al., 2007; Piñon et al., 2009). These experiments open up various possibilities to define cortical circuit formation by monitoring subplate neurite integration. Various paradigms (enucleation, infraorbital nerve cut, cross-modal plasticity, critical periods) become accessible with the use of reporter gene expressing lines from GENSAT (CTGF-GFP, Edg2-GFP, DDC-GFP in addition to the Golli-tau-eGFP). In these transgenic lines different populations of subplate neurons express the reporter gene; therefore, studying each line will give detailed information of that particular subplate population (e.g. Golli-tau-eGFP is expressed in broader population than CTGF-GFP, Aye et al., 2006, Hoerder-Suabedissen et al., 2010).

2.5 Subplate Cell Populations in Mutants with Cortical Migration Defects Our new subplate markers (Hoerder-Suabedissen et al., 2009) opened up possibilities for further identification of subplate subpopulations in comparative and neuropathological studies. We have already tested six markers in the Reeler and p35KO mice, which have known defects in SP positioning (Hoerder-Suabedissen, 2009). MoxD1, cplx3, ctgf, nurr1 and tmem163 gene expression shifted to the outermost layer in Reeler and a thin band of cells in the middle of the p35KO cortex. DDC+ cells on the other hand remained mostly in the white matter and lower half of cortex with few cells found in the outer layer in Reeler brains (Hoerder-Suabedissen et al., 2009). These findings confirm that the newly identified genes are specifically expressed in SP cells even in their altered position. Subplate has been implicated in the formation of numerous developmental malformations of the cerebral cortex such as epilepsy, schizophrenia and cerebral palsy (Eastwood and Harrison 2003; Volpe 2001). Clinical observations in preterm human suggest that neonatal hypoxia-ischemia causes selective injury in the subcortical white matter involving subplate cells, resulting in periventricular leucomalacia. The selective vulnerability of subplate cells is not understood. It is also not clear what is the causal relationship between the cellular pathologies and cognitive disorders. The combination of molecular genetics with anatomical approaches to selectively monitor and modulate subplate neurons during development could help in the understanding of causal relationships. We are currently utilising our subplate markers to investigate neuropathological changes of the subplate in animal models (hypothyroid and perinatal hypoxia-ischemia (HI) in rat) and in human histopathology (Oeschger et al., 2010).

2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits

27

In a neonatal rat model of HI, selective vulnerability of the subplate has been suggested using BrdU birthdating methods (McQuillen et al., 2003). We hypothesised that certain subplate subpopulations could be more susceptible than others and investigated the above subplate markers in a similar yet slightly milder HI model (Oeschger et al., 2008). After confirming the majority of the murine subplate markers in the postnatal rat, Franziska Oeschger started a study in collaboration with Vanessa Ginet and Anita Truttmann, University of Lausanne. To study hypoxiaischemia, 2-day old male rat pups underwent permanent occlusion of the right common carotid artery followed by a period of hypoxia (6% O2 , 1.5 h or 2 h) and were analyzed 6 days later (Oeschger et al., 2010). We found some evidence of differential changes in the expression of individual subplate markers following HI. Most prominently, the number of Nurr1+ cells was increased in some mildly affected brains but decreased in the more severe cases while Cplx3+ subplate population was more generally decreased (Oeschger et al., 2010). In parallel, hypothyroid rat models are currently being examined in collaboration with Pere Berbel (Alicante) and with the help of G Slaters (FHS student). In this model, pregnant rats undergo thyroidectomy at E17 mimicking the situation in early preterm infants, which have to rely entirely on their own hormone production, and thus presenting a highly important clinical problem (Berbel et al., 2007)). Our preliminary data on P8 rats indicate that the subplate layer is enlarged and we observed an increase of 20% in the Nurr1+ population. We are currently extending our study to other subplate subpopulations (Cplx3+, CTGF+) and other cortical layers. Subplate abnormalities have been described in several neuropathological disorders including schizophrenia, autism and periventricular leukomalacia (Eastwood and Harrison, 2003; McQuillen and Ferriero, 2005). We started to utilise our new subplate markers for human histopathological studies. Antibodies against our markers for murine subplate have been tested on human pathological specimen (CTFG, DDC, Nurr1), others are still being optimised for human tissue. For some of the markers, human in situ hybridisation probes have been generated and tested (Wang et al., 2010).

2.6 Summary Our work in understanding the involvement of subplate cells in the establishment of earliest cortical circuits is benefiting from the recent advances of molecular and genetic approaches. However, there is a lot to be accomplished. The great challenge is now to understand the combinatorial effect of lineage- and area-specific gene expression profiles during area-specific cortical circuit formation. The discovery of molecular markers will allow distinct subclassification of cells and functional dissection of early cortical circuits. Acknowledgement The laboratory of ZM was supported from grants from the MRC, Wellcome Trust, EU and Human Frontiers Science Program. We thank the Wellcome Trust Initiative in Integrative Physiology of Ion Channels (OXION) for the help with the microarray work.

28

Z. Molnár et al.

References Akerman CJ, Cline HT (2006) Depolarizing GABAergic conductances regulate the balance of excitation to inhibition in the developing retinotectal circuit in vivo. J Neurosci 26(19): 5117–5130 Allendoerfer KL, Shatz CJ (1994) The subplate, a transient neocortical structure: its role in the development of connections between thalamus and cortex. Annu Rev Neurosci 17:185–218 Antonini A, Shatz CJ (1990) Relation between putative transmitter phenotypes and connectivity of subplate neurons during cerebral cortical development. Eur J Neurosci 2(9):744–761 Arber S (2004) Subplate neurons: bridging the gap to function in the cortex. TINS 27(3):111–113 Arimatsu Y, Ishida M, Kaneko T, Ichinose S, Omori A (2003) Organization and development of corticocortical associative neurons expressing the orphan nuclear receptor Nurr1. J Comp Neurol 466(2):180–196 Aye L, Piñon MC, Horder A, Jacobs E, Campagnoni A, Molnár Z (2006) Properties of layer VIa and subplate neurons in the Golli tau eGFP (GTE) mouse. FENS Abstract for the Vienna Meeting Berbel P, Obregón MJ, Bernal J, Escobar del Rey F, Morreale de Escobar G (2007) Iodine supplementation during pregnancy: a public health challenge. Trends Endocrinol Metab 18(9):338–343. Brodmann K (1909) Vergleichende Localisationslehre der Grosshirnride in ihren Prinzipien dargestellt auf Grund des Zellenbaues. J.A. Barth, Leipzig Bicknese AR, Sheppard AM, O’Leary DD, Pearlman AL (1994) Thalamocortical axons extend along a chondroitin sulfate proteoglycan-enriched pathway coinc. with the neocort. Subplate and distinct from the eff. path. J Neurosci 14:3500–3510 Blakey D (2007) The role of neural activity in the level of thalamocortical connections. DPhil Thesis, University of Oxford, Oxford Catalano S, Shatz CJ (1998) Activity-dependent cortical target selection by thalamic axons. Science 281:559–562 Clasca F, Angelucci A, Sur M (1995) Layer-specific progr of devel in neocortical proj neur. PNAS USA 92:11145–11149 DeCarlos J, O’Leary D (1992) Growth and targeting of subplate ax. & est. of major cort pathways. J Neurosci 12:1194–1211 Dehay C, Kennedy H (2007) Cell-cycle control and cortical development. Nat Rev Neurosci 8(6):438–450 Dupont E, Hanganu IL, Kilb W, Hirsch S, Luhmann HJ (2006) Rapid developmental switch in the mechanisms driving early cortical columnar networks. Nature 439(7072):79–83 Eastwood SL, Harrison PJ (2003) Interstitial white matter neurons express less reelin and are abnormally distributed in schizophrenia: towards an integr of mol and morph asp of the neurodev hypothesis. Mol Psychiatry 8(9):769, 821–831 Friauf E, McConnell SK, Shatz CJ (1990 Aug) Functional synaptic circuits in the subplate during fetal and early postnatal development of cat visual cortex. J Neurosci 10(8):2601–2613 Friauf E, Shatz CJ (1991) Chan pat of syn input to subplate and cort plate dur dev of vis ctx. J Neurophys 66:2059–2071 Ghosh A, Antonini A, McConnell SK, Shatz CJ (1990) Requir for subplate neu in the form. of tc con. Nature 347:179–181 Hanganu IL, Kilb W, Luhmann HJ (2001) Spontaneous synaptic activity of subplate neurons in neonatal rat somatosensory cortex. Cereb Cortex 11(5):400–410 Heuer H, Christ S, Friedrichsen S et al (2003) Connective tissue growth factor: a novel marker of layer VII neurons in the rat cerebral cortex. Neuroscience 119:43–52 Hevner RF, Zecevic N (2006) Pioneer neurons and interneurons in the developing subplate: molecular markers, cell birthdays, and neurotransmitters. In: Erzurumlu RS, Guido W, Molnár Z (eds) Development and plasticity in sensory thalamus and cortex. Springer, New York, pp 1–18

2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits

29

Higashi S, Hioki K, Kurotani T, Molnár Z (2005) Functional thalamocortical synapse reorganization from subplate to layer IV during postnatal development in the Shaking rat Kawasaki: an opt rec study. J Neurosci 25(6):1395–1406 Higashi S, Molnár Z, Kurotani T, Inokawa H, Toyama K (2002) Functional thalamocortical connections develop during embryonic period in the rat: an optical recording study. Neuroscience 115:1231–1246 Hoerder A (2007) Mouse cortical subplate neurones: Mol. markers, connectivity and devel. DPhil Thesis, University of Oxford, Oxford Hoerder A, Paulsen O, Molnár Z (2006) Developmental changes in the dendritic morphology of subplate cells with known projections in the mouse cortex. FENS Abstract Hoerder-Suabedissen A, Wang WZ, Lee S, Davies KE, Goffinet AM, Raki´c S, Parnavelas J, Reim K, Nicoli´c M, Paulsen O, Molnár Z (2009 Aug) Novel markers reveal subpopulations of subplate neurons in the murine cerebral cortex. Cereb Cortex 19(8):1738–1750. E-pub 2008 Nov 13 Iwasato T, Datwani A, Wolf AM, Nishiyama H, Taguchi Y, Tonegawa S, Knöpfel T, Erzurumlu RS, Itohara S (2000) Cortex-restricted disruption of NMDAR1 impairs neuronal patterns in the barrel cortex. Nature 406(6797):726–731 Jacobs EC, Campagnoni C, Kampf K, Reyes SD, Kalra V, Handley V, Xie YY, Hong-Hu Y, Spreur V, Fisher RS, Campagnoni AT (2007 Jan). Visualization of corticofugal projections during early cortical development in a tau-GFP-transgenic mouse. Eur J Neurosci 25(1):17–30 Jethwa A, Piñon MC, Jacobs E, Campagnoni A, Molnár Z (2007) Dynamic integration of subplate neurons into the cortical barrel field circuitry during postnatal development in the Golli-taueGFP (GTE) mouse. BNA Abstr 19:49 Kanold PO (2004 Oct 5) Transient microcircuits formed by subplate neurons and their role in functional development of thalamocortical connections. Neuroreport 15(14):2149–2153. Review Kanold PO, Kara P, Reid RC, Shatz CJ (2003) Role of subplate neur in funct matur of vis cort columns. Science 301:521–525 Kanold PO, Shatz CJ (2006) Subplate neurons regulate maturation of cortical inhibition and outcome of ocular dominance plasticity. Neuron 51(5):627–638 Katz LC, Shatz CJ (1996) Synaptic activity and the construction of cortical circuits. Science 274(5290):1133–1138 Kolk SM, Whitman MC, Yun ME, Shete P, Donoghue MJ (2005) A unique subpopulation of Tbr1-expressing deep layer neurons in the developing cerebral cortex. Mol Cell Neurosci 32(1–2):200–214 Kostovic I, Rakic P (1980) Cytology and time of origin of interstitial neurons in the white matter in infant and adult human and monkey telencephalon. J Neurocytol 9:219–242 Kostovic I, Rakic P (1990) Developmental history of the transient subplate zone in the visual and somatosensory cortex of the macaque monkey and human brain. J Comp Neurol 297(3): 441–470 Little GE, López-Bendito G, Rünker AE, García N, Piñon MC, Chédotal A, Molnár Z, Mitchell KJ (2009 Apr 28) Specificity and plasticity of thalamocortical connections in Sema6A mutant mice. PLoS Biol 7(4):e98 López-Bendito G, Molnár Z (2003) Thalamocortical development. Nat Rev Neurosci 4:276–289 Luo L, Callaway EM, Svoboda K (2008) Genetic dissection of neural circuits. Neuron 57(5): 634–660 Luskin M, Shatz CJ (1985) Neurogenesis of the cat’s primary visual cortex. J Comp Neurol 242:611–631 Markram H (2004) Correlation maps allow neuronal electrical properties to be predicted from single-cell gene expression profiles in rat neocortex. Cereb Cortex 14:1310–1327 McConnell SK, Ghosh A, Shatz CJ (1989) Subplate neurons pioneer the first axon pathway from the ctx. Science 245:978–982

30

Z. Molnár et al.

McQuillen PS, Sheldon RA, Shatz CJ, Ferriero DM (2003 Apr 15) Selective vulnerability of subplate neurons after early neonatal hypoxia-ischemia. J Neurosci 23(8):3308–3315 McQuillen PS, Ferriero DM (2005) Perinatal subplate neuron injury: implications for cortical development and plasticity. Brain Pathol 15:250–260 Miller B, Sheppard AM, Bicknese AR, Pearlman AL (1995) Chondroitin sulfate proteoglycans in the developing cerebral cortex: the distribution of neurocan distinguishes forming afferent and efferent axonal pathways. JCN 355:615–628 Miyoshi G, Fishell G (2006) Directing neuron-specific transgene expression in the mouse CNS. Curr Opin Neurobiol 16(5):577–584 Molnár Z, Adams R, Blakemore C (1998a) Mechanisms underlying the establishment of topographically ordered early thalamo-cortical connections in the rat. J Neurosci 18:5723–5745 Molnár Z, Adams R, Goffinet AM, Blakemore C (1998b) The role of the first postmitotic cells in the development of thalamocortical fibre ordering in the reeler mouse. J Neurosci 18: 5746–5765 Molnár Z, Blakemore C (1995) How do thalamic axons find their way to the cortex? Trends Neurosci 18:389–397 Molnár Z, Cheung AF (2006) Towards the classification of subpopulations of layer V pyramidal projection neurons. Neurosci Res 55(2):105–115 Molnár Z, Cordery P (1999) Connections between cells of the internal capsule, thalamus and cerebral cortex in the embryonic pallium. J Comp Neurol 413:1–25 Molnár Z, Hoerder A, Wang WZ, DeProto J, Davies KE, Lee S, Paulsen O, Piñon MC, Cheung AFP (2007) Genes involved in the formation of the earliest cortical circuits. In: Bock G, Goode J (eds) Cortical development: genes and genetic abnormalities. Novartis foundation symposium 288. pp 212–229. discussion 224–229, 276–281 Molnár Z, Kurotani T, Higashi S, Toyama K (2003) Development of functional thalamocortical synapses studied with current source density analysis in whole forebrain slices. Brain Res Bull 60(4):355–372 Molyneaux BJ, Arlotta P, Menezes JR, Macklis JD (2007) Neuronal subtype spec in the ctx. Nat Rev Neurosci 8(6):427–37 Nelson SB, Sugino K, Hempel CM (2006) The problem of neuronal cell types: a physiol. genomics approach. TINS 29:339–345 Oeschger FM, Wang WZ, Ginet V, Hoerder-Suabedissen A, Truttmann AC, Molnár Z (2010) Subplate subpopulations in the hypoxic-ischemic neonatal rat brain. Anatomical Society of GB, January Meeting O’Leary DD, Chou SJ, Sahara S (2007) Area patterning of the mammalian cortex. Neuron 56(2):252–269 Pinon MC, Jacobs E, Campagnoni A, Molnar Z (2005) Development of the cortical projections from subplate neurons to the thalamus in Golli-tau-eGFP transgenic mice. Abstract for BNA meeting, Brighton Piñon MC, Jethwa A, Jacobs E, Campagnoni A, Molnár Z (2009 May 1) Dynamic integration of subplate neurons into the cortical barrel field circuitry during postnatal development in the Golli-tau-eGFP (GTE) mouse. J Physiol 587(Pt 9):1903–1915. E-pub 2009 Mar 16. Review Price DJ, Aslam S, Tasker L, Gillies K (1997) Fates of the early gener cells in the dev murine neoctx. JCN 377:414–422 Price DJ, Kennedy H, Dehay C et al (2006) The development of cortical connections. Eur J Neurosci 23:910–920 Rakic P (1976) Prenatal genesis of connections subserving ocular dominance in the rhesus monkey. Nature 261(5560):467–471 Rubenstein JL, Rakic P (1999) Genetic control of cortical development. Cerebral Cortex 9(6): 521–523 Rutherford M (2002) MRI of the neonatal brain. WB Saunders, Elsevier Science Limited, London Shatz CJ, Luskin MB (1986) The relationship between the geniculocortical afferents and their cortical target cells during development of the cat’s primary visual cortex. J Neurosci 6(12):3655–3668

2

Subplate and the Formation of the Earliest Cerebral Cortical Circuits

31

Shatz CJ, Rakic P (1981) The genesis of eff. connections from the vis. ctx of the fetal rhesus monkey. JCN 196:287–307 Shimogori T, Grove EA (2005) Fibroblast growth factor 8 regulates neocortical guidance of areaspecific thalamic innervation. J Neurosci 25:6550–6560 Thomson AM, Bannister AP (2003) Interlaminar connections in the neocortex. Cereb Cortex 13(1):5–14 Volpe JJ (2001) Neurobiology of periventricular leukomalacia in the premature infant. Pediatr Res 50:553–562 Walsh, CA (2007) Genes that control the shape and size of the human cerebral cortex. Novartis Found. Symposium 288 Wang, Franziska O, Sheena L, Zoltán M (2009) High quality RNA from multiple brain regions simultaneously acquired by laser capture microdissection. BMC Mol Biol 10(1):69 Wang WZ, Hoerder-Suabedissen A, Oeschger FM, Bayatti N, Ip BK, Lindsay S, Supramaniam V, Srinivasan L, Rutherford M, Møllgård K, Clowry GJ, Molnár Z (2010) Subplate in the developing cortex of mouse and human. Journal of Anatomy (in press)

Part II

Axonal Function

Chapter 3

Sodium Signals and Their Significance for Axonal Function Tony Kelly and Christine R. Rose

3.1 Introduction Regulation of intracellular sodium ion concentration ([Na+ ]i ) is critical for nervous system function, not only because sodium ions are the major current carriers during action potentials and excitatory postsynaptic currents in neurones, but also because many other cellular functions (e.g. intracellular Ca2+ homeostasis, intracellular pH homeostasis, reuptake of transmitters) are directly dependent on the inwardly directed Na+ gradient (Blaustein and Lederer, 1999; O’Shea, 2002; Chesler, 2003; Chen et al., 2004; Kanai and Hediger, 2004; Beart and O’Shea, 2006; also see Rose, 2003). Recent studies suggest that Na+ also acts as an intracellular signalling ion, modulating ion channel activity and altering protein transcription (see Bhattacharjee and Kaczmarek, 2005; Orlov and Hamet, 2006; Yu, 2006). In the central nervous system (CNS), bulk changes in cytoplasmic [Na+ ] are known to occur in cell bodies during pathophysiological conditions, such as ischaemia, and also likely contribute to the associated cellular dysfunction and ensuing neuronal cell death. However, dynamic measurements of [Na+ ]i in discrete cellular regions have only recently become possible due to advances in spatial and temporal resolution of cellular imaging (see Rose, 2003). Microspectrofluorescent recordings in neuronal dendrites or fine glial processes have revealed that significant changes in [Na+ ]i occur during modest physiological stimuli in compartments with high surface-to-volume ratios. However, in contrast to measurements from dendrites and somata, there is a paucity of dynamic [Na+ ]i measurements from axons; this, despite ample literature suggesting an important role of [Na+ ]i , changes in various aspects of axonal function and pathology, including action potential propagation and axonal degeneration. The first part of this chapter, therefore, briefly reviews current knowledge of [Na+ ]i transients in discrete cellular compartments, with the purpose of creating a better understanding of the possible mechanisms and roles of Na+ signals in the axonal compartment. The second part discusses Na+ signals in axons, T. Kelly (B) Department of Epileptology, Life & Brain Center, University of Bonn Medical Center, SigmundFreudstr. 25, Bonn, Germany e-mail: [email protected] D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_3, 

35

36

T. Kelly and C.R. Rose

the routes of Na+ entry along the axon and possible pathological and physiological implications of changes in axonal [Na+ ]i . Thus, it is the aim of this review to both enable and encourage further investigation of sodium in axonal function.

3.2 Sodium Signals and the Measurement of [Na+ ]i Many textbooks state that the movement of Na+ during neuronal activity has little effect on the [Na+ ]i . This statement is, however, only valid for large-volume fibres, in which electrical signalling requires only small ionic fluxes (e.g. Hodgkin and Huxley, 1952). Indeed, a calculation of the Na+ flux evoked by a single action potential in squid axons (∼1,000 μm in diameter) yields a change of the electrochemical Na+ gradient across the plasma membrane of only about 0.01% (Hille, 2001). In small compartments of central vertebrate neurones such as fine dendrites or axons with a diameter of 0.1 μm, however, the surface-to-volume ratio is much higher than in the squid axon. In this case, the calculation reveals that a single action potential will result in an increase of [Na+ ]i by as much as 10% (Hille, 2001). In contrast to large cell somata (>50 μm in diameter), which are amenable to measurements of [Na+ ]i by ion-sensitive microelectrodes, the impalement of small cellular compartments with ion-sensitive microelectrodes is not feasible. The most appropriate method for measurements of intracellular Na+ concentration in such compartments of central neurones is the use of Na+ -sensitive dyes, such as sodiumbinding benzofuran isophthalate (SBFI) or CoroNa Green. The Na+ -sensitive dye SBFI is similar to the well-known calcium-sensitive dyes, for example Fura-2 (Grynkiewicz et al., 1985; Minta and Tsien, 1989) and has been commonly used for [Na+ ]i measurements in a variety of cell types (see Rose, 2002). A distinct advantage of SBFI is the ability to calibrate ratiometric dual-excitation SBFI emission values to absolute [Na+ ]i (Rose and Ransom, 1996; also see Meier et al., 2006). In addition, SBFI loaded from the patch-pipette diffuses into the smallest cellular processes (see Rose, 2002) and thus is ideally suited for [Na+ ]i measurements from axons and dendrites. Finally, although small cellular compartments are more susceptible to photo-damage and photo-bleaching than cell somata, these effects are minimised using two-photon microscopy (see Rose, 2003).

3.2.1 Sodium Signals in Dendrites and Glial Processes 3.2.1.1 Action Potential-Induced [Na+ ]i Changes Direct demonstration of action potential-induced [Na+ ]i transients was first shown in dendrites using SBFI imaging (Jaffe et al., 1992). They showed that single Na+ action potentials caused a transient increase in [Na+ ]i in proximal dendrites of CA1 pyramidal neurones in hippocampal slices, providing evidence for backpropagation of Na+ action potentials along pyramidal cell dendrites. In addition to

3

Sodium Signals and Their Significance for Axonal Function

37

[Na+ ]i increases, dendritic Na+ spikes were also accompanied by large Ca2+ elevations because of the opening of voltage-activated Ca2+ channels. Interestingly, the spatial extent of dendritic [Na+ ]i changes following Na+ spikes was activity dependent: with increasing numbers of action potentials, [Na+ ]i increases were restricted to regions closer to the soma. This suggests that the distance to which Na+ spikes, and Ca2+ transients, invade the dendrites of CA1 neurones decreases in an activity-dependent manner, a fact since confirmed by employing dual somatic and dendritic patch-clamp recordings (Magee and Johnston, 1997). In contrast to the results obtained in hippocampal CA1 pyramidal neurones, action potentials do not elicit [Na+ ]i signals in the dendritic tree of cerebellar Purkinje neurones (see Fig. 3.1; Lasser-Ross and Ross, 1992). These studies indicate that Na+ spikes do not actively propagate along the dendrites of Purkinje cells. The increased spatial resolution of two-photon recordings permitted the measurement of [Na+ ]i from proximal dendrites and the minute dendritic spines of CA1 pyramidal neurones from hippocampal slices. In accordance with Jaffe et al. (1992), back-propagating action potentials evoked [Na+ ]i transients throughout the proximal dendritic tree and in dendritic spines (Rose et al., 1999). The action potential-induced [Na+ ]i transients reached amplitudes of 4 mM following a train of 20 action potentials and monotonically decayed with a time constant of several seconds. The mean amplitude of action potential-induced [Na+ ]i transients as well as the time to peak in spines were not significantly different from that in adjacent dendrites, indicating that Na+ accumulations might have been caused by the opening of voltage-gated Na+ channels located along the dendritic shaft as well as on the spines themselves. The above results using [Na+ ]i measurements are in good agreement with the notion that action potentials back-propagate along the dendrites of CA1 neurones and invade dendritic spines (see Nuriya et al., 2006; Araya et al., 2007). Furthermore, back-propagating action potentials in and of themselves are not sufficient to elicit large changes in [Na+ ]i within selected dendritic spines, the postsynaptic site for synaptic plasticity. However, the depolarisation and the increased [Na+ ]i in dendritic spines during back-propagating action potentials may play an important role in the coincidence detection of pre- and postsynaptic activity at synapses. 3.2.1.2 Synaptically Evoked [Na+ ]i Changes in Dendrites and Glia Dendritic spines are the major site for excitatory synaptic input in the vertebrate CNS (Harris and Kater, 1994) and many neurotransmitters act on ionotropic receptors to allow Na+ influx. Furthermore, the uptake of neurotransmitters (e.g. glutamate) requires the coupled influx of Na+ into the cell (Danbolt, 2001; Chen et al., 2004; Kanai and Hediger, 2004). Therefore, neurotransmission is expected to significantly alter the [Na+ ]i of postsynaptic neurones as well as neighbouring glia. Cerebellar Purkinje neurones. In Purkinje cells of guinea pig brain slices, synaptic stimulation of climbing fibres or parallel fibres was associated with clear [Na+ ]i increases in dendrites meditated in large part by AMPA-type receptor-operated

38

T. Kelly and C.R. Rose

Fig. 3.1 Action potential-evoked [Na+ ]i changes in axons, dendrites and somata of cerebellar neurones. Purkinje neurones in guinea pig cerebellar slices were loaded via sharp microelectrodes tip filled with 6–12 mM SBFI. Conventional wide-field fluorescence recordings were performed and changes in [Na+ ]i reported as background subtracted F/F values. Action potentials were elicited by somatically injected depolarising current pulses. Top, Image of Purkinje neurone filled with SBFI (microelectrode visible on left). Rectangles indicate selected areas from with SBFI measurements of axon (dashed), soma (dotted) and dendrite (solid) were recorded. Middle, Somatic action potentials evoked large increases in [Na+ ]i in the axon (dashed line) but failed to evoke measurable changes in the soma (dotted line) and dendrites (solid line). Bottom, Sharp microelectrode recording of a burst of action potentials evoked by somatic depolarising current injection. (Modified, with permission, from Lasser-Ross and Ross, 1992)

channels (Lasser-Ross and Ross, 1992; Callaway and Ross, 1997; Bennay et al., 2008). However, postsynaptic Na+ entry in cerebellar Purkinje neurones is not restricted to influx through AMPA receptor-operated channels. During pharmacological block of ionotropic glutamate and GABA receptors, parallel fibre stimulation

3

Sodium Signals and Their Significance for Axonal Function

39

resulted in a postsynaptic current that was associated with a rise in dendritic [Na+ ]i (Knöpfel et al., 2000). This [Na+ ]i transient was restricted to a region of the dendrite close to the stimulation electrode and sensitive to an antagonist of metabotropic glutamate receptors. These findings demonstrated that the current mediated by AMPA receptor-operated channels and metabotropic glutamate receptors leads to significant influx of Na+ into Purkinje cell dendrites. The authors suggested that the [Na+ ]i transients provide a new intracellular signalling mechanism that may contribute to synaptic plasticity at the parallel fibre–Purkinje cell synapse (Knöpfel et al., 2000; also see Yuan and Knöpfel, 2006). Hippocampal neurones. In the apical dendrites of neurones in rat hippocampal slices, modest synaptic stimulation with five stimuli evoked [Na+ ]i transients. These [Na+ ]i increases reached a maximum amplitude of ∼10 mM in 30 μm dendritic regions closest to the stimulating pipette (Rose and Konnerth, 2001). Within this dendritic region two classes of spines were distinguished based upon their response to synaptic stimulation (Rose and Konnerth, 2001). In the first class of spines, termed ‘passive’ spines, the amplitude and kinetics of [Na+ ]i transients resembled the responses in the adjacent dendritic trunk. In contrast, a second class of spines displayed activity-induced [Na+ ]i increases 2–3 times larger than those of the adjacent dendrite, reaching up to 35–40 mM. In addition, recovery to baseline [Na+ ]i followed a biexponential decay. From these results, it was concluded that this second class of spine, termed ‘active’ spines, were the sites of synaptic input and, thus, the sites of synaptic Na+ influx. In contrast to cerebellar Purkinje neurones the observed postsynaptic [Na+ ]i transients in hippocampal neurones were largely mediated by Na+ entry through NMDA receptor-operated channels, whereas Na+ entry through AMPA-receptor channels played a minor role (Rose and Konnerth, 2001). In addition, [Na+ ]i increases evoked by somatically generated short bursts of back-propagating action potentials did not significantly contribute to the [Na+ ]i increases generated by suprathreshold synaptic stimulation. However, pairing of five back-propagating action potentials with five subthreshold synaptic stimulations, two protocols that individually did not result in measurable [Na+ ]i increases, induced supralinear [Na+ ]i increases with an amplitude of ∼75% of that observed with suprathreshold synaptic stimulation. Therefore, the occurrence of postsynaptic action potentials in coincidence with presynaptic activity was the critical parameter for the induction of NMDA receptor-mediated postsynaptic [Na+ ]i transients. The results presented above (Rose and Konnerth, 2001) indicate that NMDA receptor activity may contribute yet another signal for coincidence detection, that of [Na+ ]i changes. Postsynaptic [Na+ ]i transients seem to be ideally suited for coincidence detection since they are restricted to the site of activation and exhibit all or none characteristics, in that they require the occurrence of postsynaptic action potentials in coincidence with presynaptic activity. Bergmann glial cells. A recent study by Bennay et al. (2008) revealed [Na+ ]i increases in Bergmann glial processes following short burst (5 stimuli for 100 ms) stimulation of parallel fibres or climbing fibres. The increases in [Na+ ]i were largest in the glial processes nearest to the stimulating electrode, with a maximum increase

40

T. Kelly and C.R. Rose

of up to 9 mM, and decayed to baseline with a time constant of 1–2 min. These activity-induced increases in [Na+ ]i were primarily attributed to Na+ -dependent glutamate transport and to a lesser extent CNQX-sensitive AMPA receptor-operated channels. The authors proposed that the reduction in the Na+ gradient following short bursts of synaptic activity may feedback and inhibit Na+ -dependent glutamate transport activity by 20–25% (Bennay et al., 2008). These results were the first demonstration of activity-induced [Na+ ]i increases in glial processes and are consistent with the suggestion that activity-induced increases in [Na+ ]i may modulate synaptic activity by altering glutamate uptake (also see below). Similar activity-induced [Na+ ]i changes have since been shown in the processes of hippocampal astrocytes (Langer and Rose, 2009). In addition, Na+ influx was shown to contribute to the observed loss of astrocytic processes in optic nerve white matter during a period of ischaemia (Salter and Fern, 2008). The integrity of astrocyte processes was substantially protected from 40 min ischaemia by manoeuvres that inhibited the influx of Na+ via Na+ -dependent K+ , Cl− co-transport. The involvement of Na+ influx in the loss of astrocytic processes during ischaemia is reminiscent of the well-recognised involvement of increases in axonal [Na+ ]i observed during ischaemia (see below) and may contribute to the ischaemic injury of white matter.

3.3 Sodium Signals in Axons Given the above observations that activity-evoked [Na+ ]i increases were much larger in spatially restricted cellular compartments, such as dendrites and glial processes, compared to somata, large [Na+ ]i increases also likely occur in fine axons. Consistent with this Bergman (1970) found that prolonged action potential firing decreased the reversal potential for Na+ by 24–28 mV at the nodes of Ranvier from Xenopus nerve fibres. Assuming the bath [Na+ ] of 111.5 mM remained constant at the membrane of the isolated Xenopus axons and that under resting conditions the cytosolic [Na+ ] in the 10–15 μm fibres was ∼15 mM, a decrease in ENa of 25 mV reflects an increase in [Na+ ]i to ∼40 mM. Lasser-Ross and Ross (1992) confirmed that action potentials evoke TTXsensitive [Na+ ]i increases in axons of cerebellar Purkinje cells using fluorescent imaging; furthermore, [Na+ ]i increases in Purkinje cell axons occurred in the absence of significant [Na+ ]i changes in the dendrites or somata (see Fig. 3.1). Action potential-evoked [Na+ ]i transients were also observed in the initial segment of axons from layer 5 cortical pyramidal neurones (Fleidervish et al., 2010). Furthermore, persistent [Na+ ]i increases observed in response to prolonged (0.5–1 s) subthreshold depolarising current steps were also greater in the axons compared with somata or dendrites, suggesting the additional presence of non-inactivating voltage-activated Na+ channels in axons of layer 5 cortical pyramidal neurones (Fleidervish et al., 2010). Interestingly, persistent voltage-activated Na+ channels

3

Sodium Signals and Their Significance for Axonal Function

41

have been implicated in pathological [Na+ ]i increases observed in axons in response to ischaemia (see below). Recent data from the authors, employing two-photon laser scanning microscopy of CA1 pyramidal neurones from hippocampal slices, demonstrated that modest stimulation evokes rapid transient [Na+ ]i increases of ∼9 mM in the axon initial segment (AIS) and increasing the number of action potentials elicited augments the magnitude and prolongs the duration of axonal [Na+ ]i transients (see Fig. 3.2). Increases in [Na+ ]i in the corresponding soma and basal dendrites were of reduced amplitude and slower kinetics than [Na+ ]i transients observed in the axon (Fig. 3.2). Similar results have also been observed from measurement of [Na+ ]i in Cartwheel interneurones. Bender and Trussell (2009) also found that action potential-induced [Na+ ]i transients were largest in the axon, compared to the soma

Fig. 3.2 Action potential-evoked [Na+ ]i transients in axons, basal dendrites and somata. Twophoton measurements of CA1 pyramidal neurones in mouse hippocampal slices loaded with 1 mM of the Na+ -sensitive indicator SBFI, as described by Rose et al. (1999). Action potentials were elicited with brief (10 ms) depolarising current pulses delivered at 50 Hz from the patch-pipette to the soma. Top left, Montage of 50 sections acquired at 2 μm intervals through a CA1 pyramidal neurone filled with 1 mM SBFI via the patch-pipette (patch-pipette is visible on left). Identified axon indicated with grey triangles and distinguished from neighbouring basal dendrites by the absence of dendritic spines, small diameter and extended length (in contrast to the basal dendrites, the axon extended beyond the field of view). Grey rectangles indicate the axon initial segment and the region of a neighbouring basal dendrite where measurements were performed. Bottom left, 12 somatically elicited action potentials (AP), applied at black arrowhead, evoked rapid transient increases in [Na+ ]i of ∼8 mM in the axon. Increasing the number of action potentials to 23 and 52 augmented the peak amplitude and prolonged the duration of the axonal [Na+ ]i transients. Right, In a different neurone, 22 action potentials, applied at black arrowhead, evoked larger and more rapid [Na+ ]i increases in the axon compared with [Na+ ]i transients observed in the basal dendrites or soma of the same cell

42

T. Kelly and C.R. Rose

and dendrites of Cartwheel interneurones. Furthermore, within the initial 80 μm of the axons, Na+ transients were largest between the first 20–30 μm (Bender and Trussell, 2009). The larger [Na+ ]i increases within the axon initial segment, compared with action potential-induced [Na+ ]i changes in the soma and dendrites, are consistent with a higher Na+ channel density and/or higher surface to volume ratio of the AIS. A higher density of Na+ channels in the AIS has been shown using immunohistochemical studies, estimated channel density at ∼2,500 μm−2 (see Kole et al., 2008), and [Na+ ]i imaging studies also support a greater Na+ influx in the AIS (Kole et al., 2008; Fleidervish et al., 2010). However, the recent study from Fleidervish et al. (2010), showed that a 3 fold increase in the Na+ influx was sufficient to induce the larger [Na+ ]i increases, suggesting that a 45 fold increase in Na+ channel density (see Kole et al., 2008) may not be required to elicit large [Na+ ]i in the spatially restricted AIS. Moreover, this most recent data using Na+ imaging in the AIS supports the possibility of altered Na+ channel kinetics, rather than high Na+ channel density, in the initiation of action potentials at the AIS. Increases in [Na+ ]i also occur in the more spatially restricted nodes of Ranvier of myelinated axons, although the rapid longitudinal diffusion of Na+ may hinder accurate measurements of peak [Na+ ]i amplitude (Fleidervish et al., 2010). Finally, despite the fact that the density of voltage-activated Na+ channels along unmyelinated axons is more diffuse (∼50–500 μm−2 in unmyelinated axons compared with 1,000–2,000 μm−2 at the nodes of Ranvier of myelinated axons; see Waxman and Ritchie, 1993; Hille, 2001), there is no a priori reason that action potential propagation along unmyelinated axons does not also elicit [Na+ ]i increases. In addition, activation of Na+ /Ca2+ exchange may contribute to Na+ influx along the axon. Immunofluorescent data confirm the presence of the plasmalemmal NCX1 isoform along the entire axon, including at the nodes of Ranvier of, optic nerve, sciatic nerve and dorsal root ganglion fibres (Steffensen et al., 1997). Selective loading of axon fibres with Ca2+ -sensitive indicators (e.g. Ren et al., 2000; Verbny et al., 2002) has revealed activity-induced [Ca2+ ]i transients along the axons of cerebellar Purkinje neurones (Callewaert et al., 1996), neocortical pyramidal neurones (Schiller et al., 1995), Cartwheel interneurones (Bender and Trussell, 2009), optic nerve (Lev-Ram and Grinvald, 1987; Sun and Chiu, 1999; Verbny et al., 2002; Zhang et al., 2006) and vagus nerve (Wächtler et al., 1998). Activation of voltageactivated Ca2+ channels along the axon by Na+ -dependent action potentials results in increases in axonal [Ca2+ ]i speculated to contribute to stimulation of axonal transport (Breuer et al., 1992). The recovery of increases in [Ca2+ ]i observed during the axonal propagation of action potentials involves Na+ /Ca2+ exchange (Verbny et al., 2002). Thus, activation of Na+ /Ca2+ exchange influences the magnitude and time course of [Ca2+ ]i transients along the axon and likely contributes to action potential-evoked [Na+ ]i increases in axons. In addition to action potential induced [Na+ ]i transients that occur along the length of the axon (above), [Na+ ]i transients also occur in presynaptic terminals. In presynaptic terminals of cerebellar granule cells, loaded with both Ca2+ -sensitive and Na+ -sensitive indicators, high-frequency stimulation of parallel fibres evoked

3

Sodium Signals and Their Significance for Axonal Function

43

rapid increases in [Ca2+ ]i and prolonged increases in [Na+ ]i (Regehr, 1997). The [Ca2+ ]i transients evoked after five stimuli reached a peak at ∼100 ms after the stimulation and returned to near pre-stimulus levels within a second. In contrast, the kinetics of [Na+ ]i transients are much slower. Increases in [Na+ ]i at presynaptic terminals following parallel fibre stimulation occurred in two temporally distinct phases; rapid increases in [Na+ ]i were likely mediated by voltage-activated Na+ channels and slow [Na+ ]i increases required Ca2+ influx and likely involved Na+ /Ca2+ exchange activity (Regehr, 1997). The return of [Na+ ]i to pre-stimulus levels had a decay time constant of minutes (Regehr, 1997) likely reflecting the diffusion of axoplasmic Na+ and its removal by Na+ , K+ -ATPase activity (David et al., 1997). Unlike Ca2+ , which is rapidly buffered, the diffusion coefficient of intracellular Na+ in myelinated axons (∼1,300 μm2 s−1 ) is similar to that in ionic solutions, suggesting an absence of intracellular Na+ buffers (David et al., 1997). An additional mechanism that may contribute to changes in presynaptic [Na+ ]i is synaptic transmission itself. Following the release of excitatory neurotransmitters (such as glutamate, serotonin and acetylcholine) activation of their respective presynaptic ligand-activated cation channel(s) results in Na+ influx into the presynaptic terminal. Although inhibition of ionotropic glutamate receptors minimally affected [Na+ ]i increases in presynaptic terminals of cerebellar granule cells (Regehr, 1997), presynaptic ionotropic glutamate receptors are distributed throughout the CNS (see Schenk and Matteoli, 2004; Duguid and Sjöström, 2006) and may contribute to presynaptic increases in [Na+ ]i during synaptic transmission in other regions. Additionally, activation of metabotropic glutamate receptors (mGluR) located at presynaptic terminals (see Coutinho and Knöpfel, 2002) may increase [Na+ ]i , similar to mGluR-evoked [Na+ ]i transients observed in postsynaptic dendrites. For instance, activation of group I mGluRs on the dendrites of substantia nigra dopamine neurones, cerebellar Purkinje cells and olfactory mitral cells increases [Na+ ]i possibly by activation of non-selective cation channels and/or by the activation of Na+ /Ca2+ exchange (see Knöpfel et al., 2000; Yuan and Knöpfel, 2006). The localisation of Na+ -dependent transporters responsible for transmitter uptake into presynaptic terminals (Chen et al., 2004; Kanai and Hediger, 2004) provides an additional mechanism by which neurotransmission may increase [Na+ ]i in the presynaptic terminal. The slower kinetics of Na+ -dependent transport mechanisms, compared to action potential generation, Ca2+ transients or presynaptic receptor activation may provide an additional mechanism for the prolonged duration of stimulus-evoked presynaptic [Na+ ]i increases. The above data employing fluorescent and high-resolution two-photon imaging demonstrated that activity-evoked [Na+ ]i increases occur along the axon and in the presynaptic terminal (see Fig. 3.3). Although TTX-sensitive voltage-activated Na+ channels are a major route for Na+ influx during activity-evoked [Na+ ]i transients, data suggest that Na+ -permeable ligand-gated channels and Na+ -dependent transport mechanisms also contribute. In addition, given the dependence of many cellular transport mechanism on the Na+ gradient, increases in axonal [Na+ ]i likely have an important role in axonal function and pathology.

44

T. Kelly and C.R. Rose

Fig. 3.3 Pathways for Na+ flux and possible sites of modulation by [Na+ ]i in different axonal regions. Na+ influx pathways, such as Na+ -permeable voltage- and ligand-activated ion channels and Na+ -dependent transport mechanisms, result in increases in [Na+ ]i in axons. Increases in axoplasmic [Na+ ]i results in a number of potential consequences (see text for more details). Activation of Na+ , K+ -ATPase (1) and Na+ -activated K+ channels (2) hyperpolarises the axonal membrane potential and may contribute to determining firing pattern. Elevation of [Na+ ]i also modulates the activity of Na+ -dependent transporters. For instance, stimulation of mitochondria Na+ /Ca2+ exchange (3) and inhibition of plasmalemma Na+ /Ca2+ exchange (4, 5) by increases in axonal [Na+ ]i alter [Ca2+ ]i transients along the axon and may contribute to increased release of neurotransmitters from the presynaptic terminal. In addition, elevations in presynaptic [Na+ ]i may modulate synaptic transmission by inhibiting reuptake of glutamate (6), a mechanism possibly contributing to LTD. Large prolonged elevations in [Na+ ]i may result in mitochondrial dysfunction (3) and additionally may reverse both plasmalemma Na+ /Ca2+ exchange (4, 5) and glutamate transport (6) thereby contributing, respectively, to Ca2+ - and glutamate-mediated toxicity under pathological conditions such as ischaemia. Finally, increases in [Na+ ]i have effects on downstream intracellular second messengers as evident by [Na+ ]i increases augmenting NMDA receptor-operated channels via Src kinases (7) and attenuating Ca2+ channel activity via βγ G-proteins (8)

3.3.1 Pathophysiological Implications of [Na+ ]i Changes The involvement of [Na+ ]i changes in the dysfunction and injury of axons in the CNS is increasingly well recognised and large increases in axonal [Na+ ]i are know to occur in response to ischaemia or anoxia (Stys, 1998, 2004), physical trauma (Lucas et al., 1997; Stys, 2004) and multiple sclerosis (Stys, 2005;

3

Sodium Signals and Their Significance for Axonal Function

45

Waxman, 2006). These large increases in axonal [Na+ ]i contribute to injury by compromising metabolic activity, promoting cellular swelling and exacerbating both Ca2+ -mediated toxicity and excitotoxicity mediated by glutamate (see Fig. 3.3). The response of axonal [Na+ ]i to ischaemia/anoxia illustrates the multiple mechanisms that result in large [Na+ ]i increases and the potential pathways that contribute to the subsequent irreversible failure of action potential conduction. Within minutes following the onset of anoxia or metabolic inhibition, a large depolarisation in central axons is observed (Leppanen and Stys, 1997; Stys et al., 1998) and precipitates massive movement of ions across the axoplasmic membrane (e.g. Ransom et al., 1992; also see Lipton, 1999; Somjen, 2002). Axoplasmic [Na+ ] can reach 100 mM after 60 min anoxia (Lopachin and Stys, 1995; also see Nikolaeva et al., 2005), an increase primarily attributed to the activation of non-inactivating voltage-activated Na+ channels; thus, anoxia-evoked [Na+ ]i increases are sensitive to inhibition of voltage-activated Na+ channels by local anaesthetics. In addition, evidence from mammalian central neurones suggests that the elevated [Na+ ]i activates Na+ /K+ ATPase, thereby contributing to ATP depletion and, in turn, the rapid loss of membrane potential and ion homeostasis during anoxia (Erecinska et al., 1991; Fowler and Li, 1998; Chinopoulos et al., 2000). In combination with membrane depolarisation, the accumulation of [Na+ ]i drives reverse-mode Na+ /Ca2+ exchange resulting in large increases in [Ca2+ ]i and Ca2+ -mediated toxicity. The importance of elevated [Na+ ]i in the intracellular Ca2+ overload and axonal injury following anoxia has been demonstrated in experiments in which axonal damage was greatly alleviated by removal of extracellular Na+ or the inhibition of either voltage-activated Na+ channels or reverse-mode Na+ /Ca2+ exchange activity in the optic nerve (Stys et al., 1990; Stys et al., 1992; Waxman et al., 1993; Nikolaeva et al., 2005) and spinal cord (Imaizumi et al., 1997; Li et al., 2000). Furthermore, large elevations in [Na+ ]i may contribute to glutamate-mediated excitotoxicity (see Allen et al., 2004) by releasing glutamate via reversal of glutamate transporters, observed in spinal cord white matter (Li and Stys, 2001) and in axon terminals (Tretter and Adam-Vizi, 2002), as well as increasing the sensitivity of NMDA receptors (see Yu, 2006). Along the axon, reverse glutamate transport activity and activation of AMPA receptors have been implicated in myelin degeneration (Li et al., 1999). The re-introduction of oxygenated media following 60 min of anoxia results in the return of normal ion composition in only a minority of optic nerve fibres; a majority of fibres, on the other hand, maintain high [Ca2+ ]i or further accumulate large quantities of Ca2+ (Stys and Lopachin, 1996). These data are reminiscent of data from mammalian central neurones in which increases in Na+ influx observed following anoxia may contribute to reperfusion injury (see Lipton, 1999). In hippocampal neurones, for instance, the additional Na+ influx observed following anoxia was attributed, at least in part, to Na+ /H+ exchange activity (Sheldon et al., 2004). In addition, and consistent with a role of Na+ /H+ activity in the pathology of anoxia in axons, inhibition of Na+ /H+ exchange activity with harmaline or amiloride improved the recovery of compound action potential conduction in spinal cord white matter following anoxia (Imaizumi et al., 1997). The combined elevation of [Ca2+ ]i and [Na+ ]i has been proposed to result in mitochondrial dysfunction

46

T. Kelly and C.R. Rose

during reperfusion of O2 following anoxia (see Stys, 1998). Indeed, recent evidence suggests that during oxidative stress, as occurs during the re-introduction of O2 after anoxia, elevated [Na+ ]i exacerbates the depletion of ATP and promotes the collapse of the mitochondrial membrane potential (Chinopoulos et al., 2000). This latter effect of elevated [Na+ ]i to alter cellular metabolism may additionally involve Na+ /Ca2+ exchange at the mitochondrial membrane and consequence increases in mitochondrial [Na+ ], as observed in astrocytes following glutamate-evoked [Na+ ]i increases (see Bernardinelli et al., 2006). Taken together these data suggest that increases in [Na+ ]i elicited during anoxia or ischaemia and upon oxygen reperfusion contribute to the subsequent axonal injury (see Fig. 3.3). In addition, the pathways by which large elevations in [Na+ ]i result in the deleterious effects observed in response to anoxia and ischaemia may be common to other forms of axonal injury. Indeed, in models of multiple sclerosis, the expression of voltage-activated Na+ channels was changed from Nav 1.2 channels to Nav 1.6 channels, which display more persistent currents, and Nav 1.6 channels were co-localised with Na+ /Ca2+ exchangers (Craner et al., 2004a, b; also see Waxman, 2006). Furthermore, inhibition of Na+ influx, via voltage-activated Na+ channels and Na+ /H+ exchange, during spinal cord injury in vitro reduces the subsequent deficits in action potential conduction (Agrawal and Fehlings, 1996, 1997). Similarly, inhibition of voltage-activated Na+ channels alleviates deficits in both axonal conduction and motor function following contusion spinal cord injury in vivo (Hains et al., 2004).

3.3.2 Possible Physiological Roles of [Na+ ]i Changes In contrast to the well-recognised involvement of large [Na+ ]i increases in axonal injury, the possibility that changes in [Na+ ]i mediate physiological consequences (see Fig. 3.3), has received relatively little attention. This, despite the fact that increases in axonal [Na+ ]i accompany moderate numbers of action potentials (see Figs. 3.1 and 3.2; also see Bender and Trussell, 2009) and changes in axonal [Na+ ]i are ideally placed to modulate axonal excitability. Recent data suggest that increases in axonal [Na+ ]i following trains of action potentials contribute to the repolarisation of the membrane potential. High-frequency axonal simulation at 100 Hz, elicited trains of action potentials followed by a post-tetanic hyperpolarisation (PTH) of up to 25 mV in axons (Morita et al., 1993) and at the Calyx of Held (Kim et al., 2007). Inhibition of Na+ , K+ -ATPase activity by application of ouabain inhibited the PTH and decreased the fidelity of axons to propagate high-frequency trains of action potentials, i.e. the number of failed action potentials during the stimulus train progressively increased in the presence of ouabain (Kim et al., 2007). The data suggest that Na+ , K+ -ATPase-mediated hyperpolarisations decrease the time required for recovery of Na+ channels from inactivation and thereby allow reliable axonal conduction of action potentials at high frequency. Furthermore, patch-clamp recordings from the soma of various types of central neurones revealed that action potentials

3

Sodium Signals and Their Significance for Axonal Function

47

elicit afterhyperpolarisations mediated by Na+ -activated K+ channels (Bhattacharjee and Kaczmarek, 2005). In axons of Xenopus neurones, Na+ -activated K+ channels are localised in the vicinity of voltage-activated Na+ channels at nodes of Ranvier (Koh et al., 1994) and thus ideally positioned to detect increases in [Na+ ]i during action potential propagation. Therefore, Na+ influx is not only necessary for fast action potential propagation along the axon, but increases in axoplasmic [Na+ ]i and subsequent stimulation of Na+ , K+ -ATPase activity and Na+ -activated K+ channels appear to contribute to repolarising the membrane potential and thus in determining the frequency of action potential firing in axons. In addition to regulating axonal excitability, [Na+ ]i transients are linked to [Ca2+ ]i in neuronal compartments via activation of Na+ /Ca2+ exchange (see Blaustein and Lederer, 1999); Na+ /Ca2+ exchange activity depends on membrane potential as well as the intracellular and extracellular concentrations of Na+ and Ca2+ . Thus, elevations in axoplasmic [Na+ ] during action potentials likely shape axonal [Ca2+ ]i transients; inhibition of Na+ /Ca2+ exchange, by experimentally decreasing [Na+ ]o or increasing [Na+ ]i , prolongs action potential-evoked [Ca2+ ]i transients along optic nerve axons and at central presynaptic terminals (Bouron and Reuter, 1996; Regehr, 1997; Verbny et al., 2002; Kim et al., 2005). The inhibition of Na+ /Ca2+ exchange activity likely contributes to the increased rate of neurotransmitter vesicular fusion observed under conditions of elevated [Na+ ]i in presynaptic terminals (Mulkey and Zucker, 1992; Reuter and Porzig, 1995; Bouron and Reuter, 1996; also see Scotti et al., 1999). Unlike in dendrites and dendritic spines where postsynaptic [Na+ ]i increases have been shown to modulate NMDA receptor-mediated currents (Yu and Salter, 1998, 1999) (postsynaptic effects that may contribute to the Na+ -dependent modulation of synaptic plasticity; see above; also see Linden et al., 1993), the involvement of presynaptic [Na+ ]i changes in synaptic plasticity has, as yet, received less attention. The kinetics of presynaptic evoked [Na+ ]i transients have a much slower decay than [Ca2+ ]i transients (Regehr, 1997). Thus, [Na+ ]i increases may integrate synaptic signals over a greater temporal range (100 ms to minutes), in contrast to rapidly decaying [Ca2+ ] increases. One possible mechanism whereby evoked increases in presynaptic [Na+ ]i may result in activity-dependent modulation of synaptic function is via inhibition of glutamate transport activity. At climbing fibre– Purkinje cell synapses in the cerebellum, presynaptic glutamate transporters are known to significantly contribute to the uptake of glutamate. Inhibition of glutamate transport activity at climbing fibre–Purkinje synapses results in the long-term depression (LTD) of the synapse, due to glutamate ‘spill over’ and activation of peri-synaptic mGluR (Brasnjo and Otis, 2001; Otis et al., 2004). Furthermore, climbing fibre–Purkinje synapses that exhibited reduced glutamate uptake via neuronal glutamate transporters had a greater propensity for LTD (Wadiche and Jahr, 2005). These data suggest that synapses with reduced presynaptic glutamate uptake fail to prevent ‘spill over’ of glutamate, which results in LTD via activating an mGluR-dependent mechanism. Similarly, reduced glutamate transport and ‘spill over’ of glutamate modulate GABAergic synaptic transmission (e.g. Drew et al., 2008; Szapiro and Barbour, 2007). Increases in [Na+ ]i are known to inhibit the

48

T. Kelly and C.R. Rose

activity of glutamate transporters in glial cells (see above; Bennay et al., 2008; Kelly et al., 2009) and a similar inhibition of neuronal glutamate transport activity by activity-evoked increases in presynaptic [Na+ ]i may contribute to the known involvement of Na+ in cerebellar LTD (see Linden et al., 1993). In addition, stimulating optic nerve white matter resulted in glutamate-dependent [Ca2+ ]i increases in neighbouring glia that were sensitive to the block of glutamate transport (Kriegler and Chiu, 1993; Chiu and Kriegler, 1994). The above data suggests that physiological stimuli release glutamate from the axon, possibly via reversed glutamate transport activity, alternatively released glutamate may elicit [Ca2+ ]i transients by acting on glial glutamate transporters (e.g. Rojas et al., 2007). In regard to the former possibility, reverse glutamate transport likely requires high axoplasmic [Na+ ] during action potential propagation; axonal [Na+ ]i increases may thereby provide the link between axonal activity and the activity-dependent glutamate release from axons, which in turn signals neighbouring glial cells (see Káradóttir and Attwell, 2007). In addition to the ability of changes in [Na+ ]i to exert effects on ion channels and transporters by alterations in ion gradients, recognition of Na+ as a modulator of traditional intracellular signalling pathways is increasing. In non-neuronal cells, for instance, increases in [Na+ ]i induce increased protein expression of mortalin and c-Fos, independent of [Ca2+ ]i changes (see Orlov and Hamet, 2006; Haloui et al., 2007). In central neurones, increases in [Na+ ]i increase the activity of NMDA receptor-operated channels via a Src protein-dependent mechanism (Yu and Salter, 1998, 1999) and inhibit N-type Ca2+ channels by promoting the dissociation of heterotrimeric G proteins (see Blumenstein et al., 2004; also see Rishal et al., 2003); again effects that may contribute to the aforementioned involvement of [Na+ ]i changes in synaptic plasticity. The data suggest that changes in axoplasmic [Na+ ] result in alterations to axonal function, such as firing pattern as well as neurotransmitter release and reuptake, and that the alterations persist beyond the initial transient change in [Na+ ]i . (see Fig. 3.3)

3.4 Conclusions and Outlook Taken together research to date shows that, in contrast to large-volume fibres such as the squid giant axon, activity-evoked transient increases in [Na+ ]i occur in spatially restricted cellular compartments of the mammalian CNS. In the axonal compartment, action potentials elicit larger increases in [Na+ ]i than [Na+ ]i transients observed in the corresponding soma or dendrites, reflecting the greater density of voltage-gated Na+ channels along the axon. Furthermore, the many Na+ -permeable ligand-gated ion channels and Na+ -dependent transporters likely contribute to activity-evoked increases in [Na+ ]i along the axon and at the presynaptic terminal. However, what remains largely unknown is the extent to which the various routes of Na+ entry contribute to activity-evoked axonal [Na+ ]i transients in different regions of the axons as well as in different axon fibres. In addition, and in contrast to

3

Sodium Signals and Their Significance for Axonal Function

49

well-recognised role of [Na+ ]i in axonal pathology, the physiological implications of [Na+ ]i transients in axons are still a matter for speculation. This is surprising as axonal [Na+ ]i likely serves as a sensitive barometer of axonal activity. Moreover, changes in axonal [Na+ ]i may in turn modulate many cellular processes by altering the Na+ gradient and by acting as an intracellular signalling ion. In this regard, further fluorescent studies employing model systems, which are more tolerant of photo-damage, will remain invaluable, such model systems include large synapses (e.g. the Calyx of Held and Mossy fibre boutons) and peripheral axons with a larger diameter. Finally, studies employing high-resolution two-photon microscopy will allow the investigation of Na+ transients in smaller diameter central axons. Acknowledgment Our thanks to Claudia Roderigo, whose expertise was essential in the preparation of figures, and Kate Butkus for valuable comments on previous versions of the chapter.

References Agrawal SK, Fehlings MG (1996) Mechanisms of secondary injury to spinal cord axons in vitro: role of Na+ , Na+ -K+ -ATPase, the Na+ -H+ exchanger, and the Na+ -Ca2+ exchanger. J Neurosci 16:545–552 Agrawal SK, Fehlings MG (1997) The effect of the sodium channel blocker QX-314 on recovery after acute spinal cord injury. J Neurotrauma 14:81–88 Allen NJ, Káradóttir R, Attwell D (2004) Reversal or reduction of glutamate and GABA transport in CNS pathology and therapy. Pflügers Arch 449:132–142 Araya R, Nikolenko V, Eisenthal KB, Yuste R (2007) Sodium channels amplify spine potentials. PNAS 104:12347–12352 Beart PM, O’Shea RD (2006) Transporters for L-glutamate: an update on their molecular pharmacology and pathological involvement. Br J Pharmacol 150:5–17 Bender KJ, Trussell LO (2009) Axon initial segment Ca2+ channels influence action potentials generation and timing. Neuron 61:259–271 Bennay M, Langer J, Meier SD, Kafitz KW, Rose CR (2008) Sodium signals in cerebellar Purkinje neurons and Bergmann glial cells evoked by glutamatergic synaptic transmission. Glia 56:1138–1149 Bergman C (1970) Increase of sodium concentration near the inner surface of the nodal membrane. Pflügers Arch 317:287–302 Bernardinelli Y, Azyrias G, Chatton JY (2006) In situ fluorescence imaging of glutamate-evoked mitochondrial Na+ responses in astrocytes. Glia 54:460–470 Bhattacharjee A, Kaczmarek LK (2005) For K+ channels, Na+ is the new Ca2+ . Trends Neurosci 28:422–428 Blaustein MP, Lederer WJ (1999) Sodium/calcium exchange: its physiological implications. Physiol Rev 79:763–854 Blumenstein Y, Maximyuk OP, Lozovaya N, Yatsenko NM, Kanevsky N, Krishtal O, Dascal N (2004) Intracellular Na+ inhibits voltage-dependent N-type Ca channels by a G protein βγ subunit-dependent mechanism. J Physiol 556:121–134 Bouron A, Reuter H (1996) A role of intracellular Na+ in the regulation of synaptic transmission and turnover of the vesicular pool in cultured hippocampal cells. Neuron 17:969–978 Brasnjo G, Otis TS (2001) Neuronal glutamate transporters control activation of postsynaptic metabotropic glutamate receptors and influence cerebellar long-term depression. Neuron 31:607–616 Breuer AC, Bond M, Atkinson MB (1992) Fast axonal transport is modulated by altering transaxolemmal calcium flux. Cell Calcium 13:249–262

50

T. Kelly and C.R. Rose

Callaway JC, Ross WN (1997) Spatial distribution of synaptically activated sodium concentration changes in cerebellar Purkinje neurons. J Neurophysiol 77:145–152 Callewaert G, Eilers J, Konnerth A (1996) Axonal calcium entry during fast ‘sodium’ action potentials in rat cerebellar Purkinje neurones. J Physiol 495:641–647 Chen NH, Reith M, Quick M (2004) Synaptic uptake and beyond: the sodium- and chloridedependent neurotransmitter transporter family SLC6. Pflügers Arch 447:519–531 Chesler M (2003) Regulation and modulation of pH in the brain. Physiol Rev 83:1183–1221 Chinopoulos C, Tretter L, Rozsa A, Adam-Vizi V (2000) Exacerbated responses to oxidative stress by an Na+ load in isolated nerve terminals: the role of ATP depletion and rise of [Ca2+ ]i . J Neurosci 20:2094–2103 Chiu SY, Kriegler S (1994) Neurotransmitter-mediated signalling between axons and glial cells. Glia 11:191–200 Coutinho V, Knöpfel T (2002) Metabotropic glutamate receptors: electrical and chemical signaling properties. Neuroscientist 8:551–561 Craner MJ, Hains BC, Lo AC, Black JA, Waxman SG (2004a) Co-localization of sodium channel Nav1.6 and the sodium-calcium exchanger at sites of axonal injury in the spinal cord in EAE. Brain 127:294–303 Craner MJ, Newcombe J, Black JA, Hartle C, Cuzner ML, Waxman SG (2004b) Molecular changes in neurons in multiple sclerosis: altered axonal expression of Nav1.2 and Nav1.6 sodium channels and Na+ /Ca2+ exchanger. PNAS 101:8168–8173 Danbolt NC (2001) Glutamate uptake. Prog Neurobiol 65:1–105 David G, Barrett JN, Barrett EF (1997) Spatiotemporal gradients of intra-axonal [Na+ ] after transection and resealing in lizard peripheral myelinated axons. J Physiol 498:295–307 Drew GM, Mitchell VA, Vaughan CW (2008) Glutamate spillover modulates GABAergic synaptic transmission in the rat midbrain periaqueductal grey via metabotropic glutamate receptors and endocannabinoid signalling. J Neurosci 28:808–815 Duguid I, Sjöström PJ (2006) Novel presynaptic mechanisms for coincidence detection in synaptic plasticity. Curr Opin Neurobiol 16:312–322 Erecinska M, Dagani F, Nelson D, Deas J, Silver IA (1991) Relations between intracellular ions and energy metabolism: a study with monensin in synaptosomes, neurons, and C6 glioma cells. J Neurosci 11:2410–2421 Fleidervish IA, Lasser-Ross N, Gutnick MJ, Ross WN (2010) Na+ imaging reveals little difference in action potential-evoked Na+ influx between axon and soma. Nat Neurosci doi:10.1038/nn.2574 Fowler JC, Li Y (1998) Contributions of Na+ flux and the anoxic depolarization to adenosine 5’-triphosphate levels in hypoxic/hypoglycemic rat hippocampal slices. Neuroscience 83: 717–722 Grynkiewicz G, Poenie M, Tsien RY (1985) A new generation of Ca2+ indicators with greatly improved fluorescence properties. J Biol Chem 260:3440–3450 Hains BC, Saab CY, Lo AC, Waxman SG (2004) Sodium channel blockade with phenytoin protects spinal cord axons, enhances axonal conduction, and improves functional motor recovery after contusion SCI. Exp Neurol 188:365–377 Haloui M, Taurin S, Akimova OA, Guo DF, Tremblay J, Dulin NO, Hamet P, Orlov SN (2007) [Na+ ]i -induced c-Fos expression is not mediated by activation of the 5’-promoter containing known transcriptional elements. FEBS J 274:3557–3567 Harris KM, Kater SB (1994) Dendritic spines: cellular specializations imparting both stability and flexibility to synaptic function. Ann Rev Neurosci 17:341–371 Hille B (2001) Ionic channels of excitable membranes, 3rd ed. Sinauer Associates, Sunderland Hodgkin AL, Huxley AF (1952) A quantitative description of membrane current and its application to conduction and excitation in nerve. J Physiol 117:500–544 Imaizumi T, Kocsis JD, Waxman SG (1997) Anoxic injury in the rat spinal cord: pharmacological evidence for multiple steps in Ca2+ -dependent injury of the dorsal columns. J Neurotrauma 14:299–311

3

Sodium Signals and Their Significance for Axonal Function

51

Jaffe DB, Johnston D, Lasser-Ross N, Lisman JE, Miyakawa H, Ross WN (1992) The spread of Na+ spikes determines the pattern of dendritic Ca2+ entry into hippocampal neurons. Nature 357:244–246 Kanai Y, Hediger M (2004) The glutamate/neutral amino acid transporter family SLC1: molecular, physiological and pharmacological aspects. Pflügers Arch 447:469–479 Káradóttir R, Attwell D (2007) Neurotransmitter receptors in the life and death of oligodendrocytes. Neuroscience 145:1426–1438 Kelly T, Kafitz KW, Roderigo C, Rose CR (2009) Ammonium-evoked alterations in intracellular sodium and pH reduce glial glutamate transport activity. Glia 57:921–934 Kim JH, Sizov I, Dobretsov M, von Gersdorff H (2007) Presynaptic Ca2+ buffers control the strength of a fast post-tetanic hyperpolarization mediated by the α3 Na+ /K+ -ATPase. Nat Neurosci 10:196–205 Kim MH, Korogod N, Schneggenburger R, Ho WK, Lee SH (2005) Interplay between Na+ /Ca2+ exchangers and mitochondria in Ca2+ clearance at the calyx of Held. J Neurosci 25: 6057–6065 Knöpfel T, Anchisi D, Alojado ME, Tempia F, Strata P (2000) Elevation of intradendritic sodium concentration mediated by synaptic activation of metabotropic glutamate receptors in cerebellar Purkinje cells. Eur J Neurosci 12:2199–2204 Koh DS, Jonas P, Vogel W (1994) Na+ -activated K+ channels localized in the nodal region of myelinated axons of Xenopus. J Physiol 479:183–197 Kole MH, Ilschner SU, Kampa BM, Williams SR, Ruben PC, Stuart GJ (2008) Action potential generation requires high sodium channel density in the axon initial segment. Nat Neurosci 11:178–186 Kriegler S, Chiu SY (1993) Calcium signaling of glial cells along mammalian axons. J Neurosci 13:4229–4245 Langer J, Rose CR (2009) Synaptically induced sodium signals in hippocampal astrocytes in situ. J Physiol 587:5859–5877 Lasser-Ross N, Ross WN (1992) Imaging voltage and synaptically activated sodium transients in cerebellar Purkinje cells. Proc R Soc B Biol Sci 247:35–39 Leppanen L, Stys PK (1997) Ion transport and membrane potential in CNS myelinated axons II. Effects of metabolic inhibition. J Neurophysiol 78:2095–2107 Lev-Ram V, Grinvald A (1987) Activity-dependent calcium transients in central nervous system myelinated axons revealed by the calcium indicator Fura-2. Biophys J 52:571–576 Li S, Stys PK (2001) Na+ -K+ -ATPase inhibition and depolarization induce glutamate release via reverse Na+ -dependent transport in spinal cord white matter. Neuroscience 107:675–683 Li S, Jiang Q, Stys PK (2000) Important role of reverse Na+ -Ca2+ exchange in spinal cord white matter injury at physiological temperature. J Neurophysiol 84:1116–1119 Li S, Mealing GAR, Morley P, Stys PK (1999) Novel injury mechanism in anoxia and trauma of spinal cord white matter: glutamate release via reverse Na+ -dependent glutamate transport. J Neurosci 19:16RC Linden DJ, Smeyne M, Connor JA (1993) Induction of cerebellar long-term depression in culture requires postsynaptic action of Sodium Ions. Neuron 11:1093–1100 Lipton P (1999) Ischemic cell death in brain neurons. Physiol Rev 79:1431–1568 Lopachin RM, Stys PK (1995) Elemental composition and water content of rat optic nerve myelinated axons and glial cells: effects of in vitro anoxia and reoxygenation. J Neurosci 15:6735–6746 Lucas JH, Emery DG, Rosenberg LJ (1997) Physical injury of neurons: important roles for sodium and chloride ions. Neuroscientist 3:89–101 Magee JC, Johnston D (1997) A synaptically controlled, associative signal for hebbian plasticity in hippocampal neurons. Science 275:209–213 Meier SD, Kovalchuk Y, Rose CR (2006) Properties of the new fluorescent Na+ indicator CoroNa Green: comparison with SBFI and confocal Na+ imaging. J Neurosci Methods 155: 251–259

52

T. Kelly and C.R. Rose

Minta A, Tsien RY (1989) Fluorescent indicators for cytosolic sodium. J Biol Chem 264: 19449–19457 Morita K, David G, Barrett JN, Barrett EF (1993) Posttetanic hyperpolarization produced by electrogenic Na+ -K+ pump in lizard axons impaled near their motor terminals. J Neurophysiol 70:1874–1884 Mulkey RM, Zucker RS (1992) Posttetanic potentiation at the crayfish neuromuscular junction is dependent on both intracellular calcium and sodium ion accumulation. J Neurosci 12: 4327–4336 Nikolaeva MA, Mukherjee B, Stys PK (2005) Na+ -Dependent sources of intra-axonal Ca2+ release in rat optic nerve during in vitro chemical ischemia. J Neurosci 25:9960–9967 Nuriya M, Jiang J, Nemet B, Eisenthal KB, Yuste R (2006) Imaging membrane potential in dendritic spines. PNAS 103:786–790. O’Shea RD (2002) Roles and regulation of glutamate transporters in the central nervous system. Clin Exp Pharmacol Physiol 29:1018–1023 Orlov SN, Hamet P (2006) Intracellular monovalent ions as second messengers. J Membr Biol 210:161–172 Otis TS, Brasnjo G, Dzubay JA, Pratap M (2004) Interactions between glutamate transporters and metabotropic glutamate receptors at excitatory synapses in the cerebellar cortex. Neurochem Int 45:537–544 Ransom BR, Walz W, Davis PK, Carlini WG (1992) Anoxia-induced changes in extracellular K+ and pH in mammalian central white matter. J Cereb Blood Flow Metab 12:593–602 Regehr WG (1997) Interplay between sodium and calcium dynamics in granule cell presynaptic terminals. Biophys J 73:2476–2488 Ren Y, Ridsdale A, Coderre E, Stys PK (2000) Calcium imaging in live rat optic nerve myelinated axons in vitro using confocal laser microscopy. J Neurosci Methods 102:165–176 Reuter H, Porzig H (1995) Localization and functional significance of the Na+ /Ca2+ exchanger in presynaptic boutons of hippocampal cells in culture. Neuron 15:1077–1084 Rishal I, Keren-Raifman T, Yakubovich D, Ivanina T, Dessauer CW, Slepak VZ, Dascal N (2003) Na+ promotes the dissociation between Gα GDP and Gβγ, activating G protein-gated K+ channels. J Biol Chem 278:3840–3845 Rojas H, Colina C, Ramos M, Benaim G, Jaffe EH, Caputo C, DiPolo R (2007) Na+ entry via glutamate transport activates the reverse Na+ /Ca2+ exchange and triggers Ca2+ i -induced Ca2+ release in rat cerebellar Type-1 astrocytes J Neurochem 100:1188–1202 Rose CR (2002) Na+ signals at central synapses. Neuroscientist 8:532–539 Rose CR (2003) High-resolution Na+ imaging in dendrites and spines. Pflügers Arch 446:317–321 Rose CR, Konnerth A (2001) NMDA receptor-mediated Na+ signals in spines and dendrites. J Neurosci 21:4207–4214 Rose CR, Kovalchuk Y, Eilers J, Konnerth A (1999) Two-photon Na+ imaging in spines and fine dendrites of central neurons. Pflügers Arch 439:201–207 Rose CR, Ransom BR (1996) Mechanisms of H+ and Na+ changes induced by glutamate, kainate, and D-aspartate in rat hippocampal astrocytes. J Neurosci 16:5393–5404 Salter MG, Fern R (2008) The mechanisms of acute ischemic injury in the cell processes of developing white matter astrocytes. J Cereb Blood Flow Metab 28:588–601 Schenk U, Matteoli M (2004) Presynaptic AMPA receptors: more than just ion channels? Biol Cell 96:257–260 Schiller J, Helmchen F, Sakmann B (1995) Spatial profile of dendritic calcium transients evoked by action potentials in rat neocortical pyramidal neurones. J Physiol 487:583–600 Scotti AL, Chatton Y, Reuter H (1999) Roles of Na+ -Ca2+ exchange and of mitochondria in the regulation of presynaptic Ca2+ and spontaneous glutamate release. Philos Trans R Soc B Biol Sci 354:357–364 Sheldon C, Diarra A, Cheng YM, Church J (2004) Sodium influx pathways during and after anoxia in rat hippocampal neurons. J Neurosci 24:11057–11069

3

Sodium Signals and Their Significance for Axonal Function

53

Somjen GG (2002) Ion regulation in the brain: implications for pathophysiology. Neuroscientist 8:254–267 Steffensen I, Waxman SG, Mills L, Stys PK (1997) Immunolocalization of the Na+ -Ca2+ exchanger in mammalian myelinated axons. Brain Res 776:1–9 Stys PK (1998) Anoxic and ischemic injury of myelinated axons in CNS white matter: from mechanistic concepts to therapeutics. J Cereb Blood Flow Metab 18:2–25 Stys PK (2004) White matter injury mechanisms. Curr Mol Med 4:113–130 Stys PK (2005) General mechanisms of axonal damage and its prevention. J Neurol Sci 233:3–13 Stys PK, Hubatsch DA, Leppanen LL (1998) Effects of K+ channel blockers on the anoxic response of CNS myelinated axons. Neuroreport 9:447–454 Stys PK, Lopachin RM (1996) Elemental composition and water content of rat optic nerve myelinated axons during in vitro post-anoxia reoxygenation. Neuroscience 73:1081–1090 Stys PK, Ransom BR, Waxman SG, Davis PK (1990) Role of extracellular calcium in anoxic injury of mammalian central white matter. PNAS 87:4212–4216 Stys PK, Waxman SG, Ransom BR (1992) Ionic mechanisms of anoxic injury in mammalian CNS white matter: role of Na+ channels and Na+ -Ca2+ exchanger. J Neurosci 12:430–439 Sun BB, Chiu SY (1999) N-type calcium channels and their regulation by GABAB receptors in axons of neonatal rat optic nerve. J Neurosci 19:5185–5194 Szapiro G, Barbour B (2007) Multiple climbing fibers signal to molecular layer interneurones exclusively via glutamate spillover. Nat Neurosci 10:735–742 Tretter L, Adam-Vizi V (2002) Glutamate release by an Na+ load and oxidative stress in nerve terminals: relevance to ischemia/reperfusion. J Neurochem 83:855–862 Verbny Y, Zhang CL, Chiu SY (2002) Coupling of calcium homeostasis to axonal sodium in axons of mouse optic nerve. J Neurophysiol 88:802–816 Wächtler J, Mayer C, Grafe P (1998) Activity-dependent intracellular Ca2+ transients in unmyelinated nerve fibres of the isolated adult rat vagus nerve. Pflügers Arch 435:678–686 Wadiche JI, Jahr CE (2005) Patterned expression of Purkinje cell glutamate transporters controls synaptic plasticity. Nat Neurosci 8:1329–1334 Waxman SG (2006) Axonal conduction and injury in multiple sclerosis: the role of sodium channels. Nat Rev Neurosci 7:932–941 Waxman SG, Black JA, Ransom BR, Stys PK (1993) Protection of the axonal cytoskeleton in anoxic optic nerve by decreased extracellular calcium. Brain Res 614:137–145 Waxman SG, Ritchie MJ (1993) Molecular dissection of the myelinated axon. Ann Neurol 33: 121–136 Yu XM (2006) The role of intracellular sodium in the regulation of NMDA-receptor-mediated channel activity and toxicity. Mol Neurobiol 33:63–80 Yu XM, Salter MW (1998) Gain control of NMDA-receptor currents by intracellular sodium. Nature 396:469–474 Yu XM, Salter MW (1999) Src, a molecular switch governing gain control of synaptic transmission mediated by N-methyl-D-aspartate receptors. PNAS 96:7697–7704 Yuan Q, Knöpfel T (2006) Olfactory nerve stimulation-evoked mGluR1 slow potentials, oscillations, and calcium signaling in mouse olfactory mulb mitral cells. J Neurophysiol 95: 3097–3104 Zhang CL, Wilson JA, Williams J, Chiu SY (2006) Action potentials induce uniform calcium influx in mammalian myelinated optic nerves. J Neurophysiol 96:695–709

Chapter 4

New Insights in Information Processing in the Axon Dominique Debanne and Sami Boudkkazi

4.1 Introduction The axon is defined as a long neuronal process that insures the conduction of information from the cell body to the nerve terminal. Generally, axons are highly ramified and contact several hundreds of target neurons locally or distally. But, the function of the axon is not purely limited to the conduction of the action potential from the site of initiation near the cell body to the terminal. Recent experimental findings shed new light on the functional and computational capabilities of single axons, suggesting that several different complex operations are specifically achieved along the axon. Axons integrate subthreshold synaptic potentials and therefore signal both analog and digital events. Drop of conduction or backward propagation (reflection) may occur at specific axonal branch points under certain conditions. Axonal geometry together with the biophysical properties of axonal voltage-gated channels determines the timing of propagation of the output message in different axonal branches. In addition, axons link central neurons through gap junctions that allow ultra-fast network synchrony. Moreover, local shaping of the axonal action potential may subsequently determine synaptic efficacy during repetitive stimulation. These operations have been largely described in in vitro preparations of brain tissue but evidence for these processes is still scarce in the mammalian brain in vivo. In this chapter we review the different ways in which the properties of axons can control the transmission of electrical signals. We will start with a brief discussion of the basic characteristics of propagation and how voltage-gated channels that are present in the axon shape the amplitude and duration of the action potential. Then, we will consider two novel operations achieved by the axonal membrane: signal amplification and integration and will review how axonal geometry and channel distribution can affect propagation: by introducing delays in conduction, by causing propagation failures, and by causing the reflection of the action potential.

D. Debanne (B) INSERM U641, Marseille, France; Faculté de médecine secteur nord, Université de la Méditerranée, Marseille, France e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_4, 

55

56

D. Debanne and S. Boudkkazi

4.2 Complexity of Axonal Arborization: Branch Points and Varicosities Axonal morphology is highly variable. Some axons extend locally (about 1 mm long for inhibitory interneurons) whereas others may be as long as 1 m and more. The diameter of axons varies considerably. The largest axon (squid giant axon) has a diameter close to 1 mm (Young, 1936) whereas the diameter of unmyelinated cortical axons varies between 0.08 and 0.4 μm (Westrum and Blackstad, 1962; Berbel and Innocenti, 1988). The complexity of axonal arborization is also variable. In one extreme, the cerebellar granule cell axon possesses a single T-shaped branch point that gives rise to the parallel fibers. On the other, many axons typically form an elaborate and most impressive tree. For instance, the terminal arbor of thalamocortical axons in layer IV of the cat visual cortex contains 150–275 branch points (Antonini et al., 1998). The complexity of axonal arborization is also extensive in cortical pyramidal neurons. Axons of hippocampal CA3 pyramidal cells display at least 100–200 branch points for a total axonal length of 150–300 mm, and a single cell may contact 30,000–60,000 neurons (Ishizuka et al., 1990; Major et al., 1994; Li et al., 1994). GABAergic interneurons also display complex axons. Hippocampal and cortical inhibitory interneurons emit an axon with a very dense and highly branched arborization (Gulyas et al., 1993). One obvious function of axonal divergence is to allow synchronous transmission to a wide population of target neurons within a given brain area. Thus, hippocampal basket cells synchronize the firing of principal cells through their divergent axon (Cobb et al., 1995). The second morphological feature of axons is the presence of a large number of varicosities (synaptic boutons) that are commonly distributed in an en passant, “beads-on-a-string” manner along thin axon branches. A single axon may contain several thousands of boutons (Gulyas et al., 1993; Li et al., 1994; Pinault and Deschênes, 1998). Their size varies between ∼1 μm for thin unmyelinated axons (Westrum and Blackstad, 1962; Shepherd et al., 2002) and 3–5 μm for large hippocampal mossy fiber terminals (Blackstad and Kjaerheim, 1961; Shepherd et al., 2002). Their density varies among axons and the spacing of varicosities is comprised between ∼4 and ∼6 μm in unmyelinated fibers (Shepherd et al., 2002; Shepherd and Raastad, 2003).

4.3 Functional Computation in the Axon 4.3.1 Shaping Action Potential The shape of the presynaptic action potential is of fundamental importance in determining the strength of synapses by modulating transmitter release. The waveform of the depolarization dictates the calcium signal available to trigger vesicle fusion by controlling the opening of voltage-gated calcium channels and the driving force for calcium influx. Two types of modification of the presynaptic action potential

4

New Insights in Information Processing in the Axon

57

have been reported experimentally: modifications of action potential width and/or modifications of action potential amplitude. 4.3.1.1 Activity-Dependent Broadening of Presynaptic Action Potential The duration of the presynaptic spike is not fixed and activity-dependent short-term broadening of the spike has been observed in en passant mossy fiber boutons (Geiger and Jonas, 2000). The mossy fiber-CA3 pyramidal cell synapse displays fast and synchronized transmitter release from several active zones and also shows dynamic changes in synaptic strength over a more than 10-fold range. The exceptionally large synaptic facilitation is in clear contrast with the weak facilitation (∼150%) observed at most central synapses. Granule cell axons exhibit several voltage-gated potassium channels including Kv1.1 (Rhodes et al., 1997), Kv1.2 (Sheng et al., 1993; Inda et al., 2006), and two A-type potassium channels, Kv1.4 (Sheng et al., 1992; Veh et al., 1995; Cooper et al., 1998) and Kv3.4 (Veh et al., 1995). Geiger and Jonas have shown that the action potential at the mossy fiber terminal is half as wide as that at the soma. During repetitive stimulation, the action potential gets broader in the axon terminal but not in the soma (Geiger and Jonas, 2000; Fig. 4.1). More interestingly, using simultaneous recordings from the granule cell terminal and the corresponding postsynaptic apical dendrite of a CA3 neuron Geiger and Jonas showed that action potential broadening enhanced presynaptic calcium

a

mfb

Granule cell

c

b

Voltage commands 1stAP th 10 AP th 50 AP

Synaptic currents

Fig. 4.1 Shaping of the action potential in the axon. a. A mossy fiber terminal is recorded in the whole-cell configuration and activated at a frequency of 50 Hz. b. During repetitive stimulation of the axon, the action potential becomes wider. The 10th and 15th action potentials are compared with the first action potential in the train. c. Action potential broadening potentiates transmitter release. Mossy fiber terminal (red) and the corresponding CA3 cell (blue) were recorded simultaneously. Action potential waveforms were imposed at the presynaptic terminal. The increased duration of the waveform incremented the amplitude of the synaptic current. Adapted from Geiger and Jonas (2000)

58

D. Debanne and S. Boudkkazi

influx and doubled the EPSC amplitude (Fig. 4.1). This broadening results from the inactivation of A-type K+ channels located in the membrane of the terminal. Consequently, the pronounced short-term facilitation probably results from the conjugated action of spike widening and the classical accumulation of residual calcium in the presynaptic terminal. Because ultrastructural analysis reveals A-type channel immunoreactivity not only in the terminal but also in the axonal membrane (Cooper et al., 1998), activity-dependent spike broadening might also occur in the axon.

4.3.1.2 Activity-Dependent Reduction of Presynaptic Action Potential Reduction of the amplitude of the presynaptic action potential has been reported following repetitive stimulation of invertebrate (Grossman et al., 1979a) or mammalian axons (Geiger and Jonas, 2000; Wang and Kazmarek, 1998). This decline results from sodium channel inactivation and can be amplified by low concentrations of TTX (Madeja, 2000; Brody and Yue, 2000). The consequences of sodium channel inactivation on synaptic transmission have been studied at various central synapses. Interestingly, the reduction of the sodium current by application of TTX in the nanomolar range decreases glutamatergic transmission and enhances short-term depression (Brody and Yue, 2000; Prakryia and Mennerick, 2000; He et al., 2002). In addition, the depolarization of the presynaptic terminal by raising the external potassium concentration increases paired-pulse depression at autaptic contacts of cultured hippocampal cells (He et al., 2002) and decreases paired-pulse facilitation at Schaffer collateral-CA1 synapses stimulated extracellularly (Meeks and Mennerick, 2004). In this case, the depolarization of the presynaptic axons is likely to enhance presynaptic spike attenuation. Importantly, inactivation of sodium channels by high external potassium increases the proportion of conduction failures during repetitive extracellular stimulation of Schaffer collateral axons (Meeks and Mennerick, 2004). However, these results must be interpreted carefully because apparent changes in paired-pulse ratio may simply be the result of stimulation failures produced by the reduction in presynaptic axon excitability. Interestingly, the manipulations of the sodium current mentioned above have little or no effect on GABAergic axons (Prakriya and Mennerick, 2000; He et al., 2002; Meeks and Mennerick, 2004). Riluzole, TTX, or external potassium affect neither GABAergic synaptic transmission nor short-term GABAergic plasticity. This difference between glutamatergic and GABAergic axons might result from several factors. Sodium currents in interneurons are less sensitive to inactivation and a slow recovery from inactivation has been observed for pyramidal cells but not for inhibitory interneurons (Martina and Jonas, 1997). Moreover, the density of sodium current is higher in interneurons than in pyramidal neurons (Martina et al., 2000). Thus, axons of GABAergic interneurons could be better cables for propagation than those of pyramidal cells (Forti et al., 2000; Tan and Llano, 2000). This unusual property could be important functionally: safe propagation along inhibitory axons could protect the brain from sporadic hyperactivity and prevent the development of epileptiform activity.

4

New Insights in Information Processing in the Axon

59

4.3.2 Signal Amplification Along the Axon Signal amplification is classically considered to be a neuronal operation achieved by the dendritic membrane and the cell body (Stuart and Sakmann, 1995). Whereas action potential propagation along the axon is clearly an active process that depends upon a high density of sodium channels, the process of action potential invasion into presynaptic terminals was, until recently, less well understood. This question is of primary importance because the geometrical perturbation introduced by the presynaptic terminal decrease the safety of action potential propagation and may affect the conduction time (see Sections 4.4, 4.5, and 4.6). The invasion of the spike is active at the amphibian neuromuscular junction, (Katz and Miledi, 1965) but passive at the neuromuscular junction of the mouse (Brigant and Mallart, 1982; Dreyer and Penner, 1987) and the lizard (Lindgren and Moore, 1989) and at the calyx of Held (Leao et al., 2005). This question has been recently reconsidered at hippocampal mossy fiber boutons (Engel and Jonas, 2005). In this study, Engel and Jonas showed that sodium channel density is very high at the presynaptic terminal (2,000 channels per mossy fiber bouton). In addition, sodium channels in mossy fiber boutons activate and inactivate with submillisecond kinetics. A realistic computer simulation indicates that the density of sodium channels found in the mossy fiber bouton not only amplifies the action potential but also slightly increases the conduction speed along the axon (Engel and Jonas, 2005, see also Section 4.4). It will be important to know whether different structural properties (en passant in the case of mossy fiber bouton versus terminal bouton in the case of the calyx of Held) may explain the very different effects of sodium channels at the synapses. Another mechanism of activity-dependent signal amplification has been reported at hippocampal mossy fibers (Nakamura et al., 2007). In immature hippocampus, repetitive stimulation of the mossy fiber pathway facilitates not only the synaptic transmission but also the amplitude of the presynaptic volley, the electrophysiological signature of the presynaptic action potential in the axon recorded extracellularly. This axonal facilitation is not observed in mature hippocampus. It is associated with the depolarization of mossy fibers and fully inhibited by GABAA receptor antagonists, indicating that GABA released from interneurons depolarizes the axon and increases its excitability. Because the presynaptic axon was not directly recorded in this study, further investigations will be necessary to determine whether GABAA receptor depolarization limits conduction failures or interacts with sodium channel amplification.

4.3.3 Axonal Integration Classically, the somato-dendritic compartment is considered as the locus of neuronal integration where subthreshold electrical signals originating from active synapses are temporally summated to control the production of an output message, the action potential. In this view, the axon initial segment is the final site of synaptic integration

60

D. Debanne and S. Boudkkazi

and the axon remains purely devoted to action potential conduction in a digital way. Synaptic strength can be modulated by the frequency of presynaptic action potential firing. Today, this view is now challenged by accumulating evidence showing that the axon is also able to integrate electrical signals arising from the somato-dendritic compartment of the neuron (for a review see Marder, 2006; Clark and Häusser, 2006). The story started with classic observations reported in Aplysia (Shimahara and Tauc, 1975; Shapiro et al., 1980) and the leech (Nicholls and Wallace, 1978) where, in connected pairs of neurons, the membrane potential of the presynaptic cell was found to control the efficacy of action potential-triggered synaptic transmission. In fact, constant depolarization of the soma of the presynaptic neuron facilitates synaptic transmission in a graded manner (Fig. 4.2a). Thus, the membrane potential at the cell body determines, in an analog manner, the efficacy of the digital output message (the action potential). The underlying mechanism in Aplysia neurons involves the activation of steady-state Ca2+ currents (Shapiro et al., 1980) and the overcome of propagation failures in a weakly excitable region of the neuron (Evans et al., 2003).

a

Fig. 4.2 Axonal integration. a. Graded control of synaptic efficacy by the membrane potential in a pair of connected Aplysia neurons. The hyperpolarization of the presynaptic neuron gradually reduces the amplitude of the synaptic potential. Adapted from Shimahara and Tauc (1975). b. Integration of subthreshold synaptic potential in the axon of hippocampal granule cells. Electrically evoked synaptic inputs in the dendrites of a granule cell can be detected in the mossy fiber terminal (EPreSP). Lower panel, synaptic transmission at the mossy fiber synapse was facilitated when the simulated EPreSP (“EPreSP”) was associated with a presynaptic action potential (AP + “EPreSP”). Adapted with permission from Alle and Geiger (2006)

b Granule cell mfb CA3-P EPreSP 2 mV

mfb 100 ms

‘EPreSP’ + AP ‘EPreSP’ AP mfb CA3-P

20 mV

100 pA 100 ms

4

New Insights in Information Processing in the Axon

61

This behaviour is not peculiar to invertebrate neurons and similar observations have been reported in at least three central synapses of the mammalian brain. First, at synapses established between pairs of CA3–CA3 pyramidal cells, steady-state hyperpolarization of the presynaptic neuron reduces the synaptic efficacy (Saviane et al., 2003). More recently a study published by Alle and Geiger (2006) shows, by using direct patch-clamp recording from presynaptic hippocampal mossy fiber boutons, that granule cell axons transmit analog signals in addition to action potentials. In fact, excitatory synaptic potentials evoked by local stimulation of the molecular layer in the dentate gyrus can be detected in the mossy fiber bouton located at several hundreds microns from the cell body (Fig. 4.2b). As expected from cable theory, this signal is attenuated and the EPSP waveform is much slower in the terminal than in the soma of granule cells. The salient finding here is that the space constant of the axon is much wider (∼400 μm) than initially expected. Consistent with propagation of electrical signals over very long distances, the analog facilitation of synaptic transmission has a slow time constant (2–4 s, Nicholls and Wallace, 1978; Shu et al., 2006). The functional consequence is that slow depolarizations of the membrane in somatic and dendritic regions are transmitted to the axon terminals and can influence the release of transmitter at the mossy fiber-CA3 cell synapse. Similar observations have been reported in the axon of L5 cortical pyramidal neurons recorded in wholecell configuration at distances ranging between 90 and 400 μm (Shu et al., 2006). In this case, whole-cell recording is possible on the axon because sectioning the axon produces a small enlargement of its diameter that allows positioning of a patch pipette. Here again incoming synaptic activity in the presynaptic neuron propagates down the axon and can modulate the efficacy of synaptic transmission. The modulation of synaptic efficacy by somatic potential is blocked by Ca2+ chelators (Alle and Geiger, 2006) and may therefore result from the control of background calcium levels at the presynaptic terminal (Awatramani et al., 2005). Recently, the mechanism controlling the voltage-dependent broadening of the axonal action potential has been identified in L5 pyramidal neurons (Shu et al., 2007; Kole et al., 2007). Kv1 channels are expressed at high densities in the axon initial segment (AIS; see Inda et al., 2006). With cell-attached recordings from the axon, Kole and co-workers elegantly showed that Kv1 channel density increased 10fold over the first 50 μm of the AIS. The axonal current mediated by Kv1 channels inactivates with a time constant in the second range (Shu et al., 2007; Kole et al., 2007). Pharmacological blockade or voltage inactivation of Kv1 channels produce a distance-dependent broadening of the axonal spike, as wells as an increase in synaptic strength at proximal axonal terminals (Kole et al., 2007). Thus, Kv1 channels occupy a strategic position to integrate slow subthreshold signals generated in the dendro-somatic region and control the presynaptic action potential waveform in order to finely tune synaptic coupling in local cortical circuits. The role of axonal membrane compartment is also critical in synaptic integration. The group of Alain Marty recently showed by cutting the axon of cerebellar interneurons with two-photon illumination that the axonal membrane speeds up the decay of synaptic potentials (Mejia-Gervacio et al., 2007). This effect is mainly due to passive membrane properties and the axonal compartment acts as a sink for

62

D. Debanne and S. Boudkkazi

fast synaptic currents. Functionally, axonal speeding has important consequences because it increases temporal precision of EPSP-spike coupling. The functional importance of axonal integration is clear but there are many questions left open. The three examples where hybrid (analog–digital) signaling in the axon has been observed are glutamatergic neurons [CA3 pyramidal neuron (Saviane et al., 2003), granule cell (Alle and Geiger, 2006), and L5 pyramidal neuron (Shu et al., 2006; Kole et al., 2007)]. Do axons of GABAergic interneurons also express hybrid axonal signaling? In dendrites, voltage-gated channels amplify or attenuate subthreshold EPSPs. Do axonal voltage-gated channels also influence propagation of subthreshold potentials? Finally can inhibitory postsynaptic potentials spread down the axon, and if so, how do they influence synaptic release? Now that the axons of mammalian neurons are finally becoming accessible to direct electrophysiological recording, we can expect answers to all these questions.

4.4 Propagation Failures One of the more unusual operations achieved by axons is selective conduction failure. When the action potential fails to propagate along the axon, no signal can reach the output of the cell. Conduction failure represents a powerful process that filters communication with postsynaptic neurons. Propagation failures have been observed experimentally in various axons including vertebrate spinal axons (Barron and Matthews, 1939; Krnjevic and Miledi, 1959), spiny lobster or crayfish motoneurons (Parnas, 1972; Hatt and Smith, 1976; Grossman et al., 1979a; Smith, 1980a), leech mechanosensory neurons (Van Essen, 1973; Yau, 1976; Gu, 1991; Baccus, 1998; Baccus et al., 2000), thalamocortical axons (Deschênes and Landry, 1980) rabbit nodose ganglion neurons (Ducreux et al., 1993), rat dorsal root ganglion neurons (Lüscher et al., 1994a,b), neurohypophysial axons (Dyball et al., 1988; Bielefeldt and Jackson, 1993), and hippocampal pyramidal cell axons (Debanne et al., 1997; Soleng et al., 2003; Meeks and Mennerick, 2004). However, some axons in the auditory pathways are capable of sustaining remarkably high firing rates, with perfect entrainment occurring at frequencies of up to 1 kHz (Scott et al., 2007). Several factors determine whether propagation along axons fails or succeeds.

4.4.1 Geometrical Factors: Branch Points and Swellings Although the possibility that propagation may fail at branch points was already discussed by Krnjevic and Miledi (1959), the first clear indication that propagation is perturbed by axonal branch points came from the early studies on spiny lobster, crayfish, and leech axons (Parnas, 1972; Van Essen, 1973; Yau, 1976; Grossman et al., 1979a, b; Smith, 1980a, b). The large size of invertebrate axons allowed multielectrode recordings upstream and downstream of the branch point. For example, in lobster axons, conduction across the branch point was found to fail at frequencies

4

New Insights in Information Processing in the Axon

63

a Lateral

Axone

b

Medial

Soma Axon

Fig. 4.3 Propagation failures in invertebrate neurons. a. Propagation failure at a branch point in a lobster axon. Adapted from Grossman et al. (1979a). b. Propagation failure at the junction between an axonal branch and the soma of a snail neuron (metacerebral cell). The propagation in the axonal arborization was analyzed by the local fluorescence transients due to the action potential. The recording region is indicated by an outline of a subset of individual detectors, superimposed over the fluorescence image of the neuron in situ. When the action potential was evoked by direct stimulation of the soma, it propagated actively in all axonal branches (red traces). In contrast, when the action potential was evoked by the synaptic stimulation (EPSP) of the right axonal branch (Br1), the amplitude of the fluorescent transient declined when approaching the cell body, indicating a propagation failure (black traces). Adapted with permission from Antic et al. (2000)

above 30 Hz (Fig. 4.3a, Grossman et al., 1979a). The block of conduction occurred specifically at the branch point because the parent axon and one of the daughter branches continued to conduct action potentials. Failures appeared first in the thicker daughter branch but they could be also observed in the thin branch at higher stimulus frequency. In the leech, conduction block occurs at central branch points where fine axons from the periphery meet thicker axons (Yau, 1976). Branch point failures have been observed or suspected to occur in a number of mammalian neurons (Deschênes and Landry, 1980; Ducreux et al., 1993; Debanne et al., 1997). Propagation failures also occur when the action potential enters a zone with an abrupt change in diameter. This occurs with en passant boutons (Bourque, 1990; Zhang and Jackson, 1993; Jackson and Zhang, 1995) but also when impulses propagating along the axon enter the soma (Lüscher et al., 1994a). For instance, in the megacerebral cell of the snail, propagation failures have been observed when a spike enters the cell body (Fig. 4.3b; Antic et al., 2000). These failures result because the electrical load is significantly higher on the arriving action potential and the current generated by the parent axon is not sufficient to allow the propagation process (reviewed in Segev and Schneidman, 1999). Simulations show that

64

D. Debanne and S. Boudkkazi

at geometrical irregularities the propagating action potential is usually distorted in amplitude and width, and local conduction velocity can be changed. For instance, an abrupt increase in axon diameter causes a decrease in both velocity and peak amplitude of the action potential whereas a step decrease in diameter has the opposite local effects on these two parameters (Goldstein and Rall, 1974; Parnas et al., 1976; Lüscher and Shiner, 1990a, b; Manor et al., 1991; Graham and Redman, 1994; Jackson and Zhang, 1995). In fact, the interplay between the total longitudinal current produced by the action potential and the input impedance of the axon segments ahead of the action potential determines the fate of the propagating action potential. The case of the branch point has been studied in detail (Goldstein and Rall, 1974; Goldfinger, 2000; Zhou and Chiu, 2001). The so-called 3/2 power law developed by Rall describes an ideal relationship between the geometry of mother and daughter branches (Rall, 1959; 1964; Goldstein and Rall, 1974). A geometrical parameter (the geometrical ratio, GR) has been defined as follows: GR = (d3/2 daughter 1 + d3/2 daughter 2 )/d3/2 mother where ddaughter 1 and ddaughter 2 are the diameters of the daughter branches and dmother is the diameter of the parent axon. For GR = 1, impedances match perfectly and spikes propagate in both branches. If GR>1, the combined electrical load of the daughter branches exceeds the load of the main branch. In other words, the active membrane of the mother branch may not be able to provide enough current to activate both branches. For 1
4.4.2 Frequency-Dependent Propagation Failures Conduction failures are encountered following moderate (10–50 Hz) or highfrequency (200–300 Hz) stimulation of the axon. For instance, a frequency of 20–30 Hz is sufficient to produce conduction failures at the neuromuscular terminal arborization (Krnjevic and Miledi, 1959) or at the branch point of spiny lobster motoneurons (Grossman et al., 1979a). These failures are often seen as partial spikes or spikelets that are electrotonic residues of full action potentials. The functional consequences of conduction failures might be important in vivo. For example in the leech, propagation failures produce an effect similar to that of sensory adaptation. They represent a non-synaptic mechanism that temporarily disconnects the neuron

4

New Insights in Information Processing in the Axon

65

from one defined set of postsynaptic neurons and specifically routes sensory information in the ganglion (Van Essen, 1973; Yau, 1976; Macagno et al., 1987; Gu, 1991). What are the mechanisms of frequency-dependent conduction failure? As mentioned above, the presence of a low safety conduction point such as a branch point, a bottleneck (i.e., an axon entering the soma), or an axonal swelling determines the success or failure of conduction. However, these geometrical constraints are not sufficient to fully account for all conduction failures and additional factors should be considered. The mechanisms of propagation failure can be grouped in two main categories. First, propagation may fail during repetitive axon stimulation as a result of a slight depolarization of the membrane. At spiny lobster axons, propagation failures were associated with a 10–15% reduction of the action potential amplitude in the main axon and a membrane depolarization of 1–3 mV (Grossman et al., 1979a). These observations are consistent with potassium efflux into the peri-axonal space induced by repetitive activation. In most cases, the membrane depolarization produced by external accumulation of potassium ions around the axon probably contributes to sodium channel inactivation. In fact, hyperpolarization of the axon membrane or local application of physiological saline with a low concentration of potassium in the vicinity of a block can restore propagation in crayfish axons (Smith, 1980a). Elevation of the extracellular potassium concentration produced conduction block in spiny lobster axons (Grossman et al., 1979b). However, this manipulation did not reproduce the differential block induced by repetitive stimulation, as failures occurred simultaneously in both branches (Grossman et al., 1979b). Interestingly, conduction could also be restored by the elevation of intracellular calcium concentration. Failures were also induced with a lower threshold when the electrogenic Na+ /K+ pump was blocked with ouabain. Thus, differential conduction block could be explained as follows. During high-frequency activation, potassium initially accumulates at the same rate around the parent axon and the daughter branches. Sodium and calcium accumulate more rapidly in the thin branch than in the thick branch because of the higher surface/volume ratio. Thus the Na+ /K+ pump is activated and extracellular potassium is lowered, more effectively around the thin branch (Grossman et al., 1979b). Accumulation of extracellular potassium has also been observed in the olfactory nerve (Eng and Kocsis, 1987) and in hippocampal axons (Poolos et al., 1987) and could similarly be at the origin of unreliable conduction. Propagation failures have also been reported in the axon of Purkinje neurons under high regimes of stimulation (Monsivais et al., 2005; Khaliq and Raman, 2005; Fig. 4.4a). In this case, the cell body was recorded in whole-cell configuration whereas the signal in the axon was detected in cell-attached mode at a distance up to 800 μm from the cell body. Propagation was found to be highly reliable for single spikes at frequencies below 200 Hz (failures were observed above 250 Hz). Purkinje cells fire single spikes, and complex spikes (bursts) can be elicited following stimulation of the climbing fiber. Interestingly, complex spikes did not propagate reliably in Purkinje cell axons. Generally, only the first and the last spike of the burst propagate. The failure rate of the complex spike is very sensitive to membrane potential

66

D. Debanne and S. Boudkkazi a

20 mV

50 pA

10 ms

b

Recovered IA

Inactivated IA

40 mV

pre

1 nA 2 mV

post

Fig. 4.4 Propagation failures in mammalian axons. a. Propagation failures in Purkinje cell axon. Top, fluorescent image of a Purkinje cell filled with the fluorescent dye Alexa 488. The locations of the somatic and axonal recordings are indicated schematically. Adapted with permission from Monsivais et al. (2005). b. Gating of action potential propagation by the potassium current IA . Left, at resting membrane potential, presynaptic IA was inactivated and the action potential evoked in the presynaptic cell propagated and elicited an EPSP in the postsynaptic cell. Right, following a brief hyperpolarizing pre-pulse presynaptic IA recovered from inactivation and blocked propagation. Consequently, no EPSP was evoked by the presynaptic action potential. Adapted from Debanne et al. (1997)

4

New Insights in Information Processing in the Axon

67

and systematic failures occur when the cell body is depolarized (Monsivais et al., 2005). The limit of conduction has not been yet fully explored in glutamatergic cell axons, but conduction failures have been reported when CA3 pyramidal neuron fire at 30–40 Hz during long plateau potential (Meeks et al., 2005). Thus, the conduction capacity seems much more robust in inhibitory cells compared to glutamatergic neurons. Propagation failures induced by repetitive stimulation may also result from hyperpolarization of the axon. Hyperpolarization-induced conduction block has been observed in leech (Van Essen, 1973; Yau, 1976; Macagno et al., 1987), locust (Heitler and Goodman, 1978), and mammalian axons (Bielefield and Jackson, 1993; Ducreux et al., 1993). In this case, axonal hyperpolarization opposes spike generation. Activity-dependent hyperpolarization of the axon usually results from the activation of the Na+ /K+ ATPase and/or the activation of calcium-dependent potassium channels. Unmyelinated axons in the PNS, for example vagal C-fibers, hyperpolarize in response to repeated action potentials (Ritchie and Straub, 1956; Ritchie and Straub, 1957) as a result of the intracellular accumulation of Na+ and the subsequent activation of the electrogenic Na+ /K+ pump (Ritchie and Straub, 1956; Ritchie and Straub, 1957; Beaumont et al., 2002). In crayfish axons, this hyperpolarization may amount to 5–10 mV (Beaumont et al., 2002). The blockade of the Na+ /K+ ATPase with ouabain results in axon depolarization, probably as a consequence of post-tetanic changes in extracellular potassium concentration. In the leech, hyperpolarization-dependent conduction block occurs at central branch points in all three types of mechanosensory neurons in the ganglion – touch (T), pressure (P), and nociceptive (N) neurons. In these neurons, hyperpolarization is induced by the Na+ /K+ ATPase and by cumulative activation of a calcium-activated potassium conductance. It is interesting to note that the conduction state can be changed by neuromodulatory processes. 5-HT decreases the probability for conduction block of P and T cells, probably by a reduction of the hyperpolarization (Mar and Drapeau, 1996). Hyperpolarization-dependent failures have also been reported in axons of hypothalamic neurons (from paraventricular and supraoptic nuclei) that run into the neurohypophysis. The morphology of their boutons is unusual in that their diameter varies between 5 and 15 μm (Zhang and Jackson, 1993). In single axons, propagation failures are observed at stimulation rates greater than 12 Hz and are concomitant with a hyperpolarization of 4 mV (Bielefeldt and Jackson, 1993). Here, the induced hyperpolarization of the neuron results from activation of the calcium-dependent BK potassium channel. Several recent studies indicate that the hyperpolarization produced by repetitive stimulation could be damped by hyperpolarization-induced cationic current (Ih ) (Beaumont et al., 2002; Soleng et al., 2003). This inward current is activated at resting membrane potential and produces a tonic depolarization of the axonal membrane (Beaumont et al., 2002). Thus, reduction of this current induces a hyperpolarization and perturbs propagation. The pharmacological blockade of Ih by ZD7288 or by external cesium can in fact produce more failures in Schaffer collateral axons (Soleng et al., 2003). The peculiar biophysical properties of Ih indicate

68

D. Debanne and S. Boudkkazi

that it may limit large hyperpolarizations or depolarizations produced by external and internal accumulation of ions. In fact, hyperpolarization of the axon will activate Ih which in turn produces an inward current that compensates the hyperpolarization (Beaumont et al., 2002). Reciprocally, this compensatory mechanism is also valid for depolarization by removing basal activation of Ih . In addition, activity-induced hyperpolarization of the axonal membrane may modulate the biophysical state of other channels that control the propagation.

4.4.3 Frequency-Independent Propagation Failures Action potential propagation in some axon collaterals of CA3 pyramidal neurons can be gated by activation of presynaptic A-type K+ current, independently of the frequency of stimulation (Debanne et al., 1997). Synaptic transmission between monosynaptically coupled pairs of CA3–CA3 or CA3–CA1 pyramidal cells can be blocked if a brief hyperpolarizing current pulse is applied a few milliseconds before the induction of the action potential in the presynaptic neuron (Fig. 4.4b). This regulation is observed in synaptic connections that have no transmission failures, therefore indicating that the lack of postsynaptic response is the consequence of a conduction failure along the presynaptic axon. In contrast with axonal integration where transmitter can be gradually regulated by the presynaptic membrane potential, transmission is all or none. Interestingly, failures can also be induced when the presynaptic hyperpolarizing current pulse is replaced by a somatic IPSP (Debanne et al., 1997; Kopysova and Debanne, 1998). When presynaptic cells are recorded with a microelectrode containing 4-aminopyridine (4-AP), a blocker of IA like conductances, failures are abolished, indicating that IA gates action potential propagation (see also Obaid and Salzberg, 1996). Because A-channels are partly inactivated at the resting membrane potential, their contribution during an action potential elicited from the resting membrane potential is minimal, and the action potential propagates successfully from the cell body to the nerve terminal. In contrast, A-channels recover from inactivation with a transient hyperpolarization and impede successful propagation to the terminal. Propagation failures have been induced in only 30% of cases (Debanne et al., 1997), showing that propagation is generally reliable in hippocampal axons (Mackenzie et al., 1996; Mackenzie and Murphy, 1998; Raastad and Shepherd, 2003). In particular, IA -dependent conduction failures have been found to occur at some axon collaterals but not at others (Debanne et al., 1997). Using a theoretical approach it has been shown that failures occur at branch points when A-type K+ channels are distributed in clusters near the bifurcation (Kopysova and Debanne, 1998). Perhaps because these conditions are not fulfilled in layer II/III neocortical neurons (Cox et al., 2000; Koester and Sakmann, 2000) and in dissociated hippocampal neurons (Mackenzie and Murphy, 1998), this form of gating has not been reported in these cell types. It would be interesting to explore the actual distribution of K channel clusters near branch points using immunofluorescence methods.

4

New Insights in Information Processing in the Axon

69

Functionally, this form of gating may determine part of the short-term synaptic facilitation that is observed during repetitive presynaptic stimulation. Apparent paired-pulse facilitation could be observed because the first action potential fails to propagate but not the second spike, as a result of inactivation of A-type K+ current (Debanne et al., 1999). A recent study suggests that repetitive burst-induced inactivation of A-type K+ channels in the axons of cortical cells projecting onto accumbens nucleus leads to short-term synaptic potentiation through an increased reliability of spike propagation (Casassus et al., 2005).

4.5 Ping-Pong Propagation: Reflection of Action Potential Propagation Branch points are usually considered as frequency filters, allowing separate branches of an axon to activate their synapses at different frequencies. But another way that a neuron’s branching pattern can affect impulse propagation is by reflecting the impulse (Goldstein and Rall, 1974; Ramon et al., 1975; Parnas, 1979). Reflection (or reverse propagation) occurs when an action potential is near failure (Goldstein and Rall, 1974). This form of axonal computation has been well described in leech mechanosensory neurons (Fig. 4.5a; Baccus, 1998; Baccus et al., 2000) in which an unexpected event occurs when conduction nearly becomes blocked: the action potential that has nearly failed to invade the thick branch of the principal axon sets up a local potential that propagates backward. Reflection occurs because impulses are sufficiently delayed as they travel through the branch point. Thus, when the delay exceeds the refractory period of the afferent axon, the impulse will propagate backward as well as forward, creating a reflection. This phenomenon can be identified electrophysiologically at the cell body of the P neuron because action potentials that reflect had a longer initial rising phase (or “foot”), indicating a delay in traveling through the branch point. This fast double firing in the thin branch of mechanosensory neurons has important functional consequences. It facilitates synaptic transmission at synapses formed by this axon and postsynaptic neurons by a mechanism of paired-pulse facilitation with the orthodromic spike and the antidromic action potential that reflected at the branch point (Fig. 4.4a). Reflection is not limited to P cells but also concerns T cells (Baccus, 1998). Interestingly, the facilitation of synaptic transmission also affects the chemical synapse between the P cell and the S neuron, a neuron that plays an essential role in sensitization, a non-associative form of learning (Baccus et al., 2000). Reflected propagation is not restricted to mechanosensory neurons of the leech but has also been noted in the axon of an identified snail neuron (Antic et al., 2000). Reflection has not yet been definitively reported in mammalian axons (see Isope et al., 2004), but it has been demonstrated in dendrites. In mitral cells of the mammalian olfactory bulb, both conduction failures (Chen et al., 1997) and reflection (Chen et al., 2002) have been observed for impulses that are initiated in dendrites (Fig. 4.5b). Propagation in dendrites of mitral cells is rather unusual compared to

70

D. Debanne and S. Boudkkazi

a Reflection

Full conduction Pre Post 1

Conduction block

Pre Post 1

Post 2

Pre Post 1

Post 2

Post 2

2 1 Minor field Major field

Pre Post 1 Post 2

b d 1

2

*

s 1

2

Fig. 4.5 Reflection of action potentials. a. Reflection and conduction block produce multilevel synaptic transmission in mechanosensory neurons of the leech. Left column, an action potential initiated by anterior minor field stimulation invades the whole axonal arborization (red) and evokes an EPSP in all postsynaptic cells. Middle column, following repetitive stimulation, the cell body is slightly hyperpolarized (orange) and the same stimulation induces a reflected action potential at the branch point between the left branch and the principal axon. The reflected action potential (pink arrow 2) stimulates the presynaptic terminal on postsynaptic cell 1 twice, thus enhancing synaptic transmission (arrow). Right column, when the cell body is further hyperpolarized (blue), the stimulation of the minor field now produces an action potential that fails to propagate at the branch point. The failed spike is seen as a spikelet at the cell body (upward arrow). No postsynaptic response is evoked in postsynaptic cell 2 (downward arrow). Adapted from Baccus (1998) and Baccus et al. (2000). b. Reflection of action potential propagation in the presynaptic dendrite of the mitral cell. An action potential (1) initiated in the dendrite (d) fails to propagate toward the soma (s), is then regenerated at the soma (2), and propagates back to the dendrite, thus producing a double dendritic spike (thick trace in the inset). Adapted from Chen et al. (2002)

4

New Insights in Information Processing in the Axon

71

classical dendrites. Like axons, it is highly active and no decrement in the amplitude of the AP is observed between the soma and the dendrite (Bischofberger and Jonas, 1997). In addition, mitral cell dendrites are both pre- and postsynaptic elements. “Ping-pong” propagation has been observed following near failure of dendritic action potentials evoked in distal primary dendrites (Chen et al., 2002). Forward dendritic propagation of an action potential can be evoked by an EPSP elicited by a strong stimulation of the glomerulus. This particular form of propagation may fail near the cell body when the soma is slightly hyperpolarized. For an intermediate range of membrane potential, the action potential invades the soma and may trigger a back-propagating AP which is seen as a dendritic double spike in the primary dendrite. The function of reflected propagation is not yet definitively established, but when axonal output is shut down by somatic inhibition the primary dendrite of the mitral cell may function as a local interneuron affecting its immediate environment. Reflection of fast action potentials has also been observed in dendrites of retinal ganglion cells (Velte and Masland, 1999).

4.6 Spike Timing in the Axon 4.6.1 Axonal Delay Imposed by Axonal Length Axonal conduction introduces a delay in the propagation of neuronal output and axonal arborization might transform a temporal pattern of activity in the main axon into spatial patterns in the terminals (Chung et al., 1970). Axonal delay initially depends on the velocity of the action potential in axons (generally between 0.1 m/s in unmyelinated axons and 100 m/s in large myelinated axons) which directly results from the diameter of the axon and the presence of a myelin sheath. Axonal delays may have crucial functional consequences in the integration of sensory information. In the first relay of the auditory system of the barn owl, differences in the axonal conduction delay from each ear, which in this case depends on the differences in axonal length, produce sharp temporal tuning of the binaural information that is essential for acute sound localization (Fig. 4.6a; Carr and Konishi, 1988, 1990; McAlpine and Grothe, 2003). What is the functional role of axonal delay in network behavior? Theoretical work shows that synchronization of cortical columns and network resonance both depend on axonal delay (Bush and Sejnowski, 1996; Maex and De Schutter, 2003). A recent theoretical study emphasizes the importance of axonal delay in the emergence of polysynchronization in neural networks (Izhikevich, 2006). In most computational studies of storage capacity, axonal delay is totally ignored but in fact, the interplay between axonal delays and synaptic plasticity based on timing (spike-timing-dependent plasticity, STDP) generates the emergence of polychromous groups (i.e., strongly interconnected groups of neurons that fire with millisecond precision). Most importantly, the number of groups of neurons that fire

72

D. Debanne and S. Boudkkazi

a

c ipsi

IA

contra

Pre Post

Block

b IA

3.2 ms

1

2 3 IA

Δt

2.0 ms

2 1

20 mV 1 ms

Latency (ms)

3 4 3 2 1

IA 0

IA

IA

10

20

30

40

Delay (ms)

d

Ephaptic Interaction

Delay

Resynchronization

Fig. 4.6 Axonal propagation and spike timing. a. Delay lines in the auditory system of the barn owl. Each neuron from the nucleus laminaris receives an input from each ear. Note the difference in axonal length from each side. Adapted from Carr and Konishi (1990). b. Comparison of the delay of propagation introduced by a branch point with GR > 1 (dashed traces) versus a branch point with perfect impedance matching (GR = 1, continuous traces). Upper, schematic drawing of a branched axon with three points of recording. At the branch point with GR = 8, the shape of the action potential is distorted and the propagation displays a short latency (t). Adapted from Manor et al. (1991). c. Propagation failures in hippocampal cell axons are associated with conduction delays. The presynaptic neuron was slightly hyperpolarized with constant current to remove inactivation of the A-current. A presynaptic action potential induced with a short delay after onset

4

New Insights in Information Processing in the Axon

73

synchronously exceeds the number of neurons in a network, resulting in system with massive memory capacity (Izhikevich, 2006).

4.6.2 Extra Delays Imposed by Axonal Irregularities and Ion Channels In addition to this axonal delay, local changes in the geometry of the axon produce an extra delay. The presence of axonal irregularities such as varicosities and branch points reduces conduction velocity (Fig. 4.6b). This reduction in conduction velocity occurs as a result of a high geometrical ratio (GR, see above). The degree of temporal dispersion has been simulated in the case of an axon from the somatosensory cortex of the cat (Manor et al., 1991). The delay introduced by high GR branch points could account for a delay of 0.5–1 ms (Manor et al., 1991). But this extra delay appears rather small compared to the delay imposed by the conduction in axon branches with variable lengths (in the range of 2–4 ms). A third category of delay in conduction can be introduced during repetitive stimulation or during the activation of specific ion channels. Thus, the magnitude of this delay is usually variable. It has been measured in a few cases. In lobster axons, the conduction velocity of the axon was lowered by ∼30% following repetitive stimulation (Grossman et al., 1979a). In dorsal root ganglion neurons, the latency of conducted spikes was found to be enhanced by about 1 ms following antidromic paired-pulse stimulation of the axon (Lüscher et al., 1994a). Computational studies indicate that this delay may also result from a local distortion of the action potential shape. Extra activity-dependent delays may have significant consequences on synaptic transmission. For instance, the synaptic delay was found to increase by 1–2 ms during repetitive stimulation of crayfish motor neuron (Hatt and Smith, 1976). Monosynaptic connections to motoneurons show an increase in synaptic latency concomitant with the synaptic depression induced by repetitive stimulation at 5–10 Hz which induced near-propagation failures (Streit et al., 1992). Similarly, a longer synaptic delay has been measured between connected hippocampal cells when conduction nearly fails due to reactivation of A-type potassium channels (Fig. 4.6c; Debanne et al., 1997). Thus, axonal conduction may introduce some noise into the temporal pattern of action potentials produced at the initial segment. At the scale of a nerve, delays in individual axons introduce a temporal dispersion of conduction suggesting a stuttering model of propagation (Muschol et al., 2003). 

Fig. 4.6 (continued) of the depolarizing pulse did not elicit an EPSC in the postsynaptic cell because of the large activation of IA . Increasing the delay permitted action potential propagation because IA was reduced during the action potential. For complete inactivation of IA (lower pair of traces), latency decreased. Adapted from Debanne et al. (1997). d. Ephaptic interaction in axons. Schematic representation of resynchronization of action potentials in a pair of adjacent axons. Adapted from Katz and Schmitt (1940)

74

D. Debanne and S. Boudkkazi

4.6.3 Ephaptic Interactions and Axonal Spike Synchronization Interactions between neighboring axons were first studied by Katz and Schmitt (1940, 1942) in crab. The passage of an impulse in one axonal fiber produced a subthreshold change in excitability in the adjacent fiber. As the action potential approaches in the active axon, the excitability of the resting fiber was first reduced, then quickly enhanced. This effect results from the depolarization of the resting axon by the active axon because it generates locally an extracellular potential of a few mV. Interactions of this type are called ephaptic (from the Greek for “touching onto,” Arvanitaki, 1942). They have also been observed in frog sciatic nerve (Kocsis et al., 1982). One of the most interesting features of ephaptic interaction between adjacent axons is that the conduction velocity in neighboring fibers might be unified, thus synchronizing activity in a bundle of axons. If one action potential precedes the other by a few milliseconds, it accelerates the conduction rate of the lagging action potential in the other axon (Katz and Schmitz, 1940, Fig. 4.6d). However, perfectly synchronized action potentials decrease the conduction velocity in both branches. Synchronization can only occur if the individual velocities differ only slightly and are significant for a sufficient axonal length (Katz and Schmitt, 1940). Does such synchronization also occur in mammalian axons? There is no evidence for this yet, but modeling studies indicate that the relative location of nodes of Ranvier on two adjacent myelinated axons might also determine the degree of temporal synchrony between fibers (Binczak et al., 2001; Reutskiy et al., 2003). On small unmyelinated axons, ephaptic interaction between axons is predicted to be very small (Holt and Koch, 1999) but future research in this direction might reveal a powerful means to thoroughly synchronize neuronal activity downstream of the site of action potential initiation.

4.6.4 Electric Coupling in Axons and Fast Synchronization Fast communication between neurons is assured by chemical synapses, and electrical coupling has been reported in a large number of cell types including inhibitory cortical interneurons (Hestrin and Gallareta, 2005). In the hippocampus, one type of high-frequency oscillation (100–200 Hz) called “ripple” arises from the highfrequency firing of inhibitory interneurons and phase-locked firing of many CA1 neurons (Traub et al., 2002). Some of the properties of ripple oscillation are, however, difficult to explain. First, the oscillations are so fast (near 200 Hz) that synchrony across many cells would be difficult to achieve through chemical synaptic transmission. In addition, ripples persist during pharmacological blockade of chemical transmission in vitro (Draguhn et al., 1998). While some inhibitory interneurons may synchronize a large number of pyramidal cells during the ripple (Klausberger et al., 2003), a significant part of the synchronous activity could be mediated by axo-axonal electrical synaptic contacts through gap junctions (Schmitz et al., 2001).

4

New Insights in Information Processing in the Axon

75

Antidromic stimulation of a neighboring axon elicits a small action potential, a spikelet with a fast rate of rise (near 180 mV/ms). Spikelets can be evoked at the rate of a ripple (200 Hz) and they are blocked by tetrodotoxin or by the gap junction blocker carbenoxolone. Simultaneous recording from the axon and cell body showed that the spikelet first traversed the axon prior to invading the soma and the dendrites. Finally, the labeling of pyramidal neurons with rhodamine, a small fluorescent molecule, showed dye coupling in adjacent neurons that was initiated through the axon (Schmitz et al., 2001). Thus, the function of the axon is not limited to the conduction of the impulses to the terminal, and information may process between adjacent pyramidal neurons through electrical synapses located close to their axon hillock.

4.7 Conclusion 4.7.1 Increased Computational Capabilities Axons achieve several fundamental operations that go far beyond classical propagation. Like active dendrites, axons amplify and integrate subthreshold and suprathreshold electrical signals (Engel and Jonas, 2005; Alle and Geiger, 2006; Shu et al., 2006). In addition, the output message can be routed in selective axonal pathways at a defined regime of activity. The consequences of this are not yet well understood in mammalian axons, but branch point failures may participate in the elaboration of sensory processing in invertebrate neurons (Gu, 1991). Axonal propagation may also bounce back at a branch point or at the cell body. But at present, there are only a handful of examples showing reflected propagation (Baccus, 1998; Baccus et al., 2000; Antic et al., 2000; Chen et al., 2002; Isope et al., 2004). Reflected impulses may limit the spread of the neuronal message and enhance synaptic transmission. Theoretical and experimental studies indicate that reflection of action potentials could occur in axons that display large swellings or a branch point with high GR. Moreover, axonal delay is important to set network resonance (Maex and De Schutter, 2003) and increase storage capacity in neuronal networks (Izhikevich, 2006). Finally, axonal coupling through ephaptic interactions or gap junctions may precisely synchronize network activity (Schmitz et al., 2001; Roopun et al., 2006). All these operations increase the computational capabilities of axons and affect the dynamics of synaptic coupling. Many pieces of the puzzle are, however, still missing.

4.7.2 Future Directions and Missing Pieces In the recent past, most (if not all) of our knowledge about axonal computation capabilities was derived from experiments on invertebrate neurons or from computer simulations (Segev and Schneidman, 1999). The recent democratization of direct

76

D. Debanne and S. Boudkkazi

patch-clamp recordings from the presynaptic terminal (Engel and Jonas, 2005; Alle and Geiger, 2006) or from the axon (Monsivais et al., 2005; Khaliq and Raman, 2005; Meeks et al., 2005; Shu et al., 2006; Kole et al., 2007) suggests that thin mammalian axon will yield up all its secrets in a near future. There are good reasons to believe that, combined with the development of high-resolution imaging techniques like multiphoton confocal microscopy (Cox et al., 2000; Koester and Sakmann, 2000; Forti et al., 2000) and voltage sensitive dyes (Antic et al., 2000; Muschol et al., 2003), this technique will be a powerful tool to know more about the function of axons. Axonal morphology and the subcellular localization of ion channels play a crucial role in conduction properties and propagation failures or reflected propagation may result from the presence of axonal irregularities such as varicosities and branch points. However, detailed quantitative analysis of the morphometry of single axons combined with the quantitative immunostaining of sodium channels will be needed. The use of recently developed molecular tools to target defined channel subunits toward specific axonal compartments could be of great help to determine their role in axonal propagation. Fine temporal tuning can be achieved by axons. Differences in axonal length in the terminal axonal tuft introduce delays of several milliseconds. Is temporal scaling of action potential propagation in the axonal arborization relevant to the coding of neuronal information? Differential conduction delays in axonal branches participate in precise temporal coding in the barn owl auditory system (Carr and Konishi, 1988; Carr and Konishi, 1990; McAlpine and Grothe, 2003). But the role of axonal delays has only been studied in formal neural networks (Bush and Sejnowski, 1996; Maex and De Schutter, 2003; Izhikevich, 2006), and additional work will have to be done to circumscribe its implication in hybrid or in in vivo networks. Nicotine enhances excitability of auditory thalamocortical axons and reduces both the latency and latency variability (Kawai et al., 2007). Although this process could lower the threshold for auditory perception by acting on the thalamocortical flow of information, further studies will be required to understand the underlying mechanisms at the axonal level. Local axonal interactions like ephaptic coupling and gap junction coupling allow very fast synchronization of activity in neighboring neurons. Surprisingly, little experimental effort has been devoted to ephaptic interactions. This mechanism represents a powerful means to precisely synchronize output messages of neighbouring neurons. Perhaps ephaptic interactions between parallel axons could compensate the “stuttering conduction” that is introduced by axonal varicosities and branch points (Muschol et al., 2003). The implications of these mechanisms in synchronized activity will have to be determined in axons that display favorable geometrical arrangement for ephaptic coupling (i.e., fasciculation over a sufficient axonal length). Callosal axons, mossy fibers, and Schaffer collaterals are possible candidates. Once the basic function of the axon is known, it will be important to determine whether activity-dependent, long-term regulation of ion channel activity described in the dendritic and somatic compartments (for review, see Daoudal and Debanne,

4

New Insights in Information Processing in the Axon

77

2003) is also present in the axonal compartment. There are good reasons to believe that functional plasticity exists in the axon. For instance, the repetitive stimulation of Schaffer collateral axons at 2 Hz leads to a long-lasting lowering of the antidromic activation threshold (McNaugthon et al., 1994), suggesting that axonal excitability is persistently enhanced. Furthermore, LTP and LTD are, respectively, associated with increased and decreased changes in intrinsic excitability of the presynaptic neuron (Ganguly et al., 2000; Li et al., 2004). These changes imply retrograde messengers that target the presynaptic neuron. Although these changes are detected in the cell body, one may not exclude that ion channels located in the axon are also regulated. It is therefore tempting to speculate that some of the axonal operations described in this chapter could also be modulated by network activity. Acknowledgment Supported by Institut National de la Santé et de la Recherche Médicale (Programme “Avenir”), Centre National de la Recherche Scientifique, Ministry of Research (“Actions Incitatives Jeunes Chercheurs” 5169) and Fondation pour la Recherche Médicale. We thank M. Seagar for his constant support and constructive criticisms on the manuscript. SB was supported by Ministry of Research (Doctoral grant) and European Community (LSHM-CT2004-511995, Synaptic Scaffolding Proteins Orchestrating Cortical Synapse Organisation during Development).

References Alle H, Geiger JRP (2006) Combined analog and action potential coding in hippocampal mossy fibers. Science 311:1290–1293 Antic S, Wuskell JP, Loew L, Zecevic D (2000) Functional profile of the giant metacerebral neuron of Helix aspersa: temporal and spatical dynamics of electrical activity in situ. J Physiol (Lond) 527:55–69 Antonini A, Gillespie DC, Crair MC, Stryker MP (1998) Morphology of single geniculocortical afferents and functional recovery of the visual cortex after reverse monocular deprivation in the kitten. J Neurosci 18:9896–9909 Arvanitaki A (1942) Effects evoked in an axon by the activity of a contiguous one. J Neurophysiology 5:89–108 Awatramani G, Price GD, Trussell LO (2005) Modulation of transmitter release by presynaptic resting potential and background calcium levels. Neuron 48:109–121 Baccus SA (1998) Synaptic facilitation by reflected action potentials: enhancement of transmission when nerve impulses reverse direction at axon branch points. Proc Natl Acad Sci USA 95:8345–8350 Baccus SA, Burrell BD, Sahley CL, Muller KJ (2000) Action potential reflection and failure at axon branch points cause stepwise changes in EPSPs in a neuron essential for learning. J Neurophysiol 83:1683–1700 Barron DH, Matthews BHC (1939) Intermittent conduction in the spinal cord. J Physiol (Lond) 85:73–103 Beaumont V, Zhong N, Froemke RC, Ball RW, Zucker RS (2002) Temporal synaptic tagging by Ih activation and actin: involvement in long-term facilitation and cAMP-induced synaptic enhancement. Neuron 33:601–613 Berbel P, Innocenti GM (1988) The development of the corpus callosum: a light and electromicroscopic study. J Comp Neurol 276:132–156 Bielefeldt K, Jackson MB (1993) A calcium-activated potassium channel causes frequencydependent action-potential failures in a mammalian nerve terminal. J Neurophysiol 70:284–298

78

D. Debanne and S. Boudkkazi

Binczak S, Eilbeck JC, Scott AC (2001) Ephaptic coupling between myelinated nerve fibers. Physicia D 148:159–174 Bischofberger J, Jonas P (1997) Action potential propagation into the presynaptic dendrites of rat mitral cells. J Physiol (Lond) 504:350–365 Blackstad TW, Kjaerheim Å (1961) Special axo-dendritic synapses in the hippocampal cortex: electron and light microscopic studies in the hippocampal cortex: electron and light microscopic studies on the layer of mossy fibers. J Comp Neurol 117:133–146 Bourque CW (1990) Intraterminal recordings from the rat neurohypophysis in vitro. J Physiol (Lond) 421:247–262 Brigant JL, Mallart A (1982) Presynaptic currents in mouse motor endings. J Physiol 333:619–636 Brody DL, Yue DT (2000) Release-independent short-term synaptic depression in cultured hippocampal neurons. J Neurosci 20:2480–2494 Bush P, Sejnowski T (1996) Inhibition synchronizes connected cortical neurons within and between columns in realistic networks models. J Comput Neurosci 3:91–110 Carr CE, Konishi M (1988) Axonal delay lines for time measurement in the owl’s brainstem. Proc Natl Acad Sci USA 85:8311–8315 Carr CE, Konishi M (1990) A circuit for detection of interaural time differences in the brain stem of the barn owl. J Neurosci 10:3227–3246 Casassus G, Blanchet C, Mulle C (2005) Short-term regulation of information processing at the corticoaccumbens synapse. J Neurosci 25:11504–11512 Chen WR, Midtgaard J, Shepherd GM (1997) Forward and backward propagation of dendritic impulses and their synaptic control in mitral cells. Science 278:463–467 Chen WR, Shen GY, Shepherd GM, Hines ML, Midtgaard J (2002) Multiple modes of action potential initiation and propagation in mitral cell primary dendrite. J Neurophysiol 88: 2755–2764 Chung SH, Raymond SA, Lettvin JY (1970) Multiple meaning in single visual units. Brain Behav Evol 3:72–101 Clark B, Häusser M (2006) Neural coding: hybrid analog and digital signalling in axons. Current Biol. 15:R585–R588 Cobb SR, Buhl EH, Halasy K, Paulsen O, Somogyi P (1995) Synchronization of neuronal activity in hippocampus by individual GABAergic interneurons. Nature 378:75–78 Cooper EC, Milroy A, Jan YN, Jan LY, Lowenstein DH (1998) Presynaptic localization of Kv1.4containing A-type potassium channels near excitatory synapses in the hippocampus. J Neurosci 18:965–974 Cox CL, Denk W, Tank DW, Svoboda K (2000) Action potentials reliably invade axonal arbors of rat neocortical neurons. Proc Natl Acad Sci USA 97:9724–9728 Daoudal G, Debanne D (2003) Long-term plasticity of intrinsic excitability: learning rules and mechanisms. Learn Mem 10:456–465 Debanne D (2004) Information processing in the axon. Nat Rev Neurosci 5:304–316 Debanne D, Guérineau NC, Gähwiler BH, Thompson SM (1997) Action potential propagation gated by an IA -like K+ conductance in hippocampus. Nature 389:286–289 Debanne D, Kopysova IL, Bras H, Ferrand N (1999) Gating of action potential propagation by an axonal A-like potassium conductance in the hippocampus: a new type of non-synaptic plasticity. J Physiol (Paris) 93:285–296 Deschênes M, Landry P (1980) Axonal branch diameter and spacing of nodes in the terminal arborization of identified thalamic and cortical neurons. Brain Res 191:538–544 Draguhn A, Traub RD, Schmitz D, Jefferys JGR (1998) Electrical coupling underlies highfrequency oscillations in the hippocampus in vitro. Nature 394:189–192 Dreyer F, Penner R (1987) The actions of presynaptic snake toxins on membrane currents of mouse motor nerve terminals. J Physiol 386:455–463 Ducreux C, Reynaud JC, Puizillout JJ (1993) Spike conduction properties of T-shaped C neurons in the rabbit nodose ganglion. Eur J Physiol 424:238–244 Dyball RE, Grossmann R, Leng G, Shibuki K (1988) Spike propagation and conduction failure in rat neural lobe. J Physiol (Lond) 401:241–256

4

New Insights in Information Processing in the Axon

79

Eng DL, Kocsis JD (1987) Activity-dependent changes in extracellular potassium and excitability in turtle olfactory nerve. J Neurophysiol 57:185–202 Engel D, Jonas P (2005) Presynaptic action potential amplification by voltage-gated Na+ channels in hippocampal mossy fiber boutons. Neuron 45: 405–417 Evans CG, Jing J, Rosen SC, Cropper EC (2003) Regulation of spike initiation and propagation in an Aplysia sensory neuron: gating-in via central depolarization. J Neurosci 23:2920–2931 Forti L, Pouzat C, Llano I (2000) Action potential-evoked Ca2+ signals and calcium channels in axons of developing rat cerebellar interneurones. J Physiol (Lond) 527:33–48 Ganguly K, Kiss L, Poo MM (2000) Enhancement of presynaptic neuronal excitability by correlated presynaptic and post-synaptic spiking. Nat Neurosci 3:1018–1026 Geiger JRP, Jonas P (2000) Dynamic control of presynaptic Ca2+ inflow by fast-inactivating K+ channels in hippocampal mossy fiber boutons. Neuron 28:927–939 Goldstein S, Rall W (1974) Changes of action potential shape and velocity for changing core conductor geometry. Biophys J 14:731–757 Goldfinger MD (2000) Computation of high safety factor impulse propagation at axonal branch points. NeuroReport 11:449–456 Graham B, Redman S (1994) A simulation of action potentials in synaptic boutons during presynaptic inhibition. J Neurophysiol 71:538–549 Grossman Y, Parnas I, Spira ME (1979a) Differential conduction block in branches of a bifurcating axon. J Physiol (Lond) 295:283–305 Grossman Y, Parnas I, Spira ME (1979b) Ionic mechanisms involved in differential conduction of action potentials at high frequency in a branching axon. J Physiol (Lond) 295:307–322 Gu X (1991) Effect of conduction block at axon bifurcations on synaptic transmission to different postsynaptic neurons in the leech. J Physiol (Lond) 441:755–778 Gulyas AI, Miles R, Hajos N, Freund T (1993) Precision and variability in postsynaptic target selection of inhibitory cells in the hippocampus CA3 region. Eur J Neurosci 5:1729–1751 Hatt H, Smith DO (1976) Synaptic depression related to presynaptic axon conduction block. J Physiol (Lond) 259:367–393 He Y, Zorumski CF, Mennerick S (2002) Contribution of presynaptic Na+ channel inactivation to paired-pulse synaptic depression in cultured hippocampal neurons. J Neurophysiol 87:925–936 Heitler WJ, Goodman CS (1978) Multiple sites of spike initiation in a bifurcating locust neurone. J Exp Biol 76:63–84 Hestrin S, Galaretta M (2005) Electrical synapses define networks of neocortical GABAergic neurons. Trends Neurosci 28:304–309 Holt GA, Koch C (1999) Electrical interactions via the extracellular potential near cell bodies. J Comput Neurosci 6:169–184 Inda MC, DeFelipe J, Muñoz A (2006) Voltage-gated ion channels in the axon initial segment of human cortical pyramidal cells and their relationship with chandelier cells. Proc Natl Acad Sci 103:2920–2925 Ishizuka N, Weber J, Amaral DG (1990) Organization of intrahippocampal projections originating from CA3 pyramidal cells in the rat. J Comp Neurol 295:580–623 Isope P, Franconville R, Barbour B, Ascher P (2004) Repetitive firing of rat cerebellar parallel fibres after a single stimulation. J Physiol (Lond) 554:829–839 Izhikevich EM (2006) Polychronization: computation with spikes. Neural Comput 18:245–282 Jackson MB, Zhang SJ (1995) Action potential propagation block by GABA in rat posterior pituitary nerve terminals. J Physiol (Lond) 483:597–611 Katz B, Miledi R (1965) Propagation of electrical activity in motor nerve terminal. Proc R Soc Lond B Biol Sci 161:453–482 Katz B, Schmitt O (1940) Electric interaction between two adjacent nerve fibres. J Physiol (Lond) 97:471–488 Katz B, Schmitt O (1942) A note on interaction between nerve fibers. J Physiol (Lond) 100: 369–371

80

D. Debanne and S. Boudkkazi

Kawai H, Lazar R, Metherate R (2007) Nicotinic control of axon excitability regulates thalamocortical transmission. Nat Neurosci 10:1168–1175 Khaliq ZM, Raman IM (2005) Axonal propagation of simple and complex spikes in cerebellar Purkinje neurons. J Neurosci 25:454–463 Klausberger T, Magill PJ, Marton LF, Roberts JDB, Cobden PM, Buzsaki G, Somogyi P (2003) Brain-state- and cell-type-specific firing of hippocampal interneurons in vivo. Nature 421: 844–848 Kocsis JD, Ruiz JA, Cummins KL (1982) Modulation of axonal excitability mediated by surround electrical activity: an intra-axonal study. Exp Brain Res 47:151–153 Koester HJ, Sakmann B (2000) Calcium dynamic associated with action potentials in single nerve terminals of pyramidal cells in layer 2/3 of the young rat neocortex. J Physiol (Lond) 529: 625–646 Kole MHP, Letzkus JJ, Stuart GJ (2007) Axon initiak segment Kv1 channels control axonal action potential waveform and synaptic efficacy. Neuron 55:633–647 Kopysova IL, Debanne D (1998) Critical role of axonal A-type K+ -channels and axonal geometry in the gating of action potential propagation along CA3 pyramidal cell axons: a simulation study. J Neurosci 18:7436–7451 Krnjevic K, Miledi R (1959) Presynaptic failure of neuromuscular propagation in rats. J Physiol (Lond) 149:1–22 Leao RM, Kushmerick, Pinaud R, Renden R, Li, GL, Taschenberger H, Spirou G, Levinson SR, von Gersdorff H (2005) Presynaptic Na+ channels: locus, development, and recovery from inactivation at a high fidelity synapse. J Neurosci 25:3724–3738 Li X, Somogyi P, Ylinen A, Buzsaki G (1994) The hippocampal CA3 network: an in vivo intracellular labeling study. J Comp Neurol 339:181–208 Li C, Lu J, Wu C, Duan S, Poo M (2004) Bidirectional modification of presynaptic neuronal excitability accompanying spike timing-dependent synaptic plasticity. Neuron 41:257–268 Lindgren CA, Moore JW (1989) Identification of ionic currents at presynaptic nerve endings of the lizard. J Physiol 414:201–202 Lüscher HR, Shiner JS (1990a) Computation of action potential propagation and presynaptic bouton activation in terminal arborizations of different geometries. Biophys J 58:1377–1388 Lüscher HR, Shiner JS (1990b) Simulation of action potential propagation in complex terminal arborizations. Biophys J 58:1389–1399 Lüscher C, Streit J, Quadroni R, Lüscher HR (1994a) Action potential propagation through embryonic dorsal root ganglion cells in culture. I. Influence of the cell morphology on propagation properties. J Neurophysiol 72:622–633 Lüscher C, Streit J, Lipp P, Lüscher HR (1994b) Action potential propagation through embryonic dorsal root ganglion cells in culture. II. Decrease of conduction reliability during repetitive stimulation. J Neurophysiol 72:634–644 (1994) Macagno ER, Muller KJ, Pitman RM (1987) Conduction block silences parts of a chemical synapse in the leech central nervous system. J Physiol (Lond) 387:649–664 Mackenzie PJ, Umemiya M, Murphy TH (1996) Ca2+ imaging of CNS axons in culture indicate reliable coupling between single action potentials and distal functional release sites. Neuron 16:783–795 Mackenzie PJ, Murphy TH (1998) High safety factor for action potential conduction along axons but not dendrites of cultured hippocampal and cortical neurons. J Neurophysiol 80: 2089–2101 Madeja M (2000) Do neurons have a reserve of sodium channels for the generation of action potentials? A study on acutely isolated CA1 neurons from the guinea-pig hippocampus. Eur J Neurosci 12:1–7 Maex R, De Schutter E (2003) Resonant synchronization in heterogeneous networks of inhibitory neurons. J Neurosci 23:10503–10514 Major G, Larkman AU, Jonas P, Sakmann B, Jack JJB (1994) Detailed passive cable models of whole-cell recorded CA3 pyramidal neurons in rat hippocampal slices. J Neurosci 14: 4613–4638

4

New Insights in Information Processing in the Axon

81

Manor Y, Koch C, Segev I (1991) Effect of geometrical irregularities on propagation delay in axonal trees. Biophys J 60:1424–1437 Mar A, Drapeau P (1996) Modulation of conduction block in leech mechanosensory neurons. J Neurosci 16:4335–4334 Marder E (2006) Extending influence. Nature 441:702–703 Martina M, Jonas P (1997) Functional differences in Na+ channel gating between fast-spiking interneurones and principal neurons of rat hippocampus. J Physiol (Lond) 505:593–603 Martina M, Vida I, Jonas P (2000) Distal initiation and active propagation of action potentials in interneurons dendrites. Science 287:295–300 McAlpine D, Grothe B (2003) Sound localization and delay lines – do mammals fit the model? Trends Neurosci. 26, 347–350 McNaughton BL, Shen J, Rao G, Foster TC, Barnes CA (1994) Persistent increase of hippocampal axon excitability after repetitive stimulation: dependence on N-methyl-D-aspartate receptor activity, nitric-oxyde synthase, and temperature. Proc Natl Acad Sci 91:4830–4834 Mejia-Gervacio, S, Collin T, Pouzat C, Tan YP, Llano I, Marty A (2007) Axonal speeding: shaping synaptic potentials in small neurons by the axonal membrane compartment. Neuron 53: 843–855 Meeks JP, Mennerick S (2004) The selective effects of potassium elevation on glutamate signaling and action potential conduction in hippocampus. J Neurosci 24:197–206 Meeks JP, JIang X, Mennerick S (2005) Action potential fidelity during normal and epileptiform activity in paired soma-axon recordings from rat hippocampus. J Physiol (Lond) 566: 425–441 Monsivais P, Clark BA, Roth A, Häusser M (2005) Determinants of action potential propagation in cerebellar Purkinje cell axons. J Neurosci 25:464–472 Muschol M, Kosterin P, Ichikawa M, Salzberg BM (2003) Activity-dependent depression of excitability and calcium transients in the neurohypophysis suggests a model of “stuttering conduction”. J Neurosci 23:11352–11362 Nakamura M, Sekino Y, Manabe T (2007) GABAergic interneurons facilitate mossy fiber excitability in the developing hippocampus. J Neurosci 27:1365–1373 Nicholls J, Wallace BG (1978) Modulation of transmission at an inhibitory synapse in the central nervous system of the leech. J Physiol 281:157–170 Obaid AL, Salzberg BM (1996) Micromolar 4-aminopyridine enhances invasion of a vertebrate neurosecretary terminal arborization. J Gen Physiol 107:353–368 Parnas I (1972) Differential block at high frequency of branches of a single axon innervating two muscles. J Neurophysiol 35:903–914 Parnas I, Hochstein S, Parnas H (1976) Theoretical analysis of parameters leading to frequency modulation along an inhomogeneous axon. J Neurophysiol 39:909–923 Parnas I (1979) The Neurosciences. Schmitt FO, Worden FG (eds). MIT Press, Cambridge MA, pp 499–512 Pinault D, Deschênes M (1998) Projection and innervation patterns of individual thalamic reticular axons in the thalamus of the adult rat: a three–dimensional, graphic, and morphometric analysis. J Comp Neurol 391:180–203 Poolos NP, Mauk MD, Kocsis JD (1987) Activity-evoked increases in intracellular potassium modulate presynaptic excitability in the CA1 region of the hippocampus. J Neurophysiol 58:404–416 Prakriya M, Mennerick S (2000) Selective depression of low-release probability excitatory synapses by sodium channel blockers. Neuron 26:671–682 Raastad M, Shepherd G (2003) Single-axon action potentials in the rat hippocampal cortex. J Physiol (Lond) 548:745–752 Rall W (1959) Branching dendritic trees and motoneuron membrane resistivity. Expl Neurol 1:491–527 Rall W (1964) Theoretical significance of dendritic tree for neural input-output relations. In: Reiss FP (ed) Neural theory of modeling. Standford University Press, Palo Alto, pp 73–97

82

D. Debanne and S. Boudkkazi

Ramon F, Joyner RW, Moore, JW (1975) Propagation of action potentials in inhomogeneous axon regions. Fed Proc 34:1357–1363 Reutskiy S, Rossoni E, Tirozzi B (2003) Conduction in bundles of demyelinated nerve fibers: computer simulation. Biol Cybern 89:439–448 Ritchie JM, Straub RW (1956) The after-effects of repetitive stimulation on mammalian nonmedulated fibres. J Physiol (Lond) 134:698–711 Ritchie JM, Straub RW (1957) The hyperpolarization which follows activity in mammalian nonmedulated fibres. J Physiol (Lond) 136:80–97 Rhodes KJ, Strassle BW, Monaghan MM, Bekele-Arcuri Z, Matos MF, Trimmer JS (1997) Association and colocalization of the Kvβ1 and Kvβ2 β-subunits with Kv1 α-subunits in mammalian brain K+ channel complexes. J Neurosci 17:8246–8258 Roopun AK, Middleton SJ, Cunningham MO, LeBeau FEN, Bibbig A, Whittington MA, Traub RD (2006) A beta2-frequency (20–30 Hz) oscillation in nonsynaptic networks of somatosensory cortex. Proc Natl Acad Sci USA 103:15646–15650 Saviane C, Mohajerani MH, Cherubini E (2003) An ID like current that is downregulated by Ca2+ modulates information coding at CA3-CA3 synapses in the rat hippocampus. J Physiol (Lond) 552:513–524 Schmitz D, Schuchmann S, Fisahn A, Draguhn A, Buhl EH, Petrasch-Parwez E, Dermietzel R, Heinemann U, Traub RD (2001) Axo-axonal coupling: a novel mechanism for ultrafast neuronal communication. Neuron 31:831–840 Scott LL, Hage TA, Golding NL (2007) Weak action potential backpropagation is associated wth high-frequency axonal firing capability in principal neurons of the gerbil medial superior olive. J Physiol 583:647–661 Segev I, Schneidman E (1999) Axons as computing devices: basic insights gained from models. J Physiol (Paris) 93: 263–270 Shapiro E, Castellucci VF, Kandel ER (1980) Presynaptic membrane potential affects transmitter release in an identified neuron in Aplysia by modulating the Ca2+ and K+ currents. Proc Natl Acad Sci USA 77:629–633 Sheng M, Tsaur ML, Jan YN, Jan, LY (1992) Subcellular segregation of two A-type K+ channel protein in rat central neurons. Neuron 9:243–259 Sheng M, Liao YJ, Jan YN, Jan LY (1993) Presynaptic A-current based on heteromultimeric K+ channels detected in vivo. Nature 365:72–75 Shepherd GMG, Raastad M (2003) Axonal varicosity distributions along parallel fibers: a new angle on a cerebellar circuit. The Cerebellum 2:110–113 Shepherd GMG, Raastad M, Andersen P (2002) General and variable features of varicosity spacing along unmyelinated axons in the hippocampus and cerebellum. Proc Natl Acad Sci USA 99:6340–6345 Shimahara T, Tauc L (1975) Multiple interneuronal afferents to the giant cells in Aplysia. J Physiol 247:299–319 Shu Y, Hasenstaub A, Duque A, Yu Y, McCormick DA (2006) Modulation of intracortical synaptic potentials by presynaptic somatic membrane potential. Nature 441:761–765 Shu Y, Yu Y, Yang J, McCormick DA (2007) Selective control of cortical axonal spikes by a slowly inactivating K+ current. Proc Natl Acad Sci USA 104:11453–11458 Smith DO (1980a) Mechanisms of action potential propagation failure at sites of axon branching in the crayfish. J Physiol (Lond) 301:243–259 Smith DO (1980b) Morphological aspects of the safety factor for action potential propagation at axon branch points. J Physiol (Lond) 301:261–269 Soleng AF, Chiu K, Raastad M (2003) Unmyelinated axons in the rat hippocampus hyperpolarize and activate an H current when spike frequency exceeds 1 Hz. J Physiol (Lond) 552:459–470 Streit J, Lüscher C, Lüscher HR (1992) Depression of postsynaptic potentials by high frequency stimulation in embryonic motoneurons grown in spinal cord slice cultures. J Neurophysiol 68:1793–1803 Stuart G, Sakmann B (1995) Amplification of EPSPs by axosomatic sodium channels in neocortical pyramidal neurons. Neuron 15:1065–1076

4

New Insights in Information Processing in the Axon

83

Tan YP, Llano I (2000) Modulation by K+ channels of action potential-evoked intracellular Ca2+ concentration rises in rat cerebellar basket cell axons. J Physiol (Lond) 520:65–78 Traub RD, Draguhn A, Whttington MA, Baldewed T, Bibbig A, Buhl EH, Schmitz D (2002) Axonal gap junctions between neurons: a novel source of network oscillations and perhaps epileptogenesis. Rev Neurosci 13:1–30 Van Essen DC (1973) The contribution of membrane hyperpolarization to adaptation and conduction block in sensory neurones of the leech. J Physiol (Lond) 230:509–534 Veh RW, Lichtinghagen R, Sewing S, Wunder F, Grumbach IM, Pongs O (1995) Immunohistochemical localization of five members of the Kv1 channel subunits: contrasting subcellular locations and neuron-specific co-localizations in rat brain. Eur J Neurosci 7:2189–2205 Velte TJ, Masland RH (1999) Action potentials in the dendrites of retinal ganglion cells. J Neurophysiol 81:1412–1417 Westrum LE, Blackstad TW (1962) An electron microscopic study of the stratum radiatum of the rat hippocampus (regio superior, CA1) with particular emphasis on synaptology. J Comp Neurol 119:281–309 Yau KW (1976) Receptive fields, geometry and conduction block of sensory neurons in the CNS of the leech. J Physiol (Lond) 263:513–538 Young JZ (1936) The giant nerve fibres and epistellar body of cephalopods. Q J Microsc Sci 78:367–386 Zhang SJ, Jackson MB (1993) GABA-activated chloride channels in secretory nerve endings. Science 259:531–534 Zhou L, Chiu SY (2001) Computer model for action potential propagation through branch point in myelinated nerves. J Neurophysiol 85:197–210

Chapter 5

Electrical Coupling of Axons Gunnar Birke, Dietmar Schmitz, and Andreas Draguhn

Traditionally, axons are described as cables for actively propagating output signals of neurons. In this “textbook” view, axons provide a hard-wired coupling device between excitable cells without creating autonomous computational devices. Neuronal information processing, plasticity, and dynamic functions of central networks are then functions of chemical synapses and dendritic integration. Over the past decades, several lines of evidence have challenged this traditional, rather static concept of axons. Important discoveries include dendritic backward propagation of action potentials (Eccles et al., 1966; Spruston et al., 1995; Sjöström et al., 2008), branching point failures of propagating action potentials (Debanne et al., 1997), ectopic origin of spikes at downstream locations (Stasheff et al., 1993; Avoli et al., 1998), and propagation of subthreshold synaptic potentials along axons (Alle and Geiger, 2006; Shu et al., 2006). None of these extensions, however, challenges the concept of information processing in synaptically coupled networks, including dendritic integration and oneto-one connection of neurons by axons. Here we will summarize recent evidence for neuronal network activity arising within true, autonomous axonal networks. These networks are constituted by electrical coupling between axons, providing a pathway for fast and efficient “lateral” transfer of action potentials and circumventing signal integration at chemical synapses. Thereby, electrically coupled axons can generate highly coherent patterns of action potentials in the “output” pathway of neuronal networks.

5.1 Mechanisms of Electrical Coupling In principle, electrical coupling of excitable cells can be achieved by two different mechanisms – ephaptic interactions and gap junctions. The first mechanism is

G. Birke (B) Institut für Physiologie und Pathophysiologie, Universität Heidelberg, Heidelberg, Germany e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_5, 

85

86

G. Birke et al.

based on the negative polarization of the extracellular space around excited membranes (generated by the influx of positive ions). If a nearby membrane is exposed to this negative field potential, the potential gradient across this membrane becomes smaller. This will, in turn, increase the open probability of sodium channels, ultimately triggering action potentials (Taylor and Dudek, 1984; Jefferys, 1995). Ephaptic interaction is generally considered to be rather unspecific, bypassing the precisely wired neuronal connections. Moreover, it requires high-field potential gradients, as generated by massive synchronous discharges. It is, therefore, not surprising that ephaptic transmission has been mostly discussed in pathophysiological states. It may contribute to the generation and propagation of epileptic discharges (Jefferys and Haas, 1982; Taylor and Dudek, 1982; Traub et al., 1985) or to aberrant signal propagation in damaged fibre bundles (Smith and McDonald, 1999). It should be kept in mind, however, that activity-dependent extracellular field potential changes are also present in healthy tissue. As an example, it has been suggested that potential changes in the vicinity of excited axon terminals may affect ion flow through postsynaptic receptors, contributing to synaptic plasticity in the immediately opposed postsynaptic cell (Voronin, 2000). The second, more specific mechanism of electrical signalling is based on specialized “electrical synapses”. These structures have been termed “gap junctions” due to the close membrane apposition of two adjacent cells which create a very narrow extracellular gap in electron micrographs (Revel and Karnovsky, 1967). Today we know that gap junctions contain large intercellular channels which allow for the diffusion of ions and small organic molecules. These gap junction channels are made up of connexins forming hexameric “connexons” within each cell. In order to build a functional gap junction channel, two connexons from adjacent cells have to meet by lateral diffusion to join (for review see (Bennett and Zukin, 2004)). A morphological gap junction may be made up by hundreds of such channels (Bukauskas et al., 2000), although there is no obvious lower limit for this number. Gap junction channels are characterized by their large diameter pore (∼1 nm) which is highly permeable to ions of both charges and to remarkably large organic molecules (up to ∼1 kDa). Their gating can be weakly voltage dependent and conductance can show some rectification, although effects of membrane potential are far smaller than for typical voltage-dependent ion channels. Gap junctions close upon acidification of the intracellular space (Spray et al., 1982) and upon increases in intracellular concentration of Ca2+ (Spray and Bennett, 1985; Peracchia, 2004). They are also modulated by second messenger cascades which may be activated by neurotransmitters/neuromodulators like serotonin or dopamine (Rörig et al., 1995a; Rörig and Sutor, 1996b). More than 20 isoforms of connexins have been described in the mammalian CNS (Willecke et al., 2002; Bennett and Zukin, 2004), allowing for multiple different homo- or heterotypic gap junctions (i.e. gap junctions made up by similar or different subunits, respectively). Despite some knowledge about different biophysical properties, the biological relevance of the large heterogeneity of connexins is not fully understood. Besides biophysical or regulatory differences between connexin isoforms, the differential regulation of connexin expression may be important.

5

Electrical Coupling of Axons

87

Indeed, expression studies have revealed specific, though partly overlapping, occurrence of connexin isoforms in different cell types. At present, at least five different connexins have been found in mammalian neurons: connexin (Cx)30.2, Cx36, Cx45, Cx50, and Cx57 (Condorelli et al., 2000; Rash et al., 2001a,b; Maxeiner et al., 2003; Hombach et al., 2004; Rash et al., 2005; O’Brien et al., 2006; Kreuzberg et al., 2008). Recently, another family of intercellular channel proteins has been identified in mammals. These molecules are now called pannexins and had been originally found in invertebrates (hence called “innexins”). Following their detection in the human genome by Panchin and coworkers (Panchin et al., 2000), Bruzzone and coworkers could show that pannexins form intercellular channels with similar biophysical properties as connexin-based gap junctions (Panchin et al., 2000; Bruzzone et al., 2003). Pannexin 1 and pannexin 2 are expressed in neurons (Vogt et al., 2005; Zappala et al., 2007; Zoidl et al., 2007), opening a new molecular pathway for electrical coupling. In addition, pannexin hemichannels may be involved in non-vesicular transmitter release from sensory cells (Huang et al., 2007). Activation of pannexin-1 hemichannels can also augment bursting in the hippocampus. It has been concluded that Px1 hemichannel opening is induced by NMDARs and thereby can contribute to epileptiform activity. Such hemichannels can also be constituted by connexins. The current flow through connexin26-based hemichannels in retinal horizontal cells does directly influence transmitter release from cones (Kamermans et al., 2001). This signalling mechanism shares some properties of ephaptic interactions (see above) but is spatially highly specific. Electrical coupling between neurons is not easy to analyse. Over the past decades, different and complementary experimental approaches have been chosen, many of them providing rather indirect evidence for coupling. This is not at least due to the lack of specific pharmacological tools which are far behind those used in classical synaptic physiology. A widely used paradigm is the observation of dye transfer from one neuron to a secondary cell (Andrew et al., 1982); Rörig et al., 1995a, b; Fukuda and Kosaka, 2000). In this technique, the dye is usually introduced by a microelectrode, assuming that the only way in which it could propagate into the secondary cell is through intercellular channels. However, this method may produce false-positive results (see below). More direct proofs for the existence of gap junctions are therefore based on ultrastructural evidence (Rash et al., 2000, 2001a, b; Fukuda et al., 2006; Hamzei-Sichani et al., 2007). Functional studies make use of voltage coupling in simultaneous recordings from pairs of neurons (MacVicar and Dudek, 1981; Venance et al., 2000; Landisman et al., 2002; Galarreta et al., 2004), pharmacological block of gap junctions (e.g. Draguhn et al., 1998; Dupont et al., 2006), and genetic ablation of connexins (Sohl et al., 2005). Problems and limitations of these methods will be discussed below. Nevertheless, the cumulative evidence from many different studies does provide firm evidence for electrical coupling between neurons in various mammalian brain regions, including the hippocampus, neocortex, inferior olive, thalamus, amygdala, brainstem, and retina (Dermietzel, 1998; Willecke et al., 2002; Bennett and Zukin, 2004; Sohl et al., 2005). Mostly, such gap junctions are believed to be located between the dendrites of neurons, as has been shown, e.g. for hippocampal (Fukuda and Kosaka, 2000)

88

G. Birke et al.

and neocortical interneurons (Fukuda et al., 2006). Here, however, we will focus on the special case of gap junction coupling between axons. At this strategic location, electrical synapses exert unique effects with peculiar consequences for network behaviour. Due to the low input capacitance and high excitability of axons, they can trigger action potentials with very high efficacy. This makes them fundamentally different from dendro-dentritic gap junctions where signal processing has much in common with excitatory chemical synapses. We will review the evidence for axo-axonic gap junctions, focussing on the mammalian hippocampus. We will then discuss the functional consequences of such junctions on signalling in neuronal networks. Before going into detail, however, we shall briefly summarize the peculiar experimental difficulties which arise when electrical coupling is postulated between small neuronal structures.

5.2 Methodological Issues The methods to study gap junctions are less advanced than those available for the analysis of chemical synaptic transmission. This makes it sometimes difficult to obtain reliable evidence for electrical coupling between specific neurons. Axoaxonic gap junctions are especially difficult to find, since they are located along very thin processes at some unknown distance from the soma. Moreover, efficient coupling of axons may be achieved by a very small number of gap junction channels (Traub et al., 1999). Here, we will briefly summarize the most commonly used methods for analysis of electrical coupling between neurons. One widely used approach is the application of gap junction-blocking drugs in vitro. These include long-chain alcohols, the anaesthetic drug halothane, glycyrrhetinic acid, and its derivative carbenoxolone, anandamide, oleamides, fenamate, and mefloquine (Harks et al., 2001; Harris and Bevans, 2001; Srinivas and Spray, 2003). However, all of these drugs have serious side-effects and may even lack efficacy in certain preparations (Rouach et al., 2003). Therefore, effects of such drugs are difficult to interpret. This problem is even more significant when complex preparations are studied where multiple neuronal signalling mechanisms act in parallel and the readout is indirect, e.g. alterations of network oscillations. A widely used approach is to apply a series of different gap junction blockers and measure their respective effects on a common function (e.g. a certain network oscillation). If all of them consistently block the observed behaviour, the data are taken as cumulative evidence for a causal role of gap junctions. This is, however, a dangerous argument, and the development of more specific gap junction blockers would be highly desirable. Efforts of several groups have made significant progress in this field, e.g. by defining the Cx36-specificity of mefloquine (Cruikshank et al., 2004) and by developing gap junction-mimetic peptides (Evans and Boitano, 2001; Dahl, 2007; Evans and Leybaert, 2007). A frequent functional assay for the existence of gap junctions is dye coupling. Low molecular weight dyes can pass through the wide pore of gap junction

5

Electrical Coupling of Axons

89

channels, though with potential differences between different connexin isoforms. There is, however, a long-standing debate about the so-called shish kebab-artefact, i.e. the distribution of dye traces along the shaft of the micropipette which may have hurt more than one cell during penetration of the tissue. In brain slices it is also feasible that cut dendrites or axons at the edge of the tissue slab re-join in an end-to-end conformation, leading to structural coupling independent from gap junctions. Although these methodological issues always have to be considered, dye transfer has been shown to be sensitive to biochemical modulators of gap junctions (Rörig and Sutor, 1996a, b) and to changes in pH, consistent with the gating of gap junctions (Church and Baimbridge, 1991). Dye coupling also follows the developmental time course of electrical coupling, i.e. it declines during postnatal development (Kandler and Katz, 1995; Dupont et al., 2006). Time-lapsed imaging of dye transfer (Schmitz et al., 2001; Haag and Borst, 2005) or combination of gap junction permeable and impermeable dyes (Haag and Borst, 2005) can provide additional evidence for the specificity of dye transfer. Therefore, dye coupling can still be considered a useful method, although it cannot provide direct proof for the existence of gap junctions. The two most convincing arguments for cellular coupling are electrical signal transfer in paired recordings and ultrastructural documentation of gap junctions. Paired recordings from two mutually coupled cells can be obtained in brain slices and does provide direct evidence for electrical coupling (MacVicar and Dudek, 1981; Galarreta and Hestrin, 1999; Venance et al., 2000; Deans et al., 2001; Landisman and Connors, 2005). Typically, current injection into cell 1 causes a voltage deflection in cell 1 as well as in cell 2. The ratio between both voltage changes is called “coupling coefficient” and is a measure for the efficacy of charge transfer through the gap junction. It should, however, be noted that coupling between remote axonal compartments may yield very small coupling coefficients in somatic recordings. Therefore, axonal coupling may escape detection by coupled DC voltage shifts. Structurally, the best evidence for the existence of gap junctions is electron microscopy. Gap junctions have a typical appearance in ultra-thin sections (Revel and Karnovsky, 1967). They are, however, easily missed if the section does not hit the junction exactly perpendicular to its extension. Larger parts of neuronal membranes can be screened in freeze-fracture replica which allow for the detection of connexin clusters. This method has been successfully applied to the question of axonal coupling in the hippocampus (Hamzei-Sichani et al., 2007). It is especially powerful when combined with gold-labelled antibodies against defined connexion subspecies, as done by Hamzei-Sichani and coworkers. Molecular cloning of cDNA underlying different connexin isoforms has opened the possibility to construct genetically modified animals which lack specific connexin isoforms. With respect to axonal coupling, however, these transgenic animals have not yet revealed major new insights. At the network level, sharp wave–ripple complexes are a good candidate for functional effects of axo-axonic gap junctions. Ripples are, however, largely undisturbed in several connexin “knockout” mice, including those lacking Cx36 (Maier et al., 2002; Buhl et al., 2003; Pais et al., 2003),

90

G. Birke et al.

Cx30.2 (Kreuzberg et al., 2008), and Cx45 (AD and GB, unpublished observation). In our own study on sharp wave–ripple complexes in brain slices of mice lacking Cx36 (Maier et al., 2002), we found a reduced incidence of the events which, themselves, were not different from similar events in control mice. Again, these data show that the fast synchronization at ∼200 Hz does not require expression of Cx36. At the cellular level, effects of these knockouts on electrical coupling between axons have not been directly examined. The lacking effect of connexin knockout models on ripples may be caused by promiscuous expression of different connexins as well as by compensatory expression of other molecules in constitutive knockout mice. Conditional, cell-specific or multiple knockout mice might help to reveal clearer results in the future. In contrast to ripples, network gamma oscillations are affected by deletion of Cx36 (Hormuzdi et al., 2001; Buhl et al., 2003). Here, interneurons show reduced electrical coupling, in parallel to decreased power and regularity of carbachol- and kainate-induced network gamma rhythms. This effect may, however, be caused by reduced coupling between the dendrites, rather than axons, of interneurons (Fukuda and Kosaka, 2000; Fukuda et al., 2006).

5.3 Axo-axonal Synapses Gap junctions between neuronal dendrites are now firmly established, following initial scepticism (Bennett and Pereda, 2006). Nevertheless, we are still far from fully understanding their functional impact on network activity, plasticity, and cognition in mammals. However, gap junctions between axons are even less widely accepted and understood. This may be partly due to experimental difficulties in identifying junctions at this specific location. A more important source of scepticism may be, however, that axo-axonic gap junctions challenge our canonical view of neuronal networks. Suppose that two axons of pyramidal cells are electrically coupled at some distance from the soma. An action potential in cell 1 will then cause a depolarization in axon 2. Due to the low input capacitance of axons, the rise time of the gap junction-mediated potential in axon 2 will be quite fast and, therefore, action potential threshold can be reached within very short time. Transition of spikes can be as fast as 0.2 ms (MacVicar and Dudek, 1982; Moortgat et al., 1998; Traub et al., 1999). Thus, the action potential in the primary neuron will cause a tightly correlated spike in the coupled axon. A second important feature of this coupling mechanism is that the generation of the secondary spike will be independent from dendritic signals, provided that the gap junction is located at sufficient distance from the soma (Traub et al., 1999). As a result, axonally coupled neurons function as synchronized output devices with minimal sensitivity to synaptic signal flow in the network. This makes axonal coupling fundamentally distinct from most mechanisms of neuronal communication and may restrict their distribution in the CNS. In summary, combined evidence from experiments and modelling makes it likely that axo-axonic gap junctions exist in certain neuronal networks.

5

Electrical Coupling of Axons

91

While we will focus on gap junctions linking two axonal cylinders, there are other relevant locations for axonal gap junctions, e.g. between two presynaptic terminals (Muller et al., 2005; Cuntz et al., 2007) or between the pre- and postsynaptic sites of mixed chemical–electrical synapses (Korn and Faber, 1976; Yang et al., 1990) The first case is not principally different from junctions between axonal cylinders since it follows the same topology. It may, however, specifically enhance synchrony of transmitter release and thereby potentiate the efficacy of such joined chemical synapses. Similar to junctions between the axonal cylinders, electrically coupled terminals may also elicit action potentials which propagate antidromically along the axon and, eventually, invade the soma. In contrast, mixed chemical/electrical synapses are functionally distinct from inter-axonal gap junctions. They connect an axon with its postsynaptic target cell, mediating very fast synaptic excitation which is then followed by a slower chemically mediated postsynaptic potential. The most famous example of such a mixed synapse is the Mauthner cell (Korn and Faber, 1976). Notably, the electrical synapses at Mauthner cells follow rules of activity-dependent synaptic plasticity (Yang et al., 1990). This indicates that electrical coupling takes part in the functional adaptation in neuronal circuits, similar to chemical synapses. Recent evidence indicates that mixed chemical–electrical synapses may also be present between principal neurons in the mammalian hippocampus, opening novel possibilities for signalling in the CNS (Gutierrez and Vivar, 2008). Evidence for axo-axonal gap junctions has been provided in several different species and neuronal circuits. An early finding was dye coupling between the axons of retinal horizontal cells in the rabbit (Vaney, 1993). Interestingly, recent findings show similar axo-axonic electrical coupling between visual interneurons in flies (Cuntz et al., 2007). Axons of commissural interneurons in the spinal cord of fish are coupled to axons of Mauthner cells which mediate fast motor reflexes (Yasargil and Sandri, 1990). The latter example does already point towards a possible function of axo-axonic gap junctions, namely particularly rapid signalling from one axon to the other, omitting the time needed for chemical synaptic transmission, dendritic integration, and EPSP-spike coupling. The high velocity of signal transmission between coupled axons becomes even more apparent in examples from weakly electric fish which generate ultra-fast network rhythms at frequencies of >1,000 Hz (Dye and Heiligenberg, 1987; Moortgat et al., 1998). Here, pacemaker neurons in the medullary nucleus are electrically coupled by axo-axonal gap junctions (Tokunaga et al., 1980; Dye and Heiligenberg, 1987). Electrical coupling between these cells is crucial for maintenance of the precise synchronization between discharges, allowing for spike transition within about 0.25 ms (Moortgat et al., 1998). In mammals, coupling has been shown to occur between axon terminals of parvalbumin-immunoreactive interneurons in the rat basolateral amygdala (Muller et al., 2005). Again, this mechanism opens the possibility of highly synchronized synaptic output signals. In 1998, we have hypothesized axo-axonic coupling between principal neurons in the rat hippocampus (Draguhn et al., 1998). This region is involved in the formation and consolidation of declarative and spatial memory and displays a variety of state-dependent network oscillations. Selected neurons of the principal cell layers

92

G. Birke et al.

(granule cells and pyramidal cells, respectively) fire precisely timed action potentials during these oscillations, probably constituting a temporal representation of spatial relationships in the environment (O’Keefe and Recce, 1993; Wilson and McNaughton, 1994; Draguhn et al., 2000). The fastest type of network oscillations in the hippocampus is called ripples and consists of small field potential fluctuations at ∼ 200 Hz, superimposed on propagating excitatory sharp waves (Buzsaki et al., 1992). During ripples, individual pyramidal cells fire action potentials with ∼1 ms precision both in vivo (Ylinen et al., 1995; Csicsvari et al., 1999) and in vitro (Maier et al., 2003; Both et al., 2008). This highly stringent temporal coupling between individual neurons and the underlying network rhythm is a prerequisite for the organized spatio-temporal activity pattern during ripples (Draguhn et al., 2000; Buzsaki and Draguhn, 2004). In our initial description of ultra-fast network rhythms in rat hippocampal slices we observed ripple-associated spikelets and fast prepotentials in pyramidal cells (Fig. 5.1). Spikelets are small amplitude depolarizations with rapid onset and exponential decay. Fast prepotentials were initially found as small voltage deflections at foot of action potentials which have been interpreted as somatic reflections of

Fig. 5.1 Spikelets and fast prepotentials in a CA3 pyramidal neuron accompanying fast network oscillations. a original recording. Upper trace shows field potential with three subsequent population events at ∼5 ms intervals. Lower trace shows parallel whole-cell recording from a CA3 pyramidal cell. Note that the network events are synchronous to spikelets and a fast prepotential in the cell. b Simulated antidromic action potentials and spikelets in a pyramidal cell (upper trace) and in a dendritically coupled neuron (lower trace). Note the slow waveform of coupling potentials which do not resemble the fast kinetics of experimentally observed spikelets. c Similar simulation with axonally coupled axons. Note the fast kinetics of spikelets in the secondary, coupled neuron (lower trace). Modified from (Draguhn et al., 1998)

5

Electrical Coupling of Axons

93

dendritic action potentials (Kandel and Spencer, 1961). Later on, they were also observed in CA1 pyramidal cells under epileptogenic conditions (Schwartzkroin and Prince, 1977, 1978) which have initially been interpreted as somatic reflections of dendritic action potentials (Kandel and Spencer, 1961). Later, simultaneous paired recordings from coupled hippocampal neurons revealed that fast prepotentials are generated by electrical coupling (MacVicar and Dudek, 1981, 1982). This interpretation was corroborated by experiments in calcium-free solution, where chemical synaptic transmission is largely absent but spikelets and fast prepotentials are highly abundant (Perez-Velazquez et al., 1994; Valiante et al., 1995). The potentials recorded in these studies were correlated with dye coupling and were sensitive to drugs known to block gap junctions. Since then, several studies have found that spikelets are indeed linked to the occurrence of electrical coupling (Galarreta and Hestrin, 1999; Gibson et al., 1999; Mann-Metzer and Yarom, 1999; Tamas et al., 2000; Venance et al., 2000; Schmitz et al., 2001). Thus, the occurrence of spikelets during ripples was indicative for a role of gap junctions in this highly coordinated, fast network activity. Several further experiments were consistent with this notion (Draguhn et al., 1998; Maier et al., 2003). Field oscillations were reduced by halothane, octanol, and all of which are known to block gap junctions (Harks et al., 2001; Spray et al., 2002). Moreover, intracellular alkalinization or washout of Ca2+ from the extracellular solution increased, rather than decreased the field oscillation, again consistent with the pH- and calciumsensitive gating of gap junctions. We therefore suggested that ripple-synchronous spikelets and fast prepotentials were induced by electrical coupling between pyramidal cells. In order to unravel the precise cellular mechanisms behind the generation of spikelets, we then performed computer simulations, based on a detailed compartmental model of CA3 pyramidal cells (Traub et al., 1994). However, when dendritic or somatic coupling of two pyramidal cells was introduced into this model, we failed to reproduce the fast rise times of spikelets. The slow kinetics of the resulting coupling potentials was due to the low-pass filtering effect of the resistor–capacitor (RC) circuit formed by the combined gap junction conductance and postsynaptic membrane capacitance. By exclusion, it was then found that the only localization of gap junctions reproducing the observed fast rise time was between the axons (Fig. 5.1). As mentioned above, this result can be explained by the small capacitance of the axonal membrane and by its high excitability. Small voltage deflections of the primary axon cause rapid depolarizations in the neighbouring membrane region of the secondary axon which are then rapidly amplified and propagated by activation of the densely expressed axonal sodium channels. We concluded that the somatically recorded spikelets result from axonally generated action potentials which are antidromically conducted towards the soma. Depending on the state of the soma and initial segment, these antidromic spikes may trigger full somatic action potentials. The invading antidromic spike is then reflected as preceding by “fast prepotential” or “notch” at the onset of the full somatic action potential. If the axonal action potential fails to trigger a full spike in the axon initial segment, it takes the form of spikelets or partial spikes which are basically aborted waveforms of ectopically generated action potentials.

94

G. Birke et al.

Up to this point, evidence for axonal gap junctions in the hippocampus was mostly produced by modelling, calling for more direct experimental proof. In a subsequent study, we therefore recorded from principal cells in hippocampal slices and evoked spikelets by electrical stimulation in the axonal region (Schmitz et al., 2001). Direct stimulation of the axon caused typical antidromic action potentials which rose steeply from resting potential and displayed a pronounced depolarizing afterpotential. When the stimulation pipette was moved slightly further away from the recorded cell (∼20–200 μm), we could regularly induce spikelets. Increasing stimulation intensity at this location led to the generation of full antidromic action potentials with fast prepotentials (“notches”). Importantly, these signals were elicited in both normal extracellular fluid and in calcium-free conditions, reducing the probability that they were mediated by input from chemical synapses. These findings corroborated our previous assumption that spikelets are the somatic correlate of antidromically conducted action potentials. In a more direct approach we recorded simultaneously from the axon and soma of individual CA1 pyramidal cells (Fig. 5.2). When spikelets were induced by remote extracellular stimulation, they appeared first at the axonal recording site (located ∼20 μm from the soma) before reaching the somatic pipette. Paired somatic–dendritic recordings showed that spikelets arrived even later in the apical dendrite. Thus, spikelets invade the soma and the apical dendrite of pyramial cells from the axon, consistent with the axonal coupling hypothesis. Several lines of evidence indicated that the generation of spikelets and fast prepotentials were brought about by gap junctions. They were strongly attenuated and slowed down in the presence of carbenoxolone, a known blocker of gap junctions (Davidson and Baumgarten, 1988). The same pharmacological treatment did not change intrinsic properties of neurons, directly evoked antidromic spikes, or action potentials in axon terminals of granule cells. Similar results were obtained for octanol and halothane and for intracellular acidification. Combined, these findings converge on the existence of gap junctions in axons of principal cells. Finally, dye

Fig. 5.2 Antidromic generation of spikelets. Left panel (a) shows the experimental setup with synchronous somatic and axonal whole-cell recording from a pyramidal cell and the remotely located stimulation pipette. Right panel (b) shows the recorded stimulation-induced spikelets (vertical deflection amounts to about 3 mV, traces arbitrarily scaled). Note that the signal appears in the axon before invading the soma. Modified from (Schmitz et al., 2001)

5

Electrical Coupling of Axons

95

coupling between hippocampal neurons was observed with time-lapsed imaging. Filling one cell with the fluorescent dye rhodamin-123 during whole-cell recording resulted in the transition of dye into neighbouring cells. In four neurons, the point of dye transfer into the secondary cell could be identified at 50–120 μm from the soma, between spineless processes reminiscent of axons. This is within the expected length of the unmyelinated part of these axons (Kosaka, 1980). “Seeing is believing” – that is, the ultimate proof for the existence of gap junctions in a specific structure requires direct structural evidence. Therefore, Hamzei-Sichani and coworkers (Hamzei-Sichani et al., 2007) analysed mossy fibres of rat dentate granule cells by thin-section transmission electron microscopy and by grid-mapped freeze-fracture replica immunogold labelling. Indeed, they found gap junctions as well as axonal profiles with immunoreactivity for connexin36 (Fig. 5.3). These data provide the first ultrastructural evidence for gap junctions in the axons of cortical principal cells. They should end a long debate about whether this way of communication does indeed exist.

Fig. 5.3 Structural evidence for axonal gap junctions in hippocampal mossy fibres. (a) Upper left panel shows an electron micrograph of a gap junction between two mossy fibres. Lower left (b) and right panels (c) show Cx36-immunoreactivity on an axonal profile in a freeze-fracture replica immunogold labelling. From (Hamzei-Sichani et al., 2007). Copyright (2007) National Academy of Sciences, USA

5.4 Physiological Significance of Axo-axonic Gap Junctions What makes axo-axonic gap junctions special as compared to “conventional” electrical synapses between dendrites or somata of neurons? One important property of gap junctions at this special location is that they can mediate spike transition

96

G. Birke et al.

between two neurons within the extremely short time of about 0.25 ms (Traub et al., 1999). As outlined above, this is due to the very small compartment which has to be depolarized in order to activate voltage-gated sodium channels in the immediate surrounding. These first active channels exert positive feedback on their neighbours, giving rise to full spikes. As a result, spikes propagate between axons at very high speed, synchronizing the output of coupled neurons. It is, therefore, not surprising that modelling and experimental work has indicated a role for axoaxonic gap junctions in very fast (>70 Hz) network oscillations (Traub et al., 1999, 2002; Maex and De Schutter, 2007). In addition, axo-axonic coupling provides a unique mechanism for signalling within neuronal networks: spikes which are directly generated in the secondary axon do not depend on chemical synaptic inputs and dendritic–somatic signal integration. In this situation, neurons are mere “output devices”, making them independent from afferent signal flow (Traub et al., 1999). Although we are far from understanding the impact of axo-axonic gap junctions on cognitive functions, this notion is well compatible with a present concept of memory consolidation in the hippocampus (Buzsaki, 1989). During exploration of an environment, pyramidal cells in the rat hippocampus fire at selected places and are, therefore, termed place cells (O’Keefe and Dostrovsky, 1971). This firing is entrained by underlying theta and gamma network oscillations with systematically shifting phase relationships between the population rhythm and action potentials of individual cells (O’Keefe and Recce, 1993). As a result, place cells generate sequences of firing which represent the animal’s trajectory through the explored area. During subsequent periods of slow-wave sleep or inactivity, these sequences are repeated, on top of faster (∼200 Hz) ripple oscillations (Wilson and McNaughton, 1994; Ji and Wilson, 2007). The two-stage model of spatial memory formation proposes that the re-activated cell assemblies form an output signal from the hippocampus to the neocortex, where long-term memories may be stored. Thus, the high-frequency oscillations appear during a state where the hippocampal formation does not receive major input signals. A network of axonally coupled neurons could well serve this function. It is unclear, however, how the specificity of coupling between those neurons carrying the relevant information is achieved. If axonal coupling is the key mechanism for this activity, we can assume that the underlying plasticity process involves expression, axonal sorting, membrane insertion, or gating of gap junctions. This assumption is well compatible with other processes following synaptic plasticity in the hippocampus where calcium-dependent signals do reach the nucleus and induce major changes in expression of signalling molecules (Bading, 2000). A rigorous version of this concept has been put forward by (Traub et al., 1999), using computer simulations. Neurons were randomly connected by axonal gap junctions, except for constraints of distance (up to 200 μm) and maximal number of directly coupled neurons (up to 4). When a critical threshold for coupling probability was exceeded, the resulting pseudo-random graphs contained one dominant, large cluster of neurons. This threshold was only slightly above 1, meaning that relatively few neurons have to be connected to more than one other neuron in order to form such a dominant assembly. In this simulated network, axons expressed

5

Electrical Coupling of Axons

97

highly synchronized spikes at >100 Hz, resulting in a fast network oscillation. Somatic action potentials (preceded by fast prepotentials) were less frequent, consistent with the experimental observation that pyramidal cells fire mostly only one action potential on a train of ripples (Csicsvari et al., 1999). Thus, the axonal plexus can form a largely autonomous network with perlocating ortho- and antidromic spikes. An assumption of our network model was that ectopic axonal spikes are generated spontaneously at low frequencies. This is consistent with experimental observations, at least under conditions of epileptogenesis or after strong synaptic stimulation (Stasheff et al., 1993; Traub et al., 1995; Avoli et al., 1998; Traub et al., 2001). Subsequent experimental and theoretical work showed that such fast network oscillations are also influenced by synaptic processes, most prominently by GABAergic inhibition (Traub et al., 1996; Traub and Bibbig, 2000; Traub et al., 2004). Indeed, certain hippocampal interneurons fire at high (ripple) frequencies during sharp wave–ripple complexes in vivo (Ylinen et al., 1995; Klausberger et al., 2003, 2005). Axo-axonic gap junctions have, meanwhile, been suggested to play a causal role for network patterning in various synchronized oscillations (Traub et al., 2002). They may also be involved in the generation of hypersynchronous epileptic discharges (Perez Velazquez and Carlen, 2000; Traub et al., 2001), a topic of major medical relevance. However, in order to fully understand the functional significance of electrical axo-axonic coupling we need further experiments and improved methodological approaches. This will require combined efforts from cellular, systems, genetic, and computer neuroscientists. Important questions include • • • •

Which types of neurons are axonally coupled? What is the frequency of axonal gap junctions in a given population of neurons? Which molecular constituents make up the electrical synapses? What are the precise biophysical consequences of electrical coupling between axons? • At which section(s) of the axon are neurons coupled? • Are axonal gap junctions plastic? • How are neurons selected for participation in an axonally coupled assembly? Future work will hopefully clarify these issues in appropriate experimental model systems. This will help to understand the role of this unconventional signalling mechanism in fast synchronization of transmitter release, network oscillations, and in pathophysiological states like epilepsy.

References Alle H, Geiger JR (2006) Combined analog and action potential coding in hippocampal mossy fibers. Science 311(5765):1290–1293 Andrew RD, Taylor CP et al (1982) Coupling in rat hippocampal slices: dye transfer between CA1 pyramidal cells. Brain Res Bull 8(2):211–222

98

G. Birke et al.

Avoli M, Methot M et al (1998) GABA-dependent generation of ectopic action potentials in the rat hippocampus. Eur J Neurosci 10(8):2714–2722 Bading H (2000) Transcription-dependent neuronal plasticity the nuclear calcium hypothesis. Eur J Biochem 267(17):5280–5283 Bennett MV, Pereda A (2006) Pyramid power: principal cells of the hippocampus unite! Brain Cell Biol 35(1):5–11 Bennett MV, Zukin RS (2004) Electrical coupling and neuronal synchronization in the Mammalian brain. Neuron 41(4):495–511 Both M, Bahner F et al (2008) Propagation of specific network patterns through the mouse hippocampus. Hippocampus 18(9):899–908 Bruzzone R, Hormuzdi SG et al (2003) Pannexins, a family of gap junction proteins expressed in brain. Proc Natl Acad Sci USA 100(23):13644–13649 Buhl DL, Harris KD et al (2003) Selective impairment of hippocampal gamma oscillations in connexin-36 knock-out mouse in vivo. J Neurosci 23(3):1013–1018 Bukauskas FF, Jordan K et al (2000) Clustering of connexin 43-enhanced green fluorescent protein gap junction channels and functional coupling in living cells. Proc Natl Acad Sci USA 97(6):2556–2561 Buzsaki G (1989) Two-stage model of memory trace formation: a role for “noisy” brain states. Neuroscience 31(3):551–570 Buzsaki G, Draguhn A (2004) Neuronal oscillations in cortical networks. Science 304(5679):1926–1929 Buzsaki G, Horvath Z et al (1992) High-frequency network oscillation in the hippocampus. Science 256(5059):1025–1027 Church J, Baimbridge KG (1991) Exposure to high-pH medium increases the incidence and extent of dye coupling between rat hippocampal CA1 pyramidal neurons in vitro. J Neurosci 11(10):3289–3295 Condorelli DF, Belluardo N et al (2000) Expression of Cx36 in mammalian neurons. Brain Res Brain Res Rev 32(1):72–85 Cruikshank SJ, Hopperstad M et al (2004) Potent block of Cx36 and Cx50 gap junction channels by mefloquine. Proc Natl Acad Sci USA 101(33):12364–12369 Csicsvari J, Hirase H et al (1999) Oscillatory coupling of hippocampal pyramidal cells and interneurons in the behaving Rat. J Neurosci 19(1):274–287 Cuntz H, Haag J et al (2007) Robust coding of flow-field parameters by axo-axonal gap junctions between fly visual interneurons. Proc Natl Acad Sci USA 104(24):10229–10233 Dahl G (2007) Gap junction-mimetic peptides do work, but in unexpected ways. Cell Commun Adhes 14(6):259–264 Davidson JS, Baumgarten IM (1988) Glycyrrhetinic acid derivatives: a novel class of inhibitors of gap-junctional intercellular communication. Structure-activity relationships. J Pharmacol Exp Ther 246(3):1104–1107 Deans MR, Gibson JR et al (2001) Synchronous activity of inhibitory networks in neocortex requires electrical synapses containing connexin36. Neuron 31(3):477–485 Debanne D, Guerineau NC et al (1997) Action-potential propagation gated by an axonal I(A)-like K+ conductance in hippocampus. Nature 389(6648):286–289 Dermietzel R (1998) Gap junction wiring: a ‘new’ principle in cell-to-cell communication in the nervous system? Brain Res Brain Res Rev 26(2–3):176–183 Draguhn A, Traub RD et al (2000) Ripple (approximately 200-Hz) oscillations in temporal structures. J Clin Neurophysiol 17(4):361–376 Draguhn A, Traub RD et al (1998) Electrical coupling underlies high-frequency oscillations in the hippocampus in vitro. Nature 394(6689):189–192 Dupont E, Hanganu IL et al (2006) Rapid developmental switch in the mechanisms driving early cortical columnar networks. Nature 439(7072):79–83 Dye J, Heiligenberg W (1987) Intracellular recording in the medullary pacemaker nucleus of the weakly electric fish, Apteronotus, during modulatory behaviors. J Comp Physiol [A] 161(2):187–200

5

Electrical Coupling of Axons

99

Eccles JC, Llinas R et al (1966) The action of antidromic impulses on the cerebellar Purkinje cells. J Physiol 182(2):316–345 Evans WH, Boitano S (2001) Connexin mimetic peptides: specific inhibitors of gap-junctional intercellular communication. Biochem Soc Trans 29(Pt 4):606–612 Evans WH, Leybaert L (2007) Mimetic peptides as blockers of connexin channel-facilitated intercellular communication. Cell Commun Adhes 14(6):265–273 Fukuda T, Kosaka T (2000) Gap junctions linking the dendritic network of GABAergic interneurons in the hippocampus. J Neurosci 20(4):1519–1528 Fukuda T, Kosaka T et al (2006) Gap junctions among dendrites of cortical GABAergic neurons establish a dense and widespread intercolumnar network. J Neurosci 26(13):3434–3443 Galarreta M, Erdelyi F et al (2004) Electrical coupling among irregular-spiking GABAergic interneurons expressing cannabinoid receptors. J Neurosci 24(44):9770–9778 Galarreta M, Hestrin S (1999) A network of fast-spiking cells in the neocortex connected by electrical synapses. Nature 402(6757):72–75 Gibson JR, Beierlein M et al (1999) Two networks of electrically coupled inhibitory neurons in neocortex. Nature 402(6757):75–79 Gutierrez R, Vivar C (2008) Electrical synaptic transmission in the mossy fiber-pyramidal cell synapse in healthy and epileptic rats. Soc Neurosci. Abstr. 734.11 Haag J, Borst A (2005) Dye-coupling visualizes networks of large-field motion-sensitive neurons in the fly. J Comp Physiol A Neuroethol Sens Neural Behav Physiol 191(5):445–454 Hamzei-Sichani F, Kamasawa N et al (2007) Gap junctions on hippocampal mossy fiber axons demonstrated by thin-section electron microscopy and freeze fracture replica immunogold labeling. Proc Natl Acad Sci USA 104(30):12548–12553 Harks EG, de Roos AD et al (2001) Fenamates: a novel class of reversible gap junction blockers. J Pharmacol Exp Ther 298(3):1033–1041 Harris AL, Bevans CG (2001) Exploring hemichannel permeability in vitro. Methods Mol Biol 154:357–377 Hinz B, Becher A et al (2001) Activity-dependent changes of the presynaptic synaptophysinsynaptobrevin complex in adult rat brain. Eur J Cell Biol 80(10):615–619 Hombach S, Janssen-Bienhold U et al (2004) Functional expression of connexin57 in horizontal cells of the mouse retina. Eur J Neurosci 19(10):2633–2640 Hormuzdi SG, Pais I et al (2001) Impaired electrical signaling disrupts gamma frequency oscillations in connexin 36-deficient mice. Neuron 31(3):487–495 Huang YJ, Maruyama Y et al (2007) The role of pannexin 1 hemichannels in ATP release and cell-cell communication in mouse taste buds. Proc Natl Acad Sci USA 104(15): 6436–6441 Jefferys JG (1995) Nonsynaptic modulation of neuronal activity in the brain: electric currents and extracellular ions. Physiol Rev 75(4):689–723 Jefferys JG, Haas HL (1982) Synchronized bursting of CA1 hippocampal pyramidal cells in the absence of synaptic transmission. Nature 300(5891):448–450 Ji D, Wilson MA (2007) Coordinated memory replay in the visual cortex and hippocampus during sleep. Nat Neurosci 10(1):100–107 Kamermans M, Fahrenfort I et al (2001) Hemichannel-mediated inhibition in the outer retina. Science 292(5519):1178–1180 Kandel ER, Spencer WA (1961) Electrophysiology of hippocampal neurons. IV. Fast prepotentials. J Neurophysiol 24:272–285 Kandler K, Katz LC (1995) Neuronal coupling and uncoupling in the developing nervous system. Curr Opin Neurobiol 5(1):98–105 Klausberger T, Magill PJ et al (2003) Brain-state- and cell-type-specific firing of hippocampal interneurons in vivo. Nature 421(6925):844–848 Klausberger T, Marton LF et al (2005) Complementary roles of cholecystokinin- and parvalbuminexpressing GABAergic neurons in hippocampal network oscillations. J Neurosci 25(42): 9782–9793

100

G. Birke et al.

Korn H, Faber DS (1976) Vertebrate central nervous system: same neurons mediate both electrical and chemical inhibitions. Science 194(4270):1166–1169 Kosaka T (1980) The axon initial segment as a synaptic site: ultrastructure and synaptology of the initial segment of the pyramidal cell in the rat hippocampus (CA3 region). J Neurocytol 9(6):861–882 Kreuzberg MM, Deuchars J et al (2008) Expression of connexin30.2 in interneurons of the central nervous system in the mouse. Mol Cell Neurosci 37(1):119–134 Landisman CE, Connors BW (2005) Long-term modulation of electrical synapses in the mammalian thalamus. Science 310(5755):1809–1813 Landisman CE, Long MA et al (2002) Electrical synapses in the thalamic reticular nucleus. J Neurosci 22(3):1002–1009 MacVicar BA, Dudek FE (1981) Electrotonic coupling between pyramidal cells: a direct demonstration in rat hippocampal slices. Science 213(4509):782–785 MacVicar BA, Dudek FE (1982) Electrotonic coupling between granule cells of rat dentate gyrus: physiological and anatomical evidence. J Neurophysiol 47(4):579–592 Maex R, De Schutter E (2007) Mechanism of spontaneous and self-sustained oscillations in networks connected through axo-axonal gap junctions. Eur J Neurosci 25(11): 3347–3358 Maier N, Guldenagel M et al (2002) Reduction of high-frequency network oscillations (ripples) and pathological network discharges in hippocampal slices from connexin 36-deficient mice. J Physiol 541(Pt 2):521–528 Maier N, Nimmrich V et al (2003) Cellular and network mechanisms underlying spontaneous sharp wave-ripple complexes in mouse hippocampal slices. J Physiol 550(Pt 3):873–887 Mann-Metzer P, Yarom Y (1999) Electrotonic coupling interacts with intrinsic properties to generate synchronized activity in cerebellar networks of inhibitory interneurons. J Neurosci 19(9):3298–3306 Maxeiner S, Kruger O et al (2003) Spatiotemporal transcription of connexin45 during brain development results in neuronal expression in adult mice. Neuroscience 119(3):689–700 Moortgat KT, Keller CH et al (1998) Submicrosecond pacemaker precision is behaviorally modulated: the gymnotiform electromotor pathway. Proc Natl Acad Sci USA 95(8): 4684–4689 Muller JF, Mascagni F et al (2005) Coupled networks of parvalbumin-immunoreactive interneurons in the rat basolateral amygdala. J Neurosci 25(32):7366–7376 O’Brien JJ, Li W et al (2006) Coupling between A-type horizontal cells is mediated by connexin 50 gap junctions in the rabbit retina. J Neurosci 26(45):11624–11636 O’Keefe J, Dostrovsky J (1971) The hippocampus as a spatial map. Preliminary evidence from unit activity in the freely-moving rat. Brain Res 34(1):171–175 O’Keefe J, Recce ML (1993) Phase relationship between hippocampal place units and the EEG theta rhythm. Hippocampus 3(3):317–330 Pais I, Hormuzdi SG et al (2003) Sharp wave-like activity in the hippocampus in vitro in mice lacking the gap junction protein connexin 36. J Neurophysiol 89(4):2046–2054 Panchin Y, Kelmanson I et al (2000) A ubiquitous family of putative gap junction molecules. Curr Biol 10 (13):R473–R474 Peracchia C (2004) Chemical gating of gap junction channels; roles of calcium, pH and calmodulin. Biochim Biophys Acta 1662(1–2):61–80 Perez Velazquez JL, Carlen PL (2000) Gap junctions, synchrony and seizures. Trends Neurosci 23(2):68–74 Perez-Velazquez JL, Valiante TA et al (1994) Modulation of gap junctional mechanisms during calcium-free induced field burst activity: a possible role for electrotonic coupling in epileptogenesis. J Neurosci 14(7):4308–4317 Rash JE, Davidson KG et al (2005) Ultrastructural localization of connexins (Cx36, Cx43, Cx45), glutamate receptors and aquaporin-4 in rodent olfactory mucosa, olfactory nerve and olfactory bulb. J Neurocytol 34(3–5):307–341

5

Electrical Coupling of Axons

101

Rash JE, Staines WA et al (2000) Immunogold evidence that neuronal gap junctions in adult rat brain and spinal cord contain connexin-36 but not connexin-32 or connexin-43. Proc Natl Acad Sci USA 97(13):7573–7578 Rash JE, Yasumura T et al (2001a) Identification of cells expressing Cx43, Cx30, Cx26, Cx32 and Cx36 in gap junctions of rat brain and spinal cord. Cell Commun Adhes 8(4–6):315–320 Rash JE, Yasumura T et al (2001b) Cell-specific expression of connexins and evidence of restricted gap junctional coupling between glial cells and between neurons. J Neurosci 21(6):1983–2000 Revel JP, Karnovsky MJ (1967) Hexagonal array of subunits in intercellular junctions of the mouse heart and liver. J Cell Biol 33 (3):C7–C12 Rörig B, Klausa G et al (1995a) Beta-adrenoreceptor activation reduces dye-coupling between immature rat neocortical neurones. Neuroreport 6(13):1811–1815 Rörig B, Klausa G et al (1995b) Dye coupling between pyramidal neurons in developing rat prefrontal and frontal cortex is reduced by protein kinase A activation and dopamine. J Neurosci 15(11):7386–7400 Rörig B, Sutor B (1996a) Regulation of gap junction coupling in the developing neocortex. Mol Neurobiol 12(3):225–249 Rörig B, Sutor B (1996b) Serotonin regulates gap junction coupling in the developing rat somatosensory cortex. Eur J Neurosci 8(8):1685–1695 Rouach N, Segal M et al (2003) Carbenoxolone blockade of neuronal network activity in culture is not mediated by an action on gap junctions. J Physiol 553(Pt 3):729–745 Schmitz D, Schuchmann S et al (2001) Axo-axonal coupling. a novel mechanism for ultrafast neuronal communication. Neuron 31(5):831–840 Schwartzkroin PA, Prince DA (1977) Penicillin-induced epileptiform activity in the hippocampal in vitro preparation. Ann Neurol 1(5)5:463–469 Schwartzkroin PA, Prince DA (1978) Cellular and field potential properties of epileptogenic hippocampal slices. Brain Res 147(1):117–130 Shu Y, Hasenstaub A et al (2006) Modulation of intracortical synaptic potentials by presynaptic somatic membrane potential. Nature 441(7094):761–765 Sjöström PJ, Rancz EA et al (2008) Dendritic excitability and synaptic plasticity. Physiol Rev 88(2):769–840 Smith KJ, McDonald WI (1999) The pathophysiology of multiple sclerosis: the mechanisms underlying the production of symptoms and the natural history of the disease. Philos Trans R Soc Lond B Biol Sci 354(1390):1649–1673 Sohl G, Maxeiner S et al (2005) Expression and functions of neuronal gap junctions. Nat Rev Neurosci 6(3):191–200 Spray DC, Bennett MV (1985) Physiology and pharmacology of gap junctions. Annu Rev Physiol 47:281–303 Spray DC, 1Rozental R et al (2002) Prospects for rational development of pharmacological gap junction channel blockers. Curr Drug Targets 3(6):455–464 Spray DC, Stern JH et al (1982) Gap junctional conductance: comparison of sensitivities to H and Ca ions. Proc Natl Acad Sci USA 79(2):441–445 Spruston N, Schiller Y et al (1995) Activity-dependent action potential invasion and calcium influx into hippocampal CA1 dendrites. Science 268(5208):297–300 Srinivas M, Spray DC (2003) Closure of gap junction channels by arylaminobenzoates. Mol Pharmacol 63(6):1389–1397 Stasheff SF, Hines M et al (1993) Axon terminal hyperexcitability associated with epileptogenesis in vitro. I. Origin of ectopic spikes. J Neurophysiol 70(3):961–975 Tamas G, Buhl EH et al (2000) Proximally targeted GABAergic synapses and gap junctions synchronize cortical interneurons. Nat Neurosci 3(4):366–371 Taylor CP, Dudek FE (1982) Synchronous neural afterdischarges in rat hippocampal slices without active chemical synapses. Science 218(4574):810–812 Taylor CP, Dudek FE (1984) Excitation of hippocampal pyramidal cells by an electrical field effect. J Neurophysiol 52(1):126–142

102

G. Birke et al.

Tokunaga A, Akert K et al (1980) Cell types and synaptic organization of the medullary electromotor nucleus in a constant frequency weakly electric fish, Sternarchus albifrons. J Comp Neurol 192(3):407–426 Traub RD, Bibbig A (2000) A model of high-frequency ripples in the hippocampus based on synaptic coupling plus axon-axon gap junctions between pyramidal neurons. J Neurosci 20(6):2086–2093 Traub RD, Bibbig A et al (2004) Cellular mechanisms of neuronal population oscillations in the hippocampus in vitro. Annu Rev Neurosci 27:247–278 Traub RD, Bibbig R et al (2001) Synaptic and nonsynaptic contributions to giant IPSPs and ectopic spikes induced by 4-aminopyridine in the hippocampus in vitro. J Neurophysiol 85(3): 1246–1256 Traub RD, Colling SB et al (1995) Cellular mechanisms of 4-aminopyridine-induced synchronized after-discharges in the rat hippocampal slice. J Physiol 489 ( Pt 1):127–140 Traub RD, Draguhn A et al (2002) Axonal gap junctions between principal neurons: a novel source of network oscillations, and perhaps epileptogenesis. Rev Neurosci 13(1):1–30 Traub RD, Dudek FE et al (1985) Simulation of hippocampal afterdischarges synchronized by electrical interactions. Neuroscience 14(4):1033–1038 Traub RD, Jefferys JG et al (1994) A branching dendritic model of a rodent CA3 pyramidal neurone. J Physiol 481 ( Pt 1):79–95 Traub RD, Schmitz D et al (1999) High-frequency population oscillations are predicted to occur in hippocampal pyramidal neuronal networks interconnected by axoaxonal gap junctions. Neuroscience 92(2):407–426 Traub RD, Whittington MA et al (1996) Analysis of gamma rhythms in the rat hippocampus in vitro and in vivo. J Physiol 493 ( Pt 2):471–484 Valiante TA, Perez Velazquez JL et al (1995) Coupling potentials in CA1 neurons during calciumfree-induced field burst activity. J Neurosci 15(10):6946–6956 Vaney DI (1993) The coupling pattern of axon-bearing horizontal cells in the mammalian retina. Proc Biol Sci 252(1334):93–101 Venance L, Rozov A et al (2000) Connexin expression in electrically coupled postnatal rat brain neurons. Proc Natl Acad Sci U S A 97(18):10260–10265 Vogt A, Hormuzdi SG et al (2005) Pannexin1 and Pannexin2 expression in the developing and mature rat brain. Brain Res Mol Brain Res 141(1):113–120 Voronin LL (2000) Intrasynaptic ephaptic feedback in central synapses. Neurosci Behav Physiol 30(5):575–585 Willecke K, Eiberger J et al (2002) Structural and functional diversity of connexin genes in the mouse and human genome. Biol Chem 383(5):725–737 Wilson MA, McNaughton BL (1994) Reactivation of hippocampal ensemble memories during sleep. Science 265(5172):676–679 Yang XD, Korn H et al (1990) Long-term potentiation of electrotonic coupling at mixed synapses. Nature 348(6301):542–545 Yasargil GM, Sandri C (1990) Topography and ultrastructure of commissural interneurons that may establish reciprocal inhibitory connections of the Mauthner axons in the spinal cord of the tench, Tinca tinca L. J Neurocytol 19(1):111–126 Ylinen A, Bragin A et al (1995) Sharp wave-associated high-frequency oscillation (200 Hz) in the intact hippocampus: network and intracellular mechanisms. J Neurosci 15(1 Pt 1):30–46 Zappala A, Li Volti G et al (2007) Expression of pannexin2 protein in healthy and ischemized brain of adult rats. Neuroscience 148(3):653–667 Zoidl G, Petrasch-Parwez E et al (2007) Localization of the pannexin1 protein at postsynaptic sites in the cerebral cortex and hippocampus. Neuroscience 146(1):9–16

Chapter 6

To Myelinate or Not to Myelinate? Quan Wen and Dmitri B. Chklovskii

6.1 Introduction In vertebrate brains, axons are often myelinated, i.e., wrapped in a thin myelin sheath. Myelinated and non-myelinated axons co-exist in many structures, such as the corpus callosum. This observation leads to a question: Why does nature myelinate some axons but not others? Understanding this question may shed light on the principles of brain design. About half a century ago, Rushton (1951) proposed an answer, which can be stated as follows (see Waxman and Bennett, 1972): The crucial difference between a myelinated and a non-myelinated axon is in how the conduction velocity scales with axon diameter, as shown in Fig. 6.1 (redrawn from Waxman and Bennett, 1972). In a myelinated axon, conduction velocity (CV), S, increases linearly with the axon diameter (Bullock and Horridge, 1965; Rushton, 1951) as: S = BD,

(1)

where B is the proportionality constant with dimension m/s· μm−1 , and D is the axon diameter including the myelin sheath. In a non-myelinated axon, CV, s increases with the square root of axon diameter, d (Bullock and Horridge, 1965; Hodgkin, 1954; Hursh, 1939; Rushton, 1951): s = bd1/2 ,

(2)

where b is also a proportionality constant with the dimension m/s· μm−1/2 . Equations (1) and (2) are valid only for relatively large axon diameters (solid lines in Fig. 6.1) because axon diameters much smaller than 0.1 μm have rarely been observed experimentally. According to Equations (1) and (2), there is a diameter, d0 , at which the CVs, s0 , in myelinated and non-myelinated axons are the same (Fig. 6.1): D.B. Chklovskii (B) Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, USA e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_6, 

103

104 6

5 myelinated

4

CV (ms–1)

Fig. 6.1 Conduction velocity (CV) as a function of the axon diameter for myelinated and non-myelinated axons, modified from Waxman and Bennett (1972). Axon diameters much smaller than 0.1 μm are not observed experimentally and plots in this regime are represented by dashed lines. At d0 , myelinated and non-myelinated axons have the same CV, s0

Q. Wen and D.B. Chklovskii

3 non-myelinated

2

1 (d0, s0) 0

0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 diameter (μm)

 b2 s0 = Bd0 = b d0 = . B

1

(3)

Rushton (1951) proposed that, for a given CV, nature picks an axon with a smaller diameter (Fig. 6.1). If the required CV is less than s0 , a non-myelinated axon is thinner and preferred. Otherwise, a myelinated axon is thinner and preferred. It follows from Rushton’s argument that there is a critical axon diameter (d0 ≈ 0.2 μm according to Waxman and Bennett, 1972). All axons that are thinner than d0 should be non-myelinated and those thicker than d0 should be myelinated. This prediction does not accord with experimental data (Franson and Hildebrand, 1975; Remahl and Hildebrand, 1982; Wang et al., 2008; Waxman and Swadlow, 1976). Although myelinated axons are typically thicker than the non-myelinated ones, the diameter distributions of myelinated and non-myelinated axons overlap. In this chapter, we attempt to resolve this old issue by assuming explicit expressions for costs of myelinated and non-myelinated axons. We point out that minimization of axonal cost does not necessarily lead to the existence of a critical axon diameter. We consider two possible optimization scenarios. In the first scenario, we assume that the CV for a given axon is set a priori and axonal cost is proportional to its volume. Depending on the relative cost per volume for myelinated and non-myelinated axons, the distributions of axon diameters can overlap or be separated by a gap. In the second scenario, we assume that for a given axon length, the conduction delay, i.e., the time taken for an action potential to travel between two end points is not fixed. Rather, an increase in the conduction delay invokes a finite cost. Then we search for the axon design, which optimizes the cost function including volume and conduction delay. Interestingly, the latter scenario predicts that there must be a gap in the velocity spectrum.

6

To Myelinate or Not to Myelinate?

105

6.2 Results 6.2.1 Axonal Cost with Fixed Conduction Delay In this section, we formulate a two-choice problem: given the conduction delay in an axon, should the axon be myelinated or non-myelinated? We will answer this question for a non-branching axon with fixed length by calculating the axonal cost as a function of CV. Assuming that the cost is proportional to axon volume (Cherniak, 1992; Chklovskii et al., 2002; Chklovskii and Stepanyants, 2003; Mitchison, 1991), costs of a myelinated and a non-myelinated axon per length are given by π εm = K D2 , 4 π 2 εn = k d , 4

(4) (5)

where K and k are costs per volume for myelinated and non-myelinated axons, respectively. Substituting Equations (1) and (2), we express the axonal costs in terms of CV: πS2 , 4B2 π s4 εn = k 4 . 4b

εm = K

(6) (7)

By setting εm = εn , we determine the critical CV, sc , where the costs of a myelinated and a non-myelinated axon are equal (see Fig. 6.2a):  sc =

K s0 . k

(8)

When the required CV is smaller than sc , a non-myelinated axon should be used; otherwise a myelinated axon should be used (Fig. 6.3a). This result implies that the spectrum of the axon diameters depends on the ratio of the cost per volume for a myelinated and a non-myelinated axon: • K = k. When costs per volume of a myelinated and a non-myelinated axon are equal, we recover Rushton’s result: axons with a diameter greater than the critical diameter d0 are myelinated; below the critical diameter they are non-myelinated. • K > k. When cost per volume of a myelinated axon is greater than that of a non-myelinated axon, the critical CV sc is above the intersection point s0 (Fig. 6.2b, c). In this case we expect an overlap between the diameter distributions of myelinated and non-myelinated axons. • K < k. When cost per volume of a myelinated axon is less than that of a non-myelinated axon, the critical CV sc is below the intersection point s0

106

Q. Wen and D.B. Chklovskii A 1.4 1.2

Cost

1 0.8

myelinated

0.6 0.4

non-myelinated 0.2

sc 0

0

0.2

0.4

0.6

0.8

1

1.2

CV (ms–1) D

B 5 4.5

myelinated

myelinated

2.5 2 1.5

s0

(d0, s0) 0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

non-myelinated myelinated

OVERLAP

diameter (μm)

1

E

number of axons

number of axons

non-myelinated

non-myelinated

0.5

C

(d0, s0)

(d, sc)

(D, sc)

sc

1

0

CV spectrum

3

CV (ms–1)

3.5

CV spectrum

CV (ms–1)

4

non-myelinated

GAP

myelinated

diameter (μm)

Fig. 6.2 Spectrum of axon diameters minimizing axonal cost subject to a hard conduction delay constraint (a), If CV is below the critical CV, sc , a non-myelinated axon is less costly and preferred. Otherwise, a myelinated axon is preferred (b, c). If the myelinated axon is costlier (on a per volume basis) than the non-myelinated axon, the critical CV sc > s0 and there is an overlap in the diameter distributions of myelinated and non-myelinated axons. When the ratio K/k = 16, the diameter distributions of myelinated and non-myelinated axons overlap in the regime 0.2–0.7 μm. On the right of (b), a continuous CV spectrum is assumed (d, e). If the non-myelinated axon is costlier (on a per volume basis) than the myelinated axon, sc < s0 and there is a gap in the diameter spectrum. On the right of (d), a continuous CV spectrum is assumed

6

To Myelinate or Not to Myelinate?

107

(Fig. 6.2d, e). In this case, we expect a gap in the diameter spectrum that separates the non-myelinated and myelinated axon distributions.

6.2.2 Axon with a Finite Conduction Delay Cost In the previous scenario, the conduction delay is specified a priori for an axon. In an alternative scenario, we may assume that the conduction delay only results in a finite penalty and the total cost linearly combines the cost of volume, V, and the cost of conduction delay, T, weighted by κ and α E = κV + αT,

(9)

where the unit conduction delay cost α is now set a priori for a given axon length. Then, we ask a slightly different question from that in the first scenario. If conduction delay cost is proportional to α, should the axon be myelinated or non-myelinated? To answer this question, we first determine the minimal costs for a myelinated axon and a non-myelinated axon, respectively. We then compare the two costs in order to choose the less costly axon. By rewriting the cost as a function of diameter (see Section 6.5), we show that the volume term in the cost increases with diameter while the conduction delay term decreases with diameter. By setting the derivative of the cost function with respect to diameter to zero, we find that the optimal CV in a myelinated axon is given by  S = B2/3

2α πK

1/3 .

(10)

The corresponding cost per length is given by 3α 3 εm = = 2S 2B2/3



πK 2

1/3

α 2 /3 .

(11)

In a non-myelinated axon, the optimal CV and the minimal cost per length are given by  α 1/5 , πk 5α 5 εn = = 4/5 (πk)1/5 α 4/5 . 4s 4b s = b4/5

(12) (13)

Figure 6.3d plots myelinated and non-myelinated axonal costs as a function of the unit conduction delay cost α. They are equal at the critical value α 0 . If α is less than α 0 , a non-myelinated axon is preferred. Otherwise, a myelinated axon is preferred.

108

Q. Wen and D.B. Chklovskii

A 5

C

2 (D, S)

1.5 1 0.5

(d0, v0) 0

CV (ms−1)

myelinated

4 3.5

GAP

CV (ms–1)

3 2.5

CV Spectrum

4 3.5

0

5 4.5

4.5

non-myelinated

(d, s)

3 2.5 2

myelinated

1.5 1 0.5 0

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

B

D

non-myelinated 0

0.5

1

α 1.5

2

2.5

3

4

number of axons

3.5 3

Cost

2.5

myelinated

2 C0 1.5 1 non-myelinated

0.5 0

diameter (μm)

α0 0

0.5

1

1.5

2

2.5

3

α

Fig. 6.3 Spectrum of axon diameters minimizing axonal cost for finite conduction delay cost. In this figure, we assume that K  k. a, b If K/k = 17, the diameter distributions of myelinated and non-myelinated axons overlap in the regime 0.2–0.7 μm. On the right of (a), the CV spectrum has a gap. c CV as a function of the unit conduction delay cost α. d Axonal costs as a function of the unit conduction delay cost α. At the critical α 0 , the myelinated and non-myelinated axons have the same cost, but the CVs are different. This leads to a gap in the CV spectrum when the cost of a non-myelinated axon becomes greater than that of a myelinated axon

However, unlike in the previous scenario, minimizing the axonal costs with a finite conduction delay penalty predicts a gap in the CV spectrum (Fig. 6.3). This is because the CV of a myelinated axon is different from that of a non-myelinated axon when the costs of the two types are the same (Fig. 6.3c). To see this, by setting Equations (11) and (13) equal to each other, we find that when the costs of a myelinated and non-myelinated axon are the same S=

6 s. 5

(14)

The diameter spectrum still depends on the ratio K/k. Similar to the previous scenario, we find that

6

To Myelinate or Not to Myelinate?

109

• K = 5/3 k. Axons with a diameter greater than the critical diameter d0 are myelinated; below the critical diameter, they are non-myelinated. We note that although Rushton’s (1951) result is recovered in this case, the gap in the CV spectrum always exists. • K > 5/3 k. The diameter distributions of non-myelinated and myelinated axons overlap (Fig. 6.3a, b). • K < 5/3 k. The diameter spectrum has a gap that separates non-myelinated and myelinated axon distributions (Fig. 6.4a, b).

6.2.3 Comparison with Experiments Next, we compare our theory with the available experimental data from the corpus callosum (Aboitiz et al., 1992; Wang et al., 2008; Swadlow and Waxman, 1976; Waxman and Swadlow, 1976). This region of the brain offers several advantages.

C

myelinated

(d0, s0)

myelinated

CV (sm–1)

GAP

CV (sm–1)

A

non-myelinated non-myelinated

D

Cost

number of axons

B

GAP

myelinated

C0

non-myelinated myelinated diameter (μm)

non-myelinated α 0 α

Fig. 6.4 Spectrum of axon diameters minimizing axonal cost for finite conduction delay cost. In this figure, we assume that K  k

110

Q. Wen and D.B. Chklovskii

First, myelinated and non-myelinated axons co-exist in it. Second, corpus callosum axons have different diameters but similar lengths (between 100 and 130 mm in a human brain (Aboitiz et al., 1992). As a result, the relationship between conduction delay and conduction velocity is uniform. Third, myelinated axons, in general, have the disadvantage that they cannot make en passant synapses. This disadvantage cannot play a role in the corpus callosum because there are no synapses there. Unfortunately, we are unaware of simultaneous measurements of conduction velocity and diameter for individual axons, making our comparison somewhat indirect. Although measurements have been done in studies of the peripheral nervous system dating back a long time (Boyd and Kalu, 1979; Gasser, 1950; Hursh, 1939), precise measurements of myelinated and non-myelinated axons with very small diameters (0.1–1 μm) are technically challenging. Waxman and Swadlow (1976) found that, in the visual corpus callosum, the diameter of non-myelinated axons varies between 0.08 and 0.6 μm, while the diameter of myelinated axons changes between 0.3 and 1.58 μm. They observed an overlap between the diameter distributions of non-myelinated axons and myelinated axons. The distribution of conduction velocity in the visual callosal axons (albeit different sample from the diameter measurements) ranges from 0.3 to 12.9 m/s. If we assume a square root relationship between conduction velocity and the diameter of a non-myelinated axon, the smallest conduction velocity 0.3 m/s corresponds to the smallest diameter 0.08 μm. Then, the scaling factor b in Equation (2) can be estimated as 1.06 m/s· μm−1/2 . This result is consistent with recent measurements on olfactory receptor neurons (Griff et al., 2000). In myelinated axons (Boyd and Kalu, 1979; Waxman and Bennett, 1972), the scaling factor B (see Equation 1) is between 4.5 and 6 m/s· μm−1 . These scaling factors were used in Figs. 6.2, 6.3, and 6.4. These figures show that a gap in the diameter spectrum can exist only below 0.07 μm, a value too small for non-myelinated axons to exist. Indeed, experimentally measured distributions of axon diameters exhibit an overlap between myelinated and non-myelinated axons (Wang et al., 2008; Waxman and Swadlow, 1976), suggesting that K>k. Using estimates of scaling factors, k = 1.06 m/s· μm−1/2 and K = 5 m/s· μm−1 , we can also estimate the ratio K/k. If the overlap is approximately within the range of 0.2–0.7 μm, K/k = 16 according to the first scenario and K/k = 17 according to the second scenario. The two estimates are reasonably close to each other. We must emphasize the approximate nature of these estimates because k and K are evaluated with a large degree of uncertainty. For example, using values inferred by Rushton (1951) for the largest non-myelinated axon in the peripheral nerve, k = 2.09 m/s· μm−1/2 and K = 5 m/s· μm−1 , both scenarios yield K/k = 4. Assuming that K can range between 4.5 and 6 m/s· μm−1 , our estimate of K/k is between 3 and 25. Prediction of a gap in the CV spectrum could not be verified using available experimental data. Measurements of the CV for individual axons with thin diameter (0.1–1 μm) can be very difficult. Since the first scenario does not predict a gap, testing this prediction would help us distinguish between the two scenarios.

6

To Myelinate or Not to Myelinate?

111

6.3 Discussion The square root relation between non-myelinated axon diameter and conduction velocity has proven to be followed roughly in real nerves, but the relationship does not exactly hold in different animals and different nerves. Generally, we could assume a power law between diameter and CV by rewriting Equation (2) as s = bdθ ,

(15)

where θ < 1. Then, we check how Equation (15) affects our previous arguments. In the first scenario, the critical CV sc in Equation (8) should be modified as sc =

  θ   1 K 2(1−θ) b 1−θ B. k B

(16)

But the dependence of diameter spectrum on the ratio of unit volume cost of myelinated and non-myelinated axons does not change. In the second scenario, following the procedure discussed in the main text and in Section 6.5, we found that there is also a gap in the CV spectrum. However, Equation (14) changes to S=

3s . 2+θ

(17)

In view of diameter spectrum, we have • K = (2+θ)/(3θ)k. Axons with a diameter greater than the critical diameter d0 are myelinated and below the critical diameter are non-myelinated. We note that although Rushton’s result (Rushton, 1951) is recovered in this case, the gap in the CV spectrum always exists. • K > (2+θ)/(3θ)k. The diameter distributions of non-myelinated and myelinated axons overlap. • K < (2+θ)/(3θ)k. The diameter distributions of non-myelinated and myelinated axons are separated by a gap.

6.4 Conclusions Our theory and experimental data predict that unit volume cost is different for myelinated and non-myelinated axons. This result has important implications on brain design. The reason for the difference in unit volume cost remains unclear. One could argue that myelinated fibers are costlier than non-myelinated of the same diameter because of the involvement of glia. Alternatively, one can argue the opposite because more ion channels are required to sustain membrane potential. Together with other optimization approaches, this theory will help uncovering basic principles of brain design, a crucial step in understanding brain function.

112

Q. Wen and D.B. Chklovskii

6.5 Methods According to the second scenario, given the length of an axon L, the unit cost of conduction delay α, the costs for a myelinated axon and a non-myelinated axon are given by π L Em = K D2 L + α , 4 S π 2 L En = k d L + α . 4 s

(18) (19)

By substituting Equations (1) and (2) into Equations (18) and (19), the costs per length of a myelinated and a non-myelinated axon can be formulated, respectively, as: π 1 , εm = K D2 + α 4 BD π 1 εn = k d2 + α 1/2 . 4 bd

(20) (21)

∂εn ∂εm = 0, = 0, we find the optimal diameter of the two types of By setting ∂D ∂d axons to be 

 2α 1/3 , πKB  α 2 / 5 d= . πkb

D=

(22) (23)

Substituting Equations (22) and (23) into Equations (1) and (2), we obtain the expressions for CVs (Equations (10) and (12)). Then, substituting Equations (22) and (23) into Equations (20) and (21) yields more compact expressions for the dendritic costs, as shown in Equations (11) and (13). When setting the dendritic costs for myelinated and non-myelinated axons equal to each other, we find the critical α satisfies α0 /

2 15

=

6 5



πK 2B2

1/3 

b4 πk

1/5 .

(24)

The axon diameters (i.e., Equations (22) and (23)) become the same when  α 2/15 =

2 πBK

2/3

(πbk)4/5 .

(25)

If α 0 > α or K/k > 5/3, there is an overlap in the diameter distributions of myelinated and non-myelinated axons. Otherwise there is a gap in the axon diameter spectrum.

6

To Myelinate or Not to Myelinate?

113

References Aboitiz F, Scheibel AB, Fisher RS, Zaidel E (1992) Fiber composition of the human corpus callosum. Brain Res 598:143–153 Boyd IA, Kalu KU (1979) Scaling factor relating conduction velocity and diameter for myelinated afferent nerve fibres in the cat hind limb. J Physiol 289:277–297 Bullock TH, Horridge GA (1965) Structure and function in the nervous systems of invertebrates. W. H. Freeman, San Francisco Cherniak C (1992) Local optimization of neuron arbors. Biol Cybern 66:503–510 Chklovskii DB, Schikorski T, Stevens CF (2002) Wiring optimization in cortical circuits. Neuron 34:341–347 Chklovskii DB, Stepanyants A (2003) Power-law for axon diameters at branch point. BMC Neurosci 4:18 Franson P, Hildebrand C (1975) Postnatal growth of nerve fibres in the pyramidal tract of the rabbit. Neurobiology 5:8–22 Gasser H (1950) Unmedullated fibers originating in dorsal root ganglia. J Gen Physiol 33:277–297 Griff ER, Greer CA, Margolis F, Ennis M, Shipley MT (2000) Ultrastructural characteristics and conduction velocity of olfactory receptor neuron axons in the olfactory marker protein-null mouse. Brain Res 866:227–236 Hodgkin AL (1954) A note on conduction velocity. J Physiol 125:221–224 Hursh J (1939) Conduction velocity and diameter of nerve fibers. Am J Physiol 127:131–139 Mitchison G (1991) Neuronal branching patterns and the economy of cortical wiring. Proc R Soc Lond B Biol Sci 245:151–158 Remahl S, Hildebrand C (1982) Changing relation between onset of myelination and axon diameter range in developing feline white matter. J Neurl Sci 54:33–45 Rushton WA (1951) A theory of the effects of fibre size in medullated nerve. J Physiol 115:101–122 Swadlow HA, Waxman SG (1976) Variations in conduction velocity and excitability following single and multiple impulses of visual callosal axons in the rabbit. Exp Neurol 53:128–150 Wang SSH, et al. Functional trade-offs in white matter axonal scaling. J. Neurosci 28:4047–4056 Waxman SG, Bennett MV (1972) Relative conduction velocities of small myelinated and nonmyelinated fibres in the central nervous system. Nat New Biol 238:217–219 Waxman SG, Swadlow HA. Ultrastructure of visual callosal axons in the rabbit. Exp Neurol 53:115–127

Part III

Axons and Neuronal Circuits

Chapter 7

An Axonal Perspective on Cortical Circuits Tom Binzegger, Rodney J. Douglas, and Kevan A.C. Martin

7.1 Introduction How neurons connect in the complex circuits of neocortex is one of the fundamental problems in neuroscience. Anatomical techniques are used to trace the connection pathways involving axonal and dendritic trees, and electrophysiological recordings are applied to probe the connections between two or several neurons in the circuit. More recently, optical methods such as uncaging of glutamate and calcium imaging allow the functional circuit architecture to be explored on a larger scale, but the basic difficulty remains of establishing which neurons connect to which in a network consisting of several thousand of neurons. Neither is there a consentaneous theory about the basic function of a neocortical circuit, nor is there any unanimity about the principles by which the connections in the circuit are formed. The debate about the significance of the relative role of feedforward and recurrent processing for cortical function is still undiminished (Douglas and Martin, 2007a), as is the debate about the degree of randomness or specificity involved in wiring up the neurons (Ohki and Reid, 2007). The most influential proposals about the overall circuit structure have therefore not come from direct experimental observations, but from building the circuit using the time-honoured assumption introduced by Ramón y Cajal that axons connect to dendrites whenever the two trees arborize (or overlap) in the same layer. We refer to this as the ‘Cajal circuit’ (Douglas and Martin, 2007b). What becomes clear from these studies is that our understanding of the neocortical circuits is intimately related to the precision with which we understand the organization of the dendritic and axonal branching patterns. This is particularly so for the axonal tree, which, with its complex and protruding branching pattern, is difficult to reveal and to characterize. The earliest attempts to construct an overall circuit diagram (Fig. 7.1a) were hampered by the incomplete staining of the axon of Golgi-impregnated neurons. The stained axons appeared as sparsely labelled trees and a connection had often to be T. Binzegger (B) University of Newcastle, Newcastle upon Tyne, UK e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_7, 

117

118

T. Binzegger et al.

Fig. 7.1 Local circuits of neocortex. (a) Qualitative circuit proposed by Lorente de Nó (1949), based on partially impregnated axons using the Golgi method. (b) Qualitative circuit proposed by Gilbert and Wiesel (1983) based on completely labeled axons using HRP. (c) Quantitative circuit proposed by Binzegger et al. (2004) based on digitized 3-D reconstructions of completely HRP labelled axons. Anatomical weights are indicated by grayscale and line thickness (grey: low weight, thin black: intermediate weight, bold black: large weight). Grey-filled discs indicate excitatory neurons, empty discs inhibitory neurons. Number in each disk indicates cortical layer

inferred from a few axonal arbors overlapping with the dendritic trees (Lorente de Nó, 1949; Jones, 1975; Lund and Boothe, 1975; Szentágothai, 1975; Lund et al., 1979). A more confident assessment of overlap was made possible by injecting the label horseradish peroxidase (HRP) directly into the cell body (Gilbert and Wiesel, 1979, 1983; Martin and Whitteridge, 1984). Using this method, the axon could be labelled in glorious completeness, so that what had often appeared as an isolated axonal arbor in Golgi stain developed now into richly branching structures. Because the axons proved to be very layer specific in their arborization pattern, a salient diagram of pathways between cell types emerged from the overlap between axon and dendrites (Gilbert and Wiesel, 1981; Gilbert, 1983). The resulting circuit (Fig. 7.1b) was highly influential, particularly because it was consistent with that conjectured

7

An Axonal Perspective on Cortical Circuits

119

from earlier physiological studies of cortical receptive fields (Hubel and Wiesel, 1962, 1965). The main problem with those early approaches of circuit modeling is that a large amount of subjectivity is involved in judging how much overlap is necessary for establishing a connection. In the case of the circuit diagram proposed by Lorente de Nó (1949) (Fig. 7.1a), connections were drawn between neurons for modest overlap, leading to a network with many recurrent connections between cortical layers. In contrast, the pathways in the circuit diagram of Gilbert (1983) (Fig. 7.1b) represent only the most complete overlap between axon and dendrite. Connections arising from an overlap of a given axon with the apical dendrites were also ignored. As a consequence, the diagram is simpler, essentially describing a big loop spanning cortical layers 2–6. In order to avoid potential ambiguities, quantitative approaches have been used, both in order to have objective criteria when a connection is formed, and to give some measure of the strength of overlap between dendrites and axons. Early attempts to quantify these ‘anatomical weights’ in an overall circuit diagram had to rely on very crude approximation of neuronal morphology (Braitenberg and Lauria, 1960; Krone et al., 1986; Thomas et al., 1991). Axonal and dendritic trees were modelled as simple geometrical shapes (discs, rectangles, etc.) whose dimensions were estimated from 2-D drawings of Golgi-impregnated neurons. Fully labelled neurons reconstructed in 3-D offer unprecedented detail in the quantitative characterization of axonal trees. What the analysis of these new data shows is that axons have an intricate organization with salient structural features at many levels of detail ranging from the overall gestalt of the tree down to the level of how the branches are locally organized, wriggle through the neuropil, and form boutons in order to make synaptic contacts with targets. Here we review recent results from quantitative studies of axonal morphology. Using a simple connectivity principle based on the overlap rule (‘Peters Rule’), we show that axonal complexity produces a circuit structure of exquisite complexity, and we provide a functional interpretation of this circuit.

7.2 Local Circuits The primary visual cortex of the cat (area 17) has a surface area of about 2 cm2 (Anderson et al., 1988) and a depth of about 2 mm. A typical cortical axon is confined to a vertical cylinder from pia mater to white matter whose volume occupies roughly 1% of the total volume of area 17. Within this cylinder the axon forms about 5,000 synapses, and although this seems an impressively large number, a cylinder of this size already contains about 100 times more neurons, each of which might be a potential target of the axon. Thus, we are confronted with the problem of establishing for every axon which targets are selected out of a large population of potential neurons in order to form the cortical circuit. Determining the targets for each single

120

T. Binzegger et al.

cortical axon seems an impossible task. However, the regularity of the cortex makes this problem more approachable. One of the main legacies from the Golgi era is the recognition that the morphologies of neocortical neurons are highly stereotypical. Each cortical layer consists of neurons forming a few anatomical types with distinctly different morphological features (Cajal, 1908; O’Leary, 1941; Lorente de Nó, 1949; Szentágothai, 1973; Szentágothai and Arbib, 1974; Lund et al., 1979). The ability to label axons in full using modern techniques resulted in refinements in the anatomical classification scheme, but in essence classification criteria remained the same (Martin and Whitteridge, 1984). Furthermore, it is generally assumed to be the case that while some cortical afferents may be distributed in a patchy fashion, the somata of a particular anatomical cell type are uniformly distributed within their layer of occurrence. In area 17, the Meynert cells of monkey area 17, which are readily distinguishable in layer 6 with conventional histological methods (Winfield et al., 1981), seem uniformly distributed. In general, however, direct inspection of the distribution or frequency of neurons belonging to the different anatomical cell types are, with a few exceptions, not possible due to the lack of appropriate markers which label exclusively a cell type. Thus, the regularity in organization of area 17, and in cortical areas in general (Douglas and Martin, 2004), brings a great simplification. For any arbitrary position in a cortical area, the number of neurons and the composition of cell types in a vertical cylinder remains (statistically) the same. Each axon is confronted with the same mix of potential targets and will form synapses with a small subset of them. Thus, instead of being faced with an enormous, irregular wiring diagram between the millions of neurons in the area, the uniformity of the neuropil suggests that connectivity repeats itself statistically along the cortical surface (Szentágothai, 1975; Rockel et al., 1980; Hubel and Wiesel, 1972) and can be studied locally (a ‘canonical’ circuit). The assumption, of course, is that wiring rules are generic and do not change fundamentally across the cortical sheet.

7.3 Capturing Axon Morphology Although the characteristic laminar gestalt of neurons observed in Golgiimpregnated slices has inspired many important ideas of how circuits are organized and function (Lorente de Nó, 1949; Jones, 1975; Szentágothai, 1975; Lund and Boothe, 1975; Lund et al., 1979), studies that explore the deeper intricacies of neuron-branching patterns, and in particular that of axonal trees, are still surprisingly rare, given how fundamental this is to neuronal circuits in general. Figure 7.2 shows examples of the most common cell types in cat area 17. The neurons are part of a large database of 39 neurons, each of which has been filled intracellularly with HRP during in vivo experiments. The neurons were digitized in 3-D using a computerized light microscope, that is, the spatial location of the boutons, axonal,

7

An Axonal Perspective on Cortical Circuits

121

Fig. 7.2 Example neurons of the main cell types in cat area 17. The neurons were injected with horseradish peroxidase during in vivo experiments and subsequently reconstructed in 3-D using a computerized light microscope. Dark blue and red indicate dendrites, light blue and yellow indicate axons. Cell types are indicated at the top and bottom. Abbreviations: ‘b2/3,’ ‘b4,’ ‘b5’ basket cells in layers 2/3, 4, and 5; ‘db2/3’ double bouquet cell in layer 2/3; ‘p2/3,’ ‘p4,’ ‘p5,’ ‘p6’ pyramidal cells in layers 2/3, 4, 5, and 6. ‘ss4’ spiny stellate cells in layer 4. Spiny stellate cells and pyramidal cells were further distinguished by the preferred layer of the axonal innervation (‘ss4(L4)’ (not shown), ‘ss4(L2/3),’ ‘p5(L2/3),’ ‘p5(L5/6),’ ‘p6(L4),’ and ‘p6(L5/6)’). ‘X/Y’ thalamic afferents of type X or Y. Horizontal lines indicate the approximate cortical layers L1, L2/3 (layer 2 and 3 were merged), L4, L5, and L6. Also indicated is the white matter (‘wm’). Scale bar 300 μm

and dendritic segments are represented as a list of 3-D coordinates so that further sophisticated analyses were possible. What becomes clear from these examples is that axons and dendrites, but axons in particular, are complicated spatial structures. Branches curl through the neuropil in many different directions, they are topologically arranged in a tree and spatially in characteristic vertical and horizontal patterns that help distinguish and define the cell types. Summing the length of all branches together, the total length of an axon is of the order of 40 mm, for dendrites it is 4 mm. The number of 3-D coordinate points needed to describe the composite pattern formed by the axonal and dendritic trajectories to a reasonable accuracy requires of the order of 10,000 points. Yet, despite this apparent variation and complexity, a systematic analysis of the data shows that global principles do exist to explain spatial and topological aspects of the branching patterns. This raises the attractive possibility that only a minimal set of constructions rules are needed to form the cortical circuit. The most salient feature of the overall gestalt of the axonal and dendritic trees is the laminar pattern when viewed in coronal view. Studying the dendrites of Golgiimpregnated neurons, Lorente de Nó (1949) marveled at the laminar precision with which dendrites arborized in the cortical layers. A systematic analysis based on the reconstructed neurons shows that each neuron forms most of its dendritic tree

122

T. Binzegger et al.

Fig. 7.3 Laminar innervation patterns of cortical cell types. Shown is for each cell type the fraction of synapses on boutons (left column) and the fraction of dendrites (right column) a typical neuron forms in each layer

(>75% of the total dendritic length) locally in the layer of soma, with some forming an additional richly branching apical tuft in a further layer (Fig. 7.3). HRP-labelled neurons show a similar laminar profile for the axon. Taken across all the main neuronal types, more than 40% of the boutons are formed in one layer, and more than 72% are formed in just two layers (Fig. 7.3). The consequence of the laminar specificity of axonal and dendritic trees for network organization is that only particular subtypes of neurons have dendrites, which can overlap with an axon, which imposes a non-trivial structure that is far from random or all-to-all connectivity. Even with the least specific of all wiring rules, the diagram of the possible pathways between cell types is already rather intricate (Fig. 7.1c). Additional structure comes from the axonal arborization pattern within a layer. When a local population of neurons in the superficial layer are bulk injected with a label, a dense region of axon is labelled locally around the injection site, and additional labelled axon appears in more distal, isolated patches, giving the whole pattern the appearance of a ‘daisy’ (Douglas and Martin, 2004) with the distal patches forming the petals. Although patch formation in the superficial layer of many cortical areas has been recognized as one of the most salient features of cortical organization, quantitative studies that explore and characterize their detailed organization are still rare and have mainly focused on the relationship of patches with the underlying functional maps in the visual cortex (Malach et al., 1993; Kisvárday et al., 1996; Bosking et al., 1997). The main conclusion from these studies is that patches arise from some need to connect neurons of similar receptive field properties, which explains the finding that patch organization reflects the underlying layout of the functional maps. However, the same studies invariably show that this correlation is not precise, and map layout characterizes patch organization only incompletely. For example, in the visual cortex the local patch freely innervates all neighbouring orientation domains, and a significant proportion of axon in the distal patches is not located in the iso-orientation domains. Whenever patches have been analyzed quantitatively interesting principles of organization have emerged. The typical size

7

An Axonal Perspective on Cortical Circuits

123

of the patches varies in different cortical areas, as do the basal dendritic field dendritic of superficial layer pyramidal cells, but their corresponding sizes always match (Rockland et al., 1982; Luhmann et al., 1986; Lund et al., 2003). Another scaling relationship holds for the separation distance (center-to-center) between the patches in the different areas, which is always about twice the typical diameter of the patches (Fig. 7.4). The patchy axonal distribution of superficial layer neurons is most clearly observed for the individual axons of layer 2/3 pyramidal cells (Gilbert and Wiesel, 1983; Martin and Whitteridge, 1984) (Fig. 7.5). When a clustering algorithm is used to identify objectively the bouton patches of cortical axons in cat primary visual cortex (Binzegger et al., 2007), we found that clustered bouton clouds are common to

Fig. 7.4 Clustered bouton clouds of a layer 2/3 pyramidal cell. (a) Coronal view of axonal (black) and dendritic (red) arborization patterns. Cortical layers are indicated by curved lines. (b) Bouton cloud showing the linear regions (gray dots). These boutons were excluded from cluster analysis. (c) Bouton cloud showing the clusters (color coded) which were identified by the cluster algorithm. (d) Top view of bouton cloud with identified clusters. All figures are drawn to the same scale. Scale bar 0.5 mm

124

T. Binzegger et al.

Fig. 7.5 Scaling relationship between cluster size and cluster separation for daisies in different cortical areas and species (stars), and for individual inhibitory neurons (open circles) and excitatory neurons (closed circles) of cat area 17. For the individual neurons, the distance of the proximal cluster to the next cluster was about one and a half diameters of the proximal cluster (straight line)

almost all cortical axons. Interestingly, the diameters of the identified clusters range over an order of magnitude, but nevertheless we found that single axons showed a similar scaling relationship between patch size and patch separation as was found at the population level. Thus, cluster formation is a fundamental organization principle which applies to most neurons, cell types, and cortical areas in most mammals. In this universal setting, the most basic structural role of a neuron’s patchy axonal distribution is to innervate focally discrete sites of a cortical layer that are separated by a characteristic distance of roughly two patch diameters. In a network where neurons connect indiscriminately (overlap rule), spatial separation of local populations might be important to increase the functional complexity of the network. Cortical layers might play a similar role on a larger scale. While the need for patch formation might be functionally related, the detailed organization might partly be a consequence of intrinsic growth mechanisms. A generic observation was that the number of clusters formed by an axon (1–7) is closely related to the diameter of the clusters (90–950 μm) and the number of boutons they contain (70–8,300), such that with increasing bouton number the clusters became more equal. A simple growth model can account for the same relationship (Binzegger et al., 2007). The model works by starting with a fixed reservoir of boutons from which new patches are formed by allocating a constant proportion of boutons (20%) from the reservoir. If only two patches are formed (the remaining boutons in the reservoir and the newly formed patch), the number of boutons in the patches is unequal, and since bouton density per patch was found to be constant, there is also a significant

7

An Axonal Perspective on Cortical Circuits

125

difference in patch diameter. In contrast, if seven patches are formed, the initial reservoir is depleted and all patches have roughly the same number of boutons and diameter. Axonal trees have a highly heterogenous spatial distribution of branches, which is not easily cast into a simple mathematical characterization. The cluster analysis we performed is, however, a significant step into this direction and allows to approximate the gross vertical and horizontal distribution of cortical bouton clouds by a mixture of spherical normal distributions φμ, , where the mean μ of the normal distribution codes for the spatial location of the cluster center and the covariance matrix for the spatial extent of the cluster. While most boutons are involved in forming clusters, most axons form an additional sparse web of long isolated arbors which produce linear strings of boutons. These arbors often connect clusters, but they may also simply radiate outward. This diffuse component typically contains 15% of the boutons in the cloud, and its significance is unclear. If L stands for the distribution of the boutons in the diffuse component, we can represent the bouton cloud by the formula B=

n 

bi φμi , i + L,

(1)

i=1

where bi is the number of boutons in each cluster i and n is the number of clusters. What is not captured by this model is the heterogeneous distribution of the boutons within each cluster (Fig. 7.5). The axonal branches that form the clusters are only sparsely distributed in the patch volume and the boutons formed along the branches display correlations on a small scale determined by the exact branching geometry. Detailed quantitative analysis of neuronal branching has focused mostly on the dendritic trees (Uylings and van Pelt, 2002). For the cortical axon, sporadic measurements such as branch number or branch length have been made for fully labelled 3-D reconstructions (Kisvárday et al., 1985, 1986). More systematic studies have been made for callosal and thalamic axons and different types of inhibitory cortical neurons in an attempt to quantify type specific differences in axonal branching (Tettoni et al., 1998; Gupta et al., 2000). Such differences can, perhaps not surprisingly, also be detected for the spiny and smooth cell types in the cat cortex (Binzegger et al., 2005). Branch number varied between 50 and 1,500, where smooth neurons generally form shorter branches than spiny neurons, but roughly twice as many branches. The surprising finding is that despite these differences, and the obvious differences in the overall gestalt of the different cell types, branch length and the logical or topological arrangement of the branches (‘dendrogram’) follow simple generic rules which are independent of cell type. An example dendrogram is shown in Fig. 7.6, representing the logic structure a layer 2/3 pyramidal cell axon (Fig. 7.5). Only the branching structure and the branch length are represented, and spatial information such as the location of each branch, or the direction it takes in the neuropil, is ignored. For each tree it is possible to introduce generations of branches that determine an order when branches

126

T. Binzegger et al.

Fig. 7.6 Dendrogram of the pyramidal cell axon shown in Fig. 7.5. Axonal branches are indicated in gray and are drawn vertically. Black circles indicate boutons. Branch points are indicated by horizontal gray lines

have appeared during the growth of the tree. Clearly, it is not possible to infer the real growth order from the dendrogram, but even the introduction of hypothetical generations is useful to analyze how branch statistics changes with generation. There are several branch ordering schemes possible, but the one that is particularly attractive because of its natural appeal is to start with the full tree and deduct branch order be moving backward in ‘time.’ Thus, the terminal branches in the dendrogram are assigned order 1 and after removing these branches, the new terminal branches are of order 2, and so on, until the root of the tree is reached. Astonishingly, for every cortical axon analyzed, the number of branches tripled when moving distally from one branch generation to the other, indicating topological self-similarity. A similar scaling law was found for the mean branch length per generation, but only for the lower generations, that is, the branches, that presumably form the axonal patches. Moving distally from the third to the second, and from the second to the first generation, the average branch length shortened by 70% from one generation to the other. We found that a simple three-parameter tree growth model (Galton–Watson branching process) can produce the same scaling relationship for branch number and branch length (Binzegger et al., 2005). This suggests that all cortical axons might be constrained by similar growth rules, and that they configure themselves in the 3-D space to satisfy further constraints imposed by connectivity or functional requirements.

7

An Axonal Perspective on Cortical Circuits

127

7.4 The Problem of Choice Simply from inspection of the vertical and horizontal gestalt of the axonal and dendritic morphology, it is evident that selectivity must exist in the choice of the postsynaptic cell type and in the horizontal position of the neurons of these cell types. That such specificity exists is therefore not challenged. The point of debate is if, and to what extent, an axon preferentially targets particular neurons or neuronal types beyond those implied by the gross vertical and horizontal morphology. The volume of an axonal patch (about 300 μm in diameter) contains potential target cell bodies and dendritic branches (or fragments thereof) originating from neurons of many different cell types, and the exact composition changes with vertical position of the patch (Binzegger et al., 2004). A rule needs therefore to be in place that assigns synapses formed by the axon with the dendritic segments or cell bodies in the patch volume. But what is the exact composition in the first place? While it is possible to determine the total number of neurons for each cortical layer (Beaulieu and Colonnier, 1983), the fraction that are excitatory (80–85% in each layer) or inhibitory (15–20%) (Gabbott and Somogyi, 1986), a detailed breakdown of the neuropil (cell number, axon, and dendrite) according to cell type is not possible by direct experimental observation. It is possible, however, to derive estimates based on published cell number counts and by multiplying this number with the laminar distribution of the dendrites and axons presented in Fig. 7.3. What such a quantitative estimate shows is that each layer consists of dendrites and axons originating from a unique mixture of cell types and intensity profiles (Fig. 7.7). The contribution of inhibitory cell types to each layer can be estimated from quantitative labeling studies. Inhibitory neurons are immunoreactive for parvalbumin (basket cells and chandelier cells) or calbindin (double bouquet cells, Martinotti cells, and neurogliaform cells). From these estimates it is clear that basket cells dominate in most layers, both in terms of the number of cell bodies and total length of dendrites. For the excitatory cell types no such markers exist, and estimates must be made by other means. In layer 6, for example, two types of pyramidal cells are encountered (Fig. 7.2). The subclass p6(L5/6) are characterized by an ascending axon which innervates almost exclusively the upper layers (mostly layer 4, p6(L4)), while the subclass p6(L5/6) has an axon which is contained to the deep layers 5 and 6. The subclass p6(L5/5) forms an efferent to the claustrum and can therefore be labeled exclusively by injecting a retrograde label into the claustrum. Using this method, Katz (1987) estimated that the subclass p6(L5/6) forms 25% of the total population of pyramidal cells in layer 6, and the remaining 75% is formed by the subclass p6(L4) (Fig. 7.7b). But because individual pyramidal cells p6(L5/6) tend to form most of their dendrites in the deeper layers, they contribute almost half of the dendrites in layer 6 (Fig. 7.7b). The subclass p6(L4) spreads the dendrites over several layer, contributing considerably to layers 4 and 5. A similar breakdown can be made for the boutons (or synapses) formed by the different cell types (Fig. 7.7c). The important point that follows from this analysis is that the neuropil supports an architecture where signals from many different sources

128

T. Binzegger et al. sp1 sm1

p5(2/3)

p5(5/6)

p6(4)

p6(5/6)

p2/3

p4

p4

p4

C

p6(4)

as

ss4(4)

p5(2/3) as

p2/3 as

p5(2/3)

p4

p4

p6(5/6)

L4

sm6

L6

sy

L1

p2/3

p2/3

L2/3

L5

p6(5/6)

as

L5

L1

p5(5/6)

p6(4)

L4

L6

b4

p6(4)

p6(4)

b5

sm6

sm1

ss4(2/3)

p5(2/3)

sm5

p5(5/6)

p2/3

ss4(4)

db2/3

b4

b2/3 sm2/3

B

ss4(4)

sm4

ss4(2/3)

L2/3

b2/3

sm2/3

p4

b2/3

sm4

p2/3

sm2/3

L1

ss4(2/3)

A

L2/3

b4

sy

L4

sy

L5

sy

L6

Fig. 7.7 Breakdown of the neuropil by layer and cell type. a Fraction of neurons of a given cell type. Gray-shaded areas indicate inhibitory cell types. b Fraction of dendrite formed by a given cell type. ( Fraction of synapses on boutons formed by a given cell type. Included in the estimates are the proportion of synapses from sources outside the area (‘sy’ symmetric synapses, ‘as’ asymmetric synapses)

can be mixed, even on a local level. An axonal patch located anywhere in area 17 has the possibility to simultaneously influence different cell types each of which has different physiological properties (Gupta et al., 2000) and communicates with different extrastriate areas and subcortical regions (Gilbert and Kelly, 1975). Similarly, a dendritic tree, whose branches typically span a similar volume as the axonal patch, can receive synapses from many of those cell types. The potential for multiple interactions is not restricted to cell types, but is also found for individual neurons, such that within a local population every neuron can form a synapse with every other neuron in the population. Labeling nearby layer 5 pyramidal cells in the rat slice revealed that the 3-D reconstructed axonal and dendritic trees always had close appositions between axons and dendrites (Kalisman et al., 2005). This local ‘all-to-all’ layout matches, of course, the qualitative observations of early investigators who likened the processes of Golgi-stained cortical neurons to tangled thickets like Ramon y Cajal (1989) where no structure could be

7

An Axonal Perspective on Cortical Circuits

129

observed. This is the source for the notion that cortical networks are random or diffuse (Sholl, 1959; Braitenberg and Schüz, 1998). However, an all-to-all network of synaptic connections is not realized, because the fraction of pairs of nearby layer 5 pyramidal cells that actually form synaptic connections are only 15% (Kalisman et al., 2005). More generally, experimental evidence indicates that less than 50% of axo-dendritic appositions form synapses, even if there is virtually no separation between the axonal and dendritic branches (Tamás et al., 1997). Even on the smallest scale choice exists. The density and axonal spread of thalamic afferents are such that any point in layer 4 overlaps with at least 400–800 X-axons and five times more Y-axons (Freund et al., 1985; Friedlander and Martin, 1989), and a theoretical argument shows that every point along an axon is in close apposition to more than one dendritic branch (Stepanyants et al., 2002).

7.5 Wiring Neurons The gross morphological features of the neuron morphology defined by the vertical and horizontal patchy distribution of the axon determine the main focal innervation sites. How an axonal patch connects to the various targets at those sites is regulated by the cortical connectivity rule. The rule investigated here is that the axon connects indiscriminately to the various targets. This is the quintessence of the overlap rule that has been traditonally applied to derive circuit diagrams, that is, whenever axon and dendrite overlap a connection is formed. The quantitative version of this rule adopts different forms, depending on the level of detail. The version that has been studied in most detail is termed Peters’ Rule (Braitenberg and Schüz, 1998) and has its roots in experimental observations of how the thalamic afferents connect to the spiny and smooth neurons in layer 4 of the rat cortex. Studying the postsynaptic targets of the asymmetric synapses formed by thalamic boutons, Peters and Feldman (1976) found that about 85% of the synapses were on spines (presumably from excitatory neurons) and 15% on dendritic shafts (presumably from inhibitory neurons). This proportion is similar to the spines and shafts in layer 4 which are able to receive asymmetric synapses, as can be inferred from inspecting the targets of an arbitrary selection of boutons forming asymmetric synapses in layer 4 (Peters and Feldman, 1976). The hypothesis put forward which might explain this correspondence is that ‘. . . their [the boutons of thalamic afferents] distribution with respect to postsynaptic targets may be essentially random, in the sense that no specific types of neurons receive the afferents’ (Peters and Feldman, 1976). Indeed, if the thalamic synapses are distributed randomly (i.e., indiscriminately) over the pool of possible postsynaptic targets, the thalamic afferents connect in direct proportion to the occurrence of the type-specific synaptic targets in the neuropil. It is important to note that Peters’ Rule characterizes the overall connectivity between population of neurons. The pooled synapses of all afferents innervating layer 4 (or all afferents innervating a more localized volume) distribute randomly over the dendrites of all spiny or smooth neurons in the same volume.

130

T. Binzegger et al.

No further assumption is made about how the individual afferents distribute their synapses over the target neurons. Peters’ Rule is easily extended to cortical axons and is shown to predict the correct proportions of spines and shafts contacted by the axons of spiny and smooth neurons in the mouse cortex (Braitenberg and Schüz, 1998). We extended it further to all cortical axons and derived predictions about the number of synapses (‘anatomical weight’) the axons of a particular cell type form with the dendrites of any other cortical cell type (Binzegger et al., 2004). The anatomical weights were estimated from the breakdown of the neuropil (Fig. 7.7). In essence, if a layer contains a total length of Di dendrites of cell type i, the fraction of dendrites this cell type forms in the layer are Di q= , Dk

(2)

k

where the sum is over all cortical cell types and is the total length of dendrite contained in the layer. If the axonal patch forms s synapses, Peters’ Rule dictates that the dendrites of type i in the volume receive sq of those synapses. In order to obtain all synapses the axon forms with the dendrites of type i, one has to sum over all layers where the axon forms synapses. A diagram summarizing the resulting circuit of pathways is shown in Fig. 7.1c. Peters’ Rule is feasible because the cloud of synapses formed by a population of axons is to a good approximation ‘thoroughly mixed’ (Braitenberg and Schüz, 1998) with the dendrites of the target population, such that a random distribution of synapses over the dendrites is conceivable. For individual axons this is obviously not true. The branches of an axonal patch are organized in tree-like structures, and the boutons formed by these branches do not form a homogenous cloud. So what does it mean to connect indiscriminately between individual neurons in a local population? Synapse formation between neurons is constrained to the axo-dendritic appositions between neurons, which suggests the rule to distribute randomly the pool of synapses formed by an individual axon onto the various apposition it forms with the neurons in the local population. Network connectivity is then largely determined by the cloud of axo-dendritic apposition whose organization has been extensively studied (Uttley, 1954; Liley and Wright, 1994; Stepanyants et al., 2002; Kalisman et al., 2003; Stepanyants et al., 2007). A difficulty with this approach is the need to introduce a critical distance between axonal and dendritic branches such that synapse formation is possible for branch separation smaller than the critical distance and that no synapse can be formed for a larger separation. Electron microscopy shows that the pre- and postsynaptic parts of a synapse are separated by only several nanometers (Palay, 1956), but the formation of spine necks and bouton terminaux of various lengths can significantly increase the critical distance for synapse formation. In practice, the critical distance is set globally, i.e., 2 μm for appositions between spiny neurons (Stepanyants et al., 2007), which leads to a linear dependence of the number of appositions with critical

7

An Axonal Perspective on Cortical Circuits

131

distance (Stepanyants et al., 2004; Kalisman et al., 2003). Despite these inherent ambiguities, this approach is useful because conceptually it separates the necessary aspect of synapse formation (appositions) from the actual formation of a synapse. The mapping of synapses onto the appositions has not been investigated in detail. However, because boutons appear at the synapse forming appositions of an axon, inspection of the bouton arrangement along axonal branches can indicate conspicuous clustering which would not be expected under the random setting. A systematic study of reconstructed cortical axons shows that the linear bouton arrangement along a strand of axon is not clustered, but is essentially random (homogenous Poisson process) with spiny neurons forming a bouton every 8 μm and smooth neurons every 5 μm (Anderson et al., 2002). Deviations from randomness occurred only at two places. First, there was a lack of very short interbouton distances, which can be trivially explained by the physical size of a bouton, which requires that there be a minimum center-to-center distance between neighbouring boutons formed along the same strand of axon. Second, axonal arbors involved in vertical projections or horizontal projections connecting two patches were often sparsely populated with boutons (Figs. 7.5 and 7.6). However, for the more distal branches which form the axonal patches, bouton density was constant.

7.6 Improving Peters’ Rule The concept of unspecific wiring is attractive, not least because it implies a simplicity in the rules of how to connect neurons. This simplicity contrasts strongly with the assumption that there are detailed connectivity rules. The resulting circuits are interesting because they represent the least specific connectivity structures consistent with the known quantitative neuroanatomy. But how realistic is the circuit diagram predicted by unspecific wiring? The existence of pathways between anatomical cell types can be directly tested by recording pairs of neurons that are subsequently labelled and identified anatomically. With this method, the majority of the pathways predicted by Peters’ Rule do exist (e.g., Stratford et al., 1996; Thomson and Bannister, 2003). There are a few exceptions, however. For example, no functional connection could be demonstrated from layer 3 pyramidal neurons to the upper layer 5 pyramidal cells (p5(L2/3)) and layer 4 pyramidal cells (Thomson and Bannister, 2003). While these connections may well be demonstrated in future studies (it is much harder to show the non-existence of a pathway), the clearest exception comes from anatomical studies which show that the chandelier cells form symmetric synapses exclusively with the axon initial segment of pyramidal cells (Somogyi et al., 1982). This example demonstrates two kinds of specificities that violate the assumption of indiscriminant wiring. First, chandelier cells should form synapses with smooth neurons as well, but typically no such connections are formed. Second, synapses should be distributed randomly along the dendritic branches (or soma), but the axons of chandelier cells target a specific substructure (the initial axonal segment). Similar

132

T. Binzegger et al.

substructure specificity is found for other axons of smooth neurons. For example, basket cells innervate the proximal regions of an excitatory neuron and inhibitory neurons, and double bouquet cells form synapses more distally along the dendrites of the target neuron (Tamás et al., 1997, 1998). Further substructural specificity may yet emerge from detailed maps of synapses on identified neurons. Peters’ Rule is easily adapted to incorporate this kind of specificity by allowing each cell type to specify ‘contactable regions’ on their dendrites or axons (Binzegger et al., 2004). For example, the region on pyramidal cells contactable by the chandelier cells is the initial axonal segment, and there is no region on smooth neurons which is contactable by the chandelier cells. Similar, the region on pyramidal cells contactable by the basket cells is the proximal part of the dendrites, and the region contactable by the chandelier cells is the distal part of the dendrites. Peters’ Rule predicts the existence of pathways between cell types, but it also predicts the anatomical weight associated with each pathway. For the spiny stellate cells and basket cells in layer 4 of cat area 17 the anatomical weights of their presynaptic cell types have been estimated based on experimental observations and, importantly, without the assumption that synapses distribute randomly over the contactable targets (Ahmed et al., 1994, 1997). A comparison with the weights predicted by Peters’ Rule showed good agreement (Fig. 7.8), in particular in the case of the spiny stellate cells. This suggests that Peters’ Rule, or the modified version thereof, is a very good

Fig. 7.8 Input map of a spiny stellate cell in layer 4. Shown are the proportions of synapses on the spiny stellate dendrite (black) arising from the different cell types (gray). Estimates made by Ahmed et al. (1994) are shown in black. For comparison, independent estimates based on Peters’ Rule are indicated in italics

7

An Axonal Perspective on Cortical Circuits

133

first approximation to the cortical circuit that is formed by pathways between cell types. How might deviations from Peters’ Rule originate from the local interaction of the axonal and dendritic branches? One possibility is that axonal growth is in itself specific, that is, branches avoid dendritic segments of certain subregions or types of neurons such that particular axo-dendritic combinations simply do not occur in close apposition. At the same time branches might be attracted by other types of segments, with the result that these axo-dendritic appostions occur more frequently. Thus, even if synapses are distributed randomly over the appositions, there will be a bias in the synaptic connections toward certain substructures, neurons, or cell types. For excitatory axons there is no evidence for a biased growth toward or away from dendrites. Nearby pairs of pyramidal cells in layer 5 of the rat cortex had the same number of axo-dendritic appositions, irrespectively of being functionally connected or unconnected (Kalisman et al., 2005). Similarly, in a theoretical study based on pairs of neighbouring 3-D reconstructed superficial layer neurons, the number of formed appositions was determined before and after the axon was ‘decorrelated’ from the dendrites by shifting it randomly by a small distance (Stepanyants et al., 2004). For axons from excitatory neurons no difference was detected in the number of appositions, suggesting that axons and dendrites were spatially uncorrelated. However, for the axon of inhibitory neurons the number of appositions was significantly larger than expected if the pair was functionally connected, indicating a bias in axonal growth. Inhibitory neurons form smaller axonal branches than excitatory neurons, which might enable them to correlate their trajectories more easily with the dendrites of the selected neuron. Another possible deviation from Peters’ Rule might occur if the synapses are formed only at appositions between the axon and selected dendritic segments. This does not have to result in non-random bouton placement along axons as long as the selected targets are homogeneously distributed (Anderson et al., 2002). In general, a definite validation or rejection of Peters’ Rule is difficult because computing the anatomical weights involves an accurate characterization of the composition of the neuropil and the axonal and dendritic morphology. For example, increasingly sophisticated imaging methods allow the determination of spatial input maps to a specific neuron (Dantzker and Callaway, 2000; Kozloski et al., 2001; Yabuta et al., 2001; Schubert et al., 2003; Morishima and Kawaguchi, 2006) revealing connection specificity in the sense that two different cell types within the same region receive differential input from the various sources whose axons innervate this region. These findings may reflect genuine deviations from Peters’ Rule which would have to be addressed, but systematic differences in the dendritic morphology of the target types might also be an explanation (Schubert et al., 2003).

7.7 Computation in Daisy Architectures At the heart of cortical computation is the local cortical circuit. While feedforward input to an area may drive a neuron to fire, and feedback input may modulate this activation, numerically these afferent connections are weak in comparison to the

134

T. Binzegger et al.

connections arising from within an area. The quantitative analysis of the number of synapses in area 17 suggests that no more than 30% of the asymmetric synapses on spiny dendrites are from these afferents, and that the vast majority of synapses are used in the local circuits (Binzegger et al., 2004). The local circuit forms the big loop described by Gilbert and Wiesel (1981), but the total number of synapses involved in forming the loop amounts to only 21% of all the asymmetric synapses in the circuit. An even larger number of asymmetric synapses (34%) are involved in forming connections between neurons in the same layer. More generally, the local circuit is dominated by a small number of pathways involving a large proportion of the synapses. In addition there are a large number of pathways which involve only a small number of synapses. The functional significance of such a long tail distribution of anatomical weights is not understood, but it is interesting to note that a similar distribution was also found for the functional synaptic weights between individual neurons (Song et al., 2005). The superficial layers are singled out by their position in the cortical network. They receive afferent feedback input from other cortical areas and are driven, via layer 4, by the feedforward input from the lateral geniculate nucleus. In addition, the quantitative circuit diagram (Fig. 7.1c) shows that no other layer contains so many synapses arising from the neurons within the layer itself. More than two-thirds of all excitatory synapses formed with superficial layer pyramidal cell dendrites originate from other superficial layer pyramidal cells. This suggests that the superficial layers receive peripheral sensory information as well as processed information from other cortical areas and that the massive network in the layer integrates and processes this information using the prominent system of patchy horizontal connections (‘the daisy architecture’). While these properties distinguish layer 2/3 from other layers, a wealth of anatomical studies suggest that these organizational principles are not unique to area 17, but are common to the superficial layer of all higher mammals (Douglas and Martin, 2004). We have speculated that a basic computational role of the patchy daisy architecture is to enable rich mixing of information between small population of neurons while limiting signal redundancy and spike correlation between the neurons in the network (Binzegger et al., 2007). The intuition is as follows. A population of neurons within a target patch C receives input spike trains from, say, two source populations (A and B) that are well separated spatially, and whose patchy axonal projection ‘petals’ spatially overlap with the target patch. In this way statistically independent information from A and B can be combined in population C using spike time-dependent synaptic mechanisms. If, however, populations A and B are not well separated, so that the primary bouton arborizations overlap with each other or with the petals, then the interactions between the two populations of neurons will induce correlations between their output spike trains and so reduce the efficacy of the spike time-dependent processing in C (Fig. 7.9). We consider that the conspicuous separation between the proximal patch of superficial layer neurons and the next closest patch might be a mechanism to avoid this scenario, by limiting unwanted spike correlation in the network.

7

An Axonal Perspective on Cortical Circuits

135

Fig. 7.9 The separation of patches limits unwanted spike correlation in a network of spiking neurons. Depicted is the situation where a population C (stippled line) receives projection petals (small black and grey circles) from two independent source populations A (large black circle) and B (large grey circle). In the first scenario (top), the petal formed by population A is not well separated from A, ultimately leading to an increase in correlated firing between the neurons in A and B. In the second scenario (bottom), the petal formed by A is well separated from both A and B, which does not increase correlated firing between the neurons in A and B

How the axonal patches of the individual neurons in the local source populations are mapped onto the petals in the daisy determines the degree of redundancy in the target population. One possibility (large redundancy) is that each neuron in the local population forms as many patches as there are petals and each patch is mapped onto one petal. However, the clustering study of the bouton clouds (Binzegger et al., 2007) suggests a scenario with less redundancy. The number of identified clusters per axon varied considerably in the study, even if the neurons were restricted to a single cell type such as the layer 2/3 pyramidal cells (1–7 patches). What this suggests is that patches are mapped irregularly onto the petals of the daisy, such that the patches of nearby neurons might overlap with entirely different petals. Depending on the active neurons in a local source population, information might therefore be routed to different combinations of target populations, thereby increasing the possibility to mix signals in the network in various combinations. Thus, the patchy horizontal connections in 2/3 suggest a sophisticated mechanism by which local populations of neurons exchange and mix signals. Each local population processes information from a large variety of independent signals originating from various distal sites within the layer. The local network that processes this information is in character distinctly different from the strongly structured pathways on the larger scale. The rich arborizations of the axons and dendrites of the neurons generate a diffuse architecture where any two neurons have the potential to connect, and many different circuit instantiations are possible. Both modeling and experimental data suggest that a chief signature of this network is the abundance of recurrent excitatory connections (Douglas et al., 1995;

136

T. Binzegger et al.

Holmgren et al., 2003), that is, a neuron has a high probability to receive synaptic connections from its local excitatory targets. The interplay of these recurrent connections with the local inhibitory neurons generates competitive processing such that subgroups of recurrently connected excitatory neurons tend to amplify the initial input they receive, and at the same time each subgroup uses inhibitory neurons to increasingly suppress all other subgroups in the population. Which subpopulation wins the contest for dominance depends on the initial input to each group, and which neurons form a subgroup depends on the detailed wiring and synaptic weight of the local connections (Hahnloser et al., 2000; Xie and Song, 2002). More specifically, the basic function of the local network might be to select competitively the subgroup of neurons whose a priori expectation encoded in their local weight matrix best matches the input (Hahnloser et al., 2000; Douglas and Martin, 2007a) and feed this solution back to the horizontal network until some mutual consent in form of an overall steady state in layer 2/3 is achieved. In this way, the superficial layer might evolve from a variety of potentially conflicting local inputs from a wide range of sources to the most consistent overall response. This response is fed to layer 5, the primary output layer to the subcortical areas involved in motor action. The functional model presented here is general enough so that its principles might apply universally to all cortical areas. But it is also detailed enough to give justice to the anatomical circuit complexity which exists at every spatial scale. On close inspection of the neuron geometries one finds that many aspects of circuit structure result from generic principles of organization. This suggests that quite simple construction rules might generate cortical circuit structure and the sophisticated function it subserves. Acknowledgments Preparation of the MS was supported by EU grants ‘Daisy’ FP6 2005-015803 and ‘SECO FP7–216593. Tom Binzegger has a Research Councils UK Academic Fellowship.

References Ahmed B, Allison JD, Douglas RJ, Martin KA (1997) An intracellular study of the contrastdependence of neuronal activity in cat visual cortex. Cereb Cortex 7:559–570 Ahmed B, Anderson JC, Douglas RJ, Martin KA, Nelson JC (1994) Polyneuronal innervation of spiny stellate neurons in cat visual cortex. J Comp Neurol 341:39–49 Anderson JC, Binzegger T, Douglas RJ, Martin KA (2002) Chance or design? some specific considerations concerning synaptic boutons in cat visual cortex. J Neurocytol 31:211–229 Anderson PA, Olavarria J, Sluyters RCV (1988) The overall pattern of ocular dominance bands in cat visual cortex. J Neurosci 8(6):2183–2200 Beaulieu C, Colonnier M (1983) The number of neurons in the different laminae of the binocular and monocular regions of area 17 in the cat. J Comp Neurol 217:337–344 Binzegger T, Douglas RJ, Martin KAC (2004) A quantitative map of the circuit of cat primary visual cortex. J Neurosci 24:8441–8453 Binzegger T, Douglas RJ, Martin KAC (2005) Axons in cat visual cortex are topologically selfsimilar. Cereb Cortex 15:152–165 Binzegger T, Douglas RJ, Martin KAC (2007) Stereotypical bouton clustering of individual axons in cat visual cortex. J Neurosci 27:12242–12254

7

An Axonal Perspective on Cortical Circuits

137

Bosking WH, Zhang Y, Schofield B, Fitzpatrick D (1997) Orientation selectivity and the arrangement of horizontal connections in tree shrew striate cortex. J Neurosci 17:2112–2127 Braitenberg V, Lauria F (1960) Toward a mathematical description of the grey substance of nervous systems. In: Supplemento al volume XVIII, serie X, del Nuovo Cimento, volume suppl 18 of 10 Braitenberg V, Schüz A (1998) Cortex: statistics and geometry of neuronal connections, 2nd edn. Springer, Heidelberg Cajal SR (1908) Histologie du système nerveux. CSIC, Madrid (reprinted 1972) Cajal SR (1989) Recollections of My Life (trans: Craigie EH with the assistance of Cano J). MIT Press, Cambridge Dantzker JL, Callaway EM (2000) Laminar sources of synaptic input to cortical inhibitory interneurons and pyramidal neurons. Nat Neurosci 3:701–707 Douglas RJ, Koch C, Mahowald M, Martin KA, Suarez HH (1995) Recurrent excitation in neocortical circuits. Science 269:981–985 Douglas RJ, Martin KAC (2004) Neuronal circuits of the neocortex. Annu Rev Neurosci 27: 419–451 Douglas RJ, Martin KAC (2007a) Recurrent neuronal circuits in the neocortex. Curr Biol 17: R496–R500 Douglas RJ, Martin KAC (2007b) Mapping the matrix: the ways of neocortex. Neuron 56: 226–238 Freund TF, Martin KAC, Somogyi P, Whitteridge D (1985) Innervation of cat visual areas 17 and 18 by physiologically identified X- and Y- type thalamic afferents. II. Identification of postsynaptic targets by GABA immunocytochemistry and Golgi impregnation. J Comp Neurol 242:275–291 Friedlander MJ, Martin KA (1989). Development of Y-axon innervation of cortical area 18 in the cat. J Physiol 416:183–213 Gabbott PL, Somogyi P (1986) Quantitative distribution of GABA-immunoreactive neurons in the visual cortex (area 17) of the cat. Exp Brain Res 61:323–331 Gilbert CD (1983) Microcircuitry of the visual cortex. Ann Rev Neurosci 6:217–247 Gilbert CD, Kelly JP (1975) The projections of cells in different layers of the cat’s visual cortex. J Comp Neurol 163:81–105 Gilbert CD, Wiesel TN (1979) Morphology and intracortical projections of functionally characterised neurones in the cat visual cortex. Nature 280:120–125 Gilbert CD, Wiesel TN (1981) Laminar specialization and intracortical connections in cat primary visual cortex. In: Schmitt FO, Worden FG, Adelman G, Dennis SG (eds) The organization of the cerebral cortex. MIT Press, Cambridge, pp 163–191 Gilbert CD, Wiesel TN (1983) Clustered intrinsic connections in cat visual cortex. J Neurosci 3:1116–1133 Gupta A, Wang Y, Markram H (2000) Organizing principles for a diversity of gabaergic interneurons and synapses in the neocortex. Science 287:273–278 Hahnloser RH, Sarpeshkar R, Mahowald MA, Douglas RJ, Seung HS (2000) Digital selection and analogue amplification coexist in a cortex-inspired silicon circuit. Nature 405:947–951 Holmgren C, Harkany T, Svennenfors B, Zilberter Y (2003) Pyramidal cell communication within local networks in layer 2/3 of rat neocortex. J Physiol 551:139–153 Hubel DH, Wiesel TN (1962) Receptive fields, binocular interaction and functional architecture in the cat’s visual cortex. J Physiol (Lond) 160:106–154 Hubel DH, Wiesel TN (1965) Receptive fields and functional architecture in two nonstriate visual areas (18 and 19) of the cat. J Neurophysiol 28:229–289 Hubel DH, Wiesel TN (1972) Laminar and columnar distribution of geniculo-cortical fibers in the macaque monkey. J Comp Neurol 146:421–450 Jones EG (1975) Varieties and distribution of non-pyramidal cells in the somatic sensory cortex of the squirrel monkey. J Comp Neurol 160:205–268 Kalisman N, Silberberg G, Markram H (2003) Deriving physical connectivity from neuronal morphology. Biol Cybern 88:210–218

138

T. Binzegger et al.

Kalisman N, Silberberg G, Markram H (2005) The neocortical microcircuit as a tabula rasa. Proc Natl Acad Sci USA 102:880–885 Katz LC (1987) Local circuitry of identified projection neurons in cat visual cortex brain slices. J Neurosci 7:1223–1249 Kisvárday ZF, Bonhoeffer T, Kim DS, Eysel UT (1996) Functional topography of horizontal neuronal networks in cat visual cortex (area 18). In: Aertsen A, Braitenberg V (eds) Brain theory – biological basis and computational principles. Elsevier, Amsterdam, pp 247–260 Kisvárday ZF, Cowey A, Hodgson AJ, Somogyi P (1986) The relationship between gaba immunoreactivity and labelling by local uptake of [3 h]gaba in the striate cortex of monkey. Exp Brain Res 62:89–98 Kisvárday ZF, Martin KAC, Whitteridge D, Somogyi P (1985) Synaptic connections of intracellularly filled clutch neurons: a type of small basket neuron in the visual cortex of the cat. J Comp Neurol 241:111–137 Kozloski J, Hamzei-Sichani F, Yuste R (2001) Stereotyped position of local synaptic targets in neocortex. Science 293:868–872 Krone G, Mallot H, Palm G, Schuz A (1986) Spatiotemporal receptive fields: a dynamical model derived from cortical architectonics. Proc R Soc Lond B 226:421–444 Liley DTJ, Wright JJ (1994) Intracortical connectivity of pyramidal and stellate cells: estimates of synaptic densities and coupling symmetry. Network Comput Neural Syst 5:175–189 Lorente de Nó R (1949) Physiology of the nervous system, chapter the cerebral cortex: architecture, intracortical connections, motor projections. Oxford University Press, London, pp 288–330 Luhmann HJ, Martinez Millan L, Singer W (1986) Development of horizontal intrinsic connections in cat striate cortex. Exp Brain Res 63:443–448 Lund JS, Angelucci A, Bressloff PC (2003) Anatomical substrates for functional columns in macaque monkey primary visual cortex. Cereb Cortex 13:15–24 Lund JS, Boothe RG (1975) Interlaminar connections and pyramidal neuron organisation in the visual cortex, area 17, otf the macaque monkey. J Comp Neurol 159:305–334 Lund JS, Henry GH, MacQueen CL, Harvey AR (1979) Anatomical organization of the primary visual cortex (area 17) of the cat. A comparison with area 17 of the macaque monkey. J Comp Neurol 184:599–618 Malach R, Amir Y, Harel M, Grinvald A (1993) Relationship between intrinsic connections and functional architecture revealed by optical imaging and in vivo targeted biocytin injections in primate striate cortex. Proc Natl Acad Sci 90:10469–10473 Martin KA, Whitteridge D (1984) Form, function and intracortical projections of spiny neurones in the striate visual cortex of the cat. J Physiol (Lond) 353:463–504 Morishima M, Kawaguchi Y (2006) Recurrent connection patterns of corticostriatal pyramidal cells in frontal cortex. J Neurosci 26:4394–4405 Ohki K, Reid RC (2007) Specificity and randomness in the visual cortex. Curr Opin Neurobiol 17:401–407 O’Leary JL (1941) Structure of area striata of the cat. J Comp Neurol 75:131–164 Palay SL (1956) Synapses in the central nervous system. J Biophysic Biochem Cytol Suppl 2: 193–201 Peters A, Feldman ML (1976) The projection of the lateral geniculate nucleus to area 17 of the rat cerebral cortex. i. general description. J Neurocytol 5:63–84 Rockel AJ, Hiorns RW, Powell TP (1980) The basic uniformity in structure of the neocortex. Brain 103:221–244 Rockland KS, Lund JS, Humphrey AL (1982) Anatomical binding of intrinsic connections in striate cortex of tree shrews (Tupaia glis). J Comp Neurol 209:41–58 Schubert D, Kotter R, Zilles K, Luhmann HJ, Staiger JF (2003) Cell type-specific circuits of cortical layer IV spiny neurons. J Neurosci 23:2961–2970 Sholl DA (1959) A comparative study of the neuronal packing density in the cerebral cortex. J Anat 93:143–158

7

An Axonal Perspective on Cortical Circuits

139

Somogyi P, Freund TF, Cowey A (1982) The axo-axonic interneuron in the cerebral cortex of the rat, cat and monkey. Neuroscience 7:2577–2607 Song S, Sjöström PJ, Reigl M, Nelson S, Chklovskii DB (2005) Highly nonrandom features of synaptic connectivity in local cortical circuits. PLoS Biol 3:e68 Stepanyants A, Hirsch JA, Martinez LM, Kisvárday ZF, Ferecskó AS, Chklovskii DB (2007) Local potential connectivity in cat primary visual cortex. Cereb Cortex 18:13–28 Stepanyants A, Hof PR, Chklovskii DB (2002) Geometry and structural plasticity of synaptic connectivity. Neuron 34:275–288 Stepanyants A, Tamás G, Chklovskii DB (2004) Class-specific features of neuronal wiring. Neuron 43:251–259 Stratford KJ, Tarczy-Hornoch K, Martin KA, Bannister NJ, Jack JJ (1996) Excitatory synaptic inputs to spiny stellate cells in cat visual cortex. Nature 382:258–261 Szentágothai J (1973) Synaptology of the visual cortex. In: Jung R (ed) Handbook of sensory physiology: visual centers in the brain, vol VII/3. Springer, Berlin, pp 269–324 Szentágothai J (1975) The ‘module-concept’ in cerebral cortex architecture. Brain Res 95:475–496 Szentágothai J, Arbib MA (1974) Conceptual models of neural organization. Neurosci Res Program Bull 12:313–510 Tamás G, Buhl EH, Somogyi P (1997) Fast IPSPs elicited via multiple synaptic release sites by different types of GABAergic neurones in the cat visual cortex. J Physiol (Lond) 500:715–738 Tamás G, Somogyi P, Buhl EH (1998) Differentially interconnected networks of GABAergic interneurons in the visual cortex of the cat. J Neurosci 18(11):4255–4270 Tettoni L, Gheorghita-Baechler F, Bressoud R, Welker E, Innocenti GM (1998) Constant and variable aspects of axonal phenotype in cerebral cortex. Cereb Cortex 8:543–552 Thomas E, Patton P, Wyatt RE (1991) A computational model of the vertical anatomical organization of primary visual cortex. Biol Cybern 65:189–202 Thomson AM, Bannister AP (2003) Interlaminar connections in the neocortex. Cereb Cortex 13: 5–14 Uttley AM (1954) The probability of neural connexions. Proc R Soc Biol Sci 144:229–240 Uylings HBM, van Pelt J (2002) Measures for quantifying dendritic arborizations. Network 13:397–414 Winfield DA, Brooke RN, Sloper JJ, Powell TP (1981) A combined golgi-electron microscopic study of the synapses made by the proximal axon and recurrent collaterals of a pyramidal cell in the somatic sensory cortex of the monkey. Neuroscience 6:1217–1230 Xie X, Hahnloser RH, Seung HS (2002) Selectively grouping neurons in recurrent networks of lateral inhibition. Neural Comput 14:2627–2646 Yabuta NH, Sawatari A, Callaway EM (2001) Two functional channels from primary visual cortex to dorsal visual cortical areas. Science 292:297–300

Chapter 8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns and Serve as Primary Classifiers of Interneurons in a Cortical Column Moritz Helmstaedter and Dirk Feldmeyer

The impact of a neuron on the electrical activity of other neurons is conveyed predominantly via chemical synapses between the axon of the presynaptic neuron and the soma or dendrites of the postsynaptic neurons. If different types of putative postsynaptic neurons are spatially separated, e.g., located in different regions of the neocortex, it is therefore possible to predict the synaptic connectivity of a given neuron from the spatial distribution of its axon (Fig. 8.1a). This assumption has been used extensively in tract tracing studies (Taylor and Weiss, 1965; Lasek et al., 1968; Cowan et al., 1972; LeVay and Gilbert, 1976; Chmielowska et al., 1989), and equivalently in axonal degeneration studies (Hoff, 1932; Walker, 1936; Glees, 1946; Evans and Hamlyn, 1956; Gray and Hamlyn, 1962; Colonnier, 1964; Hubel and Wiesel, 1969; Jones and Powell, 1970; White, 1979, reviewed, e.g., in White and Keller (1989) and Jones (2007) where the populations of presynaptic and postsynaptic neurons were well separated (by millimeters to centimeters). Such predictions of neuronal connectivity based on the geometry of long-range axonal projections have proven very effective in, e.g., the primary visual and primary somatosensory system (Hubel and Wiesel, 1968, 1969; Jones and Powell, 1970; Killackey, 1973; Wise and Jones, 1978; Gilbert and Wiesel, 1979; Herkenham, 1980; Chmielowska et al., 1989; Lu and Lin, 1993). For the analysis of local circuits, such as in primary sensory areas of the neocortex or in the CA1 field of the hippocampus, such a prediction is more challenging since the different types of putative postsynaptic neurons are not well segregated spatially. In these cases, the geometry of the overlap between axonal projections of putative presynaptic neurons and the dendrites of the different types of putative postsynaptic neurons has to be analyzed at a finer spatial resolution (Fig. 8.1b). Ideally, synapses between a population of presynaptic neurons and putative postsynaptic neurons can be directly identified. Available techniques for the detection of synapses include ultrastructural analysis by electron microscopy (Gray, 1959; Uchizono, 1965; Colonnier, 1968; Jones and Powell, 1970; Markram et al., 1997;

M. Helmstaedter (B) Max Planck Institute for Medical Research, Heidelberg, Germany e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_8, 

141

142

M. Helmstaedter and D. Feldmeyer A

C

x

=

B

axon

dendrite

predicted innervation

Fig. 8.1 Schematic of how axonal projections can be used to infer synaptic connectivity. a Tract tracing allows to distinguish postsynaptic targets of a given presynaptic neuron population based on the assumption that the dendrites of neurons in parts of the cortex several millimeters apart represent different types of postsynaptic targets. b At a finer spatial scale, the prediction of synaptic connectivity requires sufficient proximity of axons and dendrites. c Quantification of axo-dendritic overlap interpreted as the probability of synaptic innervation: neurite trees (top row) are converted to neurite path length density maps (middle row, unit μm per μm3 ), which can be low-pass filtered and interpolated (bottom row). The product of axonal and dendritic length density can be interpreted as the probability of establishing a synaptic contact (predicted innervation probability, for details see text)

cf. Jones, 2007; Peters, 2007; White, 2007 for recent historical accounts) or synapsespecific labeling for imaging at the light microscopic level (Bloom et al., 1979; De Camilli et al., 1983; Magrassi et al., 1987; Lichtman and Smith, 2008). It has, however, so far not been possible to detect synaptic contacts exhaustively for larger populations of pre- and postsynaptic neurons (but see White et al., 1986; Chen et al., 2006, for an exception, and Denk and Horstmann, 2004; Micheva and Smith, 2007; Helmstaedter et al., 2008a, for promising new technologies). Therefore, surrogate techniques with an intermediate spatial resolution have been used to infer synaptic connectivity from light microscopic images of axons and dendrites. The major assumption on which such techniques are based is that the probability of finding a synaptic contact is proportional to both the density of axons (belonging to the population of putative presynaptic neurons) projecting into a given region and to the density of dendrites (belonging to the population of putative postsynaptic neurons) projecting into the same region (Fig. 8.1c). The notion that axonal projections target all available dendrites in a locally random fashion was originally suggested for thalamocortical input to cortical layer 4 (Peters, 1979), extended as a principle to all thalamocortical afferents (White, 1979) and generalized as a “rule” to connectivity in the neocortex (“Peter’s rule”, e.g., Braitenberg and Schüz, 1998). The strong assumption of proportionality can actually be weakened into the following assumption: the number of synapses in a given volume is a monotonous function of the amount of axons and dendrites in that volume. This weaker formulation may allow to include more specific synaptic connectivity (i.e., exceptions to

8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

143

Peter’s rule, e.g., summarized in Braitenberg and Schüz (1998) and White (2007)). Another modification of the assumption of locally random connectivity is to consider potential synaptic connectivity on a smaller length scale (Shepherd et al., 2005; Stepanyants and Chklovskii, 2005; Stepanyants et al., 2008), i.e., to consider axons within a few micrometers around a dendrite, instead of within a whole cortical layer (0.1–1.0 mm). Independent of the exact functional relationship between axonal and dendritic density a simple extreme case will always hold true: if either of the two densities are zero, there cannot be a synapse. Therefore, the conclusion of two populations of neurons having no synaptic connections based solely on axon–dendrite overlap analysis is the strongest conclusion to draw. We will show later in this chapter how this simple notion can be used to define types of cortical interneurons based purely on their axonal projections (and their ensuing capacity to inhibit neurons located in neighboring cortical columns or not). Already the early studies by Ramón y Cajal and Lorente de Nó were based on the assumption that the projections of the axon with reference to the cortical areas, or cortical layers, were valid predictors of the neuronal connectivity between populations of neurons (Ramón y Cajal, 1904; Lorente de No, 1922, 1938, 1992; Ramón y Cajal, 1995). The analysis was focused on the laminar distribution of axonal projections from a soma-centered point of view (Sholl, 1953, 1955; Gupta et al., 2000). More recently, the analysis of axonal projections was related to the presumed outlines of cortical columns, especially in rodent barrel cortex (Feldmeyer et al., 1999; Lübke et al., 2000; Bender et al., 2003; Lübke et al., 2003). In the following, this type of analysis is summarized first. It is shown how the definition of innervation domains matched the distribution of identified synaptic contacts between synaptically coupled pairs of neurons, as established by paired intracellular recordings. Subsequently, we describe the application of this analysis to cortical interneurons.

8.1 Definition of a Cortical Column by the Geometry of Axonal Domains Cortical columns were postulated as modular structures in the primary sensory areas comprising on the order of 104 neurons. Initially, such modular structures were shown by physiological experiments mapping the coarse receptive field structure in primary sensory areas (Mountcastle, 1957, 1997). Anatomical correlates of these cortical modules were first found in primary visual cortex (Hubel and Wiesel, 1969) as alternating patterns of thalamocortical axons originating from the ipsi- or contralateral eye. In the somatosensory cortex of rodents, well-delineated differences in neuronal soma density in layer 4 were found and coined barrels (Woolsey and Van der Loos, 1970). Thalamocortical axon tracing showed distinct patches of thalamocortical afferents primarily terminating in layer 4 (Killackey, 1973; Wise and Jones, 1978). Thalamocortical projections to supra- and infragranular layers 3 and 6 were also found, but at a lower density (s. Fig. 8.2a; Wise and Jones, 1978; Herkenham, 1980; Chmielowska et al., 1989). Together, this data predicted

144

M. Helmstaedter and D. Feldmeyer

A

B

C

D

E

Pia

L 2/3

L4

L5

Fig. 8.2 Thalamocortical and intracortical axonal projections allow the definition of a cortical column based on predicted innervation domains. a Thalamocortical afferents (from VPM) are clustered in layer 4 of barrel cortex, thus defining the outline of cortical columns in layer 4 (from Chmielowska et al., 1989). b Overlay of reconstructions of synaptically connected pairs of layer 4 spiny neurons (dendrites: red, axons: blue) and L2/3 pyramidal neurons (dendrites: white). c Average axonal length density map of the L4 spiny neuron axonal projection shown in b. d Average dendritic length density map of the L2/3 pyramidal neuron dendrites shown in b. e innervation density map (the product of c and d) for the L4-to-L2/3 pyramidal neuron connection. Note how the axonal projection from L4 thus defines the outlines of a cortical column in L2/3 (b–e modified from Lübke et al., 2003, with permission)

monosynaptic connections from the thalamic projection neurons to neurons located in cortical layers 3–6, and confirmed the anatomical outlines of barrels in layer 4 as the innervation volume of thalamocortical projections.

8.2 Inference of Synaptic Connectivity Within a Cortical Column To investigate synaptic connectivity within a cortical column, an approach comparable to the thalamocortical axon tracing studies was pursued. The axons of L4 excitatory neurons were stained by intracellular dye filling of single L4 neurons (note that for the thalamocortical axon tracing, in contrast, axonal degeneration or extracellularly applied dyes were used for tracing), and the geometry of L4 axons was analyzed with reference to the barrel in layer 4 (Fig. 8.2b, Lübke et al., 2003). Evidently, axons of L4 spiny neurons were largely confined to a conical volume outlined by and centered in the barrel in layer 4 (this was true for the axon of virtually each single L4 neuron, see Fig. 8.2b). This axonal projection volume then allows to define the outlines of a cortical column in supra- and infragranular layers as the projection volume of the axons of all L4 neurons that are located within a thalamocortical innervation domain in layer 4. One way of quantifying such an average projection volume is to compute the iso-axonal density surface that contained a substantial fraction of the total axonal path length (e.g., 80% or 90%, Fig. 8.2c). Next, the analysis of axonal projection geometries can be used to infer synaptic connectivity with putative postsynaptic neurons. The predominant neuronal population in supragranular layers is pyramidal cells. Therefore, the average dendritic density of L2/3 pyramidal neurons was computed (Fig. 8.2d). Then, the probability of establishing a synaptic contact was calculated (s. Fig. 8.1c) as the product of the axonal density of putative presynaptic neurons and the dendritic density of putative

8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

145

postsynaptic neurons (Fig. 8.2e). This product density corresponds to an innvervation density (in the above sense) and its outline can again be calculated as the iso-innervation density line comprising 80% of the total innervation density (yellow outline in Fig. 8.2e). Finally, it was crucial to test whether these predicted innervation domains correspond in fact to regions with a high probability of finding synaptic contacts between the pre- and postsynaptic neurons. For that purpose, paired intracellular recordings between L4 excitatory neurons and L2/3 pyramidal neurons were made (Feldmeyer et al., 2002). The existence of a synaptic contact was tested by eliciting an action potential in the L4 neuron and recording the somatic membrane potential in the putative postsynaptic L2/3 neuron (Fig. 8.3a). Then, for those pairs of neurons that had shown monosynaptic transmission, axons and dendrites were reconstructed and synaptic contacts were searched as close appositions of L4 axon and L2/3 dendrite A

B

postsynaptic Vm (L2/3 pyramidal neuron) presynaptic Vm (L4 spiny stellate)

30 mV 0.5 mV 1 ms

10 µm

D

C

500 nm

100 µm

Fig. 8.3 Correlating predicted innervation with evidence of synapses. a Paired intracellular whole-cell recording of a L4 spiny neuron and a L2/3 pyramidal neuron. Presynaptically elicited AP (bottom) and four consecutive traces of the postsynaptically recorded somatic EPSPs (top). b Close apposition of a L4 axon and a L2/3 pyramidal dendrite detected at the light microscopic level allows the identification of putative synaptic contacts. c Proof of a synaptic contact at the electron-microscopic level (same connection as in a,b). d Overlay of the locations of putative synaptic contacts as detected by light microscopy (blue dots) with the predicted innervation domain (yellow). Approximately 90% of the contacts were in fact within the iso-innervation density line comprising 80% of the total innervation density (yellow contour line). White arrow: barrel center. Modified from Feldmeyer et al. (2002) and Lübke et al. (2003) with permission

146

M. Helmstaedter and D. Feldmeyer

at the light microscopic level (Fig. 8.3b). Since the resolution of light microscopy is not sufficient to distinguish a close apposition from a synaptic contact, electronmicroscopic images of such locations were taken, and the existence of synaptic contacts was determined based on the common ultrastructural features of synapses (Fig. 8.3c, s. Markram et al., 1997; Feldmeyer et al., 2002; Silver et al., 2003). An overlay of the predicted innervation domain and the synaptic contacts then showed that almost 90% of the synaptic contacts were within the 80% innervation domain (Fig. 8.3d, Lübke et al., 2003). We also tested whether the prediction of synaptic connectivity based on the axonal projections of L4 neurons was valid for interneurons in layer 2/3 as putative postsynaptic targets (Fig. 8.4, Helmstaedter et al., 2008b). Monosynaptic connections from L4 spiny neurons were found for most investigated types of interneurons in layer 2/3 (local, lateral, and translaminar inhibitors, see below; note, however, that the synaptic properties of these connections were target cell-specific and that at least one type of L2/3 interneurons did not seem to receive excitatory monosynaptic input from layer 4, see Dantzker and Callaway, 2000; Helmstaedter et al., 2008b). The spatial extent of the innervation domains of these connections (Fig. 8.4b) confirmed the outlines of a presumed cortical column in the supragranular layers (cf. Figs. 8.4b and 8.2e). A

B

L1

L1

L2/3 L2/3

2ms L4

1mV 25mV

L5A L5B

100 µm

L4 L5A L5B

Fig. 8.4 Innervation domain of the L4 spiny neuron-to-L2/3 interneuron connection. a Monosynaptically connected pair of L4 spiny stellate (dendrites, red, axon, blue) and L2/3 interneuron (dendrites, black, axon not shown) as demonstrated by paired whole-cell recording (right panel: presynaptic action potential followed by 10 consecutive postsynaptic responses). Note the high efficacy of the connection in spite of the substantial inter-soma distance of the neuron pair (∼300 μm). b Spatial extent of the L4-to-L2/3 interneuron innervation domain. The product of L4 axonal density (left panel) and L2/3 dendritic density (middle panel) yields the predicted innervation domain for this type of connection (right panel). Note that the outlines of this innervation domain are also aligned with the outlines of the home column (cf. Fig. 8.2e). Modified with permission from Helmstaedter et al. (2008b)

8.3 Classification of Inhibitory Interneurons Based on Their Innervation Domains with Reference to Cortical Columns While the axonal projections of excitatory neurons can thus be used to define cortical columns as thalamocortical and intracortical innervation domains, the axonal projections of interneurons can in turn be used to determine the function (or “type”)

8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

147

of the presynaptic neurons themselves. In the neocortex, a large variety of interneuron types have been described ever since Ramón y Cajal’s first drawings of “local axon” cells (Ramón y Cajal, 1904; Ramón y Cajal, 1995). As numerous as the suggested classes are the parameters proposed for the classification of interneurons. Parameters describing electrical excitability, dendritic geometry, protein expression, and axonal geometry are widely used. The large variability of most of these parameters has resulted in many attempts to detect patterns and define classes of interneurons based on the apparent similarities and differences in this parameter space (e.g., Gupta et al., 2000; Ascoli et al., 2008). Given these many possible classification parameters, it is important to first decide on the priority of parameters. Is, for example the action potential shape more important than details of the spine geometry? One way of deciding such questions is to prefer classification criteria that have clear functional relevance. In addition it would be desirable to prefer classification parameters that are both stable during the single experiment (on a timescale of minutes to hours) and show as little variability as possible between experiments (i.e., stability on a timescale of days to weeks, and between individual animals). Electrical excitability is functionally relevant (as it combines the effects of somato-dendritic morphology and the distribution of transmembrane conductances along the soma, dendrites, and axon), but potentially not very stable during an experiment (considering the dilution of the intracellular space when using low-access-resistance pipette recordings, potential mechanical effects on conductances due to impalement or movement of the recording pipette, and substantial dependence on the electrical activity of the neuron prior to, or during the experiment, see e.g., Bacci et al., 2004). It has not been shown conclusively to what extent electrical excitability is stable between experiments, especially if subtle aspects of “firing patterns” are studied (Gupta et al., 2000). The dendritic geometry is expected to be rather stable during an experiment, and, at least in the adult animal, between experiments (except for the number and geometry of spines, Grutzendler et al., 2002; Trachtenberg et al., 2002). It is however less clear what functional relevance the geometry of dendrites has beyond its direct biophysical effects (which are measured as electrical excitability; for a quantification of the relation between somato-dendritic morphology and electrical excitability see Mainen and Sejnowski, 1996 for variability between different cell types; see Schaefer et al., 2003 for effects of dendritic geometry within a pyramidal cell population; and for the possible causal effect of variability in interneuron dendrite geometry on their firing patterns, see Helmstaedter et al., 2009c). Protein expression bears functional relevance for some, but not all of the widely used antigens. Stability of protein expression is however a major concern, given the time course on which protein expression is regulated in cells, and even more so if not the expressed protein is studied, but intermediates such as mRNA. It can be argued, however, that the axonal projection of interneurons is a rather stable feature (it does not change during an experiment – at least in matured animals). The functional relevance of axonal projections follows from the arguments provided in the preceding sections. Axonal projections allow the inference of synaptic connectivity; and even if the exact functional relationship between axonal density, dendritic density, and innervation probability is not known (i.e., if there is

148

M. Helmstaedter and D. Feldmeyer

more specificity of innervation than just overlap statistics at the 50 μm scale), nonoverlap still carries information. Neurons which have no axon extending in a given postsynaptic region cannot monosynaptically inhibit neurons whose dendrites are solely located in that region. Interneurons, which have only short (but often very dense) axons, are especially amenable to such an analysis even in the context of such a small geometrical reference frame as the cortical columns (of ∼300 μm diameter). We have therefore quantified axonal projection patterns of interneurons with reference to cortical columns, in particular the barrel column in the primary somatosensory cortex of rodents. We found that in fact some interneurons can monosynaptically inhibit neurons in neighboring columns – while other types of interneurons have target structures predominantly within a cortical (barrel) column and cannot inhibit neurons in neighboring columns. These results are summarized in the following (Figs. 8.5 and 8.6, Helmstaedter et al., 2009b).

8.4 Local Versus Lateral Inhibitors in Neocortex For studies of long-range connections it is often sufficient to describe the projection targets qualitatively because the different types of putative postsynaptic target neurons are spatially well separated (s. Fig. 8.1 for a sketch illustration). If however the projection targets are to be distinguished at a much smaller spatial scale, more precise geometrical measurements are required. For the identification of inhibitory projection types within and across cortical columns, the axonal projection of interneurons was therefore quantitatively analyzed in the regions outlined by the layers and borders of cortical columns in a sample of 51 interneurons whose somata were located in layers 2/3 (Fig. 8.5). First, the total path length of the axon in layer 1 was measured with the aim of identifying interneurons with a substantial capacity of monosynaptically inhibiting postsynaptic target structures in layer 1, especially the apical tufts of pyramidal cell dendrites (Fig. 8.5a). Second, the total path length of the axon outside of the home column was measured with the aim of identifying interneurons with the capacity of monosynaptically inhibiting postsynaptic neurons in neighboring columns (Fig. 8.5c). Finally, the axonal projections within the home column were further quantified as the total axonal path length projecting to the lower layers 4– 6 (Fig. 8.5c) with the aim of identifying interneurons that have the capacity of monosynaptically inhibiting postsynaptic neurons in regions of the home column below the home layer. A human observer will in some cases easily identify neurons with a large part of their axon in layer 1 or in neighboring columns (s. Fig. 8.5a,c). In other cases, however, such projections will not be as evident. It was therefore indispensable to use quantitative measures to distinguish interneurons with differing axonal projections. The definition of axonal projection types was made using consecutive cluster analyses. First, the projection to layer 1 was identified using the absolute path

0

20

40

60

0

5 10 15 20 total axonal length in L1 [mm]

0

4 linkage distance 0

interneurons

0

4

linkage step

50

D

C

1

0

0

L5B

L5A

L4

L2/3

L1

laterality 1

0

1

2

3

4

5

translaminar

local

lateral

0

6

0

linkage step

40

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

ilnkage distance

linkage distance

% axonal length in L1

Fig. 8.5 Quantification of axonal projections of 51 L2/3 interneurons with reference to cortical columns and layers of the home column using consecutive cluster analyses. a First, the axonal projection to layer 1 (gray shaded area) was quantified and used in a cluster analysis (b) which identified a small subpopulation of interneurons with extensive axonal projection to layer 1. c Quantification of the amount of axon extended to the neighboring columns (left panel, “laterality”) and to the cortical layers below the home layer 2/3 (right panel, “verticality”). The measures of laterality and verticality were then used in a cluster analysis (d) that allowed the identification of three clusters or types of interneurons based on their axonal projection with reference to cortical columns: local, lateral, and translaminar L2/3-to-L4/5 inhibitors. Modified from Helmstaedter et al. (2009b) with permission

B

250μm

linkage distance

A

verticality

8 149

150

M. Helmstaedter and D. Feldmeyer

Fig. 8.6 Types of L2/3 interneurons determined by their axonal projections and innervation domains within and across cortical columns. a Overlay of the reconstructions of dendrites (white) and axons (green) of L2/3 interneurons, sorted by their respective types of axonal projections as quantified using a cluster analysis (see Fig. 8.5b,d). b Average axonal density maps computed from home column-aligned reconstructions as shown in a. c,d Innervation density of L2/3 interneuron axons with putative dendritic targets (L2/3 pyramidal neurons in the home and neighboring columns, c; and spiny neurons in the home column, d). Note that only lateral inhibitors have significant innervation probability of neighboring column pyramidal neurons (c, second panel); only translaminar L2/3-to-L4/5A inhibitors have significant innervation probability of L4 spiny neurons (d, third panel); and only L1 inhibitors have significant innervation probability of L2/3 pyramidal apical tuft dendrites in layer 1 (c, bottom panel). Modified from Helmstaedter et al. (2009b) with permission

length in this layer and its fraction of the total path length extending to layer 1 (Fig. 8.5b, only 4 of the 51 investigated interneurons showed substantial projections to layer 1). Then, the fraction of the axonal path length extending to neighboring columns (combined with the total horizontal extent of the axon) and the fraction of the axonal path length extending below layer 2/3 were used in a cluster analysis

8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

151

which identified three clusters (Figs. 8.5d, 8.6): largely locally projecting interneurons (“local inhibitors” with little projection beyond the home column and beyond the home layer), interneurons with substantial projection to neighboring columns (“lateral inhibitors”), and interneurons with substantial projection to lower layers 4–6 (“translaminar L2/3-to-L4/5 inhibitors”). What is the significance of these distinctions of interneuron types based purely on their axonal projection? As discussed above, the axonal projection can be used as a predictor of synaptic connectivity. The probability of establishing a synaptic contact with the most prevalent types of putative postsynaptic neurons, pyramidal cells in layer 2/3 and spiny neurons in layer 4, was therefore computed for the four types of axonal projection (“innervation density”, Fig. 8.6c,d). The innervation density maps illustrate that the distinction of the four interneuron types can be most easily made based on the neuron’s inhibitory capacity: lateral inhibitors have the capacity of monosynaptically inhibiting L2/3 pyramidal neurons in neighboring columns, which both local inhibitors and translaminar inhibitors are virtually lacking (feedforward inhibition; Fig. 8.6c). In turn, only translaminar L2/3-to-L4/5 inhibitors have a substantial capacity of inhibiting L4 spiny neurons (feedback inhibition; Fig. 8.6d). If cortical columns, defined by thalamic and intracortical innervation domains as described above, have functional relevance, i.e., if pyramidal neurons in layer 2/3 of a given home column represent different aspects of the sensory input (have different receptive fields) than pyramidal neurons in layer 2/3 of a neighboring column, then the distinction of interneurons based on their propensity to inhibit these different postsynaptic neuron populations is bound to be a functionally relevant distinction.

8.5 Subsequent Classification Steps Following the primary classification of interneurons based on their predicted connectivity within and across cortical columns, dendritic and electrical properties can be analyzed and used as subsequent (secondary, tertiary, etc.) classifiers (s. Helmstaedter et al., 2009a, c). It is however critical not to invert the order of these classifications. We found, e.g., that immunohistochemical parameters (expression of parvalbumin, calretinin, somatostatin) do not predict the axonal projection type, nor do dendritic or electrical parameters, or a combination of these (Helmstaedter et al., 2009c). As an analogy, one may consider the classification of automobiles. The car color is a highly distinctive, easy to measure, reproducible parameter of cars. It is however arguably not highly predictive of car function (which would presumably be described by engine type, engine power, car size, etc.). Distinctiveness and reproducibility alone can thus not justify the choice of parameters for classification, the chosen parameter should rather bear superior functional relevance. We therefore maintain that a classification of interneurons based on their axonal projections

152

M. Helmstaedter and D. Feldmeyer

is a primary classification and that additional classifiers should only be used in subsequent classification steps.

8.6 Summary We have summarized how the quantitative description of innervation domains in the primary somatosensory cortex allows the geometrical definition of cortical columns by means of synaptically coupled neuron ensembles. Secondly, we have illustrated how (for interneurons in layer 2/3 of a cortical column) the axonal projections can be used as a primary classifier to define types of interneurons by their capacity to inhibit neighboring columns or not. We have argued that one main conclusion to draw from the analysis of axonal domains is the detection of “zeros” in the connectivity matrix between populations of neurons; i.e., to find types of neurons that cannot monosynaptically inhibit a certain postsynaptic target population, while other types of neurons can. These conclusions hold true even if innervation is not merely locally random. Acknowledgments We are grateful to Bert Sakmann for inspiration, guidance, and continued support of many of the studies summarized in this chapter, and to the Max Planck Society and the Helmholtz Alliance for Systems Biology for funding.

References Ascoli GA et al (2008) Petilla terminology: nomenclature of features of GABAergic interneurons of the cerebral cortex. Nat Rev Neurosci 9:557–568 Bacci A, Huguenard JR, Prince DA (2004) Long-lasting self-inhibition of neocortical interneurons mediated by endocannabinoids. Nature 431:312–316 Bender KJ, Rangel J, Feldman DE (2003) Development of columnar topography in the excitatory layer 4 to layer 2/3 projection in rat barrel cortex. J Neurosci 23:8759–8770 Bloom FE, Ueda T, Battenberg E, Greengard P (1979) Immunocytochemical localization, in synapses, of protein I, an endogenous substrate for protein kinases in mammalian brain. Proc Natl Acad Sci USA 76:5982–5986 Braitenberg V, Schüz A (1998) Cortex: Statistics and Geometry of Neuronal Connectivity. Springer, Berlin Heidelberg Chen BL, Hall DH, Chklovskii DB (2006) Wiring optimization can relate neuronal structure and function. Proc Natl Acad Sci USA 103:4723–4728 Chmielowska J, Carvell GE, Simons DJ (1989) Spatial organization of thalamocortical and corticothalamic projection systems in the rat SmI barrel cortex. J Comp Neurol 285: 325–338 Colonnier M (1964) Experimental degeneration in the cerebral cortex. J Anat 98:47–53 Colonnier M (1968) Synaptic patterns on different cell types in the different laminae of the cat visual cortex. An electron microscope study. Brain Res 9:268–287 Cowan WM, Gottlieb DI, Hendrickson AE, Price JL, Woolsey TA (1972) The autoradiographic demonstration of axonal connections in the central nervous system. Brain Res 37:21–51 Dantzker JL, Callaway EM (2000) Laminar sources of synaptic input to cortical inhibitory interneurons and pyramidal neurons. Nat Neurosci 3:701–707 De Camilli P, Cameron R, Greengard P (1983) Synapsin I (protein I), a nerve terminalspecific phosphoprotein. I. Its general distribution in synapses of the central and peripheral

8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

153

nervous system demonstrated by immunofluorescence in frozen and plastic sections. J Cell Biol 96:1337–1354 Denk W, Horstmann H (2004) Serial block-face scanning electron microscopy to reconstruct threedimensional tissue nanostructure. PLoS Biol 2:e329 Evans DH, Hamlyn LH (1956) A study of silver degeneration methods in the central nervous system. J Anat 90:193–203 Feldmeyer D, Egger V, Lübke J, Sakmann B (1999) Reliable synaptic connections between pairs of excitatory layer 4 neurones within a single ‘barrel’ of developing rat somatosensory cortex. J Physiol 521:169–190 Feldmeyer D, Lübke J, Silver RA, Sakmann B (2002) Synaptic connections between layer 4 spiny neurone-layer 2/3 pyramidal cell pairs in juvenile rat barrel cortex: physiology and anatomy of interlaminar signalling within a cortical column. J Physiol 538:803–822 Gilbert CD, Wiesel TN (1979) Morphology and intracortical projections of functionally characterised neurones in the cat visual cortex. Nature 280:120–125 Glees P (1946) Terminal degeneration within the central nervous system as studied by a new silver method. J Neuropath Exp Neurol 5:54–59 Gray EG (1959) Axo-somatic and axo-dendritic synapses of the cerebral cortex: an electron microscope study. J Anat 93:420–433 Gray EG, Hamlyn LH (1962) Electron microscopy of experimental degeneration in the avian optic tectum. J Anat 96: 305, 309–316 Grutzendler J, Kasthuri N, Gan WB (2002) Long-term dendritic spine stability in the adult cortex. Nature 420:812–816 Gupta A, Wang Y, Markram H (2000) Organizing principles for a diversity of GABAergic interneurons and synapses in the neocortex. Science 287:273–278 Helmstaedter M, Briggman KL, Denk W (2008a) 3D structural imaging of the brain with photons and electrons. Curr Opin Neurobiol 18:633–641 Helmstaedter M, Staiger JF, Sakmann B, Feldmeyer D (2008b) Efficient recruitment of layer 2/3 interneurons by layer 4 input in single columns of rat somatosensory cortex. J Neurosci 28:8273–8284 Helmstaedter M, Sakmann B, Feldmeyer D (2009a) L2/3 interneuron groups defined by multiparameter analysis of axonal projection, dendritic geometry, and electrical excitability. Cereb Cortex 19:951–962 Helmstaedter M, Sakmann B, Feldmeyer D (2009b) Neuronal correlates of local, lateral, and translaminar inhibition with reference to cortical columns. Cereb Cortex 19: 926–937 Helmstaedter M, Sakmann B, Feldmeyer D (2009c) The relation between dendritic geometry, electrical excitability, and axonal projections of L2/3 interneurons in rat barrel cortex. Cereb Cortex 19:938–950 Herkenham M (1980) Laminar organization of thalamic projections to the rat neocortex. Science 207:532–535 Hoff EC (1932) Central nerve T’erminals in the mammalian spinal cord and their examination by experimental degeneration. Proc R Soc Lond B Biol Sci 111:175–188 Hubel DH, Wiesel TN (1968) Receptive fields and functional architecture of monkey striate cortex. J Physiol 195:215–243 Hubel DH, Wiesel TN (1969) Anatomical demonstration of columns in the monkey striate cortex. Nature 221:747–750 Jones EG (2007) Neuroanatomy: Cajal and after Cajal. Brain Res Rev 55:248–255 Jones EG, Powell TP (1970) An electron microscopic study of the laminar pattern and mode of termination of afferent fibre pathways in the somatic sensory cortex of the cat. Philos Trans R Soc Lond B Biol Sci 257:45–62 Killackey HP (1973) Anatomical evidence for cortical subdivisions based on vertically discrete thalamic projections from the ventral posterior nucleus to cortical barrels in the rat. Brain Res 51:326–331

154

M. Helmstaedter and D. Feldmeyer

Lasek R, Joseph BS, Whitlock DG (1968) Evaluation of a radioautographic neuroanatomical tracing method. Brain Res 8:319–336 LeVay S, Gilbert CD (1976) Laminar patterns of geniculocortical projection in the cat. Brain Res 113:1–19 Lichtman JW, Smith SJ (2008) Seeing circuits assemble. Neuron 60:441–448 Lorente de No R (1922) La corteza cerebral de ratón Trabajos del Laboratorio de Investigaciones Biológicas de la Universidad de Madrid 20:41–78 Lorente de No R (1938) Cerebral cortex: architecture, intracortical connections, motor projections (ch. 15). In: Fulton JF (ed) Physiology of the nervous system. Oxford University Press, Oxford, pp 288–313 Lorente de No R (1992) The cerebral cortex of the mouse (A first contribution – the “acoustic” cortex). (Trans: Fairén A, Regidor J, Kruger L). Somat Mot Res 9:3–36 Lu SM, Lin RC (1993) Thalamic afferents of the rat barrel cortex: a light- and electron-microscopic study using Phaseolus vulgaris leucoagglutinin as an anterograde tracer. Somatosens Mot Res 10:1–16 Lübke J, Egger V, Sakmann B, Feldmeyer D (2000) Columnar organization of dendrites and axons of single and synaptically coupled excitatory spiny neurons in layer 4 of the rat barrel cortex. J Neurosci 20:5300–5311 Lübke J, Roth A, Feldmeyer D, Sakmann B (2003) Morphometric analysis of the columnar innervation domain of neurons connecting layer 4 and layer 2/3 of juvenile rat barrel cortex. Cereb Cortex 13:1051–1063 Magrassi L, Purves D, Lichtman JW (1987) Fluorescent probes that stain living nerve terminals. J Neurosci 7:1207–1214 Mainen ZF, Sejnowski TJ (1996) Influence of dendritic structure on firing pattern in model neocortical neurons. Nature 382:363–366 Markram H, Lübke J, Frotscher M, Roth A, Sakmann B (1997) Physiology and anatomy of synaptic connections between thick tufted pyramidal neurones in the developing rat neocortex. J Physiol 500:409–440 Micheva KD, Smith SJ (2007) Array tomography: a new tool for imaging the molecular architecture and ultrastructure of neural circuits. Neuron 55:25–36 Mountcastle VB (1957) Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J Neurophysiol 20:408–434 Mountcastle VB (1997) The columnar organization of the neocortex. Brain 120 (Pt 4):701–722 Peters A (1979) Thalamic input to the cerebral cortex. Trends Neurosci 2:183–185 Peters A (2007) Golgi, Cajal, and the fine structure of the nervous system. Brain Res Rev 55: 256–263 Ramón y Cajal S (1904) Textura del sistema nervioso del hombre y de los vertebrados. Imprenta N. Moya, Madrid Ramón y Cajal S (1995) Histology of the nervous system. Oxford University Press, New York, Oxford Schaefer AT, Larkum ME, Sakmann B, Roth A (2003) Coincidence detection in pyramidal neurons is tuned by their dendritic branching pattern. J Neurophysiol 89:3143–3154 Shepherd GM, Stepanyants A, Bureau I, Chklovskii D, Svoboda K (2005) Geometric and functional organization of cortical circuits. Nat Neurosci 8:782–790 Sholl DA (1953) Dendritic organization in the neurons of the visual and motor cortices of the cat. J Anat 87:387–406 Sholl DA (1955) The organization of the visual cortex in the cat. J Anat 89:33–46 Silver RA, Lübke J, Sakmann B, Feldmeyer D (2003) High-probability uniquantal transmission at excitatory synapses in barrel cortex. Science 302:1981–1984 Stepanyants A, Chklovskii DB (2005) Neurogeometry and potential synaptic connectivity. Trends Neurosci 28:387–394 Stepanyants A, Hirsch JA, Martinez LM, Kisvarday ZF, Ferecsko AS, Chklovskii DB (2008) Local potential connectivity in cat primary visual cortex. Cereb Cortex 18:13–28

8

Axons Predict Neuronal Connectivity Within and Between Cortical Columns

155

Taylor AC, Weiss P (1965) Demonstration of axonal flow by the movement of tritium-labeled protein in mature optic nerve fibers. Proc Natl Acad Sci USA 54:1521–1527 Trachtenberg JT, Chen BE, Knott GW, Feng G, Sanes JR, Welker E, Svoboda K (2002) Longterm in vivo imaging of experience-dependent synaptic plasticity in adult cortex. Nature 420: 788–794 Uchizono K (1965) Characteristics of excitatory and inhibitory synapses in the central nervous system of the cat. Nature 207:642–643 Walker AE (1936) An experimental study of the thalamocortical projection of the macaque monkey. J Comp Neurol 64:1–39 White EL (1979) Thalamocortical synaptic relations: a review with emphasis on the projections of specific thalamic nuclei to the primary sensory areas of the neocortex. Brain Res 180:275–311 White EL (2007) Reflections on the specificity of synaptic connections. Brain Res Rev 55:422–429 White EL, Keller A (1989) Cortical circuits: synaptic organization of the cerebral cortex: structure, function, and theory. Birkhäuser, Boston White JG, Southgate E, Thomson JN, Brenner S (1986) The structure of the nervous system of the nematode caenorhabditis elegans. Philos Trans R Soc Lond B Biol Sci 314:1–340 Wise SP, Jones EG (1978) Developmental studies of thalamocortical and commissural connections in the rat somatic sensory cortex. J Comp Neurol 178:187–208 Woolsey TA, Van der Loos H (1970) The structural organization of layer IV in the somatosensory region (SI) of mouse cerebral cortex. The description of a cortical field composed of discrete cytoarchitectonic units. Brain Res 17:205–242.

Chapter 9

The Axon of Excitatory Neurons in the Neocortex: Projection Patterns and Target Specificity Joachim H.R. Lübke and Dirk Feldmeyer

9.1 Introduction The neocortex of all mammals including humans is a highly hierarchical structure with defined sensory and motor areas representing the sensory periphery. Higher level areas such as the temporal, prefrontal and frontal cortices (so-called associational cortices) without a defined sensory/motor function were added during mammalian evolution to allow a higher diversity of the neocortex’ computational properties. The neocortex is a multi-layered structure, of – in general – six horizontally oriented layers between the pial surface and the white matter. These layers are generated in an inside-first, outside-last fashion by vertical migration of the precursors of future excitatory neurons along radial glia (Luskin and Shatz, 1985; Frotscher, 1998). In addition, sensory cortices have been demonstrated to be organized in functional, vertically oriented units, the so-called cortical columns (Mountcastle, 1957; Hubel and Wiesel, 1959). Cortical layers and columns are both determining factors for the axonal projection pattern of neocortical excitatory neurons as outlined below. All cortices regardless of their different structural and functional properties consist of two major classes of neurons: principal neurons, i.e. pyramidal cells and so-called spiny stellate neurons that are excitatory and a highly heterogeneous population of GABAergic (y-amino butyric acid releasing) interneurons; excitatory neurons constitute 80–85% of the total neuronal population while GABAergic interneurons contribute to 15–20% (Micheva and Beaulieu, 1995; Somogyi et al., 1998). GABAergic interneurons have intrinsically projecting axons that are restricted to defined cortical regions; some of their axons project only locally and are largely confined to a single cortical layer and/or cortical column (Kawaguchi and Kubota, 1997; Somogyi et al., 1998; Cauli et al., 2000; Gupta et al., 2000; Porter et al., 2001; Karube et al., 2004; Ascoli et al., 2008; Helmstaedter et al., 2009a, b, c). J.H.R. Lübke (B) Research Centre Jülich GmbH, Institute of Neuroscience and Medicine, INM-2, Jülich, Germany; Department of Psychiatry and Psychotherapy, RWTH Aachen University, Aachen, Germany e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_9, 

157

158

J.H.R. Lübke and D. Feldmeyer

In contrast, principal neurons with the notable exception of spiny stellate and star pyramidal neurons in cortical layer 4 have two axonal subdomains: a vertical domain that projects throughout the cortical column, to the contralateral neocortex via the corpus callosum and to various subcortical brain regions (e.g. thalamus, striatum, pontine nuclei, spinal cord). The second axonal domain projects horizontally over a wide range, both within a given cortical region and between cortical areas within the same sensory system (e.g. from the somatosensory S1 to the S2 cortex or the visual V1 to V2–V5 cortices) or even between different cortical areas (e.g. from the somatosensory S1 to the motor cortex M1). It was long thought that pyramidal cells represent a rather stereotyped class of neurons both with respect to their dendritic configuration and their axonal arborization. There is, however, growing evidence that pyramidal cells not only in different cortical layers but also within a defined layer vary substantially in their axonal projection patterns and cellular as well as subcellular input and target specificity. This chapter will summarizes the recent literature to this point as well as the implication for neocortical connectivity. The so-called barrel field in the somatosensory cortex of rodents (Woolsey and Van der Loos, 1970) is remarkable with respect to the clearly visible somatotopic cortical representation of the sensory periphery. Here, each whisker hair and its arrangement on the rodent’s snout is represented topographically exact in form of a barrel in layer 4 of the somatosensory cortex. These barrels and their extension into the other cortical layers (‘barrel column’) are thought to be the structural correlates of cortical columns. In other sensory cortices, such an unequivocal structural correlate of a cortical column is not found. Since we would like to describe the axon in relationship to the cortical column this chapter will concentrate on the axonal domains and their target specificity in the somatosensory cortex of rodents, in particular the barrel field. For reasons of comparison we will, however, oppose the axonal morphology of excitatory somatosensory neurons with those of other cortical areas, in particular the visual cortex.

9.2 Highly Variable Axonal Domains of Neocortical Excitatory Neurons As already mentioned above principal neurons of the neocortex fall into two different classes, namely pyramidal cells in layer 2/3, 5 and 6 (for review see DeFelipe and Farinas, 1992; Thomson and Bannister, 2003) and spiny stellate neurons/star pyramidal cells that are exclusively found in layer 4 of granular, i.e. sensory cortices. Layer 4 neurons are regarded as non-projecting, local neurons and as such can be considered as excitatory, glutamatergic neocortical ‘interneurons’. It has long been thought that pyramidal cells of the neocortex constitute a rather homogeneous population of neurons based on their dendritic configuration, as revealed by Golgi studies (Ramón y Cajal, 1904; Lorente de Nó, 1949; Winkelmann et al., 1975; Feldman, 1984; Valverde, 1986; DeFelipe and Farinas, 1992), in

9

The Axon of Excitatory Neurons in the Neocortex

159

particular the characteristic polarity of the neuron with a prominent apical dendrite and a more or less symmetrical basal dendritic tree. However, in vivo and in vitro studies using intracellular dye injection and more recently genetic approaches have demonstrated a huge heterogeneity of the axonal projection pattern of pyramidal cells in different cortical layers closely related to their synaptic input and output relationships.

9.3 Layer 2/3 According to the ‘inside-first outside-last’ principle of corticogenesis (MarínPadilla, 1971; Luskin and Shatz, 1985) layer 2/3 is the last cortical layer to be generated. From a historical point of view, layer 2/3 has been regarded as two separate layers, but the usual nomenclature to date subsumes these two layers into one layer 2/3. This layer is by far the largest layer of the neocortex, in particular in sensory cortices. With respect to their somatodendritic configuration layer 2/3 pyramidal cells form the most heterogeneous population of pyramidal cells when compared to those in other cortical layers. Pyramidal cells in deep layer 2/3 (i.e. those located just above layer 4) have thick apical dendrites that terminate in a small tuft the field span of which is smaller than that of the neuron’s basal dendritic tree (Larkman and Mason, 1990; Feldmeyer et al., 2006). These apical dendrites form cluster-like arrangements that have also been observed for layer 5 pyramidal cells. These dendritic ‘bundles’, also called ‘dendrons’ have been proposed to contribute to so-called minicolumns that comprise small clusters of neurons in the cortical column (Fleischhauer et al., 1972; Peters and Walsh, 1972; Winkelmann et al., 1975; Escobar et al., 1986; DeFelipe, 2005). Pyramidal neurons located in upper layer 2/3 (i.e. those below layer 1) have only short, sometimes twinned apical dendrites with large and profusely branching apical tufts. In contrast to pyramidal cells in the lower portion of layer 2/3, the field span of these tufts exceeds that of the neuron’s basal dendritic tree (Larkman and Mason, 1990; Feldmeyer et al., 2006). Pyramidal cells in the upper portion of layer 2/3 do not form apical dendritic bundles. Neurons close to the pial surface show an aberrant morphology with an oblique to horizontally oriented ‘apical’ dendrite (Miller, 1988; Lübke et al., 2003). Furthermore, in visual cortex a very small fraction (ca. 5%) of pyramidal cell with abpial (i.e. inverted) ‘apical’ dendrites has been identified (Miller, 1988; Bueno-Lopez et al., 1991). These inverted pyramidal neurons have ascending main axons that ramify in a local plexus in layer 1 and 2/3. In addition, tracer injection studies demonstrated that these neurons have long-range intra- and inter-areal axonal collaterals (Bueno-Lopez et al., 1991; Reblet et al., 1996). All pyramidal cells in layer 2/3 project with their main axon to the contralateral hemisphere via the corpus callosum and are regarded as exclusively intracortically projecting neurons (Wise and Jones, 1976). Layer 2/3 pyramidal cells display two distinct axonal domains, one in layer 2/3 itself and one in layer 5 with a few collaterals in layer 6, as demonstrated in somatosensory (Fig. 9.1) and visual cortex. Few, vertically oriented axonal collaterals (about 10% of the total axonal length)

160

J.H.R. Lübke and D. Feldmeyer

Fig. 9.1 Layer 2/3 pyramidal cells. Axonal (blue) and dendritic configuration (red) of two layer 2/3 pyramidal neurons labelled in vivo. Panels a and c represent views from the cortical surface, panels c and d en face views showing the cortical layering. Note that the pyramidal cell shown in panels a, b has a broad axonal domain that spans the entire D row of the barrel field and projects even into adjacent cortical areas. The layer 2/3 pyramidal cell shown in panel c, d has a more circular axonal domain with a higher density of collaterals in the home column. Both pyramidal cells shown here show little if any collateralization at the level of cortical layer 4. Modified from Bruno et al. (2009) with permission

are found in layer 4 suggesting that the innervation of layer 4 neurons by these axons is at least infrequent (Gilbert and Wiesel, 1979; Martin and Whitteridge, 1984; Lübke et al., 2003; Feldmeyer et al., 2006; Larsen and Callaway, 2006; Bruno et al., 2009). During development, the axonal domain in layer 5 and 6 is generated first in accordance with the ‘inside-first outside-last’ gradient of corticogenesis. In mouse barrel cortex, this domain is already present in the first postnatal week, while

9

The Axon of Excitatory Neurons in the Neocortex

161

the second axonal domain in layer 2/3 develops only from postnatal day 10 onwards and increases gradually in size and density with ongoing maturation (Larsen and Callaway, 2006). In layer 2/3, in which about two-third of the axon resides, the axonal arborization differs markedly for deep and superficial layer 2/3 pyramidal cells. Superficial pyramidal cell had a main axon that gives rise to both a dense local plexus of axonal collaterals within the home column and a few long-range collaterals spanning several barrel columns and projecting into adjacent cortical areas S2 and M1 cortex (Porter and Sakamoto, 1988; Porter et al., 1990; Hoffer et al., 2003). Deep layer 2/3 pyramidal cells showed several obliquely ascending collaterals in layer 2/3 but their local plexus was less dense than that of superficial cells. Indeed, superficial layer 2/3 pyramidal cells had more extensive axonal arbors than deep ones, in particular in layer 2/3 (Feldmeyer et al., 2006; Bruno et al., 2009). Independent of the relative location within layer 2/3 a population of locally projecting (i.e. only into the neighbouring barrels) and a population of long-range projecting layer 2/3 pyramidal cells appear to exist. The long-range collaterals can project as far as the entire barrel field (Fig. 9.1a) and show periodically spaced clusters of short axonal collaterals. Cells with a local axon in layer 2/3 exhibit also locally projecting collaterals in the infragranular layers 5 and 6, while long-range projecting neurons did so in layers 2/3 as well as 5 and 6 (Bruno et al., 2009) (cf. Fig. 9.1). The described heterogeneity in the axonal domain of layer 2/3 pyramidal cells suggests different functions of the individual population in cortical signal processing. It is conceivable that the morphologically different neurons are part of functionally different neocortical microcircuits as has been suggested for upper and lower layer 2/3 pyramidal cells in mouse barrel cortex (Bureau et al., 2006).

9.4 Layer 4 In sensory cortices, layer 4 is the main recipient layer for thalamocortical afferents arising from the respective thalamic relay nuclei. Neurons within layer 4 represent therefore the first level of sensory signal processing and transduction in the neocortex. The somatodendritic configuration and axonal morphology of excitatory layer 4 neurons have been described in great detail for the mammalian visual and somatosensory cortex but less so for the auditory and other cortices. These excitatory neurons in layer 4 have been subdivided into two classes, the ‘classical’ spiny stellate neuron (Fig. 9.2a), which represents the majority of excitatory layer 4 neurons and the so-called star pyramidal cell (Fig. 9.2b).

9.4.1 Spiny Stellate and Star Pyramidal Neurons The first in-depth description of spiny stellate cells came from in vivo and in vitro microelectrode recordings combined with intracellular horseradish peroxidase

162

J.H.R. Lübke and D. Feldmeyer

Fig. 9.2 Barrel-related layer 4 spiny neurons. Camera lucida reconstruction of a spiny stellate cell (a) and a star pyramidal cell (b) in layer 4. The somatodendritic configuration is shown in red; the axonal collaterals are shown in blue. The outline of the barrel is shaded in light grey. Note that in particular the spiny stellate axon is largely confined to the barrel column. (a, inset) High-power micrograph of the neuron shown in (a). In this image cytochrome oxidase staining was used to reveal the outlines of the layer 4 barrel in which the biocytin-labelled neuron is located

(HRP) fillings in the visual cortex (Gilbert and Wiesel, 1979; Martin and Whitteridge, 1984; Muly and Fitzpatrick, 1992; Hirsch, 1995). Here, layer 4 is subdivided into three sublaminae because of the different termination pattern of thalamocortical afferents from the lateral geniculate nucleus of the thalamus (LGN). The axons of spiny stellate neurons in layer 4 of the visual cortex project predominantly to layers 4 and 2/3. However, these neurons display a large diversity in their axonal projection and target specificity, that is dependent on their position within layer 4. While some layer 4 axons are short and local, targeting specific sublaminae in layers 4 and 2/3 (Martin and Whitteridge, 1984; Muly and Fitzpatrick, 1992; Hirsch, 1995), others show an extensive and widespread and patchy distribution in layer 2/3 (Gilbert and Wiesel, 1979; Martin and Whitteridge, 1984; Hirsch et al.,

9

The Axon of Excitatory Neurons in the Neocortex

163

2002). In addition, the axon proceeds out of the neocortex into the white matter sending a few collaterals into layers 5 and 6 on its downwards course. An electron microscopic (EM) study showed that these axons innervate spines as the most frequent synaptic target (74%); dendritic shafts formed the remainder (26%). On the basis of ultrastructural characteristics, 8% of the target dendrites were characterized as originating from smooth GABAergic interneurons. Thus the major output of spiny stellate neurons is to other spiny neurons, probably pyramidal neurons in layer 3 and 4 (Anderson et al., 1994). Spiny stellate cells in layer 4 of the barrel field in the somatosensory cortex of rodents are strikingly different from their counterparts in the visual cortex, both with respect to their somatodendritic configuration and their axonal projection pattern. The somata of these neurons are generally located in clusters near the barrel walls, their dendrites are largely confined to a single barrel in layer 4 and display often an asymmetric orientation towards the barrel centre (Harris and Woolsey, 1981; Lübke et al., 2000a; Staiger et al., 2004; Egger et al., 2008). This is in marked contrast to the visual cortex, where spiny stellate cells were distributed in different sublaminae of layer 4 and showed a radial symmetry of their dendrites. The axonal domain of spiny stellate cells shows a similar asymmetric orientation towards the centre of the barrel column and is largely confined to the column in which it resides (Harris and Woolsey, 1983; Lübke et al., 2000a). The highest density of axonal collaterals is in layers 4 and 2/3, where the predominant target structures of these neurons are located (Figs. 9.2 and 9.3). These are mainly secondary and tertiary basal dendrites of layer 4 spiny neurons and of layer 2/3 pyramidal cells (Feldmeyer et al., 1999, 2002). Although the vast majority of spiny stellate neurons show a largely columnar axonal domain, some rare exception have been reported in which axonal collaterals projects into adjacent barrel columns where they may branch profusely. These interbarrel projections appeared to target the same region in the home and the neighbouring barrel column (Egger et al., 2008). The spatial extent and density of the axonal domain of spiny stellate cells is strongly regulated during development. In immature barrel cortex, the axon is more widespread, spanning up to several barrel columns but undergoes considerable pruning with maturation until it is mainly confined to a single barrel column (Bender et al., 2003).

9.4.2 Star Pyramidal Cells The second class of excitatory layer 4 neurons is the so-called star pyramidal cell with larger calibre vertical dendrites that terminate in layer 2/3. According to a recent study (Egger et al., 2008), star pyramidal and spiny stellate cells can be distinguished on the basis of their dendritic configuration with the most distinguishing features being the calibre of a vertically oriented dendrite, the horizontal dendritic field span and the maximal dendrite-free angle (which is a measure for

164

J.H.R. Lübke and D. Feldmeyer

the asymmetric orientation of dendrites). Furthermore, they do not exhibit a clustered arrangement and their somata are mainly found near the centre of a barrel. The axon of star pyramidal cells is also confined to the barrel column, especially at the level of layer 4. It does not show, however, the typical asymmetric orientation of the axonal domain of spiny stellate cells (Fig. 9.2b) and seems to project more diffusively within a barrel column, i.e. over wider area in layers 4 and in particular in layer 2/3. The star pyramidal axon was seen to extend well into neighbouring barrel columns (Lübke et al., 2000a, 2003; Staiger et al., 2004; Egger et al., 2008), an observation that was made more frequently for this cell type than for spiny stellate cells (see Fig. 9.3). Both spiny stellate and star pyramidal cell in layer 4 of the somatosensory cortex are clearly distinguishable from pyramidal cells in other cortical layers with respect to the columnar confinement of their axonal domain (Egger et al., 2008). This feature was also found for several classes of stellate cells in layer 4C of monkey visual cortex but to a lesser degree (Katz et al., 1989). In contrast to the somatosensory cortex where spiny stellate and star pyramidal neurons are, with a

Fig. 9.3 Layer 4 spiny neurons projecting outside barrel borders. (a, c) Layer 4 Spiny stellate cells with interbarrel axonal projections. Dendritic tree red, axon blue. schematic barrels are indicated in light grey. The scale bar applies to all reconstructions. (b, d) Projections of the above neurons into the tangential plane with the axon collaterals reduced to layer 4. The projections to the adjacent barrel(s) are fairly dense. (a, b) A reconstruction from a thalamocortical slice of a twelve day old rat, (c, d) from in vivo fillings of a spiny stellate cell of a 30 day old animal. Modified after Egger et al. (2008) with permission

9

The Axon of Excitatory Neurons in the Neocortex

165

few exceptions, rather uniform while their counterparts in the visual cortex vary substantially (Martin and Whitteridge, 1984; Lund et al., 1995; Fitzpatrick, 1996; Yabuta and Callaway, 1998). The dense axonal arborization of spiny stellate and star pyramidal cells is the basis of the high connectivity within layer 4 and between layer 4 and 2/3 (Feldmeyer et al., 1999, 2002; Lübke et al., 2003; Lefort et al., 2009). The bouton density along the entire axon is high, with no marked difference between proximal and distal axonal segments suggesting a high columnar connectivity of spiny stellate cells. However, the axons of spiny stellate cells are more ‘dense’ than those of star pyramidal cells suggesting that both cell types perform different tasks in somatosensory signal processing. It has been demonstrated that excitatory inputs to spiny stellate cells originate almost exclusively in the same barrel in layer 4 while star pyramidal cells receive also input from adjacent barrels (Schubert et al., 2003). Moreover, for primate visual cortex, spiny stellate and star pyramidal cells have been shown to receive different inputs from individual sublaminae of layer 4 (in particular layer 4C) indicating that the two cell types are integrated into functionally specialized subsystems (Yabuta et al., 2001).

9.5 Layer 5 Among the population of neocortical pyramidal cells, those in layer 5 have received the highest attention and are therefore structurally and functional extremely well characterized. Layer 5 pyramidal cells are among the largest neurons in the central nervous system of mammals as exemplified by the giant ‘motoneurons’ in the primate primary motor cortex, the so-called Betz pyramidal cells, that project to the spinal cord (Betz, 1874). A common feature of all layer 5 pyramidal cells is the presence of a prominent apical dendrite that either terminates with a terminal tuft in layer 1 or remains untufted (Larkman and Mason, 1990; Larsen and Callaway, 2006; Larsen et al., 2007). Tufted pyramidal neurons can be further subdivided in those with a small and a thick terminal tuft. In addition to this classification based on the dendritic configuration, layer 5 pyramidal cells have been subdivided into those projecting to the contralateral hemisphere via the corpus callosum and those projecting to various subcortical targets (DeFelipe and Farinas, 1992). Moreover, recent publications further subdivided layer 5 pyramidal cells into short and tall-simple corticocortical and tall-tufted corticotectal neurons as well as tall-tufted corticothalamic neurons (Larsen and Callaway, 2006; Larsen et al., 2007). The somata of tall-simple, small-tufted pyramidal cells are almost exclusively found in sublamina 5A, the short corticocortical pyramidal cells are found in deep sublamina 5B; the corticotectal and corticothalamic neurons are distributed throughout the entire sublamina 5B.

9.5.1 Sublamina 5A In the somatosensory cortex small- and thick-tufted pyramidal cells form two structurally and functionally distinct sublaminae, termed layer 5A and 5B, respectively

166

J.H.R. Lübke and D. Feldmeyer

(Ito, 1992; Hoeflinger et al., 1995; Gottlieb and Keller, 1997). Layer 5A pyramidal cells are characterized by a small and slender apical tuft with only few short branches in layer 1. The axonal configuration of these neurons is highly distinct from that of layer 2/3 and layer 5B pyramidal neurons. They exhibit a vertically projecting stem axon that gives rise to numerous ascending collaterals at the level of layer 5A, a substantial fraction of which runs in parallel to the apical dendrites. When these collaterals reach the border between layer 2/3 and 1 they fan out and form a dense horizontal plexus that spans several cortical columns; some of these collaterals form periodically spaced clusters (Manns et al., 2004; Feldmeyer et al., 2005; Shepherd et al., 2005; Larsen and Callaway, 2006; Larsen et al., 2007; Frick et al., 2008). The descending axonal collaterals form a similarly dense network in layer 5A and here in particular in the barrel column in which the L5 pyramidal cell is located (the home column). In layer 5B and 6, the stem (main) axon gives rise to a few long-range, horizontal collaterals forming a relatively sparse axonal network (Manns et al., 2004; Frick et al., 2008). The axonal projection pattern of layer 5A pyramidal cells suggest three distinct target regions. Within layer 5A itself, the axon contacts predominantly basal dendrites of other layer 5A pyramidal cells mainly in the same column; the connectivity in this region is comparatively high (∼20%) (Schubert et al., 2006; Frick et al., 2008; Lefort et al., 2009). In addition, layer 5A pyramidal cells are also connected to layer 5B and 6 pyramidal cells (Lefort et al., 2009) and are likely to establish synaptic contacts on basal dendrites. These connections may well be intra- as well as transcolumnar. The dense axonal plexus at the border between layer 2/3 and layer 1 suggests that layer 5A pyramidal neurons either target specifically pyramidal cells in upper layer 2/3 (layer 2) or terminate in the region near the terminal tuft of pyramidal cells in layers 2/3, 5A and 5B (see Fig. 9.4a). This region has been identified as the so-called Ca2+ spike initiation zone; it interacts with the Na+ action potential initiation zone in the axon of the same neuron and may be responsible for regenerative potentials critical for integrating and amplifying sensory and modulatory inputs (Larkum et al., 1999; Larkum and Zhu, 2002). It is tempting to speculate that layer 5A pyramidal cells are involved in the activation of Ca2+ spikes in pyramidal cells across several cortical columns and thereby integrate the activity of these neurons.

9.5.2 Sublamina 5B In the somatosensory cortex, thick-tufted pyramidal cells with a prominent apical dendrite are found exclusively in layer 5B while this cell type is interspersed with small-tufted and untufted pyramidal cells throughout the entire volume of layer 5 in the visual cortex (Gabbott et al., 1987). Like layer 5A pyramidal cells, those in layer 5B display a vertical and horizontal axonal domain. In contrast to layer 5A pyramidal cell, the axon of layer 5B pyramidal cell does not form a dense plexus in the home layer (i.e. layer 5B) but has a very prominent, long-range horizontal axonal

9

The Axon of Excitatory Neurons in the Neocortex

167

Fig. 9.4 Layer 5A and 5B pyramidal cells. a Reconstruction of a layer 5A pyramidal neuron with a small apical tuft. Note the dense local axonal domain and the horizontal arbors at the border of layers 1 and 2/3. b Two layer 5B pyramidal cells with dense apical dendritic tufts and long-range horizontal collaterals and a few ascending collaterals. Somatodendritic domain, red; axonal arbor, blue; cortical layering is indicated on the left

domain spanning several cortical columns with single axon collaterals seen to span over ∼4 mm in layer 5A, 5B and 6 (Markram et al., 1997, 1998; Manns et al., 2004; Larsen and Callaway, 2006; Larsen et al., 2007). The second axonal domain consists of an ascending axonal domain with several termination zones in layers 4, 2/3 and 1; these collaterals originate as second-order collaterals from horizontal projections into several adjacent columns (Fig. 9.4b, c). The bouton density along the entire axon is high, with no marked difference between proximal and distal axonal segments suggesting a high columnar and transcolumnar connectivity (Markram, 1997; Markram et al., 1997). Paired recordings from synaptically coupled layer 5B pyramidal cells have shown that synapses are predominantly established with second and third order basal dendrites and to a lesser extent with first- and second-order apical oblique dendrites within the same barrel column (Markram, 1997; Markram et al., 1997). Within the home column, the connectivity between layer 5B pyramidal cells is significantly higher than that of layer 5A pyramidal cell pairs. The axonal projection pattern suggests additional targets such as pyramidal neurons in layer 2/3, 5A and 6 which is in accordance with glutamate uncaging or paired recording studies (Staiger et al., 2004; Shepherd and Svoboda, 2005; Schubert et al., 2006; Lefort et al., 2009), however, the connectivity seems to be low. It appears that layer 5B pyramidal cells have significantly longer horizontal projections compared to layer 5A pyramidal cells. This suggests a higher integrative capacity through the synchronization of large ensembles of layer 5A and 5B pyramidal cells via these long-range horizontal collaterals and layer 2/3 pyramidal cells by the transcolumnar ascending collaterals.

168

J.H.R. Lübke and D. Feldmeyer

In the visual cortex, one striking feature of the horizontal collaterals is a patchy, clustered distribution of the layer 5B axon (Gilbert and Wiesel, 1983; Gabbott et al., 1987), which are thought to connect distant but functionally similar ocular dominance columns (Ts’o et al., 1986). While such far-reaching collaterals have not been identified for layer 5B pyramidal cells in the somatosensory cortex, it is likely that they exist here as well and serve to synchronize activity over entire barrel rows.

9.6 Layer 6 Layer 6 is present in all neocortical areas, however, its thickness is quite variable with more or less distinct sublaminae 6A and 6B. In sensory cortices, layer 6 receives thalamocortical afferents arising from specific thalamic relay nuclei. In turn, a fraction of layer 6 neurons project back to these nuclei, thus forming an excitatory feedback loop. Corticothalamically projecting layer 6 neurons are therefore involved in a direct control of their excitatory input, in contrast to layer 4 neurons. The structure of layer 6 reflects its dual origin with layer 6A as a clear derivative of the cortical plate whereas layer 6B is a heterogeneous layer with neurons originating from the subplate or the primordial plexiform layer which may develop into the white matter (Marín-Padilla, 1978). This is reflected in the markedly different sublamina size with sublamina 6A being fivefold thicker than 6B (Zhang and Deschênes, 1997). Sublamina 6A contains pyramidal cells with a vertically oriented, untufted apical dendrite that terminates in layers 3 and 4. In contrast, sublamina 6B neurons have a highly diverse dendritic configuration, with a substantial proportion of neurons with an inverted or a horizontally oriented dendritic arborization (Tömböl et al., 1975; Tömböl, 1984; Miller, 1988; Bueno-Lopez et al., 1991; Chen et al., 2009).

9.6.1 Sublamina 6A In sublamina 6A, two distinct populations of pyramidal neurons have been identified on the basis of their axonal projection pattern: the so-called corticothalamic neurons and the corticocortical projecting neurons (Zhang and Deschênes, 1997; Kumar and Ohana, 2008). The corticothalamic neurons can be further subdivided according to their intracortical projection pattern and termination zones (Zhang and Deschênes, 1997, 1998). The first subpopulation of sublamina 6A pyramidal displays a largely columnar organization of its axon with vertically oriented axonal collaterals that terminate at the layer 4/layer 2/3 border, occasionally giving rise to numerous short collaterals in layer 4 (Fig. 9.5a). These neurons project exclusively to the ventroposterior medial (VPm) nucleus of the thalamus. The somata of these VPm-projecting neurons are located in the upper to middle portion of sublamina 6A. Possible targets of these neurons are other layer 6 neurons, layer 5 and layer 4 neurons. Extracellular stimulation in layer 6A has demonstrated the existence of

9

The Axon of Excitatory Neurons in the Neocortex

169

Fig. 9.5 Axonal projection patterns of different layer 6A pyramidal neurons in the somatosensory cortex. a Layer 6A pyramidal neuron with a corticothalamic prjection to the VPm nucleus of the thalamus. b Layer 6A pyramidal neuron that projects both the VPm and the POm nucleus of the thalamus. c Layer 6A pyramidal neuron that projects corticocortically to the secondary somatosensory cortex. All layer 6A pyramidal cells have apical dendrites terminating in middle cortical layers. Somatodendritic domain, red; axonal arbor, blue. Modified after Zhang and Deschênes (1997) with permission

170

J.H.R. Lübke and D. Feldmeyer

an excitatory synaptic connection between layer 6A and layer 4 spiny stellate cells (Lee and Sherman, 2008). The axons of the second subpopulation have their major termination zone within sublamina 5A and give here rise to periodically spaced axonal clusters that can be either columnar or span several cortical (barrel) columns (Fig. 9.5b). These sublamina 6A pyramidal cells project both to the VPm and posterior medial (POm) nuclei. The cell bodies of neurons are found predominantly in middle to lower portion of sublamina 6A. Possible target neurons for these VPm and POm projecting cells are other layer 6A neurons but mainly layer 5A neurons, presumably sublamina 5A pyramidal cells that receive thalamocortical input from the POm nucleus (Bureau et al., 2006). Thus, these layer 6A pyramidal cells together with the layer 5A pyramidal cells may be involved in a thalamocortical–corticothalamic feedback loop. The third, apparently rare subpopulation consists of pyramidal neurons with more local axonal collaterals with a dense plexus in layer 6A itself and short oblique collaterals terminating in layer 5B. The subcortical targets of these neurons are heterogeneous. They may project to either the ventromedial nucleus of the thalamus or to the POm. Intracortically, their main targets are located in sublamina 6A. In contrast to corticothalamic cells, corticocortical pyramidal cells are distributed throughout the entire volume of sublamina 6A and are intermingled with the former (Zhang and Deschênes, 1997; Kumar and Ohana, 2008). The axon of corticocortical pyramidal cells forms a dense plexus in layers 5 and 6A and sends a few collaterals in sublamina 6B (Fig. 9.5c). A characteristic feature is the existence of a longrange horizontal projection to the second somatosensory (S2) cortex and/or to the motor cortex that is located exclusively in layer 6. Within S2, these axonal collaterals form periodically spaced small cluster-like arrangements. In addition, the axon sends a single fibre to the S1 cortex of the contralateral hemisphere via the corpus callosum. Corticocortical-projecting pyramidal cells are more heterogeneous with respect to both their somatodendritic configuration and their axonal projection pattern. Besides normally oriented pyramidal neurons those with an inverted or a horizontal apical dendrite exist. The horizontally oriented neurons differ in one aspect from the other corticocortical pyramidal cells: they possess several vertically ascending collaterals that terminate at the border between layer 4 and 2/3. Corticocortical projecting pyramidal neurons target mainly local layer 5 and 6 neurons. The long-range horizontal projection suggests that these neurons also innervate other layer 6 neurons over several cortical columns as well as in neighbouring cortical areas. Corticocortical pyramidal neurons are therefore involved not only in the intercolumnar signal processing but also in the coordination and balancing of sensory and motor information.

9.6.2 Sublamina 6B In contrast to sublamina 6A, relatively little is known about excitatory neurons in the polymorph sublamina 6B. Some investigators considered sublamina 6B as a distinct layer (layer 7) formed by persisting subplate neurons and present only in

9

The Axon of Excitatory Neurons in the Neocortex

171

some species (Reep and Goodwin, 1988; Clancy and Cauller, 1999; Reep, 2000). Layer 6B, as designated in this chapter, is a very thin sublamina with packed neurons of different shape and size that can be identified on the basis of its gene expression profile, at least in monkey neocortex (Watakabe et al., 2007). Neurons in sublamina 6B are densely packed and are very heterogeneous with respect to their somatodendritic configuration. Pyramidal-like neurons located at the border of sublamina 6B and 6A have vertically oriented dendrites terminating in layer 4. In contrast, neurons located near the white matter are either horizontally oriented or inverted or fusiform (Tömböl et al., 1975; Tömböl, 1984; Clancy and Cauller, 1999). The local axonal arborization of sublamina 6B neurons has been demonstrated using the Golgi technique. These studies revealed a local projection within sublamina 6B itself with a few collaterals projecting into the white matter as well as some ascending collaterals (Tömböl et al., 1975; Tömböl, 1984). Information about the long-range axonal projections of sublamina 6B neurons comes exclusively from extracellular tracing studies and therefore an identification of the exact projecting neuron type is not possible. A horizontal corticocortical projection runs parallel to the white matter in layer 6 within S1 and to the S2 and M1 cortex (Clancy and Cauller, 1999; Arimatsu and Ishida, 2002). A second dominant translaminar projection to layer 1 has also been identified (Vandevelde et al., 1996; Clancy and Cauller, 1999; Arimatsu and Ishida, 2002) and a small population of corticothalamic projecting neurons appears to exist, at least during early postnatal development (Arimatsu and Ishida, 2002).

9.7 Subcellular Axonal Targets It has long been thought that synaptic contacts established between excitatory neurons, i.e. in a cortical column, are not as target specific as described for cortical GABAergic interneurons (Somogyi et al., 1982; Tamás et al., 1998, 2000, 2003; Helmstaedter et al., 2008, 2009a,b,c); for a review see (Soltesz, 2006). There is, however, growing evidence that excitatory neurons also innervate specific regions on their target neurons, both on other excitatory and GABAergic interneurons. Characteristic for excitatory neurons is that they establish synaptic contacts exclusively on dendritic shafts or spines on the postsynaptic target neuron. Paired recordings of intralaminar (within a given cortical layer) or translaminar (between two cortical layers) excitatory connections have shown that synapses were established on basal, apical obliques and terminal tuft dendrites. Synaptic contacts are mainly found on basal dendrites and there, particularly on secondary and tertiary order dendrites (63–85% for layer 5A–layer 5A (L5A–L5A), L5B–L5B, L4–L4, L4–L2/3, L4–L5A, L2/3–L2/3 cell connections, Feldmeyer et al., 1999, 2002, 2005, 2006; Markram, 1997; Markram et al., 1997; Feldmeyer et al., 1999; Lübke et al., 2000b; Feldmeyer et al., 2002; Lübke et al., 2003; Silver et al., 2003; Feldmeyer et al., 2005, 2006; Le Bé et al., 2007; Frick et al., 2008). Only a small fraction (less than 10%) was located on

172

J.H.R. Lübke and D. Feldmeyer

primary and secondary apical oblique or terminal tuft dendrites. The majority of synaptic contacts (∼80%) were located relatively close to the soma (L4–L4: 69 μm; L4–L2/3: 67 μm; L2/3–L2/3: 91 μm; L5A–L5A: 107 μm; L5B–L5B: 147 μm). A similar distribution has been demonstrated for inputs from motor cortex and the VPm and POm thalamic nuclei using ‘channelrhodopsin-2-assisted circuit mapping’ (CRACM); however, this method does not identify the exact location of synapses (Petreanu et al., 2009). Synaptic contacts in the apical tuft have only been found for pyramidal cells in layer 5 (Markram et al., 1997; Feldmeyer et al., 2005), but there is evidence from glutamate uncaging experiments that pyramidal cells in layer 5B innervate terminal tuft dendrites from other pyramidal neurons with somata in layer 2/3 and 5. It has been shown that the proximal (close to the site of their origin near the soma) axonal collaterals of layer 2/3 pyramidal cells innervate both, inhibitory and excitatory dendritic profiles, whereas distal axonal segment of the long-range horizontal axons were seen to establish synaptic contacts only with excitatory profiles (Feldmeyer et al., 2006). Although the number of synaptic contacts varied substantially from 2 to 8 between individual excitatory connections and no strong correlation between the number of synaptic contacts and size of the EPSP amplitude was found, these connections are quite reliable as indicated by the relatively low failure rate and coefficient of variation. Furthermore, the relatively short geometric distance of synaptic contacts implies a short electrotonic distance and hence effective transmission from dendrite to soma (Lübke and Feldmeyer, 2007). Apart from the specificity of excitatory neurons with respect to their subcellular target region, it has been shown that the axons of excitatory neurons display target neuron specificity (Markram et al., 1998; Reyes et al., 1998; Kozloski et al., 2001; Beierlein et al., 2003; Koester and Johnston, 2005; Helmstaedter et al., 2008, 2009b). These findings strongly suggest that cortical connectivity, including those established by excitatory neurons, is far from being random as has long been assumed (Braitenberg and Schüz, 1991; Hellwig et al., 1994; Hellwig, 2000).

9.8 Outlook Although our knowledge about the axonal structure and projection pattern of excitatory neurons in the neocortex is steadily increasing, a number of open question remain. In particular, it is still rather unclear whether pyramidal cells in the same layer but different sensory cortices display similar axonal projection patterns. Furthermore, studies on the diversity of pyramidal neurons within and between cortical layers are just only beginning. Epigenetic factors such as the sensory experience or deprivation can also result in dramatic alterations in the axonal structure and its target specificity, even in mature animals (e.g. Bruno et al., 2009). Finally, little is known about the development of the axonal domain of identified neurons and its contribution to the establishment of defined structures such as the cortical column.

9

The Axon of Excitatory Neurons in the Neocortex

173

References Anderson JC, Douglas RJ, Martin KA, Nelson JC (1994) Synaptic output of physiologically identified spiny stellate neurons in cat visual cortex. J Comp Neurol 341:16–24 Arimatsu Y, Ishida M (2002) Distinct neuronal populations specified to form corticocortical and corticothalamic projections from layer VI of developing cerebral cortex. Neuroscience 114:1033–1045 Ascoli GA, Alonso-Nanclares L, Anderson SA, Barrionuevo G, Benavides- Piccione R, Burkhalter A, Buzsaki G, Cauli B, DeFelipe J, Fairen A, Feldmeyer D, Fishell G, Fregnac Y, Freund TF, Gardner D, Gardner EP, Goldberg JH, Helmstaedter M, Hestrin S, Karube F, Kisvarday ZF, Lambolez B, Lewis DA, Marin O, Markram H, Munoz A, Packer A, Petersen CC, Rockland KS, Rossier J, Rudy B, Somogyi P, Staiger JF, Tamas G, Thomson AM, Toledo-Rodriguez M, Wang Y, West DC, Yuste R (2008) Petilla terminology: nomenclature of features of GABAergic interneurons of the cerebral cortex. Nat Rev Neurosci 9:557–568 Beierlein M, Gibson JR, Connors BW (2003) Two dynamically distinct inhibitory networks in layer 4 of the neocortex. J Neurophysiol 90:2987–3000 Bender KJ, Rangel J, Feldman DE (2003) Development of columnar topography in the excitatory layer 4 to layer 2/3 projection in rat barrel cortex. J Neurosci 23:8759–8770 Betz W (1874) Anatomischer Nachweis zweier Gehirncentra. Centralblatt für die medizinischen Wissenschaften 12:578–580, 595–599 Braitenberg V, Schüz A (1991) Anatomy of the cortex. Statistics and geometry (Studies of Brain Function), Vol. 18. Springer, Heidelberg, Berlin, New York Bruno RM, Hahn TT, Wallace DJ, de Kock CP, Sakmann B (2009) Sensory experience alters specific branches of individual corticocortical axons during development. J Neurosci 29: 3172–3181 Bueno-Lopez JL, Reblet C, Lopez-Medina A, Gomez-Urquijo SM, Grandes P, Gondra J, Hennequet L (1991) Targets and laminar distribution of projection neurons with ‘inverted’ morphology in rabbit cortex. Eur J Neurosci 3:415–430 Bureau I, von Saint Paul F, Svoboda K (2006) Interdigitated paralemniscal and lemniscal pathways in the mouse barrel cortex. PLoS Biol 4:e382 Cauli B, Porter JT, Tsuzuki K, Lambolez B, Rossier J, Quenet B, Audinat E (2000) Classification of fusiform neocortical interneurons based on unsupervised clustering. Proc Natl Acad Sci USA 97:6144–6149 Chen CC, Abrams S, Pinhas A, Brumberg JC (2009) Morphological heterogeneity of layer VI neurons in mouse barrel cortex. J Comp Neurol 512:726–746 Clancy B, Cauller LJ (1999) Widespread projections from subgriseal neurons (layer VII) to layer I in adult rat cortex. J Comp Neurol 407:275–286 DeFelipe J (2005) Reflections on the structure of the cortical minicolumn. In: Casanova F (ed) Neocortical modularity and the cell minicolumn. Nova Science Publishers, New York, pp 57–92 DeFelipe J, Farinas I (1992) The pyramidal neuron of the cerebral cortex: morphological and chemical characteristics of the synaptic inputs. Prog Neurobiol 39:563–607 Egger V, Nevian T, Bruno RM (2008) Subcolumnar dendritic and axonal organization of spiny stellate and star pyramid neurons within a barrel in rat somatosensory Cortex. Cereb Cortex 18:876–889 Escobar MI, Pimienta H, Caviness VS Jr, Jacobson M, Crandall JE, Kosik KS (1986) Architecture of apical dendrites in the murine neocortex: dual apical dendritic systems. Neuroscience 17:975–989 Feldman ML (1984) Morphology of the neocortical pyramidal neurons. In: Peters A, Jones EG (eds) Cerebral cortex volume 1 cellular components of the cerebral cortex. Plenum Press, New York, pp 123–200 Feldmeyer D, Egger V, Lübke J, Sakmann B (1999) Reliable synaptic connections between pairs of excitatory layer 4 neurones within a single ‘barrel’ of developing rat somatosensory cortex. J Physiol-London 521(Pt 1):169–190

174

J.H.R. Lübke and D. Feldmeyer

Feldmeyer D, Lübke J, Sakmann B (2006) Efficacy and connectivity of intracolumnar pairs of layer 2/3 pyramidal cells in the barrel cortex of juvenile rats. J Physiol-London 575: 583–602 Feldmeyer D, Lübke J, Silver RA, Sakmann B (2002) Synaptic connections between layer 4 spiny neurone-layer 2/3 pyramidal cell pairs in juvenile rat barrel cortex: physiology and anatomy of interlaminar signalling within a cortical column. J Physiol-London 538: 803–822 Feldmeyer D, Roth A, Sakmann B (2005) Monosynaptic connections between pairs of spiny stellate cells in layer 4 and pyramidal cells in layer 5A indicate that lemniscal and paralemniscal afferent pathways converge in the infragranular somatosensory cortex. J Neurosci 25:3423–3431 Fitzpatrick D (1996) The functional organization of local circuits in visual cortex: insights from the study of tree shrew striate cortex. Cereb Cortex 6:329–341 Fleischhauer K, Petsche H, Wittkowski W (1972) Vertical bundles of dendrites in the neocortex. Z Anat Entwicklungsgesch 136:213–223 Frick A, Feldmeyer D, Helmstaedter M, Sakmann B (2008) Monosynaptic connections between pairs of L5A pyramidal neurons in columns of juvenile rat somatosensory cortex. Cereb Cortex 18:397–406 Frotscher M (1998) Cajal-Retzius cells, reelin, and the formation of layers. Curr Opin Neurobiol 8:570–575 Gabbott PL, Martin KA, Whitteridge D (1987) Connections between pyramidal neurons in layer 5 of cat visual cortex (area 17) J Comp Neurol 259:364–381 Gilbert CD, Wiesel TN (1979) Morphology and intracortical projections of functionally characterised neurones in the cat visual cortex. Nature 280:120–125 Gilbert CD, Wiesel TN (1983) Clustered intrinsic connections in cat visual cortex. J Neurosci 3:1116–1133 Gottlieb JP, Keller A (1997) Intrinsic circuitry and physiological properties of pyramidal neurons in rat barrel cortex. Exp Brain Res 115:47–60 Gupta A, Wang Y, Markram H (2000) Organizing principles for a diversity of GABAergic interneurons and synapses in the neocortex. Science 287:273–278 Harris RM, Woolsey TA (1981) Dendritic plasticity in mouse barrel cortex following postnatal vibrissa follicle damage. J Comp Neurol 196:357–376 Harris RM, Woolsey TA (1983) Computer-assisted analyses of barrel neuron axons and their putative synaptic contacts. J Comp Neurol 220:63–79 Hellwig B (2000) A quantitative analysis of the local connectivity between pyramidal neurons in layers 2/3 of the rat visual cortex. Biol Cybern 82:111–121 Hellwig B, Schuz A, Aertsen A (1994) Synapses on axon collaterals of pyramidal cells are spaced at random intervals: a Golgi study in the mouse cerebral cortex. Biol Cybern 71:1–12 Helmstaedter M, Sakmann B, Feldmeyer D (2009a) L2/3 interneuron groups defined by multiparameter analysis of axonal projection, dendritic geometry, and electrical excitability. Cereb Cortex 19:951–962 Helmstaedter M, Sakmann B, Feldmeyer D (2009b) Neuronal correlates of local, lateral, and translaminar inhibition with reference to cortical columns. Cereb Cortex 19:926–937 Helmstaedter M, Sakmann B, Feldmeyer D (2009c) The relation between dendritic geometry, electrical excitability, and axonal projections of L2/3 Interneurons in rat barrel cortex. Cereb Cortex 19:938–950 Helmstaedter M, Staiger JF, Sakmann B, Feldmeyer D (2008) Efficient recruitment of layer 2/3 interneurons by layer 4 input in single columns of rat somatosensory cortex. J Neurosci 28:8273–8284 Hirsch JA (1995) Synaptic integration in layer IV of the ferret striate cortex. J Physiol-London 483, 183–199 Hirsch JA, Martinez LM, Alonso JM, Desai K, Pillai C, Pierre C (2002) Synaptic physiology of the flow of information in the cat’s visual cortex in vivo. J Physiol-London 540:335–350

9

The Axon of Excitatory Neurons in the Neocortex

175

Hoeflinger BF, Bennett-Clarke CA, Chiaia NL, Killackey HP, Rhoades RW (1995) Patterning of local intracortical projections within the vibrissae representation of rat primary somatosensory cortex. J Comp Neurol 354:551–563 Hoffer ZS, Hoover JE, Alloway KD (2003) Sensorimotor corticocortical projections from rat barrel cortex have an anisotropic organization that facilitates integration of inputs from whiskers in the same row. J Comp Neurol 466:525–544 Hubel DH, Wiesel TN (1959) Receptive fields of single neurones in the cat’s striate cortex. J Physiol-London 148:574–591 Ito M (1992) Simultaneous visualization of cortical barrels and horseradish peroxidase-injected layer 5b vibrissa neurones in the rat. J Physiol-London 454:247–265 Karube F, Kubota Y, Kawaguchi Y (2004) Axon branching and synaptic bouton phenotypes in GABAergic nonpyramidal cell subtypes. J Neurosci 24:2853–2865 Katz LC, Gilbert CD, Wiesel TN (1989) Local circuits and ocular dominance columns in monkey striate cortex. J Neurosci 9:1389–1399 Kawaguchi Y, Kubota Y (1997) GABAergic cell subtypes and their synaptic connections in rat frontal cortex. Cereb Cortex 7:476–486 Koester HJ, Johnston D (2005) Target cell-dependent normalization of transmitter release at neocortical synapses. Science 308:863–866 Kozloski J, Hamzei-Sichani F, Yuste R (2001) Stereotyped position of local synaptic targets in neocortex. Science 293:868–872 Kumar P, Ohana O (2008) Inter- and intralaminar subcircuits of excitatory and inhibitory neurons in layer 6a of the rat barrel cortex. J Neurophysiol 100:1909–1922 Larkman A, Mason A (1990) Correlations between morphology and electrophysiology of pyramidal neurons in slices of rat visual cortex. I. Establishment of cell classes. J Neurosci 10:1407–1414 Larkum ME, Zhu JJ (2002) Signaling of layer 1 and whisker-evoked Ca2+ and Na+ action potentials in distal and terminal dendrites of rat neocortical pyramidal neurons in vitro and in vivo. J Neurosci 22:6991–7005 Larkum ME, Zhu JJ, Sakmann B (1999) A new cellular mechanism for coupling inputs arriving at different cortical layers. Nature 398:338–341 Larsen DD, Callaway EM (2006) Development of layer-specific axonal arborizations in mouse primary somatosensory cortex. J Comp Neurol 494:398–414 Larsen DD, Wickersham IR, Callaway EM (2007) Retrograde tracing with recombinant rabies virus reveals correlations between projection targets and dendritic architecture in layer 5 of mouse barrel cortex. Front Neural Circuits 1:5 Le Bé JV, Silberberg G, Wang Y, Markram H (2007) Morphological, electrophysiological, and synaptic properties of corticocallosal pyramidal cells in the neonatal rat neocortex. Cereb Cortex 17:2204–2213 Lee CC, Sherman SM (2008) Synaptic Properties of Thalamic and Intracortical Inputs to Layer 4 of the First- and Higher-Order Cortical Areas in the Auditory and Somatosensory Systems. J Neurophysiol 100:317–326 Lefort S, Tomm C, Floyd Sarria JC, Petersen CC (2009) The excitatory neuronal network of the C2 barrel column in mouse primary somatosensory cortex. Neuron 61:301–316 Lorente de Nó R (1949) Cerebral cortex: architecture, intracortical connections, motor projections. In: Fulton JF (ed) Physiology of the nervous system, 3rd edn. Oxford University Press, London, pp 288–313 Lübke J, Egger V, Sakmann B, Feldmeyer D (2000a) Columnar organization of dendrites and axons of single and synaptically coupled excitatory spiny neurons in layer 4 of the rat barrel cortex. J Neurosci 20:5300–5311 Lübke J, Feldmeyer D (2007) Excitatory signal flow and connectivity in a cortical column: focus on barrel cortex. Brain Struct Funct 212:3–17 Lübke J, Feldmeyer D, Silver RA, Sakmann B (2000b) Morphology of synaptic connections between spiny layer 4 neurones and layer 2/3 pyramidal cells in rat barrel cortex. Eur J Neurosci 12:14–14

176

J.H.R. Lübke and D. Feldmeyer

Lübke J, Roth A, Feldmeyer D, Sakmann B (2003) Morphometric analysis of the columnar innervation domain of neurons connecting layer 4 and layer 2/3 of juvenile rat barrel cortex. Cereb Cortex 13:1051–1063 Lund JS, Wu Q, Hadingham PT, Levitt JB (1995) Cells and circuits contributing to functional properties in area V1 of macaque monkey cerebral cortex: bases for neuroanatomically realistic models. J Anat 187:563–581 Luskin MB, Shatz CJ (1985) Neurogenesis of the cat’s primary visual cortex. J Comp Neurol 242:611–631 Manns ID, Sakmann B, Brecht M (2004) Sub- and suprathreshold receptive field properties of pyramidal neurones in layers 5A and 5B of rat somatosensory barrel cortex. J Physiol-London 556: 601–622 Marín-Padilla M (1971) Early prenatal ontogenesis of the cerebral cortex (neocortex) of the cat (Felis domestica) A Golgi study. I. The primordial neocortical organization. Z Anat Entwicklungsgesch 134:117–145 Marín-Padilla M (1978) Dual origin of the mammalian neocortex and evolution of the cortical plate. Anat Embryol (Berl) 152:109–126 Markram H (1997) A network of tufted layer 5 pyramidal neurons. Cereb Cortex 7:523–533 Markram H, Lübke J, Frotscher M, Roth A, Sakmann B (1997) Physiology and anatomy of synaptic connections between thick tufted pyramidal neurones in the developing rat neocortex. J Physiol-London 500:409–440 Markram H, Wang Y, Tsodyks M (1998) Differential signaling via the same axon of neocortical pyramidal neurons. Proc Natl Acad Sci USA 95:5323–5328 Martin KA, Whitteridge D (1984) Form, function and intracortical projections of spiny neurones in the striate visual cortex of the cat. J Physiol 353:463–504 Micheva KD, Beaulieu C (1995) Postnatal development of GABA neurons in the rat somatosensory barrel cortex: a quantitative study. Eur J Neurosci 7:419–430 Miller MW (1988) Maturation of rat visual cortex: IV. The generation, migration, morphogenesis, and connectivity of atypically oriented pyramidal neurons. J Comp Neurol 274:387–405 Mountcastle VB (1957) Modality and topographic properties of single neurons of cat’s somatic sensory cortex. J Neurophysiol 20:408–434 Muly EC, Fitzpatrick D (1992) The morphological basis for binocular and ON/OFF convergence in tree shrew striate cortex. J Neurosci 12:1319–1334 Peters A, Walsh TM (1972) A study of the organization of apical dendrites in the somatic sensory cortex of the rat. J Comp Neurol 144:253–268 Petreanu L, Mao T, Sternson SM, Svoboda K (2009) The subcellular organization of neocortical excitatory connections. Nature 457:1142–1145 Porter JT, Johnson CK, Agmon A (2001) Diverse types of interneurons generate thalamus-evoked feedforward inhibition in the mouse barrel cortex. J Neurosci 21:2699–2710 Porter LL, Sakamoto K (1988) Organization and synaptic relationships of the projection from the primary sensory to the primary motor cortex in the cat. J Comp Neurol 271: 387–396 Porter LL, Sakamoto T, Asanuma H (1990) Morphological and physiological identification of neurons in the cat motor cortex which receive direct input from the somatic sensory cortex. Exp Brain Res 80:209–212 Ramón y Cajal S (1904) Textura del Sistema Nervioso del Hombre y de los Vertebrados. Imprenta de Nicolás Moya, Madrid Reblet C, Blanco I, Mendizabal-Zubiaga J, Gutierrez-Ibarluzea I, Bueno-Lopez JL (1996) Development of inverted cells in infragranular layers of the rabbit visual cortex. Int J Dev Biol Suppl 1:145S–146S Reep RL (2000) Cortical layer VII and persistent subplate cells in mammalian brains. Brain Behav Evol 56:212–234 Reep RL, Goodwin GS (1988) Layer VII of rodent cerebral cortex. Neurosci Lett 90:15–20

9

The Axon of Excitatory Neurons in the Neocortex

177

Reyes A, Lujan R, Rozov A, Burnashev N, Somogyi P, Sakmann B (1998) Target-cell-specific facilitation and depression in neocortical circuits. Nat Neurosci 1:279–285 Schubert D, Kotter R, Luhmann HJ, Staiger JF (2006) Morphology, electrophysiology and functional input connectivity of pyramidal neurons characterizes a genuine layer Va in the primary somatosensory cortex. Cereb Cortex 16:223–236 Schubert D, Kotter R, Zilles K, Luhmann HJ, Staiger JF (2003) Cell type-specific circuits of cortical layer IV spiny neurons. J Neurosci 23:2961–2970 Shepherd GM, Stepanyants A, Bureau I, Chklovskii D, Svoboda K (2005) Geometric and functional organization of cortical circuits. Nat Neurosci 8:782–790 Shepherd GM, Svoboda K (2005) Laminar and columnar organization of ascending excitatory projections to layer 2/3 pyramidal neurons in rat barrel cortex. J Neurosci 25:5670–5679 Silver RA, Lübke J, Sakmann B, Feldmeyer D (2003) High-probability uniquantal transmission at excitatory synapses in barrel cortex. Science 302:1981–1984 Soltesz I (2006) Diversity in the neuronal machine: order and variability in interneuronal microcircuits. Oxford University Press, Oxford, New York Somogyi P, Freund TF, Cowey A (1982) The axo-axonic interneuron in the cerebral cortex of the rat, cat and monkey. Neuroscience 7:2577–2607 Somogyi P, Tamás G, Lujan R, Buhl EH (1998) Salient features of synaptic organisation in the cerebral cortex. Brain Res Brain Res Rev 26:113–135 Staiger JF, Flagmeyer I, Schubert D, Zilles K, Kotter R, Luhmann HJ (2004) Functional diversity of layer IV spiny neurons in rat somatosensory cortex: quantitative morphology of electrophysiologically characterized and biocytin labeled cells. Cereb Cortex 14: 690–701 Tamás G, Buhl EH, L˝orincz A, Somogyi P (2000) Proximally targeted GABAergic synapses and gap junctions synchronize cortical interneurons. Nat Neurosci 3:366–371 Tamás G, L˝orincz A, Simon A, Szabadics J (2003) Identified sources and targets of slow inhibition in the neocortex. Science 299:1902–1905 Tamás G, Somogyi P, Buhl EH (1998) Differentially interconnected networks of GABAergic interneurons in the visual cortex of the cat. J Neurosci 18:4255–4270 Thomson AM, Bannister AP (2003) Interlaminar connections in the neocortex. Cereb Cortex 13: 5–14 Tömböl T (1984) Layer VI cells. In: Peters A, Jones EG (eds) Cerebral cortex. Plenum Press, New York, London, pp. 479–519 Tömböl T, Hajdu F, Somogyi G (1975) Identification of the Golgi picture of the layer VI corticgeniculate projection neurons. Exp Brain Res 24:107–110 Ts’o DY, Gilbert CD, Wiesel TN (1986) Relationships between horizontal interactions and functional architecture in cat striate cortex as revealed by cross-correlation analysis. J Neurosci 6:1160–1170 Valverde F (1986) Intrinsic neocortical organization: some comparative aspects. Neuroscience 18:1–23 Vandevelde IL, Duckworth E, Reep RL (1996) Layer VII and the gray matter trajectories of corticocortical axons in rats. Anat Embryol (Berl) 194:581–593 Watakabe A, Ichinohe N, Ohsawa S, Hashikawa T, Komatsu Y, Rockland KS, Yamamori T (2007) Comparative analysis of layer-specific genes in mammalian neocortex. Cereb Cortex 17:1918–1933 Winkelmann E, Brauer K, Berger U (1975) Zur columnaren Organisation von Pyramidenzellen im visuellen Cortex der Albinoratte. Z Mikrosk Anat Forsch 89:239–256 Wise SP, Jones EG (1976) The organization and postnatal development of the commissural projection of the rat somatic sensory cortex. J Comp Neurol 168:313–343 Woolsey TA, Van der Loos H (1970) The structural organization of layer IV in the somatosensory region (SI) of mouse cerebral cortex. The description of a cortical field composed of discrete cytoarchitectonic units. Brain Res 17:205–242

178

J.H.R. Lübke and D. Feldmeyer

Yabuta NH, Callaway EM (1998) Functional streams and local connections of layer 4C neurons in primary visual cortex of the macaque monkey. J Neurosci 18:9489–9499 Yabuta NH, Sawatari A, Callaway EM (2001) Two functional channels from primary visual cortex to dorsal visual cortical areas. Science 292:297–300 Zhang ZW, Deschênes M (1997) Intracortical axonal projections of lamina VI cells of the primary somatosensory cortex in the rat: a single-cell labeling study. J Neurosci 17:6365–6379 Zhang ZW, Deschênes M (1998) Projections to layer VI of the posteromedial barrel field in the rat: a reappraisal of the role of corticothalamic pathways. Cereb Cortex 8:428–436

Part IV

Axons and Degeneration/Regeneration

Chapter 10

Axon Degeneration: Mechanisms and Consequences Lucy J. Broom and V. Hugh Perry

10.1 Introduction Historically the focus of research into neurodegenerative diseases has been the cell body, but more recently this is shifting to acknowledge the strong contribution of axon and synapse degeneration to neurodegenerative diseases. Axonal loss is evident in traumatic injury of the spinal cord as well as neurodegenerative diseases such as amyotrophic lateral sclerosis (ALS), multiple sclerosis (MS) and toxic neuropathy. For example, end stage multiple sclerosis plaques show up to 60% loss of axons and the extent of axonal damage is most pronounced in active demyelinating plaques, although demyelination is not a prerequisite for axonal injury. This suggests that axonal loss is an early event in disease pathology and contributes to the functional deficits experienced by the patient. While therapeutic interventions which protect the cell body in models of spinal cord injury, ALS and MS do show some efficacy ameliorating clinical symptoms, directing protection to the axon may prove to be of considerable value. However, this requires a greater understanding of axonal biology, the extent to which the axon can function as an independent cellular compartment under normal physiological conditions, and also the pathways that are involved in the initiation of degeneration.

10.2 The Axon as a Functionally Distinct Compartment Classically, protein synthesis has been attributed exclusively to the cell body compartment, with proteins subsequently being transported to their required location (Droz and Leblond, 1963). We classify these axonally transported proteins according to their speed of transport. Fast transport applies to vesicles and membranous organelles, which travel along microtubules at a mean rate of 50–200 mm per day. L.J. Broom (B) CNS Inflammation group, University of Southampton, Southampton, UK e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_10, 

181

182

L.J. Broom and V.H. Perry

Slow transport of cytoskeletal and cytoplasmic proteins occurs at a rate of 0.1–3 mm per day (Campenot and Eng, 2000). The problem with transport-dependent distribution is an obvious inconsistency that lies in the fact that many of the largest axons, such as those of motor neurons, can be up to a metre in length. It is unlikely that proteins could survive transport to such distal regions of the nervous system, for example tubulin has a half-life of 1 week in rat brain neurons (Forgue and Dahl, 1978), and if transported at a maximum rate of 3 mm per day is only likely to survive a few centimetres before being degraded. Since axonal proteins do not possess extended half-lives it is logical that proteins can be synthesised locally in the axon (Nixon, 1980). Following axotomy, the earliest regenerative sprouting at the axonal tip can occur within 24 h of the injury and suggests a local axonal response (Friede and Bischhausen, 1980; Pan et al., 2003). Supportive evidence for this hypothesis was sought and the presence of tRNA (Koenig, 1979), initiation factors (Zheng et al., 2001) and mRNA (Koenig, 1979; Bassell et al., 1998) has been demonstrated within vertebrate axons. Ribosomes were identified sporadically along axons of developing and mature neurons in early studies (Tennyson, 1970; Zelena, 1970; Bunge, 1973), but have since been confirmed in the initial axon segment as well as intermittently along the length of the axon, forming focal centres of translational activity (Steward and Ribak, 1986; Pannese and Ledda, 1991; Koenig et al., 2000). Conclusive evidence of mRNA translation within mature axons had proved difficult until in vitro studies using a dorsal root ganglion (DRG) crush injury model investigated protein synthesis in the separated axon during subsequent regeneration (Twiss et al., 2000; Zheng et al., 2001). Electron microscopic analysis of dorsal root ganglia (DRG) cultures with uranyl acetate detected electron-dense particles free within the axoplasm as well as on a modified type of ER structure. These particles were similar to those observed on rough ER in the cell body and were therefore be assumed to represent ribosomes. The functional activity of these ribosomes has since been verified through their association with a number of cytoskeletal protein mRNAs including beta-actin and neurofilament (Zheng et al., 2001). Axons also have the capacity to export proteins and insert them at the cell surface, which hints at the existence of a Golgi apparatus. It is possible that an unconventional Golgi may be present but to date no evidence of post-translational glycosylation has been reported to occur in axons (Brittis et al., 2002). In contrast, axonal lipid biosynthesis has been demonstrated (Posse de Chaves et al., 1995) and coupled with the now accepted local protein synthesis; the axon should now be considered as a metabolically active neuronal compartment. Characterisation of the range of axonally synthesised proteins is in its infancy but cytoskeletal proteins are thought to be the main group. Embryonic mRNAs identified to date include microtubule-associated protein (MAP) tau (Litman et al., 1993), beta-actin (Bassell et al., 1998) and actin-depolymerising factor (Lee and Hollenbeck, 2003). Using a ‘Campenot chamber’ to culture developing sympathetic rat neurons Eng et al. were able to show axonal uptake of radio-labelled methionine, which could then be tracked through its incorporation into newly synthesised axonal proteins including actin and tubulin (Eng et al., 1999). Similarly, in

10

Axon Degeneration: Mechanisms and Consequences

183

adult mammalian axons mRNA encoding beta-actin, tubulin and neurofilament (NF) subunits seem to predominate (Zheng et al., 2001). By far the most comprehensive list of axonally synthesised proteins has been produced by Willis et al., using a proteomic approach to analyse DRG axonal preparations. This in vitro study identified 40 proteins that were axonally synthesised and proved that the axon is capable of considerable autonomy (Willis et al., 2005). It is currently estimated that axonally synthesised proteins account for 5% of total neuronal protein synthesis (Lee and Hollenbeck, 2003). The role of local protein synthesis has largely been considered in the context of developing neurons and regenerating neurons where it confers distinct advantages, allowing rapid, flexible and efficient responses. One of the best studied roles for axonal protein synthesis is in responses to guidance cues. Xenopus retinal ganglion cell (RGC) axons separated from their cell bodies show an increased level of axonal protein synthesis following exposure to the attractant sema3A or the repellent netrin-1. In this model the orientation of the newly formed growth cone is dependent on local protein synthesis since chemotactic responses of the axon can be blocked by application of a protein synthesis inhibitor (Campbell and Holt, 2001). It is possible that local synthesis and presentation of specific receptors may confer regional specialisation enabling selective responses to guidance cues such as netrin-1 and sema3A. This may be advantageous in allowing spatially restricted control and enabling induction of more efficient responses compared to those possible if reliant on bulk axonal transport. Similar experiments with axotomised rat DRGs support the hypothesis that increased axonal protein synthesis is necessary for growth cone formation (Campbell and Holt, 2001; Brittis et al., 2002; Verma et al., 2005), which involves major remodelling of the cytoskeleton (Mueller, 1999). This process can be initiated within 24 h of axotomy, making axonal synthesis of these components essential for rapid remodelling. Tracking the incorporation of radio-labelled leucine into axotomised axons shows that regeneration is driven by synthesis of new proteins at the site of injury rather than redistribution of existing proteins. Local protein synthesis is therefore clearly important in development because the formation of a new growth cone is associated with a rapid increase in protein synthesis and an abundance of ribosomal protein, capped RNA and translation initiation factors (eIF-4E and eIF-4EBP1) (Campbell and Holt, 2001, 2003). The complement of protein synthesis machinery within different neuronal populations may also offer an explanation for variation in axonal regenerative and chemotactic capabilities. Accordingly, adult retinal axons have low levels of ribosomes (detected by ribosomal protein P0 immunoreactivity) and show limited regenerative efforts, while sensory axons have high levels and implement successful regenerative mechanisms following injury (Verma et al., 2005). The role of axonal protein synthesis under normal conditions is not known and current knowledge is restricted to development and under conditions of stress/injury. Axonal injury immediately induces a rapid burst of action potentials, and following a short delay a pathway of communication from the site of injury to the cell body is activated. This mechanism has two phases, first an interruption to the normal flow of trophic factors including a 10-fold reduction in retrograde transport of nerve

184

L.J. Broom and V.H. Perry

growth factor (NGF) (Raivich et al., 1991). Second, retrograde transport of newly modified proteins originating from the site of injury (Perlson et al., 2004). The first evidence of this communication came from studies in Aplysia, where microinjection of lesion-induced axoplasmic proteins initiated growth and survival responses in the neuronal cell body (Zhang and Ambron, 2000). The mechanism of transport was then identified as nuclear localisation signal (NLS) dependent, following the observation that the SV40 NLS targeted axonal proteins for retrograde transport and nuclear uptake (Schmied and Ambron, 1997; Wellmann et al., 2001). The NLS is known to bind to importins, which are expressed throughout neuronal axons. Of the two isoforms, importin α expressed in control and injured sciatic nerve axons is associated with dynein (Hanz et al., 2003), while importin β is only detected in the axoplasm of lesioned nerves. The precise mechanism of this signalling pathway therefore involves a rapid increase in local axonal importin β synthesis and formation of an importin α/β dimer. This dimer then binds with high affinity to axoplasmic proteins containing the NLS and co-ordinates their transport to the cell body in association with the dynein motor protein (Hanz et al., 2003). This supports a second role for axonal synthesis in transmitting ‘distress signals’ to the cell body following injury, a function that relies entirely on the ability of axons to activate or modify proteins at the site of injury and up-regulate importin, allowing these high-affinity complexes to be efficiently transported along dynein transport machinery. Possible candidates for cargo molecules include vimentin fragments complexed to activated ERKs (Perlson et al., 2005) as well as the activating transcription factors ATF2 and ATF3 (Lindwall and Kanje, 2005). However, these molecules do not posses a classical NLS and may therefore require an adaptor protein to facilitate retrograde transport by this pathway.

10.3 Axon Degeneration 10.3.1 Morphological Changes Following nerve injury an acute phase of axonal degeneration occurs in the regions immediately proximal and distal to the site of injury. There is a rapid transient depolarisation of the axonal membrane at the time of injury, which is quickly normalised by resealing at the site of injury. This is followed by a short delay during which retrograde transport continues and results in a focal accumulation of organelles, dense bodies and vesicular bodies at the proximal end of the disconnected nerve fibres (Zelena et al., 1968; Griffin et al., 1977). Other accumulating proteins include amyloid precursor protein (APP), non-phosphorylated neurofilaments (NP-NF), metabotropic glutamate receptors and the pore-forming subunit of N-type calcium channels. The distal segment of the nerve fibre then undergoes Wallerian degeneration, a process that was first described by Waller in 1850, see Figure 10.1. The spatiotemporal aspects of Wallerian degeneration are highly variable and are modified by a number of factors, for example axons of increasing calibres are affected

10

Axon Degeneration: Mechanisms and Consequences

Fig. 10.1 A schematic of Wallerian degeneration following traumatic injury to a neuron, such as a crush or transection

185 Axon injury

differently, with large-calibre axons degenerating before fine calibre axons. The type of insult, whether acute or chemical, the origin of the nerve, peripheral or central, as well as the species itself, all have a bearing on the rate of Wallerian degeneration. The morphological changes are, however, conserved between all axons (LoPachin et al., 1990). The earliest morphological events present as focal swelling and beading of axons (Waller, 1850). Rapid disruption of the cytoskeleton results in a granular debris of breakdown products (George and Griffin, 1994). In the CNS the residual axolemma and mitochondria can frequently be observed in axons, and viewed at an electron microscopy level the internal cytoskeletal structure is found to have already disintegrated (Franson and Ronnevi, 1984; George and Griffin, 1994). Components of the cytoskeleton do not disintegrate in a uniform fashion, microtubules fragment first and appear to be incorporated into spheroid bodies within a few hours, while neurofilaments are lost later and are degraded by the UPS. The axolemma itself is not spared and undergoes blebbing and flipping of phosphatidyl serine lipids to the external surface (Sievers et al., 2003), a process that is catalysed by scramblase and flippase enzymes (Seigneuret and Devaux, 1984; Zhou et al., 1997). The fine details of the progression of axonal degeneration can now be visualised using the latest technology and a transgenic YFP mouse (Beirowski et al., 2004), which expresses the fluorescent protein in approximately 3% of its motor and sensory nerve fibres. Using this transgenic mouse individual axons can be monitored in order to determine the direction and rate of degeneration over time. In wildtype mice degeneration appears in a wave-like procession with a sharp boundary between fragmented and non-fragmented cytoskeleton. The direction is determined by the lesion type and this may fit with differences in the speed of degeneration with transected axons undergoing a rapid asynchronous anterograde degeneration, while crush injuries degenerate in a retrograde direction (Beirowski et al., 2004). This is contrary to previous understanding, which assumed a uniform direction of degeneration occurring at different rates.

186

L.J. Broom and V.H. Perry

10.3.2 Protein Expression Changes Protein changes underlying axonal degeneration are very poorly understood. This is mostly due to the inaccessibility of the axon itself, being ensheathed by glial cells, which make it difficult to distinguish between changes in the axon and the ensheathing cells. This can be achieved to some extent by immunostaining for a specific protein of interest but most molecular biology techniques that analyse quantitative protein changes or isolate and study different protein groups are thwarted by the mix of cell types within the experimental sample. Here we summarise some of the proteins that appear to show altered expression either distally or proximally to the site of axonal injury: RhoA: In situ assay of Rho GTPase activity shows neuron-specific activation of RhoA following spinal cord injury in rats. RhoA is dependent on the p75 receptor which transduces signals from myelin-derived inhibitors of axonal regeneration such as NogoA, MAG and MOG. Its activation was specifically distal to the lesion site and may therefore be involved in axon degeneration mechanisms. Follow-up in vitro experiments on DRG explants were able to block the effect of NogoA signalling with a Rho kinase inhibitor, and in vivo, Rho kinase inhibition resulted in a 40% increase in preserved axons 5 days after spinal cord injury, compared to untreated controls (Yamagishi et al., 2005). Roundabout (Robo) 2: Control DRG express Robo2 mRNA at a consistently low level but following axotomy a significant proximal upregulation is detected. Through immunohistochemical techniques Robo2 expression was localised to axons and appeared to be maintained during the early stages of degeneration particularly within large-calibre axons. Interestingly, spinal cord axotomy did not induce the same effect and it is therefore likely to be a specific response in peripheral nerves (Yi et al., 2006). Early upregulation suggests a role for Robo2 in mediating axonal damage, however, it is also known to control migration of developing neural cells, which may be relevant to later regenerative efforts (Kidd et al., 1998). Brain-derived neurotrophic factor (BDNF): BDNF is normally synthesised in 20–30% of spinal sensory neurons, specifically those of a small to medium diameter. The expression of BDNF at the mRNA and protein level shows a significant increase in DRG following injury. This upregulation is rapid and sustained for at least 14 days, however, the picture is complicated by a ‘phenotypic switching’ in neurons with axons of different calibres (Zhou et al., 1999). Small diameter neurons that normally express BDNF downregulate it, while larger diameter neurons upregulate its expression, a finding which is based on immunohistochemical analyses. The trigger in large neurons is attributed to calcium influx into the axotomised neuron, which induces BDNF expression through a CREB-mediated pathway. In contrast BDNF expressed in small neurons is controlled by NGF signalling via the TrkA receptor. Axotomy of these small neurons deprives them of trophic factors transported retrogradely from their targets resulting in the NGF-TrkA-induced expression of BDNF being switched off. This effect seems to correlate with the differential rates of degeneration observed in small- and large-calibre axons, with small

10

Axon Degeneration: Mechanisms and Consequences

187

ones degenerating first and larger calibre axons persisting for longer, which may be partly dependent on trophic factors. Some of the protein changes that have been characterised in relation to axonal degeneration involve the reorganisation and disintegration of cytoskeletal components including neurofilaments and microtubules. These, however, are the downstream consequences of other as yet unknown protein changes and protein activation pathways. The cytoskeleton itself is highly dynamic and its reorganisation and eventual breakdown are all tightly regulated by phosphorylation and activation of regulatory proteins. Some of the earliest morphological changes can be detected as early as 4 h but the cue prior to this and the signalling pathways resulting in remodelling of the cytoskeleton, as well as changes to organelles and the axolemma, are still undiscovered.

10.3.3 Wlds A mutant mouse exhibiting slow Wallerian degeneration was discovered in 1989 (Lunn et al., 1989). Mice possessing this spontaneous mutation present with a 10-fold delayed degeneration of distal axons following a crush injury, see Figure 10.2. The distal segment of peripheral motor axons shows a remarkable capacity to conducting action potentials for more than 2 weeks after separation from the cell body, while the motor nerve terminals continue to release neurotransmitter for more than 4 days. In subsequent years the mutation responsible for this phenotype was narrowed down to a 85 kb tandem triplication on chromosome 4 (Lyon et al., 1993; Coleman et al., 1998). This region contained a chimeric gene comprising exons derived from the 5 -end of ubiquitin factor E4b (Ube4b), nicotinamide mononucleotide adenylytransferase (Nmnat), and ribosome-binding protein 7 (Rbp7). The chimeric gene remains in frame and thus a fusion protein is transcribed and translated, which is apparently responsible for the protection of injured neurons, see Figure 10.2. Transgenic replication of the Wlds phenotype through expression of Nmnat/Ube4b showed that this portion is the minimum requirement to confer axonal protection (Conforti et al., 2000; Mack et al., 2001). In isolation neither the Ube4b or Nmnat domains are sufficient to fully produce the Wlds phenotype (Laser et al., 2006). The Wlds phenotype does not, however, extend to protection of neuronal cell bodies following axonal injury. Retinal ganglion cells (RGC) specifically labelled by fluoro-gold show comparable rates of degeneration in Wlds mutant and control mice at all times point up to 7 days post-optic nerve crush. TUNEL staining of the ganglion cell layer also confirmed comparable levels of apoptotic cells in control and Wlds mice (Wang et al., 2006). This study adds to evidence from Wlds mutant mice subject to a stereotaxic injection of a catecholaminergic neurotoxin. This procedure generates a Parkinson disease-like pathology in wild-type mice, but on a Wlds background degeneration of substantia nigra dopaminergic cell bodies proceeds as normal while dopaminergic fibres remain intact, indicating an independent mechanism of degeneration within axons and cell bodies (Sajadi et al., 2004).

188

L.J. Broom and V.H. Perry 18 AA

A 70 AA Ube4b

285 AA Nmnat Wlds chimeric protein

B

Fig. 10.2 The Wlds protein and axonal protection. (a) The amino acid composition of the chimeric protein responsible for delaying degeneration of injured axons in the Wlds mouse. (b) Toluidinestained 1 μm semithin sections of sciatic nerve illustrating the normal appearance of axons in wild-type mice (panel b), and Wallerian degeneration of distal axons in wild-type mice 7 days posttransection (panel c). This is in stark contrast to panel a, which shows the preservation of axon structure in distal portions of the sciatic nerve from a Wlds mouse, 7 days after transection

As described earlier the pattern of Wallerian degeneration in YFP-expressing wild-type mice appears in a wave-like procession with a sharp boundary between fragmented and non-fragmented cytoskeleton. The direction is determined by the lesion, with transected axons undergoing a rapid asynchronous anterograde degeneration, while crush injuries degenerate in a retrograde direction. The Wlds mutant was also analysed and a more synchronous degeneration was observed that progressed at a rate 100-fold slower than wild-type animals and is described as a ‘slow anterograde axon decay’ (Beirowski et al., 2004). So a distinct pattern of degeneration is evident in wild-type compared to Wlds animals, which further complicates efforts to understand the underlying mechanisms. This does, however, highlight an interesting parallel between axonal degeneration after a crush injury and the ‘dying back’ pathology associated with neurodegenerative diseases and suggests that acutely induced Wallerian degeneration is a good model for addressing more chronic degenerative conditions. It is proposed that degeneration in the Wlds mutant

10

Axon Degeneration: Mechanisms and Consequences

189

follows a separate process and could be a case of gradual protein degradation without replacement and eventual loss of viability. It is unclear what factors may control this or the critical levels that must be reached for degeneration to eventually occur. If this were the case it does not explain the ability of the Wlds protein to protect against neurodegeneration in a series of animal models. Transgenic crosses with the Wlds mutant mouse are protective in the progressive motor neuropathy (pmn) model of motor neuron degeneration (Ferri et al., 2003), the Po knockout dysmyelinating mutant (Samsam et al., 2003) and paclitaxel (taxol)-mediated toxic neuropathy (Wang et al., 2002). This clearly demonstrates that the protective mechanism of the Wlds phenotype is able to counter in vivo mechanisms of neurodegenerative disease, and by understanding the mode of Wlds protection we may find new approaches to discovering neurodegenerative pathways and potential targets for therapeutic intervention. The Wlds protein is localised to the nucleus and surprisingly it is largely, if not entirely absent from axons where its neuroprotective effects are exerted (Fang et al., 2005). It is therefore likely that a downstream factor is responsible for mediating the protective effect through alteration of structural proteins or modulation of signalling pathways. Possible modes of action are continuing to be investigated and hypotheses are based on the functions of the two parent components of the chimeric Wlds protein. Proposed mechanisms have included ubiquitination of transcription factors to alter their stability or RNA processing and also modulation of NAD synthesis with downstream effects on poly-ADP ribosylation and protein activity. These changes would potentially prime the axons to resist degeneration through either altered gene expression or post-translational modification of proteins. However, findings from a series of studies have proven to be highly contradictory and lead us little closer to understanding this phenomenon. The proposed nuclear targets of Wlds include SIRT1, a NAD-dependent deacetylase and VCP/p97, which plays a role in ER-associated degradation (Araki et al., 2004; Laser et al., 2006). The pathways interacting with these two proteins have since become the focus of much investigation. The greatest focus has been on a hypothesis involving NAD, Nmnat and Sirt1. Lentiviral expression of Nmnat1 or Wlds in DRG explant cultures produced a 10-fold increase in intact neurites at 72 h following axotomy (Araki et al., 2004). Mutations to abolish the enzymatic activity of Nmnat resulted in a return to the wild-type phenotype and seemed to suggest an NAD-dependent mechanism of protection. Indeed, exogenous exposure to NAD 24 h prior to injury was able to mimic the protective effect of Wlds . The downstream pathway was determined as Sirt1-dependent because siRNA knockdown or inhibition using sirtinol was able to block NAD-dependent axonal protection (Araki et al., 2004). More recent studies have been unable to reproduce these results and call into question the relevance of the Sirt1 pathway. Herpes simplex viral expression of Nmnat or Wlds still produced a protective phenotype in DRG explants from a Sirt1 null mouse, indicating a Sirt1-independent mechanism (Wang et al., 2005). The same group also attempted to reproduce the in vitro NAD-dependent axonal protection but found that higher concentrations of 1–20 mM NAD were required to produce a protective effect compared to the 0.1–1 mM NAD used by Araki et al.

190

L.J. Broom and V.H. Perry

Both studies did agree on the ability of NAD exposure to mimic the Wlds phenotype but similar effects were also be produced by application of methyl-pyruvate, indicating that local bioenergetics are important factors in axonal degeneration (Wang et al., 2005). This supports an alternative NAD-dependent hypothesis acting through its role in ATP-synthesising redox reactions within the mitochondria. Conforti et al. sought to reproduce the initial study using a lentiviral-based expression system and found that Wlds expression protected neurites from vincristine toxicity for up to 13 days but Nmnat conferred only an intermediate level of protection at 7 days and was not significantly different from controls at 13 days (Conforti et al., 2007). This intermediate level of protection was also conferred by lentiviral expression of Ube4b, but no additive effect was observed by combining Ube4b and Nmnat expression. They were also unable to demonstrate any protective effects of direct NAD application (1 mM) either prior to or after exposure to vincristine. The differences may be due to the change in type of injury but the theory behind this line of investigation is still debatable since increased levels of NAD are not characteristic of the Wlds animals and protection by Nmnat does not translate to in vivo studies. Transgenic Nmnat/YFP mice show axonal degeneration that is indistinguishable from wild-type mice following a sciatic nerve lesion (Conforti et al., 2007). All labelled axons were fragmented by 52 h posttransection, whereas axons in Wlds mice remain continuous after 20 days (Beirowski et al., 2004). DRG explant cultures were used for in vitro confirmation and again Nmnat transgenic neurons exhibited axonal degeneration at an equivalent rate to wild-type axons. Further evidence from a second in vivo model shows that overexpression of dNmnat in Drosophila delays axonal degeneration after exposure to intense light and this protection is thought to be independent of NAD synthesis because enzymatically inactive dNmnat is equally protective (MacDonald et al., 2006). Combining the conclusions from each of these studies results in a somewhat confused story, but all of the most recent and in vivo findings tend to support an NAD-independent mechanism of protection in the Wlds mutant. Downstream effects of the first 70 amino acids of Ube4b, the other Wlds protein component, are proposed through binding of valosin-containing protein (VCP/p97), which causes a redistribution of the protein to the nucleus (Laser et al., 2006). VCP is part of a family of AAA type ATPases and has multiple functions including a role in the UPS (Dai and Li, 2001). Through its binding of polyubiquitin chains VCP has the potential to alter UPS-mediated turnover of transcription factors and in turn alter axonal-acting pathways. Evidence for the role of the UPS in axonal degeneration is discussed in more detail later in this chapter.

10.3.4 Pathways Involved in Axonal Degeneration We can draw a lot of information about the degenerative processes of wild-type axons from our observations of the intricacies of degenerative processes in Wlds animals. Wallerian degeneration is restricted to the distal portion of the axon

10

Axon Degeneration: Mechanisms and Consequences

191

in wild-type animals, this occurs after a short delay and is rapid, but in Wlds animals the axon is sealed off from the cell body and remains functional for up to 21 days. The survival of the axon is therefore not dependent on proteins translated and transported from the cell body. It follows that the effectors of Wallerian degeneration must be intrinsic to the axon in order for the delayed, spontaneous and active dismantling of the axon to proceed. The mechanisms of Wallerian degeneration are unique to the axon, since Wlds does not confer protection to the cell body and the pathways involved are not shared by those controlling other axon remodelling events. For example, developmental pruning is achieved through distal-to-proximal retraction of axons, but while Wlds protects injury-induced degeneration of transected RGC, developmental pruning of the same RGC axons continues as normal (Hoopfer et al., 2006). Similarly, young motor axons in Wlds mutants undergo remodelling at the neuromuscular junction while being protected from Wallerian degeneration (Parson et al., 1997). Synapse elimination at the developing neuromuscular junction, which occurs via so-called axon shedding (Bishop et al., 2004), is also maintained in the Wlds mutant. We can find some similarities between these distinct processes including the need for a cell-autonomous program that is independent of caspase-mediated apoptosis and morphologically all involve microtubule fragmentation at an early stage (Raff et al., 2002; Zhai et al., 2003). However, their underlying mechanisms are diverse; UPS inhibition can completely block developmental axon pruning whereas Wallerian degeneration is only delayed (Watts et al., 2003; Hoopfer et al., 2006). The triggers for these two degenerative processes are, however, distinct and may account for initiation of distinct mechanisms. Wallerian degeneration is triggered by an axonal insult and subsequent events are restricted to the distal portion, without the need for communication with the cell body. In contrast, developmental pruning may be triggered by any number of factors, such as patterned neuronal activity, local signals, transcriptional regulation or a combination of these factors. 10.3.4.1 Mitogen-Activated Protein Kinases (MAPKs) MAPKs are a family of serine/threonine specific kinases whose function is the transduction of extracellular stimuli into altered gene expression. The three subfamilies; extracellular signal-regulated kinase (ERK), c-Jun NH2 -terminal kinase JNK and p38 kinase, are all candidates for initiating the early events after nerve lesion which include a rapid induction of several genes and progressive expression of growth factors. At 30 min following a crush injury the extracellular signal-related kinase Erk1/2 cascade is already activated (phosphorylated) within distal and proximal segments (Sheu et al., 2000). Upregulation of ERK phosphorylation continues until 1 month post-injury within both segments but the cellular source of these changes was not identified. Our understanding has since been enhanced by immunohistochemical analysis of sciatic nerve tissue at several time points post-injury to establish the cellular origins of ERK activation. Proximal activation of ERK was mainly within axons, identified by co-localisation with neurofilaments, but distally, axons were only responsible for immediate ERK activation. S-100 immunopositive Schwann

192

L.J. Broom and V.H. Perry

cells forming bands of Bungner were found responsible for the continuation of ERK activation (Agthong et al., 2006). Possible functions of this ERK activation include stimulation of Schwann cell proliferation and mediation of semaphorinrepulsive activity, but these are related to regenerative efforts (Bagnard et al., 2004). The upstream signals and role of early ERK activation within the axon are as yet undetermined. 10.3.4.2 Mitochondria Wallerian degeneration is an active process, and as such, alterations to mitochondrial function may be a key to its progression and may drive many of the changes associated with this type of axon loss. Changes in mitochondrial membrane potential can be tracked using the fluorescent dye, JC-1, which accumulates in the mitochondrial matrix and shifts its emission wavelength depending on the membrane potential. In cultured DRG explants, axotomy induces a red to green shift in fluorescence which indicates a drop in potential to more depolarised values and therefore a compromise in mitochondrial function (Sievers et al., 2003). Loss of mitochondrial membrane potential is a characteristic feature of apoptotic cell death (Ankarcrona et al., 1995) and this depolarisation precedes the releases of a number of factors including caspases and cytochrome c, but this does not seem to be true for Wallerian degeneration (Deshmukh et al., 2000; Finn et al., 2000). In the DRG axotomy model the spatiotemporal progression of mitochondrial depolarisation mirrors that of membrane blebbing and phosphatidylserine exposure, but the consequence of this is unclear. Certainly it does not induce caspase activation but it may result in energy depletion and contribute to the reversal of the sodium/calcium exchanger allowing a net influx of calcium into the axon. Mitochondria are also a major source of free radicals, and a reduced rate of oxidative phosphorylation and electron transfer may lead to greater release of these damaging molecules and contribute to the intra-axonal spread of degeneration, as hypothesised to occur in multiple sclerosis (Andrews et al., 2005). The ability of dysfunctional mitochondria to directly cause axonal degeneration is evident from a mutation in SPG7, which results in the loss of function of paraplegin, progressive retrograde degeneration of long ascending and descending tracts of the spinal cord, optic neuropathy and peripheral axonopathy (Ferreirinha et al., 2004). In the mouse knockout model of this disease axon degeneration exhibits features of Wallerian degeneration and the primary ultrastructural change is the appearance of abnormal mitochondria several months prior to the onset of axonal loss. This demonstrates that the dysfunction of the mitochondria alone is sufficient to cause axonal degeneration. The authors hypothesise that the distance from the cell body requires that mitochondria persist for longer before replacement and distal regions may therefore be less able to cope with mitochondrial insults and subsequent energy depletion. 10.3.4.3 Calcium/Calpains The earliest stages of Wallerian degeneration are initiated by sodium ions, which increase in concentration within the axoplasm during the rapid transient membrane

10

Axon Degeneration: Mechanisms and Consequences

193

depolarisation that immediately follows injury. This is quickly normalised by resealing of the axon until a later unknown event is thought to trigger the reversal of the Na+ /Ca2+ exchanger and ion specific transport of calcium into the axon (Xie and Barrett, 1991; Craner et al., 2004). Influx of calcium into the axon occurs through an intact axolemma, rather than a result of loss of integrity and subsequent membrane leakage, because specific ion channel inhibitors such as dihydropyridine calcium channel antagonists and bepridil can delay axon degeneration by 4 days (George et al., 1995). This calcium entry is necessary because Wallerian degeneration does not proceed in culture neuron that are maintained in an extracellular calcium concentration of less than 200 μm. Downstream events which are mediated by calcium include degradation of cytoskeletal proteins and activation of calpains but not caspases (Finn et al., 2000). Calpains are calcium-dependent cysteine proteases which are activated by low-micromolar calcium levels (μ-calpain, calpain I) or by millimolar calcium levels (M-calpain, calpain II) (Saido et al., 1994; Shields et al., 1998). Known targets of calpains include axonal cytoskeletal proteins such as spectrin, MAP-1, tau, tubulin and neurofilaments. Myelin components including MBP, MAG and PLP can also be degraded by calpains (Shields and Banik, 1999; Schaecher et al., 2001; Ray et al., 2003). With respect to axonal degeneration, calpain action is thought to result in granular degradation of the axoplasmic proteins. Numerous models of axonal damage including stretch and crush injuries, as well as acute axonal injury including ischaemia and spinal cord injury have all indicated a critical role for calpains (Finn et al., 2000; Yamashima, 2000). In a clinical setting calpains are associated with active MS lesions where axon damage is clearly evident (Shields et al., 1998). These findings all implicate calpains as effectors in axonal degeneration. In vitro studies support this hypothesis since the calcium chelator (EGTA) or calpain blocker (ALLN) can delay axonal degradation, protecting both neurofilaments and microtubules (Zhai et al., 2003). However, the protective effects of these pharmacological approaches were not equal, calcium chelation produced higher levels of protection and this may be indicative of a second or multiple calcium-dependent pathways contributing to Wallerian degeneration. The axonal protection evoked by calcium chelation is minimal compared to that of the Wlds mutant. In this study more than 90% of control SCG axons had degenerated by 8 h post-axotomy and EGTA treatment delayed degeneration by 2-fold, but in vitro Wlds -expressing neurones are capable of surviving intact for as long as 7 days, representing a 20-fold delay (Buckmaster et al., 1995). It therefore appears that while calcium-dependent pathways are critical to the mechanism of cytoskeletal disintegration they are not the ultimate controlling factor in axon degeneration.

10.3.4.4 Ubiquitin Proteasome System The UPS is responsible for selective protein degradation and is involved in the pathogenesis of a number of neurodegenerative diseases. Genetic mutations of UPS components are known to cause neurodegenerative conditions. Mutations in the E3 ligase, Parkin, are associated with a recessive form of juvenile parkinsonism (Kitada et al., 1998; Shimura et al., 2000), and mutations in ubiquitin C-terminal hydrolase

194

L.J. Broom and V.H. Perry

(UCH-L1), which catalyses the release of ubiquitin from its substrate, are associated with gracile axon dystrophy (Gad) (Mukoyama et al., 1989; Saigoh et al., 1999). Furthermore, failure of the UPS may lead to accumulation and subsequent deposition of polyubiquitinated proteins to form the ubiquitin-positive inclusions that are characteristic of neurodegenerative diseases, e.g. neurofibrillary tangles in Alzheimer’s disease and Lewy bodies in Parkinson’s disease (Lowe et al., 1993). The UPS has therefore been of great interest as a potential route for degradation of axonal proteins during Wallerian degeneration. It is unclear whether modification of UPS degradation in these diseases is beneficial, deposition of proteins in ubiquitin-positive inclusions may be protective by sequestering harmful accumulating proteins in one place, but alternatively inappropriate sequestering of proteins may be detrimental to essential cellular pathways and survival. Inhibition of the UPS significantly delays injury-induced axon degeneration. In vitro studies using cultures of primary neurons showed that reversible or irreversible inhibition, using MG132 and lactacystin, respectively, could delay early events such as axon fragmentation, but the eventual degeneration of injured axon was not prevented by these means. Axon degeneration following an optic nerve crush can be delayed by application of a proteasome inhibitor (Zhai et al., 2003). Neurofilament and tubulin staining showed that as a percentage of initial levels, UPS inhibition improved overall fibre density by 50% at 3-day post-injury compared to controls, but again this is only a small delay compared to the Wlds mutant. In vitro analyses point to a UPS involvement in specific aspects of axon degeneration. Microtubule structures disintegrate as early as 4-h post-injury but levels of beta-tubulin are unaltered at this time point. The stability of microtubule structures are maintained by a balance of stabilising and de-stabilising factors (Garcia and Cleveland, 2001), many of which are targets for the UPS, ubiquitin-dependent degradation of MAPs or regulatory molecules may therefore lead to microtubule disintegration. In contrast neurofilament disintegration at the later time point of 8–12 h does involve loss of neurofilament subunits. Neurofilaments are thought to be de-phosphorylated and then degraded in an ubiquitin-dependent manner because pharmacological inhibition of the UPS delays the loss of neurofilament immunoreactivity in transected SCGs (Meller et al., 1994; Zhai et al., 2003). One of the roles of the UPS in axonal degeneration therefore appears to be disassembly and degradation of cytoskeletal components. The Ube4b component of the Wlds fusion protein regulates the polyubiquitination of proteins targeted for degradation by the proteasome (Koegl et al., 1999; Hatakeyama et al., 2001). Ufd2, the yeast homolog has been identified as a positive survival factor under stress conditions, however, any direct interaction between the Wlds protein and UPS function has yet to be determined. Interference with efficient protein ubiquitination and degradation is not compatible with the protective nature of this mutant, since abnormal accumulation of polyubiquitinated proteins is commonly associated with neurodegenerative diseases. A gain of function related to protein ubiquitination is possible but unlikely to account for the Wlds phenomenon since the Wlds protein lacks a functional U-box motif. So despite having pieced together a few parts of the mechanisms underlying Wallerian degeneration, the controlling pathways and how the Wlds protein interferes with these are far from being fully understood.

10

Axon Degeneration: Mechanisms and Consequences

195

10.4 Lessons from Anuclear ‘Models’ of Degeneration In addition to the information that can be gained from neurodegenerative disease and models of axon injury a natural model of nuclear-independent degeneration exists in the form of erythrocytes that undergo programmed cell death known as eryptosis. The average life span of an erythrocyte is 120 days and cells must be cleared from peripheral blood by reticuloendothelial cells in a manner which does not release potentially harmful intracellular contents. Erythrocytes are devoid of nuclei and mitochondria, the elements which are crucial for classic mechanisms of apoptosis, since they are the source of apoptotic enzymes and effector molecules. Erythrocytes must therefore undergo a non-apoptotic form of cell death. The morphological characteristics of eryptosis are largely comparable to those of degenerating axons and apoptotic cells, including breakdown of plasma membrane asymmetry, i.e. phosphatidylserine exposure, cell shrinkage and membrane blebbing (Berg et al., 2001; Bratosin et al., 2001). The mechanism of eryptosis is also highly dependent on calcium. Under normal circumstances erythrocytes show limited channel activity and the membrane is permeable to calcium via a ‘calcium pump leak’ (Etzion et al., 1993). Upon exposure to osmotic or oxidative stress, non-selective cation channels are opened (Huber et al., 2001; Duranton et al., 2002), which are also calcium permeable (Kaestner et al., 2000). Calcium appears to be the central controlling molecule in eryptosis and mirrors Wallerian degeneration in this feature as well as sharing morphological characteristics caused by calcium-activated pathways. Calcium influx activates scramblase, a calcium-sensitive enzyme that flips phospatidylserine lipids to the external membrane (Zhou et al., 2002) and this action favours binding to macrophage receptors and binding to receptors on the vascular wall thereby impeding microcirculation and facilitating degradation (Boas et al., 1998; Fadok et al., 2000; Eda and Sherman, 2002). Calcium-sensitive potassium channels then are activated which allows hyperpolarisation of the cell membrane and drives chloride efflux in parallel with K+ ; this loss of KCl results in cell shrinkage (Brugnara et al., 1993; Franco et al., 1996). In a positive feedback pathway cell shrinkage induces release of platelet-activating factor (PAF), a phospholipids mediator, which stimulates the breakdown of sphingomyelin and release of ceramide (Lang et al., 2005). Downstream effects are cell shrinkage and phosphatidylserine exposure. PAF may also activate potassium channels (Garay and Braquet, 1986) thereby sensitizing the channels to the effects of cytosolic calcium (Rivera et al., 2002). Other consequences of increased intracellular calcium levels include modification of the cytoskeleton (Tanaka et al., 1991; Nunomura et al., 1997) and activation of transglutaminase which catalyses protein crosslinking (Anderson et al., 1977; Bratton, 1993). Calcium-dependent enzymes such as kinases, phosphatases (Cohen and Gascard, 1992; Minetti et al., 1996), phospholipases (Smith et al., 2001) and proteases (calpains) (Anderson et al., 1977) are also stimulated. Apoptosis in nucleated cells is implemented by caspases and since procaspase-3 and procaspase-8 are expressed by erythrocytes (Berg et al., 2001), their involvement has not been overlooked. However, in the absence of mitochondria and the essential apoptosome components Apaf-1, cytochrome c and caspase-9, activation of these caspases is not thought to be possible. Indeed, treatment with a variety

196

L.J. Broom and V.H. Perry

of apoptotic stimuli does not induce activation of caspases-3 or -8. Interestingly, in vitro treatment of erythrocytes with cysteine protease inhibitors does block calciuminduced eryptosis, but this does not involve processing of procaspase 3 (Bratosin et al., 2001). Eryptosis is also blocked by a calpain/cysteine protease inhibitor, which confirms that calpains are the major calcium-dependent protease activated in eryptosis. The targets of calpain are structural components including spectrin, keratins and actin, and also signal transducers, for example Bax, PKC and FAK (Schaecher et al., 2001; Ray et al., 2003). Calpains may also be the main calciumdependent degradative enzyme activated during Wallerian degeneration (see earlier section). Breakdown of structural proteins and cleavage of signal transducers represent a further link between these two mechanisms. By applying our knowledge of the mechanisms underlying eryptosis to what we know of Wallerian degeneration there is the potential for uncovering new avenues of investigation.

10.5 Cellular Responses in Axon Degeneration 10.5.1 Ensheathing Cells Following axonal injury and degeneration, a series of later events are centred around the responses from surrounding and infiltrating cell types, including glial cells and macrophages. In the peripheral nervous system Schwann cells sequester myelin debris and fragment their own myelin sheaths to form characteristic myelin ovoids. Schwann cells also undergo a period of rapid proliferation reaching a peak at day 3, at this time they align with the basal lamina and form bands of Bungner, which serve as a guide for regenerating nerve fibres (Stoll et al., 1989; Liu et al., 1995). In contrast the oligodendrocytes of the CNS appear quiescent following nerve injury and myelin clearance is significantly delayed compared to the PNS response; myelin and axonal debris remain for several months following CNS injury (Bignami and Ralston, 1969). The degeneration of myelin studied using either EM or general stains, such as luxol fast blue, only looks at the morphological characteristics, which progress from looser wrapping of myelin around the axon to segmentation and eventual formation of ovoids (George and Griffin, 1994). If the protein detail of myelin degeneration is examined a lot more can be understood about this process. Myelin sheaths are crudely separated into three features: a periaxonal membrane comprising mainly of myelin-associated glycoprotein (MAG) and NogoA; a compact middle region of myelin basic protein (MBP) and an abaxonal membrane of myelin oligodendrocyte glycoprotein (MOG) and NogoA. Each of these regions seems to have a slightly different mode of degeneration (Buss and Schwab, 2003). The inner component degenerates very rapidly, showing a similar time course to cytoskeletal disintegration in the axon it ensheaths. The outer layers remain intact for much longer and the first signs of degeneration coincide with macrophage recruitment at around 7 days. This implies a close relationship between the axon and its apposing myelin

10

Axon Degeneration: Mechanisms and Consequences

197

membrane and their mechanism of degeneration may also be similar in nature. Potentially, the proteases responsible for axonal degeneration may then go on to attack the inner myelin membrane following breakdown of the axonal membrane integrity. Meanwhile, the outer myelin layers show a delayed degeneration and persist in the CNS for an extended length of time, despite the presence of activated microglia. Whether this failure of clearance is due to the microglia or a feature of the myelin debris is not known. It does seem to be dependent on the environment of the CNS because in vitro the addition of peripheral macrophages to optic nerve does not result in a substantial increase in myelin uptake (Kuhlmann et al., 2002). Previous studies have hinted at a lack of macrophage-activating factor in CNS tissue or indeed the presence of an inhibitory factor which may account for the different rates of myelin clearance (Reichert and Rotshenker, 1996). Another possibility is that the unknown factor is intrinsic to the ensheathing cell. In the PNS, Schwann cells undergo dramatic changes in response to axonal injury, including dedifferentiation, proliferation, modification of their cell surface and secretion of molecules (Martini and Schachner, 1988; Meyer et al., 1992; Carroll et al., 1997). In comparison CNS oligodendrocytes appear quiescent (Ludwin, 1990). They do shed their myelin sheath but seem to take no further role in either axonal or myelin degeneration and phagocytosis. Oligodendrocytes may therefore be key to impeding myelin recognition and subsequent phagocytosis of debris in the CNS.

10.5.2 Macrophages The importance of infiltrating monocytes was first described by Ramon y Cajal (1928). The role of these cells is multifactorial but one of their key functions is the clearance of myelin debris. Macrophages begin to infiltrate and migrate to the myelin ovoids at maximal levels between days 4 and 7, and myelin clearance is completed in less than 2 weeks in the PNS (Perry et al., 1987; Bruck, 1997; Stoll and Jander, 1999). Within the PNS, endoneurial macrophages are formed of two populations. First, the resident macrophages that account for approximately 9% of this endoneurial macrophage population and second, blood-borne (haematogenous) monocytes that rapidly migrate to the site of injury and distal segment (Griffin et al., 1993). Once in situ they differentiate into macrophages and proceed to engulf myelin debris (Griffin et al., 1992; Perry and Brown, 1992; Friede and Bruck, 1993; Goodrum et al., 1994). In contrast, injured nerves of the CNS recruit very few macrophages from the circulation instead, resident microglia mount a reduced and delayed response resulting in slower clearance of debris (Perry et al., 1987; Stoll et al., 1989; Lawson et al., 1994). One of the clearest demonstrations of this central versus peripheral distinction comes from studies of the spinal dorsal root, which has portions within both the peripheral and the central nervous system. Wallerian degeneration of the dorsal root sees macrophage invasion into the PNS portion progressively increasing from 24 h

198

L.J. Broom and V.H. Perry

onwards, reaching statistically significant levels after day 3 (Avellino et al., 1995). Levels are maintained for 14 days post-axotomy, at which point myelin clearance is complete. Meanwhile the CNS portion demonstrates very limited recruitment of macrophages, at day 14 a small but significant number were detectable. Similar results are obtained when comparing crush injuries in entirely separate regions, i.e. sciatic nerve versus optic nerve and the extent of macrophage invasion correlates positively with the degree of myelin breakdown. The contribution of resident PNS macrophages to myelin clearance is thought to be minimal but important during early stages because myelin phagocytosis is evident at post-injury day 2 and haematogenous macrophage invasion does not commence until day 4. The contribution of resident macrophages may therefore be greater than they are credited with as they also undergo a significant period of proliferation after nerve injury (Mueller et al., 2001). This is supported by experiments where myelin removal from degenerating peripheral nerves occurs in the absence of macrophages. Whole-body irradiation can be used to prevent macrophage recruitment, and following peripheral nerve injury the rate of loss of MBP proceeds identically to that in non-irradiated mice for up to 5 days. Thereafter myelin removal is slower in macrophage-deprived nerves but complete removal from some Schwann cells is still possible. Macrophages therefore appear to have a limited role in early clearance of myelin, but accelerate its removal at later stages of degeneration (Perry et al., 1995). This mirrors events in the CNS where rapid proliferation of microglia peaks at day 2–3 post-injury, although their efficiency at myelin clearance is still significantly lower taking many weeks to reach completion (Lawson et al., 1994). We have clear evidence for macrophage engulfment of myelin debris, but what role they could play in axonal events is not something which has received much if any investigation. The time course of macrophage recruitment in the PNS and CNS would argue against any significant role because the degeneration of an injured axon proceeds well before significant tissue infiltration by other cell types. There remains the possibility of resident macrophage involvement but the accessibility of the axon itself, while still ensheathed, would surely prevent any significant interaction other than via the ensheathing cell. In trying to assess how macrophages might ‘sense’ and respond to axon degeneration we can learn some lessons from the macrophage and glial responses to axonal injury in Drosophila. This model organism exhibits classical Wallerian degeneration following axotomy and subsequent macrophage recruitment to degenerating axons of the CNS and PNS reveals the same split in recruitment and debris clearance rates (Aldskogius and Kozlova, 1998). A study by MacDonald et al. used the Drosophila olfactory system to further investigate the specificity and molecular requirements of the macrophage response to axonal degeneration. Axons of the olfactory receptor neurons (ORN) project to the antennal lobe of the brain, ablation of either the third antennal segments or maxillary palps will completely remove ORN cell bodies and fully transect the antennal or maxillary nerve, respectively. Genetic manipulation of this system is easy and allows marker proteins to be observed by driving coexpression of green fluorescent protein (GFP). Clearance of a specific protein can then be traced by removal of GFP-positive material. Monitoring the glial response

10

Axon Degeneration: Mechanisms and Consequences

199

to axotomy showed initial morphological changes very early in the degenerative process (day 1), and this involved an extension of the glial membrane towards the axon (MacDonald et al., 2006). The response was local and not accompanied by any increase in glial numbers. The specificity of glial responses was investigated by ablation of the maxillary palps instead of the third antennal segment. This reduces the number of degenerating ORNs from 600 to 60 and reduces the number of glomeruli that are innervated from 44/50 to 6 (Couto et al., 2005; Fishilevich and Vosshall, 2005). At day 1 post-injury glial membranes were enriched and Draper immunoreactivity was increased in six glomeruli in the ventro-medial antennal lobe which are assumed to correspond to those innervated by the maxillary nerve. The reaction is as early as 4 h post-injury and this correlates with the early signs of axon degeneration. The authors conclude that the transected axons trigger a selective recruitment of local glia by releasing a molecular signal. This signal is probably detected by Draper, the homolog of cell corpse engulfment gene ced-1 in Caenorhabditis elegans (Zhou et al., 2001), that is required for glial engulfment of apoptotic neuronal cells (Freeman et al., 2003). This theory was tested in Draper mutants, drpr5 and using RNAi downstream of draper, and was confirmed by the apparent blockade of glia recruitment to the injured axon. Interestingly, the upregulation of Draper immunoreactivity was suppressed in Wlds expressing Drosophila and this may account for the lack of glial recruitment in this mutant. The identity of the axonal signal, that is required for glial recognition and recruitment to an injured axon, can only be speculated. The signal is likely to be a short range and non-diffusible molecule, since only local glial are recruited leaving more distant populations unaffected even at later time points after injury. It must also be highly specific for the injured axons because neighbouring uninjured axons are not targeted by recruited glia, they only clear severed axons. Furthermore, Wlds expression blocks recruitment to severed axons but not to adjacent wild-type severed axons, implying a cell-autonomous level of protection. It is very possible that such a system of axon–glial signalling underlies recruitment of microglia to injured axons of the CNS and macrophages to axons of the PNS. Candidate glial receptors include MEGF10, which was identified by its sequence homology to ced-1 and draper and has since been validated as a ced-1-type engulfment receptor (Nagase et al., 2001; Hamon et al., 2006). Under normal conditions microglia are negative for macrophage scavenger receptors (MSR), but it can be induced by a variety of neuronal injuries (Bell et al., 1994). Ligands of the MSR include oxidised lowdensity lipoprotein, which can target macrophage uptake of myelin debris (Cuzner and Norton, 1996; Mosley and Cuzner, 1996), but potential early axonal signals are yet to be identified.

10.6 Concluding Remarks The most important message to take from this chapter is that neurodegeneration should be considered in terms of the individual compartments that are affected and not as a generic ‘neuronal death’. Each compartment, whether the cell body, axon or synapse is subject to its own unique environment. They are influenced by

200

L.J. Broom and V.H. Perry

interactions with different populations of supporting cells and have a particular internal homeostasis to maintain as well as individual vulnerabilities related to energy requirements and structure. It follows that each compartment can evoke a unique response to injury and this is evident as a particular spatial and temporal course of events associated with their degeneration. For example, the initial signal emanating from the site of an axonal injury results in signalling cascades, which transmit this message to more distal regions as well as to the cell body and results in the spread of structural disintegration within the confines of the distal region. These signalling cascades are co-ordinated in an axon autonomous manner and do not rely on direct communication with the cell body, they are also targeted to specific effector molecules in the axon since they do not trigger apoptosis of the cell body. The existence of specific control mechanisms and degenerative pathways for the cell body, axon and synapse is therefore a real possibility and future research should acknowledge this.

References Agthong S, Kaewsema A, Tanomsridejchai N, Chentanez V (2006) Activation of MAPK ERK in peripheral nerve after injury. BMC Neurosci 7:45 Aldskogius H, Kozlova EN (1998) Central neuron-glial and glial-glial interactions following axon injury. Prog Neurobiol 55(1):1–26 Anderson DR, Davis JL, Carraway KL (1977) Calcium-promoted changes of the human erythrocyte membrane. Involvement of spectrin, transglutaminase, and a membrane-bound protease. J Biol Chem 252(19):6617–6623 Andrews HE, Nichols PP, Bates D, Turnbull DM (2005) Mitochondrial dysfunction plays a key role in progressive axonal loss in Multiple Sclerosis. Med Hypotheses 64(4):669–677 Ankarcrona M, Dypbukt JM, Bonfoco E, Zhivotovsky B, Orrenius S, Lipton SA, Nicotera P (1995) Glutamate-induced neuronal death: a succession of necrosis or apoptosis depending on mitochondrial function. Neuron 15(4):961–973 Araki T, Sasaki Y, Milbrandt J (2004) Increased nuclear NAD biosynthesis and SIRT1 activation prevent axonal degeneration. Science 305(5686):1010–1013 Avellino AM, Hart D, Dailey AT, MacKinnon M, Ellegala D, Kliot M (1995) Differential macrophage responses in the peripheral and central nervous system during Wallerian degeneration of axons. Exp Neurol 136(2):183–198 Bagnard D, Sainturet N, Meyronet D, Perraut M, Miehe M, Roussel G, Aunis D, Belin MF, Thomasset N (2004) Differential MAP kinases activation during semaphorin3A-induced repulsion or apoptosis of neural progenitor cells. Mol Cell Neurosci 25(4):722–731 Bassell GJ, Zhang H, Byrd AL, Femino AM, Singer RH, Taneja KL, Lifshitz LM, Herman IM, Kosik KS (1998) Sorting of beta-actin mRNA and protein to neurites and growth cones in culture. J Neurosci 18(1):251–265 Beirowski B, Berek L, Adalbert R, Wagner D, Grumme DS, Addicks K, Ribchester RR, Coleman MP (2004) Quantitative and qualitative analysis of Wallerian degeneration using restricted axonal labelling in YFP-H mice. J Neurosci Methods 134(1):23–35 Bell MD, Lopez-Gonzalez R, Lawson L, Hughes D, Fraser I, Gordon S, Perry VH (1994) Upregulation of the macrophage scavenger receptor in response to different forms of injury in the CNS. J Neurocytol 23(10):605–613 Berg CP, Engels IH, Rothbart A, Lauber K, Renz A, Schlosser SF, Schulze-Osthoff K, Wesselborg S (2001) Human mature red blood cells express caspase-3 and caspase-8, but are devoid of mitochondrial regulators of apoptosis. Cell Death Differ 8(12):1197–1206

10

Axon Degeneration: Mechanisms and Consequences

201

Bignami A, Ralston HJ 3rd (1969) The cellular reaction to Wallerian degeneration in the central nervous system of the cat. Brain Res 13(3):444–461 Bishop DL, Misgeld T, Walsh MK, Gan WB, Lichtman JW (2004) Axon branch removal at developing synapses by axosome shedding. Neuron 44(4):651–661 Boas FE, Forman L, Beutler E (1998) Phosphatidylserine exposure and red cell viability in red cell aging and in hemolytic anemia. Proc Natl Acad Sci USA 95(6):3077–3081 Bratosin D, Leszczynski S, Sartiaux C, Fontaine O, Descamps J, Huart JJ, Poplineau J, Goudaliez F, Aminoff D, Montreuil J (2001) Improved storage of erythrocytes by prior leukodepletion: flow cytometric evaluation of stored erythrocytes. Cytometry 46(6):351–356 Bratton DL (1993) Release of platelet activation factor from activated neutrophils. Transglutaminase-dependent enhancement of transbilayer movement across the plasma membrane. J Biol Chem 268(5):3364–3373 Brittis PA, Lu Q, Flanagan JG (2002) Axonal protein synthesis provides a mechanism for localized regulation at an intermediate target. Cell 110(2):223–235 Bruck W (1997) The role of macrophages in Wallerian degeneration. Brain Pathol 7(2):741–752 Brugnara C, De Franceschi L, Alper SL (1993). Ca(2+)-activated K+ transport in erythrocytes. Comparison of binding and transport inhibition by scorpion toxins. J Biol Chem 268(12): 8760–8768 Buckmaster EA, Perry VH, Brown MC (1995) The rate of Wallerian degeneration in cultured neurons from wild type and C57BL/WldS mice depends on time in culture and may be extended in the presence of elevated K+ levels. Eur J Neurosci 7(7):1596–1602 Bunge MB (1973) Fine structure of nerve fibers and growth cones of isolated sympathetic neurons in culture. J Cell Biol 56(3):713–735 Buss A, Schwab ME (2003) Sequential loss of myelin proteins during Wallerian degeneration in the rat spinal cord. Glia 42(4):424–432 Campbell DS, Holt CE (2001). Chemotropic responses of retinal growth cones mediated by rapid local protein synthesis and degradation. Neuron 32(6):1013–1026 Campbell DS, Holt CE (2003) Apoptotic pathway and MAPKs differentially regulate chemotropic responses of retinal growth cones. Neuron 37(6):939–952 Campenot RB, Eng H (2000) Protein synthesis in axons and its possible functions. J Neurocytol 29(11–12):793–798 Carroll SL, Miller ML, Frohnert PW, Kim SS, Corbett JA (1997) Expression of neuregulins and their putative receptors, ErbB2 and ErbB3, is induced during Wallerian degeneration. J Neurosci 17(5):1642–1659 Cohen CM, Gascard P (1992) Regulation and post-translational modification of erythrocyte membrane and membrane-skeletal proteins. Semin Hematol 29(4):244–292 Coleman MP, Conforti L, Buckmaster EA, Tarlton A, Ewing RM, Brown MC, Lyon MF, Perry VH (1998). An 85-kb tandem triplication in the slow Wallerian degeneration (Wlds) mouse. Proc Natl Acad Sci USA 95(17):9985–9990 Conforti L, Fang G, Beirowski B, Wang MS, Sorci L, Asress S, Adalbert R, Silva A, Bridge K, Huang XP, Magni G, Glass JD, Coleman MP (2007) “NAD(+) and axon degeneration revisited: Nmnat1 cannot substitute for Wld(S) to delay Wallerian degeneration”. Cell Death Differ 14(1):116–127. Conforti L, Tarlton A, Mack TG, Mi W, Buckmaster EA, Wagner D, Perry VH, Coleman MP (2000) A Ufd2/D4Cole1e chimeric protein and overexpression of Rbp7 in the slow Wallerian degeneration (WldS) mouse. Proc Natl Acad Sci USA 97(21):11377–11382 Couto A, Alenius M, Dickson BJ (2005) Molecular, anatomical, and functional organization of the Drosophila olfactory system. Curr Biol 15(17):1535–1547 Craner MJ, Newcombe J, Black JA, Hartle C, Cuzner ML, Waxman SG (2004) Molecular changes in neurons in multiple sclerosis: altered axonal expression of Nav1.2 and Nav1.6 sodium channels and Na+/Ca2+ exchanger. Proc Natl Acad Sci USA 101(21): 8168–8173 Cuzner ML, Norton WT (1996) Biochemistry of demyelination. Brain Pathol 6(3):231–242

202

L.J. Broom and V.H. Perry

Dai RM, Li CC (2001) Valosin-containing protein is a multi-ubiquitin chain-targeting factor required in ubiquitin-proteasome degradation. Nat Cell Biol 3(8):740–744 Deshmukh M, Kuida K, Johnson EM Jr (2000) Caspase inhibition extends the commitment to neuronal death beyond cytochrome c release to the point of mitochondrial depolarization. J Cell Biol 150(1):131–143 Droz B, Leblond CP (1963) Axonal migration of proteins in the central nervous system and peripheral nerves as shown by radioautography. J Comp Neurol 121:325–346 Duranton C, Huber SM, Lang F (2002) Oxidation induces a Cl(–)-dependent cation conductance in human red blood cells. J Physiol 539(Pt 3):847–855 Eda S, Sherman IW (2002) Cytoadherence of malaria-infected red blood cells involves exposure of phosphatidylserine. Cell Physiol Biochem 12(5–6):373–384 Eng H, Lund K, Campenot RB (1999) Synthesis of beta-tubulin, actin, and other proteins in axons of sympathetic neurons in compartmented cultures. J Neurosci 19(1):1–9 Etzion Z, Tiffert T, Bookchin RM, Lew VL (1993) Effects of deoxygenation on active and passive Ca2+ transport and on the cytoplasmic Ca2+ levels of sickle cell anemia red cells. J Clin Invest 92(5):2489–2498 Fadok VA, Bratton DL, Rose DM, Pearson A, Ezekewitz RA, Henson PM (2000) A receptor for phosphatidylserine-specific clearance of apoptotic cells. Nature 405(6782):85–90 Fang C, Bernardes-Silva M, Coleman MP, Perry VH (2005) The cellular distribution of the Wld s chimeric protein and its constituent proteins in the CNS. Neuroscience 135(4):1107–1118 Ferreirinha F, Quattrini A, Pirozzi M, Valsecchi V, Dina G, Broccoli V, Auricchio A, Piemonte F, Tozzi G, Gaeta L, Casari G, Ballabio A, Rugarli EI (2004) Axonal degeneration in paraplegindeficient mice is associated with abnormal mitochondria and impairment of axonal transport. J Clin Invest 113(2):231–242 Ferri A, Sanes JR, Coleman MP, Cunningham JM, Kato AC (2003) Inhibiting axon degeneration and synapse loss attenuates apoptosis and disease progression in a mouse model of motoneuron disease. Curr Biol 13(8):669–673 Finn JT, Weil M, Archer F, Siman R, Srinivasan A, Raff MC (2000) Evidence that Wallerian degeneration and localized axon degeneration induced by local neurotrophin deprivation do not involve caspases. J Neurosci 20(4):1333–1341 Fishilevich E, Vosshall LB (2005) Genetic and functional subdivision of the Drosophila antennal lobe. Curr Biol 15(17):1548–1553 Forgue ST, Dahl JL (1978) The turnover rate of tubulin in rat brain. J Neurochem 31(5):1289–1297 Franco RS, Palascak M, Thompson H, Rucknagel DL, Joiner CH (1996) Dehydration of transferrin receptor-positive sickle reticulocytes during continuous or cyclic deoxygenation: role of KCl cotransport and extracellular calcium. Blood 88(11):4359–4365 Franson P, Ronnevi LO (1984) Myelin breakdown and elimination in the posterior funiculus of the adult cat after dorsal rhizotomy: a light and electron microscopic qualitative and quantitative study. J Comp Neurol 223(1):138–151 Freeman MR, Delrow J, Kim J, Johnson E, Doe CQ (2003) Unwrapping glial biology: Gcm target genes regulating glial development, diversification, and function. Neuron 38(4):567–580 Friede RL, Bischhausen R (1980) The fine structure of stumps of transected nerve fibers in subserial sections. J Neurol Sci 44(2–3):181–203 Friede RL, Bruck W (1993) Macrophage functional properties during myelin degradation. Adv Neurol 59:327–336 Garay R, Braquet P (1986) Involvement of K+ movements in the membrane signal induced by PAF-acether. Biochem Pharmacol 35(16):2811–2815 Garcia ML, Cleveland DW (2001) Going new places using an old MAP: tau, microtubules and human neurodegenerative disease. Curr Opin Cell Biol 13(1):41–48 George EB, Glass JD, Griffin JW (1995) Axotomy-induced axonal degeneration is mediated by calcium influx through ion-specific channels. J Neurosci 15(10):6445–6452 George R, Griffin JW (1994) The proximo-distal spread of axonal degeneration in the dorsal columns of the rat. J Neurocytol 23(11):657–667

10

Axon Degeneration: Mechanisms and Consequences

203

Goodrum KJ, McCormick LL, Schneider B (1994) Group B streptococcus-induced nitric oxide production in murine macrophages is CR3 (CD11b/CD18) dependent. Infect Immun 62(8):3102–3107 Griffin JW, George R, Ho T (1993) Macrophage systems in peripheral nerves. A review. J Neuropathol Exp Neurol 52(6):553–560 Griffin JW, George R, Lobato C, Tyor WR, Yan LC, Glass JD (1992) Macrophage responses and myelin clearance during Wallerian degeneration: relevance to immune-mediated demyelination. J Neuroimmunol 40(2–3):153–165 Griffin JW, Price DL, Engel WK, Drachman DB (1977) The pathogenesis of reactive axonal swellings: role of axonal transport. J Neuropathol Exp Neurol 36(2):214–227 Hamon Y, Trompier D, Ma Z, Venegas V, Pophillat M, Mignotte V, Zhou Z, Chimini G (2006) Cooperation between engulfment receptors: the case of ABCA1 and MEGF10. PLoS ONE 1:e120 Hanz S, Perlson E, Willis D, Zheng JQ, Massarwa R, Huerta JJ, Koltzenburg M, Kohler M, van-Minnen J, Twiss JL, Fainzilber M (2003) Axoplasmic importins enable retrograde injury signaling in lesioned nerve. Neuron 40(6):1095–1104 Hatakeyama S, Yada M, Matsumoto M, Ishida N, Nakayama KI (2001) U box proteins as a new family of ubiquitin-protein ligases. J Biol Chem 276(35):33111–33120 Hoopfer ED, McLaughlin T, Watts RJ, Schuldiner O, O’Leary DD, Luo L (2006) Wlds protection distinguishes axon degeneration following injury from naturally occurring developmental pruning. Neuron 50(6):883–895 Huber SM, Gamper N, Lang F (2001) Chloride conductance and volume-regulatory nonselective cation conductance in human red blood cell ghosts. Pflugers Arch 441(4):551–558 Kaestner L, Christophersen P, Bernhardt I, Bennekou P (2000) The non-selective voltage-activated cation channel in the human red blood cell membrane: reconciliation between two conflicting reports and further characterisation. Bioelectrochemistry 52(2):117–125 Kidd T, Brose K, Mitchell KJ, Fetter RD, Tessier-Lavigne M, Goodman CS, Tear G (1998) Roundabout controls axon crossing of the CNS midline and defines a novel subfamily of evolutionarily conserved guidance receptors. Cell 92(2):205–215 Kitada T, Asakawa S, Hattori N, Matsumine H, Yamamura Y, Minoshima S, Yokochi M, Mizuno Y, Shimizu N (1998) Mutations in the parkin gene cause autosomal recessive juvenile parkinsonism. Nature 392(6676):605–608 Koegl M, Hoppe T, Schlenker S, Ulrich HD, Mayer TU, Jentsch S (1999) A novel ubiquitination factor, E4, is involved in multiubiquitin chain assembly. Cell 96(5):635–644 Koenig E (1979) Ribosomal RNA in Mauthner axon: implications for a protein synthesizing machinery in the myelinated axon. Brain Res 174(1):95–107 Koenig E, Martin R, Titmus M, Sotelo-Silveira JR (2000) Cryptic peripheral ribosomal domains distributed intermittently along mammalian myelinated axons. J Neurosci 20(22):8390–8400 Kuhlmann T, Wendling U, Nolte C, Zipp F, Maruschak B, Stadelmann C, Siebert H, Bruck W (2002) Differential regulation of myelin phagocytosis by macrophages/microglia, involvement of target myelin, Fc receptors and activation by intravenous immunoglobulins. J Neurosci Res 67(2):185–190 Lang PA, Kempe DS, Tanneur V, Eisele K, Klarl BA, Myssina S, Jendrossek V, Ishii S, Shimizu T, Waidmann M, Hessler G, Huber SM, Lang F, Wieder T (2005) Stimulation of erythrocyte ceramide formation by platelet-activating factor. J Cell Sci 118(Pt 6):1233–1243 Laser H, Conforti L, Morreale G, Mack TG, Heyer M, Haley JE, Wishart TM, Beirowski B, Walker SA, Haase G, Celik A, Adalbert R, Wagner D, Grumme D, Ribchester RR, Plomann M, Coleman MP (2006) The slow Wallerian degeneration protein, WldS, binds directly to VCP/p97 and partially redistributes it within the nucleus. Mol Biol Cell 17(3):1075–1084 Lawson LJ, Frost L, Risbridger J, Fearn S, Perry VH (1994) Quantification of the mononuclear phagocyte response to Wallerian degeneration of the optic nerve. J Neurocytol 23(12):729–744 Lee SK, Hollenbeck PJ (2003) Organization and translation of mRNA in sympathetic axons. J Cell Sci 116(Pt 21):4467–4478

204

L.J. Broom and V.H. Perry

Lindwall C, Kanje M (2005) Retrograde axonal transport of JNK signaling molecules influence injury induced nuclear changes in p-c-Jun and ATF3 in adult rat sensory neurons. Mol Cell Neurosci 29(2):269–282 Litman P, Barg J, Rindzoonski L, Ginzburg I (1993) Subcellular localization of tau mRNA in differentiating neuronal cell culture: implications for neuronal polarity. Neuron 10(4):627–638 Liu HM, Yang LH, Yang YJ (1995) Schwann cell properties: 3. C-fos expression, bFGF production, phagocytosis and proliferation during Wallerian degeneration. J Neuropathol Exp Neurol 54(4):487–496 LoPachin RM, Jr, LoPachin VR, Saubermann AJ (1990) Effects of axotomy on distribution and concentration of elements in rat sciatic nerve. J Neurochem 54(1):320–332 Lowe J, Mayer RJ, Landon M (1993) Ubiquitin in neurodegenerative diseases. Brain Pathol 3(1):55–65 Ludwin SK (1990) Oligodendrocyte survival in Wallerian degeneration. Acta Neuropathol (Berl) 80(2):184–191 Lunn ER, Perry VH, Brown MC, Rosen H, Gordon S (1989) Absence of Wallerian Degeneration does not Hinder Regeneration in Peripheral Nerve. Eur J Neurosci 1(1):27–33 Lyon MF, Ogunkolade BW, Brown MC, Atherton DJ, Perry VH (1993) A gene affecting Wallerian nerve degeneration maps distally on mouse chromosome 4. Proc Natl Acad Sci USA 90(20):9717–9720 MacDonald JM, Beach MG, Porpiglia E, Sheehan AE, Watts RJ, Freeman MR (2006) The Drosophila cell corpse engulfment receptor Draper mediates glial clearance of severed axons. Neuron 50(6):869–881 Mack TG, Reiner M, Beirowski B, Mi W, Emanuelli M, Wagner D, Thomson D, Gillingwater T, Court F, Conforti L, Fernando FS, Tarlton A, Andressen C, Addicks K, Magni G, Ribchester RR, Perry VH, Coleman MP (2001) Wallerian degeneration of injured axons and synapses is delayed by a Ube4b/Nmnat chimeric gene. Nat Neurosci 4(12):1199–1206 Martini R, Schachner M (1988) Immunoelectron microscopic localization of neural cell adhesion molecules (L1, N-CAM, and myelin-associated glycoprotein) in regenerating adult mouse sciatic nerve. J Cell Biol 106(5):1735–1746 Meller D, Eysel UT, Schmidt-Kastner R (1994) Transient immunohistochemical labelling of rat retinal axons during Wallerian degeneration by a monoclonal antibody to neurofilaments. Brain Res 648(1):162–166 Meyer M, Matsuoka I, Wetmore C, Olson L, Thoenen H (1992) Enhanced synthesis of brainderived neurotrophic factor in the lesioned peripheral nerve: different mechanisms are responsible for the regulation of BDNF and NGF mRNA. J Cell Biol 119(1):45–54 Minetti G, Piccinini G, Balduini C, Seppi C, Brovelli A (1996) Tyrosine phosphorylation of band 3 protein in Ca2+/A23187-treated human erythrocytes. Biochem J 320(Pt 2):445–450 Mosley K, Cuzner ML (1996) Receptor-mediated phagocytosis of myelin by macrophages and microglia: effect of opsonization and receptor blocking agents. Neurochem Res 21(4): 481–487 Mueller BK (1999) Growth cone guidance: first steps towards a deeper understanding. Annu Rev Neurosci 22:351–388 Mueller M, Wacker K, Ringelstein EB, Hickey WF, Imai Y, Kiefer R (2001) Rapid response of identified resident endoneurial macrophages to nerve injury. Am J Pathol 159(6): 2187–2197 Mukoyama M, Yamazaki K, Kikuchi T, Tomita T (1989) Neuropathology of gracile axonal dystrophy (GAD) mouse. An animal model of central distal axonopathy in primary sensory neurons. Acta Neuropathol (Berl) 79(3):294–299 Nagase T, Nakayama M, Nakajima D, Kikuno R, Ohara O (2001) Prediction of the coding sequences of unidentified human genes. XX. The complete sequences of 100 new cDNA clones from brain which code for large proteins in vitro. DNA Res 8(2):85–95 Nixon RA (1980) Protein degradation in the mouse visual system. I. Degradation of axonally transported and retinal proteins. Brain Res 200(1):69–83

10

Axon Degeneration: Mechanisms and Consequences

205

Nunomura W, Takakuwa Y, Tokimitsu R, Krauss SW, Kawashima M, Mohandas N (1997) Regulation of CD44-protein 4.1 interaction by Ca2+ and calmodulin. Implications for modulation of CD44-ankyrin interaction. J Biol Chem 272(48):30322–30328 Pan YA, Misgeld T, Lichtman JW, Sanes JR (2003) Effects of neurotoxic and neuroprotective agents on peripheral nerve regeneration assayed by time-lapse imaging in vivo. J Neurosci 23(36):11479–11488 Pannese E, Ledda M (1991) Ribosomes in myelinated axons of the rabbit spinal ganglion neurons. J Submicrosc Cytol Pathol 23(1):33–38 Parson SH, Mackintosh CL, Ribchester RR (1997) Elimination of motor nerve terminals in neonatal mice expressing a gene for slow Wallerian degeneration (C57Bl/Wlds. Eur J Neurosci 9(8):1586–1592 Perlson E, Hanz S, Ben-Yaakov K, Segal-Ruder Y, Seger R, Fainzilber M (2005) Vimentindependent spatial translocation of an activated MAP kinase in injured nerve. Neuron 45(5): 715–726 Perlson E, Hanz S, Medzihradszky KF, Burlingame AL, Fainzilber M (2004) From snails to sciatic nerve: retrograde injury signaling from axon to soma in lesioned neurons. J Neurobiol 58(2):287–294 Perry VH, Brown MC (1992) Macrophages and nerve regeneration. Curr Opin Neurobiol 2(5): 679–682 Perry VH, Brown MC, Gordon S (1987) The macrophage response to central and peripheral nerve injury. A possible role for macrophages in regeneration. J Exp Med 165(4):1218–1223 Perry VH, Tsao JW, Fearn S, Brown MC (1995) Radiation-induced reductions in macrophage recruitment have only slight effects on myelin degeneration in sectioned peripheral nerves of mice. Eur J Neurosci 7(2):271–280 Posse de Chaves E, Vance DE, Campenot RB, Vance JE (1995) Axonal synthesis of phosphatidylcholine is required for normal axonal growth in rat sympathetic neurons. J Cell Biol 128(5):913–918 Raff MC, Whitmore AV, Finn JT (2002) Axonal self-destruction and neurodegeneration. Science 296(5569):868–871 Raivich G, Hellweg R, Kreutzberg GW (1991) NGF receptor-mediated reduction in axonal NGF uptake and retrograde transport following sciatic nerve injury and during regeneration. Neuron 7(1):151–164 Ramon y Cajal, S. (1928) Degeneration and regeneration of the nervous system. Volume 2. Haffner Publishing Co. New York, New York, USA Ray SK, Matzelle DD, Sribnick EA, Guyton MK, Wingrave JM, Banik NL (2003) Calpain inhibitor prevented apoptosis and maintained transcription of proteolipid protein and myelin basic protein genes in rat spinal cord injury. J Chem Neuroanat 26(2):119–124 Reichert F, Rotshenker S (1996) Deficient activation of microglia during optic nerve degeneration. J Neuroimmunol 70(2):153–161 Rivera A, Jarolim P, Brugnara C (2002) Modulation of Gardos channel activity by cytokines in sickle erythrocytes. Blood 99(1):357–603 Saido TC, Nagao S, Shiramine M, Tsukaguchi M, Yoshizawa T, Sorimachi H, Ito H, Tsuchiya T, Kawashima S, Suzuki K (1994) Distinct kinetics of subunit autolysis in mammalian m-calpain activation. FEBS Lett 346(2–3):263–267 Saigoh K, Wang YL, Suh JG, Yamanishi T, Sakai Y, Kiyosawa H, Harada T, Ichihara N, Wakana S, Kikuchi T, Wada K (1999) Intragenic deletion in the gene encoding ubiquitin carboxy-terminal hydrolase in gad mice. Nat Genet 23(1):47–51 Sajadi A, Schneider BL, Aebischer P (2004) Wlds-mediated protection of dopaminergic fibers in an animal model of Parkinson disease. Curr Biol 14(4):326–330 Samsam M, Mi W, Wessig C, Zielasek J, Toyka KV, Coleman MP, Martini R (2003) The Wlds mutation delays robust loss of motor and sensory axons in a genetic model for myelin-related axonopathy. J Neurosci 23(7):2833–2839

206

L.J. Broom and V.H. Perry

Schaecher KE, Shields DC, Banik NL (2001) Mechanism of myelin breakdown in experimental demyelination: a putative role for calpain. Neurochem Res 26(6):731–737 Schmied R, Ambron RT (1997) A nuclear localization signal targets proteins to the retrograde transport system, thereby evading uptake into organelles in aplysia axons. J Neurobiol 33(2):151–160 Seigneuret M, Devaux PF (1984) ATP-dependent asymmetric distribution of spin-labeled phospholipids in the erythrocyte membrane: relation to shape changes. Proc Natl Acad Sci USA 81(12):3751–3755 Sheu JY, Kulhanek DJ, Eckenstein FP (2000) Differential patterns of ERK and STAT3 phosphorylation after sciatic nerve transection in the rat. Exp Neurol 166(2):392–402 Shields DC, Banik NL (1999) Pathophysiological role of calpain in experimental demyelination. J Neurosci Res 55(5):533–541 Shields DC, Tyor WR, Deibler GE, Banik NL (1998) Increased calpain expression in experimental demyelinating optic neuritis: an immunocytochemical study. Brain Res 784(1–2):299–304 Shimura H, Hattori N, Kubo S, Mizuno Y, Asakawa S, Minoshima S, Shimizu N, Iwai K, Chiba T, Tanaka K, Suzuki T (2000) Familial Parkinson disease gene product, parkin, is a ubiquitinprotein ligase. Nat Genet 25(3):302–305 Sievers C, Platt N, Perry VH, Coleman MP, Conforti L (2003) Neurites undergoing Wallerian degeneration show an apoptotic-like process with Annexin V positive staining and loss of mitochondrial membrane potential. Neurosci Res 46(2):161–169 Smith SK, Farnbach AR, Harris FM, Hawes AC, Jackson LR, Judd AM, Vest RS, Sanchez S, Bell JD (2001) Mechanisms by which intracellular calcium induces susceptibility to secretory phospholipase A2 in human erythrocytes. J Biol Chem 276(25):22732–22741 Steward O, Ribak CE (1986) Polyribosomes associated with synaptic specializations on axon initial segments: localization of protein-synthetic machinery at inhibitory synapses. J Neurosci 6(10):3079–3085 Stoll G, Griffin JW, Li CY, Trapp BD (1989) Wallerian degeneration in the peripheral nervous system: participation of both Schwann cells and macrophages in myelin degradation. J Neurocytol 18(5):671–683 Stoll G, Jander S (1999) The role of microglia and macrophages in the pathophysiology of the CNS. Prog Neurobiol 58(3):233–247 Tanaka T, Kadowaki K, Lazarides E, Sobue K (1991) Ca2(+)-dependent regulation of the spectrin/actin interaction by calmodulin and protein 4.1. J Biol Chem 266(2): 1134–1140 Tennyson VM (1970) The fine structure of the axon and growth cone of the dorsal root neuroblast of the rabbit embryo. J Cell Biol 44(1):62–79 Twiss JL, Smith DS, Chang B, Shooter EM (2000) Translational control of ribosomal protein L4 mRNA is required for rapid neurite regeneration. Neurobiol Dis 7(4):416–428 Verma P, Chierzi S, Codd AM, Campbell DS, Meyer RL, Holt CE, Fawcett JW (2005) Axonal protein synthesis and degradation are necessary for efficient growth cone regeneration. J Neurosci 25(2):331–342 Waller A (1850) Experiments on the section of glossopharyngeal and hypoglossal nerves of the frog and observations of the alternatives produced thereby in the structure of their primitive fibres. Philos Trans R Soc Lond Biol 140:423 Wang AL, Yuan M, Neufeld AH (2006) Degeneration of neuronal cell bodies following axonal injury in Wld(S) mice. J Neurosci Res 84(8):1799–1807 Wang J, Zhai Q, Chen Y, Lin E, Gu W, McBurney MW, He Z (2005) A local mechanism mediates NAD-dependent protection of axon degeneration. J Cell Biol 170(3):349–355 Wang MS, Davis AA, Culver DG, Glass JD (2002) WldS mice are resistant to paclitaxel (taxol) neuropathy. Ann Neurol 52(4):442–447 Watts RJ, Hoopfer ED, Luo L (2003) Axon pruning during Drosophila metamorphosis: evidence for local degeneration and requirement of the ubiquitin-proteasome system. Neuron 38(6): 871–885

10

Axon Degeneration: Mechanisms and Consequences

207

Wellmann H, Kaltschmidt B, Kaltschmidt C (2001) Retrograde transport of transcription factor NF-kappa B in living neurons. J Biol Chem 276(15):11821–11829 Willis D, Li KW, Zheng JQ, Chang JH, Smit A, Kelly T, Merianda TT, Sylvester J, van Minnen J, Twiss JL (2005) Differential transport and local translation of cytoskeletal, injury-response, and neurodegeneration protein mRNAs in axons. J Neurosci 25(4):778–791 Xie XY, Barrett JN (1991) Membrane resealing in cultured rat septal neurons after neurite transection: evidence for enhancement by Ca(2+)-triggered protease activity and cytoskeletal disassembly. J Neurosci 11(10):3257–3367 Yamagishi S, Fujitani M, Hata K, Kitajo K, Mimura F, Abe H, Yamashita T (2005) Wallerian degeneration involves Rho/Rho-kinase signaling. J Biol Chem 280(21):20384–20388 Yamashima T (2000) Implication of cysteine proteases calpain, cathepsin and caspase in ischemic neuronal death of primates. Prog Neurobiol 62(3):273–295 Yi XN, Zheng LF, Zhang JW, Zhang LZ, Xu YZ, Luo G, Luo XG (2006) Dynamic changes in Robo2 and Slit1 expression in adult rat dorsal root ganglion and sciatic nerve after peripheral and central axonal injury. Neurosci Res 56(3):314–321 Zelena J (1970) Ribosome-like particles in myelinated axons of the rat. Brain Res 24(2):359–363 Zelena J, Lubinska L, Gutmann E (1968) Accumulation of organelles at the ends of interrupted axons. Z Zellforsch Mikrosk Anat 91(2):200–219 Zhai Q, Wang J, Kim A, Liu Q, Watts R, Hoopfer E, Mitchison T, Luo L, He Z (2003) Involvement of the ubiquitin-proteasome system in the early stages of Wallerian degeneration. Neuron 39(2):217–225 Zhang XP, Ambron RT (2000) Positive injury signals induce growth and prolong survival in Aplysia neurons. J Neurobiol 45(2):84–94 Zheng JQ, Kelly TK, Chang B, Ryazantsev S, Rajasekaran AK, Martin KC, Twiss JL (2001) A functional role for intra-axonal protein synthesis during axonal regeneration from adult sensory neurons. J Neurosci 21(23):9291–9303 Zhou J, Liao J, Xing C (1997) Determination of changes in membrane phospholipids of rat lung and liver tissues from acute heat stress by high performance liquid chromatography. Se Pu 15(1):77–78 Zhou Q, Zhao J, Wiedmer T, Sims PJ (2002) Normal hemostasis but defective hematopoietic response to growth factors in mice deficient in phospholipid scramblase 1. Blood 99(11):4030–4038 Zhou XF, Chie ET, Deng YS, Zhong JH, Xue Q, Rush RA, Xian CJ (1999) Injured primary sensory neurons switch phenotype for brain-derived neurotrophic factor in the rat. Neuroscience 92(3):841–853 Zhou Z, Hartwieg E, Horvitz HR (2001) CED-1 is a transmembrane receptor that mediates cell corpse engulfment in C. elegans. Cell 104(1):43–56

Chapter 11

Regeneration After CNS Lesion: Help from the Immune System? Sven Hendrix and Robert Nitsch

11.1 Introduction Mechanical injury to the central nervous system (CNS), e.g., during surgical interventions in the brain is followed by an inflammatory response, which is a multi-phase process orchestrated by a complex network of immune and CNS cells interlinked by reciprocal signaling. Oversimplified, the CNS response is characterized by a quick, highly controlled burst of acute inflammatory defense followed by a long-term remodeling phase. There is increasing evidence that immune cells and inflammatory mediators play differential roles in the acute versus chronic phases after CNS injury (Fabry et al., 2008). Interestingly, at least one or two phases of T-cell infiltration have been described in different CNS trauma models suggesting differential functions of T cells in the acute and remodeling phase (Hendrix and Nitsch, 2007). Not surprisingly, a long-standing debate continues regarding the functional role of these invading T cells in neuronal damage and/or repair responses of the injured CNS. Here, we briefly review the acute defense and chronic repair processes of the CNS after traumatic injury, as well as the contradictory results regarding the beneficial or detrimental effects of T cells in the treatment of traumatic CNS damage. We also summarize as yet unanswered questions in relation to the particularly beneficial role of a specific T-cell subpopulation, namely T-helper cells type 2 (Th2 cells), in the context of CNS lesions. Finally, we will briefly review potential mechanisms that may explain beneficial T-cell effects during CNS wound healing.

S. Hendrix (B) Department of Functional Morphology & BIOMED Institute, Hasselt University – Campus Diepenbeek, Agoralaan Gebouw C, BE 3590, DIEPENBEEK, Belgium e-mail: [email protected]

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1_11, 

209

210

S. Hendrix and R. Nitsch

11.2 Wound Healing Responses of the CNS 11.2.1 Inflammation in Autoimmune Neurodegeneration Versus CNS Repair After Trauma To understand the conflicting reports on T-cell effects in the CNS during pathological processes, it is crucial to differentiate between the etiopathogenesis of autoimmune diseases and the regenerative remodeling of the CNS following mechanical damage. During autoimmune processes such as multiple sclerosis (MS), T cells represent major causative agents of the disease. In contrast, mechanical damage such as surgical interventions in brain and spinal cord leads to substantial inflammatory responses (Babcock et al., 2003; Chen et al., 2003; Schnell et al., 1999) (Fig. 11.1), which have many of the characteristics as classical wound healing responses, e.g., in the skin (Park and Barbul, 2004). Thus, in contrast to an autoimmune context, in which inflammation is considered to be primarily detrimental, in a repair context inflammation may be an important part of CNS regeneration. However, it is feasible to expect that – in addition to the T-cell-mediated immune attack – regenerative responses such as regulatory immune reactions and axonal regeneration play a major role, e.g., in the remission phase of autoimmune CNS diseases such as MS and its animal models.

Traumatic Insult Damaged cells Release of ATP, glutamate, + 2+ K ,Ca , HSPs Disturbed ion homeostasis Impaired energy metabolism

BBB breakdown

Resting microglia

Quiescent astroctyes

Chronic Remodeling Reactive gliosis / glial scar formation

Leukocyte infiltration

Proinflammatory mediators/ mediators of secondary injury e.g. IL-1β, IL-6, TNFa, ROS

Neutrophils Activation Macrophages Mediator Activation, release Proliferation Phagocytosis Mediator release T cells AG recognition/ Activation Mediator release Protective Autoimmunity ?

Activation

Resting endothelial cells

Acute Defense

Activated endothelial cells BBB permability ↑ Adhesion molecules ↑ Mediator release Activated microglia / infiltrating macrophages Migration, Proliferation Phagocytosis, MHC ↑ Mediator release Activated astrocytes Migration, Proliferation intermediate filaments ↑

Reactive astrocytes, reactive microglia/ macrophages reactive OPCs, fibroblasts, meningeal cells, ECM

Reestablishment of BBB integrity Trophic support Limitation of secondary injury

Neuronal survival

Production of growth Inhibiting ECM Physical barrier for neuronal outgrowth

Neuronal regeneration

Antiinflammatory mediators neurotrophic factors, growth factors e.g. IL-4, IL-10, IL-13, BDNF

Fig. 11.1 Key cellular and biochemical players during inflammatory and remodeling phases after traumatic injury to the central nervous system

11

T-Helper Cells and Neuroregeneration

211

11.2.2 Multi-phase Responses of the CNS After Trauma The T-cell response during and after traumatic injuries to the CNS is part of a multi-phase process of acute defense followed by long-term remodeling processes orchestrated by a complex network of immune and CNS cells interlinked by reciprocal signaling. In order to develop preliminary concepts of how T-cell-based therapies may exert beneficial effects, a more thorough understanding of the complex interactions of immune cells during the different phases of acute and chronic CNS reactions to injury is needed. There is increasing evidence that in the acute phase of traumatic injury, the increased access of CD4+ T cells, innate mediators, and complement components into the CNS may be harmful, while these cells and factors may exert beneficial effects in the chronic repair processes (Fabry et al., 2008). Thus, one major step to develop T-cell-based therapies is to understand the regulatory switches from the highly acute phase toward the chronic remodeling phase. However, it is important to keep in mind that the acute and the chronic phases of the CNS response to injury are composed of several subphases of differential immune and astroglial reactions creating a big challenge to find the therapeutic window for specific treatments. Although equivalent inflammatory stimuli elicit a much weaker response in the CNS compared to non-neuronal tissues, i.e., inflammation is restricted in space and magnitude (Andersson et al., 1992; Galea et al., 2007; Schnell et al., 1997), the CNS shares key features of wound healing responses with many other mammalian tissues. CNS repair involves not only CNS resident cells but also a broad variety of non-CNS cell populations. The recruitment, activation, and action of these participants are orchestrated by the elaborately regulated spatio-temporal expression and/or release of cell-surface molecules, cytokines, chemokines, and other pro- and anti-inflammatory mediators, as well as their appropriate receptors, resulting in an extraordinarily complex network of interlinked and reciprocal signaling (Fig. 11.1), which is far from completely understood. In most trauma models there is a specific timing of immune cell infiltration and activation after trauma. A simplified synopsis of the timing of immune cell invasion and astroglial activation after a stab wound in the brain of mice is represented in Fig. 11.2. The complex temporal correlation between inflammatory cascades, secondary neurodegenerative events, and functional recovery in rodents after spinal cord injury has been comprehensively summarized in a previous review (Donnelly and Popovich, 2008) including anatomical and functional outcomes (such as de- and remyelination, axonal sprouting/plasticity, and locomotor recovery), microglia activation, intraspinal leukocyte accumulation, expression of proinflammatory cytokines, reactive oxygen species, neurotrophic factors and blood–brain barrier (BBB) permeability. In mechanical CNS lesions, the initial traumatic insult causes immediate cell necrosis, the level of which depends on impact severity, cell dysfunction of the injured cells, and leads to disruption of the blood–brain barrier (BBB) integrity. This results in the release of intracellular components from necrotic neurons and glial cells, disturbances of extracellular ion homeostasis and the entrance of blood

212

S. Hendrix and R. Nitsch

Platelet Activated astrocyte

Relative cell number

Neutrophil

Mikroglia/ Macrophage T cell

0

1

2

3

4

5 6 7 8 9 10 11 12 13 14 Days after Brain Injury

Fig. 11.2 Immune cell invasion and astroglial activation after brain trauma in mice

components physiologically excluded from CNS, all of which can trigger microglial and astrocytic activation and rapid inflammatory mediator release (Nicole et al., 2005; Illes et al., 1996; Kurpius et al., 2007; Weinstein et al., 2005). Furthermore, endothelial cells becomes activated and enhance expression of adhesion molecules necessary for leukocyte recruitment. They increase BBB permeability and can release vasoactive and proinflammatory mediators such as nitric oxide and endothelin-1, further influencing local metabolic impairment, glial activation, and leukocyte infiltration. In turn, rapidly synthesized inflammatory mediators from activated microglia, astrocytes, and invading leukocytes may perpetuate glial activation in an autocrine/paracrine manner, enhance BBB permeability and endothelial adhesion molecule expression and leukocyte chemotaxis, thus amplifying the inflammatory response. This process takes place very rapidly, e.g., significant microglial IL-1β production has been shown to take place as early as 15 min after CNS injury (Herx and Yong, 2001). Given the often redundant and pleiotrophic nature of the possible initiating factors involved, it is difficult to establish a clear causative cascade of glial activation after CNS injury. Shortly after a traumatic insult there is a prominent glial activation and increase in a variety of proinflammatory cytokines, chemokines, and cell adhesion molecules necessary for leukocyte recruitment in most experimental CNS lesion paradigms, resulting in a substantial postlesional CNS inflammation (Babcock et al., 2003; Ahn et al., 2006; Bye et al., 2007; Hang et al., 2005; Quintana et al., 2005; Rostworowski et al., 1997; Schnell et al., 1999). But despite many similarities, some substantial differences in wound healing responses exist between species, strains,

11

T-Helper Cells and Neuroregeneration

213

and lesion paradigms, especially regarding postlesional inflammatory phenotype (Schnell et al., 1999; Campbell et al., 2002; Fitch et al., 1999; Sroga et al., 2003). Rapidly after traumatic CNS insult an increasing number of resident microglia around the lesion site become activated, being particularly sensitive to microenvironmental disturbances (review in Kreutzberg, 1996). They alter their cellular morphology, becoming less ramified and/or transforming into an amoeboid phagocytic phenotype. Concomitantly, profound changes in gene expression take place, resulting in the upregulation of cell-surface molecules, such as CD11b, MHC antigens, receptors for various cytokines and chemokines, as well as production of a large variety of proinflammatory mediators, such as IL-1β, IL-6, TNF-α, TGF-β, reactive oxygen species, nitric oxide, and arachidonic-acid intermediates, among many others (reviewed in (Streit et al., 1999; Hanisch, 2002)). Activated microglia and infiltrating peripheral macrophages migrate to the lesion site and proliferate. Depending on the lesion paradigm they display a peak proliferation rate at 2–5 days after injury, reaching maximum quantity at about 7–14 days, slowly decreasing thereafter, and persisting in elevated numbers for up to 1 month (Vela et al., 2002; Hakamada-Taguchi et al., 2003; Hampton et al., 2004; Holmin et al., 1995; Ma et al., 2004). The microglia/macrophage action in CNS wound healing response has historically been considered to be a “double-edged sword,” eliciting both neurodegenerative and neuroprotective/pro-regenerative effects. On the one hand, released microglial proinflammatory mediators have been shown to be able to directly and indirectly damage and kill neighboring cells like neurons and oligodendrocytes, suggesting a deleterious role in postlesional CNS inflammation (Boje and Arora, 1992; Gibbons and Dragunow, 2006; He et al., 2002; Merrill et al., 1993; Munch et al., 2003). Other studies have failed to detect neurotoxic effects of microglia but rather suggest neuroprotective and regenerative functions by the release of neurotrophic factors or the modulation of astrocytic function (Nagata et al., 1993; Del Rio et al., 1997; Eskes et al., 2002). However, many of the underlying studies have been carried out in vitro and the respective conclusions do not necessarily apply on in vivo conditions. Reactive microglia are not a homogeneous cell population and respond differently to varying stimulating conditions, which have led some authors to propose a graded microglial response (review in Streit et al., 1999). Indeed, a previous study supports the idea of a graded response and suggests that the changed balance of released factors resulting from this gradual activation might contribute to the switch from a neuroprotective to a neurotoxic phenotype (Schneider et al., 2007). However, more work is needed to correlate the differential effects of invading macrophages and microglial cells with the differential phases of the acute and chronic CNS response to trauma and to determine those factors and neuroimmune interactions that switch these cells from an acute proinflammatory and/or degenerative phenotype to a pro-regenerative or wound healing-like phenotype. Thus, the modulating influence of the phase-specific micro-environmental conditions, as well as the distinct properties of microglia/macrophage subpopulations, might account for the controversial results obtained in different experimental in vivo approaches aiming at elucidating microglial/macrophage influence on

214

S. Hendrix and R. Nitsch

neuronal degeneration and regeneration, respectively. For example, transferring activated autologous macrophages into lesioned CNS tissue and depletion of hematogenous macrophages prior to CNS lesion showed markedly improved neuronal regeneration. In contrast, the non-traumatic activation of microglia/macrophages in CNS tissue provoked a clear neurodegenerative pathology (Franzen et al., 1998; Jones et al., 2002; Moalem et al., 1999; Popovich et al., 1999). Another hypothesis is based on the assumption that microglial function in the CNS, in accordance with those of other glial cells, is primarily homeostatic, aiming at the perseverance of CNS integrity and function. Here, injury-induced microglial neurotoxicity is not thought to represent non-specific “bystander damage” but rather a purposeful, physiological removal of neurons compromised in function and viability (reviewed in Streit, 2002). At a slightly later time point than microglia, astrocytes begin appreciably responding to the activating stimuli. Microglial IL-1β release has been implicated as one of the major factors initiating early astrocyte activation after CNS injury (Herx and Yong, 2001). Reactive astrocytes become hypertrophic, migrate to the lesion site, and also proliferate to a certain extent. Like activated microglia, they also significantly alter gene expression, resulting in the upregulation of intermediate filament proteins, GFAP, vimentin, and a number of cell-surface molecules, in the synthesis and release of proinflammatory mediators like TNF-α, IL-1β, IL-6, MCP-1, NO, and in an increased production of growth factors and neurotrophic factors, such as NGF, BDNF, bFGF, and CNTF (review Ridet et al., 1997; Eddleston and Mucke, 1993; Liberto et al., 2004). Within a few days they wall off the lesion site by creating a dense cell plexus with interdigitated processes morphologically described as anisomorphic gliosis. However, it should be kept in mind that reactive astrocytes display a considerable heterogeneity in response patterns to CNS lesions. Astrocytes distal from the lesion site or in lesion paradigms with less disturbances like denervating lesion paradigms with subsequent axonal and dendritic degeneration are also activated in terms of hypertrophy and increased GFAP synthesis but remain more or less evenly distributed within the respective tissue, a process termed isomorphic gliosis. These astrocytes have also been shown to differ noticeably in their secreted mediators compared to astrocytes-forming anisomorphic gliosis. The factors determining the evolution of either gliotic tissue remain largely unknown (review in Ridet et al., 1997; Silver and Miller, 2004). It has become clear that astrocytes accomplish pivotal functions after CNS injury in preventing further tissue damage, reducing excitotoxic neuronal death by uptake of extracellular glutamate, promoting neuronal survival by trophic support and reestablishing the integrity of the BBB to limit inflammatory cell invasion. Several studies in which reactive astrocytes were ablated after CNS injury have shown the supportive effects of astrocytes to be impressive, resulting in a dramatic increase in leukocyte infiltration, neuronal death, and axonal degeneration, respectively (Bush et al., 1999; Faulkner et al., 2004). On the other hand, the astrocytic response in CNS wound healing is to a great extent responsible for the formation of the characteristic anisomorphic gliotic tissue often referred to as “glial scar,” which is widely regarded as one of the major obstacles to neuronal regeneration in

11

T-Helper Cells and Neuroregeneration

215

the injured CNS. It consists mainly of a dense network of reactive astrocytes, but reactive microglia/macrophages, reactive OPCs/NG2-glia and, depending on insult severity, meningeal cells and fibroblasts also participate in its formation (review in Fawcett and Asher, 1999; Stichel and Muller, 1998). These cells produce a variety of surface-bound and/or deposited extracellular matrix molecules, such as fibronectin, collagen, semaphorins sema3A and sema4D and the chondroitin sulpfte proteoglycans (CSPG) neurocan, phosphacan, versican, and NG2, among others (Asher et al., 2002; Camand et al., 2004; Liesi and Kauppila, 2002; Tang et al., 2003). Several studies have suggested that this “glial scar” and its environment might be largely responsible for unsuccesful neuronal regeneration due to the particular inhibitory effects of many of its ECM components, as well as associated myelin-derived molecules, on axonal outgrowth in vitro (Dou and Levine, 1994; He and Koprivica, 2004; Moreau-Fauvarque et al., 2003; Niclou et al., 2003; Schmalfeldt et al., 2000). Consistent with these observations, the degradation of ECM components in the glial scar, inhibition of their synthesis, or neutralization of inhibition-mediating epitopes by the application of specific antibodies led to clearly improved neuronal regeneration and functional recovery (Cafferty et al., 2007; Bradbury et al., 2002; Kawano et al., 2005; Merkler et al., 2003; Schnell and Schwab, 1990; Stichel et al., 1999). Similar to repair processes in other mammalian tissues, polymorphonuclear neutrophils (PMN) are the first leukocytes to invade the injured tissue, with a peak presence in the CNS at about 24–48 h after lesion (Schnell et al., 1999; Clark et al., 1994; Youssef et al., 2002). They are thought to contribute to host defence with their bactericidal and phagocytotic properties but can also exacerbate tissue injury by occluding microvasculature and thus aggravating local ischemia, production of free radicals and leukotriens, which may increase BBB permeability, facilitating further leukocyte infiltration, and favoring brain edema formation and secretion of proteolytic enzymes, which can damage neurons (Tonai et al., 2001; Dinkel et al., 2004; Taoka et al., 1998; Taoka et al., 1997). However, the exact pathogenic potential of PMNs in CNS injury remains to be clarified. Studies diminishing PMN infiltration in mechanical CNS lesions failed to show significant improvements regarding postlesional pathology or neurological outcome (Weaver et al., 2006; Whalen et al., 1999). In addition, a few studies suggest that neutrophils may even modulate CNS inflammation. Neutrophil-derived proteases can cleave IL-2 and IL-6 receptor ligand-binding chains and reduce IL-6 bioactivity and thus may influence IL-2 and IL-6 signaling during inflammatory reactions (Bank et al., 1999; Bank et al., 1999). In models of autoimmune brain inflammation it has been shown that they can also regulate T-cell responses (Zehntner et al., 2005). Peripheral macrophage infiltration after CNS injury is also evident in a variety of lesion models. They invade the lesioned tissue, become activated, and complement CNS resident microglia. Since both fractions are neither morphologically nor immunohistochemically sufficiently distinguishable from each other, it is difficult to determine the actual proportion of peripheral macrophage infiltration in the reactive microglia/macrophage population observed (review in Guillemin and Brew, 2004). Two slightly different approaches in the entorhinal cortex lesion model

216

S. Hendrix and R. Nitsch

using flow cytometric analysis, one distinguishing infiltrating macrophages from resident microglia by differential expression of cell-surface molecule CD45 and the other by the use of GFP-expressing bone marrow chimeras, respectively, showed a rather moderate accumulation of invading macrophages in the zone of anterograde degeneration (Babcock et al., 2003; Wirenfeldt et al., 2005). In the spinal cord injury model, the use of bone marrow chimeric rats not only revealed a considerable proportion of invading macrophages at later stages but also showed that zones of axonal degeneration are predominantly populated by intrinsic microglia in contrast to damaged gray matter, which is infiltrated by peripheral macrophages (Popovich and Hickey, 2001). These results provide further evidence supporting the hypothesis of functional heterogeneity and distinct governing signals of microglia and macrophages, respectively. Other traumatic CNS lesion paradigms still require discrimination of microglia/macrophage subpopulations. Finally, over the last decade, the role of invading T cells has been the subject of an ongoing heated debate, which we will review in the following chapters. Furthermore, we will provide an overview of potential mechanisms by which T cells may mediate beneficial effects after traumatic CNS injury.

11.3 The Adaptive Immune System in CNS Injury 11.3.1 T Cells in the Healthy and Injured CNS The CNS is routinely and effectively surveyed by the immune system (Engelhardt and Ransohoff, 2005). For a long time a concept of a “passive” immune privilege has dominated the literature to explain the high threshold for initiating lymphocyte responses within the CNS (Carson et al., 2006). According to the classical view three conditions are responsible for the development and maintenance of the CNS immune privilege: (1) Reduction of T-cell entry into the CNS by the BBB; (2) reduction of T-cell exposure to CNS antigens due to the absence of draining lymphatic vessels; and (3) reduced ability of microglial and astrocytes to present antigens to T cells. However, this view has been challenged by increasing evidence that the CNS actively interacts with the immune system to specifically regulate T-cell responses on a local and systemic level (Galea et al., 2007; Carson et al., 2006; Tian et al., 2009). Activated lymphocytes enter the CNS in the absence of overt inflammatory disease (Hickey et al., 1991; Wekerle et al., 1987), however, in the healthy brain only low numbers of T lymphocytes are present (Hickey, 1999), T-cell numbers are elevated during a strong immunological response in the body, even if the nervous system is itself not involved (Hickey and Kimura, 1987), suggesting that the immunosurveillance of the entire body including the CNS is increased after immunological challenge. Human central memory CD4+ T cells enter the cerebrospinal fluid, across either the choroids plexus veins or the meningeal blood vessels, to monitor the subarachnoid space, retaining the capacity to either

11

T-Helper Cells and Neuroregeneration

217

initiate local immune reactions or return to secondary lymphoid organs (Kivisakk et al., 2003). Interestingly, antigen specificity of T cells is irrelevant for their accumulation in CNS lesions, since they enter the healthy CNS parenchyma within hours after injection. However, only CNS-specific cells remain there or cyclically reenter the brain (Hickey et al., 1991). Approximately 24 h after mechanical lesion increased numbers of T cells infiltrate the brain (Fig. 11.2; Kramer and Hendrix, unpublished observations) or the spinal cord of rodents and reach a peak around 2–3 weeks after lesion; however, the timing of T-cell invasion, timing of the peak, and number of peaks are hugely dependent on the model and the mouse or rat strain (Babcock et al., 2003; Donnelly and Popovich, 2008; Schnell et al., 1999). A second, late phase of T-cell infiltration after spinal cord injury has been reported in mice exposed to mouse hepatitis virus in a conventional breeding facility, while this late-phase response was absent in mice kept in specific pathogen-free facilities (Schnell et al., 1997). It is not clear whether the absence of the second peak in specific pathogen-free mice represents an artifact in an artificially clean mouse population.

11.3.2 Self-Reactive T Cells in CNS Injury For decades it is known that autoreactive T cells are found in rodents and men after traumatic CNS injury. Activated CNS-reactive T cells are considered to be key players in the induction and maintenance of autoimmune diseases such as multiple sclerosis and experimental autoimmune encephalomyelitis. However, they are also found in per se non-autoimmune conditions such as experimental spinal cord injury in rodents (Jones et al., 2002; Popovich et al., 1996), experimental and clinical nerve trauma (Olsson et al., 1993, 1992), and in spinal cord injury and stroke patients (Wang et al., 1992; Kil et al., 1999). Seven days after spinal cord injury in the rat, autoreactive T cells can be isolated from peripheral lymph nodes. These T cells react with myelin basic protein (MBP), an important protein of the myelin sheath. These CNS-reactive T cells can exacerbate axonal injury, demyelination, and functional loss after spinal cord injury in Lewis rats and transgenic mice with high levels of MBP-reactive T cells (Jones et al., 2002; Popovich et al., 1996). When these autoreactive T cells are injected into healthy animals, they develop a phenotype reminiscent of experimental autoimmune encephalomyelitis (EAE) (Popovich et al., 1996). The authors interpret these data as a clear indication that autoimmune T cells contribute to neurodestruction and demyelination during this transient phase of T-cell activation. However, there is considerable debate as to whether autoreactive T cells damage the CNS or whether they exert neuroprotective effects and support neuroregeneration (Popovich and Longbrake, 2008; Schwartz and Ziv, 2008). Notably, T-cell subtypes have never been systematically investigated for subtype-specific impact on neuroprotection and regeneration (Hendrix and Nitsch, 2007). Furthermore, there are only a few studies investigating the differential effects of CD4+ T cells in the different phases of

218

S. Hendrix and R. Nitsch

acute and chronic CNS response after trauma (for a review see Fabry et al., 2008), however, without specifically determining the effects of different subtypes of CD4+ T cells.

11.3.3 Neuronal Injury in T Cell-Deficient Mice Clarifying the role of T cells in neuronal injury by analyzing neuronal lesion models in T-cell-deficient mice with and without replenishing T cells seems a logical step. However, the published results have not solved the controversy. Fee et al. (2003) reported that aseptic cerebral injury is attenuated in T-celldeficient RAG1(−/−) mice and replenishment of activated CD4 T cells exacerbated cerebral injury in RAG1(−/−) mice (Fee et al., 2003). In contrast, after experimental axotomy of the facial nerve of immunodeficient SCID mice, the survival of facial motor neurons was severely impaired compared to immunocompetent wild-type mice. Reconstitution of SCID mice with wild-type splenocytes containing T and B cells restored the survival of facial motor neurons in these mice to the level of the wild-type controls (Serpe et al., 1999). Thus, in one model, replenishment of T cells exacerbated the injury and in another model it increased cell survival. It is important to note that these studies are difficult to compare for a number of reasons. The mouse lines display profound differences in their immune systems, the lesion is either in the CNS or in the peripheral nervous system (PNS), the parameters evaluated are different, and either activated CD4 T cells or mixed splenocytes have been administered without further characterization of the T-cell subtypes. However, it is notoriously difficult to interpret the results of replenishment of T cells into immunocompromised mice since the changed immunological milieu may impair normal T-cell functions and may e.g., suppress regulatory subpopulations.

11.3.4 The Controversy About “Protective Autoimmunity” There is abundant evidence that autoimmune T cells play a detrimental role in the CNS (McFarland and Martin, 2007; Stinissen et al., 1997). Interestingly, in contrast to these data and in contrast to the general notion that autoimmune T cells are primarily harmful to the CNS (particularly hold by multiple sclerosis researchers), there is increasing evidence that T cells may also exert beneficial effects in a wound healing context following CNS trauma. Michal Schwartz’s group has proposed the concept of “protective autoimmunity,” whereby autoreactive T cells directed to specific self-antigens are recognized as “the physiologic fighting force against acute and chronic neurodegenerative conditions” (Moalem et al., 1999; Schwartz, 2005). In their landmark publication, Moalem et al. (1999), reported that injection of activated anti-MBP T cells protects retinal ganglion cells after partial optic nerve crush in rats. In this model, activated rat T cells specific for MBP, heatshock protein (HSP), and OVA were injected intraperitoneally into rats. All types of activated T cells accumulated in the injured optic nerve but not in the healthy contralateral optic nerve

11

T-Helper Cells and Neuroregeneration

219

(Moalem et al., 1999; Hirschberg et al., 1998). Using a neurotracer the degree of primary and secondary damage to optic nerve axons and their attached retinal ganglion cells was analyzed immediately after the injury and again after 2 weeks. The percentage of labeled retinal ganglion cells reflects the viable axons, which are still able to transport the neurotracer. Only autoreactive MBP-specific T cells were able to protect retinal ganglion cells from secondary damage – in contrast to T cells specific for the non-self antigen OVA or T cells specific for HSP, which is not restricted to the CNS. Thus, the accumulation of activated T cells in the lesion site appears to be antigen-independent, while neuroprotection appears to be a characteristic of T cells specific for CNS antigens. In a multitude of follow-up studies published, Michal Schwartz’s group extended the concept of “protective autoimmunity” to many other CNS and PNS lesion paradigms (reviewed in Hendrix and Nitsch, 2007; Schwartz, 2005). Several other laboratories have reported protective and/or pro-regenerative effects of T cells or vaccine strategies (selected studies are summarized in Hendrix and Nitsch, 2007). The therapeutic outcome of this concept might be the administration of autoreactive T cells or a vaccination strategy with CNS antigens to expand autoreactive T-cell clones. Such approaches have been proposed as a potential clinical therapy for a variety of neurodegenerative conditions, including spinal cord injury (Hauben et al., 2001; Schwartz et al., 1999), Alzheimer’s disease (Janus et al., 2000; Morgan et al., 2000; Schenk et al., 1999), glaucoma (Fisher et al., 2001), amyotrophic lateral sclerosis (Angelov et al., 2003), and Parkinson’s disease (Benner et al., 2004). However, the concept of “protective autoimmunity” and corresponding clinical approaches have been severely criticized (Jones et al., 2002; Jones et al., 2004; Popovich and Jones, 2003). Promising results in animal models may not be reproducible in human trials. For example, one potential therapeutic application in the case of Alzheimer’s disease involves immunizing patients against the amyloid-β peptide, the major proteinaceous component that characterizes plaques in the brains of patients with this disease. Experimental data in mice (Janus et al., 2000; Morgan et al., 2000; Schenk et al., 1999) led to the development and clinical trials of a vaccine, AN-1792, based on this approach. Unfortunately, several patients in the trial developed CNS inflammation and the vaccine was withdrawn from human trials (Bishop et al., 2002; Check, 2002). Therefore, further research is necessary to understand the underlying mechanism of protective autoimmunity. Suggested potential mechanisms include the local production of neurotrophins and cytokines by T cells. These are said to cause microglia to buffer toxic mediators such as glutamate, produce growth factors and remove growth inhibitors, for example, by phagocytosis of myelin (Schwartz, 2005).

11.4 T-Helper Cell Subpopulations in CNS Injury and Repair? One possible explanation for T-cell injections and vaccination strategies leading to conflicting results may be that different subtypes of T-cells are responsible for distinct T-cell effects. Here, we review increasing evidence that a specific T-helper cell subpopulation is particularly beneficial in the context of CNS lesions.

220

S. Hendrix and R. Nitsch

11.4.1 Analysis of the Th1/Th2 Ratio T-helper cell differentiation and associated effector responses have been intensively studied over the last three decades, initiated by the identification and characterization of cell-mediated versus humoral immunity (Cherwinski et al., 1987; Parish, 1972). In 1986 it was formally shown by Mosmann et al. that mouse CD4+ T cells could be subdivided on the basis of cytokine secretion patterns into two subsets, designated T-helper cells type 1 (Th1) and type 2 (Th2) (Mosmann et al., 1986). This conclusion was later extended to human CD4+ T-cells (Parronchi et al., 1991; Wierenga et al., 1990) and had an enormous impact on basic and applied immunology (Liew, 2002). After initial activation by immunogenic peptides presented by antigen-presenting cells (APC), naive Th lymphocytes can differentiate into the phenotypically distinct memory Th1 or Th2 cells. Th1 cells are characterized by their marker cytokines such as interleukin-2 (IL-2) and interferon-γ (IFN-γ). They activate macrophages and are very effective in controlling infection with intracellular pathogens. In contrast, the marker cytokines of Th2 cells are IL-4, IL-5, IL-10, and IL-13. These stimulate B cells to produce antibodies and play a major role in eradicating helminths and extracellular parasites. In addition, other CD4+ T-cell populations have been described, such as Th17 T cells that secrete IL-17 and IL-6, as well as regulatory T-cell populations that secrete high levels of IL-10 and TGF-β and display the potential to modulate both Th1 and Th2 responses (comprehensive reviews in Weaver et al., 2006; Biedermann et al., 2004; Shevach, 2002; Weiner, 2001). However, when comparing studies using Th1 and/or Th2 cells, it is important to keep in mind that there are several methods to determine the Th-status under different experimental conditions (Hendrix and Nitsch, 2007). One method is the analysis of selected Th1 and Th2 cytokines in affected tissues or in the supernatant of stimulated cells isolated from local lymph nodes, spleen, thymus, or blood. However, these cytokines are not specific to Th1 or Th2 cells. For example, the Th1 cytokines INF-γ and tumor necrosis factor-α (TNF-α) are expressed by other cell types such as natural killer cells and macrophages, while the Th2 cytokines IL-4, IL-5, and IL-13 are also expressed by mast cells. Other investigators analyze Th1 and Th2 cytokines (for example Con A-stimulated pooled splenocytes or blood-derived CD4+ T cells). Another critical point is the calculation of Th1/Th2 ratio, e.g, in order to determine whether an experimental test group of animals or humans shows a shift towards a Th1- or Th2-dominated state compared to controls. Again, there are several methods such as dividing the number of Th1 cytokine-producing T cells by the number of Th2 cytokine-producing T cells detected, for example, by FACS analysis. Another method is determining the concentration of selected Th1 and Th2 cytokines in the supernatant of isolated and cultured T cells, for example, by ELISA or capturedbead assay. In the first case, a Th2 shift is the result of reduced numbers of Th1 cells or increased numbers of Th2 cells or both. In the second case, a Th2 shift is either the result of decreased Th1 cytokine secretion or increased Th2 cytokine secretion or both.

11

T-Helper Cells and Neuroregeneration

221

The systemic Th1/Th2 balance is dependent on age and genetic background (Caruso et al., 1996, 1997; Dayan et al., 2000; Engwerda et al., 1996; Ernst et al., 1993; Frasca et al., 1997; Ginaldi et al., 1999; Kariv et al., 1992; Kubo and Cinader, 1990; Sandmand et al., 2002; Segal et al., 1997; Shearer, 1997). During aging, the Th1/Th2 status passes through three distinct phases (Shearer, 1997). A net Th2 dominance marks the period from birth to young adulthood, followed by a relatively stable Th1/Th2 equilibrium during the reproductive years, except during the period of pregnancy. During the post-reproductive years a net Th2 dominance again emerges (Sandmand et al., 2002; Ginaldi et al., 1999). Analysis of healthy, untreated rodents has revealed important strain differences between the base-line Th1/Th2 status (Hendrix and Nitsch, 2007) and an age-dependent switch from a Th1 to a Th2 status (Dayan et al., 2000). These strain- and age-dependent differences make comparing different experiments or studies particularly difficult (Hendrix and Nitsch, 2007).

11.4.2 Are Subpopulations of T Cells Responsible for Neuroprotective and Pro-regenerative Effects? Surprisingly, only a few studies have further characterized those T-cell subtypes exerting beneficial effects in the context of CNS injury. For example, in a model of murine entorhinal–hippocampal brain slice cultures it could be demonstrated in vitro that Th2 cells prevented or even reversed Th1-induced upregulation of the inflammatory marker ICAM-1 on microglial cells (Gimsa et al., 2001). Furthermore, in a similar model, both Th1 and Th2 cells were able to support neuronal survival, provided there was no cell–cell contact with the slices. Th2 cells displayed a significantly higher protective effect compared to Th1 cells (Wolf et al., 2002). An in vivo study by the Schwartz group demonstrates that injections of autoimmune Th1 cells (characterized by IFNγ and IL-10 secretion patterns) lead to neuroprotection compared to PBS injection (Kipnis et al., 2002). Unfortunately, no Th2 cells were used as controls in this study. In a spinal cord hemisection model the regeneration of descending fiber tracts was significantly reduced in signal transducers and activators of transcription 6 (STAT6)-deficient mice compared to wild-type controls (Fraidakis et al., 2007). Since STAT6 signaling is crucial for IL-4 signaling and the development of Th2 cells (Kaplan et al., 1996) these data suggest that a Th2 response is – at least in part – necessary for spontaneous regeneration. However, there was no difference in the locomotion scores between wild-type and STAT6-deficient mice (Fraidakis et al., 2007). Interestingly, in a model of facial nerve axotomy STAT6-mediated CD4+ T-cell differentiation into the Th2 subset has been shown to play a key role in the survival of facial motoneurons (Deboy et al., 2006). These studies suggest that both Th1 and, to a greater extent, Th2 cells may support neuroprotection and regeneration. Interestingly, several studies have shown that potent inducers of a systemic Th2 shift such as Th2-inducing adjuvants, glatiramer

222

S. Hendrix and R. Nitsch

acetate, or members of the statin family improve outcome after CNS injury (Hendrix and Nitsch, 2007).

11.5 CNS Injury-Induced Immunosuppression Results in a Systemic Th2 Shift A potential beneficial role of Th2 cells in neuroregeneration is illustrated by the fact that CNS injury such as stroke is associated with a systemic Th2 shift. CNS injury induces a substantial modulation of the immune system, which leads to secondary immunodeficiency [CNS injury-induced immunodepression (CIDS)] (Meisel et al., 2005). CIDS results in increased susceptibility to infection, the leading cause of morbidity and mortality in patients with acute CNS injury. Interestingly, this immunodepression is characterized by a catecholamine-driven systemic Th2 shift in mice (Prass et al., 2003), which leads to impaired cellular immune responses and decreased IFN-γ production by blood lymphocytes, while humoral immune responses were less affected (reviewed in Meisel et al., 2005; Dirnagl et al., 2007). In a mouse model of stroke, the adoptive transfer of IFN-γ-producing lymphocytes from healthy littermates or treatment with recombinant IFN-γ greatly diminished bacterial burden demonstrating that a Th2 switch plays a causal role in CIDS persistence and consecutive infections (Prass et al., 2003). Thus, it is feasible that CIDS may not be limited to counterbalancing excessive inflammation in the brain. Since CNS injury is associated with the activation of autoreactive T cells (see above), the associated immunodepression may suppress Th1-mediated cellular immune responses that induce autoimmune brain inflammation (similar to the immunization models using Th2-inducing adjuvants) (Dirnagl et al., 2007). However, more studies are necessary to determine the Th1/Th2 status of the CNS after mechanical injury. Two studies using a spinal cord injury model suggest that the adaptive immune response is biased toward a Th1 phenotype (Jones et al., 2002); Ankeny et al., 2006). Since the systemic Th1/Th2 balance is dependent on age and genetic background (Caruso et al., 1996; Castle et al., 1997; Dayan et al., 2000; Engwerda et al., 1996; Ernst et al., 1993; Frasca et al., 1997; Ginaldi et al., 1999; Kariv et al., 1992; Kubo and Cinader, 1990; Sandmand et al., 2002; Segal et al., 1997; Shearer, 1997), it is feasible to expect phase-specific changes of the Th1/Th2 status when the CNS passes through distinct phases of acute response and chronic remodeling.

11.6 Are Beneficial T-Cell Effects Exclusively Attributable to Th2 Cells? Importantly, in the two studies of the David lab summarized above, IFA or Alum as adjuvants potently promoted axon regeneration but prevented the development of EAE, in contrast to CFA (Huang et al., 1999; Sicotte et al., 2003). Thus, it is tempting to speculate that a downregulation of Th1 responses and a shift toward Th2

11

T-Helper Cells and Neuroregeneration

223

responses after CNS damage is a physiological wound healing reaction following CNS injury in order to prevent the development of autoimmune diseases such as EAE and MS. Briefly, after CNS injury a systemic Th2 shift may be important in downregulating Th1- or Th17-driven cellular immune responses in order to prevent autoimmune diseases such as EAE or MS. The price for the systemic Th2 shift is increased susceptibility to infection. An additional Th2 shift may further increase neuroprotection and regeneration. In summary, there are several lines of evidence that a Th2 cells are beneficial in the context of the injured CNS: 1. Th2 cells support neuronal survival better than Th1 cells in vitro (Wolf et al., 2002). 2. Th2 cells suppress Th1-induced inflammatory signals in brain slices in vitro (Gimsa et al., 2001). 3. STAT6-deficient mice show reduced axon regeneration after spinal cord injury (although no differences in the locomotion scores were detectable) (Fraidakis et al., 2007) and STAT6/IL-4-mediated signaling supports the survival of facial motoneurons in a model of facial nerve axotomy (Deboy et al., 2006). 4. CNS injury induces systemic immunosuppression characterized by a systemic Th2 switch, which impairs cellular immune responses (Meisel et al., 2005). 5. In vaccination models for the treatment of CNS injury, Th2-inducing adjuvants such as IFA and Alum promote axon regeneration better than the Th1-inducing adjuvant CFA (Huang et al., 1999; Sicotte et al., 2003). 6. In vaccination models for the treatment of CNS damage, Th2-inducing adjuvants prevent the development of autoimmune diseases such as EAE (Huang et al., 1999; Sicotte et al., 2003). 7. Potent inducers of a systemic Th2 switch such as GA and statins support neuroprotection and/or regeneration (Lu et al., 2004; Pannu et al., 2005). However, there is good evidence that beneficial T-cell effects have been described for both Th1 and Th2 cells; thus, neuroprotection and regeneration appear to be not exclusively limited to Th2 cells. There is increasing evidence in vitro and in vivo that Th2 cells or a systemic Th2 shift may further increase neuroprotection and axon regeneration after CNS injury – but most importantly, a Th2 shift seems to prevent EAE after immunization with CNS antigens. Future studies are desperately needed to determine the effects of these T-cell subpopulations in the different phases of acute and chronic CNS responses to trauma.

11.7 How May T Cells Contribute to CNS Repair? Although there is much controversy about the role of T cells in the context of CNS injury, there is increasing acceptance that T-cell-based therapies such as vaccination strategies may be powerful methods to improve the clinical outcome after CNS

224

S. Hendrix and R. Nitsch

injury but also heated debate about developing therapies without understanding the precise mechanisms of T-cell contribution to CNS repair (Hendrix and Nitsch, 2007; Popovich and Longbrake, 2008; Schwartz and Ziv, 2008). Several mechanisms have been proposed but await still further investigation – in particular in relation to the different phases of CNS reaction to trauma:

11.7.1 Neurotrophin Secretion by T Cells Invading T cells may exert neuroprotective and pro-regenerative effects in the CNS by locally secreting growth factors such as neurotrophins in lesioned neural tissue (Hammarberg et al., 2000; Kerschensteiner et al., 1999).

11.7.2 Modulation of Microglia Functions It has been suggested that the reduced capacity of the CNS to regenerate damaged neurites may be a drawback to CNS immune privilege (Bechmann, 2005): the weak costimulation and elimination of T cells in the rather anti-inflammatory milieu of the CNS may reduce the ability of T cells to stimulate microglial functions such as phagocytosis of myelin debris, which contains potent inhibitors of axonal regeneration (Buchli and Schwab, 2005). Thus, stimulating T-cell responses, e.g., by vaccination, may support microglial phagocytosis (Gimsa et al., 2000). It has been also suggested that autoimmune T cells may influence microglial cells to buffer glutamate and to produce growth factors after CNS damage (Schwartz, 2005).

11.7.3 Modulation of Endogenous and Exogenous Stem Cells The Schwartz group suggested that autoreactive T cells may modulate endogenous stem cells and support the effect of therapeutically applied stem cells. They reported that T-cell deficiency leads to impaired hippocampal neurogenesis and that transgenic mice with most T cells autoreactive for a CNS antigen (MBP) display higher levels of neurogenesis (Ziv et al., 2006). They also showed an influence of CNS-specific T cells on adult neurogenesis of GABAergic neurons in the spinal cord (Shechter et al., 2007). Finally, they demonstrated a synergic therapeutic effect of their T-cell-based vaccination strategy and the transplantation of exogenous adult neural stem/progenitor cells into the cerebrospinal fluid leading to increased functional recovery after spinal cord injury (Ziv et al., 2006).

11.7.4 Anti-inflammatory T-Cell Effects In contrast to the concept of protective autoimmunity it has been proposed that beneficial T-cell effects may be rather a result of T-cell-induced suppression of

11

T-Helper Cells and Neuroregeneration

225

detrimental CNS inflammation – in particular because in certain models immunosuppression, e.g., by methylprednisolone may promote healing after CNS trauma (Fabry et al., 2008; Popovich and Longbrake, 2008).

11.7.5 Direct Stimulation of Axon Growth by T-Cell Cytokines Finally, cytokines secreted or induced by T cells such as IL-1β, IL-4, IL-6, TNF-α, and IFN-γ may directly influence axonal outgrowth (Hendrix and Peters, 2007). Recently, we demonstrated that classical Th1 and Th2 cytokines specifically modulate neurotrophin-induced axonal outgrowth from dorsal root ganglia (DRG) (Golz et al., 2006). Thus, axonal regeneration after neuronal damage may be mediated via a cytokine/neurotrophin axis (Hendrix and Peters, 2007) and T cells may be important players in providing the necessary cytokines.

11.8 Future Perspectives It is important to note that the effects of T cells and in particular of the different T-cell subtypes have not been correlated with the different phases of acute and chronic CNS responses to trauma. There is evidence that Th1 as well as Th2 cells may exert beneficial effects, but there is still no accepted concept or experimental evidence that demonstrates by which mechanism these beneficial effects of T cells are mediated. Additionally, a systematic investigation of the influence of the Th1/Th2 status on neuroprotection and axon regeneration has not yet been carried out. Therefore, it would be advantageous to compare systematically the outcomes of Th1- and Th2inducing conditions in both lesion development after CNS lesion and the results of therapeutic interventions such as T-cell injections and vaccine strategies. In particular, the role of these T-cell subtypes in the different phases of acute and chronic CNS response to trauma should be systematically investigated in order to reveal whether contradictory findings may be attributable to changing roles of different T-cell subpopulations during different phases of CNS lesion and repair.

References Ahn YH, Lee G, Kang SK (2006) Molecular insights of the injured lesions of rat spinal cords: Inflammation, apoptosis, and cell survival. Biochem Biophys Res Commun 348:560–570 Andersson PB, Perry VH, Gordon S (1992) The acute inflammatory response to lipopolysaccharide in CNS parenchyma differs from that in other body tissues. Neuroscience 48:169–186 Angelov DN et al (2003) Therapeutic vaccine for acute and chronic motor neuron diseases: implications for amyotrophic lateral sclerosis. Proc Natl Acad Sci USA 100: 4790–4795 Ankeny DP, Lucin KM, Sanders VM, McGaughy VM, Popovich PG (2006) Spinal cord injury triggers systemic autoimmunity: evidence for chronic B lymphocyte activation and lupus-like autoantibody synthesis. J Neurochem 99:1073–1087

226

S. Hendrix and R. Nitsch

Asher RA et al (2002) Versican is upregulated in CNS injury and is a product of oligodendrocyte lineage cells. J Neurosci 22:2225–2236 Babcock AA, Kuziel WA, Rivest S, Owens T (2003) Chemokine expression by glial cells directs leukocytes to sites of axonal injury in the CNS. J Neurosci 23:7922–7930 Bank U et al (1999) Selective proteolytic cleavage of IL-2 receptor and IL-6 receptor ligand binding chains by neutrophil-derived serine proteases at foci of inflammation. J Interferon Cytokine Res 19:1277–1287 Bank U, Kupper B, Reinhold D, Hoffmann T, Ansorge S (1999) Evidence for a crucial role of neutrophil-derived serine proteases in the inactivation of interleukin-6 at sites of inflammation. FEBS Lett 461:235–240 Bechmann I (2005) Failed central nervous system regeneration: a downside of immune privilege? Neuromolecular Med 7:217–228 Benner EJ et al (2004) Therapeutic immunization protects dopaminergic neurons in a mouse model of Parkinson’s disease. Proc Natl Acad Sci U S A 101:9435–9440 Biedermann T, Rocken M, Carballido JM (2004) TH1 and TH2 lymphocyte development and regulation of TH cell-mediated immune responses of the skin. J Investig Dermatol Symp Proc 9:5–14 Bishop GM, Robinson SR, Smith MA, Perry G, Atwood CS (2002) Call for Elan to publish Alzheimer’s trial details. Nature 416:677 Boje KM, Arora PK (1992) Microglial-produced nitric oxide and reactive nitrogen oxides mediate neuronal cell death. Brain Res 587:250–256 Bradbury EJ et al (2002) Chondroitinase ABC promotes functional recovery after spinal cord injury. Nature 416:636–640 Buchli AD, Schwab ME (2005) Inhibition of Nogo: a key strategy to increase regeneration, plasticity and functional recovery of the lesioned central nervous system. Ann Med 37: 556–567 Bush TG et al (1999) Leukocyte infiltration, neuronal degeneration, and neurite outgrowth after ablation of scar-forming, reactive astrocytes in adult transgenic mice. Neuron 23:297–308 Bye N et al (2007) Transient neuroprotection by minocycline following traumatic brain injury is associated with attenuated microglial activation but no changes in cell apoptosis or neutrophil infiltration. Exp Neurol204:220–233 Cafferty WB, Yang SH, Duffy PJ, Li S, Strittmatter SM (2007) Functional axonal regeneration through astrocytic scar genetically modified to digest chondroitin sulfate proteoglycans. J Neurosci 27:2176–2185 Camand E, Morel MP, Faissner A, Sotelo C, Dusart I (2004) Long-term changes in the molecular composition of the glial scar and progressive increase of serotoninergic fibre sprouting after hemisection of the mouse spinal cord. Eur J Neurosci 20:1161–1176 Campbell SJ, Wilcockson DC, Butchart AG, Perry VH, Anthony DC (2002) Altered chemokine expression in the spinal cord and brain contributes to differential interleukin-1beta-induced neutrophil recruitment. J Neurochem 83:432–441 Carson MJ, Doose JM, Melchior B, Schmid CD, Ploix CC (2006) CNS immune privilege: hiding in plain sight. Immunol Rev 213:48–65 Caruso C et al (1996) Cytokine production pathway in the elderly. Immunol Res 15:84–90 Castle S, Uyemura K, Wong W, Modlin R, Effros R (1997) Evidence of enhanced type 2 immune response and impaired upregulation of a type 1 response in frail elderly nursing home residents. Mech Ageing Dev 94:7–16 Check E (2002) Nerve inflammation halts trial for Alzheimer’s drug. Nature 415:462 Chen S, Pickard JD, Harris NG (2003) Time course of cellular pathology after controlled cortical impact injury. Exp Neurol 182:87–102 Cherwinski HM, Schumacher JH, Brown KD, Mosmann TR (1987) Two types of mouse helper T cell clone. III. Further differences in lymphokine synthesis between Th1 and Th2 clones revealed by RNA hybridization, functionally monospecific bioassays, and monoclonal antibodies. J Exp Med 166:1229–1244

11

T-Helper Cells and Neuroregeneration

227

Clark RS, Schiding JK, Kaczorowski SL, Marion DW, Kochanek PM (1994) Neutrophil accumulation after traumatic brain injury in rats: comparison of weight drop and controlled cortical impact models. J Neurotrauma 11:499–506 Dayan M et al (2000) Effect of aging on cytokine production in normal and experimental systemic lupus erythematosus-afflicted mice. Exp Gerontol 35:225–236 Deboy CA et al (2006) Immune-mediated neuroprotection of axotomized mouse facial motoneurons is dependent on the IL-4/STAT6 signaling pathway in CD4(+) T cells. Exp Neurol 201:212–224 Del Rio JA et al (1997) A role for Cajal-Retzius cells and reelin in the development of hippocampal connections. Nature 385:70–74 Dinkel K, Dhabhar FS, Sapolsky RM (2004) Neurotoxic effects of polymorphonuclear granulocytes on hippocampal primary cultures. Proc Natl Acad Sci USA 101:331–336 Dirnagl U et al (2007) Stroke-induced immunodepression: experimental evidence and clinical relevance. Stroke 38:770–773 Donnelly DJ, Popovich PG (2008) Inflammation and its role in neuroprotection, axonal regeneration and functional recovery after spinal cord injury. Exp Neurol 209:378–388 Dou CL, Levine JM (1994) Inhibition of neurite growth by the NG2 chondroitin sulfate proteoglycan. J Neurosci 14:7616–7628 Eddleston M, Mucke L (1993) Molecular profile of reactive astrocytes–implications for their role in neurologic disease. Neuroscience 54:15–36 Engelhardt B, Ransohoff RM (2005) The ins and outs of T-lymphocyte trafficking to the CNS: anatomical sites and molecular mechanisms. Trends Immunol 26: 485–495 Engwerda CR, Fox BS, Handwerger BS (1996) Cytokine production by T lymphocytes from young and aged mice. J Immunol 156:3621–3630 Ernst DN, Weigle WO, Hobbs MV (1993) Aging and lymphokine production by T cell subsets. Stem Cells 11:487–498 Eskes C, Honegger P, Juillerat-Jeanneret L, Monnet-Tschudi F (2002) Microglial reaction induced by noncytotoxic methylmercury treatment leads to neuroprotection via interactions with astrocytes and IL-6 release. Glia 37:43–52 Fabry Z, Ling C, Swartz K, Fee, D (2008) Innate and adaptive immune responses in the CNS: role of innate mediators, T cells, and the complement system in traumatic injuries of the CNS. In: Ibarra A (ed) The role of immune cells in neurodegenerative diseases. Research Signpost, Kerala, pp 43–68 Faulkner JR et al (2004) Reactive astrocytes protect tissue and preserve function after spinal cord injury. J Neurosci 24:2143–2155 Fawcett JW, Asher RA (1999) The glial scar and central nervous system repair. Brain Res Bull 49:377–391 Fee D et al (2003) Activated/effector CD4+ T cells exacerbate acute damage in the central nervous system following traumatic injury. J Neuroimmunol 136:54–66 Fisher J et al (2001) Vaccination for neuroprotection in the mouse optic nerve: implications for optic neuropathies. J Neurosci 21:136–142 Fitch MT, Doller C, Combs CK, Landreth GE, Silver J (1999) Cellular and molecular mechanisms of glial scarring and progressive cavitation: in vivo and in vitro analysis of inflammationinduced secondary injury after CNS trauma. J Neurosci 19:8182–8198 Fraidakis MJ, Kiyotani T, Pernold K, Bergstrom J, Olson L (2007) Recovery from spinal cord injury in tumor necrosis factor-alpha, signal transducers and activators of transcription 4 and signal transducers and activators of transcription 6 null mice. Neuroreport 18: 185–189 Franzen R et al (1998) Effects of macrophage transplantation in the injured adult rat spinal cord: a combined immunocytochemical and biochemical study. J Neurosci Res 51: 316–327 Frasca D et al (1997) Regulation of cytokine production in aging: use of recombinant cytokines to upregulate mitogen-stimulated spleen cells. Mech Ageing Dev 93:157–169

228

S. Hendrix and R. Nitsch

Galea I, Bechmann I, Perry VH (2007) What is immune privilege (not)? Trends Immunol 28:12–18 Gibbons HM, Dragunow M (2006) Microglia induce neural cell death via a proximity-dependent mechanism involving nitric oxide. Brain Res 1084:1–15 Gimsa U, Peter SV, Lehmann K, Bechmann I, Nitsch R (2000) Axonal damage induced by invading T cells in organotypic central nervous system tissue in vitro: involvement of microglial cells. Brain Pathol 10:365–377 Gimsa U, Wolf SA, Haas D, Bechmann I, Nitsch R (2001) Th2 cells support intrinsic antiinflammatory properties of the brain. J Neuroimmunol 119:73–80 Ginaldi L et al (1999) The immune system in the elderly: III. Innate immunity. Immunol Res 20:117–126 Golz G et al (2006) The cytokine/neurotrophin axis in peripheral axon outgrowth. Eur J Neurosci 24:2721–2730 Guillemin GJ, Brew BJ (2004) Microglia, macrophages, perivascular macrophages, and pericytes: a review of function and identification. J Leukoc Biol 75:388–397 Hakamada-Taguchi R et al (2003) Inhibition of hydroxymethylglutaryl-coenzyme a reductase reduces Th1 development and promotes Th2 development. Circ Res 93:948–956 Hammarberg H et al (2000) Neuroprotection by encephalomyelitis: rescue of mechanically injured neurons and neurotrophin production by CNS-infiltrating T and natural killer cells. J Neurosci 20:5283–5291 Hampton DW, Rhodes KE, Zhao C, Franklin RJ, Fawcett JW (2004) The responses of oligodendrocyte precursor cells, astrocytes and microglia to a cortical stab injury, in the brain. Neuroscience 127:813–820 Hang CH, Shi JX, Li JS, Wu W, Yin HX (2005) Concomitant upregulation of nuclear factor-kB activity, proinflammatory cytokines and ICAM-1 in the injured brain after cortical contusion trauma in a rat model. Neurol India 53:312–317 Hanisch UK (2002) Microglia as a source and target of cytokines. Glia 40:140–155 Hauben E et al (2001) Vaccination with a Nogo-A-derived peptide after incomplete spinal-cord injury promotes recovery via a T-cell-mediated neuroprotective response: comparison with other myelin antigens. Proc Natl Acad Sci USA 98:15173–12178 He Z, Koprivica V (2004) The Nogo signaling pathway for regeneration block. Annu Rev Neurosci 27:341–368 He BP, Wen W, Strong MJ (2002) Activated microglia (BV-2) facilitation of TNF-alpha-mediated motor neuron death in vitro. J Neuroimmunol 128:31–38 Hendrix S, Nitsch R (2007) The role of T helper cells in neuroprotection and regeneration. J Neuroimmunol 184:100–112 Hendrix S, Peters EM (2007) Neuronal plasticity and neuroregeneration in the skin – The role of inflammation. J Neuroimmunol 184:113–126 Herx LM, Yong VW (2001) Interleukin-1 beta is required for the early evolution of reactive astrogliosis following CNS lesion. J Neuropathol Exp Neurol 60:961–971 Hickey WF (1999) Leukocyte traffic in the central nervous system: the participants and their roles. Semin Immunol 11:125–137 Hickey WF, Hsu BL, Kimura H (1991) T-lymphocyte entry into the central nervous system. J Neurosci Res 28:254–260 Hickey WF, Kimura H (1987) Graft-vs.-host disease elicits expression of class I and class II histocompatibility antigens and the presence of scattered T lymphocytes in rat central nervous system. Proc Natl Acad Sci USA 84:2082–2086 Hirschberg DL et al (1998) Accumulation of passively transferred primed T cells independently of their antigen specificity following central nervous system trauma. J Neuroimmunol 89: 88–96 Holmin S, Mathiesen T, Shetye J, Biberfeld P (1995) Intracerebral inflammatory response to experimental brain contusion. Acta Neurochir (Wien) 132:110–119 Huang DW, McKerracher L, Braun PE, David S (1999) A therapeutic vaccine approach to stimulate axon regeneration in the adult mammalian spinal cord. Neuron 24:639–647

11

T-Helper Cells and Neuroregeneration

229

Illes P, Norenberg W, Gebicke-Haerter PJ (1996) Molecular mechanisms of microglial activation. B. Voltage- and purinoceptor-operated channels in microglia. Neurochem Int 29: 13–24 Janus C et al (2000) A beta peptide immunization reduces behavioural impairment and plaques in a model of Alzheimer’s disease. Nature 408:979–982 Jones TB et al (2002) Pathological CNS autoimmune disease triggered by traumatic spinal cord injury: implications for autoimmune vaccine therapy. J Neurosci 22:2690–2700 Jones TB et al (2004) Passive or active immunization with myelin basic protein impairs neurological function and exacerbates neuropathology after spinal cord injury in rats. J Neurosci 24:3752–3761 Kaplan MH, Schindler U, Smiley ST, Grusby MJ (1996) Stat6 is required for mediating responses to IL-4 and for development of Th2 cells. Immunity 4:313–319 Kariv I, Ferguson FG, Confer FL (1992) Age- and strain-related differences in murine spleen cell responses to different activation signals. Cell Immunol 140:67–80 Kawano H, Li HP, Sango K, Kawamura K, Raisman G (2005) Inhibition of collagen synthesis overrides the age-related failure of regeneration of nigrostriatal dopaminergic axons. J Neurosci Res 80:191–202 Kerschensteiner M et al (1999) Activated human T cells, B cells, and monocytes produce brainderived neurotrophic factor in vitro and in inflammatory brain lesions: a neuroprotective role of inflammation? J Exp Med 189:865–870 Kil K et al (1999) T cell responses to myelin basic protein in patients with spinal cord injury and multiple sclerosis. J Neuroimmunol 98:201–207 Kipnis J et al (2002) Myelin specific Th1 cells are necessary for post-traumatic protective autoimmunity. J Neuroimmunol 130:78–85 Kivisakk P et al (2003) Human cerebrospinal fluid central memory CD4+ T cells: evidence for trafficking through choroid plexus and meninges via P-selectin. Proc Natl Acad Sci USA 100:8389–8394 Kreutzberg GW (1996) Microglia: a sensor for pathological events in the CNS. Trends Neurosci 19:312–318 Kubo M, Cinader B (1990) Polymorphism of age-related changes in interleukin (IL) production: differential changes of T helper subpopulations, synthesizing IL 2, IL 3 and IL 4. Eur J Immunol 20:1289–1296 Kurpius D, Nolley EP, Dailey ME (2007) Purines induce directed migration and rapid homing of microglia to injured pyramidal neurons in developing hippocampus. Glia 55:873–884 Liberto CM, Albrecht PJ, Herx LM, Yong VW, Levison SW (2004) Pro-regenerative properties of cytokine-activated astrocytes. J Neurochem 89:1092–1100 Liesi P, Kauppila T (2002) Induction of type IV collagen and other basement-membrane-associated proteins after spinal cord injury of the adult rat may participate in formation of the glial scar. Exp Neurol 173:31–45 Liew FY (2002) T(H)1 and T(H)2 cells: a historical perspective. Nat Rev Immunol 2:55–60 Lu D et al (2004) Atorvastatin reduces neurological deficit and increases synaptogenesis, angiogenesis, and neuronal survival in rats subjected to traumatic brain injury. J Neurotrauma 21:21–32 Ma M, Wei P, Wei T, Ransohoff RM, Jakeman LB (2004) Enhanced axonal growth into a spinal cord contusion injury site in a strain of mouse (129X1/SvJ) with a diminished inflammatory response. J Comp Neurol 474:469–486 McFarland HF, Martin R (2007) Multiple sclerosis: a complicated picture of autoimmunity. Nat Immunol 8:913–919 Meisel C, Schwab JM, Prass K, Meisel A, Dirnagl U (2005) Central nervous system injury-induced immune deficiency syndrome. Nat Rev Neurosci 6:775–786 Merkler D et al (2003) Rapid induction of autoantibodies against Nogo-A and MOG in the absence of an encephalitogenic T cell response: implication for immunotherapeutic approaches in neurological diseases. Faseb J 17:2275–2277

230

S. Hendrix and R. Nitsch

Merrill JE, Ignarro LJ, Sherman MP, Melinek J, Lane TE (1993) Microglial cell cytotoxicity of oligodendrocytes is mediated through nitric oxide. J Immunol 151:2132–2141 Moalem G et al (1999) Autoimmune T cells protect neurons from secondary degeneration after central nervous system axotomy. Nat Med 5:49–55 Moreau-Fauvarque C et al (2003) The transmembrane semaphorin Sema4D/CD100, an inhibitor of axonal growth, is expressed on oligodendrocytes and upregulated after CNS lesion. J Neurosci 23:9229–9239 Morgan D et al (2000) A beta peptide vaccination prevents memory loss in an animal model of Alzheimer’s disease. Nature 408:982–985 Mosmann TR, Cherwinski H, Bond MW, Giedlin MA, Coffman RL (1986) Two types of murine helper T cell clone. I. Definition according to profiles of lymphokine activities and secreted proteins. J Immunol 136:2348–2357 Munch G et al (2003) Microglial activation induces cell death, inhibits neurite outgrowth and causes neurite retraction of differentiated neuroblastoma cells. Exp Brain Res 150:1–8 Nagata K, Takei N, Nakajima K, Saito H, Kohsaka S (1993) Microglial conditioned medium promotes survival and development of cultured mesencephalic neurons from embryonic rat brain. J Neurosci Res 34:357–363 Niclou SP, Franssen EH, Ehlert EM, Taniguchi M, Verhaagen J (2003) Meningeal cell-derived semaphorin 3A inhibits neurite outgrowth. Mol Cell Neurosci 24:902–912 Nicole O et al (2005) Activation of protease-activated receptor-1 triggers astrogliosis after brain injury. J Neurosci 25:4319–4329 Olsson T et al (1992) Facial nerve transection causes expansion of myelin autoreactive T cells in regional lymph nodes and T cell homing to the facial nucleus. Autoimmunity 13: 117–126 Olsson T et al (1993) Autoreactive T and B cell responses to myelin antigens after diagnostic sural nerve biopsy. J Neurol Sci 117:130–139 Pannu R, Barbosa E, Singh AK, Singh I (2005) Attenuation of acute inflammatory response by atorvastatin after spinal cord injury in rats. J Neurosci Res 79:340–350 Parish CR (1972) The relationship between humoral and cell-mediated immunity. Transplant Rev 13:35–66 Park JE, Barbul A (2004) Understanding the role of immune regulation in wound healing. Am J Surg 187:11S–16S Parronchi P et al (1991) Allergen- and bacterial antigen-specific T-cell clones established from atopic donors show a different profile of cytokine production. Proc Natl Acad Sci USA 88:4538–4542 Popovich PG et al (1999) Depletion of hematogenous macrophages promotes partial hindlimb recovery and neuroanatomical repair after experimental spinal cord injury. Exp Neurol 158:351–365 Popovich PG, Hickey WF (2001) Bone marrow chimeric rats reveal the unique distribution of resident and recruited macrophages in the contused rat spinal cord. J Neuropathol Exp Neurol 60:676–685 Popovich PG, Jones TB (2003) Manipulating neuroinflammatory reactions in the injured spinal cord: back to basics. Trends Pharmacol Sci 24:13–17 Popovich PG, Longbrake EE. (2008) Can the immune system be harnessed to repair the CNS? Nat Rev Neurosci 9:481–493 Popovich PG, Stokes BT, Whitacre CC. (1996) Concept of autoimmunity following spinal cord injury: possible roles for T lymphocytes in the traumatized central nervous system. J Neurosci Res 45:349–363 Prass K et al. (2003) Stroke-induced immunodeficiency promotes spontaneous bacterial infections and is mediated by sympathetic activation reversal by poststroke T helper cell type 1-like immunostimulation. J Exp Med 198:725–736 Quintana A et al (2005) Differential role of tumor necrosis factor receptors in mouse brain inflammatory responses in cryolesion brain injury. J Neurosci Res 82:701–716

11

T-Helper Cells and Neuroregeneration

231

Ridet JL, Malhotra SK, Privat A, Gage FH (1997) Reactive astrocytes: cellular and molecular cues to biological function. Trends Neursci 20:570–577 Rostworowski M, Balasingam V, Chabot S, Owens T, Yong VW (1997) Astrogliosis in the neonatal and adult murine brain post-trauma: elevation of inflammatory cytokines and the lack of requirement for endogenous interferon-gamma. J Neurosci 17:3664–3674 Sandmand M et al (2002) Is ageing associated with a shift in the balance between Type 1 and Type 2 cytokines in humans? Clin Exp Immunol 127:107–114 Schenk D et al (1999) Immunization with amyloid-beta attenuates Alzheimer-disease-like pathology in the PDAPP mouse. Nature 400:173–177 Schmalfeldt M, Bandtlow CE, Dours-Zimmermann MT, Winterhalter KH, Zimmermann DR (2000) Brain derived versican V2 is a potent inhibitor of axonal growth. J Cell Sci 113 (Pt 5), 807–816 Schneider LA, Schlenner SM, Feyerabend TB, Wunderlin M, Rodewald HR (2007) Molecular mechanism of mast cell mediated innate defense against endothelin and snake venom sarafotoxin. J Exp Med 204:2629–2639 Schnell L, Fearn S, Klassen H, Schwab ME, Perry VH (1999) Acute inflammatory responses to mechanical lesions in the CNS: differences between brain and spinal cord. Eur J Neurosci 11:3648–3658 Schnell L, Schneider R, Berman MA, Perry VH, Schwab ME (1997) Lymphocyte recruitment following spinal cord injury in mice is altered by prior viral exposure. Eur J Neurosci 9: 1000–1007 Schnell L, Schwab ME (1990) Axonal regeneration in the rat spinal cord produced by an antibody against myelin-associated neurite growth inhibitors. Nature 343:269–272 Schwartz M (2005) T-cell-based vaccination against neurodegeneration: a new therapeutic approach. Retina 25:S33–S35 Schwartz M, Cohen I, Lazarov-Spiegler O, Moalem G, Yoles E (1999) The remedy may lie in ourselves: prospects for immune cell therapy in central nervous system protection and repair. J Mol Med 77:713–717 Schwartz M, Ziv Y (2008) Immunity to self and self-maintenance: a unified theory of brain pathologies. Trends Immunol 29:211–219 Segal R et al (1997) Effect of aging on cytokine production in normal and experimental systemic lupus erythematosus afflicted mice. Mech Ageing Dev 96:47–58 Serpe CJ, Kohm AP, Huppenbauer CB, Sanders VM, Jones KJ (1999) Exacerbation of facial motoneuron loss after facial nerve transection in severe combined immunodeficient (SCID) mice. J Neurosci 19:RC7 Shearer GM (1997) Th1/Th2 changes in aging. Mech Ageing Dev 94:1–5 Shechter R, Ziv Y, Schwartz M (2007) New GABAergic interneurons supported by myelin-specific T cells are formed in intact adult spinal cord. Stem Cells 25:2277–2282 Shevach EM (2002) CD4+ CD25+ suppressor T cells: more questions than answers. Nat Rev Immunol 2:389–400 Sicotte M et al (2003) Immunization with myelin or recombinant Nogo-66/MAG in alum promotes axon regeneration and sprouting after corticospinal tract lesions in the spinal cord. Mol Cell Neurosci 23:251–263 Silver J, Miller JH (2004) Regeneration beyond the glial scar. Nat Rev Neurosci 5:146–156 Sroga JM, Jones TB, Kigerl KA, McGaughy VM, Popovich PG (2003) Rats and mice exhibit distinct inflammatory reactions after spinal cord injury. J Comp Neurol 462:223–240 Stichel CC et al (1999) Inhibition of collagen IV deposition promotes regeneration of injured CNS axons. Eur J Neurosci 11:632–646 Stichel CC, Muller HW (1998) The CNS lesion scar: new vistas on an old regeneration barrier. Cell Tissue Res 294:1–9 Stinissen P, Raus J, Zhang J (1997) Autoimmune pathogenesis of multiple sclerosis: role of autoreactive T lymphocytes and new immunotherapeutic strategies. Crit Rev Immunol 17:33–75

232

S. Hendrix and R. Nitsch

Streit WJ (2002) Microglia as neuroprotective, immunocompetent cells of the CNS. Glia 40: 133–139 Streit WJ, Walter SA, Pennell NA (1999) Reactive microgliosis. Prog Neurobiol 57:563–581 Tang X, Davies JE, Davies SJ (2003) Changes in distribution, cell associations, and protein expression levels of NG2, neurocan, phosphacan, brevican, versican V2, and tenascin-C during acute to chronic maturation of spinal cord scar tissue. J Neurosci Res 71:427–444 Taoka Y et al (1997) Role of neutrophils in spinal cord injury in the rat. Neuroscience 79: 1177–1182 Taoka Y, Okajima K, Murakami K, Johno M, Naruo M (1998) Role of neutrophil elastase in compression-induced spinal cord injury in rats. Brain Res 799:264–269 Tian L, Rauvala H, Gahmberg CG (2009) Neuronal regulation of immune responses in the central nervous system. Trends Immunol 30:91–99 Tonai T et al (2001) A neutrophil elastase inhibitor (ONO-5046) reduces neurologic damage after spinal cord injury in rats. J Neurochem 78:1064–1072 Vela JM, Yanez A, Gonzalez B, Castellano B (2002) Time course of proliferation and elimination of microglia/macrophages in different neurodegenerative conditions. J Neurotrauma 19: 1503–1520 Wang WZ et al (1992) Myelin antigen reactive T cells in cerebrovascular diseases. Clin Exp Immunol 88:157–162 Weaver CT, Harrington LE, Mangan PR, Gavrieli M, Murphy KM (2006) Th17: an effector CD4 T cell lineage with regulatory T cell ties. Immunity 24:677–688 Weiner HL (2001) Induction and mechanism of action of transforming growth factor-beta-secreting Th3 regulatory cells. Immunol Rev 182:207–214 Weinstein JR et al (2005) Unraveling thrombin’s true microglia-activating potential: markedly disparate profiles of pharmaceutical-grade and commercial-grade thrombin preparations. J Neurochem 95:1177–1187 Wekerle H, Sun D, Oropeza-Wekerle RL, Meyermann R (1987) Immune reactivity in the nervous system: modulation of T-lymphocyte activation by glial cells. J Exp Biol 132:43–57 Whalen MJ et al (1999) Neutrophils do not mediate blood-brain barrier permeability early after controlled cortical impact in rats. J Neurotrauma 16:583–594 Wierenga EA et al (1990) Evidence for compartmentalization of functional subsets of CD2+ T lymphocytes in atopic patients. J Immunol 144:4651–4656 Wirenfeldt M et al (2005) Reactive microgliosis engages distinct responses by microglial subpopulations after minor central nervous system injury. J Neurosci Res 82:507–514 Wolf SA et al (2002) Neuroprotection by T-cells depends on their subtype and activation state. J Neuroimmunol 133:72–80 Youssef S et al (2002) The HMG-CoA reductase inhibitor, atorvastatin, promotes a Th2 bias and reverses paralysis in central nervous system autoimmune disease. Nature 420:78–84 Zehntner SP et al (2005) Neutrophils that infiltrate the central nervous system regulate T cell responses. J Immunol 174:5124–5131 Ziv Y et al (2006) Immune cells contribute to the maintenance of neurogenesis and spatial learning abilities in adulthood. Nat Neurosci 9:268–275 Ziv Y, Avidan H, Pluchino S, Martino G, Schwartz M (2006) Synergy between immune cells and adult neural stem/progenitor cells promotes functional recovery from spinal cord injury. Proc Natl Acad Sci USA 103:13174–13179

Index

A Action potential, 35–48, 57–60, 69–71, 74–76, 90–94 Adaptive immune system in CNS injury controversy about “protective autoimmunity,” 218–219 neuronal injury in T cell-deficient mice, 218 self-reactive T cells in CNS injury, 217–218 T cells in healthy and injured CNS, 216–217 Anti-inflammatory T-cell effects, 224–225 Anuclear ‘models’ of degeneration, 195–196 Axo-axonal synapses, 90–95 Axo-axonic gap junctions, physiological significance of, 95–97 Axon, insights in information processing in, 55 complexity of axonal arborization, 56 functional computation in axon axonal integration, 59–62 shaping action potential, 56–58 signal amplification along axon, 59 ping-pong propagation, 69–71 propagation failures, 62 branch points and swellings, 62–64 frequency-dependent propagation failures, 64–68 frequency-independent propagation failures, 68–69 spike timing in axon axonal delay imposed by axonal length, 71–73 delays imposed by axonal irregularities and ion channels, 73 electric coupling in axons and fast synchronization, 74–75 ephaptic interactions and axonal spike synchronization, 74 Axonal arborization, complexity of, 56

Axonal computation, 55, 69, 75 Axonal domains, 143–144, 158–159, 163–164 definition of cortical column by geometry of, 143–144 of neocortical excitatory neurons, 158–159 Axonal function, sodium signals and their significance for, 35–36 sodium signals and measurement of [Na+ ]i, 36 sodium signals in dendrites and glial processes, 36–40 sodium signals in axons, 40–44 pathophysiological implications of [Na+ ]i changes, 44–46 possible physiological roles of [Na+ ]i changes, 46–48 Axonal gap junction, 91, 94–95 Axonal integration, 59–62 Axonal loss, 181, 192 Axonal morphology, 56, 76, 158 Axonal navigation, basic principles of, 3–4 Axonal perspective on cortical circuits, 117–119 capturing axon morphology, 120–126 circuits, 119–120 computation in daisy architectures, 134–136 improving Peters’ rule, 131–133 problem of choice, 127–129 wiring neurons, 129–131 Axonal projection pattern, 157–159, 166–170 Axonal spike synchronization, ephaptic interactions and, 74 Axon degeneration, 181 axon as functionally distinct compartment, 181–184 cellular responses in axon degeneration ensheathing cells, 196–197 macrophages, 197–199

D. Feldmeyer, J.H.R. Lübke (eds.), New Aspects of Axonal Structure and Function, C Springer Science+Business Media, LLC 2010 DOI 10.1007/978-1-4419-1676-1, 

233

234 Axon degeneration (cont.) lessons from anuclear ‘models’ of degeneration, 195–196 morphological changes, 184–185 pathways involved in axonal degeneration, 190–191 calcium/calpains, 192–193 mitochondria, 192 mitogen-activated protein kinases (MAPKs), 191–192 ubiquitin proteasome system, 193–194 protein expression changes, 186–187 Wlds , 187–190 Axon growth by T-cell cytokines, direct stimulation of, 225 Axon of excitatory neurons in neocortex, 157–158 highly variable axonal domains of neocortical excitatory neurons, 158–159 layer 2/3, 159–161 layer 4, 161 spiny stellate and star pyramidal neurons, 161–163 star pyramidal cells, 163–165 layer 5, 165 sublamina 5A, 165–166 sublamina 5B, 166–168 layer 6, 168 sublamina 6A, 168–170 sublamina 6B, 170–171 subcellular axonal targets, 171–172 Axons action potential-evoked [Na+ ]i transients, 40–41 boutons, 56 defined, 55 dendrogram of pyramidal cell, 126 diameter, 103–112 electrical coupling of, 85 axo-axonal synapses, 90–95 mechanisms of electrical coupling, 85–88 methodological issues, 88–90 physiological significance of axo-axonic gap junctions, 95–97 propagation/stimulation, 65 shaping of action potential in, 57 signal amplification along, 59 sodium signals in, 40–44 pathophysiological implications of [Na+ ]i changes, 44–46

Index possible physiological roles of [Na+ ]i changes, 46–48 spike timing in axonal delay imposed by axonal length, 71–73 delays imposed by axonal irregularities and ion channels, 73 electric coupling in axons and fast synchronization, 74–75 ephaptic interactions and axonal spike synchronization, 74 B Barrel cortex, 25 C Calcium/calpains, 192–193 Cellular responses in axon degeneration ensheathing cells, 196–197 macrophages, 197–199 Cerebral cortical circuits, subplate and formation of earliest, 19–22 clinical importance of understanding of early cortical circuits, 22–23 manipulation of sensory periphery alters subplate integration into barrel field, 25–26 microarray screen of murine subplate neurons, 24 studies of subplate neuron integration into cortical/extracortical circuitry in reporter gene expressing mouse models, 24–25 subplate cell populations in mutants with cortical migration defects, 26–27 Cerebral cortical development, 19, 22 CNS T cells in healthy and injured, 216–217 wound healing responses of inflammation in autoimmune neurodegeneration vs. CNS repair after trauma, 210 multi-phase responses of CNS after trauma, 211–216 CNS injury adaptive immune system in controversy about “protective autoimmunity,” 218–219 neuronal injury in T cell-deficient mice, 218 self-reactive T cells in CNS injury, 217–218 T cells in healthy and injured CNS, 216–217

Index and repair, T-helper cell subpopulations in, 219 analysis of Th1/Th2 ratio, 220 subpopulations of T cells for neuroprotective/pro-regenerative effects, 221–222 CNS injury-induced immunosuppression results in systemic Th2 shift, 222 CNS trauma, 209, 218, 225 Commissural axon guidance, molecular aspects of, 3 basic principles of axonal navigation, 3–4 commissural neurons as a model system to study axonal pathfinding, 4 commissural axon guidance in Drosophila melanogaster, 11–14 commissural axon guidance in vertebrate spinal cord, 4–11 Commissural neurons as model system to study axonal pathfinding, 4 commissural axon guidance in Drosophila melanogaster, 11–14 commissural axon guidance in vertebrate spinal cord, 4–11 Conduction velocity, 64, 73–74, 103, 110 as function of axon diameter for myelinated and non-myelinated axons, 104t Connexins, 86–87, 90 Cortical circuits, axonal perspective on, 117–119 capturing axon morphology, 120–126 circuits, 119–120 computation in daisy architectures, 134–136 problem of choice, 127–129 wiring neurons, 129–131 Cortical columns, 143–144 by geometry of axonal domains, definition of, 143–144 inference of synaptic connectivity within, 144–146 neuronal connectivity within and between, 141–143 classification of inhibitory interneurons, 146–148 definition of cortical column by geometry of axonal domains, 143–144 inference of synaptic connectivity within cortical column, 144–146

235 local versus lateral inhibitors in neocortex, 148–151 subsequent classification steps, 151–152 D Degeneration, anuclear ‘models’ of, 195–196 Dendrites, 35–43, 71, 121–122, 127–130, 132–135 Dendrites, sodium signals in, 36–40 Dendrogram of pyramidal cell axon, 126 Drosophila melanogaster, commissural axon guidance in, 11–14 guidance across midline, 12–13 guidance toward midline, 11 longitudinal projection after crossing, 13–14 E Electrical coupling, 87–88 mechanisms of, 85–88 Electrical coupling of axons, 85 axo-axonal synapses, 90–95 mechanisms of electrical coupling, 85–88 methodological issues, 88–90 physiological significance of axo-axonic gap junctions, 95–97 Excitatory neurons in neocortex, axon of, 157–158 highly variable axonal domains of neocortical excitatory neurons, 158–159 layer 2/3, 159–161 layer 4, 161 spiny stellate and star pyramidal neurons, 161–163 star pyramidal cells, 163–165 layer 5, 165 sublamina 5A, 165–166 sublamina 5B, 166–168 layer 6, 168 sublamina 6A, 168–170 sublamina 6B, 170–171 subcellular axonal targets, 171–172 F Floorplate, 4–5 guidance across, 6–9 guidance toward, 5–6 G Glia, synaptically evoked [Na+ ]i changes in dendrites and, 37–40 Glial processes, sodium signals in, 36–40

236 H Hippocampus, 59, 74, 87, 141 I Immunosuppression, 222 Inhibitory interneurons, classification of, 146–148 Innervations, 24–25, 144–148, 150–151 Interneurons, 4, 56, 58–59, 74, 146–148, 158 M Mitochondria, 44–45, 192 Mitogen-activated protein kinases (MAPKs), 191–192 Myelinated axon, 42–43, 74, 103–104 Myelinate/non-myelinated axons, 103–104 axonal cost with fixed conduction delay, 105–107 axon with finite conduction delay cost, 107–109 comparison with experiments, 109–110 N Neocortex, 87, 96, 117, 157 local circuits of, 118 local versus lateral inhibitors in, 148–151 Neocortical excitatory neurons, axonal domains of, 158–159 Netrins, 11–12 Neuronal connectivity, 141–152 Neuronal injury in T cell-deficient mice, 218 Neuroregeneration, T-helper cells and, 209 Non-myelinated axon, 103–108, 110–112 P Peters’ Rule, 129–133 Ping-pong propagation, 69–71 Prepotentials, 92–94 Propagation failures, insights in information processing in axon, 62 branch points and swellings, 62–64 frequency-dependent propagation failures, 64–68 frequency-independent propagation failures, 68–69 Protective autoimmunity, 218–219 Pyramidal cell axon, dendrogram of, 126 Pyramidal cells, 36, 92–94, 123–129, 131–135, 158–170

Index R Reflected propagation, 69, 71, 75–76 Regeneration after CNS lesion, see T-helper cells and neuroregeneration Robo proteins, 12 S Sodium signals and their significance for axonal function, 35–36 sodium signals and measurement of [Na+ ]i, 36 sodium signals in dendrites and glial processes, 36–40 sodium signals in axons, 40–44 pathophysiological implications of [Na+ ]i changes, 44–46 possible physiological roles of [Na+ ]i changes, 46–48 Spikelets, 75, 92–94 Spike timing, 71–75 in axon axonal delay imposed by axonal length, 71–73 delays imposed by axonal irregularities and ion channels, 73 electric coupling in axons and fast synchronization, 74–75 ephaptic interactions and axonal spike synchronization, 74 Spinal cord injury, 46, 186, 216–217, 222 Spiny stellate, 161–163 cell, 121, 132, 162–165 Squid giant axon, 56 Star pyramidal cells, 163–165 Star pyramidal neurons, 161–163 Subplate neurons, 20–21, 24, 170 Synaptic activity, 37, 39–40, 61 Synaptic connectivity within cortical column, inference of, 144–146 T Target neuron specificity, 172 T cell-deficient mice, neuronal injury in, 218 T cells contribute to CNS repair, 223–224 anti-inflammatory T-cell effects, 224–225 direct stimulation of axon growth by T-cell cytokines, 225 modulation of endogenous and exogenous stem cells, 224 modulation of microglia functions, 224 neurotrophin secretion by T cells, 224 Th1, 220–223 Th2, 220–223

Index Th2 cells, T-cell effects attributable to, 222–223 T-helper cells and neuroregeneration, 209 adaptive immune system in CNS injury controversy about “protective autoimmunity,” 218–219 neuronal injury in T cell-deficient mice, 218 self-reactive T cells in CNS injury, 217–218 T cells in healthy and injured CNS, 216–217 CNS injury-induced immunosuppression results in systemic Th2 shift, 222 T-cell effects attributable to Th2 cells, 222–223 T cells contribute to CNS repair, 223–224 anti-inflammatory T-cell effects, 224–225 direct stimulation of axon growth by T-cell cytokines, 225 modulation of endogenous and exogenous stem cells, 224 modulation of microglia functions, 224 neurotrophin secretion by T cells, 224 T-helper cell subpopulations in CNS injury and repair, 219 analysis of Th1/Th2 ratio, 220 subpopulations of T cells for neuroprotective/pro-regenerative effects, 221–222

237 wound healing responses of CNS inflammation in autoimmune neurodegeneration vs. CNS repair after trauma, 210 multi-phase responses of CNS after trauma, 211–216 U Ubiquitin proteasome system, 193–194 V Vertebrate spinal cord, commissural axon guidance in, 4–11 guidance across floorplate, 6–9 guidance toward floorplate, 5–6 rostral turning and longitudinal projection after crossing, 9–11 Visual cortex, 119, 122–123 Voltage-dependent ion channels, 86 W Wallerian degeneration, 184–185, 187–188, 190–198 Wound healing responses of CNS inflammation in autoimmune neurodegeneration vs. CNS repair after trauma, 210 multi-phase responses of CNS after trauma, 211–216