• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
Neuroscience - Purves et al 3ed

Neuroscience - Purves et al 3ed






Total Views
Views on SlideShare
Embed Views



0 Embeds 0

No embeds


Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
Post Comment
Edit your comment

    Neuroscience - Purves et al 3ed Neuroscience - Purves et al 3ed Document Transcript

    • NEUROSCIENCEThird EditionPurves3/eFM 5/13/04 12:59 PM Page i
    • Purves3/eFM 5/13/04 12:59 PM Page ii
    • NEUROSCIENCE: Third EditionCopyright © 2004 by Sinauer Associates, Inc. All rights reserved.This book may not be reproduced in whole or in part without permission.Address inquiries and orders toSinauer Associates, Inc.23 Plumtree RoadSunderland, MA 01375 U.S.A.www.sinauer.comFAX: 413-549-1118orders@sinauer.compublish@sinauer.comLibrary of Congress Cataloging-in-Publication DataNeuroscience / edited by Dale Purves ... [et al.].— 3rd ed.p. ; cm.Includes bibliographical references and index.ISBN 0-87893-725-0 (casebound : alk. paper)1. Neurosciences.[DNLM: 1. Nervous System Physiology. 2. Neurochemistry.WL 102 N50588 2004] I. Purves, Dale.QP355.2.N487 2004612.8—dc22 2004003973Printed in U.S.A.5 4 3 2 1THE COVERDorsal view of the human brain.(Courtesy of S. Mark Williams.)Purves3/eFM 5/13/04 12:59 PM Page iv
    • George J. Augustine, Ph.D.Dona M. Chikaraishi, Ph.D.Michael D. Ehlers, M.D., Ph.D.Gillian Einstein, Ph.D.David Fitzpatrick, Ph.D.William C. Hall, Ph.D.Erich Jarvis, Ph.D.Lawrence C. Katz, Ph.D.Julie Kauer, Ph.D.Anthony-Samuel LaMantia, Ph.D.James O. McNamara, M.D.Richard D. Mooney, Ph.D.Miguel A. L. Nicolelis, M.D., Ph.D.Dale Purves, M.D.Peter H. Reinhart, Ph.D.Sidney A. Simon, Ph.D.J. H. Pate Skene, Ph.D.James Voyvodic, Ph.D.Leonard E. White, Ph.D.S. Mark Williams, Ph.D.UNIT EDITORSUNIT I: George J. AugustineUNIT II: David FitzpatrickUNIT III: William C. HallUNIT IV: Anthony-Samuel LaMantiaUNIT V: Dale PurvesContributorsPurves3/eFM 5/13/04 12:59 PM Page v
    • Purves3/eFM 5/13/04 12:59 PM Page vi
    • 1. Studying the Nervous Systems of Humans and Other Animals 1UNIT I NEURAL SIGNALING2. Electrical Signals of Nerve Cells 313. Voltage-Dependent Membrane Permeability 474. Channels and Transporters 695. Synaptic Transmission 936. Neurotransmitters, Receptors, and Their Effects 1297. Molecular Signaling within Neurons 165UNIT II SENSATION AND SENSORY PROCESSING8. The Somatic Sensory System 1899. Pain 20910. Vision: The Eye 22911. Central Visual Pathways 25912. The Auditory System 28313. The Vestibular System 31514. The Chemical Senses 337UNIT III MOVEMENT AND ITS CENTRAL CONTROL15. Lower Motor Neuron Circuits and Motor Control 37116. Upper Motor Neuron Control of the Brainstem and Spinal Cord 39317. Modulation of Movement by the Basal Ganglia 41718. Modulation of Movement by the Cerebellum 43519. Eye Movements and Sensory Motor Integration 45320. The Visceral Motor System 469UNIT IV THE CHANGING BRAIN21. Early Brain Development 50122. Construction of Neural Circuits 52123. Modification of Brain Circuits as a Result of Experience 55724. Plasticity of Mature Synapses and Circuits 575UNIT V COMPLEX BRAIN FUNCTIONS25. The Association Cortices 61326. Language and Speech 63727. Sleep and Wakefulness 65928. Emotions 68729. Sex, Sexuality, and the Brain 71130. Memory 733APPENDIX A THE BRAINSTEM AND CRANIAL NERVES 755APPENDIX B VASCULAR SUPPLY, THE MENINGES, AND THE VENTRICULAR SYSTEM 763Contents in BriefPurves3/eFM 5/13/04 12:59 PM Page vii
    • Chapter 1 Studying the Nervous Systemsof Humans and Other Animals 1Overview 1Genetics, Genomics, and the Brain 1The Cellular Components of the Nervous System 2Neurons 4Neuroglial Cells 8Cellular Diversity in the Nervous System 9Neural Circuits 11Overall Organization of the Human NervousSystem 14Neuroanatomical Terminology 16The Subdivisions of the Central Nervous System 18Organizational Principles of Neural Systems 20Functional Analysis of Neural Systems 23Analyzing Complex Behavior 24BOX A Brain Imaging Techniques 25Summary 26ContentsUnit I NEURAL SIGNALINGChapter 2 Electrical Signalsof Nerve Cells 31Overview 31Electrical Potentials across Nerve Cell Membranes 31How Ionic Movements Produce Electrical Signals 34The Forces That Create Membrane Potentials 36Electrochemical Equilibrium in an Environment withMore Than One Permeant Ion 38The Ionic Basis of the Resting Membrane Potential 40BOX A The Remarkable Giant Nerve Cellsof Squid 41The Ionic Basis of Action Potentials 43BOX B Action Potential Formand Nomenclature 44Summary 45Chapter 3 Voltage-Dependent MembranePermeability 47Overview 47Ionic Currents Across Nerve Cell Membranes 47BOX A The Voltage Clamp Method 48Two Types of Voltage-Dependent Ionic Current 49Two Voltage-Dependent Membrane Conductances 52Reconstruction of the Action Potential 54Long-Distance Signaling by Means of ActionPotentials 56BOX B Threshold 57BOX C Passive Membrane Properties 60The Refractory Period 61Increased Conduction Velocity as a Resultof Myelination 63Summary 65BOX D Multiple Sclerosis 66Preface xviAcknowledgments xviiSupplements to Accompany NEUROSCIENCE xviiiPurves3/eFM 5/13/04 12:59 PM Page viii
    • Chapter 4 Channels and Transporters 69Overview 69Ion Channels Underlying Action Potentials 69BOX A The Patch Clamp Method 70The Diversity of Ion Channels 73BOX B Expression of Ion Channels in XenopusOocytes 75Voltage-Gated Ion Channels 76Ligand-Gated Ion Channels 78Stretch- and Heat-Activated Channels 78The Molecular Structure of Ion Channels 79BOX C Toxins That Poison Ion Channels 82BOX D Diseases Caused by Altered IonChannels 84Active Transporters Create and Maintain IonGradients 86Functional Properties of the Na+/K+Pump 87The Molecular Structure of the Na+/K+Pump 89Summary 90Chapter 5 Synaptic Transmission 93Overview 93Electrical Synapses 93Signal Transmission at Chemical Synapses 96Properties of Neurotransmitters 96BOX A Criteria That Define aNeurotransmitter 99Quantal Release of Neurotransmitters 102Release of Transmitters from Synaptic Vesicles 103Local Recycling of Synaptic Vesicles 105The Role of Calcium in Transmitter Secretion 107BOX B Diseases That Affect the PresynapticTerminal 108Molecular Mechanisms of Transmitter Secretion 110Neurotransmitter Receptors 113BOX C Toxins That Affect TransmitterRelease 115Postsynaptic Membrane Permeability Changes duringSynaptic Transmission 116Excitatory and Inhibitory Postsynaptic Potentials 121Summation of Synaptic Potentials 123Two Families of Postsynaptic Receptors 124Summary 126Chapter 6 Neurotransmitters and TheirReceptors 129Overview 129Categories of Neurotransmitters 129Acetylcholine 129BOX A Addiction 134BOX B Neurotoxins that Act on PostsynapticReceptors 136Glutamate 137BOX C Myasthenia Gravis: An AutoimmuneDisease of Neuromuscular Synapses 140GABA and Glycine 143BOX D Excitotoxicity Following Acute BrainInjury 145The Biogenic Amines 147BOX E Biogenic Amine Neurotransmitters andPsychiatric Disorders 148ATP and Other Purines 152Peptide Neurotransmitters 153Unconventional Neurotransmitters 157BOX F Marijuana and the Brain 160Summary 161Chapter 7 Molecular Signaling withinNeurons 165Overview 165Strategies of Molecular Signaling 165The Activation of Signaling Pathways 167Receptor Types 168G-Proteins and Their Molecular Targets 170Second Messengers 172Second Messenger Targets: Protein Kinases andPhosphatases 175Nuclear Signaling 178Examples of Neuronal Signal Transduction 181Summary 184Contents ixPurves3/eFM 5/13/04 12:59 PM Page ix
    • x ContentsChapter 8 The Somatic Sensory System 189Overview 189Cutaneous and Subcutaneous Somatic SensoryReceptors 189Mechanoreceptors Specialized to Receive TactileInformation 192Differences in Mechanosensory Discrimination acrossthe Body Surface 193BOX A Receptive Fields and Sensory Mapsin the Cricket 195BOX B Dynamic Aspects of Somatic SensoryReceptive Fields 196Mechanoreceptors Specialized for Proprioception 197Active Tactile Exploration 199The Major Afferent Pathway for MechanosensoryInformation: The Dorsal Column–Medial LemniscusSystem 199The Trigeminal Portion of the MechanosensorySystem 202BOX C Dermatomes 202The Somatic Sensory Components of the Thalamus 203The Somatic Sensory Cortex 203Higher-Order Cortical Representations 206BOX D Patterns of Organization within theSensory Cortices: Brain Modules 207Summary 208Chapter 9 Pain 209Overview 209Nociceptors 209Transduction of Nociceptive Signals 211BOX A Capsaicin 212Central Pain Pathways 213BOX B Referred Pain 215BOX C A Dorsal Column Pathway for VisceralPain 218Sensitization 220BOX D Phantom Limbs and Phantom Pain 222Descending Control of Pain Perception 224The Placebo Effect 224The Physiological Basis of Pain Modulation 225Summary 227Chapter 10 Vision:The Eye 229Overview 229Anatomy of the Eye 229The Formation of Images on the Retina 231BOX A Myopia and Other Refractive Errors 232The Retina 234Phototransduction 236BOX B Retinitis Pigmentosa 239Functional Specialization of the Rod and ConeSystems 240BOX C Macular Degeneration 243Anatomical Distribution of Rods and Cones 244Cones and Color Vision 245BOX D The Importance of Context in ColorPerception 247Retinal Circuits for Detecting LuminanceChange 249BOX E The Perception of Light Intensity 250Contribution of Retinal Circuits to LightAdaptation 254Summary 257Chapter 11 Central Visual Pathways 259Overview 259Central Projections of Retinal Ganglion Cells 259BOX A The Blind Spot 262The Retinotopic Representation of the Visual Field 263Visual Field Deficits 267The Functional Organization of the Striate Cortex 269The Columnar Organization of the Striate Cortex 271BOX B Random Dot Stereograms and RelatedAmusements 272Division of Labor within the Primary VisualPathway 275BOX C Optical Imaging of Functional Domains inthe Visual Cortex 276The Functional Organization of Extrastriate VisualAreas 278Summary 281Chapter 12 The Auditory System 283Overview 283Sound 283The Audible Spectrum 284Unit II SENSATION AND SENSORY PROCESSINGPurves3/eFM 5/13/04 12:59 PM Page x
    • Chapter 15 Lower Motor Neuron Circuitsand Motor Control 371Overview 371Neural Centers Responsible for Movement 371Motor Neuron–Muscle Relationships 373The Motor Unit 375The Regulation of Muscle Force 377The Spinal Cord Circuitry Underlying Muscle StretchReflexes 379A Synopsis of Auditory Function 285BOX A Four Causes of Acquired Hearing Loss 285BOX B Music 286The External Ear 287The Middle Ear 289The Inner Ear 289BOX C Sensorineural Hearing Loss and CochlearImplants 290BOX D The Sweet Sound of Distortion 295Hair Cells and the Mechanoelectrical Transduction ofSound Waves 294Two Kinds of Hair Cells in the Cochlea 300Tuning and Timing in the Auditory Nerve 301How Information from the Cochlea Reaches Targets inthe Brainstem 303Integrating Information from the Two Ears 303Monaural Pathways from the Cochlear Nucleus to theLateral Lemniscus 307Integration in the Inferior Colliculus 307The Auditory Thalamus 308The Auditory Cortex 309BOX E Representing Complex Sounds in theBrains of Bats and Humans 310Summary 313Chapter 13 The Vestibular System 315Overview 315The Vestibular Labyrinth 315Vestibular Hair Cells 316The Otolith Organs: The Utricle and Saccule 317BOX A A Primer on Vestibular Navigation 318BOX B Adaptation and Tuning of Vestibular HairCells 320How Otolith Neurons Sense Linear Forces 322The Semicircular Canals 324How Semicircular Canal Neurons Sense AngularAccelerations 325BOX C Throwing Cold Water on the VestibularSystem 326Central Pathways for Stabilizing Gaze, Head, andPosture 328Vestibular Pathways to the Thalamus and Cortex 331BOX D Mauthner Cells in Fish 332Summary 333Chapter 14 The Chemical Senses 337Overview 337The Organization of the Olfactory System 337Olfactory Perception in Humans 339Physiological and Behavioral Responses toOdorants 341The Olfactory Epithelium and Olfactory ReceptorNeurons 342BOX A Olfaction, Pheromones, and Behavior inthe Hawk Moth 344The Transduction of Olfactory Signals 345Odorant Receptors 346Olfactory Coding 348The Olfactory Bulb 350BOX B Temporal“Coding”of OlfactoryInformation in Insects 350Central Projections of the Olfactory Bulb 353The Organization of the Taste System 354Taste Perception in Humans 356Idiosyncratic Responses to Tastants 357The Organization of the Peripheral Taste System 359Taste Receptors and the Transduction of TasteSignals 360Neural Coding in the Taste System 362Trigeminal Chemoreception 363Summary 366Contents xiUnit III MOVEMENT AND ITS CENTRAL CONTROLPurves3/eFM 5/13/04 12:59 PM Page xi
    • xii ContentsThe Influence of Sensory Activity on Motor Behavior381Other Sensory Feedback That Affects MotorPerformance 382BOX A Locomotion in the Leech and the Lamprey384Flexion Reflex Pathways 387Spinal Cord Circuitry and Locomotion 387BOX B The Autonomy of Central PatternGenerators: Evidence from the LobsterStomatogastric Ganglion 388The Lower Motor Neuron Syndrome 389BOX C Amyotrophic Lateral Sclerosis 391Summary 391Chapter 16 Upper Motor Neuron Controlof the Brainstem and SpinalCord 393Overview 393Descending Control of Spinal Cord Circuitry:General Information 393Motor Control Centers in the Brainstem: Upper MotorNeurons That Maintain Balance and Posture 397BOX A The Reticular Formation 398The Corticospinal and Corticobulbar Pathways:Upper Motor Neurons That Initiate ComplexVoluntary Movements 402BOX B Descending Projections to Cranial NerveMotor Nuclei and Their Importancein Diagnosing the Cause of MotorDeficits 404Functional Organization of the Primary Motor Cortex405BOX C What Do Motor Maps Represent? 408The Premotor Cortex 411BOX D Sensory Motor Talents and CorticalSpace 410Damage to Descending Motor Pathways: The UpperMotor Neuron Syndrome 412BOX E Muscle Tone 414Summary 415Chapter 17 Modulation of Movement bythe Basal Ganglia 417Overview 417Projections to the Basal Ganglia 417Projections from the Basal Ganglia to Other BrainRegions 422Evidence from Studies of Eye Movements 423Circuits within the Basal Ganglia System 424BOX A Huntington’s Disease 426BOX B Parkinson’s Disease: An Opportunity forNovel Therapeutic Approaches 429BOX C Basal Ganglia Loops and Non-MotorBrain Functions 432Summary 433Chapter 18 Modulation of Movement bythe Cerebellum 435Overview 435Organization of the Cerebellum 435Projections to the Cerebellum 438Projections from the Cerebellum 440Circuits within the Cerebellum 441BOX A Prion Diseases 444Cerebellar Circuitry and the Coordination of OngoingMovement 445Futher Consequences of Cerebellar Lesions 448Summary 449BOX B Genetic Analysis of Cerebellar Function 450Chapter 19 Eye Movements and SensoryMotor Integration 453Overview 453What Eye Movements Accomplish 453The Actions and Innervation of Extraocular Muscles454BOX A The Perception of Stabilized RetinalImages 456Types of Eye Movements and Their Functions 457Neural Control of Saccadic Eye Movements 458BOX B Sensory Motor Integration in theSuperior Colliculus 462Neural Control of Smooth Pursuit Movements 466Neural Control of Vergence Movements 466Summary 467Chapter 20 The Visceral Motor System 469Overview 469Early Studies of the Visceral Motor System 469Distinctive Features of the Visceral Motor System 470The Sympathetic Division of the Visceral MotorSystem 471The Parasympathetic Division of the Visceral MotorSystem 476The Enteric Nervous System 479Sensory Components of the Visceral Motor System 480Purves3/eFM 5/13/04 12:59 PM Page xii
    • Chapter 21 Early Brain Development 501Overview 501The Initial Formation of the Nervous System:Gastrulation and Neurulation 501The Molecular Basis of Neural Induction 503BOX A Stem Cells: Promise and Perils 504BOX B Retinoic Acid:Teratogen and InductiveSignal 506Formation of the Major Brain Subdivisions 510BOX C Homeotic Genes and Human BrainDevelopment 513BOX D Rhombomeres 514Genetic Abnormalities and Altered Human BrainDevelopment 515The Initial Differentiation of Neurons and Glia 516BOX E Neurogenesis and Neuronal Birthdating517The Generation of Neuronal Diversity 518Neuronal Migration 520BOX F Mixing It Up: Long-Distance NeuronalMigration 524Summary 525Chapter 22 Construction of NeuralCircuits 527Overview 527The Axonal Growth Cone 527Non-Diffusible Signals for Axon Guidance 528BOX A Choosing Sides: Axon Guidance at theOptic Chiasm 530Diffusible Signals for Axon Guidance:Chemoattraction and Repulsion 534The Formation of Topographic Maps 537Selective Synapse Formation 539BOX B Molecular Signals That Promote SynapseFormation 542Trophic Interactions and the Ultimate Size of NeuronalPopulations 543Further Competitive Interactions in the Formation ofNeuronal Connections 545Molecular Basis of Trophic Interactions 547BOX C Why Do Neurons Have Dendrites? 548BOX D The Discovery of BDNF and theNeurotrophin Family 552Neurotrophin Signaling 553Summary 554Chapter 23 Modification of Brain Circuitsas a Result of Experience 557Overview 557Critical Periods 557BOX A Built-In Behaviors 558The Development of Language:Example of a Human Critical Period 559BOX B Birdsong 560Critical Periods in Visual System Development 562Effects of Visual Deprivation on Ocular Dominance 563BOX C Transneuronal Labeling with RadioactiveAmino Acids 564Visual Deprivation and Amblyopia in Humans 568Mechanisms by which Neuronal Activity Affects theDevelopment of Neural Circuits 569Cellular and Molecular Correlates of Activity-Dependent Plasticity during Critical Periods 572Evidence for Critical Periods in Other SensorySystems 572Summary 573Contents xiiiUnit IV THE CHANGING BRAINCentral Control of Visceral Motor Functions 483BOX A The Hypothalamus 484Neurotransmission in the Visceral Motor System 487BOX B Horner’s Syndrome 488BOX C Obesity and the Brain 490Visceral Motor Reflex Functions 491Autonomic Regulation of Cardiovascular Function 491Autonomic Regulation of the Bladder 493Autonomic Regulation of Sexual Function 496Summary 498Purves3/eFM 5/13/04 12:59 PM Page xiii
    • xiv ContentsChapter 25 The Association Cortices 613Overview 613The Association Cortices 613An Overview of Cortical Structure 614Specific Features of the Association Cortices 615BOX A A More Detailed Look at CorticalLamination 617Lesions of the Parietal Association Cortex: Deficits ofAttention 619Lesions of the Temporal Association Cortex:Deficits of Recognition 622Lesions of the Frontal Association Cortex: Deficits ofPlanning 623BOX B Psychosurgery 625“Attention Neurons” in the Monkey Parietal Cortex 626“Recognition Neurons” in the Monkey TemporalCortex 627“Planning Neurons” in the Monkey Frontal Cortex 630BOX C Neuropsychological Testing 632BOX D Brain Size and Intelligence 634Summary 635Chapter 26 Language and Speech 637Overview 637Language Is Both Localized and Lateralized 637Aphasias 638BOX A Speech 640BOX B Do Other Animals Have Language? 642BOX C Words and Meaning 645A Dramatic Confirmation of Language Lateralization646Anatomical Differences between the Right and LeftHemispheres 648Mapping Language Functions 649BOX D Language and Handedness 650The Role of the Right Hemisphere in Language 654Sign Language 655Summary 656Chapter 27 Sleep and Wakefulness 659Overview 659Why Do Humans (and Many Other Animals) Sleep?659BOX A Styles of Sleep in Different Species 661Unit V COMPLEX BRAIN FUNCTIONSChapter 24 Plasticity of Mature Synapsesand Circuits 575Overview 575Synaptic Plasticity Underlies Behavioral Modificationin Invertebrates 575BOX A Genetics of Learning and Memory in theFruit Fly 581Short-Term Synaptic Plasticity in the MammalianNervous System 582Long-Term Synaptic Plasticity in the MammalianNervous System 583Long-Term Potentiation of Hippocampal Synapses 584Molecular Mechanisms Underlying LTP 587BOX B Dendritic Spines 590Long-Term Synaptic Depression 592BOX C Silent Synapses 594Changes in Gene Expression Cause EnduringChanges in Synaptic Function during LTP andLTD 597Plasticity in the Adult Cerebral Cortex 599BOX D Epilepsy:The Effect of PathologicalActivity on Neural Circuitry 600Recovery from Neural Injury 602Generation of Neurons in the Adult Brain 605BOX E Why Aren’t We More Like Fish andFrogs? 606Summary 609Purves3/eFM 5/13/04 12:59 PM Page xiv
    • The Circadian Cycle of Sleep and Wakefulness 662Stages of Sleep 665BOX B Molecular Mechanisms of BiologicalClocks 666BOX C Electroencephalography 668Physiological Changes in Sleep States 671The Possible Functions of REM Sleep and Dreaming671Neural Circuits Governing Sleep 674BOX D Consciousness 675Thalamocortical Interactions 679Sleep Disorders 681BOX E Drugs and Sleep 682Summary 684Chapter 28 Emotions 687Overview 687Physiological Changes Associated with Emotion 687The Integration of Emotional Behavior 688BOX A Facial Expressions: Pyramidal andExtrapyramidal Contributions 690The Limbic System 693BOX B The Anatomy of the Amygdala 696The Importance of the Amygdala 697BOX C The Reasoning Behind an ImportantDiscovery 698The Relationship between Neocortex and Amygdala701BOX D Fear and the Human Amygdala:A Case Study 702BOX E Affective Disorders 704Cortical Lateralization of Emotional Functions 705Emotion, Reason, and Social Behavior 707Summary 708Chapter 29 Sex,Sexuality,and the Brain 711Overview 711Sexually Dimorphic Behavior 711What Is Sex? 712BOX A The Development of Male and FemalePhenotypes 714Hormonal Influences on Sexual Dimorphism 715BOX B The Case of Bruce/Brenda 716The Effect of Sex Hormones on Neural Circuitry 718BOX C The Actions of Sex Hormones 718Other Central Nervous System DimorphismsSpecifically Related to Reproductive Behaviors 720Brain Dimorphisms Related to Cognitive Function 728Hormone-Sensitive Brain Circuits in Adult Animals 729Summary 731Chapter 30 Memory 733Overview 733Qualitative Categories of Human Memory 733Temporal Categories of Memory 734BOX A Phylogenetic Memory 735The Importance of Association in Information Storage736Forgetting 738BOX B Savant Syndrome 739Brain Systems Underlying Declarative MemoryFormation 741BOX C Clinical Cases That Reveal the AnatomicalSubstrate for Declarative Memories 742Brain Systems Underlying Long-Term Storage ofDeclarative Memory 746Brain Systems Underlying Nondeclarative Learningand Memory 748Memory and Aging 749BOX D Alzheimer’s Disease 750Summary 753Appendix A The Brainstem and CranialNerves 755Appendix B Vascular Supply,the Meninges,and the Ventricular System 763The Blood Supply of the Brain and Spinal Cord 763The Blood-Brain Barrier 766BOX A Stroke 767The Meninges 768The Ventricular System 770GlossaryIllustration Source ReferencesIndexContents xvPurves3/eFM 5/13/04 12:59 PM Page xv
    • Whether judged in molecular, cellular, systemic, behavioral, or cogni-tive terms, the human nervous system is a stupendous piece of bio-logical machinery. Given its accomplishments—all the artifacts ofhuman culture, for instance—there is good reason for wanting tounderstand how the brain and the rest of the nervous system works.The debilitating and costly effects of neurological and psychiatric dis-ease add a further sense of urgency to this quest. The aim of this bookis to highlight the intellectual challenges and excitement—as well asthe uncertainties—of what many see as the last great frontier of bio-logical science. The information presented should serve as a startingpoint for undergraduates, medical students, graduate students in theneurosciences, and others who want to understand how the humannervous system operates. Like any other great challenge, neuro-science should be, and is, full of debate, dissension, and considerablefun. All these ingredients have gone into the construction of the thirdedition of this book; we hope they will be conveyed in equal measureto readers at all levels.PrefacePurves3/eFM 5/13/04 12:59 PM Page xvi
    • We are grateful to numerous colleagues who provided helpful contri-butions, criticisms and suggestions to this and previous editions. Weparticularly wish to thank Ralph Adolphs, David Amaral, Eva Anton,Gary Banker, Bob Barlow, Marlene Behrmann, Ursula Bellugi, DanBlazer, Bob Burke, Roberto Cabeza, Nell Cant, Jim Cavanaugh, JohnChapin, Milt Charlton, Michael Davis, Rob Deaner, Bob Desimone,Allison Doupe, Sasha du Lac, Jen Eilers, Anne Fausto-Sterling,Howard Fields, Elizabeth Finch, Nancy Forger, Jannon Fuchs,Michela Gallagher, Dana Garcia, Steve George, the late Patricia Gold-man-Rakic, Mike Haglund, Zach Hall, Kristen Harris, Bill Henson,John Heuser, Jonathan Horton, Ron Hoy, Alan Humphrey, Jon Kaas,Jagmeet Kanwal, Herb Killackey, Len Kitzes, Arthur Lander, StoryLandis, Simon LeVay, Darrell Lewis, Jeff Lichtman, Alan Light, SteveLisberger, Donald Lo, Arthur Loewy, Ron Mangun, Eve Marder,Robert McCarley, Greg McCarthy, Jim McIlwain, Chris Muly, VicNadler, Ron Oppenheim, Larysa Pevny, Michael Platt, FranckPolleux, Scott Pomeroy, Rodney Radtke, Louis Reichardt, Marnie Rid-dle, Jamie Roitman, Steve Roper, John Rubenstein, Ben Rubin, DavidRubin, Josh Sanes, Cliff Saper, Lynn Selemon, Carla Shatz, Bill Snider,Larry Squire, John Staddon, Peter Strick, Warren Strittmatter, JoeTakahashi, Richard Weinberg, Jonathan Weiner, Christina Williams,Joel Winston, and Fulton Wong. It is understood, of course, that anyerrors are in no way attributable to our critics and advisors.We also thank the students at Duke University Medical School aswell as many other students and colleagues who provided sugges-tions for improvement of the last edition. Finally, we owe specialthanks to Robert Reynolds and Nate O’Keefe, who labored long andhard to put the third edition together, and to Andy Sinauer, GraigDonini, Carol Wigg, Christopher Small, Janice Holabird, and the restof the staff at Sinauer Associates for their outstanding work and highstandards.AcknowledgmentsPurves3/eFM 5/13/04 12:59 PM Page xvii
    • For the StudentSylvius for Neuroscience:A Visual Glossary of Human Neuroanatomy (CD-ROM)S. Mark Williams, Leonard E. White, and Andrew C. MaceSylvius for Neuroscience: A Visual Glossary of Human Neuroanatomy,included in every copy of the textbook, is an interactive CD referenceguide to the structure of the human nervous system. By entering acorresponding page number from the textbook, students can quicklysearch the CD for any neuroanatomical structure or term and viewcorresponding images and animations. Descriptive information isprovided with all images and animations. Additionally, students cantake notes on the content and share these with other Sylvius users.Sylvius is an essential study aid for learning basic human neuro-anatomy.Sylvius for Neuroscience features:• Over 400 neuroanatomical structures and terms.• High-resolution images.• Animations of pathways and 3-D reconstructions.• Definitions and descriptions.• Audio pronunciations.• A searchable glossary.• Categories of anatomical structures and terms (e.g., cranialnerves, spinal cord tracts, lobes, cortical areas, etc.), that can beeasily browsed. In addition, structures can be browsed by text-book chapter.Supplements to Accompany NEUROSCIENCEThird EditionPurves3/eFM 5/13/04 1:00 PM Page xviii
    • • Images and text relevant to the textbook: Icons in the textbookindicate specific content on the CD. By entering a textbook pagenumber, students can automatically load the relevant imagesand text.• A history feature that allows the student to quickly reloadrecently viewed structures.• A bookmark feature that adds bookmarks to structures of in-terest; bookmarks are automatically stored on the student’scomputer.• A notes feature that allows students to type notes for anyselected structure; notes are automatically saved on the stu-dent’s computer and can be shared among students andinstructors (i.e., imported and exported).• A self-quiz mode that allows for testing on structure identifica-tion and functional information.• A print feature that formats images and text for printed output.• An image zoom tool.For the InstructorInstructor’s Resource CD (ISBN 0-87893-750-1)This expanded resource includes all the figures and tables from thetextbook in JPEG format, reformatted and relabeled for optimal read-ability. Also included are ready-to-use PowerPoint®presentations ofall figures and tables. In addition, new for the Third Edition, theInstructor’s Resource CD includes a set of short-answer study ques-tions for each chapter in Microsoft®Word®format.Overhead Transparencies (ISBN 0-87893-751-X)This set includes 100 illustrations (approximately 150 transparencies),selected from throughout the textbook for teaching purposes. Theseare relabeled and optimized for projection in classrooms.Supplements xixPurves3/eFM 5/13/04 1:00 PM Page xix
    • Purves3/eFM 5/13/04 1:00 PM Page xx
    • OverviewNeuroscience encompasses a broad range of questions about how nervoussystems are organized, and how they function to generate behavior. Thesequestions can be explored using the analytical tools of genetics, molecularand cell biology, systems anatomy and physiology, behavioral biology, andpsychology. The major challenge for a student of neuroscience is to integratethe diverse knowledge derived from these various levels of analysis into amore or less coherent understanding of brain structure and function (onehas to qualify this statement because so many questions remain unan-swered). Many of the issues that have been explored successfully concernhow the principal cells of any nervous system—neurons and glia—performtheir basic functions in anatomical, electrophysiological, and molecularterms. The varieties of neurons and supporting glial cells that have beenidentified are assembled into ensembles called neural circuits, and these cir-cuits are the primary components of neural systems that process specifictypes of information. Neural systems comprise neurons and circuits in anumber of discrete anatomical locations in the brain. These systems subserveone of three general functions. Sensory systems represent information aboutthe state of the organism and its environment, motor systems organize andgenerate actions; and associational systems link the sensory and motor sidesof the nervous system, providing the basis for “higher-order” functions suchas perception, attention, cognition, emotions, rational thinking, and othercomplex brain functions that lie at the core of understanding human beings,their history and their future.Genetics, Genomics, and the BrainThe recently completed sequencing of the genome in humans, mice, the fruitfly Drosophila melanogaster, and the nematode worm Caenorhabditis elegans isperhaps the logical starting point for studying the brain and the rest of thenervous system; after all, this inherited information is also the starting pointof each individual organism. The relative ease of obtaining, analyzing, andcorrelating gene sequences with neurobiological observations has facilitateda wealth of new insights into the basic biology of the nervous system. In par-allel with studies of normal nervous systems, the genetic analysis of humanpedigrees with various brain diseases has led to a widespread sense that itwill soon be possible to understand and treat disorders long consideredbeyond the reach of science and medicine.A gene consists of DNA sequences called exons that are transcribed into amessenger RNA and subsequently a protein. The set of exons that definesChapter 11Studying theNervous Systemsof Humans andOther AnimalsPurves01 5/13/04 1:02 PM Page 1
    • 2 Chapter OneFigure 1.1 Estimates of the number ofgenes in the human genome, as well asin the genomes of the mouse, the fruitfly Drosophila melanogaster, and thenematode worm Caenorhabditis elegans.the transcript of any gene is flanked by upstream (or 5′) and downstream (or3′) regulatory sequences that control gene expression. In addition, sequencesbetween exons—called introns—further influence transcription. Of theapproximately 35,000 genes in the human genome, a majority are expressedin the developing and adult brain; the same is true in mice, flies, andworms—the species commonly used in modern genetics (and increasingly inneuroscience) (Figure 1.1). Nevertheless, very few genes are uniquely ex-pressed in neurons, indicating that nerve cells share most of the basic struc-tural and functional properties of other cells. Accordingly, most “brain-specific” genetic information must reside in the remainder of nucleic acidsequences—regulatory sequences and introns—that control the timing,quantity, variability and cellular specificity of gene expression.One of the most promising dividends of sequencing the human genomehas been the realization that one or a few genes, when altered (mutated), canbegin to explain some aspects of neurological and psychiatric diseases.Before the “postgenomic era” (which began following completion of thesequencing of the human genome), many of the most devastating brain dis-eases remained largely mysterious because there was little sense of how orwhy the normal biology of the nervous system was compromised. The iden-tification of genes correlated with disorders such as Huntington’s disease,Parkinson’s disease, Alzheimer’s disease, major depression, and schizophre-nia has provided a promising start to understanding these pathologicalprocesses in a much deeper way (and thus devising rational therapies).Genetic and genomic information alone do not completely explain howthe brain normally works or how disease processes disrupt its function. Toachieve these goals it is equally essential to understand the cell biology,anatomy, and physiology of the brain in health as well as disease.The Cellular Components of the Nervous SystemEarly in the nineteenth century, the cell was recognized as the fundamentalunit of all living organisms. It was not until well into the twentieth century,however, that neuroscientists agreed that nervous tissue, like all otherorgans, is made up of these fundamental units. The major reason was thatthe first generation of “modern” neurobiologists in the nineteenth centuryhad difficulty resolving the unitary nature of nerve cells with the micro-scopes and cell staining techniques that were then available. This inade-Number of genes0 50,00040,00030,00020,00010,000HumanMouseD. melanogasterC. elegansPurves01 5/13/04 1:02 PM Page 2
    • quacy was exacerbated by the extraordinarily complex shapes and extensivebranches of individual nerve cells, which further obscured their resemblanceto the geometrically simpler cells of other tissues (Figures 1.2–1.4). As aresult, some biologists of that era concluded that each nerve cell was con-nected to its neighbors by protoplasmic links, forming a continuous nervecell network, or reticulum. The “reticular theory” of nerve cell communica-tion, which was championed by the Italian neuropathologist Camillo Golgi(for whom the Golgi apparatus in cells is named), eventually fell from favorand was replaced by what came to be known as the “neuron doctrine.” Themajor proponents of this new perspective were the Spanish neuroanatomistSantiago Ramón y Cajal and the British physiologist Charles Sherrington.The contrasting views represented by Golgi and Cajal occasioned a spir-ited debate in the early twentieth century that set the course of modern neu-roscience. Based on light microscopic examination of nervous tissue stainedwith silver salts according to a method pioneered by Golgi, Cajal arguedpersuasively that nerve cells are discrete entities, and that they communicateStudying the Nervous Systems of Humans and Other Animals 3AxonCellbodyDendritesDendrites(C) Retinal ganglion cell(F) Cerebellar Purkinje cellsAxonCellbody(A) Neurons in mesencephalicnucleus of cranial nerve VAxons**Cellbodies(B) Retinalbipolar cellDendritesDendritesCell bodyAxonCell bodyAxon Cell bodyDendrites(D) Retinal amacrine cell(E) Cortical pyramidal cell**Figure 1.2 Examples of the rich varietyof nerve cell morphologies found in thehuman nervous system. Tracings arefrom actual nerve cells stained byimpregnation with silver salts (the so-called Golgi technique, the method usedin the classical studies of Golgi andCajal). Asterisks indicate that the axonruns on much farther than shown. Notethat some cells, like the retinal bipolarcell, have a very short axon, and thatothers, like the retinal amacrine cell,have no axon at all. The drawings arenot all at the same scale.Purves01 5/13/04 1:02 PM Page 3
    • 4 Chapter Onewith one another by means of specialized contacts that Sherrington called“synapses.” The work that framed this debate was recognized by the awardof the Nobel Prize for Physiology or Medicine in 1906 to both Golgi andCajal ( the joint award suggests some ongoing concern about just who wascorrect, despite Cajal’s overwhelming evidence). The subsequent work ofSherrington and others demonstrating the transfer of electrical signals atsynaptic junctions between nerve cells provided strong support of the “neu-ron doctrine,” but challenges to the autonomy of individual neuronsremained. It was not until the advent of electron microscopy in the 1950sthat any lingering doubts about the discreteness of neurons were resolved.The high-magnification, high-resolution pictures that could be obtained withthe electron microscope clearly established that nerve cells are functionallyindependent units; such pictures also identified the specialized cellular junc-tions that Sherrington had named synapses (see Figures 1.3 and 1.4).The histological studies of Cajal, Golgi, and a host of successors led to thefurther consensus that the cells of the nervous system can be divided intotwo broad categories: nerve cells (or neurons), and supporting cells calledneuroglia (or simply glia; see Figure 1.5). Nerve cells are specialized for elec-trical signaling over long distances, and understanding this process repre-sents one of the more dramatic success stories in modern biology (and thesubject of Unit I of this book). Supporting cells, in contrast, are not capable ofelectrical signaling; nevertheless, they have several essential functions in thedeveloping and adult brain.NeuronsNeurons and glia share the complement of organelles found in all cells,including the endoplasmic reticulum and Golgi apparatus, mitochondria,and a variety of vesicular structures. In neurons, however, these organellesare often more prominent in distinct regions of the cell. In addition to thedistribution of organelles and subcellular components, neurons and glia arein some measure different from other cells in the specialized fibrillar ortubular proteins that constitute the cytoskeleton (Figures 1.3 and 1.4).Although many of these proteins—isoforms of actin, tubulin, and myosin, aswell as several others—are found in other cells, their distinctive organizationin neurons is critical for the stability and function of neuronal processes andsynaptic junctions. The filaments, tubules, vesicular motors, and scaffoldingproteins of neurons orchestrate the growth of axons and dendrites; the traf-ficking and appropiate positioning of membrane components, organelles,and vesicles; and the active processes of exocytosis and endocytosis thatunderlie synaptic communication. Understanding the ways in which thesemolecular components are used to insure the proper development and func-tion of neurons and glia remains a primary focus of modern neurobiology.The basic cellular organization of neurons resembles that of other cells;however, they are clearly distinguished by specialization for intercellularcommunication. This attribute is apparent in their overall morphology, in thespecific organization of their membrane components for electrical signaling,and in the structural and functional intricacies of the synaptic contactsbetween neurons (see Figures 1.3 and 1.4). The most obvious sign of neu-ronal specialization for communication via electrical signaling is the exten-sive branching of neurons. The most salient aspect of this branching for typ-ical nerve cells is the elaborate arborization of dendrites that arise from theneuronal cell body (also called dendritic branches or dendritic processes). Den-drites are the primary target for synaptic input from other neurons and arePurves01 5/13/04 1:02 PM Page 4
    • Studying the Nervous Systems of Humans and Other Animals 5MitochondrionEndoplasmicreticulumAxonsRibosomesGolgiapparatusNucleusDendriteSoma(A) (B) Axon (C) Synaptic endings (terminal boutons)(D) Myelinated axons(G) Myelinated axon and node of Ranvier(F) Neuronal cell body (soma)(E) DendritesFEBDGCFigure 1.3 The major light and electron microscopical features of neurons. (A) Dia-gram of nerve cells and their component parts. (B) Axon initial segment (blue)entering a myelin sheath (gold). (C) Terminal boutons (blue) loaded with synapticvesicles (arrowheads) forming synapses (arrows) with a dendrite (purple).(D) Transverse section of axons (blue) ensheathed by the processes of oligodendro-cytes (gold). (E) Apical dendrites (purple) of cortical pyramidal cells. (F) Nerve cellbodies (purple) occupied by large round nuclei. (G) Portion of a myelinated axon(blue) illustrating the intervals between adjacent segments of myelin (gold) referredto as nodes of Ranvier (arrows). (Micrographs from Peters et al., 1991.)Purves01 5/13/04 1:02 PM Page 5
    • 6 Chapter OneFigure 1.4 Distinctive arrangement ofcytoskeletal elements in neurons. (A)The cell body, axons, and dendrites aredistinguished by the distribution oftubulin (green throughout cell) versusother cytoskeletal elements—in thiscase, Tau (red), a microtubule-bindingprotein found only in axons. (B) Thestrikingly distinct localization of actin(red) to the growing tips of axonal anddendritic processes is shown here incultured neuron taken from the hip-pocampus. (C) In contrast, in a culturedepithelial cell, actin (red) is distributedin fibrils that occupy most of the cellbody. (D) In astroglial cells in culture,actin (red) is also seen in fibrillar bun-dles. (E) Tubulin (green) is seenthroughout the cell body and dendritesof neurons. (F) Although tubulin is amajor component of dendrites, extend-ing into spines, the head of the spine isenriched in actin (red). (G) The tubulincomponent of the cytoskeleton in non-neuronal cells is arrayed in filamentousnetworks. (H–K) Synapses have a dis-tinct arrangement of cytoskeletal ele-ments, receptors, and scaffold proteins.(H) Two axons (green; tubulin) frommotor neurons are seen issuing twobranches each to four muscle fibers. Thered shows the clustering of postsynapticreceptors (in this case for the neuro-transmitter acetylcholine). (I) A higherpower view of a single motor neuronsynapse shows the relationship betweenthe axon (green) and the postsynapticreceptors (red). (J) The extracellularspace between the axon and its targetmuscle is shown in green. (K) The clus-tering of scaffolding proteins (in thiscase, dystrophin) that localize receptorsand link them to other cytoskeletal ele-ments is shown in green. (A courtesy ofY. N. Jan; B courtesy of E. Dent and F.Gertler; C courtesy of D. Arneman andC. Otey; D courtesy of A. Gonzales andR. Cheney; E from Sheng, 2003; F fromMatus, 2000; G courtesy of T. Salmon etal.; H–K courtesy of R. Sealock.)(A) (B) (C)(D)(E) (G)(F)(H) (I)(J) (K)Purves01 5/13/04 1:02 PM Page 6
    • also distinguished by their high content of ribosomes as well as specificcytoskeletal proteins that reflect their function in receiving and integratinginformation from other neurons. The spectrum of neuronal geometriesranges from a small minority of cells that lack dendrites altogether to neu-rons with dendritic arborizations that rival the complexity of a mature tree(see Figure 1.2). The number of inputs that a particular neuron receivesdepends on the complexity of its dendritic arbor: nerve cells that lack den-drites are innervated by (thus, receive electrical signals from) just one or afew other nerve cells, whereas those with increasingly elaborate dendritesare innervated by a commensurately larger number of other neurons.The synaptic contacts made on dendrites (and, less frequently, on neu-ronal cell bodies) comprise a special elaboration of the secretory apparatusfound in most polarized epithelial cells. Typically, the presynaptic terminalis immediately adjacent to a postsynaptic specialization of the target cell(see Figure 1.3). For the majority of synapses, there is no physical continuitybetween these pre- and postsynaptic elements. Instead, pre- and postsynap-tic components communicate via secretion of molecules from the presynap-tic terminal that bind to receptors in the postsynaptic specialization. Thesemolecules must traverse an interval of extracellular space between pre- andpostsynaptic elements called the synaptic cleft. The synaptic cleft, however,is not simply a space to be traversed; rather, it is the site of extracellular pro-teins that influence the diffusion, binding, and degradation of moleculessecreted by the presynaptic terminal (see Figure 1.4). The number of synap-tic inputs received by each nerve cell in the human nervous system variesfrom 1 to about 100,000. This range reflects a fundamental purpose of nervecells, namely to integrate information from other neurons. The number ofsynaptic contacts from different presynaptic neurons onto any particular cellis therefore an especially important determinant of neuronal function.The information conveyed by synapses on the neuronal dendrites is inte-grated and “read out” at the origin of the axon, the portion of the nerve cellspecialized for signal conduction to the next site of synaptic interaction (seeFigures 1.2 and 1.3). The axon is a unique extension from the neuronal cellbody that may travel a few hundred micrometers (µm; usually calledmicrons) or much farther, depending on the type of neuron and the size ofthe species. Moreover, the axon also has a distinct cytoskeleton whose ele-ments are central for its functional integrity (see Figure 1.4). Many nervecells in the human brain (as well as that of other species) have axons nomore than a few millimeters long, and a few have no axons at all.Relatively short axons are a feature of local circuit neurons or interneu-rons throughout the brain. The axons of projection neurons, however, extendto distant targets. For example, the axons that run from the human spinalcord to the foot are about a meter long. The electrical event that carries sig-nals over such distances is called the action potential, which is a self-regen-erating wave of electrical activity that propagates from its point of initiationat the cell body (called the axon hillock) to the terminus of the axon wheresynaptic contacts are made. The target cells of neurons include other nervecells in the brain, spinal cord, and autonomic ganglia, and the cells of mus-cles and glands throughout the body.The chemical and electrical process by which the information encoded byaction potentials is passed on at synaptic contacts to the next cell in a path-way is called synaptic transmission. Presynaptic terminals (also called syn-aptic endings, axon terminals, or terminal boutons) and their postsynaptic spe-cializations are typically chemical synapses, the most abundant type ofStudying the Nervous Systems of Humans and Other Animals 7Purves01 5/13/04 1:02 PM Page 7
    • 8 Chapter Onesynapse in the nervous system. Another type, the electrical synapse, is farmore rare (see Chapter 5). The secretory organelles in the presynaptic termi-nal of chemical synapses are synaptic vesicles (see Figure 1.3), which aregenerally spherical structures filled with neurotransmitter molecules. Thepositioning of synaptic vesicles at the presynaptic membrane and theirfusion to initiate neurotransmitter release is regulated by a number of pro-teins either within or associated with the vesicle. The neurotransmittersreleased from synaptic vesicles modify the electrical properties of the targetcell by binding to neurotransmitter receptors (Figure 1.4), which are local-ized primarily at the postsynaptic specialization.The intricate and concerted activity of neurotransmitters, receptors,related cytoskeletal elements, and signal transduction molecules are thus thebasis for nerve cells communicating with one another, and with effector cellsin muscles and glands.Neuroglial CellsNeuroglial cells—also referred to as glial cells or simply glia—are quite dif-ferent from nerve cells. Glia are more numerous than neurons in the brain,outnumbering them by a ratio of perhaps 3 to 1. The major distinction is thatglia do not participate directly in synaptic interactions and electrical signal-ing, although their supportive functions help define synaptic contacts andmaintain the signaling abilities of neurons. Although glial cells also havecomplex processes extending from their cell bodies, these are generally lessprominent than neuronal branches, and do not serve the same purposes asaxons and dendrites (Figure 1.5).(B) Oligodendrocyte(A) AstrocyteCellbody Glialprocesses(D) (E) (F) (G)(C) Microglial cellFigure 1.5 Varieties of neuroglialcells. Tracings of an astrocyte (A), anoligodendrocyte (B), and a microglialcell (C) visualized using the Golgimethod. The images are at approxi-mately the same scale. (D) Astrocytes intissue culture, labeled (red) with anantibody against an astrocyte-specificprotein. (E) Oligodendroglial cells intissue culture labeled with an antibodyagainst an oligodendroglial-specificprotein. (F) Peripheral axon are en-sheathed by myelin (labeled red) exceptat a distinct region called the node ofRanvier. The green label indicates ionchannels concentrated in the node; theblue label indicates a molecularly dis-tinct region called the paranode. (G)Microglial cells from the spinal cord,labeled with a cell type-specific anti-body. Inset: Higher-magnificationimage of a single microglial cell labeledwith a macrophage-selective marker.(A–C after Jones and Cowan, 1983; D, Ecourtesy of A.-S. LaMantia; F courtesyof M. Bhat; G courtesy of A. Light; insetcourtesy of G. Matsushima.)Purves01 5/13/04 1:03 PM Page 8
    • The term glia (from the Greek word meaning “glue”) reflects the nine-teenth-century presumption that these cells held the nervous systemtogether in some way. The word has survived, despite the lack of any evi-dence that binding nerve cells together is among the many functions of glialcells. Glial roles that are well-established include maintaining the ionicmilieu of nerve cells, modulating the rate of nerve signal propagation, mod-ulating synaptic action by controlling the uptake of neurotransmitters at ornear the synaptic cleft, providing a scaffold for some aspects of neural devel-opment, and aiding in (or impeding, in some instances) recovery fromneural injury.There are three types of glial cells in the mature central nervous system:astrocytes, oligodendrocytes, and microglial cells (see Figure 1.5). Astro-cytes, which are restricted to the brain and spinal cord, have elaborate localprocesses that give these cells a starlike appearance (hence the prefix“astro”). A major function of astrocytes is to maintain, in a variety of ways,an appropriate chemical environment for neuronal signaling. Oligodendro-cytes, which are also restricted to the central nervous system, lay down alaminated, lipid-rich wrapping called myelin around some, but not all,axons. Myelin has important effects on the speed of the transmission of elec-trical signals (see Chapter 3). In the peripheral nervous system, the cells thatelaborate myelin are called Schwann cells.Finally, microglial cells are derived primarily from hematopoietic precur-sor cells (although some may be derived directly from neural precursorcells). They share many properties with macrophages found in other tissues,and are primarily scavenger cells that remove cellular debris from sites ofinjury or normal cell turnover. In addition, microglia, like their macrophagecounterparts, secrete signaling molecules—particularly a wide range ofcytokines that are also produced by cells of the immune system—that canmodulate local inflammation and influence cell survival or death. Indeed,some neurobiologists prefer to categorize microglia as a type of macrophage.Following brain damage, the number of microglia at the site of injuryincreases dramatically. Some of these cells proliferate from microglia residentin the brain, while others come from macrophages that migrate to the injuredarea and enter the brain via local disruptions in the cerebral vasculature.Cellular Diversity in the Nervous SystemAlthough the cellular constituents of the human nervous system are in manyways similar to those of other organs, they are unusual in their extraordi-nary numbers: the human brain is estimated to contain 100 billion neuronsand several times as many supporting cells. More importantly, the nervoussystem has a greater range of distinct cell types—whether categorized bymorphology, molecular identity, or physiological activity—than any otherorgan system (a fact that presumably explains why so many different genesare expressed in the nervous system; see above). The cellular diversity of anynervous system—including our own—undoubtedly underlies the the capac-ity of the system to form increasingly complicated networks to mediateincreasingly sophisticated behaviors.For much of the twentieth century, neuroscientists relied on the same setof techniques developed by Cajal and Golgi to describe and categorize thediversity of cell types in the nervous system. From the late 1970s onward,however, new technologies made possible by the advances in cell and mole-cular biology provided investigators with many additional tools to discernthe properties of neurons (Figure 1.6). Whereas general cell staining methodsStudying the Nervous Systems of Humans and Other Animals 9Purves01 5/13/04 1:03 PM Page 9
    • 10 Chapter Oneshowed mainly differences in cell size and distribution, antibody stains andprobes for messenger RNA added greatly to the appreciation of distinctivetypes of neurons and glia in various regions of the nervous system. At thesame time, new tract tracing methods using a wide variety of tracing sub-stances allowed the interconnections among specific groups of neurons to be(A) (B) (C) (D)(E) (F) (G) (H)(I) (J) (K) (L)(M) (N) (O) (P)Purves01 5/13/04 1:03 PM Page 10
    • explored much more fully. Tracers can be introduced into either living orfixed tissue, and are transported along nerve cell processes to reveal theirorigin and termination. More recently, genetic and neuroanatomical meth-ods have been combined to visualize the expression of fluorescent or othertracer molecules under the control of regulatory sequences of neural genes.This approach, which shows individual cells in fixed or living tissue inremarkable detail, allows nerve cells to be identified by both their transcrip-tional state and their structure. Finally, ways of determining the molecularidentity and morphology of nerve cells can be combined with measurementsof their physiological activity, thus illuminating structure–function relation-ships. Examples of these various approaches are shown in Figure 1.6.Neural CircuitsNeurons never function in isolation; they are organized into ensembles orneural circuits that process specific kinds of information and provide thefoundation of sensation, perception and behavior. The synaptic connectionsthat define such circuits are typically made in a dense tangle of dendrites,axons terminals, and glial cell processes that together constitute what iscalled neuropil (the suffix -pil comes from the Greek word pilos, meaning“felt”; see Figure 1.3). The neuropil is thus the region between nerve cellbodies where most synaptic connectivity occurs.Although the arrangement of neural circuits varies greatly according tothe function being served, some features are characteristic of all such ensem-bles. Preeminent is the direction of information flow in any particular circuit,which is obviously essential to understanding its purpose. Nerve cells thatStudying the Nervous Systems of Humans and Other Animals 11Figure 1.6 Structural diversity in the nervous system demonstrated with cellularand molecular markers. First row: Cellular organization of different brain regionsdemonstrated with Nissl stains, which label nerve and glial cell bodies. (A) Thecerebral cortex at the boundary between the primary and secondary visual areas. (B)The olfactory bulbs. (C) Differences in cell density in cerebral cortical layers. (D)Individual Nissl-stained neurons and glia at higher magnification. Second row: Clas-sical and modern approaches to seeing individual neurons and their processes. (E)Golgi-labeled cortical pyramidal cells. (F) Golgi-labeled cerebellar Purkinje cells. (G)Cortical interneuron labeled by intracellular injection of a fluorescent dye. (H) Reti-nal neurons labeled by intracellular injection of fluorescent dye. Third row: Cellularand molecular approaches to seeing neural connections and systems. (I) At top, anantibody that detects synaptic proteins in the olfactory bulb; at bottom, a fluorescentlabel shows the location of cell bodies. (J) Synaptic zones and the location of Purk-inje cell bodies in the cerebellar cortex labeled with synapse-specific antibodies(green) and a cell body marker (blue). (K) The projection from one eye to the lateralgeniculate nucleus in the thalamus, traced with radioactive amino acids (the brightlabel shows the axon terminals from the eye in distinct layers of the nucleus). (L)The map of the body surface of a rat in the somatic sensory cortex, shown with amarker that distinguishes zones of higher synapse density and metabolic activity.Fourth row: Peripheral neurons and their projections. (M) An autonomic neuronlabeled by intracellular injection of an enzyme marker. (N) Motor axons (green) andneuromuscular synapses (orange) in transgenic mice genetically engineered toexpress fluorescent proteins. (O) The projection of dorsal root ganglia to the spinalcord, demonstrated by an enzymatic tracer. (P) Axons of olfactory receptor neuronsfrom the nose labeled in the olfactory bulb with a vital fluorescent dye. (G courtesyof L. C. Katz; H courtesy of C. J. Shatz; N,O courtesy of W. Snider and J. Lichtman;all others courtesy of A.-S. LaMantia and D. Purves.)▲Purves01 5/13/04 1:03 PM Page 11
    • 12 Chapter Onecarry information toward the brain or spinal cord (or farther centrally withinthe spinal cord and brain) are called afferent neurons; nerve cells that carryinformation away from the brain or spinal cord (or away from the circuit inquestion) are called efferent neurons. Interneurons or local circuit neuronsonly participate in the local aspects of a circuit, based on the short distancesover which their axons extend. These three functional classes—afferent neu-rons, efferent neurons, and interneurons—are the basic constituents of allneural circuits.A simple example of a neural circuit is the ensemble of cells that subservesthe myotatic spinal reflex (the “knee-jerk” reflex; Figure 1.7). The afferentneurons of the reflex are sensory neurons whose cell bodies lie the dorsalroot ganglia and whose peripheral axons terminate in sensory endings inskeletal muscles (the ganglia that serve this same of function for much of thehead and neck are called cranial nerve ganglia; see Appendix A). The centralaxons of these afferent sensory neurons enter the the spinal cord where theyterminate on a variety of central neurons concerned with the regualtion ofmuscle tone, most obviously the motor neurons that determine the activity ofthe related muscles. These neurons constitute the efferent neurons as well asinterneurons of the circuit. One group of these efferent neurons in the ventralhorn of the spinal cord projects to the flexor muscles in the limb, and theother to extensor muscles. Spinal cord interneurons are the third element ofthis circuit. The interneurons receive synaptic contacts from sensory afferentneurons and make synapses on the efferent motor neurons that project to theSensory(afferent)axonInterneuronMotor(efferent)axonsMusclesensoryreceptorFlexormuscleExtensormuscle2C2B2A13A3B4Hammer tap stretchestendon, which, in turn,stretches sensoryreceptors in leg extensormuscleLegextends(C) Interneuron synapseinhibits motor neuronto flexor muscles(B) Sensory neuron alsoexcites spinal interneuron(A) Sensory neuron synapseswith and excites motorneuron in the spinal cord(B) Flexor muscle relaxesbecause the activity of itsmotor neurons has beeninhibited(A) Motor neuron conductsaction potential tosynapses on extensormuscle fibers, causingcontraction1 2 3 4Figure 1.7 A simple reflex circuit, theknee-jerk response (more formally, themyotatic reflex), illustrates severalpoints about the functional organizationof neural circuits. Stimulation of periph-eral sensors (a muscle stretch receptor inthis case) initiates receptor potentialsthat trigger action potentials that travelcentrally along the afferent axons of thesensory neurons. This information stim-ulates spinal motor neurons by meansof synaptic contacts. The action poten-tials triggered by the synaptic potentialin motor neurons travel peripherally inefferent axons, giving rise to muscle con-traction and a behavioral response. Oneof the purposes of this particular reflexis to help maintain an upright posture inthe face of unexpected changes.Purves01 5/13/04 1:03 PM Page 12
    • flexor muscles; therefore they are capable of modulating the input–outputlinkage. The excitatory synaptic connections between the sensory afferentsand the extensor efferent motor neurons cause the extensor muscles to con-tract; at the same time, the interneurons activated by the afferents areinhibitory, and their activation diminishes electrical activity in flexor efferentmotor neurons and causes the flexor muscles to become less active (Figure1.8). The result is a complementary activation and inactivation of the syner-gist and antagonist muscles that control the position of the leg.A more detailed picture of the events underlying the myotatic or any othercircuit can be obtained by electrophysiological recording (Figure 1.9). Thereare two basic approaches to measuring the electrical activity of a nerve cell:extracellular recording (also referred to as single-unit recording), where anelectrode is placed near the nerve cell of interest to detect its activity; andintracellular recording, where the electrode is placed inside the cell. Extracel-lular recordings primarily detect action potentials, the all-or-nothing changesin the potential across nerve cell membranes that convey information fromone point to another in the nervous system. This sort of recording is particu-larly useful for detecting temporal patterns of action potential activity andrelating those patterns to stimulation by other inputs, or to specific behavioralevents. Intracellular recordings can detect the smaller, graded potentialchanges that trigger action potentials, and thus allow a more detailed analy-sis of communication between neurons within a circuit. These graded trig-gering potentials can arise at either sensory receptors or synapses and arecalled receptor potentials or synaptic potentials, respectively.For the myotatic circuit, electrical activity can be measured both extracellu-larly and intracellularly, thus defining the functional relationships betweenneurons in the circuit. The pattern of action potential activity can be measuredfor each element of the circuit (afferents, efferents, and interneurons) before,during, and after a stimulus (see Figure 1.8). By comparing the onset, dura-tion, and frequency of action potential activity in each cell, a functional pictureof the circuit emerges. As a result of the stimulus, the sensory neuron is trig-gered to fire at higher frequency (i.e., more action potentials per unit time).This increase triggers a higher frequency of action potentials in both the exten-sor motor neurons and the interneurons. Concurrently, the inhibitory synapsesmade by the interneurons onto the flexor motor neurons cause the frequencyof action potentials in these cells to decline. Using intracellular recording, it ispossible to observe directly the potential changes underlying the synaptic con-nections of the myotatic reflex circuit (see Figure 1.9).Studying the Nervous Systems of Humans and Other Animals 13Sensory(afferent)axonInterneuronMotor(efferent)axonsMotor neuron(extensor)InterneuronSensory neuronHammertapLegextendsMotor neuron(flexor)Figure 1.8 Relative frequency of actionpotentials (indicated by individual verti-cal lines) in different components of themyotatic reflex as the reflex pathway isactivated. Notice the modulatory effectof the interneuron.Purves01 5/13/04 1:03 PM Page 13
    • 14 Chapter OneOverall Organization of the Human Nervous SystemWhen considered together, circuits that process similar types of informationcomprise neural systems that serve broader behavioral purposes. The mostgeneral functional distinction divides such collections into sensory systemsthat acquire and process information from the environment (e.g., the visualsystem or the auditory system, see Unit II), and motor systems that respondto such information by generating movements and other behavior (see UnitIII). There are, however, large numbers of cells and circuits that lie betweenthese relatively well-defined input and output systems. These are collec-tively referred to as associational systems, and they mediate the most com-plex and least well-characterized brain functions (see Unit V).In addition to these broad functional distinctions, neuroscientists andneurologists have conventionally divided the vertebrate nervous systemanatomically into central and peripheral components (Figure 1.10). The cen-tral nervous system, typically referred to as the CNS, comprises the brain(cerebral hemispheres, diencephalon, cerebellum, and brainstem) and thespinal cord (see Appendix A for more information about the gross anatomi-cal features of the CNS). The peripheral nervous system (PNS) includes thesensory neurons that link sensory receptors on the body surface or deeperwithin it with relevant processing circuits in the central nervous system. Themotor portion of the peripheral nervous system in turn consists of two com-ponents. The motor axons that connect the brain and spinal cord to skeletal(C) InterneuronInterneuronSensoryneuron(A) Sensory neuronMotor neuron(flexor)(D) Motor neuron (flexor)Motorneuron(extensor)(B) Motor neuron (extensor)Microelectrodeto measuremembrane potentialRecordRecordRecordRecordMembranepotential(mV)Membranepotential(mV)Membranepotential(mV)Membranepotential(mV)Time (ms)Activate excitatorysynapseActivate excitatorysynapseActivate inhibitorysynapseActionpotentialActionpotentialSynapticpotentialActionpotentialSynapticpotentialFigure 1.9 Intracellularly recordedresponses underlying the myotaticreflex. (A) Action potential measured ina sensory neuron. (B) Postsynaptic trig-gering potential recorded in an extensormotor neuron. (C) Postsynaptic trigger-ing potential in an interneuron. (D)Postsynaptic inhibitory potential in aflexor motor neuron. Such intracellularrecordings are the basis for understand-ing the cellular mechanisms of actionpotential generation, and the sensoryreceptor and synaptic potentials thattrigger these conducted signals.Purves01 5/13/04 1:03 PM Page 14
    • muscles make up the somatic motor division of the peripheral nervous sys-tem, whereas the cells and axons that innervate smooth muscles, cardiacmuscle, and glands make up the visceral or autonomic motor division.Those nerve cell bodies that reside in the peripheral nervous system arelocated in ganglia, which are simply local accumulations of nerve cell bodies(and supporting cells). Peripheral axons are gathered into bundles callednerves, many of which are enveloped by the glial cells of the peripheral ner-vous system called Schwann cells. In the central nervous system, nerve cellsare arranged in two different ways. Nuclei are local accumulations of neu-rons having roughly similar connections and functions; such collections arefound throughout the cerebrum, brainstem and spinal cord. In contrast, cor-tex (plural, cortices) describes sheet-like arrays of nerve cells (again, consultAppendix A for additional information and illustrations). The cortices of thecerebral hemispheres and of the cerebellum provide the clearest example ofthis organizational principle.Axons in the central nervous system are gathered into tracts that are moreor less analogous to nerves in the periphery. Tracts that cross the midline ofthe brain are referred to as commissures. Two gross histological terms dis-tinguish regions rich in neuronal cell bodies versus regions rich in axons.Gray matter refers to any accumulation of cell bodies and neuropil in thebrain and spinal cord (e.g., nuclei or cortices), whereas white matter, namedfor its relatively light appearance resulting from the lipid content of myelin,refers to axon tracts and commissures.Studying the Nervous Systems of Humans and Other Animals 15SENSORYCOMPONENTSCerebral hemispheres, diencephalon,cerebellum, brainstem, and spinal cord(analysis and integration ofsensory and motor information)(B)(A)MOTORCOMPONENTSINTERNALANDEXTERNALENVIRONMENTSensory gangliaand nerves (sympathetic,parasympathetic,and entericdivisions)VISCERALMOTORSYSTEMSOMATICMOTORSYSTEMSensory receptors(at surface andwithin the body)Autonomicgangliaand nervesMotor nervesSmooth muscles,cardiac muscles,and glandsSkeletal (striated)musclesEFFECTORSCentralnervoussystemPeripheralnervoussystemCentralnervous systemPeripheralnervous systemCranial nervesSpinal nervesBrainSpinal cordFigure 1.10 The major components ofthe nervous system and their functionalrelationships. (A) The CNS (brain andspinal cord) and PNS (spinal and cranialnerves). (B) Diagram of the major com-ponents of the central and peripheralnervous systems and their functionalrelationships. Stimuli from the environ-ment convey information to processingcircuits within the brain and spinal cord,which in turn interpret their significanceand send signals to peripheral effectorsthat move the body and adjust theworkings of its internal organs.Purves01 5/13/04 1:03 PM Page 15
    • 16 Chapter OneThe organization of the visceral motor division of the peripheral nervoussystem is a bit more complicated (see Chapter 20). Visceral motor neurons inthe brainstem and spinal cord, the so-called preganglionic neurons, formsynapses with peripheral motor neurons that lie in the autonomic ganglia.The motor neurons in autonomic ganglia innervate smooth muscle, glands,and cardiac muscle, thus controlling most involuntary (visceral) behavior. Inthe sympathetic division of the autonomic motor system, the ganglia liealong or in front of the vertebral column and send their axons to a variety ofperipheral targets. In the parasympathetic division, the ganglia are foundwithin the organs they innervate. Another component of the visceral motorsystem, called the enteric system, is made up of small ganglia as well asindividual neurons scattered throughout the wall of the gut. These neuronsinfluence gastric motility and secretion.Neuroanatomical TerminologyDescribing the organization of any neural system requires a rudimentaryunderstanding of anatomical terminology. The terms used to specify locationin the central nervous system are the same as those used for the grossanatomical description of the rest of the body (Figure 1.11). Thus, anteriorand posterior indicate front and back (head and tail); rostral and caudal,toward the head and tail; dorsal and ventral, top and bottom (back and belly);and medial and lateral, at the midline or to the side. Nevertheless, the com-parison between these coordinates in the body versus the brain can be con-fusing. For the entire body these anatomical terms refer to the long axis,which is straight. The long axis of the central nervous system, however, hasa bend in it. In humans and other bipeds, a compensatory tilting of the ros-tral–caudal axis for the brain is necessary to properly compare body axes tobrain axes. Once this adjustment has been made, the other axes for the braincan be easily assigned.The proper assignment of the anatomical axes then dictates the standardplanes for histological sections or live images (see Box A) used to study theinternal anatomy of the brain (see Figure 1.11B). Horizontal sections (alsoreferred to as axial or transverse sections) are taken parallel to the rostral–caudal axis of the brain; thus, in an individual standing upright, such sectionsare parallel to the ground. Sections taken in the plane dividing the two hemi-spheres are sagittal, and can be further categorized as midsagittal andparasagittal, according to whether the section is near the midline (midsagittal)Figure 1.11 A flexure in the long axis of the nervous system arose as humansevolved upright posture, leading to an approximately 120° angle between the longaxis of the brainstem and that of the forebrain The consequences of this flexure foranatomical terminology are indicated in (A). The terms anterior, posterior, superior,and inferior refer to the long axis of the body, which is straight. Therefore, these termsindicate the same direction for both the forebrain and the brainstem. In contrast, theterms dorsal, ventral, rostral, and caudal refer to the long axis of the central nervoussystem. The dorsal direction is toward the back for the brainstem and spinal cord,but toward the top of the head for the forebrain. The opposite direction is ventral.The rostral direction is toward the top of the head for the brainstem and spinal cord,but toward the face for the forebrain. The opposite direction is caudal. (B) The majorplanes of section used in cutting or imaging the brain. (C) The subdivisions and com-ponents of the central nervous system. (Note that the position of the brackets on theleft side of the figure refers to the vertebrae, not the spinal segments.)▲Purves01 5/13/04 1:03 PM Page 16
    • or more lateral (parasagittal). Sections in the plane of the face are called coro-nal or frontal. Different terms are usually used to refer to sections of thespinal cord. The plane of section orthogonal to the long axis of the cord iscalled transverse, whereas sections parallel to the long axis of the cord arecalled longitudinal. In a transverse section through the human spinal cord,the dorsal and ventral axes and the anterior and posterior axes indicate thesame directions (see Figure 1.11). Tedious though this terminology may be, itStudying the Nervous Systems of Humans and Other Animals 17(B)Posterior(behind)Superior(above)Anterior(in front of)Inferior(below)CaudalLongitudinalaxis of theforebrainLongitudinal axisof the brainstemand spinal cord(A)RostralCaudalHorizontalCoronal SagittalDorsalVentralDorsalVentralDorsalVentralDorsalVentralSpinal cordCervicalenlargementLumbarenlargementCaudaequinaC 12345678T 1T 1CervicalnervesThoracicnervesLumbarnerves(C)SacralnervesCoccygealnerveT 123456789101112L 12345S 1345Coc 12MedullaPonsMidbrainDiencephalonCerebrumCerebellumPurves01 5/13/04 1:03 PM Page 17
    • 18 Chapter Oneis essential for understanding the basic subdivisions of the nervous system(Figure 1.11C).The Subdivisions of the Central Nervous SystemThe central nervous system (defined as the brain and spinal cord) is usuallyconsidered to have seven basic parts: the spinal cord, the medulla, the pons,the cerebellum, the midbrain, the diencephalon, and the cerebral hemi-spheres (see Figures 1.10 and 1.11C). Running through all of these subdivi-sons are fluid-filled spaces called ventricles (a detailed account of the ven-tricular system can be found in Appendix B). These ventricles are theremnants of the continuous lumen initially enclosed by the neural plate as itrounded to become the neural tube during early development (see Chapter21). Variations in the shape and size of the mature ventricular space are char-acteristic of each adult brain region. The medulla, pons, and midbrain arecollectively called the brainstem and they surround the 4th ventricle(medulla and pons) and cerebral aqueduct (midbrain). The diencephalonand cerebral hemispheres are collectively called the forebrain, and theyenclose the 3rd and lateral ventricles, respectively. Within the brainstem arethe cranial nerve nuclei that either receive input from the cranial sensoryganglia mentioned earlier via the cranial sensory nerves, or give rise toaxons that constitute the cranial motor nerves (see Appendix A).The brainstem is also a conduit for several major tracts in the central ner-vous system that relay sensory information from the spinal cord and brain-stem to the forebrain, or relay motor commands from forebrain back tomotor neurons in the brainstem and spinal cord. Accordingly, detailedknowledge of the consequences of damage to the brainstem provides neu-rologists and other clinicians an essential tool in the localization and diagno-sis of brain injury. The brainstem contains numerous additional nuclei thatare involved in a myriad of important functions including the control ofheart rate, respiration, blood pressure, and level of consciousness. Finally,one of the most prominent features of the brainstem is the cerebellum,which extends over much of its dorsal aspect. The cerebellum is essential forthe coordination and planning of movements (see Chapter 18) as well aslearning motor tasks and storing that information (see Chapter 30).There are several anatomical subdivisions of the forebrain. The most obvi-ous anatomical structures are the prominent cerebral hemispheres (Figure1.12). In humans, the cerebral hemispheres (the outermost portions of whichare continuous, highly folded sheets of cortex) are proportionally larger thanin any other mammal, and are characterized by the gyri (singular, gyrus) orcrests of folded cortical tissue, and sulci (singular, sulcus) the grooves thatdivide gyri from one another (as pictured on the cover of this book, forexample). Although gyral and sulcal patterns vary from individual to indi-vidual, there are some fairly consistent landmarks that help divide the hemi-spheres into four lobes. The names of the lobes are derived from the cranialbones that overlie them: occipital, temporal, parietal, and frontal. A key fea-ture of the surface anatomy of the cerebrum is the central sulcus locatedFigure 1.12 Gross anatomy of the forebrain (A) Cerebral hemisphere surfaceanatomy, showing the four lobes of the brain and the major sulci and gyri. The ven-tricular system and basal ganglia can also be seen in this phantom view. (B) Mid-sagittal view showing the location of the hippocampus, amygdala, thalamus andhypothalamus.Purves01 5/13/04 1:03 PM Page 18
    • Studying the Nervous Systems of Humans and Other Animals 19Precentralgyrus(A)(B)(E)(D)(F)(C)PostcentralgyrusCentral sulcusParieto-occipitalsulcusPreoccipitalnotchLateral(Sylvian) fissureCerebralhemisphereCerebellumBrainstemSpinalcordCerebellumCingulategyrusParieto-occipital sulcusSpinal cordCingulatesulcusDiencephalonCorpuscallosumAnteriorcommissureBrainstemMidbrainPonsMedullaCalcarinesulcusCentralsulcusCorpuscallosumCaudatePutamenInternalcapsuleWhitematterOpticchiasmBasal forebrainnucleiAnteriorcommissureTemporallobeCerebral cortex(gray matter)AmygdalaCorpuscallosumLateralventricleFornixThirdventricleHippocampusMammillarybodyLateralventricle(temporalhorn)ThalamusCaudatePutamenGlobuspallidusTail ofcaudatenucleusBasal gangliaInternalcapsuleFrontallobeTemporallobeParietallobeOccipitallobeFrontallobeTemporallobeParietallobeOccipitallobeLevel of sectionshown in (E)Level of sectionshown in (F)Purves01 5/13/04 1:03 PM Page 19
    • 20 Chapter Oneroughly halfway between the rostral and caudal poles of the hemispheres(Figure 1.12A). This prominent sulcus divides the frontal lobe at the rostralend of the hemisphere from the more caudal parietal lobe. Prominent oneither side of the central sulcus are the pre- and postcentral gyri. These gyriare also functionally significant in that the precentral gyrus contains the pri-mary motor cortex important for the control of movement, and the postcen-tral gyrus contains the primary somatic sensory cortex which is importantfor the bodily senses (see below).The remaining subdivisions of the forebrain lie deeper in the cerebralhemispheres (Figure 1.12B). The most prominent of these is the collection ofdeep structures involved in motor and cognitive processes collectivelyreferred to as the basal ganglia. Other particularly important structures arethe hippocampus and amygdala in the temporal lobes (these are vital sub-strates for memory and emotional behavior, respectively), and the olfactorybulbs (the central stations for processing chemosensory information arisingfrom receptor neurons in the nasal cavity) on the anterior–inferior aspect ofthe frontal lobes. Finally, the thalamus lies in the diencephalon and is a crit-ical relay for sensory information (although it has many other functions aswell); the hypothalamus, which as the name implies lies below the thala-mus, is the central organizing structure for the regulation of the body’smany homeostatic functions (e.g., feeding, drinking, thermoregulation).This rudimentary description of some prominent anatomical landmarksprovides a framework for understanding how neurons resident in a numberof widely distributed and distinct brain structures communicate with oneanother to define neural systems dedicated to encoding, processing andrelaying specific sorts of information about aspects of the organism’s envi-ronment, and then initiating and coordinating appropriate behavioralresponses.Organizational Principles of Neural SystemsThese complex perceptual and motor capacities of the brain reflect the inte-grated function of various neural systems. The processing of somatic sensoryinformation (arising from receptors in the skin, subcutaneous tissues, andthe musculoskeletal system that respond to physical deformation at thebody surface or displacement of muscles and joints) provides a convenientexample. These widely distributed structures that participate in generatingsomatic sensations are referred to as the somatic sensory system (Figure1.13). The components in the peripheral nervous system include the recep-tors distributed throughout the skin as well as in muscles and tendons, therelated neurons in dorsal root ganglia, and neurons in some cranial ganglia.The central nervous system components include neurons in the spinal cord,as well as the long tracts of their axons that originate in the spinal cord,travel through the brainstem, and ultimately terminate in distinct relaynuclei in the thalamus in the diencephalon. The still-higher targets of thethalamic neurons are the cortical areas around the postcentral gyrus that arecollectively referred to as the somatic sensory cortex. Thus, the somatic sen-sory system includes specific populations of neurons in practically everysubdivision of the nervous system.Two further principles of neural system organization are evident in thesomatic sensory system: topographic organization and the prevalence ofparallel pathways (see Figure 1.13). As the name implies, topography refersto a mapping function—in this case a map of the body surface that can bediscerned within the various structures that constitute the somatic sensoryPurves01 5/13/04 1:03 PM Page 20
    • system. Thus, adjacent areas on the body surface are mapped toadjacent regions in nuclei, in white matter tracts, and in the thal-amic and cortical targets of the system. Beginning in the periph-ery, the cells in each dorsal root ganglion define a discrete der-matome (the area of the skin innervated by the processes of cellsfrom a single dorsal root). In the spinal cord, from caudal to ros-tral, the dermatomes are represented in corresponding regionsof the spinal cord from sacral (back) to lumbar (legs) to thoracic(chest) and cervical (arms and shoulders) (see Figures 1.13 and1.11C). This so-called somatotopy is maintained in the somaticsensory tracts in spinal cord and brainstem that convey infor-mation to the relevant forebrain structures of the somatic sen-sory system (Figure 1.14).Parallel pathways refer to the segregation of nerve cell axonsthat process the distinct stimulus attributes that comprise a par-ticular sensory, motor, or cognitive modality. For somatic sensa-tion, the stimulus attributes relayed via parallel pathways arepain, temperature, touch, pressure, and proprioception (the senseof joint or limb position). From the dorsal root ganglia, throughStudying the Nervous Systems of Humans and Other Animals 21Centralnervous systemPeripheralnervous systemSensoryreceptorsfor bodySensoryreceptorsfor faceSensoryreceptorThalamusCerebral cortexSomatic sensorycortex(A)(B)BrainstemSpinalcordThalamusCerebral cortexSomatic sensorycortexBrainstemSpinalcordDorsal rootgangliaDorsal rootganglia (DRG)Mechanical sensationTrigeminalgangliaTrigeminalganglionTrigeminalgangliaPain and temperatureMechanical sensationPain and temperatureCervicalThoracicLumbarSacralFigure 1.13 The anatomical and functional organi-zation of the somatic sensory system. Central ner-vous system components of the somatic sensory sys-tem are found in the spinal cord, brainstem,thalamus, and cerebral cortex. (A) Somatosensoryinformation from the body surface is mapped ontodorsal root ganglia (DRG), schematically depictedhere as attachments to the spinal cord. The variousshades of purple indicate correspondence betweenregions of the body and the DRG that relay informa-tion from the body surface to the central nervoussystem. Information from the head and neck isrelayed to the CNS via the trigeminal ganglia. (B)Somatosensory information travels from the peri-pheral sensory receptors via parallel pathways formechanical sensation and for the sensation of painand temperature. These parallel pathways relaythrough the spinal cord and brainstem, ultimatelysending sensory information to the thalamus, fromwhich it is relayed to the somatic sensory cortex inthe postcentral gyrus (indicated in blue in the imageof the whole brain; MRI courtesy of L. E. White, J.Vovoydic, and S. M. Williams).Purves01 5/13/04 1:03 PM Page 21
    • 22 Chapter Onethe spinal cord and brainstem, and on to the somatic sensory cortex, thesesubmodalities are kept largely segregated. Thus anatomically, biochemically,and physiologically distinct neurons transduce, encode, and relay pain, tem-perature, and mechanical information. Although this information is subse-quently integrated to provide unitary perception of the relevant stimuli, neu-rons and circuits in the somatic sensory system are clearly specialized toprocess discrete aspects of somatic sensation.This basic outline of the organization of the somatic system is representa-tive of the principles pertinent to understanding any neural system. It will inevery case be pertinent to consider the anatomical distribution of neural cir-cuits dedicated to a particular function, how the function is represented or“mapped” onto the neural elements within the system, and how distinctstimulus attributes are segregated within subsets of neurons that comprisethe system. Such details provide a framework for understanding how activ-ity within the system provides a representation of relevant stimulus, therequired motor response, and higher order cognitive correlates.Somatic sensorycortexShoulderNeckHeadNeckArmHandDigitsThumbEyesNoseFaceLipsJawTongueThroatToesGenitaliaFeetLegTrunkLateral MedialFigure 1.14 Somatotopic organization of sensory information. (Top) The locationsof primary and secondary somatosensory cortical areas on the lateral surface of thebrain. (Bottom) Cortical representation of different regions of skin.Purves01 5/13/04 1:03 PM Page 22
    • Functional Analysis of Neural SystemsA wide range of physiological methods is now available to evaluate the elec-trical (and metabolic) activity of the neuronal circuits that make up a neuralsystem. Two approaches, however, have been particularly useful in defininghow neural systems represent information. The most widely used method issingle-cell, or single-unit electrophysiological recording with microelec-trodes (see above; this method often records from several nearby cells inaddition to the one selected, providing further useful information). The useof microelectrodes to record action potential activity provides a cell-by-cellanalysis of the organization topographic maps (Figure 1.15), and can givespecific insight into the type of stimulus to which the neuron is “tuned” (i.e.,the stimulus that elicits a maximal change in action potential activity fromthe baseline state). Single-unit analysis is often used to define a neuron’sreceptive field—the region in sensory space (e.g., the body surface, or a spe-cialized structure such as the retina) within which a specific stimulus elicitsthe greatest action potential response. This approach to understandingneural systems was introduced by Stephen Kuffler and Vernon Mountcastlein the early 1950s and has now been used by several generations of neuro-scientists to evaluate the relationship between stimuli and neuronal re-sponses in both sensory and motor systems. Electrical recording techniquesStudying the Nervous Systems of Humans and Other Animals 23(A) (B)Somatic sensorycortexReceptive field(surround)Activity of cortical neuronPeriod ofstimulationCentralsulcusPostcentralgyrusRecordReceptivefield (center)Touch in the surroundof receptive fielddecreases cell firingTouch outside ofreceptive field hasno effectTouch in the centerof receptive fieldincreases cell firingFigure 1.15 Single-unit electrophysiological recording from cortical pyramidalneuron, showing the firing pattern in response to a specific peripheral stimulus.(A) Typical experimental set-up. (B) Defining neuronal receptive fields.Purves01 5/13/04 1:03 PM Page 23
    • 24 Chapter Oneat the single-cell level have now been extended and refined to include singleand simultaneous multiple cell analysis in animals performing complex cog-nitive tasks, intracellular recordings in intact animals, and the use of patchelectrodes to detect and monitor the activity of the individual membranemolecules that ultimately underlie neural signaling (see Unit I).The second major area in which remarkable technical advances have beenmade is functional brain imaging in human subjects (and to a lesser extentanimals), which has revolutionized the functional understanding of neuralsystems over the last two decades (Box A). Unlike electrical methods ofrecording neural activity, which are invasive in the sense of having to exposethe brain and insert electrodes into it, functional imaging is noninvasive andthus applicable to both patients and normal human subjects. Moreover, func-tional imaging allows the simultaneous evaluation of multiple brain struc-tures (which is possible but obviously difficult with electrical recordingmethods). The tasks that can be evaluated with functional imaging permit afar more ambitious and integrative approach to studying the operations of aneural system.Over the last 20 years, these noninvasive methods have allowed neurosci-entists to evaluate the representation of an enormous number of complexhuman behaviors, and at the same time have provided diagnostic tools thatare used more and more routinely. Many of the resulting observations haveconfirmed inferences about functional localization and the organization ofneural systems that were originally based on the study of neurologicalpatients who exhibited altered behavior after stroke or other forms of braininjury. Others findings, however, have given new insights into the wayneural systems function in the human brain.Analyzing Complex BehaviorMany of the most widely heralded advances in modern neuroscience haveinvolved reducing the complexity of the brain to more readily analyzedcomponents—i.e., genes, molecules, or cells. Nevertheless, the brain func-tions as a whole, and the study of more complex (and, some might argue,more interesting) brain functions such as perception, language, emotion,memory, and consciousness remain a central challenge for contemporaryneuroscientists. In recognition of this challenge, over the last 20 years or so afield called cognitive neuroscience has emerged that is specifically devotedto understanding these issues (see Unit V). This evolution has also rejuve-nated the field of neuroethology (which is devoted to observing complexbehaviors of animals in their native environments—for example, social com-munication in birds and non-human primates), and has encouraged thedevelopment of tasks to better evaluate the genesis of complex behaviors inhuman subjects. When used in combination with functional imaging, welldesigned behavioral tasks can facilitate identification of brain networksdevoted to specific complex functions, including language skills, mathemat-ical and musical ability, emotional responses, aesthetic judgments, andabstract thinking. Carefully constructed behavioral tasks can also be used tostudy the pathology of complex brain diseases that compromise cognition,such Alzheimer’s disease, schizophrenia, and depression.In short, new or revitalized efforts to study higher brain functions withincreasingly powerful techniques offer ways of beginning to understandeven the most complex aspects of human behavior.Purves01 5/13/04 1:03 PM Page 24
    • Studying the Nervous Systems of Humans and Other Animals 25Box ABrain Imaging TechniquesIn the 1970s, computerized tomography,or CT, opened a new era in noninvasiveimaging by introducing the use of com-puter processing technology to helpprobe the living brain. Prior to CT, theonly brain imaging technique availablewas standard X-ray film, which has poorsoft tissue contrast and involves rela-tively high radiation exposure.The CT approach uses a narrow X-raybeam and a row of very sensitive detec-tors placed on opposite sides of the headto probe just a small portion of tissue at atime with limited radiation exposure (seeFigure A). In order to make an image, theX-ray tube and detectors rotate aroundthe head to collect radiodensity informa-tion from every orientation around a nar-row slice. Computer processing tech-niques then calculate the radiodensity ofeach point within the slice plane, produc-ing a tomographic image (tomo means“cut” or “slice”). If the patient is slowlymoved through the scanner while the X-ray tube rotates in this way, a three-dimensional radiodensity matrix can becreated, allowing images to be computedfor any plane through the brain. CT scanscan readily distinguish gray matter andwhite matter, differentiate the ventriclesquite well, and show many other brainstructures with a spatial resolution of sev-eral millimeters.Brain imaging took another large stepforward in the 1980s with the develop-ment of magnetic resonance imaging(MRI). MRI is based on the fact that thenuclei of some atoms act as spinningmagnets, and that if they are placed in astrong magnetic field they will line upwith the field and spin at a frequencythat is dependent on the field strength. Ifthey then receive a brief radiofrequencypulse tuned to their spinning frequencythey are knocked out of alignment withthe field, and subsequently emit energyin an oscillatory fashion as they gradu-ally realign themselves with the field. Thestrength of the emitted signal depends onhow many nuclei are involved in thisprocess. To get spatial information inMRI, the magnetic field is distortedslightly by imposing magnetic gradientsalong three different spatial axes so thatonly nuclei at certain locations are tunedto the detector’s frequency at any giventime. Almost all MRI scanners use detec-tors tuned to the radio frequencies ofspinning hydrogen nuclei in water mole-cules, and thus create images based onthe distribution of water in different tis-sues. Careful manipulation of magneticfield gradients and radiofrequency pulsesmake it possible to construct extraordi-narily detailed images of the brain at anylocation and orientation with sub-mil-limeter resolution.The strong magnetic field and radio-frequency pulses used in MRI scanningare harmless, making this techniquecompletely noninvasive (although metalobjects in or near a scanner are a safetyconcern) (see Figure B). MRI is alsoextremely versatile because, by changingthe scanning parameters, images basedon a wide variety of different contrastmechanisms can be generated. For exam-ple, conventional MR images take advan-tage of the fact that hydrogen in differenttypes of tissue (e.g., gray matter, whitematter, cerebrospinal fluid) have slightlydifferent realignment rates, meaning thatsoft tissue contrast can be manipulatedsimply by adjusting when the realigninghydrogen signal is measured. Differentparameter settings can also be used togenerate images in which gray andwhite matter are invisible but in whichthe brain vasculature stands out in sharpdetail. Safety and versatility have madeMRI the technique of choice for imagingbrain structure in most applications.Imaging functional variations in theliving brain has also become possiblewith the recent development of tech-niques for detecting small, localizedX-raysourceX-raydetector(A) In computerized tomography, the X-ray source and detectors are moved around thepatient’s head. The inset shows a horizontal CT section of a normal adult brain. (continued)Purves01 5/13/04 1:03 PM Page 25
    • 26 Chapter OneSummaryThe brain can be studied by methods that range from genetics and molecu-lar biology to behavioral testing of normal human subjects. In addition to anever-increasing store of knowledge about the anatomical organization of thenervous system, many of the brightest successes of modern neurosciencehave come from understanding nerve cells as the basic structural and func-tional unit of the nervous system. Studies of the distinct cellular architectureand molecular components of neurons and glia have revealed much aboutBox A (continued)Brain Imaging Techniqueschanges in metabolism or cerebral bloodflow. To conserve energy, the brain regu-lates its blood flow such that active neu-rons with relatively high metabolicdemands receive more blood than rela-tively inactive neurons. Detecting andmapping these local changes in cerebralblood flow forms the basis for threewidely used functional brain imagingtechniques: positron emission tomogra-phy (PET), single-photon emissioncomputerized tomography (SPECT),and functional magnetic resonanceimaging (fMRI).In PET scanning, unstable positron-emitting isotopes are incorporated intodifferent reagents (including water, pre-cursor molecules of specific neurotrans-mitters, or glucose) and injected into thebloodstream. Labeled oxygen and glu-cose quickly accumulate in more meta-bolically active areas, and labeled trans-mitter probes are taken up selectively byappropriate regions. As the unstable iso-tope decays, it results in the emission oftwo positrons moving in opposite direc-tions. Gamma ray detectors placedaround the head register a “hit” onlywhen two detectors 180° apart reactsimultaneously. Images of tissue isotopedensity can then be generated (much theway CT images are calculated) showingthe location of active regions with a spa-tial resolution of about 4 mm. Dependingon the probe injected, PET imaging canbe used to visualize activity-dependentchanges in blood flow, tissue metabolism,or biochemical activity. SPECT imaging issimilar to PET in that it involves injectionor inhalation of a radiolabeled compound(for example, 133Xe or 123I-labelediodoamphetamine), which produce pho-tons that are detected by a gamma cam-era moving rapidly around the head.Functional MRI, a variant of MRI,currently offers the best approach forvisualizing brain function based on localmetabolism. fMRI is predicated on thefact that hemoglobin in blood slightlydistorts the magnetic resonance proper-ties of hydrogen nuclei in its vicinity, and(B) In MRI scanning, the head is placed in the center of a large magnet. A radiofrequencyantenna coil is placed around the head for exciting and recording the magnetic resonance sig-nal. For fMRI, stimuli can be presented using virtual reality video goggles and stereo head-phones while inside the scanner.Purves01 5/13/04 1:03 PM Page 26
    • their individual functions, as well as providing a basis for understandinghow nerve cells are organized into circuits, and circuits into systems thatprocess specific types of information pertinent to perception and action.Goals that remain include understanding how basic molecular genetic phe-nomena are linked to cellular, circuit, and system functions; understandinghow these processes go awry in neurological and psychiatric diseases; andbeginning to understand the especially complex functions of the brain thatmake us human.Studying the Nervous Systems of Humans and Other Animals 27the amount of magnetic distortionchanges depending on whether thehemoglobin has oxygen bound to it.When a brain area is activated by a spe-cific task it begins to use more oxygenand within seconds the brain microvas-culature responds by increasing the flowof oxygen-rich blood to the active area.These changes in the concentration ofoxygen and blood flow lead to localizedblood oxygenation level-dependent(BOLD) changes in the magnetic reso-nance signal. Such fluctuations aredetected using statistical image process-ing techniques to produce maps of task-dependent brain function (see Figure C).Because fMRI uses signals intrinsic to thebrain without any radioactivity, repeatedobservations can be made on the sameindividual—a major advantage overimaging methods such as PET. The spa-tial resolution (2–3 mm) and temporalresolution (a few seconds) of fMRI arealso superior to other functional imagingtechniques. MRI has thus emerged as thetechnology of choice for probing both thestructure and function of the livinghuman brain.ReferencesHUETTEL, S. A., A. W. SONG AND G. MCCARTHY(2004) Functional Magnetic Resonance Imaging.Sunderland, MA: Sinauer Associates.OLDENDORF, W. AND W. OLDENDORF JR. (1988)Basics of Magnetic Resonance Imaging. Boston:Kluwer Academic Publishers.RAICHLE, M. E. (1994) Images of the mind:Studies with modern imaging techniques.Ann. Rev. Psychol. 45: 333–356.SCHILD, H. (1990) MRI Made Easy (…Well,Almost). Berlin: H. Heineman.in progressRight Left Tumor(C) MRI images of an adult patient with a brain tumor, with fMRI activity during a handmotion task superimposed (left hand activity is shown in yellow, right hand activity in green).At right is a three-dimensional surface reconstructed view of the same data.Purves01 5/13/04 1:03 PM Page 27
    • 28 Chapter OneAdditional ReadingBRODAL, P. (1992) The Central Nervous System:Structure and Function. New York: Oxford Uni-versity Press.CARPENTER, M. B. AND J. SUTIN (1983) HumanNeuroanatomy, 8th Ed. Baltimore, MD: Wil-liams and Wilkins.ENGLAND, M. A. AND J. WAKELY (1991) ColorAtlas of the Brain and Spinal Cord: An Introduc-tion to Normal Neuroanatomy. St. Louis: MosbyYearbook.GIBSON, G. AND S. MUSE (2001) A Primer ofGenome Science. Sunderland, MA: SinauerAssociates.HAINES, D. E. (1995) Neuroanatomy: An Atlas ofStructures, Sections, and Systems, 2nd Ed. Balti-more: Urban and Schwarzenberg.MARTIN, J. H. (1996) Neuroanatomy: Text andAtlas, 2nd Ed. Stamford, CT: Appleton andLange.NATURE VOL. 409, NO. 6822 (2001) Issue ofFebruary 16. Special issue on the humangenome.NETTER, F. H. (1983) The CIBA Collection ofMedical Illustrations, Vols. I and II. A. Brassand R. V. Dingle (eds.). Summit, NJ: CIBAPharmaceutical Co.PETERS, A., S. L. PALAY AND H. DE F. WEBSTER(1991) The Fine Structure of the Nervous System:Neurons and Their Supporting Cells, 3rd Ed.New York: Oxford University Press.POSNER, M. I. AND M. E. RAICHLE (1997) Imagesof Mind, 2nd Ed. New York: W. H. Freeman &Co.RAMÓN Y CAJAL, S. (1984) The Neuron and theGlial Cell. (Transl. by J. de la Torre and W. C.Gibson.) Springfield, IL: Charles C. Thomas.RAMÓN Y CAJAL, S. (1990) New Ideas on theStructure of the Nervous System in Man and Ver-tebrates. (Transl. by N. Swanson and L. W.Swanson.) Cambridge, MA: MIT Press.SCIENCE VOL. 291, NO. 5507 (2001) Issue ofFebruary 16. Special issue on the humangenome.SHEPHERD, G. M. (1991) Foundations of the Neu-ron Doctrine. History of Neuroscience Series,No. 6. Oxford: Oxford University Press.WAXMAN, S. G. AND J. DEGROOT (1995) Correla-tive Neuroanatomy, 22nd Ed. Norwalk, CT:Appleton and Lange.Purves01 5/13/04 1:03 PM Page 28
    • Neural SignalingIPurves02 5/13/04 1:20 PM Page 29
    • Calcium signaling in a cerebel-lar Purkinje neuron. An elec-trode was used to fill the neu-ron with a fluorescent calciumindicator dye. This dyerevealed the release of intra-cellular calcium ions (color)produced by the actions of thesecond messenger IP3. (Cour-tesy of Elizabeth A. Finch andGeorge J. Augustine.)UNIT INEURAL SIGNALING2 Electrical Signals of Nerve Cells3 Voltage-Dependent Membrane Permeability4 Channels and Transporters5 Synaptic Transmission6 Neurotransmitters, Receptors, and Their Effects7 Molecular Signaling within NeuronsThe brain is remarkably adept at acquiring, coordinating, and dis-seminating information about the body and its environment. Suchinformation must be processed within milliseconds, yet it also can bestored away as memories that endure for years. Neurons within thecentral and peripheral nervous systems perform these functions bygenerating sophisticated electrical and chemical signals. This unitdescribes these signals and how they are produced. It explains howone type of electrical signal, the action potential, allows informationto travel along the length of a nerve cell. It also explains how othertypes of signals—both electrical and chemical—are generated at syn-aptic connections between nerve cells. Synapses permit informationtransfer by interconnecting neurons to form the circuitry on whichneural processing depends. Finally, it describes the intricate bio-chemical signaling events that take place within neurons. Appreciat-ing these fundamental forms of neuronal signaling provides a foun-dation for appreciating the higher-level functions considered in therest of the book.The cellular and molecular mechanisms that give neurons theirunique signaling abilities are also targets for disease processes thatcompromise the function of the nervous system. A working knowl-edge of the cellular and molecular biology of neurons is thereforefundamental to understanding a variety of brain pathologies, andfor developing novel approaches to diagnosing and treating these alltoo prevalent problems.Purves02 5/13/04 1:21 PM Page 30
    • OverviewNerve cells generate electrical signals that transmit information. Althoughneurons are not intrinsically good conductors of electricity, they haveevolved elaborate mechanisms for generating these signals based on theflow of ions across their plasma membranes. Ordinarily, neurons generate anegative potential, called the resting membrane potential, that can be mea-sured by recording the voltage between the inside and outside of nerve cells.The action potential transiently abolishes the negative resting potential andmakes the transmembrane potential positive. Action potentials are propa-gated along the length of axons and are the fundamental signal that carriesinformation from one place to another in the nervous system. Still othertypes of electrical signals are produced by the activation of synaptic contactsbetween neurons or by the actions of external forms of energy on sensoryneurons. All of these electrical signals arise from ion fluxes brought about bynerve cell membranes being selectively permeable to different ions, andfrom the non-uniform distribution of these ions across the membrane.Electrical Potentials across Nerve Cell MembranesNeurons employ several different types of electrical signal to encode andtransfer information. The best way to observe these signals is to use an intra-cellular microelectrode to measure the electrical potential across the neu-ronal plasma membrane. A typical microelectrode is a piece of glass tubingpulled to a very fine point (with an opening of less than 1 µm diameter) andfilled with a good electrical conductor, such as a concentrated salt solution.This conductive core can then be connected to a voltmeter, such as an oscil-loscope, to record the transmembrane voltage of the nerve cell.The first type of electrical phenomenon can be observed as soon as amicroelectrode is inserted through the membrane of the neuron. Upon enter-ing the cell, the microelectrode reports a negative potential, indicating thatneurons have a means of generating a constant voltage across their mem-branes when at rest. This voltage, called the resting membrane potential,depends on the type of neuron being examined, but it is always a fraction ofa volt (typically –40 to –90 mV).The electrical signals produced by neurons are caused by responses tostimuli, which then change the resting membrane potential. Receptor poten-tials are due to the activation of sensory neurons by external stimuli, such aslight, sound, or heat. For example, touching the skin activates Pacinian cor-puscles, receptor neurons that sense mechanical disturbances of the skin.These neurons respond to touch with a receptor potential that changes theresting potential for a fraction of a second (Figure 2.1A). These transientChapter 231Electrical Signalsof Nerve CellsPurves02 5/13/04 1:21 PM Page 31
    • 32 Chapter Twochanges in potential are the first step in generating the sensation of vibra-tions (or “tickles”) of the skin in the somatic sensory system (Chapter 8).Similar sorts of receptor potentials are observed in all other sensory neuronsduring transduction of sensory signals (Unit II).Another type of electrical signal is associated with communicationbetween neurons at synaptic contacts. Activation of these synapses generatessynaptic potentials, which allow transmission of information from one neu-ron to another. An example of such a signal is shown in Figure 2.1B. In thiscase, activation of a synaptic terminal innervating a hippocampal pyramidalneuron causes a very brief change in the resting membrane potential in thepyramidal neuron. Synaptic potentials serve as the means of exchanginginformation in complex neural circuits in both the central and peripheralnervous systems (Chapter 5).The use of electrical signals—as in sending electricity over wires to pro-vide power or information—presents a series of problems in electrical engi-neering. A fundamental problem for neurons is that their axons, which canbe quite long (remember that a spinal motor neuron can extend for a meteror more), are not good electrical conductors. Although neurons and wires–60−70–50−60Membranepotential(mV)Membranepotential(mV)40−60Membranepotential(mV)ActivatesynapseTouch skinActivatemotorneuronRecordStimulateRecordStimulate(A) Receptor potential(B) Synaptic potentialTime (ms)Time (ms)Time (ms)(C) Action potentialRecordFigure 2.1 Types of neuronal electricalsignals. In all cases, microelectrodes areused to measure changes in the restingmembrane potential during the indi-cated signals. (A) A brief touch causes areceptor potential in a Pacinian corpus-cle in the skin. (B) Activation of a synap-tic contact onto a hippocampal pyrami-dal neuron elicits a synaptic potential.(C) Stimulation of a spinal reflex pro-duces an action potential in a spinalmotor neuron.Purves02 5/13/04 1:21 PM Page 32
    • are both capable of passively conducting electricity, the electrical propertiesof neurons compare poorly to an ordinary wire. To compensate for this defi-ciency, neurons have evolved a “booster system” that allows them to con-duct electrical signals over great distances despite their intrinsically poorelectrical characteristics. The electrical signals produced by this booster sys-tem are called action potentials (which are also referred to as “spikes” or“impulses”). An example of an action potential recorded from the axon of aspinal motor neuron is shown in Figure 2.1C.One way to elicit an action potential is to pass electrical current across themembrane of the neuron. In normal circumstances, this current would begenerated by receptor potentials or by synaptic potentials. In the laboratory,however, electrical current suitable for initiating an action potential can bereadily produced by inserting a second microelectrode into the same neuronand then connecting the electrode to a battery (Figure 2.2A). If the currentdelivered in this way makes the membrane potential more negative (hyper-polarization), nothing very dramatic happens. The membrane potential sim-ply changes in proportion to the magnitude of the injected current (centralpart of Figure 2.2B). Such hyperpolarizing responses do not require anyunique property of neurons and are therefore called passive electricalresponses. A much more interesting phenomenon is seen if current of theopposite polarity is delivered, so that the membrane potential of the nervecell becomes more positive than the resting potential (depolarization). Inthis case, at a certain level of membrane potential, called the thresholdpotential, an action potential occurs (see right side of Figure 2.2B).The action potential, which is an active response generated by the neuron,is a brief (about 1 ms) change from negative to positive in the transmem-Electrical Signals of Nerve Cells 33Neuron(A)Microelectrodeto measuremembranepotentialMicroelectrodeto inject currentRecordStimulate−50TimeCurrent(nA)Membranepotential(mV)−65−1000+400−2+2ThresholdDepolarizationHyperpolarizationRestingpotentialAction potentialsInsertmicroelectrodePassive responses(B)Figure 2.2 Recording passive andactive electrical signals in a nerve cell.(A) Two microelectrodes are insertedinto a neuron; one of these measuresmembrane potential while the otherinjects current into the neuron. (B) In-serting the voltage-measuring micro-electrode into the neuron reveals a nega-tive potential, the resting membranepotential. Injecting current through thecurrent-passing microelectrode altersthe neuronal membrane potential.Hyperpolarizing current pulses produceonly passive changes in the membranepotential. While small depolarizing cur-rents also elict only passive responses,depolarizations that cause the mem-brane potential to meet or exceedthreshold additionally evoke actionpotentials. Action potentials are activeresponses in the sense that they are gen-erated by changes in the permeability ofthe neuronal membrane.Purves02 5/13/04 1:21 PM Page 33
    • 34 Chapter TwoFigure 2.3 Ion transporters and ionchannels are responsible for ionic move-ments across neuronal membranes.Transporters create ion concentrationdifferences by actively transporting ionsagainst their chemical gradients. Chan-nels take advantage of these concentra-tion gradients, allowing selected ions tomove, via diffusion, down their chemi-cal gradients.brane potential. Importantly, the amplitude of the action potential is inde-pendent of the magnitude of the current used to evoke it; that is, larger cur-rents do not elicit larger action potentials. The action potentials of a givenneuron are therefore said to be all-or-none, because they occur fully or not atall. If the amplitude or duration of the stimulus current is increased suffi-ciently, multiple action potentials occur, as can be seen in the responses tothe three different current intensities shown in Figure 2.2B (right side). It fol-lows, therefore, that the intensity of a stimulus is encoded in the frequencyof action potentials rather than in their amplitude. This arrangement differsdramatically from receptor potentials, whose amplitudes are graded in pro-portion to the magnitude of the sensory stimulus, or synaptic potentials,whose amplitude varies according to the number of synapses activated andthe previous amount of synaptic activity.Because electrical signals are the basis of information transfer in the ner-vous system, it is essential to understand how these signals arise. Remarkably,all of the neuronal electrical signals described above are produced by similarmechanisms that rely upon the movement of ions across the neuronal mem-brane. The remainder of this chapter addresses the question of how nerve cellsuse ions to generate electrical potentials. Chapter 3 explores more specificallythe means by which action potentials are produced and how these signalssolve the problem of long-distance electrical conduction within nerve cells.Chapter 4 examines the properties of membrane molecules responsible forelectrical signaling. Finally, Chapters 5–7 consider how electrical signals aretransmitted from one nerve cell to another at synaptic contacts.How Ionic Movements Produce Electrical SignalsElectrical potentials are generated across the membranes of neurons—and,indeed, all cells—because (1) there are differences in the concentrations of spe-cific ions across nerve cell membranes, and (2) the membranes are selectivelypermeable to some of these ions. These two facts depend in turn on two dif-ferent kinds of proteins in the cell membrane (Figure 2.3). The ion concentra-tion gradients are established by proteins known as active transporters,which, as their name suggests, actively move ions into or out of cells againsttheir concentration gradients. The selective permeability of membranes isION TRANSPORTERS ION CHANNELS1 IonbindsIonsIon transporters−Actively move ions againstconcentration gradient−Create ion concentrationgradients2 Ion transportedacross membraneInsideOutsideIon channels−Allow ions to diffuse downconcentration gradient−Cause selective permeabilityto certain ionsIon diffusesthrough channelNeuronalNeuronalmembranemembraneNeuronalmembranePurves02 5/13/04 1:21 PM Page 34
    • due largely to ion channels, proteins that allow only certain kinds of ions tocross the membrane in the direction of their concentration gradients. Thus,channels and transporters basically work against each other, and in so doingthey generate the resting membrane potential, action potentials, and the syn-aptic potentials and receptor potentials that trigger action potentials. Thestructure and function of these channels and transporters are described inChapter 4.To appreciate the role of ion gradients and selective permeability in gener-ating a membrane potential, consider a simple system in which an artificialmembrane separates two compartments containing solutions of ions. In sucha system, it is possible to determine the composition of the two solutions and,thereby, control the ion gradients across the membrane. For example, take thecase of a membrane that is permeable only to potassium ions (K+). If the con-centration of K+on each side of this membrane is equal, then no electricalpotential will be measured across it (Figure 2.4A). However, if the concentra-tion of K+is not the same on the two sides, then an electrical potential will begenerated. For instance, if the concentration of K+on one side of the mem-brane (compartment 1) is 10 times higher than the K+concentration on theother side (compartment 2), then the electrical potential of compartment 1will be negative relative to compartment 2 (Figure 2.4B). This difference inelectrical potential is generated because the potassium ions flow down theirconcentration gradient and take their electrical charge (one positive chargeper ion) with them as they go. Because neuronal membranes contain pumpsthat accumulate K+in the cell cytoplasm, and because potassium-permeablechannels in the plasma membrane allow a transmembrane flow of K+, ananalogous situation exists in living nerve cells. A continual resting efflux ofK+is therefore responsible for the resting membrane potential.In the hypothetical case just described, an equilibrium will quickly bereached. As K+moves from compartment 1 to compartment 2 (the initialconditions on the left of Figure 2.4B), a potential is generated that tends toimpede further flow of K+. This impediment results from the fact that theElectrical Signals of Nerve Cells 35VoltmeterV = 0InitiallyV = 0V1−2=−58 mV+++++–––––log[K+]2[K+]1 (mM)−116−580MembranepotentialV1−2(mV)−2 −1 0100 10 1[K+]111 mM KClNo net flux of K+Net flux of K+from 1 to 2Flux of K+from 1 to 2balanced byopposing membranepotentialInitial conditions At equilibrium21 mM KCl(A) (B) (C)110 mM KCl21 mM KCl110 mM KCl21 mM KClSlope = 58 mV pertenfold change inK+gradientPermeable to K+Figure 2.4 Electrochemical equilib-rium. (A) A membrane permeable onlyto K+(yellow spheres) separates com-partments 1 and 2, which contain theindicated concentrations of KCl. (B)Increasing the KCl concentration in com-partment 1 to 10 mM initially causes asmall movement of K+into compartment2 (initial conditions) until the electromo-tive force acting on K+balances theconcentration gradient, and the netmovement of K+becomes zero (at equi-librium). (C) The relationship betweenthe transmembrane concentration gradi-ent ([K+]2/[K+]1) and the membranepotential. As predicted by the Nernstequation, this relationship is linear whenplotted on semi-logarithmic coordinates,with a slope of 58 mV per tenfold differ-ence in the concentration gradient.Purves02 5/13/04 1:21 PM Page 35
    • 36 Chapter Twopotential gradient across the membrane tends to repel the positive potas-sium ions that would otherwise move across the membrane. Thus, as com-partment 2 becomes positive relative to compartment 1, the increasing posi-tivity makes compartment 2 less attractive to the positively charged K+. Thenet movement (or flux) of K+will stop at the point (at equilibrium on theright of Figure 2.4B) where the potential change across the membrane (therelative positivity of compartment 2) exactly offsets the concentration gradi-ent (the tenfold excess of K+in compartment 1). At this electrochemicalequilibrium, there is an exact balance between two opposing forces: (1) theconcentration gradient that causes K+to move from compartment 1 to com-partment 2, taking along positive charge, and (2) an opposing electrical gra-dient that increasingly tends to stop K+from moving across the membrane(Figure 2.4B). The number of ions that needs to flow to generate this electri-cal potential is very small (approximately 10–12moles of K+per cm2of mem-brane, or 1012K+ions). This last fact is significant in two ways. First, itmeans that the concentrations of permeant ions on each side of the mem-brane remain essentially constant, even after the flow of ions has generatedthe potential. Second, the tiny fluxes of ions required to establish the mem-brane potential do not disrupt chemical electroneutrality because each ionhas an oppositely charged counter-ion (chloride ions in the example shownin Figure 2.4) to maintain the neutrality of the solutions on each side of themembrane. The concentration of K+remains equal to the concentration ofCl–in the solutions in compartments 1 and 2, meaning that the separation ofcharge that creates the potential difference is restricted to the immediatevicinity of the membrane.The Forces That Create Membrane PotentialsThe electrical potential generated across the membrane at electrochemicalequilibrium, the equilibrium potential, can be predicted by a simple for-mula called the Nernst equation. This relationship is generally expressed aswhere EX is the equilibrium potential for any ion X, R is the gas constant, T isthe absolute temperature (in degrees on the Kelvin scale), z is the valence(electrical charge) of the permeant ion, and F is the Faraday constant (theamount of electrical charge contained in one mole of a univalent ion). Thebrackets indicate the concentrations of ion X on each side of the membraneand the symbol ln indicates the natural logarithm of the concentration gradi-ent. Because it is easier to perform calculations using base 10 logarithms andto perform experiments at room temperature, this relationship is usuallysimplified towhere log indicates the base 10 logarithm of the concentration ratio. Thus,for the example in Figure 2.4B, the potential across the membrane at electro-chemical equilibrium isThe equilibrium potential is conventionally defined in terms of the potentialdifference between the reference compartment, side 2 in Figure 2.4, and theother side. This approach is also applied to biological systems. In this case,EK2158 logKK58 log11058 mV=[ ][ ]= = −zEX2158 logXX=[ ][ ]zERTzFX21lnXX=[ ][ ]Purves02 5/13/04 1:21 PM Page 36
    • the outside of the cell is the conventional reference point (defined as zeropotential). Thus, when the concentration of K+is higher inside than out, aninside-negative potential is measured across the K+-permeable neuronalmembrane.For a simple hypothetical system with only one permeant ion species, theNernst equation allows the electrical potential across the membrane at equi-librium to be predicted exactly. For example, if the concentration of K+onside 1 is increased to 100 mM, the membrane potential will be –116 mV.More generally, if the membrane potential is plotted against the logarithm ofthe K+concentration gradient ([K]2/[K]1), the Nernst equation predicts a lin-ear relationship with a slope of 58 mV (actually 58/z) per tenfold change inthe K+gradient (Figure 2.4C).To reinforce and extend the concept of electrochemical equilibrium, con-sider some additional experiments on the influence of ionic species and ionicpermeability that could be performed on the simple model system in Figure2.4. What would happen to the electrical potential across the membrane (thepotential of side 1 relative to side 2) if the potassium on side 2 were replacedwith 10 mM sodium (Na+) and the K+in compartment 1 were replaced by 1mM Na+? No potential would be generated, because no Na+could flowacross the membrane (which was defined as being permeable only to K+).However, if under these ionic conditions (10 times more Na+in compartment2) the K+-permeable membrane were to be magically replaced by a mem-brane permeable only to Na+, a potential of +58 mV would be measured atequilibrium. If 10 mM calcium (Ca2+) were present in compartment 2 and 1mM Ca2+in compartment 1, and a Ca2+-selective membrane separated thetwo sides, what would happen to the membrane potential? A potential of+29 mV would develop, because the valence of calcium is +2. Finally, whatwould happen to the membrane potential if 10 mM Cl–were present in com-partment 1 and 1 mM Cl–were present in compartment 2, with the two sidesseparated by a Cl–-permeable membrane? Because the valence of this anionis –1, the potential would again be +58 mV.The balance of chemical and electrical forces at equilibrium means thatthe electrical potential can determine ionic fluxes across the membrane, justas the ionic gradient can determine the membrane potential. To examine theinfluence of membrane potential on ionic flux, imagine connecting a batteryacross the two sides of the membrane to control the electrical potential acrossthe membrane without changing the distribution of ions on the two sides(Figure 2.5). As long as the battery is off, things will be just as in Figure 2.4,with the flow of K+from compartment 1 to compartment 2 causing a nega-tive membrane potential (Figure 2.5A, left). However, if the battery is used tomake compartment 1 initially more negative relative to compartment 2, therewill be less K+flux, because the negative potential will tend to keep K+incompartment 1. How negative will side 1 need to be before there is no netflux of K+? The answer is –58 mV, the voltage needed to counter the tenfolddifference in K+concentrations on the two sides of the membrane (Figure2.5A, center). If compartment 1 is initially made more negative than –58 mV,then K+will actually flow from compartment 2 into compartment 1, becausethe positive ions will be attracted to the more negative potential of compart-ment 1 (Figure 2.5A, right). This example demonstrates that both the direc-tion and magnitude of ion flux depend on the membrane potential. Thus, insome circumstances the electrical potential can overcome an ionic concentra-tion gradient.The ability to alter ion flux experimentally by changing either the poten-tial imposed on the membrane (Figure 2.5B) or the transmembrane concen-Electrical Signals of Nerve Cells 37Purves02 5/13/04 1:21 PM Page 37
    • 38 Chapter Twotration gradient for an ion (see Figure 2.4C) provides convenient tools forstudying ion fluxes across the plasma membranes of neurons, as will be evi-dent in many of the experiments described in the following chapters.Electrochemical Equilibrium in an Environment withMore Than One Permeant IonNow consider a somewhat more complex situation in which Na+and K+areunequally distributed across the membrane, as in Figure 2.6A. What wouldhappen if 10 mM K+and 1 mM Na+were present in compartment 1, and 1mM K+and 10 mM Na+in compartment 2? If the membrane were perme-able only to K+, the membrane potential would be –58 mV; if the membranewere permeable only to Na+, the potential would be +58 mV. But whatwould the potential be if the membrane were permeable to both K+andNa+? In this case, the potential would depend on the relative permeability ofthe membrane to K+and Na+. If it were more permeable to K+, the potentialwould approach –58 mV, and if it were more permeable to Na+, the potentialwould be closer to +58 mV. Because there is no permeability term in theNernst equation, which only considers the simple case of a single permeantion species, a more elaborate equation is needed that takes into account boththe concentration gradients of the permeant ions and the relative permeabil-ity of the membrane to each permeant species.Such an equation was developed by David Goldman in 1943. For the casemost relevant to neurons, in which K+, Na+, and Cl–are the primary perme-ant ions, the Goldman equation is writtenwhere V is the voltage across the membrane (again, compartment 1 relativeto the reference compartment 2) and P indicates the permeability of theVP P PP P P=[ ] + [ ] + [ ][ ] + [ ] + [ ]58 logK Na ClK Na ClK 2 Na 2 Cl 1K 1 Na 1 Cl 2V1−2= −58 mVV1−2= 0 mV V1−2= −116 mV+++ ++++−−−−−−−Net flux of K+from 1 to 2No net fluxof K+Battery off Battery on Battery on(A)(B)110 mM KCl21 mM KCl110 mM KCl21 mM KCl Membrane potentialV1−2 (mV)NetfluxofK+No net fluxof K+−58 00−1161221Net flux of K+from 1 to 2Net flux of K+from 2 to 1110 mM KCl21 mM KClNet flux of K+from 2 to 1Battery Battery BatteryFigure 2.5 Membrane potential influ-ences ion fluxes. (A) Connecting a bat-tery across the K+-permeable membraneallows direct control of membranepotential. When the battery is turned off(left), K+ions (yellow) flow simplyaccording to their concentration gradi-ent. Setting the initial membrane poten-tial (V1–2) at the equilibrium potentialfor K+(center) yields no net flux of K+,while making the membrane potentialmore negative than the K+equilibriumpotential (right) causes K+to flowagainst its concentration gradient. (B)Relationship between membrane poten-tial and direction of K+flux.Purves02 5/13/04 1:21 PM Page 38
    • Figure 2.6 Resting and action poten-tials entail permeabilities to differentions. (A) Hypothetical situation inwhich a membrane variably permeableto Na+(red) and K+(yellow) separatestwo compartments that contain bothions. For simplicity, Cl–ions are notshown in the diagram. (B) Schematicrepresentation of the membrane ionicpermeabilities associated with restingand action potentials. At rest, neuronalmembranes are more permeable to K+(yellow) than to Na+(red); accordingly,the resting membrane potential is nega-tive and approaches the equilibriumpotential for K+, EK. During an actionpotential, the membrane becomes verypermeable to Na+(red); thus the mem-brane potential becomes positive andapproaches the equilibrium potential forNa+, ENa. The rise in Na+permeability istransient, however, so that the mem-brane again becomes primarily perme-able to K+(yellow), causing the poten-tial to return to its negative restingvalue. Notice that at the equilibriumpotential for a given ion, there is no netflux of that ion across the membrane.membrane to each ion of interest. The Goldman equation is thus anextended version of the Nernst equation that takes into account the relativepermeabilities of each of the ions involved. The relationship between the twoequations becomes obvious in the situation where the membrane is perme-able only to one ion, say, K+; in this case, the Goldman expression collapsesback to the simpler Nernst equation. In this context, it is important to notethat the valence factor (z) in the Nernst equation has been eliminated; this iswhy the concentrations of negatively charged chloride ions, Cl–, have beeninverted relative to the concentrations of the positively charged ions [remem-ber that –log (A/B) = log (B/A)].If the membrane in Figure 2.6A is permeable to K+and Na+only, theterms involving Cl–drop out because PCl is 0. In this case, solution of theGoldman equation yields a potential of –58 mV when only K+is permeant,+58 mV when only Na+is permeant, and some intermediate value if bothions are permeant. For example, if K+and Na+were equally permeant, thenthe potential would be 0 mV.With respect to neural signaling, it is particularly pertinent to ask whatwould happen if the membrane started out being permeable to K+, and thentemporarily switched to become most permeable to Na+. In this circum-stance, the membrane potential would start out at a negative level, becomepositive while the Na+permeability remained high, and then fall back to anegative level as the Na+permeability decreased again. As it turns out, thislast case essentially describes what goes on in a neuron during the genera-tion of an action potential. In the resting state, PK of the neuronal plasmamembrane is much higher than PNa; since, as a result of the action of iontransporters, there is always more K+inside the cell than outside (Table 2.1),the resting potential is negative (Figure 2.6B). As the membrane potential isdepolarized (by synaptic action, for example), PNa increases. The transientincrease in Na+permeability causes the membrane potential to become evenmore positive (red region in Figure 2.6B), because Na+rushes in (there ismuch more Na+outside a neuron than inside, again as a result of ionpumps). Because of this positive feedback loop, an action potential occurs.The rise in Na+permeability during the action potential is transient, how-ever; as the membrane permeability to K+is restored, the membrane poten-tial quickly returns to its resting level.Electrical Signals of Nerve Cells 39RestingpotentialRepolarizationActionpotential0MembranepotentialTimePNa>> PKPNa PNaPK>>PNa PK>>PNaEKENaVoltmeter10 mM KCl1 mM NaClVariable permeabilityto Na+and K+1 mM KCl10 mM NaCl(A) (B)Na+permeableK+permeable1 2Purves02 5/13/04 1:21 PM Page 39
    • 40 Chapter TwoArmed with an appreciation of these simple electrochemical principles, itwill be much easier to understand the following, more detailed account ofhow neurons generate resting and action potentials.The Ionic Basis of the Resting Membrane PotentialThe action of ion transporters creates substantial transmembrane gradientsfor most ions. Table 2.1 summarizes the ion concentrations measureddirectly in an exceptionally large nerve cell found in the nervous system ofthe squid (Box A). Such measurements are the basis for stating that there ismuch more K+inside the neuron than out, and much more Na+outside thanin. Similar concentration gradients occur in the neurons of most animals,including humans. However, because the ionic strength of mammalianblood is lower than that of sea-dwelling animals such as squid, in mammalsthe concentrations of each ion are several times lower. These transporter-dependent concentration gradients are, indirectly, the source of the restingneuronal membrane potential and the action potential.Once the ion concentration gradients across various neuronal membranesare known, the Nernst equation can be used to calculate the equilibriumpotential for K+and other major ions. Since the resting membrane potentialof the squid neuron is approximately –65 mV, K+is the ion that is closest tobeing in electrochemical equilibrium when the cell is at rest. This factimplies that the resting membrane is more permeable to K+than to the otherions listed in Table 2.1, and that this permeability is the source of restingpotentials.It is possible to test this guess, as Alan Hodgkin and Bernard Katz did in1949, by asking what happens to the resting membrane potential if the con-centration of K+outside the neuron is altered. If the resting membrane werepermeable only to K+, then the Goldman equation (or even the simplerNernst equation) predicts that the membrane potential will vary in propor-tion to the logarithm of the K+concentration gradient across the membrane.Assuming that the internal K+concentration is unchanged during the exper-iment, a plot of membrane potential against the logarithm of the external K+concentration should yield a straight line with a slope of 58 mV per tenfoldchange in external K+concentration at room temperature (see Figure 2.4C).(The slope becomes about 61 mV at mammalian body temperatures.)TABLE 2.1Extracellular and Intracellular Ion ConcentrationsConcentration (mM)Ion Intracellular ExtracellularSquid neuronPotassium (K+) 400 20Sodium (Na+) 50 440Chloride (Cl–) 40–150 560Calcium (Ca2+) 0.0001 10Mammalian neuronPotassium (K+) 140 5Sodium (Na+) 5–15 145Chloride (Cl–) 4–30 110Calcium (Ca2+) 0.0001 1–2Purves02 5/13/04 1:21 PM Page 40
    • Electrical Signals of Nerve Cells 41Box AThe Remarkable Giant Nerve Cells of SquidMany of the initial insights into how ionconcentration gradients and changes inmembrane permeability produce electri-cal signals came from experiments per-formed on the extraordinarily largenerve cells of the squid. The axons ofthese nerve cells can be up to 1 mm indiameter—100 to 1000 times larger thanmammalian axons. Thus, squid axonsare large enough to allow experimentsthat would be impossible on most othernerve cells. For example, it is not difficultto insert simple wire electrodes insidethese giant axons and make reliable elec-trical measurements. The relative ease ofthis approach yielded the first intracellu-lar recordings of action potentials fromnerve cells and, as discussed in the nextchapter, the first experimental measure-ments of the ion currents that produceaction potentials. It also is practical toextrude the cytoplasm from giant axonsand measure its ionic composition (seeTable 2.1). In addition, some giant nervecells form synaptic contacts with othergiant nerve cells, producing very largesynapses that have been extraordinarilyvaluable in understanding the funda-mental mechanisms of synaptic trans-mission (see Chapter 5).Giant neurons evidently evolved insquid because they enhanced survival.These neurons participate in a simpleneural circuit that activates the contrac-tion of the mantle muscle, producing ajet propulsion effect that allows the squidto move away from predators at aremarkably fast speed. As discussed inChapter 3, larger axonal diameter allowsfaster conduction of action potentials.Thus, presumably these huge nerve cellshelp squid escape more successfullyfrom their numerous enemies.Today—nearly 70 years after their dis-covery by John Z. Young at UniversityCollege London—the giant nerve cells ofsquid remain useful experimental sys-tems for probing basic neuronal functions.ReferencesLLINÁS, R. (1999) The Squid Synapse: A Modelfor Chemical Transmission. Oxford: OxfordUniversity Press.YOUNG, J. Z. (1939) Fused neurons and syn-aptic contacts in the giant nerve fibres ofcephalopods. Phil. Trans. R. Soc. Lond.229(B): 465–503.Brain1st-levelneuron2nd-levelneuronSquid giant axon = 800 µm diameterMammalian axon = 2 µm diameter3rd-levelneuron(A) (B) (C)StellateganglionPresynaptic(2nd level)Postsynaptic(3rd level)StellatenerveGiant axonSmaller axonsCrosssection1 mmStellatenerve withgiant axon1 mm(A) Diagram of a squid, showing the location of its giant nerve cells. Different colors indi-cate the neuronal components of the escape circuitry. The first- and second-level neuronsoriginate in the brain, while the third-level neurons are in the stellate ganglion and inner-vate muscle cells of the mantle. (B) Giant synapses within the stellate ganglion. The sec-ond-level neuron forms a series of fingerlike processes, each of which makes an extraordi-narily large synapse with a single third-level neuron. (C) Structure of a giant axon of athird-level neuron lying within its nerve. The enormous difference in the diameters of asquid giant axon and a mammalian axon are shown below.Purves02 5/13/04 1:21 PM Page 41
    • 42 Chapter TwoWhen Hodgkin and Katz carried out this experiment on a living squidneuron, they found that the resting membrane potential did indeed changewhen the external K+concentration was modified, becoming less negative asexternal K+concentration was raised (Figure 2.7A). When the external K+concentration was raised high enough to equal the concentration of K+inside the neuron, thus making the K+equilibrium potential 0 mV, the rest-ing membrane potential was also approximately 0 mV. In short, the restingmembrane potential varied as predicted with the logarithm of the K+con-centration, with a slope that approached 58 mV per tenfold change in K+concentration (Figure 2.7B). The value obtained was not exactly 58 mVbecause other ions, such as Cl–and Na+, are also slightly permeable, andthus influence the resting potential to a small degree. The contribution ofthese other ions is particularly evident at low external K+levels, again aspredicted by the Goldman equation. In general, however, manipulation ofthe external concentrations of these other ions has only a small effect,emphasizing that K+permeability is indeed the primary source of the restingmembrane potential.In summary, Hodgkin and Katz showed that the inside-negative restingpotential arises because (1) the membrane of the resting neuron is more per-meable to K+than to any of the other ions present, and (2) there is more K+inside the neuron than outside. The selective permeability to K+is caused byK+-permeable membrane channels that are open in resting neurons, and the(A)(B)0−20−60−40−800−20−60−40−80Restingmembranepotential(mV)Restingmembranepotential(mV)2 5 10 20 50 100 200 500[K+]out (mM)3.5 mMK+10 mMK+20 mMK+50 mMK+200 mMK+450 mMK+Time (min)1050Slope = 58 mV pertenfold change inK+gradientFigure 2.7 Experimental evidence that the resting membrane potential of a squidgiant axon is determined by the K+concentration gradient across the membrane.(A) Increasing the external K+concentration makes the resting membrane potentialmore positive. (B) Relationship between resting membrane potential and externalK+concentration, plotted on a semi-logarithmic scale. The straight line represents aslope of 58 mV per tenfold change in concentration, as given by the Nernst equa-tion. (After Hodgkin and Katz, 1949.)Purves02 5/13/04 1:21 PM Page 42
    • Figure 2.8 The role of sodium in thegeneration of an action potential in asquid giant axon. (A) An action poten-tial evoked with the normal ion concen-trations inside and outside the cell. (B)The amplitude and rate of rise of theaction potential diminish when externalsodium concentration is reduced to one-third of normal, but (C) recover whenthe Na+is replaced. (D) While theamplitude of the action potential isquite sensitive to the external concentra-tion of Na+, the resting membranepotential (E) is little affected by chang-ing the concentration of this ion. (AfterHodgkin and Katz, 1949.)large K+concentration gradient is, as noted, produced by membrane trans-porters that selectively accumulate K+within neurons. Many subsequentstudies have confirmed the general validity of these principles.The Ionic Basis of Action PotentialsWhat causes the membrane potential of a neuron to depolarize during anaction potential? Although a general answer to this question has been given(increased permeability to Na+), it is well worth examining some of theexperimental support for this concept. Given the data presented in Table 2.1,one can use the Nernst equation to calculate that the equilibrium potentialfor Na+(ENa) in neurons, and indeed in most cells, is positive. Thus, if themembrane were to become highly permeable to Na+, the membrane poten-tial would approach ENa. Based on these considerations, Hodgkin and Katzhypothesized that the action potential arises because the neuronal mem-brane becomes temporarily permeable to Na+.Taking advantage of the same style of ion substitution experiment theyused to assess the resting potential, Hodgkin and Katz tested the role of Na+in generating the action potential by asking what happens to the actionpotential when Na+is removed from the external medium. They found thatlowering the external Na+concentration reduces both the rate of rise of theaction potential and its peak amplitude (Figure 2.8A–C). Indeed, when theyexamined this Na+dependence quantitatively, they found a more-or-less lin-ear relationship between the amplitude of the action potential and the loga-rithm of the external Na+concentration (Figure 2.8D). The slope of this rela-Electrical Signals of Nerve Cells 43(D)(A)(B)(C)100+40Membranepotential(mV)Membranepotential(mV)Membranepotential(mV)−40−800+40Time (ms)−40−800+40−40−8000 1 2 3Time (ms)0 1 2 3Time (ms)0 1 2 3Actionpotentialamplitude(mV)Restingmembranepotential(mV)8040602050 100 200 500 10000−40−20−60−8050 100 200[Na+]out (mM)[Na+]out (mM)500 1000ControlLow [Na+]RecoverySlope = 58 mV pertenfold change inNa+gradient(E)Purves02 5/13/04 1:21 PM Page 43
    • 44 Chapter TwoBox BAction Potential Form and NomenclatureThe action potential of the squid giantaxon has a characteristic shape, or wave-form, with a number of different phases(Figure A). During the rising phase, themembrane potential rapidly depolarizes.In fact, action potentials cause the mem-brane potential to depolarize so muchthat the membrane potential transientlybecomes positive with respect to theexternal medium, producing an over-shoot. The overshoot of the action poten-tial gives way to a falling phase in whichthe membrane potential rapidly repolar-izes. Repolarization takes the membranepotential to levels even more negativethan the resting membrane potential fora short time; this brief period of hyper-polarization is called the undershoot.Although the waveform of the squidaction potential is typical, the details ofthe action potential form vary widelyfrom neuron to neuron in different ani-mals. In myelinated axons of vertebratemotor neurons (Figure B), the actionpotential is virtually indistinguishablefrom that of the squid axon. However,the action potential recorded in the cellbody of this same motor neuron (FigureC) looks rather different. Thus, the actionpotential waveform can vary even withinthe same neuron. More complex actionpotentials are seen in other central neu-rons. For example, action potentialsrecorded from the cell bodies of neuronsin the mammalian inferior olive (a regionof the brainstem involved in motor con-trol) last tens of milliseconds (Figure D).These action potentials exhibit a pro-nounced plateau during their fallingphase, and their undershoot lasts evenlonger than that of the motor neuron.One of the most dramatic types of actionpotentials occurs in the cell bodies ofcerebellar Purkinje neurons (Figure E).These potentials have several complexphases that result from the summation ofmultiple, discrete action potentials.The variety of action potential wave-forms could mean that each type of neu-ron has a different mechanism of actionpotential production. Fortunately, how-ever, these diverse waveforms all resultfrom relatively minor variations in thescheme used by the squid giant axon.For example, plateaus in the repolariza-tion phase result from the presence ofion channels that are permeable to Ca2+,and long-lasting undershoots result fromthe presence of additional types of mem-brane K+channels. The complex actionpotential of the Purkinje cell results fromthese extra features plus the fact that dif-ferent types of action potentials are gen-erated in various parts of the Purkinjeneuron—cell body, dendrites, andaxons—and are summed together inrecordings from the cell body. Thus, thelessons learned from the squid axon areapplicable to, and indeed essential for,understanding action potential genera-tion in all neurons.ReferencesBARRETT, E. F. AND J. N. BARRETT (1976) Sepa-ration of two voltage-sensitive potassiumcurrents, and demonstration of a tetro-dotoxin-resistant calcium current in frogmotoneurones. J. Physiol. (Lond.) 255:737–774.DODGE, F. A. AND B. FRANKENHAEUSER (1958)Membrane currents in isolated frog nervefibre under voltage clamp conditions. J.Physiol. (Lond.) 143: 76–90.HODGKIN, A. L. AND A. F. HUXLEY (1939)Action potentials recorded from inside anerve fibre. Nature 144: 710–711.LLINÁS, R. AND M. SUGIMORI (1980) Electro-physiological properties of in vitro Purkinjecell dendrites in mammalian cerebellar slices.J. Physiol. (Lond.) 305: 197–213.LLINÁS, R. AND Y. YAROM (1981) Electrophysi-ology of mammalian inferior olivary neu-rones in vitro. Different types of voltage-dependent ionic conductances. J. Physiol.(Lond.) 315: 549–567.(A) (B) (C) (D) (E)0−400 2 4 6 8 4 4 6 832 2 2010 30 4010 0 0 0 50 100 150+40Membranepotential(mV)Time (ms)Overshoot phaseFalling phaseUndershoot phaseRisingphase(A) The phases of an action potential of the squid giant axon. (B) Action potential recordedfrom a myelinated axon of a frog motor neuron. (C) Action potential recorded from the cellbody of a frog motor neuron. The action potential is smaller and the undershoot prolonged incomparison to the action potential recorded from the axon of this same neuron (B). (D) Actionpotential recorded from the cell body of a neuron from the inferior olive of a guinea pig. Thisaction potential has a pronounced plateau during its falling phase. (E) Action potentialrecorded from the cell body of a Purkinje neuron in the cerebellum of a guinea pig. (A afterHodgkin and Huxley, 1939; B after Dodge and Frankenhaeuser, 1958; C after Barrett and Bar-rett, 1976; D after Llinás and Yarom, 1981; E after Llinás and Sugimori, 1980.)Purves02 5/13/04 1:21 PM Page 44
    • tionship approached a value of 58 mV per tenfold change in Na+concentra-tion, as expected for a membrane selectively permeable to Na+. In contrast,lowering Na+concentration had very little effect on the resting membranepotential (Figure 2.8E). Thus, while the resting neuronal membrane is onlyslightly permeable to Na+, the membrane becomes extraordinarily perme-able to Na+during the rising phase and overshoot phase of the actionpotential (see Box B for an explanation of action potential nomenclature).This temporary increase in Na+permeability results from the opening ofNa+-selective channels that are essentially closed in the resting state. Mem-brane pumps maintain a large electrochemical gradient for Na+, which is inmuch higher concentration outside the neuron than inside. When the Na+channels open, Na+flows into the neuron, causing the membrane potentialto depolarize and approach ENa.The time that the membrane potential lingers near ENa (about +58 mV)during the overshoot phase of an action potential is brief because theincreased membrane permeability to Na+itself is short-lived. The membranepotential rapidly repolarizes to resting levels and is actually followed by atransient undershoot. As will be described in Chapter 3, these latter eventsin the action potential are due to an inactivation of the Na+permeability andan increase in the K+permeability of the membrane. During the undershoot,the membrane potential is transiently hyperpolarized because K+permeabil-ity becomes even greater than it is at rest. The action potential ends whenthis phase of enhanced K+permeability subsides, and the membrane poten-tial thus returns to its normal resting level.The ion substitution experiments carried out by Hodgkin and Katz pro-vided convincing evidence that the resting membrane potential results froma high resting membrane permeability to K+, and that depolarization duringan action potential results from a transient rise in membrane Na+permeabil-ity. Although these experiments identified the ions that flow during anaction potential, they did not establish how the neuronal membrane is able tochange its ionic permeability to generate the action potential, or what mech-anisms trigger this critical change. The next chapter addresses these issues,documenting the surprising conclusion that the neuronal membrane poten-tial itself affects membrane permeability.SummaryNerve cells generate electrical signals to convey information over substantialdistances and to transmit it to other cells by means of synaptic connections.These signals ultimately depend on changes in the resting electrical potentialacross the neuronal membrane. A resting potential occurs because nerve cellmembranes are permeable to one or more ion species subject to an electro-chemical gradient. More specifically, a negative membrane potential at restresults from a net efflux of K+across neuronal membranes that are predomi-nantly permeable to K+. In contrast, an action potential occurs when a tran-sient rise in Na+permeability allows a net flow of Na+in the opposite direc-tion across the membrane that is now predominantly permeable to Na+. Thebrief rise in membrane Na+permeability is followed by a secondary, tran-sient rise in membrane K+permeability that repolarizes the neuronal mem-brane and produces a brief undershoot of the action potential. As a result ofthese processes, the membrane is depolarized in an all-or-none fashion dur-ing an action potential. When these active permeability changes subside, themembrane potential returns to its resting level because of the high restingmembrane permeability to K+.Electrical Signals of Nerve Cells 45Purves02 5/13/04 1:21 PM Page 45
    • 46 Chapter TwoAdditional ReadingReviewsHODGKIN, A. L. (1951) The ionic basis of elec-trical activity in nerve and muscle. Biol. Rev.26: 339–409.HODGKIN, A. L. (1958) The Croonian Lecture:Ionic movements and electrical activity ingiant nerve fibres. Proc. R. Soc. Lond. (B) 148:1–37.Important Original PapersBAKER, P. F., A. L. HODGKIN AND T. I. SHAW(1962) Replacement of the axoplasm of giantnerve fibres with artificial solutions. J. Phys-iol. (London) 164: 330–354.COLE, K. S. AND H. J. CURTIS (1939) Electricimpedence of the squid giant axon duringactivity. J. Gen. Physiol. 22: 649–670.GOLDMAN, D. E. (1943) Potential, impedence,and rectification in membranes. J. Gen. Phys-iol. 27: 37–60.HODGKIN, A. L. AND P. HOROWICZ (1959) Theinfluence of potassium and chloride ions onthe membrane potential of single musclefibres. J. Physiol. (London) 148: 127–160.HODGKIN, A. L. AND B. KATZ (1949) The effectof sodium ions on the electrical activity of thegiant axon of the squid. J. Physiol. (London)108: 37–77.HODGKIN, A. L. AND R. D. KEYNES (1953) Themobility and diffusion coefficient of potas-sium in giant axons from Sepia. J. Physiol.(London) 119: 513–528.KEYNES, R. D. (1951) The ionic movementsduring nervous activity. J. Physiol. (London)114: 119–150.BooksHODGKIN, A. L. (1967) The Conduction of theNervous Impulse. Springfield, IL: Charles C.Thomas.HODGKIN, A. L. (1992) Chance and Design.Cambridge: Cambridge University Press.JUNGE, D. (1992) Nerve and Muscle Excitation,3rd Ed. Sunderland, MA: Sinauer Associates.KATZ, B. (1966) Nerve, Muscle, and Synapse.New York: McGraw-Hill.Purves02 5/13/04 1:21 PM Page 46
    • OverviewThe action potential, the primary electrical signal generated by nerve cells,reflects changes in membrane permeability to specific ions. Present under-standing of these changes in ionic permeability is based on evidenceobtained by the voltage clamp technique, which permits detailed characteri-zation of permeability changes as a function of membrane potential andtime. For most types of axons, these changes consist of a rapid and transientrise in sodium (Na+) permeability, followed by a slower but more prolongedrise in potassium (K+) permeability. Both permeabilities are voltage-depen-dent, increasing as the membrane potential depolarizes. The kinetics andvoltage dependence of Na+and K+permeabilities provide a complete expla-nation of action potential generation. Depolarizing the membrane potentialto the threshold level causes a rapid, self-sustaining increase in Na+perme-ability that produces the rising phase of the action potential; however, theNa+permeability increase is short-lived and is followed by a slower increasein K+permeability that restores the membrane potential to its usual negativeresting level. A mathematical model that describes the behavior of theseionic permeabilities predicts virtually all of the observed properties of actionpotentials. Importantly, this same ionic mechanism permits action potentialsto be propagated along the length of neuronal axons, explaining how electri-cal signals are conveyed throughout the nervous system.Ionic Currents Across Nerve Cell MembranesThe previous chapter introduced the idea that nerve cells generate electricalsignals by virtue of a membrane that is differentially permeable to variousion species. In particular, a transient increase in the permeability of the neu-ronal membrane to Na+initiates the action potential. This chapter considersexactly how this increase in Na+permeability occurs. A key to understand-ing this phenomenon is the observation that action potentials are initiatedonly when the neuronal membrane potential becomes more positive than athreshold level. This observation suggests that the mechanism responsiblefor the increase in Na+permeability is sensitive to the membrane potential.Therefore, if one could understand how a change in membrane potentialactivates Na+permeability, it should be possible to explain how actionpotentials are generated.The fact that the Na+permeability that generates the membrane potentialchange is itself sensitive to the membrane potential presents both conceptualand practical obstacles to studying the mechanism of the action potential. Apractical problem is the difficulty of systematically varying the membraneChapter 347Voltage-DependentMembranePermeabilityPurves03 5/13/04 1:29 PM Page 47
    • 48 Chapter Threepotential to study the permeability change, because such changes in mem-brane potential will produce an action potential, which causes further,uncontrolled changes in the membrane potential. Historically, then, it wasnot really possible to understand action potentials until a technique wasdeveloped that allowed experimenters to control membrane potential andsimultaneously measure the underlying permeability changes. This tech-Box AThe Voltage Clamp MethodBreakthroughs in scientific research oftenrely on the development of new tech-nologies. In the case of the action poten-tial, detailed understanding came onlyafter of the invention of the voltageclamp technique by Kenneth Cole in the1940s. This device is called a voltageclamp because it controls, or clamps,membrane potential (or voltage) at anylevel desired by the experimenter. Themethod measures the membrane poten-tial with a microelectrode (or other typeof electrode) placed inside the cell (1),and electronically compares this voltageto the voltage to be maintained (calledthe command voltage) (2). The clamp cir-cuitry then passes a current back into thecell though another intracellular elec-trode (3). This electronic feedback circuitholds the membrane potential at the de-sired level, even in the face of permeabil-ity changes that would normally alter themembrane potential (such as those gen-erated during the action potential). Mostimportantly, the device permits thesimultaneous measurement of the cur-rent needed to keep the cell at a givenvoltage (4). This current is exactly equalto the amount of current flowing acrossthe neuronal membrane, allowing directmeasurement of these membrane cur-rents. Therefore, the voltage clamp tech-nique can indicate how membranepotential influences ionic current flowacross the membrane. This informationgave Hodgkin and Huxley the keyinsights that led to their model for actionpotential generation.Today, the voltage clamp methodremains widely used to study ionic cur-rents in neurons and other cells. Themost popular contemporary version ofthis approach is the patch clamp tech-nique, a method that can be applied tovirtually any cell and has a resolutionhigh enough to measure the minute elec-trical currents flowing through single ionchannels (see Box A in Chapter 4).ReferencesCOLE, K. S. (1968) Membranes, Ions andImpulses: A Chapter of Classical Biophysics.Berkeley, CA: University of California Press.CommandvoltageCurrent-passingelectrodeVoltageclampamplifierMeasurecurrentMeasureVmRecordingelectrode4 The current flowing backinto the axon, and thusacross its membrane,can be measured hereSquidaxonSalinesolution2 Voltage clamp amplifiercompares membranepotential to the desired(command) potential3 When Vm is different from the commandpotential, the clamp amplifier injects currentinto the axon through a second electrode.This feedback arrangement causes themembrane potential to become the sameas the command potential1 One internal electrode measuresmembrane potential (Vm) and isconnected to the voltage clampamplifier+−ReferenceelectrodeVoltage clamp technique for studying mem-brane currents of a squid axon.Purves03 5/13/04 1:29 PM Page 48
    • nique, the voltage clamp method (Box A), provides the information neededto define the ionic permeability of the membrane at any level of membranepotential.In the late 1940s, Alan Hodgkin and Andrew Huxley working at the Uni-versity of Cambridge used the voltage clamp technique to work out the per-meability changes underlying the action potential. They again chose to usethe giant neuron of the squid because its large size (up to 1 mm in diameter;see Box A in Chapter 2) allowed insertion of the electrodes necessary forvoltage clamping. They were the first investigators to test directly thehypothesis that potential-sensitive Na+and K+permeability changes areboth necessary and sufficient for the production of action potentials.Hodgkin and Huxley’s first goal was to determine whether neuronalmembranes do, in fact, have voltage-dependent permeabilities. To addressthis issue, they asked whether ionic currents flow across the membranewhen its potential is changed. The result of one such experiment is shown inFigure 3.1. Figure 3.1A illustrates the currents produced by a squid axonwhen its membrane potential, Vm, is hyperpolarized from the resting level of–65 mV to –130 mV. The initial response of the axon results from the redistri-bution of charge across the axonal membrane. This capacitive current isnearly instantaneous, ending within a fraction of a millisecond. Aside fromthis brief event, very little current flows when the membrane is hyperpolar-ized. However, when the membrane potential is depolarized from –65 mV to0 mV, the response is quite different (Figure 3.1B). Following the capacitivecurrent, the axon produces a rapidly rising inward ionic current (inwardrefers to a positive charge entering the cell—that is, cations in or anions out),which gives way to a more slowly rising, delayed outward current. The factthat membrane depolarization elicits these ionic currents establishes that themembrane permeability of axons is indeed voltage-dependent.Two Types of Voltage-Dependent Ionic CurrentThe results shown in Figure 3.1 demonstrate that the ionic permeability ofneuronal membranes is voltage-sensitive, but the experiments do not iden-tify how many types of permeability exist, or which ions are involved. Asdiscussed in Chapter 2 (see Figure 2.5), varying the potential across a mem-brane makes it possible to deduce the equilibrium potential for the ionicfluxes through the membrane, and thus to identify the ions that are flowing.Voltage-Dependent Membrane Permeability 49(A) (B)−65−130 −130Membranecurrent(mA/cm2)Membranepotential(mV)00 1 2 3Time (ms)CapacitivecurrentTransient inward currentDelayedoutward current40 2Time (ms)4+10−1+1−10−6501 365 mV HyperpolarizationCapacitive current65 mV DepolarizationOutwardInwardOutwardInwardFigure 3.1 Current flow across a squidaxon membrane during a voltage clampexperiment. (A) A 65 mV hyperpolariza-tion of the membrane potential pro-duces only a very brief capacitive cur-rent. (B) A 65 mV depolarization of themembrane potential also produces abrief capacitive current, which is fol-lowed by a longer lasting but transientphase of inward current and a delayedbut sustained outward current. (AfterHodgkin et al., 1952.)Purves03 5/13/04 1:29 PM Page 49
    • 50 Chapter ThreeBecause the voltage clamp method allows the membrane potential to bechanged while ionic currents are being measured, it was a straightforwardmatter for Hodgkin and Huxley to determine ionic permeability by examin-ing how the properties of the early inward and late outward currentschanged as the membrane potential was varied (Figure 3.2). As alreadynoted, no appreciable ionic currents flow at membrane potentials more neg-ative than the resting potential. At more positive potentials, however, thecurrents not only flow but change in magnitude. The early current has a U-shaped dependence on membrane potential, increasing over a range ofdepolarizations up to approximately 0 mV but decreasing as the potential isdepolarized further. In contrast, the late current increases monotonicallywith increasingly positive membrane potentials. These different responses tomembrane potential can be seen more clearly when the magnitudes of thetwo current components are plotted as a function of membrane potential, asin Figure 3.3.The voltage sensitivity of the early inward current gives an importantclue about the nature of the ions carrying the current, namely, that no cur-rent flows when the membrane potential is clamped at +52 mV. For thesquid neurons studied by Hodgkin and Huxley, the external Na+concentra-tion is 440 mM, and the internal Na+concentration is 50 mM. For this con-centration gradient, the Nernst equation predicts that the equilibrium poten-Time (ms)0−22467Membranecurrent(mA/cm2)0 2 4 6 8 0 2 4 6 8 0 2 4 6 8 8 80 2 4 6 0 2 4 6−50−25−260255075Membranepotential(mV)+260+52+65Figure 3.2 Current produced by membrane depolarizations to several differentpotentials. The early current first increases, then decreases in magnitude as thedepolarization increases; note that this current is actually reversed in polarity atpotentials more positive than about +55 mV. In contrast, the late current increasesmonotonically with increasing depolarization. (After Hodgkin et al., 1952.)Membranecurrent(mA/cm2)Late001.02.03.050Membrane potential (mV)−50−100EarlyFigure 3.3 Relationship between current amplitude and membrane potential,taken from experiments such as the one shown in Figure 3.2. Whereas the late out-ward current increases steeply with increasing depolarization, the early inward cur-rent first increases in magnitude, but then decreases and reverses to outward cur-rent at about +55 mV (the sodium equilibrium potential). (After Hodgkin et al.,1952.)Purves03 5/13/04 1:29 PM Page 50
    • tial for Na+should be +55 mV. Recall further from Chapter 2 that at the Na+equilibrium potential there is no net flux of Na+across the membrane, evenif the membrane is highly permeable to Na+. Thus, the experimental obser-vation that no current flows at the membrane potential where Na+cannotflow is a strong indication that the early inward current is carried by entry ofNa+into the axon.An even more demanding way to test whether Na+carries the earlyinward current is to examine the behavior of this current after removingexternal Na+. Removing the Na+outside the axon makes ENa negative; if thepermeability to Na+is increased under these conditions, current should flowoutward as Na+leaves the neuron, due to the reversed electrochemical gra-dient. When Hodgkin and Huxley performed this experiment, they obtainedthe result shown in Figure 3.4. Removing external Na+caused the earlyinward current to reverse its polarity and become an outward current at amembrane potential that gave rise to an inward current when external Na+was present. This result demonstrates convincingly that the early inwardcurrent measured when Na+is present in the external medium must be dueto Na+entering the neuron.Notice that removal of external Na+in the experiment shown in Figure 3.4has little effect on the outward current that flows after the neuron has beenkept at a depolarized membrane voltage for several milliseconds. This fur-ther result shows that the late outward current must be due to the flow of anion other than Na+. Several lines of evidence presented by Hodgkin, Huxley,and others showed that this late outward current is caused by K+exiting theneuron. Perhaps the most compelling demonstration of K+involvement isthat the amount of K+efflux from the neuron, measured by loading the neu-ron with radioactive K+, is closely correlated with the magnitude of the lateoutward current.Taken together, these experiments using the voltage clamp show thatchanging the membrane potential to a level more positive than the restingpotential produces two effects: an early influx of Na+into the neuron, fol-lowed by a delayed efflux of K+. The early influx of Na+produces a transientinward current, whereas the delayed efflux of K+produces a sustained out-ward current. The differences in the time course and ionic selectivity of thetwo fluxes suggest that two different ionic permeability mechanisms are acti-vated by changes in membrane potential. Confirmation that there are indeedtwo distinct mechanisms has come from pharmacological studies of drugsthat specifically affect these two currents (Figure 3.5). Tetrodotoxin, an alka-loid neurotoxin found in certain puffer fish, tropical frogs, and salamanders,blocks the Na+current without affecting the K+current. Conversely, tetra-ethylammonium ions block K+currents without affecting Na+currents. Thedifferential sensitivity of Na+and K+currents to these drugs provides strongadditional evidence that Na+and K+flow through independent permeabilitypathways. As discussed in Chapter 4, it is now known that these pathwaysare ion channels that are selectively permeable to either Na+or K+. In fact,tetrodotoxin, tetraethylammonium, and other drugs that interact with spe-Voltage-Dependent Membrane Permeability 51Membranepotential(mV)Membranecurrent(mA/cm2)Early currentis inward00 2 4Time (ms)6 8−1+10−1+10−1+1250−25−50−75460 mM Na+460 mM Na+Early currentis outwardEarly currentis inward againNa+-freeFigure 3.4 Dependence of the early inward current on sodium. In the presence ofnormal external concentrations of Na+, depolarization of a squid axon to 0 mV pro-duces an inward initial current. However, removal of external Na+causes the initialinward current to become outward, an effect that is reversed by restoration of exter-nal Na+. (After Hodgkin and Huxley, 1952a.)Purves03 5/13/04 1:29 PM Page 51
    • 52 Chapter ThreeFigure 3.5 Pharmacological separationof Na+and K+currents into sodium andpotassium components. Panel (1) showsthe current that flows when the mem-brane potential of a squid axon is depo-larized to 0 mV in control conditions. (2)Treatment with tetrodotoxin causes theearly Na+currents to disappear butspares the late K+currents. (3) Additionof tetraethylammonium blocks the K+currents without affecting the Na+cur-rents. (After Moore et al., 1967 and Arm-strong and Binstock, 1965.)cific types of ion channels have been extraordinarily useful tools in charac-terizing these channel molecules (see Chapter 4).Two Voltage-Dependent Membrane ConductancesThe next goal Hodgkin and Huxley set for themselves was to describe Na+and K+permeability changes mathematically. To do this, they assumed thatthe ionic currents are due to a change in membrane conductance, defined asthe reciprocal of the membrane resistance. Membrane conductance is thusclosely related, although not identical, to membrane permeability. Whenevaluating ionic movements from an electrical standpoint, it is convenient todescribe them in terms of ionic conductances rather than ionic permeabili-ties. For present purposes, permeability and conductance can be consideredsynonymous. If membrane conductance (g) obeys Ohm’s Law (which statesthat voltage is equal to the product of current and resistance), then the ioniccurrent that flows during an increase in membrane conductance is given byIion = gion (Vm – Eion)where Iion is the ionic current, Vm is the membrane potential, and Eion is theequilibrium potential for the ion flowing through the conductance, gion. Thedifference between Vm and Eion is the electrochemical driving force acting onthe ion.Hodgkin and Huxley used this simple relationship to calculate the depen-dence of Na+and K+conductances on time and membrane potential. Theyknew Vm, which was set by their voltage clamp device (Figure 3.6A), andcould determine ENa and EK from the ionic concentrations on the two sides0−75−50−25250−1+1(1)(2)(3)0−1+1Membranecurrent(mA/cm2)005Time (ms)Time (ms)100Time (ms)5 10 5 10Membranepotential(mV)Add tetrodotoxinAdd tetraethyl-ammoniumNa+ currentblockedK+currentblockedPurves03 5/13/04 1:29 PM Page 52
    • of the axonal membrane (see Table 2.1). The currents carried by Na+andK+—INa and IK—could be determined separately from recordings of themembrane currents resulting from depolarization (Figure 3.6B) by measur-ing the difference between currents recorded in the presence and absence ofexternal Na+(as shown in Figure 3.4). From these measurements, Hodgkinand Huxley were able to calculate gNa and gK (Figure 3.6C,D), from whichthey drew two fundamental conclusions. The first conclusion is that the Na+and K+conductances change over time. For example, both Na+and K+con-ductances require some time to activate, or turn on. In particular, the K+con-ductance has a pronounced delay, requiring several milliseconds to reach itsmaximum (Figure 3.6D), whereas the Na+conductance reaches its maximummore rapidly (Figure 3.6C). The more rapid activation of the Na+conduc-tance allows the resulting inward Na+current to precede the delayed out-ward K+current (see Figure 3.6B). Although the Na+conductance risesrapidly, it quickly declines, even though the membrane potential is kept at adepolarized level. This fact shows that depolarization not only causes theNa+conductance to activate, but also causes it to decrease over time, or inac-tivate. The K+conductance of the squid axon does not inactivate in this way;thus, while the Na+and K+conductances share the property of time-depen-dent activation, only the Na+conductance inactivates. (Inactivating K+conductances have since been discovered in other types of nerve cells; seeChapter 4.) The time courses of the Na+and K+conductances are voltage-Voltage-Dependent Membrane Permeability 53Membranepotential(mV)02−246MembranecurrentmA/cm230201006040200Na+conductancemSiemens/cm2K+conductancemSiemens/cm244−2−2723Time (ms)0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6 0 2 4 6(A)(B)(C)(D)8 8 88 8−50−75−2502550−39++Figure 3.6 Membrane conductancechanges underlying the action potentialare time- and voltage-dependent. Depo-larizations to various membrane poten-tials (A) elicit different membrane cur-rents (B). Below are shown the Na+(C)and K+(D) conductances calculatedfrom these currents. Both peak Na+con-ductance and steady-state K+conduc-tance increase as the membrane poten-tial becomes more positive. In addition,the activation of both conductances, aswell as the rate of inactivation of theNa+conductance, occur more rapidlywith larger depolarizations. (AfterHodgkin and Huxley, 1952b.)Purves03 5/13/04 1:29 PM Page 53
    • 54 Chapter ThreeFigure 3.7 Depolarization increasesNa+and K+conductances of the squidgiant axon. The peak magnitude of Na+conductance and steady-state value ofK+conductance both increase steeply asthe membrane potential is depolarized.(After Hodgkin and Huxley, 1952b.)dependent, with the speed of both activation and inactivation increasing atmore depolarized potentials. This finding accounts for more rapid timecourses of membrane currents measured at more depolarized potentials.The second conclusion derived from Hodgkin and Huxley’s calculationsis that both the Na+and K+conductances are voltage-dependent—that is,both conductances increase progressively as the neuron is depolarized. Fig-ure 3.7 illustrates this by plotting the relationship between peak value of theconductances (from Figure 3.6C,D) against the membrane potential. Notethe similar voltage dependence for each conductance; both conductances arequite small at negative potentials, maximal at very positive potentials, andexquisitely dependent on membrane voltage at intermediate potentials. Theobservation that these conductances are sensitive to changes in membranepotential shows that the mechanism underlying the conductances somehow“senses” the voltage across the membrane.All told, the voltage clamp experiments carried out by Hodgkin and Hux-ley showed that the ionic currents that flow when the neuronal membrane isdepolarized are due to three different voltage-sensitive processes: (1) activa-tion of Na+conductance, (2) activation of K+conductance, and (3) inactiva-tion of Na+conductance.Reconstruction of the Action PotentialFrom their experimental measurements, Hodgkin and Huxley were able toconstruct a detailed mathematical model of the Na+and K+conductancechanges. The goal of these modeling efforts was to determine whether theNa+and K+conductances alone are sufficient to produce an action potential.Using this information, they could in fact generate the form and time courseof the action potential with remarkable accuracy (Figure 3.8A). Further, theHodgkin-Huxley model predicted other features of action potential behaviorin the squid axon, such as how the delay before action potential generationchanges in response to stimulating currents of different intensities (Figure3.8B,C). The model also predicted that the axon membrane would becomerefractory to further excitation for a brief period following an action poten-tial, as was experimentally observed.The Hodgkin-Huxley model also provided many insights into how actionpotentials are generated. Figure 3.8A shows a reconstructed action potential,together with the time courses of the underlying Na+and K+conductances.The coincidence of the initial increase in Na+conductance with the rapid ris-ing phase of the action potential demonstrates that a selective increase inMembrane potential (mV)05101520Conductance(mSiemens/cm2)−60−80 −40 −20 0 20 40Na+Membrane potential (mV)010203040Conductance(mSiemens/cm2)−60−80 −40 −20 0 20 40K+Purves03 5/13/04 1:29 PM Page 54
    • Na+conductance is responsible for action potential initiation. The increase inNa+conductance causes Na+to enter the neuron, thus depolarizing themembrane potential, which approaches ENa. The rate of depolarization sub-sequently falls both because the electrochemical driving force on Na+decreases and because the Na+conductance inactivates. At the same time,depolarization slowly activates the voltage-dependent K+conductance, caus-ing K+to leave the cell and repolarizing the membrane potential toward EK.Because the K+conductance becomes temporarily higher than it is in theresting condition, the membrane potential actually becomes briefly morenegative than the normal resting potential (the undershoot). The hyperpo-larization of the membrane potential causes the voltage-dependent K+con-ductance (and any Na+conductance not inactivated) to turn off, allowing themembrane potential to return to its resting level.Voltage-Dependent Membrane Permeability 55K+Na+(B)Time (ms)010203005075250+20−20−60−40−80+40(A)Membranepotential(mV)Membranepotential(mV)ConductancemSiemens/cm2ACTION POTENTIALS OF SQUID AXON0 1 2 3 40 1 2 3 40 1 2 3 40 1 2 3 4Time (ms)0−25−50−755025StimuluscurrentMembranepotential(mV)2 3 4(C)Time (ms)0−25−50−755025MATHEMATICAL MODEL BASED ON Na+AND K+CONDUCTANCES0 1 0 0 12 3 412 3 4Figure 3.8 Mathematical reconstruction of the action potential. (A) Reconstructionof an action potential (black curve) together with the underlying changes in Na+(red curve) and K+(yellow curve) conductance. The size and time course of theaction potential were calculated using only the properties of gNa and gK measured involtage clamp experiments. Real action potentials evoked by brief current pulses ofdifferent intensities (B) are remarkably similar to those generated by the mathemati-cal model (C). The reconstructed action potentials shown in (A) and (C) differ induration because (A) simulates an action potential at 19°C, whereas (C) simulatesan action potential at 6°C. (After Hodgkin and Huxley, 1952d.)Purves03 5/13/04 1:29 PM Page 55
    • 56 Chapter ThreeThis mechanism of action potential generation represents a positive feed-back loop: Activating the voltage-dependent Na+conductance increases Na+entry into the neuron, which makes the membrane potential depolarize,which leads to the activation of still more Na+conductance, more Na+entry,and still further depolarization (Figure 3.9). Positive feedback continuesunabated until Na+conductance inactivation and K+conductance activationrestore the membrane potential to the resting level. Because this positivefeedback loop, once initiated, is sustained by the intrinsic properties of theneuron—namely, the voltage dependence of the ionic conductances—theaction potential is self-supporting, or regenerative. This regenerative qualityexplains why action potentials exhibit all-or-none behavior (see Figure 2.1),and why they have a threshold (Box B). The delayed activation of the K+con-ductance represents a negative feedback loop that eventually restores themembrane to its resting state.Hodgkin and Huxley’s reconstruction of the action potential and all itsfeatures shows that the properties of the voltage-sensitive Na+and K+con-ductances, together with the electrochemical driving forces created by iontransporters, are sufficient to explain action potentials. Their use of bothempirical and theoretical methods brought an unprecedented level of rigorto a long-standing problem, setting a standard of proof that is achieved onlyrarely in biological research.Long-Distance Signaling by Means of Action PotentialsThe voltage-dependent mechanisms of action potential generation alsoexplain the long-distance transmission of these electrical signals. Recall fromChapter 2 that neurons are relatively poor conductors of electricity, at leastcompared to a wire. Current conduction by wires, and by neurons in theabsence of action potentials, is called passive current flow (Box C). The pas-sive electrical properties of a nerve cell axon can be determined by measur-ing the voltage change resulting from a current pulse passed across theaxonal membrane (Figure 3.10A). If this current pulse is not large enough togenerate action potentials, the magnitude of the resulting potential changedecays exponentially with increasing distance from the site of current injec-tion (Figure 3.10B). Typically, the potential falls to a small fraction of its ini-tial value at a distance of no more than a couple of millimeters away fromthe site of injection (Figure 3.10C). The progressive decrease in the amplitudeof the induced potential change occurs because the injected current leaks outacross the axonal membrane; accordingly, less current is available to changethe membrane potential farther along the axon. Thus, the leakiness of theaxonal membrane prevents effective passive transmission of electrical signalsin all but the shortest axons (those 1 mm or less in length). Likewise, theleakiness of the membrane slows the time course of the responses measuredat increasing distances from the site where current was injected (Figure3.10D).FAST POSITIVECYCLEIncreaseNa+ currentOpen Na+channelsSLOW NEGATIVECYCLEIncreaseK+ currentOpen K+channelsDepolarizemembranepotentialHyperpolarizesDepolarizesmoreFigure 3.9 Feedback cycles responsible for membrane potential changes duringan action potential. Membrane depolarization rapidly activates a positive feedbackcycle fueled by the voltage-dependent activation of Na+conductance. This phe-nomenon is followed by the slower activation of a negative feedback loop as depo-larization activates a K+conductance, which helps to repolarize the membranepotential and terminate the action potential.Purves03 5/13/04 1:29 PM Page 56
    • Voltage-Dependent Membrane Permeability 57Box BThresholdAn important—and potentially puz-zling—property of the action potential isits initiation at a particular membranepotential, called threshold. Indeed,action potentials never occur without adepolarizing stimulus that brings themembrane to this level. The depolarizing“trigger” can be one of several events: asynaptic input, a receptor potential gen-erated by specialized receptor organs,the endogenous pacemaker activity ofcells that generate action potentials spon-taneously, or the local current that medi-ates the spread of the action potentialdown the axon.Why the action potential “takes off”at a particular level of depolarization canbe understood by comparing the under-lying events to a chemical explosion(Figure A). Exogenous heat (analogousto the initial depolarization of the mem-brane potential) stimulates an exother-mic chemical reaction, which producesmore heat, which further enhances thereaction (Figure B). As a result of thispositive feedback loop, the rate of thereaction builds up exponentially—thedefinition of an explosion. In any suchprocess, however, there is a threshold,that is, a point up to which heat can besupplied without resulting in an explo-sion. The threshold for the chemicalexplosion diagrammed here is the pointat which the amount of heat suppliedexogenously is just equal to the amountof heat that can be dissipated by the cir-cumstances of the reaction (such asescape of heat from the beaker).The threshold of action potential initi-ation is, in principle, similar (Figure C).There is a range of “subthreshold” depo-larization, within which the rate ofincreased sodium entry is less than therate of potassium exit (remember that themembrane at rest is highly permeable toK+, which therefore flows out as themembrane is depolarized). The point atwhich Na+inflow just equals K+outflowrepresents an unstable equilibrium anal-ogous to the ignition point of an explo-sive mixture. The behavior of the mem-brane at threshold reflects this instability:The membrane potential may linger atthe threshold level for a variable periodbefore either returning to the restinglevel or flaring up into a full-blownaction potential. In theory at least, ifthere is a net internal gain of a single Na+ion, an action potential occurs; con-versely, the net loss of a single K+ionleads to repolarization. A more precisedefinition of threshold, therefore, is thatvalue of membrane potential, in depolar-izing from the resting potential, at whichthe current carried by Na+entering theneuron is exactly equal to the K+currentthat is flowing out. Once the triggeringevent depolarizes the membrane beyondthis point, the positive feedback loop ofNa+entry on membrane potential closesand the action potential “fires.”Because the Na+and K+conductanceschange dynamically over time, thethreshold potential for producing anaction potential also varies as a conse-quence of the previous activity of theneuron. For example, following an actionpotential, the membrane becomes tem-porarily refractory to further excitationbecause the threshold for firing an actionpotential transiently rises. There is, there-fore, no specific value of membranepotential that defines the threshold for agiven nerve cell in all circumstances.(A) (C)(B)Some heatescapesAdditionalheatproducedHeatsourceK+loss repolarizesmembrane potentialIncrease in Na+permeabilityNa+entryACTIONPOTENTIALDepolarizationof membraneHeat escapeslows reactionExothermicreactionIncrease inreaction rateCHEMICALEXPLOSIONHeatA positive feedback loop underlying the action potentialexplains the phenomenon of threshold.Purves03 5/13/04 1:29 PM Page 57
    • 58 Chapter ThreeIf the experiment shown in Figure 3.10 is repeated with a depolarizingcurrent pulse large enough to produce an action potential, the result is dra-matically different (Figure 3.11A). In this case, an action potential occurswithout decrement along the entire length of the axon, which in humans−50−65−65−62−59−60−55Membranepotential(mV)Membranepotential(mV)ThresholdCurrentinjectionelectrode0 1010 20 30 40 0 10 20 30 40Time (ms)0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 400−0.5 0 0.5 1.0 1.5 2.0 2.520 30 40Distance along axon (mm)(A)(B)(C)Resting potentialPotentialrecordingelectrodesRecord Record Record Record Record Record RecordStimulateAxon100 20 30 40−65−60Membranepotential(mV)Time (ms) 2.50Distance from currentinjection (mm)(D)−0−1Current(nA)+11 mmFigure 3.10 Passive current flow in anaxon. (A) Experimental arrangement forexamining the local flow of electricalcurrent in an axon. A current-passingelectrode produces a subthresholdchange in membrane potential, whichspreads passively along the axon. (B)Potential responses recorded at the posi-tions indicated by microelectrodes. Withincreasing distance from the site of cur-rent injection, the amplitude of thepotential change is attenuated. (C) Rela-tionship between the amplitude ofpotential responses and distance. (D)Superimposed responses (from B) tocurrent pulse, measured at indicateddistances along axon. Note that theresponses develop more slowly atgreater distances from the site of currentinjection, for reasons explained in BoxC. (After Hodgkin and Rushton, 1938.)Purves03 5/13/04 1:29 PM Page 58
    • may be a distance of a meter or more (Figure 3.11B). Thus, action potentialssomehow circumvent the inherent leakiness of neurons.How, then, do action potentials traverse great distances along such a poorpassive conductor? The answer is in part provided by the observation thatthe amplitude of the action potentials recorded at different distances is con-stant. This all-or-none behavior indicates that more than simple passive flowof current must be involved in action potential propagation. A second cluecomes from examination of the time of occurrence of the action potentialsrecorded at different distances from the site of stimulation: Action potentialsoccur later and later at greater distances along the axon (Figure 3.11B). Thus,the action potential has a measurable rate of transmission, called the con-duction velocity. The delay in the arrival of the action potential at succes-sively more distant points along the axon differs from the case shown in Fig-ure 3.10, in which the electrical changes produced by passive current flowoccur at more or less the same time at successive points.The mechanism of action potential propagation is easy to grasp once oneunderstands how action potentials are generated and how current passivelyflows along an axon (Figure 3.12). A depolarizing stimulus—a synapticpotential or a receptor potential in an intact neuron, or an injected currentpulse in an experiment—locally depolarizes the axon, thus opening the volt-age-sensitive Na+channels in that region. The opening of Na+channelscauses inward movement of Na+, and the resultant depolarization of themembrane potential generates an action potential at that site. Some of thelocal current generated by the action potential will then flow passively downVoltage-Dependent Membrane Permeability 59ms−65−65−50−50−252500PotentialrecordingelectrodesCurrentinjectionelectrodeAxonms ms ms msms0 2 6 84ms0 2 6 84 0 2 6 84 0 2 6 84 0 2 6 84 0 2 6 84 0 2 6 84Membranepotential(mV)Record Record Record Record Record Record RecordStimulateMembranepotential(mV)−0.5 0 0.5 1.0 1.5 2.0 2.5Distance along axon (mm)(C)(B)(A)ThresholdResting potential1 mmFigure 3.11 Propagation of an actionpotential. (A) In this experimentalarrangement, an electrode evokes anaction potential by injecting a supra-threshold current. (B) Potentialresponses recorded at the positions indi-cated by microelectrodes. The amplitudeof the action potential is constant alongthe length of the axon, although thetime of appearance of the action poten-tial is delayed with increasing distance.(C) The constant amplitude of an actionpotential (solid black line) measured atdifferent distances.Purves03 5/13/04 1:29 PM Page 59
    • 60 Chapter Threethe axon, in the same way that subthreshold currents spread along the axon(see Figure 3.10). Note that this passive current flow does not require themovement of Na+along the axon but, instead, occurs by a shuttling ofcharge, somewhat similar to what happens when wires passively conductelectricity by transmission of electron charge. This passive current flowdepolarizes the membrane potential in the adjacent region of the axon, thusopening the Na+channels in the neighboring membrane. The local depolar-ization triggers an action potential in this region, which then spreads againin a continuing cycle until the end of the axon is reached. Thus, action poten-tial propagation requires the coordinated action of two forms of currentBox CPassive Membrane PropertiesThe passive flow of electrical currentplays a central role in action potentialpropagation, synaptic transmission, andall other forms of electrical signaling innerve cells. Therefore, it is worthwhileunderstanding in quantitative terms howpassive current flow varies with distancealong a neuron. For the case of a cylindri-cal axon, such as the one depicted in Fig-ure 3.10, subthreshold current injectedinto one part of the axon spreads pas-sively along the axon until the current isdissipated by leakage out across the axonmembrane. The decrement in the currentflow with distance (Figure A) is describedby a simple exponential function:Vx = V0 e–x/λwhere Vx is the voltage response at anydistance x along the axon, V0 is the volt-age change at the point where current isinjected into the axon, e is the base ofnatural logarithms (approximately 2.7),and λ is the length constant of the axon.As evident in this relationship, the lengthconstant is the distance where the initialvoltage response (V0) decays to 1/e (or37%) of its value. The length constant isthus a way to characterize how far pas-sive current flow spreads before it leaksout of the axon, with leakier axons hav-ing shorter length constants.The length constant depends uponthe physical properties of the axon, inparticular the relative resistances of theplasma membrane (rm), the intracellularaxoplasm (ri), and the extracellularmedium (r0). The relationship betweenthese parameters is:Hence, to improve the passive flow ofcurrent along an axon, the resistance ofthe plasma membrane should be as highas possible and the resistances of theaxoplasm and extracellular mediumshould be low.Another important consequence ofthe passive properties of neurons is thatcurrents flowing across a membrane donot immediately change the membranepotential. For example, when a rectangu-lar current pulse is injected into the axonshown in the experiment illustrated inFigure 3.10A, the membrane potentialdepolarizes slowly over a few millisec-onds and then repolarizes over a similartime course when the current pulse ends(see Figure 3.10D). These delays inchanging the membrane potential aredue to the fact that the plasma mem-λ =+rr rm0 i1. 1 2 3 4 5−5 −4 −3 −2 −1Distance from current injection (mm)37%VX= V0e−x/λVX/V0λ λ(A) Spatial decay of membrane potential along a cylindrical axon. A current pulse injected atone point in the axon (0 mm) produces voltage responses (Vx) that decay exponentially withdistance. The distance where the voltage response is 1/e of its initial value (V0) is the lengthconstant, λ.Purves03 5/13/04 1:29 PM Page 60
    • flow—the passive flow of current as well as active currents flowing throughvoltage-dependent ion channels. The regenerative properties of Na+channelopening allow action potentials to propagate in an all-or-none fashion byacting as a booster at each point along the axon, thus ensuring the long-dis-tance transmission of electrical signals.The Refractory PeriodRecall that the depolarization that produces Na+channel opening alsocauses delayed activation of K+channels and Na+channel inactivation, lead-Voltage-Dependent Membrane Permeability 61brane behaves as a capacitor, storing theinitial charge that flows at the beginningand end of the current pulse. For thecase of a cell whose membrane potentialis spatially uniform, the change in themembrane potential at any time, Vt, afterbeginning the current pulse (Figure B)can also be described by an exponentialrelationship:Vt = V∞(1 − e−t/τ)where V∞ is the steady-state value of themembrane potential change, t is the timeafter the current pulse begins, and τ isthe membrane time constant. The timeconstant is thus defined as the timewhen the voltage response (Vt) rises to1 − (1/e) (or 63%) of V∞. After the currentpulse ends, the membrane potentialchange also declines exponentiallyaccording to the relationshipVt = V∞ e−t/τDuring this decay, the membrane poten-tial returns to 1/e of V∞ at a time equal tot. For cells with more complex geome-tries than the axon in Figure 3.10, thetime courses of the changes in mem-brane potential are not simple exponen-tials, but nonetheless depend on themembrane time constant. Thus, the timeconstant characterizes how rapidly cur-rent flow changes the membrane poten-tial. The membrane time constant alsodepends on the physical properties of thenerve cell, specifically on the resistance(rm) and capacitance (cm) of the plasmamembrane such that:τ = rmcmThe values of rm and cm depend, in part,on the size of the neuron, with largercells having lower resistances and largercapacitances. In general, small nerve cellstend to have long time constants andlarge cells brief time constants.ReferencesHODGKIN, A. L. AND W. A. H. RUSHTON (1938)The electrical constants of a crustacean nervefibre. Proc. R. Soc. Lond. 133: 444–479.JOHNSTON, D. AND S. M.-S. WU (1995) Founda-tions of Cellular Neurophysiology. Cambridge,MA: MIT Press.RALL, W. (1977) Core conductor theory andcable properties of neurons. In Handbook ofPhysiology, Section 1: The Nervous System,Vol. 1: Cellular Biology of Neurons. E. R. Kan-del (ed.). Bethesda, MD: American Physio-logical Society, pp. 39– 5 10 15 20 25 30 35 40Time (ms)37%63%Vt= V∞e−t/τVt= V∞(1 – e−t/τ)V∞/Vτ−0−1Current(nA)+1τ τ(B) Time course of potential changes produced in a spatially uniform cell by a current pulse.The rise and fall of the membrane potential (Vt) can be described as exponential functions,with the time constant τ defining the time required for the response to rise to 1 – (1/e) of thesteady-state value (V∞), or to decline to 1/e of V∞.Purves03 5/13/04 1:29 PM Page 61
    • 62 Chapter ThreeAxonPoint A Point B Point CPoint A Point B Point CPoint A Point B Point CNa+Na+Na+Na+ Na+channelK+channel Membrane1 Na+ channels locally open inresponse to stimulus, generatingan action potential here2 Some depolarizing currentpassively flows down axonK+K+K+Na+Na+Na+K+K+K+Na+3 Local depolarization causes neighboringNa+channels to open and generates anaction potential here4 Upstream Na+channels inactivate, whileK+channels open. Membrane potentialrepolarizes and axon is refractory here5 The process is repeated, propagatingthe action potential along the axonNa+t=2t=3t=3t=2t=1Point APoint BPoint Ct=1Stimulate0 mV−65 Resting potentialThreshold0−65 Resting potentialThreshold0−65 Resting potentialThresholdPurves Neuroscience 3EPyramis StudiosP3_312Figure 3.12 Action potential conduction requires both active and passivecurrent flow. Depolarization opens Na+channels locally and produces anaction potential at point A of the axon (time t = 1). The resulting inwardcurrent flows passively along the axon, depolarizing the adjacent region(point B) of the axon. At a later time (t = 2), the depolarization of the adja-cent membrane has opened Na+channels at point B, resulting in the initia-tion of the action potential at this site and additional inward current thatagain spreads passively to an adjacent point (point C) farther along theaxon. At a still later time (t = 3), the action potential has propagated evenfarther. This cycle continues along the full length of the axon. Note that asthe action potential spreads, the membrane potential repolarizes due to K+channel opening and Na+channel inactivation, leaving a “wake” of refrac-toriness behind the action potential that prevents its backward propaga-tion (panel 4). The panel to the left of this figure legend shows the timecourse of membrane potential changes at the points indicated.Purves03 5/13/04 1:29 PM Page 62
    • ing to repolarization of the membrane potential as the action potentialsweeps along the length of an axon (see Figure 3.12). In its wake, the actionpotential leaves the Na+channels inactivated and K+channels activated for abrief time. These transitory changes make it harder for the axon to producesubsequent action potentials during this interval, which is called the refrac-tory period. Thus, the refractory period limits the number of action poten-tials that a given nerve cell can produce per unit time. As might be expected,different types of neurons have different maximum rates of action potentialfiring due to different types and densities of ion channels. The refractorinessof the membrane in the wake of the action potential also explains why actionpotentials do not propagate back toward the point of their initiation as theytravel along an axon.Increased Conduction Velocity as a Result of MyelinationThe rate of action potential conduction limits the flow of information withinthe nervous system. It is not surprising, then, that various mechanisms haveevolved to optimize the propagation of action potentials along axons.Because action potential conduction requires passive and active flow of cur-rent (see Figure 3.12), the rate of action potential propagation is determinedby both of these phenomena. One way of improving passive current flow isto increase the diameter of an axon, which effectively decreases the internalresistance to passive current flow (see Box C). The consequent increase inaction potential conduction velocity presumably explains why giant axonsevolved in invertebrates such as squid, and why rapidly conducting axons inall animals tend to be larger than slowly conducting ones.Another strategy to improve the passive flow of electrical current is toinsulate the axonal membrane, reducing the ability of current to leak out ofthe axon and thus increasing the distance along the axon that a given localcurrent can flow passively (see Box C). This strategy is evident in the myeli-nation of axons, a process by which oligodendrocytes in the central nervoussystem (and Schwann cells in the peripheral nervous system) wrap the axonin myelin, which consists of multiple layers of closely opposed glial mem-branes (Figure 3.13; see also Chapter 1). By acting as an electrical insulator,myelin greatly speeds up action potential conduction (Figure 3.14). Forexample, whereas unmyelinated axon conduction velocities range fromabout 0.5 to 10 m/s, myelinated axons can conduct at velocities of up to 150m/s. The major reason underlying this marked increase in speed is that thetime-consuming process of action potential generation occurs only at spe-cific points along the axon, called nodes of Ranvier, where there is a gap inthe myelin wrapping (see Figure 1.4F). If the entire surface of an axon wereinsulated, there would be no place for current to flow out of the axon andaction potentials could not be generated. As it happens, an action potentialgenerated at one node of Ranvier elicits current that flows passively withinthe myelinated segment until the next node is reached. This local currentflow then generates an action potential in the neighboring segment, and thecycle is repeated along the length of the axon. Because current flows acrossthe neuronal membrane only at the nodes (see Figure 3.13), this type ofpropagation is called saltatory, meaning that the action potential jumpsfrom node to node. Not surprisingly, loss of myelin, as occurs in diseasessuch as multiple sclerosis, causes a variety of serious neurological problems(Box D).Voltage-Dependent Membrane Permeability 63Purves03 5/13/04 1:29 PM Page 63
    • AxonMyelin sheathNode ofRanvier(A) Myelinated axonNa+NaNa+Na+NaNa+Na+Na+Na+K+K+K+K+K+(B) Action potential propagationt=1Na+Na+t=1.5t=2OligodendrocytePoint A Point B Point CPoint A Point B Point CPoint A Point B Point Ct=2t=1.5t=1Point APoint BPoint C0 mV−65 Resting potentialThreshold0−65 Resting potentialThreshold0−65 Resting potentialThresholdPurves Neuroscience 3EPyramis StudiosP3_313120103Figure 3.13 Saltatory action potential conduction along a myeli-nated axon. (A) Diagram of a myelinated axon. (B) Local current inresponse to action potential initiation at a particular site flowslocally, as described in Figure 3.12. However, the presence of myelinprevents the local current from leaking across the internodal mem-brane; it therefore flows farther along the axon than it would in theabsence of myelin. Moreover, voltage-gated Na+channels are presentonly at the nodes of Ranvier (K+channels are present at the nodes ofsome neurons, but not others). This arrangement means that thegeneration of active, voltage-gated Na+currents need only occur atthese unmyelinated regions. The result is a greatly enhanced velocityof action potential conduction. The panel to the left of this figure leg-end shows the time course of membrane potential changes at thepoints indicated.Purves03 5/13/04 1:29 PM Page 64
    • SummaryThe action potential and all its complex properties can be explained by time-and voltage-dependent changes in the Na+and K+permeabilities of neu-ronal membranes. This conclusion derives primarily from evidence obtainedby a device called the voltage clamp. The voltage clamp technique is an elec-tronic feedback method that allows control of neuronal membrane potentialVoltage-Dependent Membrane Permeability 65Myelinated axonUnmyelinated axont=2t=1t=3Figure 3.14 Comparison of speed ofaction potential conduction in unmyeli-nated (upper) and myelinated (lower)axons.Purves03 5/13/04 1:29 PM Page 65
    • 66 Chapter ThreeBox DMultiple SclerosisMultiple sclerosis (MS) is a disease of thecentral nervous system characterized bya variety of clinical problems arisingfrom multiple regions of demyelinationand inflammation along axonal path-ways. The disorder commonly beginsbetween ages 20 and 40, characterized bythe abrupt onset of neurological deficitsthat typically persist for days or weeksand then remit. The clinical courseranges from patients with no persistentneurological loss, some of whom experi-ence only occasional later exacerbations,to others who progressively deteriorateas a result of extensive and relentlesscentral nervous system involvement.The signs and symptoms of MS aredetermined by the location of theaffected regions. Particularly commonare monocular blindness (due to lesionsof the optic nerve), motor weakness orparalysis (due to lesions of the corti-cospinal tracts), abnormal somatic sensa-tions (due to lesions of somatic sensorypathways, often in the posteriorcolumns), double vision (due to lesionsof medial longitudinal fasciculus), anddizziness (due to lesions of vestibularpathways). Abnormalities are oftenapparent in the cerebrospinal fluid,which usually contains an abnormalnumber of cells associated with inflam-mation and an increased content of anti-bodies (a sign of an altered immuneresponse). The diagnosis of MS generallyrelies on the presence of a neurologicalproblem that remits and then returns atan unrelated site. Confirmation cansometimes be obtained from magneticresonance imaging (MRI), or functionalevidence of lesions in a particular path-way by abnormal evoked potentials. Thehistological hallmark of MS at post-mortem exam is multiple lesions at dif-ferent sites showing loss of myelin asso-ciated with infiltration of inflammatorycells and, in some instances, loss of axonsthemselves.The concept of MS as a demyelinatingdisease is deeply embedded in the clini-cal literature, although precisely how thedemyelination translates into functionaldeficits is poorly understood. The loss ofthe myelin sheath surrounding manyaxons clearly compromises action poten-tial conduction, and the abnormal pat-terns of nerve conduction that result pre-sumably produce most of the clinicaldeficits in the disease. However, MS mayhave effects that extend beyond loss ofthe myelin sheath. It is clear that someaxons are actually destroyed, probably asa result of inflammatory processes in theoverlying myelin and/or loss of trophicsupport of the axon by oligodendrocytes.Thus, axon loss also contributes to thefunctional deficits in MS, especially in thechronic, progressive forms of the disease.The ultimate cause of MS remainsunclear. The immune system undoubt-edly contributes to the damage and newimmunoregulatory therapies providesubstantial benefits to many patients.Precisely how the immune system is acti-vated to cause the injury is not known.The most popular hypothesis is that MSis an autoimmune disease (i.e., a diseasein which the immune system attacks thebody’s proper constituents). The fact thatimmunization of experimental animalswith any one of several molecular con-stituents of the myelin sheath can inducea demyelinating disease (called experi-mental allergic encephalomyelitis) showsthat an autoimmune attack on themyelin membrane is sufficient to pro-duce a picture similar to MS. A possibleexplanation of the human disease is thata genetically susceptible individualbecomes transiently infected (by a minorviral illness, for example) with a microor-ganism that expresses a molecule struc-turally similar to a component of myelin.An immune response to this antigen ismounted to attack the invader, but thefailure of the immune system to discrim-inate between the foreign protein andself results in destruction of otherwisenormal myelin, a scenario occurring inmice infected with Theiler’s virus.An alternative hypothesis is that MSis caused by a persistent infection by avirus or other microorganism. In thisinterpretation, the immune system’songoing efforts to get rid of the pathogencause the damage to myelin. Tropicalspastic paraparesis (TSP) provides aprecedent for this idea. TSP is a diseasecharacterized by the gradual progressionof weakness of the legs and impairedcontrol of bladder function associatedwith increased deep tendon reflexes anda positive Babinski sign (see Chapter 16).This clinical picture is similar to that ofrapidly advancing MS. TSP is known tobe caused by persistent infection with aretrovirus (human T lymphotropicvirus-1). This precedent notwithstand-ing, proving the persistent viral infectionhypothesis for MS requires unambigu-ous demonstration of the presence of avirus. Despite periodic reports of a virusassociated with MS, convincing evidencehas not been forthcoming. In sum, MSremains a daunting clinical challenge.ReferencesADAMS, R. D. AND M. VICTOR (2001) Principlesof Neurology, 7th Ed. New York: McGraw-Hill, pp. 954–982.MILLER, D. H. AND 9 OTHERS. (2003) A con-trolled trial of natalizumab for relapsing mul-tiple sclerosis. N. Engl. J. Med. 348: 15–23.ZANVIL, S. S. AND L. STEINMAN (2003) Diversetargets for intervention during inflammatoryand neurodegenerative phases of multiplesclerosis. Neuron 38: 685–688.Purves03 5/13/04 1:29 PM Page 66
    • and, simultaneously, direct measurement of the voltage-dependent fluxes ofNa+and K+that produce the action potential. Voltage clamp experimentsshow that a transient rise in Na+conductance activates rapidly and theninactivates during a sustained depolarization of the membrane potential.Such experiments also demonstrate a rise in K+conductance that activates ina delayed fashion and, in contrast to the Na+conductance, does not inacti-vate. Mathematical modeling of the properties of these conductances indi-cates that they, and they alone, are responsible for the production of all-or-none action potentials in the squid axon. Action potentials propagate alongthe nerve cell axons initiated by the voltage gradient between the active andinactive regions of the axon by virtue of the local current flow. In this way,action potentials compensate for the relatively poor passive electrical prop-erties of nerve cells and enable neural signaling over long distances. Theseclassical electrophysiological findings provide a solid basis for consideringthe functional and ultimately molecular variations on neural signaling takenup in the next chapter.Voltage-Dependent Membrane Permeability 67Additional ReadingReviewsARMSTRONG, C. M. AND B. HILLE (1998) Volt-age-gated ion channels and electricalexcitability. Neuron 20: 371–80.NEHER, E. (1992) Ion channels for communica-tion between and within cells. Science 256:498–502.Important Original PapersARMSTRONG, C. M. AND L. BINSTOCK (1965)Anomalous rectification in the squid giantaxon injected with tetraethylammonium chlo-ride. J. Gen. Physiol. 48: 859–872.HODGKIN, A. L. AND A. F. HUXLEY (1952a) Cur-rents carried by sodium and potassium ionsthrough the membrane of the giant axon ofLoligo. J. Physiol. 116: 449–472.HODGKIN, A. L. AND A. F. HUXLEY (1952b) Thecomponents of membrane conductance in thegiant axon of Loligo. J. Physiol. 116: 473–496.HODGKIN, A. L. AND A. F. HUXLEY (1952c) The-dual effect of membrane potential on sodiumconductance in the giant axon of Loligo. J.Physiol. 116: 497–506.HODGKIN, A. L. AND A. F. HUXLEY (1952d) Aquantitative description of membrane currentand its application to conduction and excita-tion in nerve. J. Physiol. 116: 507–544.HODGKIN, A. L. AND W. A. H. RUSHTON (1938)The electrical constants of a crustacean nervefibre. Proc. R. Soc. Lond. 133: 444–479.HODGKIN, A. L., A. F. HUXLEY AND B. KATZ(1952) Measurements of current–voltage rela-tions in the membrane of the giant axon ofLoligo. J. Physiol. 116: 424–448.MOORE, J. W., M. P. BLAUSTEIN, N. C. ANDER-SON AND T. NARAHASHI (1967) Basis oftetrodotoxin’s selectivity in blockage of squidaxons. J. Gen. Physiol. 50: 1401–1411.BooksAIDLEY, D. J. AND P. R. STANFIELD (1996) IonChannels: Molecules in Action. Cambridge:Cambridge University Press.HILLE, B. (2001) Ion Channels of Excitable Mem-branes, 3rd Ed. Sunderland, MA: SinauerAssociates.JOHNSTON, D. AND S. M.-S. WU (1995) Founda-tions of Cellular Neurophysiology. Cambridge,MA: MIT Press.JUNGE, D. (1992) Nerve and Muscle Excitation,3rd Ed. Sunderland, MA: Sinauer Associates.Purves03 5/13/04 1:29 PM Page 67
    • Purves03 5/13/04 1:29 PM Page 68
    • OverviewThe generation of electrical signals in neurons requires that plasma mem-branes establish concentration gradients for specific ions and that thesemembranes undergo rapid and selective changes in the membrane perme-ability to these ions. The membrane proteins that create and maintain iongradients are called active transporters, whereas other proteins called ionchannels give rise to selective ion permeability changes. As their nameimplies, ion channels are transmembrane proteins that contain a specializedstructure, called a pore, that permits particular ions to cross the neuronalmembrane. Some of these channels also contain other structures that are ableto sense the electrical potential across the membrane. Such voltage-gatedchannels open or close in response to the magnitude of the membrane poten-tial, allowing the membrane permeability to be regulated by changes in thispotential. Other types of ion channels are gated by extracellular chemicalsignals such as neurotransmitters, and some by intracellular signals such assecond messengers. Still others respond to mechanical stimuli, temperaturechanges, or a combination of such effects. Many types of ion channels havenow been characterized at both the gene and protein level, resulting in theidentification of a large number of ion channel subtypes that are expresseddifferentially in neuronal and non-neuronal cells. The specific expressionpattern of ion channels in each cell type can generate a wide spectrum ofelectrical characteristics. In contrast to ion channels, active transporters aremembrane proteins that produce and maintain ion concentration gradients.The most important of these is the Na+pump, which hydrolyzes ATP to reg-ulate the intracellular concentrations of both Na+and K+. Other active trans-porters produce concentration gradients for the full range of physiologicallyimportant ions, including Cl–, Ca2+, and H+. From the perspective of electri-cal signaling, active transporters and ion channels are complementary:Transporters create the concentration gradients that help drive ion fluxesthrough open ion channels, thus generating electrical signals.Ion Channels Underlying Action PotentialsAlthough Hodgkin and Huxley had no knowledge of the physical nature ofthe conductance mechanisms underlying action potentials, they nonethelessproposed that nerve cell membranes have channels that allow ions to passselectively from one side of the membrane to the other (see Chapter 3).Based on the ionic conductances and currents measured in voltage clampexperiments, the postulated channels had to have several properties. First,because the ionic currents are quite large, the channels had to be capable ofallowing ions to move across the membrane at high rates. Second, becauseChapter 469Channels andTransportersPurves04 5/13/04 1:41 PM Page 69
    • 70 Chapter FourA wealth of new information about ionchannels resulted from the invention ofthe patch clamp method in the 1970s.This technique is based on a very simpleidea. A glass pipette with a very smallopening is used to make tight contactwith a tiny area, or patch, of neuronalmembrane. After the application of asmall amount of suction to the back ofthe pipette, the seal between pipette andmembrane becomes so tight that no ionscan flow between the pipette and themembrane. Thus, all the ions that flowwhen a single ion channel opens mustflow into the pipette. The resulting elec-trical current, though small, can be mea-sured with an ultrasensitive electronicamplifier connected to the pipette. Basedon the geometry involved, this arrange-ment usually is called the cell-attachedpatch clamp recording method. As with theconventional voltage clamp method, thepatch clamp method allows experimen-tal control of the membrane potential tocharacterize the voltage dependence ofmembrane currents.Although the ability to record cur-rents flowing through single ion chan-nels is an important advantage of thecell-attached patch clamp method, minortechnical modifications yield still otheradvantages. For example, if the mem-brane patch within the pipette is dis-rupted by briefly applying strong suc-tion, the interior of the pipette becomescontinuous with the cytoplasm of thecell. This arrangement allows measure-ments of electrical potentials and cur-rents from the entire cell and is thereforecalled the whole-cell recording method. Thewhole-cell configuration also allows dif-fusional exchange between the pipetteand the cytoplasm, producing a conve-nient way to inject substances into theinterior of a “patched” cell.Two other variants of the patch clampmethod originate from the finding thatonce a tight seal has formed between themembrane and the glass pipette, smallpieces of membrane can be pulled awayfrom the cell without disrupting the seal;this yields a preparation that is free ofthe complications imposed by the rest ofthe cell. Simply retracting a pipette thatis in the cell-attached configurationcauses a small vesicle of membrane toremain attached to the pipette. By expos-ing the tip of the pipette to air, the vesicleopens to yield a small patch of mem-brane with its (former) intracellular sur-RetractpipetteStrongpulse ofsuctionCytoplasm is continuouswith pipette interiorMildsuctionTight contact betweenpipette and membraneExposeto airCytoplasmicdomain accessibleEnds ofmembraneannealExtracellulardomain accessibleWhole-cell recordingOutside-out recordingInside-out recordingRecording pipetteCell-attached recordingFour configurations in patch clamp measurements of ionic currents.Box AThe Patch Clamp MethodPurves04 5/13/04 1:41 PM Page 70
    • the ionic currents depend on the electrochemical gradient across the mem-brane, the channels had to make use of these gradients. Third, because Na+and K+flow across the membrane independently of each other, differentchannel types had to be capable of discriminating between Na+and K+,allowing only one of these ions to flow across the membrane under the rele-vant conditions. Finally, given that the conductances are voltage-dependent,the channels had to be able to sense the voltage drop across the membrane,opening only when the voltage reached appropriate levels. While this con-cept of channels was highly speculative in the 1950s, later experimentalwork established beyond any doubt that transmembrane proteins calledvoltage-sensitive ion channels indeed exist and are responsible for all of theionic conductance phenomena described in Chapter 3.The first direct evidence for the presence of voltage-sensitive, ion-selectivechannels in nerve cell membranes came from measurements of the ionic cur-rents flowing through individual ion channels. The voltage-clamp apparatusused by Hodgkin and Huxley could only resolve the aggregate current result-ing from the flow of ions through many thousands of channels. A techniquecapable of measuring the currents flowing through single channels wasdevised in 1976 by Erwin Neher and Bert Sakmann at the Max Planck Insti-tute in Goettingen. This remarkable approach, called patch clamping (BoxA), revolutionized the study of membrane currents. In particular, the patchclamp method provided the means to test directly Hodgkin and Huxley’sproposals about the characteristics of ion channels.Currents flowing through Na+channels are best examined in experimentalcircumstances that prevent the flow of current through other types of chan-nels that are present in the membrane (e.g., K+channels). Under such condi-tions, depolarizing a patch of membrane from a squid giant axon causes tinyinward currents to flow, but only occasionally (Figure 4.1). The size of thesecurrents is minuscule—approximately l–2 pA (i.e., 10–12ampere), which isorders of magnitude smaller than the Na+currents measured by voltageclamping the entire axon. The currents flowing through single channels arecalled microscopic currents to distinguish them from the macroscopic cur-rents flowing through a large number of channels distributed over a muchmore extensive region of surface membrane. Although microscopic currentsare certainly small, a current of 1 pA nonetheless reflects the flow of thou-sands of ions per millisecond. Thus, as predicted, a single channel can letmany ions pass through the membrane in a very short time.Channels and Transporters 71face exposed. This arrangement, calledthe inside-out patch recording configura-tion, allows the measurement of single-channel currents with the added benefitof making it possible to change themedium to which the intracellular sur-face of the membrane is exposed. Thus,the inside-out configuration is particu-larly valuable when studying the influ-ence of intracellular molecules on ionchannel function. Alternatively, if thepipette is retracted while it is in thewhole-cell configuration, a membranepatch is produced that has its extracellu-lar surface exposed. This arrangement,called the outside-out recording configu-ration, is optimal for studying how chan-nel activity is influenced by extracellularchemical signals, such as neurotransmit-ters (see Chapter 5). This range of possi-ble configurations makes the patchclamp method an unusually versatiletechnique for studies of ion channelfunction.ReferencesHAMILL, O. P., A. MARTY, E. NEHER, B. SAK-MANN AND F. J. SIGWORTH (1981) Improvedpatch-clamp techniques for high-resolutioncurrent recording from cells and cell-freemembrane patches. Pflügers Arch. 391:85–100.LEOIS, R. A. AND J. L. RAE (1998) Low-noisepatch-clamp techniques. Meth. Enzym. 293:218–266.SIGWORTH, F. J. (1986) The patch clamp ismore useful than anyone had expected. Fed.Proc. 45: 2673–2677.Purves04 5/13/04 1:41 PM Page 71
    • 72 Chapter FourSeveral observations further proved that the microscopic currents in Fig-ure 4.1B are due to the opening of single, voltage-activated Na+channels.First, the currents are carried by Na+; thus, they are directed inward whenthe membrane potential is more negative than ENa, reverse their polarity atENa, are outward at more positive potentials, and are reduced in size whenthe Na+concentration of the external medium is decreased. This behaviorexactly parallels that of the macroscopic Na+currents described in Chapter 3.Second, the channels have a time course of opening, closing, and inactivatingthat matches the kinetics of macroscopic Na+currents. This correspondenceis difficult to appreciate in the measurement of microscopic currents flowingthrough a single open channel, because individual channels open and closein a stochastic (random) manner, as can be seen by examining the individualtraces in Figure 4.1B. However, repeated depolarization of the membranepotential causes each Na+channel to open and close many times. When thecurrent responses to a large number of such stimuli are averaged together,the collective response has a time course that looks much like the macro-scopic Na+current (Figure 4.1C). In particular, the channels open mostly atthe beginning of a prolonged depolarization, showing that they subse-quently inactivate, as predicted from the macroscopic Na+current (compareFigures 4.1C and 4.1D). Third, both the opening and closing of the channelsare voltage-dependent; thus, the channels are closed at –80 mV but openwhen the membrane potential is depolarized. In fact, the probability thatany given channel will be open varies with membrane potential (Figure4.1E), again as predicted from the macroscopic Na+conductance (see Figure3.7). Finally, tetrodotoxin, which blocks the macroscopic Na+current (seeBox C), also blocks microscopic Na+currents. Taken together, these resultsshow that the macroscopic Na+current measured by Hodgkin and Huxleydoes indeed arise from the aggregate effect of many thousands of micro-scopic Na+currents, each representing the opening of a single voltage-sensi-tive Na+channel.Patch clamp experiments have also revealed the properties of the channelsresponsible for the macroscopic K+currents associated with action poten-tials. When the membrane potential is depolarized (Figure 4.2A), micro-scopic outward currents (Figure 4.2B) can be observed under conditions thatblock Na+channels. The microscopic outward currents exhibit all the fea-tures expected for currents flowing through action-potential-related K+channels. Thus, the microscopic currents (Figure 4.2C), like their macro-scopic counterparts (Figure 4.2D), fail to inactivate during brief depolariza-tions. Moreover, these single-channel currents are sensitive to ionic manipu-Figure 4.1 Patch clamp measurements of ionic currents flowing through singleNa+channels in a squid giant axon. In these experiments, Cs+was applied to theaxon to block voltage-gated K+channels. Depolarizing voltage pulses (A) applied toa patch of membrane containing a single Na+channel result in brief currents (B,downward deflections) in the seven successive recordings of membrane current(INa). (C) The sum of many such current records shows that most channels open inthe initial 1–2 ms following depolarization of the membrane, after which the proba-bility of channel openings diminishes because of channel inactivation. (D) A macro-scopic current measured from another axon shows the close correlation between thetime courses of microscopic and macroscopic Na+currents. (E) The probability of anNa+channel opening depends on the membrane potential, increasing as the mem-brane is depolarized. (B,C after Bezanilla and Correa, 1995; D after Vandenburg andBezanilla, 1991; E after Correa and Bezanilla, 1994.)ClosedOpen00 5 10 15200−600−800−400−200MacroscopicINa(pA)(B)(A)(D)−800−40Membranepotential(mV)MicroscopicINaMembrane potential (mV)Time (ms)0 5 10 15Time (ms)0 5 10 15Time (ms)0 5 10 15Time (ms)00.20.4−60−80 −40 –20 0 20 40 60(E)0.60.8ProbabilityofNa+channelopening0−0.8−0.40.4SummedmicroscopicINa(pA)(C)2 pAPurves04 5/13/04 1:41 PM Page 72
    • lations and drugs that affect the macroscopic K+currents and, like themacroscopic K+currents, are voltage-dependent (Figure 4.2E). This andother evidence shows that macroscopic K+currents associated with actionpotentials arise from the opening of many voltage-sensitive K+channels.In summary, patch clamping has allowed direct observation of micro-scopic ionic currents flowing through single ion channels, confirming thatvoltage sensitive Na+and K+channels are responsible for the macroscopicconductances and currents that underlie the action potential. Measurementsof the behavior of single ion channels has also provided some insight intothe molecular attributes of these channels. For example, single channel stud-ies show that the membrane of the squid axon contains at least two types ofchannels—one selectively permeable to Na+and a second selectively perme-able to K+. Both channel types are voltage-gated, meaning that their openingis influenced by membrane potential (Figure 4.3). For each channel, depolar-ization increases the probability of channel opening, whereas hyperpolariza-tion closes them (see Figures 4.1E and 4.2E). Thus, both channel types musthave a voltage sensor that detects the potential across the membrane (Figure4.3). However, these channels differ in important respects. In addition totheir different ion selectivities, depolarization also inactivates the Na+chan-nel but not the K+channel, causing Na+channels to pass into a nonconduct-ing state. The Na+channel must therefore have an additional molecularmechanism responsible for inactivation. And, as expected from the macro-scopic behavior of the Na+and K+currents described in Chapter 3, thekinetic properties of the gating of the two channels differs. This informationabout the physiology of single channels set the stage for subsequent studiesof the molecular diversity of ion channels in various cell types, and of theirdetailed functional characteristics.The Diversity of Ion ChannelsMolecular genetic studies, in conjunction with the patch clamp method andother techniques, have led to many additional advances in understandingion channels. Genes encoding Na+and K+channels, as well as many otherchannel types, have now been identified and cloned. A surprising fact thathas emerged from these molecular studies is the diversity of genes that codefor ion channels. Well over 100 ion channel genes have now been discovered,a number that could not have been anticipated from early studies of ionchannel function. To understand the functional significance of this multitudeof ion channel genes, the channels can be selectively expressed in well-Channels and Transporters 73Figure 4.2 Patch clamp measurements of ionic currents flowing through single K+channels in a squid giant axon. In these experiments, tetrodotoxin was applied tothe axon to block voltage-gated Na+channels. Depolarizing voltage pulses (A)applied to a patch of membrane containing a single K+channel results in brief cur-rents (B, upward deflections) whenever the channel opens. (C) The sum of such cur-rent records shows that most channels open with a delay, but remain open for theduration of the depolarization. (D) A macroscopic current measured from anotheraxon shows the correlation between the time courses of microscopic and macro-scopic K+currents. (E) The probability of a K+channel opening depends on themembrane potential, increasing as the membrane is depolarized. (B and C afterAugustine and Bezanilla, in Hille 1992; D after Augustine and Bezanilla, 1990; Eafter Perozo et al., 1991.)(A)(B)(C)−100500−50Membranepotential(mV)MicroscopicIKSummedmicroscopicIK(pA)MacroscopicIK(mA/cm2)Membrane potential (mV)0−60−80 −40 −20 0 20 40 60(E)(D)ProbabilityofK+channelopening0.60.40.20 10 20 30 40Time (ms)0 10 20 30 40Time (ms)0 10 20 30 40Time (ms)0 10 20 30 40Time (ms)012310OpenClosed 2 pAPurves04 5/13/04 1:41 PM Page 73
    • 74 Chapter FourFigure 4.3 Functional states of voltage-gated Na+and K+channels. The gates ofboth channels are closed when themembrane potential is hyperpolarized.When the potential is depolarized, volt-age sensors (indicated by +) allow thechannel gates to open—first the Na+channels and then the K+channels. Na+channels also inactivate during pro-longed depolarization, whereas manytypes of K+channels do not.defined experimental systems, such as in cultured cells or frog oocytes (BoxB), and then studied with patch clamping and other physiological tech-niques. Such studies have found many voltage-gated channels that respondto membrane potential in much the same way as the Na+and K+channelsthat underlie the action potential. Other channels, however, are gated bychemical signals that bind to extracellular or intracellular domains on theseproteins and are insensitive to membrane voltage. Still others are sensitive tomechanical displacement, or to changes in temperature.Further magnifying this diversity of ion channels are a number of mecha-nisms that can produce functionally different types of ion channels from asingle gene. Ion channel genes contain a large number of coding regions thatcan be spliced together in different ways, giving rise to channel proteins thatcan have dramatically different functional properties. RNAs encoding ionchannels also can be edited, modifying their base composition after tran-scription from the gene. For example, editing the RNA encoding of somereceptors for the neurotransmitter glutamate (Chapter 6) changes a singleamino acid within the receptor, which in turn gives rise to channels that dif-fer in their selectivity for cations and in their conductance. Channel proteinscan also undergo posttranslational modifications, such as phosphorylationby protein kinases (see Chapter 7), which can further change their functionalcharacteristics. Thus, although the basic electrical signals of the nervous sys-tem are relatively stereotyped, the proteins responsible for generating thesesignals are remarkably diverse, conferring specialized signaling properties tomany of the neuronal cell types that populate the nervous system. Thesechannels also are involved in a broad range of neurological diseases.+ +++ ++ + ++ +Na+CHANNELClosed Inactivated ClosedOpen InactivatingNa+Na+K+CHANNELK+K+Closed Closed ClosedOpen Open−100500−50Membranepotential(mV)0 5 10 15Time (ms)Purves04 5/13/04 1:41 PM Page 74
    • Channels and Transporters 75Box BExpression of Ion Channels in Xenopus OocytesBridging the gap between the sequenceof an ion channel gene and understand-ing channel function is a challenge. Tomeet this challenge, it is essential to havean experimental system in which thegene product can be expressed effi-ciently, and in which the function of theresulting channel can be studied withmethods such as the patch clamp tech-nique. Ideally, the vehicle for expressionshould be readily available, have fewendogenous channels, and be largeenough to permit mRNA and DNA to bemicroinjected with ease. Oocytes (imma-ture eggs) from the clawed African frog,Xenopus laevis (Figure A), fulfill all thesedemands. These huge cells (approxi-mately 1 mm in diameter; Figure B) areeasily harvested from the femaleXenopus. Work performed in the 1970s byJohn Gurdon, a developmental biologist,showed that injection of exogenousmRNA into frog oocytes causes them tosynthesize foreign protein in prodigiousquantities. In the early 1980s, RicardoMiledi, Eric Barnard, and other neurobi-ologists demonstrated that Xenopusoocytes could express exogenous ionchannels, and that physiological meth-ods could be used to study the ionic cur-rents generated by the newly-synthe-sized channels (Figure C).As a result of these pioneering stud-ies, heterologous expression experimentshave now become a standard way ofstudying ion channels. The approach hasbeen especially valuable in decipheringthe relationship between channel struc-ture and function. In such experiments,defined mutations (often affecting a sin-gle nucleotide) are made in the part ofthe channel gene that encodes a struc-ture of interest; the resulting channelproteins are then expressed in oocytes toassess the functional consequences ofthe mutation.The ability to combine molecular andphysiological methods in a single cellsystem has made Xenopus oocytes apowerful experimental tool. Indeed, thissystem has been as valuable to contem-porary studies of voltage-gated ionchannels as the squid axon was to suchstudies in the 1950s and 1960s.ReferencesGUNDERSEN, C. B., R. MILEDI AND I. PARKER(1984) Slowly inactivating potassium chan-nels induced in Xenopus oocytes by messen-ger ribonucleic acid from Torpedo brain. J.Physiol. (Lond.) 353: 231–248.GURDON, J. B., C. D. LANE, H. R. WOODLANDAND G. MARBAIX (1971) Use of frog eggs andoocytes for the study of messenger RNA andits translation in living cells. Nature 233:177–182.STÜHMER, W. (1998) Electrophysiologicalrecordings from Xenopus oocytes. Meth.Enzym. 293: 280–300.SUMIKAWA, K., M. HOUGHTON, J. S. EMTAGE, B.M. RICHARDS AND E. A. BARNARD (1981)Active multi-subunit ACh receptor assem-bled by translation of heterologous mRNAin Xenopus oocytes. Nature 292: 862– 864.(A)(B)(C)PHOTO: PU04BXCB.tifas in 2/eTime (s)−100−500Membranepotential(mV)K+current(µA)000.2 0.4 0.6 0.8 1.0501234(A) The clawed African frog, Xenopus laevis.(B) Several oocytes from Xenopus highlight-ing the dark coloration of the original poleand the lighter coloration of the vegetal pole.(Courtesy of P. Reinhart.) (C) Results of avoltage clamp experiment showing K+cur-rents produced following injection of K+channel mRNA into an oocyte. (After Gun-dersen et al., 1984.)Purves04 5/13/04 1:41 PM Page 75
    • 76 Chapter FourVoltage-Gated Ion ChannelsVoltage-gated ion channels that are selectively permeable to each of themajor physiological ions—Na+, K+, Ca2+, and Cl–—have now been discov-ered (Figure 4.4 A–D). Indeed, many different genes have been discoveredfor each type of voltage-gated ion channel. An example is the identificationof 10 human Na+channel genes. This finding was unexpected because Na+channels from many different cell types have similar functional properties,consistent with their origin from a single gene. It is now clear, however, thatall of these Na+channel genes (called SCN genes) produce proteins that dif-fer in their structure, function, and distribution in specific tissues. Forinstance, in addition to the rapidly inactivating Na+channels discovered byHodgkin and Huxley in squid axon, a voltage-sensitive Na+channel thatdoes not inactivate has been identified in mammalian axons. As might beexpected, this channel gives rise to action potentials of long duration and isa target of local anesthetics such as benzocaine and lidocaine.Other electrical responses in neurons entail the activation of voltage-gatedCa2+channels (Figure 4.4B). In some neurons, voltage-gated Ca2+channelsgive rise to action potentials in much the same way as voltage-sensitive Na+channels. In other neurons, Ca2+channels control the shape of action poten-tials generated primarily by Na+conductance changes. More generally, byaffecting intracellular Ca2+concentrations, the activity of Ca2+channels reg-ulates an enormous range of biochemical processes within cells (see Chapter7). Perhaps the most important of the processes regulated by voltage-sensi-tive Ca2+channels is the release of neurotransmitters at synapses (see Chap-ter 5). Given these crucial functions, it is perhaps not surprising that 16 dif-ferent Ca2+channel genes (called CACNA genes) have been identified. LikeNa+channels, Ca2+channels differ in their activation and inactivation prop-erties, allowing subtle variations in both electrical and chemical signalingprocesses mediated by Ca2+. As a result, drugs that block voltage-gated Ca2+channels are especially valuable in treating a variety of conditions rangingfrom heart disease to anxiety disorders.By far the largest and most diverse class of voltage-gated ion channels arethe K+channels (Figure 4.4C). Nearly 100 K+channel genes are now known,and these fall into several distinct groups that differ substantially in theiractivation, gating, and inactivation properties. Some take minutes to inacti-vate, as in the case of squid axon K+channels studied by Hodgkin and Hux-ley (Figure 4.5A). Others inactivate within milliseconds, as is typical of mostvoltage-gated Na+channels (Figure 4.5B). These properties influence the(A) Na+channel(B) Ca2+channel(C) K+channel(D) Cl−channel(E) Neurotransmitterreceptor(F) Ca2+-activatedK+ channel(G) Cyclic nucleotidegated channelOutsideInsideNa+ GlutamateCa2+K+K+Na+K+VOLTAGE-GATED CHANNELS LIGAND-GATED CHANNELScAMPcAMPcGMPNa++Ca2++K++Cl−+VoltagesensorFigure 4.4 Types of voltage-gated ionchannels. Examples of voltage-gatedchannels include those selectively per-meable to Na+(A), Ca2+(B), K+(C), andCl–(D). Ligand-gated ion channelsinclude those activated by the extracel-lular presence of neurotransmitters,such as glutamate (E). Other ligand-gated channels are activated by intracel-lular second messengers, such as Ca2+(F) or the cyclic nucleotides, cAMP andcGMP (G).Purves04 5/13/04 1:41 PM Page 76
    • Figure 4.5 Diverse properties of K+channels. Different types of K+channelswere expressed in Xenopus oocytes (seeBox B), and the voltage clamp methodwas used to change the membranepotential (top) and measure the result-ing currents flowing through each typeof channel. These K+channels varymarkedly in their gating properties, asevident in their currents (left) and con-ductances (right). (A) KV2.1 channelsshow little inactivation and are closelyrelated to the delayed rectifier K+chan-nels involved in action potential repolar-ization. (B) KV4.1 channels inactivateduring a depolarization. (C) HERGchannels inactivate so rapidly that cur-rent flows only when inactivation israpidly removed at the end of a depolar-ization. (D) Inward rectifying K+chan-nels allow more K+current to flow athyperpolarized potentials than at depo-larized potentials. (E) Ca2+-activated K+channels open in response to intracellu-lar Ca2+ions and, in some cases, mem-brane depolarization. (F) K+channelswith two pores usually respond tochemical signals, such as pH, ratherthan changes in membrane potential.Channels and Transporters 77Membranepotential(mV)−60−3003050−90−120Time (ms)0 100 200 300−100 0 100Membrane potential (mV)10−100 0 100Membrane potential (mV)10−100 0 100Membrane potential (mV)10−100 0 100Membrane potential (mV)10−100 0 100Membrane potential (mV)106 7 8pH10(A) KV2.1(B) KV4.1(C) HERG(D) Inwardrectifier(E) Ca2+-activated(F) 2-poreK+current(µA)K+current(µA)K+current(µA)K+current(µA)K+conductanceK+conductanceK+conductanceK+conductanceK+conductanceK+conductanceShawK+current(µA)K+current(µA)Time (ms)0 100 200 300pH 8pH 610 µM Ca2+1 µM Ca2++50 mV+50 mV+50 mV−120 mV−120 mV−120 mV−120 mV+50 mV−120 mV10µM Ca2+1µM Ca2++50 mV+50 mVPurves04 5/13/04 1:41 PM Page 77
    • 78 Chapter Fourduration and rate of action potential firing, with important consequences foraxonal conduction and synaptic transmission. Perhaps the most importantfunction of K+channels is the role they play in generating the resting mem-brane potential (see Chapter 2). At least two families of K+channels that areopen at substantially negative membrane voltage levels contribute to settingthe resting membrane potential (Figure 4.5D).Finally, several types of voltage-gated Cl–channel have been identified(see Figure 4.4D). These channels are present in every type of neuron, wherethey control excitability, contribute to the resting membrane potential, andhelp regulate cell volume.Ligand-Gated Ion ChannelsMany types of ion channels respond to chemical signals (ligands) ratherthan to changes in the membrane potential (Figure 4.4E–G). The mostimportant of these ligand-gated ion channels in the nervous system is theclass activated by binding neurotransmitters (Figure 4.4E). These channelsare essential for synaptic transmission and other forms of cell-cell signalingphenomena discussed in Chapters 5–7. Whereas the voltage-gated ion chan-nels underlying the action potential typically allow only one type of ion topermeate, channels activated by extracellular ligands are usually less selec-tive, allowing two or more types of ions to pass through the channel pore.Other ligand-gated channels are sensitive to chemical signals arisingwithin the cytoplasm of neurons (see Chapter 7), and can be selective forspecific ions such as K+or Cl–, or permeable to all physiological cations.Such channels are distinguised by ligand-binding domains on their intracel-lular surfaces that interact with second messengers such as Ca2+, the cyclicnucleotides cAMP and cGMP, or protons. Examples of channels that respondto intracellular cues include Ca2+-activated K+channels (Figure 4.4.F), thecyclic nucleotide gated cation channel (Figure 4.4G), or acid-sensing ionchannels (ASICs). The main function of these channels is to convert intracel-lular chemical signals into electrical information. This process is particularlyimportant in sensory transduction, where channels gated by cyclicnucleotides convert odors and light, for example, into electrical signals.Although many of these ligand-gated ion channels are located in the cellsurface membrane, others are in membranes of intracellular organelles suchas mitochondria or the endoplasmic reticulum . Some of these latter chan-nels are selectively permeable to Ca2+and regulate the release of Ca2+fromthe lumen of the endoplasmic reticulum into the cytoplasm, where this sec-ond messenger can then trigger a spectrum of cellular responses such asdescribed in Chapter 7.Stretch- and Heat-Activated ChannelsStill other ion channels respond to heat or membrane deformation. Heat-activated ion channels, such as some members of the transient receptorpotential (TRP) gene family, contribute to the sensations of pain and temper-ature and help mediate inflammation (see Chapter 9). These channels areoften specialized to detect specific temperature ranges, and some are evenactivated by cold. Other ion channels respond to mechanical distortion ofthe plasma membrane and are the basis of stretch receptors and neuromus-cular stretch reflexes (see Chapters 8, 15 and 16). A specialized form of thesechannels enables hearing by allowing auditory hair cells to respond to soundwaves (see Chapter 12).Purves04 5/13/04 1:41 PM Page 78
    • In summary, this tremendous variety of ion channels allows neurons togenerate electrical signals in response to changes in membrane potential,synaptic input, intracellular second messengers, light, odors, heat, sound,touch, and many other stimuli.The Molecular Structure of Ion ChannelsUnderstanding the physical structure of ion channels is obviously the key tosorting out how they actually work. Until recently, most information aboutchannel structure was derived indirectly from studies of the amino acidcomposition and physiological properties of these proteins. For example, agreat deal has been learned by exploring the functions of particular aminoacids within the proteins using mutagenesis and the expression of suchchannels in Xenopus oocytes (see Box B). Such studies have discovered a gen-eral transmembrane architecture common to all the major ion channel fami-lies. Thus, these molecules are all integral membrane proteins that span theplasma membrane repeatedly. Na+(and Ca2+) channel proteins, consist ofrepeating motifs of 6 membrane-spanning regions that are repeated 4 times,for a total of 24 transmembrane regions (Figure 4.6A,B). Na+(or Ca2+) chan-nels can be produced by just one of these proteins, although other accessoryproteins, called β subunits, can regulate the function of these channels.K+channel proteins typically span the membrane six times (Figure 4.6C),Channels and Transporters 79NNNNNNNNNNCCCCCCCCC CI II III IV I II III IV(A) Na+CHANNELβ subunitβ subunit(B) Ca2+CHANNEL(F) 2-pore(C) Kv and HERGK+CHANNELS(D) Inward rectifier (E) Ca2+-activated(G) Cl−CHANNELβ subunitFigure 4.6 Topology of the principalsubunits of voltage-gated Na+, Ca2+,K+, and Cl–channels. Repeating motifsof Na+(A) and Ca2+(B) channels arelabeled I, II, III, and IV; (C–F) K+chan-nels are more diverse. In all cases, foursubunits combine to form a functionalchannel. (G) Chloride channels arestructurally distinct from all othervoltage-gated channels.Purves04 5/13/04 1:41 PM Page 79
    • 80 Chapter Fourthough there are some K+channels, such as a bacterial channel and somemammalian channels, that span the membrane only twice (Figure 4.6D), andothers that span the membrane four times (Figure 4.6F) or seven times (Fig-ure 4.6E). Each of these K+channel proteins serves as a channel subunit,with 4 of these subunits typically aggregating to form a single functional ionchannel.Other imaginative mutagenesis experiments have provided informationabout how these proteins function. Two membrane-spanning domains of allion channels appear to form a central pore through which ions can diffuse,and one of these domains contains a protein loop that confers an ability toselectivity allow certain ions to diffuse through the channel pore (Figure 4.7).As might be expected, the amino acid composition of the pore loop differsamong channels that conduct different ions. These distinct structural fea-tures of channel proteins also provide unique binding sites for drugs and forvarious neurotoxins known to block specific subclasses of ion channels (BoxC). Furthermore, many voltage gated ion channels contain a distinct type oftransmembrane helix containing a number of positively charged amino acidsalong one face of the helix (Figures 4.6 and 4.7). This structure evidentlyserves as a sensor that detects changes in the electrical potential across themembrane. Membrane depolarization influences the charged amino acidssuch that the helix undergoes a conformational change, which in turn allowsthe channel pore to open. One suggestion is that the helix rotates to causethe pore to open (Figure 4.7). Other types of mutagenesis experiments havedemonstrated that one end of certain K+channels plays a key role in channelinactivation. This intracellular structure (labeled “N” in Figure 4.6C) canplug the channel pore during prolonged depolarization.More recently, very direct information about the structural underpinningsof ion channel function has come from X-ray crystallography studies of bac-terial K+channels (Figure 4.8). This molecule was chosen for analysisbecause the large quantity of channel protein needed for crystallographycould be obtained by growing large numbers of bacteria expressing this mol-ecule. The results of such studies showed that the channel is formed by sub-units that each cross the plasma membrane twice; between these two mem-brane-spanning structures is a loop that inserts into the plasma membrane(Figure 4.8A). Four of these subunits are assembled together to form a chan-Membrane depolarizationcauses charged helix torotatePore closed Pore openDepolarizeHyperpolarizeIonfluxFigure 4.7 A charged voltage sensorpermits voltage-dependent gating of ionchannels. The process of voltage activa-tion may involve the rotation of a posi-tively charged transmembrane domain.This movement causes a change in theconformation of the pore loop, enablingthe channel to conduct specific ions.Purves04 5/13/04 1:41 PM Page 80
    • nel (Figure 4.8B). In the center of the assembled channel is a narrow openingthrough the protein that allows K+to flow across the membrane. This open-ing is the channel pore and is formed by the protein loop, as well as by themembrane-spanning domains. The structure of the pore is well suited forconducting K+ions (Figure 4.8C). The narrowest part is near the outsidemouth of the channel and is so constricted that only a non-hydrated K+ioncan fit through the bottleneck. Larger cations, such as Cs+, cannot traversethis region of the pore, and smaller cations such as Na+cannot enter the porebecause the “walls” of the pore are too far apart to stabilize a dehydratedNa+ion. This part of the channel complex is responsible for the selective per-meability to K+and is therefore called the selectivity filter. The sequence ofamino acids making up part of this selectivity filter is often referred to as theK+channel “signature sequence”. Deeper within the channel is a water-filledcavity that connects to the interior of the cell. This cavity evidently collectsK+from the cytoplasm and, utilizing negative charges from the protein,Channels and Transporters 81(A)(C)(B)Pore helixOuter helixInner helixOuter helixInner helixSelectivity filterPore loopSIDE VIEW TOP VIEWK+ionin poreK+ionsPoreSelectivityfilterWater-filledcavityNegativelychargedpore helixFigure 4.8 Structure of a simple bacterial K+channel determined bycrystallography. (A) Structure of one subunit of the channel, which con-sists of two membrane-spanning domains and a pore loop that insertsinto the membrane. (B) Three-dimensional arrangement of four subunits(each in a different color) to form a K+channel. The top view illustrates aK+ion (green) within the channel pore. (C) The permeation pathway ofthe K+channel consists of a large aqueous cavity connected to a narrowselectivity filter. Helical domains of the channel point negative charges(red) toward this cavity, allowing K+ions (green) to become dehydratedand then move through the selectivity filter. (A, B from Doyle et al., 1998;C after Doyle et al., 1998.)Purves04 5/13/04 1:41 PM Page 81
    • 82 Chapter FourBox CToxins That Poison Ion ChannelsGiven the importance of Na+and K+channels for neuronal excitation, it is notsurprising that a number of organismshave evolved channel-specific toxins asmechanisms for self-defense or for cap-turing prey. A rich collection of naturaltoxins selectively target the ion channelsof neurons and other cells. These toxinsare valuable not only for survival, but forstudying the function of cellular ionchannels. The best-known channel toxinis tetrodotoxin, which is produced by cer-tain puffer fish and other animals.Tetrodotoxin produces a potent and spe-cific obstruction of the Na+channelsresponsible for action potential genera-tion, thereby paralyzing the animalsunfortunate enough to ingest it.Saxitoxin, a chemical homologue oftetrodotoxin produced by dinoflagel-lates, has a similar action on Na+chan-nels. The potentially lethal effects of eat-ing shellfish that have ingested these“red tide” dinoflagellates are due to thepotent neuronal actions of saxitoxin.Scorpions paralyze their prey byinjecting a potent mix of peptide toxinsthat also affect ion channels. Amongthese are the a-toxins, which slow theinactivation of Na+channels (Figure A1);exposure of neurons to these toxins pro-longs the action potential (Figure A2),thereby scrambling information flowwithin the nervous system of the soon-to-be-devoured victim. Other peptides inscorpion venom, called b-toxins, shift thevoltage dependence of Na+channel acti-vation (Figure B). These toxins cause Na+channels to open at potentials muchmore negative than normal, disruptingaction potential generation. Some alka-loid toxins combine these actions, bothremoving inactivation and shifting activa-tion of Na+channels. One such toxin isbatrachotoxin, produced by a species offrog; some tribes of South AmericanIndians use this poison on their arrowtips. A number of plants produce similartoxins, including aconitine, from butter-cups; veratridine, from lilies; and a num-ber of insecticidal toxins produced byplants such as chrysanthemums andrhododendrons.Potassium channels have also beentargeted by toxin-producing organisms.Peptide toxins affecting K+channelsinclude dendrotoxin, from wasps; apamin,from bees; and charybdotoxin, yet anothertoxin produced by scorpions. All of thesetoxins block K+channels as their primaryaction; no toxin is known to affect theactivation or inactivation of these chan-nels, although such agents may simplybe awaiting discovery.ReferencesCAHALAN, M. (1975) Modification of sodiumchannel gating in frog myelinated nervefibers by Centruroides sculpturatus scorpionvenom. J. Physiol. (Lond.) 244: 511–534.NARAHASHI, T. (2000) Neuroreceptors and ionchannels as the basis for drug action: Presentand future. J. Pharmacol. Exptl. Therapeutics294: 1–26.SCHMIDT, O. AND H. SCHMIDT (1972) Influenceof calcium ions on the ionic currents of nodesof Ranvier treated with scorpion venom.Pflügers Arch. 333: 51–61.(A)(1) (B)Membranepotential(mV)Membrane potential (mV)Membranepotential(mV)Na+current(nA/cm2)Time (ms)Time (ms)Time (s)+500−40−80+400−40−80−120000 20 40 60 80 0 20 40 60 800 8 1042 4 6−75ControlTreated withscorpion toxinTreated withscorpion toxinControl−25−50+25NormalizedNa+conductanceControlTreatedwithscorpiontoxin(2)(A) Effects of toxin treatment on frog axons.(1) α-Toxin from the scorpion Leiurusquinquestriatus prolongs Na+currentsrecorded with the voltage clamp method. (2)As a result of the increased Na+current, α-toxin greatly prolongs the duration of theaxonal action potential. Note the change intimescale after treating with toxin. (B) Treat-ment of a frog axon with β-toxin fromanother scorpion, Centruroides sculpturatus,shifts the activation of Na+channels, so thatNa+conductance begins to increase at poten-tials much more negative than usual. (A afterSchmidt and Schmidt, 1972; B after Cahalan,1975.)Purves04 5/13/04 1:41 PM Page 82
    • Figure 4.9 Structural features of K+channel gating. (A) Voltage sensing mayinvolve paddle-like structures of thechannel. These paddles reside withinthe lipid bilayer of the plasma mem-brane and may respond to changes inmembrane potential by moving throughthe membrane. The gating charges thatsense membrane potential are indicatedby red “plus” signs. (B) Structure of K+channels in closed (left) and open(right) conformations. Three of the fourchannel subunits are shown. Openingof the pore of the channel involves kink-ing of a transmembrane domain at thepoint indicated in red, which thendilates the pore. (A after Jiang et al.,2003; B after MacKinnon, 2003).allows K+ions to become dehydrated so they can enter the selectivity filter.These “naked” ions are then able to move through four K+binding siteswithin the selectivity filter to eventually reach the extracellular space (recallthat the normal concentration gradient drives K+out of cells). On average,two K+ions reside within the selectivity filter at any moment, with electro-static repulsion between the two ions helping to speed their transit throughthe selectivity filter, thereby permitting rapid ion flux through the channel.Crystallographic studies have also determined the structure of the voltagesensor in another type of bacterial K+channel. Such studies indicate that thesensor is at the interface between proteins and lipid on the cytoplasmic sur-face of the channel, leading to the suggestion that the sensor is a paddle-likestructure that moves through the membrane to gate the opening of the chan-nel pore (Figure 4.9A), rather than being a rotating helix buried within theion channel protein (as in Figure 4.7). Crystallographic work has alsorevealed the molecular basis of the rapid transitions between the closed andthe open state of the channel during channel gating. By comparing datafrom K+channels crystallized in what is believed to be closed and open con-formations (Figure 4.9B), it appears that channels gate by a conformationalchange in one of the transmembrane helices lining the channel pore. Pro-ducing a “kink” in one of these helices increases the opening from the cen-tral water-filled pore to the intracellular space, thereby permitting ion fluxes.Channels and Transporters 83(A)(B)(A)Closed OpenClosedDepolarizeHyperpolarizeOpenPurves04 5/13/04 1:41 PM Page 83
    • 84 Chapter FourSeveral genetic diseases, collectivelycalled channelopathies, result from smallbut critical alterations in ion channelgenes. The best-characterized of these dis-eases are those that affect skeletal musclecells. In these disorders, alterations in ionchannel proteins produce either myotonia(muscle stiffness due to excessive electri-cal excitability) or paralysis (due to insuf-ficient muscle excitability). Other disor-ders arise from ion channel defects inheart, kidney, and the inner ear.Channelopathies associated with ionchannels localized in brain are muchmore difficult to study. Nonetheless,voltage-gated Ca2+channels have re-cently been implicated in a range of neu-rological diseases. These include episodicataxia, spinocerebellar degeneration,night blindness, and migraine head-aches. Familial hemiplegic migraine (FHM)is characterized by migraine attacks thattypically last one to three days. Duringsuch episodes, patients experience severeheadaches and vomiting. Several muta-tions in a human Ca2+channel have beenidentified in families with FHM, eachhaving different clinical symptoms. Forexample, a mutation in the pore-formingregion of the channel produces hemi-plegic migraine with progressive cerebel-lar ataxia, whereas other mutations causeonly the usual FHM symptoms. Howthese altered Ca2+channel propertieslead to migraine attacks is not known.Episodic ataxia type 2 (EA2) is a neuro-logical disorder in which affected indi-viduals suffer recurrent attacks of abnor-mal limb movements and severe ataxia.These problems are sometimes accompa-NNNCCC(B) Na+CHANNELNC(A) Ca2+CHANNEL(C) K+CHANNEL (D) Cl−CHANNELCNCFHMEA2CSNBGEFSMyotoniaParalysisEA1BFNC MyotoniaParalysisβ subunitI II III IVI II III IVBox DDiseases Caused by Altered Ion ChannelsGenetic mutations in (A) Ca2+channels, (B)Na+channels, (C) K+channels, and (D) Cl–channels that result in diseases. Red regionsindicate the sites of these mutations; the redcircles indicate mutations. (After Lehmann-Horn and Jurkat-Kott, 1999.)Purves04 5/13/04 1:41 PM Page 84
    • In short, ion channels are integral membrane proteins with characteristicfeatures that allow them to assemble into multimolecular aggregates. Collec-tively, these structures allow channels to conduct ions, sense the transmem-brane potential, to inactivate, and to bind to various neurotoxins. A combi-nation of physiological, molecular biological and crystallographic studieshas begun to provide a detailed physical picture of K+channels. This workhas now provided considerable insight into how ions are conducted fromone side of the plasma membrane to the other, how a channel can be selec-tively permeable to a single type of ion, how they are able to sense changesin membrane voltage, and how they gate the opening of their pores. It islikely that other types of ion channels will be similar in their functionalarchitecture. Finally, this sort of work has illuminated how mutations in ionchannel genes can lead to a variety of neurological disorders (Box D).Channels and Transporters 85nied by vertigo, nausea, and headache.Usually, attacks are precipitated by emo-tional stress, exercise, or alcohol and lastfor a few hours. The mutations in EA2cause Ca2+channels to be truncated atvarious sites, which may cause the clini-cal manifestations of the disease by pre-venting the normal assembly of Ca2+channels in the membrane.X-linked congenital stationary nightblindness (CSNB) is a recessive retinal dis-order that causes night blindness,decreased visual acuity, myopia, nystag-mus, and strabismus. Complete CSNBcauses retinal rod photoreceptors to benonfunctional. Incomplete CSNB causessubnormal (but measurable) functioningof both rod and cone photoreceptors.Like EA2, the incomplete type of CSNBis caused by mutations producing trun-cated Ca2+channels. Abnormal retinalfunction may arise from decreased Ca2+currents and neurotransmitter releasefrom photoreceptors (see Chapter 11).A defect in brain Na+channels causesgeneralized epilepsy with febrile seizures(GEFS) that begins in infancy and usu-ally continues through early puberty.This defect has been mapped to twomutations: one on chromosome 2 thatencodes an α subunit for a voltage-gatedNa+channel, and the other on chromo-some 19 that encodes a Na+channel βsubunit. These mutations cause a slow-ing of Na+channel inactivation (see fig-ure above), which may explain the neu-ronal hyperexcitability underlying GEFS.Another type of seizure, benign famil-ial neonatal convulsion (BFNC), is due toK+channel mutations. This disease ischaracterized by frequent brief seizurescommencing within the first week of lifeand disappearing spontaneously withina few months. The mutation has beenmapped to at least two voltage-gated K+channel genes. A reduction in K+currentflow through the mutated channels prob-ably accounts for the hyperexcitabilityassociated with this defect. A related dis-ease, episodic ataxia type 1 (EA1), hasbeen linked to a defect in another type ofvoltage-gated K+channel. EA1 is charac-terized by brief episodes of ataxia. Mu-tant channels inhibit the function ofother, non-mutant K+channels and mayproduce clinical symptoms by impairingaction potential repolarization. Muta-tions in the K+channels of cardiac mus-cle are responsible for the irregular heart-beat of patients with long Q-T syndrome.Numerous genetic disorders affect thevoltage-gated channels of skeletal mus-cle and are responsible for a host of mus-cle diseases that either cause muscleweakness (paralysis) or muscle contrac-tion (myotonia).ReferencesBARCHI, R. L. (1995) Molecular pathology ofthe skeletal muscle sodium channel. Ann.Rev. Physiol. 57: 355–385.BERKOVIC, S. F. AND I. E. SCHEFFER (1997) Epi-lepsies with single gene inheritance. BrainDevelop. 19 :13–28.COOPER, E. C. AND L. Y. JAN (1999) Ion chan-nel genes and human neurological disease:Recent progress, prospects, and challenges.Proc. Natl. Acad. Sci. USA 96: 4759–4766.DAVIES, N. P. AND M. G. HANNA (1999) Neuro-logical channelopathies: Diagnosis and ther-apy in the new millennium. Ann. Med. 31:406–420.JEN, J. (1999) Calcium channelopathies in thecentral nervous system. Curr. Op. Neurobiol.9: 274–280.LEHMANN-HORN, F. AND K. JURKAT-ROTT(1999) Voltage-gated ion channels and hered-itary disease. Physiol. Rev. 79: 1317–1372.OPHOFF, R. A., G. M. TERWINDT, R. R. FRANTSAND M. D. FERRARI (1998) P/Q-type Ca2+channel defects in migraine, ataxia andepilepsy. Trends Pharm. Sci. 19: 121–127.Mutations in Na+channels slow the rate ofinactivation of Na+currents. (After Barchi,1995.)Na+current(nA)Wild typeNa+channelmutants0 5 10Time (ms)−80−40040Membranepotential(mV)Purves04 5/13/04 1:41 PM Page 85
    • 86 Chapter FourActive Transporters Create and Maintain Ion GradientsUp to this point, the discussion of the molecular basis of electrical signalinghas taken for granted the fact that nerve cells maintain ion concentrationgradients across their surface membranes. However, none of the ions ofphysiological importance (Na+, K+, Cl–, and Ca2+) are in electrochemicalequilibrium. Because channels produce electrical effects by allowing one ormore of these ions to diffuse down their electrochemical gradients, therewould be a gradual dissipation of these concentration gradients unless nervecells could restore ions displaced during the current flow that occurs as aresult of both neural signaling and the continual ionic leakage that occurs atrest. The work of generating and maintaining ionic concentration gradientsfor particular ions is carried out by a group of plasma membrane proteinsknown as active transporters.Active transporters carry out this task by forming complexes with theions that they are translocating. The process of ion binding and unbindingfor transport typically requires several milliseconds. As a result, ion translo-cation by active transporters is much slower than ion movement throughchannels: Recall that ion channels can conduct thousands of ions across amembrane each millisecond. In short, active transporters gradually storeenergy in the form of ion concentration gradients, whereas the opening ofion channels rapidly dissipates this stored energy during relatively briefelectrical signaling events.Several types of active transporter have now been identified (Figure 4.10).Although the specific jobs of these transporters differ, all must translocateions against their electrochemical gradients. Moving ions uphill requires theconsumption of energy, and neuronal transporters fall into two classes basedon their energy sources. Some transporters acquire energy directly from thehydrolysis of ATP and are called ATPase pumps (Figure 4.10, left). The mostprominent example of an ATPase pump is the Na+pump (or, more properly,the Na+/K+ATPase pump), which is responsible for maintaining transmem-brane concentration gradients for both Na+and K+(Figure 4.10A). Anotheris the Ca2+pump, which provides one of the main mechanisms for removingCa2+from cells (Figure 4.10B). The second class of active transporter does notuse ATP directly, but depends instead on the electrochemical gradients ofother ions as an energy source. This type of transporter carries one or moreions up its electrochemical gradient while simultaneously taking another ion(most often Na+) down its gradient. Because at least two species of ions areATPase PUMPS ION EXCHANGERS(A) Na+/K+ pumpInsideOutside(B) Ca2+ pump (C) Na+/Ca2+exchanger(E) Na+/H+exchanger(D) Cl−/HCO3−exchangerATPADPCl−Na+Na+Ca2+HCO3−H+Na+K+ATPADP(F) Na+/neurotransmittertransporterNa+H+Ca2+GABA,DopamineFigure 4.10 Examples of ion trans-porters found in cell membranes. (A,B)Some transporters are powered by thehydrolysis of ATP (ATPase pumps),whereas others (C–F) use the electro-chemical gradients of co-transportedions as a source of energy (ion ex-changers).Purves04 5/13/04 1:41 PM Page 86
    • involved in such transactions, these transporters are usually called ionexchangers (Figure 4.10, right). An example of such a transporter is theNa+/Ca2+exchanger, which shares with the Ca2+pump the important job ofkeeping intracellular Ca2+concentrations low (Figure 4.10C). Anotherexchanger in this category regulates both intracellular Cl–concentration andpH by swapping intracellular Cl–for another extracellular anion, bicarbon-ate (Figure 4.10D). Other ion exchangers, such as the Na+/H+exchanger(Figure 4.10E), also regulate intracellular pH, in this case by acting directlyon the concentration of H+. Yet other ion exchangers are involved in trans-porting neurotransmitters into synaptic terminals (Figure 4.10F), asdescribed in Chapter 6. Although the electrochemical gradient of Na+(orother counter ions) is the proximate source of energy for ion exchangers,these gradients ultimately depend on the hydrolysis of ATP by ATPasepumps, such as the Na+/K+ATPase pump.Functional Properties of the Na+/K+PumpOf these various transporters, the best understood is the Na+/K+pump. Theactivity of this pump is estimated to account for 20–40% of the brain’senergy consumption, indicating its importance for brain function. The Na+pump was first discovered in neurons in the 1950s, when Richard Keynes atCambridge University used radioactive Na+to demonstrate the energy-dependent efflux of Na+from squid giant axons. Keynes and his collabora-tors found that this efflux ceased when the supply of ATP in the axon wasinterrupted by treatment with metabolic poisons (Figure 4.11A, point 4).Other conditions that lower intracellular ATP also prevent Na+efflux. Theseexperiments showed that removing intracellular Na+requires cellularmetabolism. Further studies with radioactive K+demonstrated that Na+efflux is associated with simultaneous, ATP-dependent influx of K+. Theseopposing fluxes of Na+and K+are operationally inseparable: Removal ofexternal K+greatly reduces Na+efflux (Figure 4.11, point 2), and vice versa.These energy-dependent movements of Na+and K+implicated an ATP-hydrolyzing Na+/K+pump in the generation of the transmembrane gradi-ents of both Na+and K+. The exact mechanism responsible for these fluxesof Na+and K+is still not entirely clear, but the pump is thought to alter-nately shuttle these ions across the membranes in a cycle fueled by the trans-fer of a phosphate group from ATP to the pump protein (Figure 4.11B).Additional quantitative studies of the movements of Na+and K+indicatethat the two ions are not pumped at identical rates: The K+influx is onlyabout two-thirds the Na+efflux. Thus, the pump apparently transports twoK+into the cell for every three Na+that are removed (see Figure 4.11B). Thisstoichiometry causes a net loss of one positively charged ion from inside ofthe cell during each round of pumping, meaning that the pump generates anelectrical current that can hyperpolarize the membrane potential. For thisreason, the Na+/K+pump is said to be electrogenic. Because pumps actmuch more slowly than ion channels, the current produced by the Na+/K+pump is quite small. For example, in the squid axon, the net current gener-ated by the pump is less than 1% of the current flowing through voltage-gated Na+channels and affects the resting membrane potential by only amillivolt or less.Although the electrical current generated by the activity of the Na+/K+pump is small, under special circumstances the pump can significantlyinfluence the membrane potential. For instance, prolonged stimulation ofChannels and Transporters 87Purves04 5/13/04 1:41 PM Page 87
    • 88 Chapter Four0 50 100Time (min)Na+efflux(logarithmicscale)(A)(B)150 200 250 300ATPADPDephosphorylation-induced conformationalchange leads to K+releaseConformational changecauses Na+release and K+binding2. Phosphorylation3.1. Na+binding1 Efflux of Na+ 3 Recovery whenK+is restored5 Recovery whenATP is restored2 Na+efflux reducedby removal ofexternal K+4 Efflux decreased by metabolicinhibitors, such as dinitrophenol,which block ATP synthesis4.OutsideInsideNa+K+K+Na+PiPiPiFigure 4.11 Ionic movements due tothe Na+/K+pump. (A) Measurement ofradioactive Na+efflux from a squidgiant axon. This efflux depends onexternal K+and intracellular ATP. (B) Amodel for the movement of ions by theNa+/K+pump. Uphill movements ofNa+and K+are driven by ATP, whichphosphorylates the pump. These fluxesare asymmetrical, with three Na+car-ried out for every two K+brought in. (Aafter Hodgkin and Keynes, 1955; B afterLingrel et al., 1994.)Purves04 5/13/04 1:41 PM Page 88
    • small unmyelinated axons produces a substantial hyperpolarization (Figure4.12). During the period of stimulation, Na+enters through voltage-gatedchannels and accumulates within the axons. As the pump removes this extraNa+, the resulting current generates a long-lasting hyperpolarization. Sup-port for this interpretation comes from the observation that conditions thatblock the Na+/K+pump—for example, treatment with ouabain, a plant gly-coside that specifically inhibits the pump—prevent the hyperpolarization.The electrical contribution of the Na+/K+pump is particularly significant inthese small-diameter axons because their large surface-to-volume ratiocauses intracellular Na+concentration to rise to higher levels than it wouldin other cells. Nonetheless, it is important to emphasize that, in most cir-cumstances, the Na+/K+pump plays no part in generating the action poten-tial and has very little direct effect on the resting potential.The Molecular Structure of the Na+/K+PumpThese observations imply that the Na+and K+pump must exhibit severalmolecular properties: (1) It must bind both Na+and K+; (2) it must possesssites that bind ATP and receive a phosphate group from this ATP; and (3) itmust bind ouabain, the toxin that blocks this pump (Figure 4.13A). A varietyof studies have now identified the aspects of the protein that account forthese properties of the Na+/K+pump. This pump is a large, integral mem-brane protein made up of at least two subunits, called α and β. The primarysequence shows that the α subunit spans the membrane 10 times, with mostof the molecule found on the cytoplasmic side, whereas the β subunit spansthe membrane once and is predominantly extracellular. Although a detailedaccount of the functional domains of the Na+/K+pump is not yet available,some parts of the amino acid sequence have identified functions (Figure4.13B). One intracellular domain of the protein is required for ATP bindingChannels and Transporters 8910Time (min)2005mVMembranepotential(mV)−60−80−20−40−60−800Individualaction potentialsTrains of action potentialsTime (s)OuabainPoststimulushyperpolarizationblockedOuabainbinding siteNNCCPhosphorylationsiteATPADP + PiMembraneMembraneOuabainbindingsiteOutside2 K+3 Na+InsideOutside(A) (B)InsideMembraneATPbindingsiteMembraneMembraneMembraneNa+and K+bindingβ subunitα subunitFigure 4.12 The electrogenic transport of ions by the Na+/K+pump can influencemembrane potential. Measurements of the membrane potential of a small unmyeli-nated axon show that a train of action potentials is followed by a long-lastinghyperpolarization. This hyperpolarization is blocked by ouabain, indicating that itresults from the activity of the Na+/K+pump. (After Rang and Ritchie, 1968.)Figure 4.13 Molecular structure of theNa+/K+pump. (A) General features ofthe pump. (B) The molecule spans themembrane 10 times. Amino acidresidues thought to be important forbinding of ATP, K+, and ouabain arehighlighted. (After Lingrel et al., 1994.)Purves04 5/13/04 1:41 PM Page 89
    • 90 Chapter Fourand hydrolysis, and the amino acid phosphorylated by ATP has been identi-fied. Another extracellular domain may represent the binding site forouabain. However, the sites involved in the most critical function of thepump—the movement of Na+and K+—have not yet been defined. Nonethe-less, altering certain membrane-spanning domains (red in Figure 4.13B)impairs ion translocation; moreover, kinetic studies indicate that both ionsbind to the pump at the same site. Because these ions move across the mem-brane, it is likely that this site traverses the plasma membrane; it is alsolikely that the site has a negative charge, since both Na+and K+are posi-tively charged. The observation that removing negatively charged residuesin a membrane-spanning domain of the protein (pale yellow in Figure 4.13B)greatly reduces Na+and K+binding provides at least a hint about the ion-translocating domain of the transporter molecule.SummaryIon transporters and channels have complementary functions. The primarypurpose of transporters is to generate transmembrane concentration gradi-ents, which are then exploited by ion channels to generate electrical signals.Ion channels are responsible for the voltage-dependent conductances ofnerve cell membranes. The channels underlying the action potential are inte-gral membrane proteins that open or close ion-selective pores in response tothe membrane potential, allowing specific ions to diffuse across the mem-brane. The flow of ions through single open channels can be detected as tinyelectrical currents, and the synchronous opening of many such channelsgenerates the macroscopic currents that produce action potentials. Molecularstudies show that such voltage-gated channels have highly conserved struc-tures that are responsible for features such as ion permeation and voltagesensing, as well as the features that specify ion selectivity and toxin sensitiv-ity. Other types of channels are sensitive to chemical signals, such as neuro-transmitters or second messengers, or to heat or membrane deformation. Alarge number of ion channel genes create channels with a correspondinglywide range of functional characteristics, thus allowing different types of neu-rons to have a remarkable spectrum of electrical properties. Ion transporterproteins are quite different in both structure and function. The energyneeded for ion movement against a concentration gradient (e.g., in main-taining the resting potential) is provided either by the hydrolysis of ATP orby the electrochemical gradient of co-transported ions. The Na+/K+pumpproduces and maintains the transmembrane gradients of Na+and K+, whileother transporters are responsible for the electrochemical gradients for otherphysiologically important ions, such as Cl–, Ca2+, and H+. Together, trans-porters and channels provide a reasonably comprehensive molecular expla-nation for the ability of neurons to generate electrical signals.Purves04 5/13/04 1:41 PM Page 90
    • Additional ReadingReviewsARMSTRONG, C. M. AND B. HILLE (1998) Volt-age-gated ion channels and electrical excit-ability. Neuron 20: 371–380.BEZANILLA, F. AND A. M. CORREA (1995) Single-channel properties and gating of Na+and K+channels in the squid giant axon. In Cephalo-pod Neurobiology, N. J. Abbott, R. Williamsonand L. Maddock (eds.). New York: OxfordUniversity Press, pp. 131–151.CATTERALL, W. A. (1988) Structure and func-tion of voltage-sensitive ion channels. Science242: 50–61.ISOM, L. L., K. S. DE JONGH AND W. A. CATTER-ALL (1994) Auxiliary subunits of voltage-gatedion channels. Neuron 12: 1183–1194.JAN, L. Y. AND Y. N. JAN (1997) Voltage-gatedand inwardly rectifying potassium channels.J. Physiol. 505: 267–282.JENTSCH, T. J., T. FRIEDRICH, A. SCHRIEVER ANDH. YAMADA (1999) The CLC chloride channelfamily. Pflügers Archiv. 437: 783–795.KAPLAN, J. H. (2002) Biochemistry of Na,K-ATPase. Annu. Rev. Biochem. 71: 511–535.KRISHTAL, O. (2003). The ASICs: Signalingmolecules? Modulators? Trends Neurosci, 26:477–483.LINGREL, J. B., J. VAN HUYSSE, W. O’BRIEN, E.JEWELL-MOTZ, R. ASKEW AND P. SCHULTHEIS(1994) Structure-function studies of the Na, K-ATPase. Kidney Internat. 45: S32–S39.MACKINNON, R. (2003) Potassium channels.FEBS Lett. 555: 62–65.NEHER, E. (1992) Nobel lecture: Ion channelsfor communication between and within cells.Neuron 8: 605–612.PATAPOUTIAN, A., A. M. PEIER, G. M. STORY ANDV. VISWANATH (2003). ThermoTRP channelsand beyond: Mechanisms of temperature sen-sation. Nat. Rev. Neurosci. 4: 529–539.SEEBURG, P. H. (2002). A-to-I editing: New andold sites, functions and speculations. Neuron35: 17–20.SKOU, J. C. (1988) Overview: The Na,K pump.Meth. Enzymol. 156: 1–25.Important Original PapersANTZ, C. AND 7 OTHERS (1997) NMR structureof inactivation gates from mammalian volt-age-dependent potassium channels. Nature385: 272–275.BEZANILLA, F., E. PEROZO, D. M. PAPAZIAN ANDE. STEFANI (1991) Molecular basis of gatingcharge immobilization in Shaker potassiumchannels. Science 254: 679–683.BOULTER, J. AND 6 OTHERS (1990) Molecularcloning and functional expression of gluta-mate receptor subunit genes. Science 249:1033–1037.CATERINA, M. J., M. A. SCHUMACHER, M. TOMI-NAGA, T. A. ROSEN, J. D. LEVINE AND D. JULIUS(1997) The capsaicin receptor: A heat-activatedion channel in the pain pathway. Nature 389:816–824.CHA, A., G. E. SNYDER, P. R. SELVIN AND F.BEZANILLA (1999) Atomic scale movement ofthe voltage-sensing region in a potassiumchannel measured via spectroscopy. Nature402: 809–813.DOYLE, D. A. AND 7 OTHERS (1998) The struc-ture of the potassium channel: Molecularbasis of K+conduction and selectivity. Science280: 69–77.FAHLKE, C., H. T. YU, C. L. BECK, T. H. RHODESAND A. L. GEORGE JR. (1997) Pore-forming seg-ments in voltage-gated chloride channels.Nature 390: 529–532.HO, K. AND 6 OTHERS (1993) Cloning andexpression of an inwardly rectifying ATP-reg-ulated potassium channel. Nature 362: 31–38.HODGKIN, A. L. AND R. D. KEYNES (1955)Active transport of cations in giant axonsfrom Sepia and Loligo. J. Physiol. 128: 28–60.HOSHI, T., W. N. ZAGOTTA AND R. W. ALDRICH(1990) Biophysical and molecular mechanismsof Shaker potassium channel inactivation. Sci-ence 250: 533–538.JIANG, Y. AND 6 OTHERS (2003) X-ray structureof a voltage-dependent K+channel. Nature423: 33–41.LLANO, I., C. K. WEBB AND F. BEZANILLA (1988)Potassium conductance of squid giant axon.Single-channel studies. J. Gen. Physiol. 92:179–196.MIKAMI, A. AND 7 OTHERS (1989) Primary struc-ture and functional expression of the cardiacdihydropyridine-sensitive calcium channel.Nature 340: 230–233.NODA, M. AND 6 OTHERS (1986) Expression offunctional sodium channels from clonedcDNA. Nature 322: 826–828.NOWYCKY, M. C., A. P. FOX AND R. W. TSIEN(1985) Three types of neuronal calcium chan-nel with different calcium agonist sensitivity.Nature 316: 440–443.PAPAZIAN, D. M., T. L. SCHWARZ, B. L. TEMPEL,Y. N. JAN AND L. Y. JAN (1987) Cloning ofgenomic and complementary DNA fromShaker, a putative potassium channel genefrom Drosophila. Science 237: 749–753.RANG, H. P. AND J. M. RITCHIE (1968) On theelectrogenic sodium pump in mammaliannon-myelinated nerve fibres and its activationby various external cations. J. Physiol. 196:183–221.SIGWORTH, F. J. AND E. NEHER (1980) Single Na+channel currents observed in cultured ratmuscle cells. Nature 287: 447–449.THOMAS, R. C. (1969) Membrane current andintracellular sodium changes in a snail neu-rone during extrusion of injected sodium. J.Physiol. 201: 495–514.TOYOSHIMA, C., M. NAKASAKO, H. NOMURAAND H. OGAWA (2000) Crystal structure of thecalcium pump of sarcoplasmic reticulum at2.6 Å resolution. Nature 405: 647–655.VANDERBERG, C. A. AND F. BEZANILLA (1991) Asodium channel model based on single chan-nel, macroscopic ionic, and gating currents inthe squid giant axon. Biophys. J. 60:1511–1533.WALDMANN, R., G. CHAMPIGNY, F. BASSILANA,C. HEURTEAUX AND M. LAZDUNSKI (1997) Aproton-gated cation channel involved in acid-sensing. Nature 386: 173–177.WEI, A. M., A. COVARRUBIAS, A. BUTLER, K.BAKER, M. PAK AND L. SALKOFF (1990) K+cur-rent diversity is produced by an extendedgene family conserved in Drosophila andmouse. Science 248: 599–603.YANG, N., A. L. GEORGE JR. AND R. HORN (1996)Molecular basis of charge movement in volt-age-gated sodium channels. Neuron 16:113–22.BooksAIDLEY, D. J. AND P. R. STANFIELD (1996) IonChannels: Molecules in Action. Cambridge:Cambridge University Press.ASHCROFT, F. M. (2000) Ion Channels and Dis-ease. Boston: Academic Press.HILLE, B. (2001) Ion Channels of Excitable Mem-branes, 3rd Ed. Sunderland, MA: SinauerAssociates.JUNGE, D. (1992) Nerve and Muscle Excitation,3rd Ed. Sunderland, MA: Sinauer Associates.NICHOLLS, D. G. (1994) Proteins, Transmittersand Synapses. Oxford: Blackwell ScientificSIEGEL, G. J., B. W. AGRANOFF, R. W. ALBERS, S.K. FISHER AND M. D. UHLER (1999) Basic Neuro-chemistry. Philadelphia: Lippincott-Raven.Channels and Transporters 91Purves04 5/13/04 1:41 PM Page 91
    • Purves04 5/13/04 1:41 PM Page 92
    • OverviewThe human brain contains at least 100 billion neurons, each with the abilityto influence many other cells. Clearly, sophisticated and highly efficientmechanisms are needed to enable communication among this astronomicalnumber of elements. Such communication is made possible by synapses, thefunctional contacts between neurons. Two different types of synapse—elec-trical and chemical—can be distinguished on the basis of their mechanism oftransmission. At electrical synapses, current flows through gap junctions,which are specialized membrane channels that connect two cells. In contrast,chemical synapses enable cell-to-cell communication via the secretion ofneurotransmitters; these chemical agents released by the presynaptic neu-rons produce secondary current flow in postsynaptic neurons by activatingspecific receptor molecules. The total number of neurotransmitters is notknown, but is well over 100. Virtually all neurotransmitters undergo a simi-lar cycle of use: synthesis and packaging into synaptic vesicles; release fromthe presynaptic cell; binding to postsynaptic receptors; and, finally, rapidremoval and/or degradation. The secretion of neurotransmitters is triggeredby the influx of Ca2+through voltage-gated channels, which gives rise to atransient increase in Ca2+concentration within the presynaptic terminal. Therise in Ca2+concentration causes synaptic vesicles to fuse with the presynap-tic plasma membrane and release their contents into the space between thepre- and postsynaptic cells. Although it is not yet understood exactly howCa2+triggers exocytosis, specific proteins on the surface of the synaptic vesi-cle and elsewhere in the presynaptic terminal mediate this process. Neuro-transmitters evoke postsynaptic electrical responses by binding to membersof a diverse group of neurotransmitter receptors. There are two major classesof receptors: those in which the receptor molecule is also an ion channel, andthose in which the receptor and ion channel are separate molecules. Thesereceptors give rise to electrical signals by transmitter-induced opening orclosing of the ion channels. Whether the postsynaptic actions of a particularneurotransmitter are excitatory or inhibitory is determined by the ionic per-meability of the ion channel affected by the transmitter, and by the concen-tration of permeant ions inside and outside the cell.Electrical SynapsesAlthough there are many kinds of synapses within the human brain, theycan be divided into two general classes: electrical synapses and chemicalsynapses. Although they are a distinct minority, electrical synapses arefound in all nervous systems, permitting direct, passive flow of electricalcurrent from one neuron to another.Chapter 593SynapticTransmissionPurves05 5/13/04 2:27 PM Page 93
    • 94 Chapter FiveThe structure of an electrical synapse is shown schematically in Figure5.1A. The “upstream” neuron, which is the source of current, is called thepresynaptic element, and the “downstream” neuron into which this currentflows is termed postsynaptic. The membranes of the two communicatingneurons come extremely close at the synapse and are actually linkedtogether by an intercellular specialization called a gap junction. Gap junc-tions contain precisely aligned, paired channels in the membrane of the pre-and postsynaptic neurons, such that each channel pair forms a pore (see Fig-ure 5.2A). The pore of a gap junction channel is much larger than the poresof the voltage-gated ion channels described in the previous chapter. As aresult, a variety of substances can simply diffuse between the cytoplasm ofthe pre- and postsynaptic neurons. In addition to ions, substances that dif-fuse through gap junction pores include molecules with molecular weightsas great as several hundred daltons. This permits ATP and other importantintracellular metabolites, such as second messengers (see Chapter 7), to betransferred between neurons.Electrical synapses thus work by allowing ionic current to flow passivelythrough the gap junction pores from one neuron to another. The usualsource of this current is the potential difference generated locally by theaction potential (see Chapter 3). This arrangement has a number of interest-ing consequences. One is that transmission can be bidirectional; that is, cur-rent can flow in either direction across the gap junction, depending on whichmember of the coupled pair is invaded by an action potential (althoughCytoplasmMitochondrionMicrotubulePresynapticneuronPresynapticneuronPostsynapticneurotransmitterreceptorGap junction channelsGapjunctionPresynapticmembranePostsynapticmembraneSynaptic vesiclePresynaptic membranePostsynapticmembraneSynapticcleftSynapticvesicle fusingPostsynapticneuronPostsynapticneuronIons flow throughgap junction channelsNeurotransmitter releasedIons flow throughpostsynaptic channelsP N i 3E(A) ELECTRONIC SYNAPSE (B) CHEMICAL SYNAPSEFigure 5.1 Electrical and chemical syn-apses differ fundamentally in theirtransmission mechanisms. (A) At electri-cal synapses, gap junctions between pre-and postsynaptic membranes permitcurrent to flow passively through inter-cellular channels (blowup). This currentflow changes the postsynaptic mem-brane potential, initiating (or in someinstances inhibiting) the generation ofpostsynaptic action potentials. (B) Atchemical synapses, there is no intercel-lular continuity, and thus no direct flowof current from pre- to postsynaptic cell.Synaptic current flows across the post-synaptic membrane only in response tothe secretion of neurotransmitters,which open or close postsynaptic ionchannels after binding to receptor mole-cules (blowup).Purves05 5/13/04 2:27 PM Page 94
    • some types of gap junctions have special features that render their transmis-sion unidirectional). Another important feature of the electrical synapse isthat transmission is extraordinarily fast: because passive current flow acrossthe gap junction is virtually instantaneous, communication can occur with-out the delay that is characteristic of chemical synapses.These features are apparent in the operation of the first electrical synapseto be discovered, which resides in the crayfish nervous system. A postsynap-tic electrical signal is observed at this synapse within a fraction of a millisec-ond after the generation of a presynaptic action potential (Figure 5.2). In fact,at least part of this brief synaptic delay is caused by propagation of theaction potential into the presynaptic terminal, so that there may be essen-tially no delay at all in the transmission of electrical signals across the syn-apse. Such synapses interconnect many of the neurons within the circuit thatallows the crayfish to escape from its predators, thus minimizing the timebetween the presence of a threatening stimulus and a potentially life-savingmotor response.A more general purpose of electrical synapses is to synchronize electricalactivity among populations of neurons. For example, the brainstem neuronsthat generate rhythmic electrical activity underlying breathing are synchro-nized by electrical synapses, as are populations of interneurons in the cere-bral cortex, thalamus, cerebellum, and other brain regions. Electrical trans-mission between certain hormone-secreting neurons within the mammalianhypothalamus ensures that all cells fire action potentials at about the sametime, thus facilitating a burst of hormone secretion into the circulation. Thefact that gap junction pores are large enough to allow molecules such as ATPand second messengers to diffuse intercellularly also permits electrical syn-apses to coordinate the intracellular signaling and metabolism of coupledcells. This property may be particularly important for glial cells, which formlarge intracellular signaling networks via their gap junctions.Synaptic Transmission 95(A) (B)Membranepotential(mV)0 1Time (ms)32 4PostsynapticneuronPresynapticneuronConnexonsPresynapticcell membranePores connectingcytoplasm of twoneuronsPostsynapticcell membrane20 nm20 nm20 nm3.5 nm3.5 nm3.5 nm0−2525−50025−50−25Brief (~0.1 ms)synaptic delayFigure 5.2 Structure and function ofgap junctions at electrical synapses. (A)Gap junctions consist of hexameric com-plexes formed by the coming together ofsubunits called connexons, which arepresent in both the pre- and postsynap-tic membranes. The pores of the chan-nels connect to one another, creatingelectrical continuity between the twocells. (B) Rapid transmission of signalsat an electrical synapse in the crayfish.An action potential in the presynapticneuron causes the postsynaptic neuronto be depolarized within a fraction of amillisecond. (B after Furshpan and Pot-ter, 1959.)Purves05 5/13/04 2:27 PM Page 95
    • 96 Chapter FiveSignal Transmission at Chemical SynapsesThe general structure of a chemical synapse is shown schematically in Figure5.1B. The space between the pre- and postsynaptic neurons is substantiallygreater at chemical synapses than at electrical synapses and is called the syn-aptic cleft. However, the key feature of all chemical synapses is the presenceof small, membrane-bounded organelles called synaptic vesicles within thepresynaptic terminal. These spherical organelles are filled with one or moreneurotransmitters, the chemical signals secreted from the presynaptic neu-ron, and it is these chemical agents acting as messengers between the com-municating neurons that gives this type of synapse its name.Transmission at chemical synapses is based on the elaborate sequence ofevents depicted in Figure 5.3. The process is initiated when an action poten-tial invades the terminal of the presynaptic neuron. The change in mem-brane potential caused by the arrival of the action potential leads to theopening of voltage-gated calcium channels in the presynaptic membrane.Because of the steep concentration gradient of Ca2+across the presynapticmembrane (the external Ca2+concentration is approximately 10–3M, where-as the internal Ca2+concentration is approximately 10–7M), the opening ofthese channels causes a rapid influx of Ca2+into the presynaptic terminal,with the result that the Ca2+concentration of the cytoplasm in the terminaltransiently rises to a much higher value. Elevation of the presynaptic Ca2+concentration, in turn, allows synaptic vesicles to fuse with the plasma mem-brane of the presynaptic neuron. The Ca2+-dependent fusion of synapticvesicles with the terminal membrane causes their contents, most importantlyneurotransmitters, to be released into the synaptic cleft.Following exocytosis, transmitters diffuse across the synaptic cleft andbind to specific receptors on the membrane of the postsynaptic neuron. Thebinding of neurotransmitter to the receptors causes channels in the postsyn-aptic membrane to open (or sometimes to close), thus changing the ability ofions to flow into (or out of) the postsynaptic cells. The resulting neurotrans-mitter-induced current flow alters the conductance and (usually) the mem-brane potential of the postsynaptic neuron, increasing or decreasing theprobability that the neuron will fire an action potential. In this way, informa-tion is transmitted from one neuron to another.Properties of NeurotransmittersThe notion that electrical information can be transferred from one neuron tothe next by means of chemical signaling was the subject of intense debatethrough the first half of the twentieth century. A key experiment that sup-ported this idea was performed in 1926 by German physiologist Otto Loewi.Acting on an idea that allegedly came to him in the middle of the night,Loewi proved that electrical stimulation of the vagus nerve slows the heart-beat by releasing a chemical signal. He isolated and perfused the hearts oftwo frogs, monitoring the rates at which they were beating (Figure 5.4). Hisexperiment collected the perfusate flowing through the stimulated heart andtransferred this solution to the second heart. When the vagus nerve to thefirst heart was stimulated, the beat of this heart slowed. Remarkably, eventhough the vagus nerve of the second heart had not been stimulated, its beatalso slowed when exposed to the perfusate from the first heart. This resultshowed that the vagus nerve regulates the heart rate by releasing a chemicalthat accumulates in the perfusate. Originally referred to as “vagus sub-stance,” the agent was later shown to be acetylcholine (ACh). ACh is nowknown to be a neurotransmitter that acts not only in the heart but at a vari-Purves05 5/13/04 2:27 PM Page 96
    • ety of postsynaptic targets in the central and peripheral nervous systems,preeminently at the neuromuscular junction of striated muscles and in thevisceral motor system (see Chapters 6 and 20).Over the years, a number of formal criteria have emerged that definitivelyidentify a substance as a neurotransmitter (Box A). These have led to theidentification of more than 100 different neurotransmitters, which can beSynaptic Transmission 97AcrossdendriteMyelinTransmitterreceptorIonsPostsynapticcurrent flowSynapticvesicle92TransmittermoleculesCa2+Opening or closing ofpostsynaptic channels8Transmitter binds toreceptor molecules inpostsynaptic membrane7Influx of Ca2+through channels46Transmitter is synthesizedand then stored in vesicles1Transmitter is releasedinto synaptic cleft viaexocytosisRetrieval of vesicularmembrane from plasmamembrane10Ca2+causes vesicles to fusewith presynaptic membrane5Postsynaptic current causesexcitatory or inhibitorypostsynaptic potential thatchanges the excitability ofthe postsynaptic cellDepolarization of presynapticterminal causes opening ofvoltage-gated Ca2+channels3An action potential invadesthe presynaptic terminalTransmittermoleculesFigure 5.3 Sequence of eventsinvolved in transmission at a typicalchemical synapse.Purves05 5/13/04 2:27 PM Page 97
    • 98 Chapter Fiveclassified into two broad categories: small-molecule neurotransmitters andneuropeptides (Chapter 6). Having more than one transmitter diversifies thephysiological repertoire of synapses. Multiple neurotransmitters can pro-duce different types of responses on individual postsynaptic cells. For exam-ple, a neuron can be excited by one type of neurotransmitter and inhibitedby another type of neurotransmitter. The speed of postsynaptic responsesproduced by different transmitters also differs, allowing control of electricalsignaling over different time scales. In general, small-molecule neurotrans-mitters mediate rapid synaptic actions, whereas neuropeptides tend to mod-ulate slower, ongoing synaptic functions.Until relatively recently, it was believed that a given neuron producedonly a single type of neurotransmitter. It is now clear, however, that manytypes of neurons synthesize and release two or more different neurotrans-mitters. When more than one transmitter is present within a nerve terminal,the molecules are called co-transmitters. Because different types of transmit-ters can be packaged in different populations of synaptic vesicles, co-trans-mitters need not be released simultaneously. When peptide and small-mole-cule neurotransmitters act as co-transmitters at the same synapse, they aredifferentially released according to the pattern of synaptic activity: low-fre-quency activity often releases only small neurotransmitters, whereas high-frequency activity is required to release neuropeptides from the same pre-synaptic terminals. As a result, the chemical signaling properties of suchsynapses change according to the rate of activity.Effective synaptic transmission requires close control of the concentrationof neurotransmitters within the synaptic cleft. Neurons have therefore devel-oped a sophisticated ability to regulate the synthesis, packaging, release, andHeart 1Heart 2VagusnerveStimulate vagusnerve of heart 1(A) (B)ContractionforceContractionforceInhibitory effectof vagus transferredHeartbeatslowedHeart 1Heart 2Time (s)Time (s)StimulateSolutiontransferredto heart 2Figure 5.4 Loewi’s experiment demonstrating chemical neurotransmission. (A)Diagram of experimental setup. (B) Where the vagus nerve of an isolated frog’sheart was stimulated, the heart rate decreased (upper panel). If the perfusion fluidfrom the stimulated heart was transferred to a second heart, its rate decreased aswell (lower panel).Purves05 5/13/04 2:27 PM Page 98
    • Synaptic Transmission 99Box ACriteria That Define a NeurotransmitterThree primary criteria have been used toconfirm that a molecule acts as a neuro-transmitter at a given chemical synapse.1. The substance must be present withinthe presynaptic neuron. Clearly, a chemicalcannot be secreted from a presynapticneuron unless it is present there. Becauseelaborate biochemical pathways arerequired to produce neurotransmitters,showing that the enzymes and precur-sors required to synthesize the substanceare present in presynaptic neurons pro-vides additional evidence that the sub-stance is used as a transmitter. Note,however, that since the transmitters glu-tamate, glycine, and aspartate are alsoneeded for protein synthesis and othermetabolic reactions in all neurons, theirpresence is not sufficient evidence toestablish them as neurotransmitters.2. The substance must be released inresponse to presynaptic depolarization, andthe release must be Ca2+-dependent.Another essential criterion for identify-ing a neurotransmitter is to demonstratethat it is released from the presynapticneuron in response to presynaptic elec-trical activity, and that this releaserequires Ca2+influx into the presynapticterminal. Meeting this criterion is techni-cally challenging, not only because itmay be difficult to selectively stimulatethe presynaptic neurons, but alsobecause enzymes and transporters effi-ciently remove the secreted neurotrans-mitters.3. Specific receptors for the substancemust be present on the postsynaptic cell. Aneurotransmitter cannot act on its targetunless specific receptors for the trans-mitter are present in the postsynapticmembrane. One way to demonstratereceptors is to show that application ofexogenous transmitter mimics the post-synaptic effect of presynaptic stimula-tion. A more rigorous demonstration isto show that agonists and antagoniststhat alter the normal postsynapticresponse have the same effect when thesubstance in question is applied exoge-nously. High-resolution histologicalmethods can also be used to show thatspecific receptors are present in the post-synaptic membrane (by detection ofradioactively labeled receptor antibod-ies, for example).Fulfilling these criteria establishesunambiguously that a substance is usedas a transmitter at a given synapse. Prac-tical difficulties, however, have pre-vented these standards from beingapplied at many types of synapses. It isfor this reason that so many substancesmust be referred to as “putative”neurotransmitters.(1) (2) (3)Postsynaptic cell2 NeurotransmitterreleasedCa2+Ca2+3 Neurotransmitterreceptors activatedApplication oftransmitter, agonists,or antagonistsActionpotentialPresynapticterminal1 Neuro-transmitterpresentDemonstrating the identity of a neurotransmitter at a synapse requires showing (1) its pres-ence, (2) its release, and (3) the postsynaptic presence of specific receptors.Purves05 5/13/04 2:27 PM Page 99
    • 100 Chapter Fivedegradation (or removal) of neurotransmitters to achieve the desired levelsof transmitter molecules. The synthesis of small-molecule neurotransmittersoccurs locally within presynaptic terminals (Figure 5.5A). The enzymesneeded to synthesize these transmitters are produced in the neuronal cellbody and transported to the nerve terminal cytoplasm at 0.5–5 millimeters aday by a mechanism called slow axonal transport. The precursor moleculesrequired to make new molecules of neurotransmitter are usually taken intothe nerve terminal by transporters found in the plasma membrane of the ter-minal. The enzymes synthesize neurotransmitters in the cytoplasm of thepresynaptic terminal and the transmitters are then loaded into synaptic vesi-cles via transporters in the vesicular membrane (see Chapter 4). For somesmall-molecule neurotransmitters, the final steps of synthesis occur insidethe synaptic vesicles. Most small-molecule neurotransmitters are packagedin vesicles 40 to 60 nm in diameter, the centers of which appear clear in elec-tron micrographs; accordingly, these vesicles are referred to as small clear-core vesicles (Figure 5.5B). Neuropeptides are synthesized in the cell body ofa neuron, meaning that the peptide is produced a long distance away fromits site of secretion (Figure 5.5C). To solve this problem, peptide-filled vesi-cles are transported along an axon and down to the synaptic terminal viafast axonal transport. This process carries vesicles at rates up to 400mm/day along cytoskeletal elements called microtubules (in contrast to theslow axonal transport of the enzymes that synthesize small-molecule trans-mitters). Microtubules are long, cylindrical filaments, 25 nm in diameter, pre-sent throughout neurons and other cells. Peptide-containing vesicles aremoved along these microtubule “tracks” by ATP-requiring “motor” proteinssuch as kinesin. Neuropeptides are packaged into synaptic vesicles thatrange from 90 to 250 nm in diameter. These vesicles are electron-dense inelectron micrographs—hence they are referred to as large dense-core vesi-cles (Figure 5.5D).After a neurotransmitter has been secreted into the synaptic cleft, it mustbe removed to enable the postsynaptic cell to engage in another cycle of syn-Figure 5.5 Metabolism of small-molecule and peptide transmitters. (A) Small-mol-ecule neurotransmitters are synthesized at nerve terminals. The enzymes necessaryfor neurotransmitter synthesis are made in the cell body of the presynaptic cell (1)and are transported down the axon by slow axonal transport (2). Precursors aretaken up into the terminals by specific transporters, and neurotransmitter synthesisand packaging take place within the nerve endings (3). After vesicle fusion andrelease (4), the neurotransmitter may be enzymatically degraded. The reuptake ofthe neurotransmitter (or its metabolites) starts another cycle of synthesis, packaging,release, and removal (5). (B) Small clear-core vesicles at a synapse between an axonterminal (AT) and a dendritic spine (Den) in the central nervous system. Such vesi-cles typically contain small-molecule neurotransmitters. (C) Peptide neurotransmit-ters, as well as the enzymes that modify their precursors, are synthesized in the cellbody (1). Enzymes and propeptides are packaged into vesicles in the Golgi appara-tus. During fast axonal transport of these vesicles to the nerve terminals (2), theenzymes modify the propeptides to produce one or more neurotransmitter peptides(3). After vesicle fusion and exocytosis, the peptides diffuse away and are degradedby proteolytic enzymes (4). (D) Large dense-core vesicles in a central axon terminal(AT) synapsing onto a dendrite (Den). Such vesicles typically contain neuropeptidesor, in some cases, biogenic amines. (B and D from Peters, Palay, and Webster, 1991.)▲Purves05 5/13/04 2:27 PM Page 100
    • Enzymes(A)RERNucleusGolgiapparatusMicrotubulesTerminalAxonTransportof precursorsinto terminal1Synthesisof enzymesin cell body2Slow axonaltransportof enzymes4Release anddiffusionof neuro–transmitter Neuro-transmitterNeuro-transmitterDiffusion andDiffusion anddegradationdegradationDiffusion anddegradation(C)1Synthesis ofneurotransmitterprecursors andenzymes2Transport of enzymesand peptideprecursors downmicrotubule tracks3Enzymes modifyprecursors toproduce peptideneurotransmitter4Neurotransmitterdiffuses away andis degraded byproteolytic enzymes(B) (D)53Synthesis andpackaging ofneurotransmitterPrecursorATATDenDenATAT DenDenATDenAT Den0.5 mmPurves05 5/13/04 2:27 PM Page 101
    • 102 Chapter Fiveaptic transmission. The removal of neurotransmitters involves diffusionaway from the postsynaptic receptors, in combination with reuptake intonerve terminals or surrounding glial cells, degradation by specific enzymes,or a combination of these mechanisms. Specific transporter proteins removemost small-molecule neurotransmitters (or their metabolites) from the syn-aptic cleft, ultimately delivering them back to the presynaptic terminal forreuse.Quantal Release of NeurotransmittersMuch of the evidence leading to the present understanding of chemical syn-aptic transmission was obtained from experiments examining the release ofACh at neuromuscular junctions. These synapses between spinal motor neu-rons and skeletal muscle cells are simple, large, and peripherally located,making them particularly amenable to experimental analysis. Such synapsesoccur at specializations called end plates because of the saucer-like appear-ance of the site on the muscle fiber where the presynaptic axon elaborates itsterminals (Figure 5.6A). Most of the pioneering work on neuromusculartransmission was performed by Bernard Katz and his collaborators at Uni-versity College London during the 1950s and 1960s, and Katz has beenwidely recognized for his remarkable contributions to understanding synap-tic transmission. Though he worked primarily on the frog neuromuscularjunction, numerous subsequent experiments have confirmed the applicabil-ity of his observations to transmission at chemical synapses throughout thenervous system.When an intracellular microelectrode is used to record the membranepotential of a muscle cell, an action potential in the presynaptic motor neu-ron can be seen to elicit a transient depolarization of the postsynaptic musclefiber. This change in membrane potential, called an end plate potential(EPP), is normally large enough to bring the membrane potential of the mus-cle cell well above the threshold for producing a postsynaptic action poten-tial (Figure 5.6B). The postsynaptic action potential triggered by the EPPcauses the muscle fiber to contract. Unlike the case for electrical synapses,there is a pronounced delay between the time that the presynaptic motorneuron is stimulated and when the EPP occurs in the postsynaptic musclecell. Such a delay is characteristic of all chemical synapses.One of Katz’s seminal findings, in studies carried out with Paul Fatt in1951, was that spontaneous changes in muscle cell membrane potentialoccur even in the absence of stimulation of the presynaptic motor neuron(Figure 5.6C). These changes have the same shape as EPPs but are much+500−50−100Postsynapticmembranepotential(mV)(B) Stimulatemotor axonTime (ms)20 4 6End platepotential (EPP)ActionpotentialPostsynapticmembranepotential(mV)(C)Time (ms)1 mV0 200 400MEPPPostsynapticmembranepotential(mV)1 mV(D) Stimulatemotor axonTime (ms)0 20 40 60 80 100SpontaneousMEPPSubthreshold EPP(A)StimulateaxonRecordpostsynapticmembranepotentialMuscle cellRecordStimulateAxonThresholdFigure 5.6 Synaptic transmission at the neuromuscular junction. (A) Experimentalarrangement, typically using the muscle of a frog or rat. The axon of the motor neu-ron innervating the muscle fiber is stimulated with an extracellular electrode, whilean intracellular microelectrode is inserted into the postsynaptic muscle cell to recordits electrical responses. (B) End plate potentials (EPPs) evoked by stimulation of amotor neuron are normally above threshold and therefore produce an action poten-tial in the postsynaptic muscle cell. (C) Spontaneous miniature EPPs (MEPPs) occurin the absence of presynaptic stimulation. (D) When the neuromuscular junction isbathed in a solution that has a low concentration of Ca2+, stimulating the motorneuron evokes EPPs whose amplitudes are reduced to about the size of MEPPs.(After Fatt and Katz, 1952.)Purves05 5/13/04 2:27 PM Page 102
    • smaller (typically less than 1 mV in amplitude, compared to an EPP of per-haps 40 or 50 mV). Both EPPs and these small, spontaneous events are sensi-tive to pharmacological agents that block postsynaptic acetylcholine recep-tors, such as curare (see Box B in Chapter 6). These and other parallelsbetween EPPs and the spontaneously occurring depolarizations led Katzand his colleagues to call these spontaneous events miniature end platepotentials, or MEPPs.The relationship between the full-blown end plate potential and MEPPswas clarified by careful analysis of the EPPs. The magnitude of the EPP pro-vides a convenient electrical assay of neurotransmitter secretion from amotor neuron terminal; however, measuring it is complicated by the need toprevent muscle contraction from dislodging the microelectrode. The usualmeans of eliminating muscle contractions is either to lower Ca2+concentra-tion in the extracellular medium or to partially block the postsynaptic AChreceptors with the drug curare. As expected from the scheme illustrated inFigure 5.3, lowering the Ca2+concentration reduces neurotransmitter secre-tion, thus reducing the magnitude of the EPP below the threshold for post-synaptic action potential production and allowing it to be measured moreprecisely. Under such conditions, stimulation of the motor neuron producesvery small EPPs that fluctuate in amplitude from trial to trial (Figure 5.6D).These fluctuations give considerable insight into the mechanisms responsi-ble for neurotransmitter release. In particular, the variable evoked responsein low Ca2+is now known to result from the release of unit amounts of AChby the presynaptic nerve terminal. Indeed, the amplitude of the smallestevoked response is strikingly similar to the size of single MEPPs (compareFigure 5.6C and D). Further supporting this similarity, increments in the EPPresponse (Figure 5.7A) occur in units about the size of single MEPPs (Figure5.7B). These “quantal” fluctuations in the amplitude of EPPs indicated toKatz and colleagues that EPPs are made up of individual units, each equiva-lent to a MEPP.The idea that EPPs represent the simultaneous release of many MEPP-likeunits can be tested statistically. A method of statistical analysis based on theindependent occurrence of unitary events (called Poisson statistics) predictswhat the distribution of EPP amplitudes would look like during a largenumber of trials of motor neuron stimulation, under the assumption thatEPPs are built up from unitary events like MEPPs (see Figure 5.7B). The dis-tribution of EPP amplitudes determined experimentally was found to be justthat expected if transmitter release from the motor neuron is indeed quantal(the red curve in Figure 5.7A). Such analyses confirmed the idea that releaseof acetylcholine does indeed occur in discrete packets, each equivalent to aMEPP. In short, a presynaptic action potential causes a postsynaptic EPPbecause it synchronizes the release of many transmitter quanta.Release of Transmitters from Synaptic VesiclesThe discovery of the quantal release of packets of neurotransmitter immedi-ately raised the question of how such quanta are formed and dischargedinto the synaptic cleft. At about the time Katz and his colleagues were usingphysiological methods to discover quantal release of neurotransmitter, elec-tron microscopy revealed, for the first time, the presence of synaptic vesiclesin presynaptic terminals. Putting these two discoveries together, Katz andothers proposed that synaptic vesicles loaded with transmitter are the sourceof the quanta. Subsequent biochemical studies confirmed that synaptic vesi-Synaptic Transmission 103Purves05 5/13/04 2:27 PM Page 103
    • 104 Chapter Five(B)(A)MEPP amplitude (mV)EPP amplitude (mV)NumberofMEPPsNumberofEPPs00102030051015200.4 0.80 0.4 0.81.2 1.6 2.0 2.4 2.8Prediction ofstatistical modelNo EPP inresponse tostimulationFigure 5.7 Quantized distribution of EPP ampli-tudes evoked in a low Ca2+solution. Peaks of EPPamplitudes (A) tend to occur in integer multiplesof the mean amplitude of MEPPs, whose ampli-tude distribution is shown in (B). The leftmost barin the EPP amplitude distribution shows trials inwhich presynaptic stimulation failed to elicit anEPP in the muscle cell. The red curve indicates theprediction of a statistical model based on theassumption that the EPPs result from the indepen-dent release of multiple MEPP-like quanta. Theobserved match, including the predicted numberof failures, supports this interpretation. (AfterBoyd and Martin, 1955.)cles are the repositories of transmitters. These studies have shown that AChis highly concentrated in the synaptic vesicles of motor neurons, where it ispresent at a concentration of about 100 mM. Given the diameter of a small,clear-core synaptic vesicle (∼50 nm), approximately 10,000 molecules of neu-rotransmitter are contained in a single vesicle. This number correspondsquite nicely to the amount of ACh that must be applied to a neuromuscularjunction to mimic a MEPP, providing further support for the idea thatquanta arise from discharge of the contents of single synaptic vesicles.To prove that quanta are caused by the fusion of individual synaptic vesi-cles with the plasma membrane, it is necessary to show that each fusedvesicle causes a single quantal event to be recorded postsynaptically. Thischallenge was met in the late 1970s, when John Heuser, Tom Reese, and col-leagues correlated measurements of vesicle fusion with the quantal contentof EPPs at the neuromuscular junction. In their experiments, the number ofvesicles that fused with the presynaptic plasma membrane was measuredby electron microscopy in terminals that had been treated with a drug (4-aminopyridine, or 4-AP) that enhances the number of vesicle fusion eventsproduced by single action potentials (Figure 5.8A). Parallel electrical mea-surements were made of the quantal content of the EPPs elicited in thisway. A comparison of the number of synaptic vesicle fusions observed withthe electron microscope and the number of quanta released at the synapseshowed a good correlation between these two measures (Figure 5.8B).These results remain one of the strongest lines of support for the idea that aquantum of transmitter release is due to a synaptic vesicle fusing with thepresynaptic membrane. Subsequent evidence, based on other means ofmeasuring vesicle fusion, has left no doubt about the validity of this generalinterpretation of chemical synaptic transmission. Very recent work hasidentified structures within the presynaptic terminal that connect vesicles tothe plasma membrane and may be involved in membrane fusion (Figure5.8C).Purves05 5/13/04 2:27 PM Page 104
    • Local Recycling of Synaptic VesiclesThe fusion of synaptic vesicles causes new membrane to be added to theplasma membrane of the presynaptic terminal, but the addition is not per-manent. Although a bout of exocytosis can dramatically increase the surfacearea of presynaptic terminals, this extra membrane is removed within a fewminutes. Heuser and Reese performed another important set of experi-ments showing that the fused vesicle membrane is actually retrieved andtaken back into the cytoplasm of the nerve terminal (a process called endo-cytosis). The experiments, again carried out at the frog neuromuscular junc-tion, were based on filling the synaptic cleft with horseradish peroxidase(HRP), an enzyme that can be made to produce a dense reaction productthat is visible in an electron microscope. Under appropriate experimentalconditions, endocytosis could then be visualized by the uptake of HRP intothe nerve terminal (Figure 5.9). To activate endocytosis, the presynaptic ter-minal was stimulated with a train of action potentials, and the subsequentfate of the HRP was followed by electron microscopy. Immediately follow-Synaptic Transmission 1050(B)(A)(C)4-AP concentration: 10−3M10−4M10−5MNumber of quanta releasedNumberofvesiclesfusing100001000200030004000500060002000 3000 4000 5000 6000Figure 5.8 Relationship of synapticvesicle exocytosis and quantal transmit-ter release. (A) A special electron micro-scopical technique called freeze-fracturemicroscopy was used to visualize thefusion of synaptic vesicles in presynap-tic terminals of frog motor neurons. Left:Image of the plasma membrane of anunstimulated presynaptic terminal.Right: Image of the plasma membrane ofa terminal stimulated by an actionpotential. Stimulation causes theappearance of dimple-like structuresthat represent the fusion of synapticvesicles with the presynaptic membrane.The view is as if looking down on therelease sites from outside the presynap-tic terminal. (B) Comparison of thenumber of observed vesicle fusions tothe number of quanta released by a pre-synaptic action potential. Transmitterrelease was varied by using a drug (4-AP) that affects the duration of the pre-synaptic action potential, thus changingthe amount of calcium that enters dur-ing the action potential. The diagonalline is the 1:1 relationship expected ifeach vesicle that opened released a sin-gle quantum of transmitter. (C) Finestructure of vesicle fusion sites of frogpresynaptic terminals. Synaptic vesiclesare arranged in rows and are connectedto each other and to the plasma mem-brane by a variety of proteinaceousstructures (yellow). Green structures inthe presynaptic membrane, correspond-ing to the rows of particles seen in (A),are thought to be Ca2+channels. (A andB from Heuser et al., 1979; C after Har-low et al., 2001)Purves05 5/13/04 2:27 PM Page 105
    • 106 Chapter Fiveing stimulation, the HRP was found within special endocytotic organellescalled coated vesicles (Figure 5.9A,B). A few minutes later, however, thecoated vesicles had disappeared and the HRP was found in a differentorganelle, the endosome (Figure 5.9C). Finally, within an hour after stimu-lating the terminal, the HRP reaction product appeared inside synaptic vesi-cles (Figure 5.9D).These observations indicate that synaptic vesicle membrane is recycledwithin the presynaptic terminal via the sequence summarized in Figure 5.9E.In this process, called the synaptic vesicle cycle, the retrieved vesicular mem-brane passes through a number of intracellular compartments—such ascoated vesicles and endosomes—and is eventually used to make new synap-tic vesicles. After synaptic vesicles are re-formed, they are stored in a reservepool within the cytoplasm until they need to participate again in neurotrans-mitter release. These vesicles are mobilized from the reserve pool, docked atthe presynaptic plasma membrane, and primed to participate in exocytosisonce again. More recent experiments, employing a fluorescent label ratherthan HRP, have determined the time course of synaptic vesicle recycling.These studies indicate that the entire vesicle cycle requires approximately 1minute, with membrane budding during endocytosis requiring 10–20 sec-EndosomeDockingBuddingPrimingFusion1 msecCa2+(A)(E)Horseradish peroxidase (HRP)BrieflystimulatepresynapticterminalWash awayextracellular HRP;wait 5 minutes 1 hour later(B) (C) (D)Coated pits andcoated vesiclescontain HRPEndosomecontains HRPSynaptic vesiclescontain HRP12 34Synapticvesicles fuseBuddingEndocytosisExocytosis10–20sec1 minFigure 5.9 Local recycling of synaptic vesicles inpresynaptic terminals. (A) Horseradish peroxidase(HRP) introduced into the synaptic cleft is used tofollow the fate of membrane retrieved from the pre-synaptic plasma membrane. Stimulation of endocy-tosis by presynaptic action potentials causes HRP tobe taken up into the presynaptic terminals via apathway that includes (B) coated vesicles and (C)endosomes. (D) Eventually, the HRP is found innewly formed synaptic vesicles. (E) Interpretation ofthe results shown in A–D. Calcium-regulated fusionof vesicles with the presynaptic membrane is fol-lowed by endocytotic retrieval of vesicular mem-brane via coated vesicles and endosomes, and sub-sequent re-formation of new synaptic vesicles.(After Heuser and Reese, 1973.)Purves05 5/13/04 2:27 PM Page 106
    • onds of this time. As can be seen from the 1-millisecond delay in transmissionfollowing excitation of the presynaptic terminal (see Figure 5.6B), membranefusion during exocytosis is much more rapid than budding during endocyto-sis. Thus, all of the recycling steps interspersed between membrane buddingand subsequent refusion of a vesicle are completed in less than a minute.The precursors to synaptic vesicles originally are produced in the endo-plasmic reticulum and Golgi apparatus in the neuronal cell body. Because ofthe long distance between the cell body and the presynaptic terminal in mostneurons, transport of vesicles from the soma would not permit rapid replen-ishment of synaptic vesicles during continuous neural activity. Thus, localrecycling is well suited to the peculiar anatomy of neurons, giving nerve ter-minals the means to provide a continual supply of synaptic vesicles. Asmight be expected, defects in synaptic vesicle recycling can cause severeneurological disorders, some of which are described in Box B.The Role of Calcium in Transmitter SecretionAs was apparent in the experiments of Katz and others described in the pre-ceding sections, lowering the concentration of Ca2+outside a presynapticmotor nerve terminal reduces the size of the EPP (compare Figure 5.6B andD). Moreover, measurement of the number of transmitter quanta releasedunder such conditions shows that the reason the EPP gets smaller is thatlowering Ca2+concentration decreases the number of vesicles that fuse withthe plasma membrane of the terminal. An important insight into how Ca2+regulates the fusion of synaptic vesicles was the discovery that presynapticterminals have voltage-sensitive Ca2+channels in their plasma membranes(see Chapter 4).The first indication of presynaptic Ca2+channels was provided by Katzand Ricardo Miledi. They observed that presynaptic terminals treated withtetrodotoxin (which blocks Na+channels; see Chapter 3) could still producea peculiarly prolonged type of action potential. The explanation for this sur-prising finding was that current was still flowing through Ca2+channels,substituting for the current ordinarily carried by the blocked Na+channels.Subsequent voltage clamp experiments, performed by Rodolfo Llinás andothers at a giant presynaptic terminal of the squid (Figure 5.10A), confirmedSynaptic Transmission 107(A) CONTROL(B)Postsynapticmembranepotential (mV)Presynapticcalciumcurrent(µA/cm2)Presynapticmembranepotential (mV)–75–50–250200–750–25–500Time (ms)0 3CADMIUM ADDED96 12–3 –3 0 3 96 12PresynapticneuronVpre IprePostsynapticneuronPostsynaptic membranepotentialRecordVoltageclampFigure 5.10 The entry of Ca2+throughthe specific voltage-dependent calciumchannels in the presynaptic terminalscauses transmitter release. (A) Experi-mental setup using an extraordinarilylarge synapse in the squid. The voltageclamp method detects currents flowingacross the presynaptic membrane whenthe membrane potential is depolarized.(B) Pharmacological agents that blockcurrents flowing through Na+and K+channels reveal a remaining inward cur-rent flowing through Ca2+channels.This influx of calcium triggers transmit-ter secretion, as indicated by a change inthe postsynaptic membrane potential.Treatment of the same presynaptic ter-minal with cadmium, a calcium channelblocker, eliminates both the presynapticcalcium current and the postsynapticresponse. (After Augustine and Eckert,1984.)Purves05 5/13/04 2:27 PM Page 107
    • 108 Chapter FiveBox BDiseases That Affect the Presynaptic TerminalVarious steps in the exocytosis and endo-cytosis of synaptic vesicles are targets ofa number of rare but debilitating neuro-logical diseases. Many of these are myas-thenic syndromes, in which abnormaltransmission at neuromuscular synapsesleads to weakness and fatigability ofskeletal muscles (see Box B in Chapter 7).One of the best-understood examples ofsuch disorders is the Lambert-Eatonmyasthenic syndrome (LEMS), an occa-sional complication in patients with cer-tain kinds of cancers. Biopsies of muscletissue removed from LEMS patientsallow intracellular recordings identical tothose shown in Figure 5.6. Such record-ings have shown that when a motor neu-ron is stimulated, the number of quantacontained in individual EPPs is greatlyreduced, although the amplitude ofspontaneous MEPPs is normal. Thus,LEMS impairs evoked neurotransmitterrelease, but does not affect the size ofindividual quanta.Several lines of evidence indicate thatthis reduction in neurotransmitter releaseis due to a loss of voltage-gated Ca2+channels in the presynaptic terminal ofmotor neurons (see figure). Thus, thedefect in neuromuscular transmissioncan be overcome by increasing the extra-cellular concentration of Ca2+, andanatomical studies indicate a lower den-sity of Ca2+channel proteins in the pre-synaptic plasma membrane. The loss ofpresynaptic Ca2+channels in LEMSapparently arises from a defect in theimmune system. The blood of LEMSpatients has a very high concentration ofantibodies that bind to Ca2+channels,and it seems likely that these antibodiesare the primary cause of LEMS. Forexample, removal of Ca2+channel anti-bodies from the blood of LEMS patientsby plasma exchange reduces muscleweakness. Similarly, immunosuppres-sant drugs also can alleviate LEMSsymptoms. Perhaps most telling, inject-ing these antibodies into experimentalanimals elicits muscle weakness andabnormal neuromuscular transmission.Why the immune system generates anti-bodies against Ca2+channels is not clear.Most LEMS patients have small-cell car-cinoma, a form of lung cancer that maysomehow initiate the immune responseto Ca2+channels. Whatever the origin,the binding of antibodies to Ca2+chan-nels causes a reduction in Ca2+channelcurrents. It is this antibody-induceddefect in presynaptic Ca2+entry thataccounts for the muscle weakness associ-ated with LEMS.Congenital myasthenic syndromesare genetic disorders that also causemuscle weakness by affecting neuromus-cular transmission. Some of these syn-dromes affect the acetylcholinesterasethat degrades acetylcholine in the synap-tic cleft, whereas others arise fromautoimmune attack of acetylcholinereceptors (see Box C in Chapter 6). How-ever, a number of congenital myasthenicsyndromes arise from defects in acetyl-choline release due to altered synapticvesicle traffic within the motor neuronterminal. Neuromuscular synapses insome of these patients have EPPs withreduced quantal content, a deficit that isespecially prominent when the synapseis activated repeatedly. Electronmicroscopy shows that presynapticmotor nerve terminals have a greatlyreduced number of synaptic vesicles. Thedefect in neurotransmitter release evi-dently results from an inadequate num-ber of synaptic vesicles available forrelease during sustained presynapticactivity. The origins of this shortage ofsynaptic vesicles is not clear, but couldresult either from an impairment inendocytosis in the nerve terminal (seefigure) or from a reduced supply of vesi-cles from the motor neuron cell body.Still other patients suffering fromfamilial infantile myasthenia appear tohave neuromuscular weakness thatarises from reductions in the size of indi-vidual quanta, rather than the number ofquanta released. Motor nerve terminalsfrom these patients have synaptic vesi-cles that are normal in number, butsmaller in diameter. This finding sug-gests a different type of genetic lesionthat somehow alters formation of newsynaptic vesicles following endocytosis,thereby leading to less acetylcholine ineach vesicle.Another disorder of synaptic trans-mitter release results from poisoning byanaerobic Clostridium bacteria. Thisgenus of microorganisms produces someEndosomeDockingBuddingPrimingFusionCa2+BuddingImpaired endocytosisin congenital myasthenicsyndromesBotulinum and tetanustoxins affect SNARE proteinsinvolved in vesicle fusionLEMS attackspresynaptic Ca2+channelsPresynaptic targets of several neurologicaldisorders.Purves05 5/13/04 2:27 PM Page 108
    • the presence of voltage-gated Ca2+channels in the presynaptic terminal (Fig-ure 5.10B). Such experiments showed that the amount of neurotransmitterreleased is very sensitive to the exact amount of Ca2+that enters. Further,blockade of these Ca2+channels with drugs also inhibits transmitter release(Figure 5.10B, right). These observations all confirm that the voltage-gatedCa2+channels are directly involved in neurotransmission. Thus, presynapticaction potentials open voltage-gated Ca2+channels, with a resulting influx ofCa2+.That Ca2+entry into presynaptic terminals causes a rise in the concentra-tion of Ca2+within the terminal has been documented by microscopic imag-ing of terminals filled with Ca2+-sensitive fluorescent dyes (Figure 5.11A).The consequences of the rise in presynaptic Ca2+concentration for neuro-transmitter release has been directly shown in two ways. First, microinjec-tion of Ca2+into presynaptic terminals triggers transmitter release in theabsence of presynaptic action potentials (Figure 5.11B). Second, presynapticmicroinjection of calcium chelators (chemicals that bind Ca2+and keep itsconcentration buffered at low levels) prevents presynaptic action potentialsfrom causing transmitter secretion (Figure 5.11C). These results provebeyond any doubt that a rise in presynaptic Ca2+concentration is both nec-essary and sufficient for neurotransmitter release. Thus, as is the case formany other forms of neuronal signaling (see Chapter 7), Ca2+serves as a sec-ond messenger during transmitter release.While Ca2+is a universal trigger for transmitter release, not all transmit-ters are released with the same speed. For example, while secretion of AChSynaptic Transmission 109of the most potent toxins known, includ-ing several botulinum toxins and tetanustoxin. Both botulism and tetanus arepotentially deadly disorders.Botulism can occur by consumingfood containing Clostridium bacteria orby infection of wounds with the sporesof these ubiquitous organisms. In eithercase, the presence of the toxin can causeparalysis of peripheral neuromuscularsynapses due to abolition of neurotrans-mitter release. This interference withneuromuscular transmission causesskeletal muscle weakness, in extremecases producing respiratory failure dueto paralysis of the diaphragm and othermuscles required for breathing. Botu-linum toxins also block synapses inner-vating the smooth muscles of severalorgans, giving rise to visceral motor dys-function.Tetanus typically results from the con-tamination of puncture wounds byClostridium bacteria that produce tetanustoxin. In contrast to botulism, tetanuspoisoning blocks the release of inhibitorytransmitters from interneurons in thespinal cord. This effect causes a loss ofsynaptic inhibition on spinal motor neu-rons, producing hyperexcitation of skele-tal muscle and tetanic contractions inaffected muscles (hence the name of thedisease).Although their clinical consequencesare dramatically different, clostridial tox-ins have a common mechanism of action(see figure). Tetanus toxin and botulinumtoxins work by cleaving the SNARE pro-teins involved in fusion of synaptic vesi-cles with the presynaptic plasma mem-brane (see Box C). This proteolytic actionpresumably accounts for the block oftransmitter release at the afflicted syn-apses. The different actions of these tox-ins on synaptic transmission at excitatorymotor versus inhibitory synapses appar-ently results from the fact that these tox-ins are taken up by different types ofneurons: Whereas the botulinum toxinsare taken up by motor neurons, tetanustoxin is preferentially targeted tointerneurons. The basis for this differen-tial uptake of toxins is not known, butpresumably arises from the presence ofdifferent types of toxin receptors on thetwo types of neurons.ReferencesENGEL, A. G. (1991) Review of evidence forloss of motor nerve terminal calcium chan-nels in Lambert-Eaton myasthenic syndrome.Ann. N.Y. Acad. Sci. 635: 246–258.ENGEL, A. G. (1994) Congenital myasthenicsyndromes. Neurol. Clin. 12: 401–437.LANG, B. AND A. VINCENT (2003) Autoantibod-ies to ion channels at the neuromuscularjunction. Autoimmun. Rev. 2: 94–100.MASELLI, R. A. (1998) Pathogenesis of humanbotulism. Ann. N.Y. Acad. Sci. 841: 122–139.Purves05 5/13/04 2:27 PM Page 109
    • 110 Chapter FiveFigure 5.11 Evidence that a rise in pre-synaptic Ca2+concentration triggerstransmitter release from presynaptic ter-minals. (A) Fluorescence microscopymeasurements of presynaptic Ca2+con-centration at the squid giant synapse(see Figure 5.8A). A train of presynapticaction potentials causes a rise in Ca2+concentration, as revealed by a dye(called fura-2) that fluoresces morestrongly when the Ca2+concentrationincreases. (B) Microinjection of Ca2+intoa squid giant presynaptic terminal trig-gers transmitter release, measured as adepolarization of the postsynaptic mem-brane potential. (C) Microinjection ofBAPTA, a Ca2+chelator, into a squidgiant presynaptic terminal preventstransmitter release. (A from Smith et al.,1993; B after Miledi, 1971; C after Adleret al., 1991.)from motor neurons requires only a fraction of a millisecond (see Figure 5.6),release of neuropeptides require high-frequency bursts of action potentialsfor many seconds. These differences in the rate of release probably arisefrom differences in the spatial arrangement of vesicles relative to presynapticCa2+channels. This perhaps is most evident in cases where small moleculesand peptides serve as co-transmitters (Figure 5.12). Whereas the small, clear-core vesicles containing small-molecule transmitters are typically docked atthe plasma membrane in advance of Ca2+entry, large dense core vesiclescontaining peptide transmitters are farther away from the plasma membrane(see Figure 5.5D). At low firing frequencies, the concentration of Ca2+mayincrease only locally at the presynaptic plasma membrane, in the vicinity ofopen Ca2+channels, limiting release to small-molecule transmitters from thedocked small, clear-core vesicles. Prolonged high-frequency stimulationincreases the Ca2+concentration throughout the presynaptic terminal,thereby inducing the slower release of neuropeptides.Molecular Mechanisms of Transmitter SecretionPrecisely how an increase in presynaptic Ca2+concentration goes on to trig-ger vesicle fusion and neurotransmitter release is not understood. However,many important clues have come from molecular studies that have identifiedand characterized the proteins found on synaptic vesicles and their binding−75−50−25025CONTROL(C)(B)(A)INJECT Ca2+ BUFFERPHOTOwith line overlayTime (s)−65−640 1 2 3Postsynapticmembranepotential(mV)Presynapticmembranepotential(mV)Postsynapticmembranepotential(mV)0−75−50−250251 2 3 4 5Time (ms)0 1 2 3 4 5Ca2+injection4250µmCa2+Purves05 5/13/04 2:27 PM Page 110
    • partners on the presynaptic plasma membrane and cytoplasm (Figure 5.13).Most, if not all, of these proteins act at one or more steps in the synaptic vesi-cle cycle. Although a complete molecular picture of neurotransmitter releaseis still lacking, the roles of several proteins involved in vesicle fusion havebeen deduced.Several of the proteins important for neurotransmitter release are alsoinvolved in other types of membrane fusion events common to all cells. Forexample, two proteins originally found to be important for the fusion ofvesicles with membranes of the Golgi apparatus, the ATPase NSF (NEM-sen-sitive fusion protein) and SNAPs (soluble NSF-attachment proteins), are alsoinvolved in priming synaptic vesicles for fusion. These two proteins work byregulating the assembly of other proteins that are called SNAREs (SNAPreceptors). One of these SNARE proteins, synaptobrevin, is in the mem-brane of synaptic vesicles, while two other SNARE proteins called syntaxinand SNAP-25 are found primarily on the plasma membrane. These SNAREproteins can form a macromolecular complex that spans the two mem-branes, thus bringing them into close apposition (Figure 5.14A). Such anarrangement is well suited to promote the fusion of the two membranes, andseveral lines of evidence suggest that this is what actually occurs. Oneimportant observation is that toxins that cleave the SNARE proteins blockneurotransmitter release (Box C). In addition, putting SNARE proteins intoartificial lipid membranes and allowing these proteins to form complexeswith each other causes the membranes to fuse. Many other proteins, such asSynaptic Transmission 111Localizedincrease in Ca2+concentrationSmall-moleculeneurotransmitterin small clear-core vesiclesLow-frequencystimulationMore diffuseincrease in Ca2+concentrationHigh-frequencystimulationRelease of bothtypes of transmitterPreferential release of small-molecule neurotransmitterNeuropeptidein large dense-core vesiclesFigure 5.12 Differential release of neu-ropeptide and small-molecule co-trans-mitters. Low-frequency stimulationpreferentially raises the Ca2+concentra-tion close to the membrane, favoring therelease of transmitter from small clear-core vesicles docked at presynaptic spe-cializations. High-frequency stimulationleads to a more general increase in Ca2+,causing the release of peptide neuro-transmitters from large dense-core vesi-cles, as well as small-molecule neuro-transmitters from small clear-corevesicles.Purves05 5/13/04 2:27 PM Page 111
    • 112 Chapter FiveSynapticvesicleSynaptic vesiclemembraneSynapticcleftCytoplasmPlasma membrane ofpresynaptic terminalSV2Rab 3RabphilinSynaptophysinCysteinestring proteinGTP-binding proteinsMiscellaneous important proteinsProteins that form channels, transporters, or receptorsCa2+-binding proteinsSNARE-associated proteinsProteins involved in endocytosisCa2+channelNeurexin I CLICa2+/CaMdependent proteinkinase IISynapsinDOC2RIMSynaptotagminSyndapinDynaminClathrinAP–2AP180AmphiphysinAuxilinSynaptojaninHsc70SnapinTomosynSNAPSynaptobrevinSyntaxinSNAP–25SyntaphilinComplexinNSFnSec1Figure 5.13 Presynaptic proteins implicated in neurotransmitter release. Structuresadapted from Brunger (2001) and Brodsky et al. (2001).Purves05 5/13/04 2:27 PM Page 112
    • complexin, nSec-1, snapin, syntaphilin, and tomosyn, bind to the SNAREsand presumably regulate the formation or disassembly of this complex.Because the SNARE proteins do not bind Ca2+, still other molecules mustbe responsible for Ca2+regulation of neurotransmitter release. Several pre-synaptic proteins, including calmodulin, CAPS, and munc-13, are capable ofbinding Ca2+. However, the leading candidate for Ca2+regulation of neuro-transmitter release is synaptotagmin, a protein found in the membrane ofsynaptic vesicles. Synaptotagmin binds Ca2+at concentrations similar tothose required to trigger vesicle fusion within the presynaptic terminal. Itmay act as a Ca2+sensor, signaling the elevation of Ca2+within the terminaland thus triggering vesicle fusion. In support of this idea, alterations of theproperties of synaptotagmin in the presynaptic terminals of mice, fruit flies,squid, and other experimental animals impair Ca2+-dependent neurotrans-mitter release. In fact, deletion of only one of the 19 synaptotagmin genes ofmice is a lethal mutation, causing the mice to die soon after birth. How Ca2+binding to synaptotagmin could lead to exocytosis is not yet clear. It isknown that Ca2+changes the chemical properties of synaptotagmin, allow-ing it to insert into membranes and to bind to other proteins, including theSNAREs. A plausible model is that the SNARE proteins bring the two mem-branes close together, and that Ca2+-induced changes in synaptotagmin thenproduce the final fusion of these membranes (Figure 5.14B).Still other proteins appear to be involved at subsequent steps of the syn-aptic vesicle cycle (Figure 5.14C). For example, the protein clathrin isinvolved in endocytotic budding of vesicles from the plasma membrane.Clathrin forms structures that resemble geodesic domes (Figure 5.14D);these structures form coated pits that initiate membrane budding. Assemblyof individual clathrin triskelia (so named because of their 3-legged appear-ance) into coats is aided by several other accessory proteins, such as AP2,AP180 and amphiphysin. The coats increase the curvature of the buddingmembrane until it forms a coated vesicle-like structure. Another protein,called dynamin, is at least partly responsible for the final pinching-off ofmembrane to convert the coated pits into coated vesicles. The coats are thenremoved by an ATPase, Hsc70, with another protein called auxilin servingas a co-factor. Other proteins, such as synaptojanin, are also important forvesicle uncoating. Several lines of evidence indicate that the proteinsynapsin, which reversibly binds to synaptic vesicles, may cross-link newlyformed vesicles to the cytoskeleton to keep the vesicles tethered within thereserve pool. Mobilization of these reserve pool vesicles is caused by phos-phorylation of synapsin by proteins kinases (Chapter 7), which allowssynapsin to dissociate from the vesicles, thus freeing the vesicles to maketheir way to the plasma membrane.In summary, a complex cascade of proteins, acting in a defined temporaland spatial order, allows neurons to secrete transmitters. Although thedetailed mechanisms responsible for transmitter secretion are not completelyclear, rapid progress is being made toward this goal.Neurotransmitter ReceptorsThe generation of postsynaptic electrical signals is also understood in con-siderable depth. Such studies began in 1907, when the British physiologistJohn N. Langley introduced the concept of receptor molecules to explain thespecific and potent actions of certain chemicals on muscle and nerve cells.Much subsequent work has shown that receptor molecules do indeedaccount for the ability of neurotransmitters, hormones, and drugs to alter theSynaptic Transmission 113Purves05 5/13/04 2:27 PM Page 113
    • 114 Chapter FiveEndosomeDockingBuddingUncoatingSynapsinDynaminHsc 70AuxilinSynaptojaninPriming FusionCa2+BuddingSNAREsNSF SNAPs(C)(D)SynaptotagminClathrintriskelionClathrin(1) Vesicle docks(2) SNARE complexes form to pullmembranes together(3) Entering Ca2+binds to synaptotagmin(4) Ca2+-bound synaptotagmin catalyzesmembrane fusionSyntaxinSynaptobrevinSNAP-25VesicleCa2+channelSynaptotagminCa2+Presynaptic plasma membraneSynapticvesiclemembraneSyntaxinSynaptobrevinSNAP-25(A)Clathrincoat(B)SynaptotagminFigure 5.14 Molecular mechanisms of neurotransmitter release. (A) Struc-ture of the SNARE complex. The vesicular SNARE, synaptobrevin (blue),forms a helical complex with the plasma membrane SNAREs syntaxin (red)and SNAP-25 (green). Also shown is the structure of synaptotagmin, a vesic-ular Ca2+-binding protein. (B) A model for Ca2+-triggered vesicle fusion.SNARE proteins on the synaptic vesicle and plasma membranes form a com-plex (as in A) that brings together the two membranes. Ca2+then binds tosynaptotagmin, causing the cytoplasmic region of this protein to insert intothe plasma membrane, bind to SNAREs and catalyze membrane fusion. (C)Roles of presynaptic proteins in synaptic vesicle cycling. (D) Individualclathrin triskelia (left) assemble together to form membrane coats (right)involved in membrane budding during endocytosis. (A after Sutton et al.,1998; C after Sudhof, 1995; D after Marsh and McMahon, 2001.)Purves05 5/13/04 2:27 PM Page 114
    • Synaptic Transmission 115Box CToxins That Affect Transmitter ReleaseSeveral important insights about themolecular basis of neurotransmittersecretion have come from analyzing theactions of a series of biological toxinsproduced by a fascinating variety oforganisms. One family of such agents isthe clostridial toxins responsible for bot-ulism and tetanus (see Box B). Cleverand patient biochemical work has shownthat these toxins are highly specific pro-teases that cleave presynaptic SNAREproteins (see figure). Tetanus toxin andbotulinum toxin (types B, D, F, and G)specifically cleave the vesicle SNAREprotein, synaptobrevin. Other botulinumtoxins are proteases that cleave syntaxin(type C) and SNAP-25 (types A and E),SNARE proteins found on the presynap-tic plasma membrane. Destruction ofthese presynaptic proteins is the basis forthe actions of the toxins on neurotrans-mitter release. The evidence described inthe text also implies that these three syn-aptic SNARE proteins are somehowimportant in the process ofvesicle–plasma membrane fusion.Another toxin that targets neurotrans-mitter release is α-latrotoxin, a proteinfound in the venom of the female blackwidow spider. Application of this mole-cule to neuromuscular synapses causes amassive discharge of synaptic vesicles,even when Ca2+is absent from the extra-cellular medium. While it is not yet clearhow this toxin triggers Ca2+-independentexocytosis, α-latrotoxin binds to two dif-ferent types of presynaptic proteins thatmay mediate its actions. One group ofbinding partners for α-latrotoxin is theneurexins, a group of integral membraneproteins found in presynaptic terminals(see Figure 5.13). Several lines of evi-dence implicate binding to neurexins inat least some of the actions of α-latro-toxin. Because the neurexins bind tosynaptotagmin, a vesicular Ca2+-bindingprotein that is known to be important inexocytosis, this interaction may allow α-latrotoxin to bypass the usual Ca2+requirement for triggering vesicle fusion.Another type of presynaptic protein thatcan bind to α-latrotoxin is called CL1(based on its previous names, Ca2+-inde-pendent receptor for latrotoxin and lat-rophilin-1). CL1 is a relative of the G-pro-tein-coupled receptors that mediate theactions of neurotransmitters and otherextracellular chemical signals (see Chap-ter 7). Thus, the binding of α-latrotoxinto CL1 is thought to activate an intracel-lular signal transduction cascade thatmay be involved in the Ca2+-indepen-dent actions of α-latrotoxin. While morework is needed to establish the roles ofneurexins and CL1 in the actions of α-latrotoxin definitively, effects on thesetwo proteins probably account for thepotent presynaptic actions of this toxin.Still other toxins produced by snakes,snails, spiders, and other predatory ani-mals are known to affect transmitterrelease, but their sites of action have yetto be identified. Based on the precedentsdescribed here, it is likely that these bio-logical poisons will continue to providevaluable tools for elucidating the molec-ular basis of neurotransmitter release,just as they will continue to enable thepredators to feast on their prey.ReferencesKRASNOPEROV, V. G. AND 10 OTHERS (1997) α-Latrotoxin stimulates exocytosis by the inter-action with a neuronal G-protein-coupledreceptor. Neuron 18: 925–937.MONTECUCCO, C. AND G. SCHIAVO (1994)Mechanism of action of tetanus and botu-linum neurotoxins. Mol. Microbiol. 13: 1–8.SCHIAVO, G., M. MATTEOLI AND C. MONTE-CUCCO (2000) Neurotoxins affecting neuro-exocytosis. Physiol. Rev. 80: 717–766.SUGITA, S., M. KHVOCHTEV AND T. C. SUDHOF(1999) Neurexins are functional α-latrotoxinreceptors. Neuron 22: 489–496.BoTX−GBoTX−DBoTX−FBoTX−ABoTX−CBoTX−EBoTX−BTeTXSyntaxinSynaptobrevinSNAP-25Presynaptic plasma membraneSynaptic vesiclemembraneCleavage of SNARE proteins by clostridial toxins. Indicated are the sites of proteolysis bytetanus toxin (TeTX) and various types of botulinum toxin (BoTX). (After Sutton et al., 1998.)Purves05 5/13/04 2:27 PM Page 115
    • 116 Chapter Fivefunctional properties of neurons. While it has been clear since Langley’s daythat receptors are important for synaptic transmission, their identity anddetailed mechanism of action remained a mystery until quite recently. It isnow known that neurotransmitter receptors are proteins embedded in theplasma membrane of postsynaptic cells. Domains of receptor molecules thatextend into the synaptic cleft bind neurotransmitters that are released intothis space by the presynaptic neuron. The binding of neurotransmitters,either directly or indirectly, causes ion channels in the postsynaptic mem-brane to open or close. Typically, the resulting ion fluxes change the mem-brane potential of the postsynaptic cell, thus mediating the transfer of infor-mation across the synapse.Postsynaptic Membrane Permeability Changes during SynapticTransmissionJust as studies of the neuromuscular synapse paved the way for understand-ing neurotransmitter release mechanisms, this peripheral synapse has beenequally valuable for understanding the mechanisms that allow neurotrans-mitter receptors to generate postsynaptic signals. The binding of ACh to post-synaptic receptors opens ion channels in the muscle fiber membrane. Thiseffect can be demonstrated directly by using the patch clamp method (seeBox A in Chapter 4) to measure the minute postsynaptic currents that flowwhen two molecules of individual ACh bind to receptors, as Erwin Neherand Bert Sakmann first did in 1976. Exposure of the extracellular surface of apatch of postsynaptic membrane to ACh causes single-channel currents toflow for a few milliseconds (Figure 5.15A). This shows that ACh binding toits receptors opens ligand-gated ion channels, much in the way that changesin membrane potential open voltage-gated ion channels (Chapter 4).The electrical actions of ACh are greatly multiplied when an action poten-tial in a presynaptic motor neuron causes the release of millions of moleculesof ACh into the synaptic cleft. In this more physiological case, the transmit-ter molecules bind to many thousands of ACh receptors packed in a densearray on the postsynaptic membrane, transiently opening a very large num-ber of postsynaptic ion channels. Although individual ACh receptors onlyopen briefly, (Figure 5.15B1), the opening of a large number of channels issynchronized by the brief duration during which ACh is secreted from pre-synaptic terminals (Figure 5.15B2,3). The macroscopic current resulting fromthe summed opening of many ion channels is called the end plate current,or EPC. Because the current flowing during the EPC is normally inward, itcauses the postsynaptic membrane potential to depolarize. This depolarizingchange in potential is the EPP (Figure 5.15C), which typically triggers a post-synaptic action potential by opening voltage-gated Na+and K+channels (seeFigure 5.6B).The identity of the ions that flow during the EPC can be determined viathe same approaches used to identify the roles of Na+and K+fluxes in thecurrents underlying action potentials (Chapter 3). Key to such an analysis isidentifying the membrane potential at which no current flows during trans-mitter action. When the potential of the postsynaptic muscle cell is controlledby the voltage clamp method (Figure 5.16A), the magnitude of the membranepotential clearly affects the amplitude and polarity of EPCs (Figure 5.16B).Thus, when the postsynaptic membrane potential is made more negativethan the resting potential, the amplitude of the EPC becomes larger, whereasthis current is reduced when the membrane potential is made more positive.At approximately 0 mV, no EPC is detected, and at even more positive poten-Purves05 5/13/04 2:27 PM Page 116
    • tials, the current reverses its polarity, becoming outward rather than inward(Figure 5.16C). The potential where the EPC reverses, about 0 mV in the caseof the neuromuscular junction, is called the reversal potential.As was the case for currents flowing through voltage-gated ion channels(see Chapter 3), the magnitude of the EPC at any membrane potential isgiven by the product of the ionic conductance activated by ACh (gACh) andthe electrochemical driving force on the ions flowing through ligand-gatedchannels. Thus, the value of the EPC is given by the relationshipEPC = gACh(Vm – Erev)where Erev is the reversal potential for the EPC. This relationship predictsthat the EPC will be an inward current at potentials more negative than Erevbecause the electrochemical driving force, Vm – Erev, is a negative number.Further, the EPC will become smaller at potentials approaching Erev becausethe driving force is reduced. At potentials more positive than Erev, the EPC isoutward because the driving force is reversed in direction (that is, positive).Because the channels opened by ACh are largely insensitive to membranevoltage, gACh will depend only on the number of channels opened by ACh,which depends in turn on the concentration of ACh in the synaptic cleft.Synaptic Transmission 1172 µM Acetylcholine (ACh)I (pA)0 2 4 6 8 10Time (ms)1220 Channel closedChannel closedChannel openChannel openChannel open 0000600,000020024200,00010300,000Micropipette(A) Patch clamp measurement of single ACh receptor current(1) SINGLE OPEN CHANNEL(B) Currents produced by:Outside-outmembrane patchACh receptorAChNa+0 2–2 4 6 8 10 12 14Time (ms)(3) ALL CHANNELS OPEN1Numberof openchannelsNumberof openchannelsNumberof openchannels(2) FEW OPEN CHANNELSMembranrecurrent(pA)−90−70−100−800 2 4 6 8 10 12 14–2(C) Postsynaptic potential change (EPP) produced by EPCTime (ms)Membranepotential(mV)ACh release bystimulating motor neuronChannel closed2Figure 5.15 Activation of ACh receptors at neuromuscular syn-apses. (A) Outside-out patch clamp measurement of single AChreceptor currents from a patch of membrane removed from thepostsynaptic muscle cell. When ACh is applied to the extracellu-lar surface of the membrane clamped at negative voltages, therepeated brief opening of a single channel can be seen as down-ward deflections corresponding to inward current (i.e., positiveions flowing into the cell). (B) Synchronized opening of manyACh-activated channels at a synapse being voltage-clamped atnegative voltages. (1) If a single channel is examined during therelease of ACh from the presynaptic terminal, the channel openstransiently. (2) If a number of channels are examined together, ACh release opensthe channels almost synchronously. (3) The opening of a very large number of post-synaptic channels produces a macroscopic EPC. (C) In a normal muscle cell (i.e.,not being voltage-clamped), the inward EPC depolarizes the postsynaptic musclecell, giving rise to an EPP. Typically, this depolarization generates an action poten-tial (not shown).Purves05 5/13/04 2:27 PM Page 117
    • 118 Chapter FiveThus, the magnitude and polarity of the postsynaptic membrane potentialdetermines the direction and amplitude of the EPC solely by altering the dri-ving force on ions flowing through the receptor channels opened by ACh.When Vm is at the reversal potential, Vm – Erev is equal to 0 and there is nonet driving force on the ions that can permeate the receptor-activated chan-nel. As a result, the identity of the ions that flow during the EPC can bededuced by observing how the reversal potential of the EPC compares to theequilibrium potential for various ion species (Figure 5.17). For example, ifACh were to open an ion channel permeable only to K+, then the reversal(B) Effect of membrane voltage on postsynaptic end plate currents4200200100−100−200−3006(C)Postsynaptic membrane potential (mV)100200−1000−200−300−110 −60 +700EPCamplitude(nA)EPC(nA)Time (ms)420 6 420 6 420 6EK ECl ENa(A) Scheme for voltage clamping postsynaptic muscle fiberPostsynapticmuscle fiberPresynapticterminalsAxon ofpresynapticmotor neuron Voltage clampamplifierVoltage-measuringelectrodeCurrent-passingelectrode−110 mVStimulate−110 −60 +700 −110 −60 +700Reversalpotential−60 mV 0 mV +70 mVStimulate presynaptic axonStimulate presynaptic axonStimulate presynaptic axonStimulate presynaptic axon(D)Lower external [Na+]shifts reversalpotential to left(E)Higher external [K+]shifts reversalpotential to rightFigure 5.16 The influence of the postsynaptic membrane potentialon end plate currents. (A) A postsynaptic muscle fiber is voltageclamped using two electrodes, while the presynaptic neuron is electri-cally stimulated to cause the release of ACh from presynaptic termi-nals. This experimental arrangement allows the recording of macro-scopic EPCs produced by ACh. (B) Amplitude and time course ofEPCs generated by stimulating the presynaptic motor neuron whilethe postsynaptic cell is voltage clamped at four different membranepotentials. (C) The relationship between the peak amplitude of EPCsand postsynaptic membrane potential is nearly linear, with a reversalpotential (the voltage at which the direction of the current changesfrom inward to outward) close to 0 mV. Also indicated on this graphare the equilibrium potentials of Na+, K+, and Cl–ions. (D) Loweringthe external Na+concentration causes EPCs to reverse at more nega-tive potentials. (E) Raising the external K+concentration makes thereversal potential more positive. (After Takeuchi and Takeuchi, 1960.)Purves05 5/13/04 2:27 PM Page 118
    • potential of the EPC would be at the equilibrium potential for K+, which fora muscle cell is close to –100 mV (Figure 5.17A). If the ACh-activated chan-nels were permeable only to Na+, then the reversal potential of the currentwould be approximately +70 mV, the Na+equilibrium potential of musclecells (Figure 5.17B); if these channels were permeable only to Cl–, then thereversal potential would be approximately –50 mV (Figure 5.17C). By thisreasoning, ACh-activated channels cannot be permeable to only one of theseions, because the reversal potential of the EPC is not near the equilibriumpotential for any of them (see Figure 5.16C). However, if these channels werepermeable to both Na+and K+, then the reversal potential of the EPC wouldbe between +70 mV and –100 mV (Figure 5.17D).The fact that EPCs reverse at approximately 0 mV is therefore consistentwith the idea that ACh-activated ion channels are almost equally permeableto both Na+and K+. This was tested in 1960, by the husband and wife teamof Akira and Noriko Takeuchi, by altering the extracellular concentration ofthese two ions. As predicted, the magnitude and reversal potential of theEPC was changed by altering the concentration gradient of each ion. Lower-ing the external Na+concentration, which makes ENa more negative, pro-duces a negative shift in Erev (Figure 5.16D), whereas elevating external K+concentration, which makes EK more positive, causes Erev to shift to a morepositive potential (Figure 5.16E). Such experiments confirm that the ACh-activated ion channels are in fact permeable to both Na+and K+.Even though the channels opened by the binding of ACh to its receptorsare permeable to both Na+and K+, at the resting membrane potential theEPC is generated primarily by Na+influx (Figure 5.18). If the membranepotential is kept at EK, the EPC arises entirely from an influx of Na+becauseat this potential there is no driving force on K+(Figure 5.18A). At the usualmuscle fiber resting membrane potential of –90 mV, there is a small drivingforce on K+, but a much greater one on Na+. Thus, during the EPC, muchmore Na+flows into the muscle cell than K+flows out (Figure 5.18B); it is thenet influx of positively charged Na+that constitutes the inward current mea-sured as the EPC. At the reversal potential of about 0 mV, Na+influx and K+efflux are exactly balanced, so no current flows during the opening of chan-nels by ACh binding (Figure 5.18C). At potentials more positive than Erev thebalance reverses; for example, at ENa there is no influx of Na+and a largeefflux of K+because of the large driving force on Na+(Figure 5.18D). Evenmore positive potentials cause efflux of both Na+and K+and produce aneven larger outward EPC.Were it possible to measure the EPP at the same time as the EPC (ofcourse, the voltage clamp technique prevents this by keeping membranepotential constant), the EPP would be seen to vary in parallel with the ampli-tude and polarity of the EPC (Figures 5.18E,F). At the usual postsynapticresting membrane potential of –90 mV, the large inward EPC causes thepostsynaptic membrane potential to become more depolarized (see FigureSynaptic Transmission 119Figure 5.17 The effect of ion channel selectivity on the reversal potential. Voltageclamping a postsynaptic cell while activating presynaptic neurotransmitter releasereveals the identity of the ions permeating the postsynaptic receptors being acti-vated. (A) The activation of postsynaptic channels permeable only to K+results incurrents reversing at EK, near –100 mV. (B) The activation of postsynaptic Na+chan-nels results in currents reversing at ENa, near +70 mV. (C) Cl–-selective currentsreverse at ECl, near –50 mV. (D) Ligand-gated channels that are about equally per-meable to both K+and Na+show a reversal potential near 0 mV.(A)−150 −100 100−50 500Membrane potentialOnly K+selective channel open100200300−1000000−200−300EPCamplitude(nA)(B) Only Na+selective channel open−150 −100 100−50 500Membrane potential100200300−100−200−300EPCamplitude(nA)Only Cl−selective channel open(C)−150 −100 100−50 500Membrane potential100200300−100−200−300EPCamplitude(nA)(D) Cation non-selective channel open−150 −100 100−50 500Membrane potential100200300−100−200−300EPCamplitude(nA)K+effluxK+influxErev= EKErev= ENaErev= EClErev= 0Na+effluxNa+influxCl−influxCl−effluxCationeffluxCationinfluxPurves05 5/13/04 2:27 PM Page 119
    • 120 Chapter FiveFigure 5.18 Na+and K+movementsduring EPCs and EPPs. (A–D) Each ofthe postsynaptic potentials (Vpost) indi-cated at the left results in different rela-tive fluxes of net Na+and K+(ionfluxes). These ion fluxes determine theamplitude and polarity of the EPCs,which in turn determine the EPPs. Notethat at about 0 mV the Na+flux isexactly balanced by an opposite K+flux,resulting in no net current flow, andhence no change in the membranepotential. (E) EPCs are inward currentsat potentials more negative than Erevand outward currents at potentialsmore positive than Erev. (F) EPPs depo-larize the postsynaptic cell at potentialsmore negative than Erev. At potentialsmore positive than Erev, EPPs hyperpo-larize the cell.5.18F). However, at 0 mV, the EPP reverses its polarity, and at more positivepotentials, the EPP is hyperpolarizing. Thus, the polarity and magnitude ofthe EPC depend on the electrochemical driving force, which in turn deter-mines the polarity and magnitude of the EPP. EPPs will depolarize when themembrane potential is more negative than Erev, and hyperpolarize when themembrane potential is more positive than Erev. The general rule, then, is thatEPPsEPCsNET ION FLUXESPostsynapticmembranepotential−100 −90 +700Postsynaptic membrane potentialEPCpeakamplitude(nA)EPPpeakamplitude(mV)0Postsynaptic membrane potential−100 −90 +70Na+Outsidecell−90 mVNa+0 mV(Erev)+70 mV(ENa)K+ACh-activatedchannelAChInsidecellEK ENa EK ENaNa+K+K+−100 mV(EK)(A)(B)(C)(D)(E) (F)OutwardInwardDepolarizingHyper−polarizingPurves05 5/13/04 2:27 PM Page 120
    • the action of a transmitter drives the postsynaptic membrane potential toward Erevfor the particular ion channels being activated.Although this discussion has focused on the neuromuscular junction, sim-ilar mechanisms generate postsynaptic responses at all chemical synapses.The general principle is that transmitter binding to postsynaptic receptorsproduces a postsynaptic conductance change as ion channels are opened (orsometimes closed). The postsynaptic conductance is increased if—as at theneuromuscular junction—channels are opened, and decreased if channels areclosed. This conductance change typically generates an electrical current, thepostsynaptic current (PSC), which in turn changes the postsynaptic mem-brane potential to produce a postsynaptic potential (PSP). As in the specificcase of the EPP at the neuromuscular junction, PSPs are depolarizing if theirreversal potential is more positive than the postsynaptic membrane potentialand hyperpolarizing if their reversal potential is more negative.The conductance changes and the PSPs that typically accompany themare the ultimate outcome of most chemical synaptic transmission, conclud-ing a sequence of electrical and chemical events that begins with the inva-sion of an action potential into the terminals of a presynaptic neuron. Inmany ways, the events that produce PSPs at synapses are similar to thosethat generate action potentials in axons; in both cases, conductance changesproduced by ion channels lead to ionic current flow that changes the mem-brane potential (see Figure 5.18).Excitatory and Inhibitory Postsynaptic PotentialsPSPs ultimately alter the probability that an action potential will be producedin the postsynaptic cell. At the neuromuscular junction, synaptic actionincreases the probability that an action potential will occur in the postsynap-tic muscle cell; indeed, the large amplitude of the EPP ensures that an actionpotential always is triggered. At many other synapses, PSPs similarlyincrease the probability of firing a postsynaptic action potential. However,still other synapses actually decrease the probability that the postsynaptic cellwill generate an action potential. PSPs are called excitatory (or EPSPs) if theyincrease the likelihood of a postsynaptic action potential occurring, andinhibitory (or IPSPs) if they decrease this likelihood. Given that most neu-rons receive inputs from both excitatory and inhibitory synapses, it is impor-tant to understand more precisely the mechanisms that determine whether aparticular synapse excites or inhibits its postsynaptic partner.The principles of excitation just described for the neuromuscular junctionare pertinent to all excitatory synapses. The principles of postsynaptic inhi-bition are much the same as for excitation, and are also quite general. In bothcases, neurotransmitters binding to receptors open or close ion channels inthe postsynaptic cell. Whether a postsynaptic response is an EPSP or an IPSPdepends on the type of channel that is coupled to the receptor, and on theconcentration of permeant ions inside and outside the cell. In fact, the onlydistinction between postsynaptic excitation and inhibition is the reversalpotential of the PSP in relation to the threshold voltage for generating actionpotentials in the postsynaptic cell.Consider, for example, a neuronal synapse that uses glutamate as thetransmitter. Many such synapses have receptors that, like the ACh receptorsat neuromuscular synapses, open ion channels that are nonselectively per-meable to cations (see Chapter 6). When these glutamate receptors are acti-vated, both Na+and K+flow across the postsynaptic membrane, yielding anErev of approximately 0 mV for the resulting postsynaptic current. If the rest-Synaptic Transmission 121Purves05 5/13/04 2:27 PM Page 121
    • 122 Chapter Fiveing potential of the postsynaptic neuron is –60 mV, the resulting EPSP willdepolarize by bringing the postsynaptic membrane potential toward 0 mV.For the hypothetical neuron shown in Figure 5.19A, the action potentialthreshold voltage is –40 mV. Thus, a glutamate-induced EPSP will increasethe probability that this neuron produces an action potential, defining thesynapse as excitatory.As an example of inhibitory postsynaptic action, consider a neuronal syn-apse that uses GABA as its transmitter. At such synapses, the GABA recep-tors typically open channels that are selectively permeable to Cl–and theaction of GABA causes Cl–to flow across the postsynaptic membrane. Con-sider a case where ECl is –70 mV, as is typical for many neurons, so that thepostsynaptic resting potential of –60 mV is less negative than ECl. The result-ing positive electrochemical driving force (Vm – Erev) will cause negativelycharged Cl–to flow into the cell and produce a hyperpolarizing IPSP (Figure5.19B). This hyperpolarizing IPSP will take the postsynaptic membraneaway from the action potential threshold of –40 mV, clearly inhibiting thepostsynaptic cell.Surprisingly, inhibitory synapses need not produce hyperpolarizingIPSPs. For instance, if ECl were –50 mV instead of –70 mV, then the negativeelectrochemical driving force would cause Cl–to flow out of the cell and pro-duce a depolarizing IPSP (Figure 5.19C). However, the synapse would stillbe inhibitory: Given that the reversal potential of the IPSP still is more nega-tive than the action potential threshold (–40 mV), the depolarizing IPSPwould inhibit because the postsynaptic membrane potential would be keptmore negative than the threshold for action potential initiation. Another wayto think about this peculiar situation is that if another excitatory input ontothis neuron brought the cell’s membrane potential to –41 mV, just belowthreshold for firing an action potential, the IPSP would then hyperpolarizethe membrane potential toward –50 mV, bringing the potential away fromthe action potential threshold. Thus, while EPSPs depolarize the postsynap-tic cell, IPSPs can hyperpolarize or depolarize; indeed, an inhibitory conduc-tance change may produce no potential change at all and still exert aninhibitory effect by making it more difficult for an EPSP to evoke an actionpotential in the postsynaptic cell.Although the particulars of postsynaptic action can be complex, a simplerule distinguishes postsynaptic excitation from inhibition: An EPSP has areversal potential more positive than the action potential threshold, whereasThresholdmVTime (ms)−60−40−50−70−1100 1 2 3 4 0 1 2 3 4 0 1 2 3 4+500Activate GABA synapseActivate GABA synapseErevErevErevVrestENa(A) (B) (C) (D)EKActivate glutamatesynapseActionpotentialEPSPIPSPIPSPErev > threshold=excitatoryErev < threshold=inhibitoryFigure 5.19 Reversal potentials andthreshold potentials determine postsyn-aptic excitation and inhibition. (A) If thereversal potential for a PSP (0 mV) ismore positive than the action potentialthreshold (–40 mV), the effect of a trans-mitter is excitatory, and it generatesEPSPs. (B) If the reversal potential for aPSP is more negative than the actionpotential threshold, the transmitter isinhibitory and generate IPSPs. (C) IPSPscan nonetheless depolarize the postsyn-aptic cell if their reversal potential isbetween the resting potential and theaction potential threshold. (D) The gen-eral rule of postsynaptic action is: If thereversal potential is more positive thanthreshold, excitation results; inhibitionoccurs if the reversal potential is morenegative than threshold.Purves05 5/13/04 2:27 PM Page 122
    • Figure 5.20 Summation of postsynap-tic potentials. (A) A microelectroderecords the postsynaptic potentials pro-duced by the activity of two excitatorysynapses (E1 and E2) and an inhibitorysynapse (I). (B) Electrical responses tosynaptic activation. Stimulating eitherexcitatory synapse (E1 or E2) produces asubthreshold EPSP, whereas stimulatingboth synapses at the same time (E1 +E2) produces a suprathreshold EPSPthat evokes a postsynaptic action poten-tial (shown in blue). Activation of theinhibitory synapse alone (I) results in ahyperpolarizing IPSP. Summing thisIPSP (dashed red line) with the EPSP(dashed yellow line) produced by oneexcitatory synapse (E1 + I) reduces theamplitude of the EPSP (orange line),while summing it with the suprathresh-old EPSP produced by activating syn-apses E1 and E2 keeps the postsynapticneuron below threshold, so that noaction potential is evoked.an IPSP has a reversal potential more negative than threshold (Figure 5.19D).Intuitively, this rule can be understood by realizing that an EPSP will tend todepolarize the membrane potential so that it exceeds threshold, whereas anIPSP will always act to keep the membrane potential more negative than thethreshold potential.Summation of Synaptic PotentialsThe PSPs produced at most synapses in the brain are much smaller thanthose at the neuromuscular junction; indeed, EPSPs produced by individualexcitatory synapses may be only a fraction of a millivolt and are usually wellbelow the threshold for generating postsynaptic action potentials. How,then, can such synapses transmit information if their PSPs are subthreshold?The answer is that neurons in the central nervous system are typically inner-vated by thousands of synapses, and the PSPs produced by each active syn-apse can sum together—in space and in time—to determine the behavior ofthe postsynaptic neuron.Consider the highly simplified case of a neuron that is innervated by twoexcitatory synapses, each generating a subthreshold EPSP, and an inhibitorysynapse that produces an IPSP (Figure 5.20A). While activation of either oneof the excitatory synapses alone (E1 or E2 in Figure 5.20B) produces a sub-Synaptic Transmission 123RecordExcitatory (E1)Inhibitory (I)Excitatory (E2)CellbodyDendrites(A)(B)VrestTime (ms)−40−60−20+200EPSP (SynapseE1 or E2)Summed EPSPs(Synapses E1 + E2)IPSP(Synapse I)ThresholdPostsynapticmembranepotential(mV)SummedEPSP + IPSP(SynapsesE1 + I)SummedEPSPs + IPSP(SynapsesE1 + I +E2)PostsynapticmembranepotentialAxonPurves05 5/13/04 2:27 PM Page 123
    • 124 Chapter FiveFigure 5.21 Events from neurotrans-mitter release to postsynaptic excitationor inhibition. Neurotransmitter releaseat all presynaptic terminals on a cellresults in receptor binding, whichcauses the opening or closing of specificion channels. The resulting conductancechange causes current to flow, whichmay change the membrane potential.The postsynaptic cell sums (or inte-grates) all of the EPSPs and IPSPs,resulting in moment-to-moment controlof action potential generation.threshold EPSP, activation of both excitatory synapses at about the sametime causes the two EPSPs to sum together. If the sum of the two EPSPs (E1+ E2) depolarizes the postsynaptic neuron sufficiently to reach the thresholdpotential, a postsynaptic action potential results. Summation thus allowssubthreshold EPSPs to influence action potential production. Likewise, anIPSP generated by an inhibitory synapse (I) can sum (algebraically speaking)with a subthreshold EPSP to reduce its amplitude (E1 + I) or can sum withsuprathreshold EPSPs to prevent the postsynaptic neuron from reachingthreshold (E1 + I + E2).In short, the summation of EPSPs and IPSPs by a postsynaptic neuronpermits a neuron to integrate the electrical information provided by all theinhibitory and excitatory synapses acting on it at any moment. Whether thesum of active synaptic inputs results in the production of an action potentialdepends on the balance between excitation and inhibition. If the sum of allEPSPs and IPSPs results in a depolarization of sufficient amplitude to raisethe membrane potential above threshold, then the postsynaptic cell will pro-duce an action potential. Conversely, if inhibition prevails, then the postsyn-aptic cell will remain silent. Normally, the balance between EPSPs and IPSPschanges continually over time, depending on the number of excitatory andinhibitory synapses active at a given moment and the magnitude of the cur-rent at each active synapse. Summation is therefore a neurotransmitter-induced tug-of-war between all excitatory and inhibitory postsynaptic cur-rents; the outcome of the contest determines whether or not a postsynapticneuron fires an action potential and, thereby, becomes an active element inthe neural circuits to which it belongs (Figure 5.21).Two Families of Postsynaptic ReceptorsThe opening or closing of postsynaptic ion channels is accomplished in dif-ferent ways by two broad families of receptor proteins. The receptors in onefamily—called ionotropic receptors—are linked directly to ion channels (theGreek tropos means to move in response to a stimulus). These receptors con-tain two functional domains: an extracellular site that binds neurotransmit-ters, and a membrane-spanning domain that forms an ion channel (Figure5.22A). Thus ionotropic receptors combine transmitter-binding and channelfunctions into a single molecular entity (they are also called ligand-gatedion channels to reflect this concatenation). Such receptors are multimersmade up of at least four or five individual protein subunits, each of whichcontributes to the pore of the ion channel.The second family of neurotransmitter receptors are the metabotropicreceptors, so called because the eventual movement of ions through a chan-nel depends on one or more metabolic steps. These receptors do not have ionchannels as part of their structure; instead, they affect channels by the activa-tion of intermediate molecules called G-proteins (Figure 5.22B). For this rea-son, metabotropic receptors are also called G-protein-coupled receptors.Metabotropic receptors are monomeric proteins with an extracellular domainthat contains a neurotransmitter binding site and an intracellular domain thatbinds to G-proteins. Neurotransmitter binding to metabotropic receptors acti-vates G-proteins, which then dissociate from the receptor and interact directlywith ion channels or bind to other effector proteins, such as enzymes, thatmake intracellular messengers that open or close ion channels. Thus, G-pro-teins can be thought of as transducers that couple neurotransmitter bindingto the regulation of postsynaptic ion channels. The postsynaptic signalingevents initiated by metabotropic receptors are taken up in detail in Chapter 7.Postsynapticcells excitedor inhibitedPostsynapticpotentialchangesConductancechange causescurrent flowIon channelsopen or closeReceptorbindingNeurotransmitterreleaseSummation determineswhether or not anaction potential occursPurves05 5/13/04 2:27 PM Page 124
    • These two families of postsynaptic receptors give rise to PSPs with verydifferent time courses, producing postsynaptic actions that range from lessthan a millisecond to minutes, hours, or even days. Ionotropic receptors gen-erally mediate rapid postsynaptic effects. Examples are the EPP produced atneuromuscular synapses by ACh (see Figure 5.15), EPSPs produced at cer-tain glutamatergic synapses (Figure 5.19A), and IPSPs produced at certainGABAergic synapses (Figure 5.19B). In all three cases, the PSPs arise withina millisecond or two of an action potential invading the presynaptic terminaland last for only a few tens of milliseconds or less. In contrast, the activationof metabotropic receptors typically produces much slower responses, rang-ing from hundreds of milliseconds to minutes or even longer. The compara-tive slowness of metabotropic receptor actions reflects the fact that multipleproteins need to bind to each other sequentially in order to produce the finalphysiological response. Importantly, a given transmitter may activate bothionotropic and metabotropic receptors to produce both fast and slow PSPs atthe same synapse.Perhaps the most important principle to keep in mind is that the responseelicited at a given synapse depends upon the neurotransmitter released andthe postsynaptic complement of receptors and associated channels. The mol-ecular mechanisms that allow neurotransmitters and their receptors to gen-erate synaptic responses are considered in the next chapter.Synaptic Transmission 125NeurotransmitterbindsNeurotransmitterOutside cellInside cellIons(A) Ligand-gated ion channels (B) G-protein-coupled receptors1NeurotransmitterbindsNeurotrans-mitterIonsIons flowacross membraneG-protein isactivatedG-proteinReceptorG-protein subunits orintracellular messengersmodulate ion channelsIntracellularmessengersβγEffector proteinα3 312αChannelopens2Ions flowacrossmembrane5Ionchannelopens4Figure 5.22 A neurotransmitter can affect the activity of a postsynaptic cell via twodifferent types of receptor proteins: ionotropic or ligand-gated ion channels, andmetabotropic or G-protein-coupled receptors. (A) Ligand-gated ion channels com-bine receptor and channel functions in a single protein complex. (B) Metabotropicreceptors usually activate G-proteins, which modulate ion channels directly or indi-rectly through intracellular effector enzymes and second messengers.Purves05 5/13/04 2:27 PM Page 125
    • 126 Chapter FiveSummarySynapses communicate the information carried by action potentials fromone neuron to the next in neural circuits. The cellular mechanisms thatunderlie synaptic transmission are closely related to the mechanisms thatgenerate other types of neuronal electrical signals, namely ion flow throughmembrane channels. In the case of electrical synapses, these channels aregap junctions; direct but passive flow of current through the gap junctions isthe basis for transmission. In the case of chemical synapses, channels withsmaller and more selective pores are activated by the binding of neurotrans-mitters to postsynaptic receptors after release from the presynaptic terminal.The large number of neurotransmitters in the nervous system can be dividedinto two broad classes: small-molecule transmitters and neuropeptides. Neu-rotransmitters are synthesized from defined precursors by regulated enzy-matic pathways, packaged into one of several types of synaptic vesicle, andreleased into the synaptic cleft in a Ca2+-dependent manner. Many synapsesrelease more than one type of neurotransmitter, and multiple transmitterscan even be packaged within the same synaptic vesicle. Transmitter agentsare released presynaptically in units or quanta, reflecting their storagewithin synaptic vesicles. Vesicles discharge their contents into the synapticcleft when the presynaptic depolarization generated by the invasion of anaction potential opens voltage-gated calcium channels, allowing Ca2+toenter the presynaptic terminal. How calcium triggers neurotransmitterrelease is not yet established, but synaptotagmin, SNAREs, and a number ofother proteins found within the presynaptic terminal are clearly involved.Postsynaptic receptors are a diverse group of proteins that transduce bind-ing of neurotransmitters into electrical signals by opening or closing post-synaptic ion channels. The postsynaptic currents produced by the synchro-nous opening or closing of ion channels changes the conductance of thepostsynaptic cell, thus increasing or decreasing its excitability. Conductancechanges that increase the probability of firing an action potential are excita-tory, whereas those that decrease the probability of generating an actionpotential are inhibitory. Because postsynaptic neurons are usually innervatedby many different inputs, the integrated effect of the conductance changesunderlying all EPSPs and IPSPs produced in a postsynaptic cell at anymoment determines whether or not the cell fires an action potential. Twobroadly different families of neurotransmitter receptors have evolved tocarry out the postsynaptic signaling actions of neurotransmitters. The post-synaptic effects of neurotransmitters are terminated by the degradation ofthe transmitter in the synaptic cleft, by transport of the transmitter back intocells, or by diffusion out of the synaptic cleft.Purves05 5/13/04 2:27 PM Page 126
    • Additional ReadingReviewsAUGUSTINE, G. J. (2001) How does calciumtrigger neurotransmitter release? Curr. Opin.Neurobiol. 11: 320–326.BENNETT, M. V. L. (2000) Electrical synapses, apersonal perspective (or history). Brain Res.Rev. 32: 16–28.BRODSKY, F. M., C. Y. CHEN, C. KNUEHL, M. C.TOWLER AND D. E. WAKEHAM (2001) Biologicalbasket weaving: Formation and function ofclathrin-coated vesicles. Annu. Rev. Cell. Dev.Biol. 17: 517–568.BRUNGER, A. T. (2001) Structure of proteinsinvolved in synaptic vesicle fusion in neurons.Annu. Rev. Biophys. Biomol. Struct. 30:157–171.CARLSSON, A. (1987) Perspectives on the dis-covery of central monoaminergic neurotrans-mission. Annu. Rev. Neurosci. 10: 19–40.CHANGEUX, J.-P. (1993) Chemical signaling inthe brain. Sci. Am. 269 (May): 58–62.EMSON, P. C. (1979) Peptides as neurotransmit-ter candidates in the CNS. Prog. Neurobiol.13: 61–116.GALARRETA, M. AND S. HESTRIN (2001) Electri-cal synapses between GABA-releasinginterneurons. Nature Rev. Neurosci. 2:425–433.JAHN, R., T. LANG AND T. C. SÜDHOF (2003)Membrane fusion. Cell 112: 519–533.KUPFERMANN, I. (1991) Functional studies ofcotransmission. Physiol. Rev. 71: 683–732.MARSH, M. AND H. T. MCMAHON (1999) Thestructural era of endocytosis. Science 285:215–220.MURTHY, V. N. AND P. DE CAMILLI (2003) Cellbiology of the presynaptic terminal. Annu.Rev. Neurosci. 26: 701–728.ROTHMAN, J. E. (1994) Mechanisms of intracel-lular protein transport. Nature 372: 55–63.SÜDHOF, T. C. (1995) The synaptic vesicle cycle:A cascade of protein-protein interactions.Nature 375: 645–653.TUCKER, W. C. AND E. R. CHAPMAN (2002) Roleof synaptotagmin in Ca2+triggered exocyto-sis. Biochem. J. 366: 1–13.Important Original PapersADLER, E., M. ADLER, G. J. AUGUSTINE, M. P.CHARLTON AND S. N. DUFFY (1991) Alien intra-cellular calcium chelators attenuate neuro-transmitter release at the squid giant synapse.J. Neurosci. 11: 1496–1507.AUGUSTINE, G. J. AND R. ECKERT (1984) Diva-lent cations differentially support transmitterrelease at the squid giant synapse. J. Physiol.(Lond.) 346: 257–271.BOYD, I. A. AND A. R. MARTIN (1955) The end-plate potential in mammalian muscle. J. Phys-iol. (Lond.) 132: 74–91.CURTIS, D. R., J. W. PHILLIS AND J. C. WATKINS(1959) Chemical excitation of spinal neurons.Nature 183: 611–612.DALE, H. H., W. FELDBERG AND M. VOGT (1936)Release of acetylcholine at voluntary motornerve endings. J. Physiol. 86: 353–380.DEL CASTILLO, J. AND B. KATZ (1954) Quantalcomponents of the end plate potential. J.Physiol. (Lond.) 124: 560–573.FATT, P. AND B. KATZ (1951) An analysis of theend plate potential recorded with an intracel-lular electrode. J. Physiol. (Lond.) 115:320–370.FATT, P. AND B. KATZ (1952) Spontaneous sub-threshold activity at motor nerve endings. J.Physiol. (Lond.) 117: 109–128.FURSHPAN, E. J. AND D. D. POTTER (1959) Trans-mission at the giant motor synapses of thecrayfish. J. Physiol. (Lond.) 145: 289–325.GEPPERT, M. AND 6 OTHERS (1994) Synaptotag-min I: A major Ca2+sensor for transmitterrelease at a central synapse. Cell 79: 717–727.GIBSON, J. R., M. BEIERLEIN AND B. W. CONNORS.(1999) Two networks of electrically coupledinhibtory neurons in neocortex. Nature 402:75–79.HARRIS, B. A., J. D. ROBISHAW, S. M. MUMBYAND A. G. GILMAN (1985) Molecular cloning ofcomplementary DNA for the alpha subunit ofthe G protein that stimulates adenylatecyclase. Science 229: 1274–1277.HEUSER, J. E. AND 5 OTHERS (1979) Synapticvesicle exocytosis captured by quick freezingand correlated with quantal transmitterrelease. J. Cell Biol. 81: 275–300.HEUSER, J. E. AND T. S. REESE (1973) Evidencefor recycling of synaptic vesicle membraneduring transmitter release at the frog neuro-muscular junction. J. Cell Biol. 57: 315–344.HÖKFELT, T., O. JOHANSSON, A. LJUNGDAHL, J.M. LUNDBERG AND M. SCHULTZBERG (1980) Pep-tidergic neurons. Nature 284: 515–521.JONAS, P., J. BISCHOFBERGER AND J. SANDKUHLER(1998) Corelease of two fast neurotransmittersat a central synapse. Science 281: 419–424.LOEWI, O. (1921) Über humorale übertrag-barheit der herznervenwirkung. PflügersArch. 189: 239–242.MILEDI, R. (1973) Transmitter release inducedby injection of calcium ions into nerve termi-nals. Proc. R. Soc. Lond. B 183: 421–425.NEHER, E. AND B. SAKMANN (1976) Single-channel currents recorded from membrane ofdenervated frog muscle fibres. Nature260:799-802.REKLING, J. C., X. M. SHAO AND J. L. FELDMAN(2000) Electrical coupling and excitatory syn-aptic transmission between rhythmogenic res-piratory neurons in the preBotzinger complex.J. Neurosci. 20: RC113: 1–5.SMITH, S. J., J. BUCHANAN, L. R. OSSES, M. P.CHARLTON AND G. J. AUGUSTINE (1993) Thespatial distribution of calcium signals in squidpresynaptic terminals. J. Physiol. (Lond.) 472:573–593.SOSSIN, W. S., A. SWEET-CORDERO AND R. H.SCHELLER (1990) Dale’s hypothesis revisited:Different neuropeptides derived from a com-mon prohormone are targeted to differentprocesses. Proc. Natl. Acad. Sci. U.S.A. 87:4845–4548.SUTTON, R. B., D. FASSHAUER, R. JAHN AND A. T.BRÜNGER (1998) Crystal structure of a SNAREcomplex involved in synaptic exocytosis at 2.4Å resolution. Nature 395: 347–353.TAKEUCHI, A. AND N. TAKEUCHI (1960) One thepermeability of end-plate membrane duringthe action of transmitter. J. Physiol. (Lond.)154: 52–67.WICKMAN, K. AND 7 OTHERS (1994) Recombi-nant Gβγ activates the muscarinic-gated atrialpotassium channel IKACh. Nature 368: 255–257.BooksBRADFORD, H. F. (1986) Chemical Neurobiology.New York: W. H. Freeman.COOPER, J. R., F. E. BLOOM AND R. H. ROTH(1991) The Biochemical Basis of Neuropharmacol-ogy. New York: Oxford University Press.HALL, Z. (1992) An Introduction to MolecularNeurobiology. Sunderland, MA: Sinauer Asso-ciates.KATZ, B. (1966) Nerve, Muscle, and Synapse.New York: McGraw-Hill.KATZ, B. (1969) The Release of Neural Transmit-ter Substances. Liverpool: Liverpool UniversityPress.LLINÁS, R. R. (1999) The Squid Giant Synapse: AModel for Chemical Synaptic Transmission.Oxford: Oxford University Press.NICHOLLS, D. G. (1994) Proteins, Transmitters,and Synapses. Oxford: Blackwell.PETERS, A., S. L. PALAY AND H. DEF. WEBSTER(1991) The Fine Structure of the Nervous System:Neurons and their Supporting Cells. 3rd edition.Oxford: Oxford University Press.Synaptic Transmission 127Purves05 5/13/04 2:27 PM Page 127
    • Purves05 5/13/04 2:27 PM Page 128
    • OverviewFor the most part, neurons in the human brain communicate with oneanother by releasing chemical messengers called neurotransmitters. A largenumber of neurotransmitters are now known and more remain to be discov-ered. Neurotransmitters evoke postsynaptic electrical responses by bindingto members of a diverse group of proteins called neurotransmitter receptors.There are two major classes of receptors: those in which the receptor mole-cule is also an ion channel, and those in which the receptor and ion channelare separate molecules. The former are called ionotropic receptors or ligand-gated ion channels, and give rise to fast postsynaptic responses that typicallylast only a few milliseconds. The latter are called metabotropic receptors,and they produce slower postsynaptic effects that may endure much longer.Abnormalities in the function of neurotransmitter systems contribute to awide range of neurological and psychiatric disorders. As a result, many neu-ropharmacological therapies are based on drugs that affect neurotransmitterrelease, binding, and/or removal.Categories of NeurotransmittersMore than 100 different agents are known to serve as neurotransmitters. Thislarge number of transmitters allows for tremendous diversity in chemicalsignaling between neurons. It is useful to separate this panoply of transmit-ters into two broad categories based simply on size (Figure 6.1). Neuropep-tides are relatively large transmitter molecules composed of 3 to 36 aminoacids. Individual amino acids, such as glutamate and GABA, as well as thetransmitters acetylcholine, serotonin, and histamine, are much smaller thanneuropeptides and have therefore come to be called small-molecule neuro-transmitters. Within the category of small-molecule neurotransmitters, thebiogenic amines (dopamine, norepinephrine, epinephrine, serotonin, andhistamine) are often discussed separately because of their similar chemicalproperties and postsynaptic actions. The particulars of synthesis, packaging,release, and removal differ for each neurotransmitter (Table 6.1). This chap-ter will describe some of the main features of these transmitters and theirpostsynaptic receptors.AcetylcholineAs mentioned in the previous chapter, acetylcholine (ACh) was the first sub-stance identified as a neurotransmitter. In addition to the action of ACh asthe neurotransmitter at skeletal neuromuscular junctions (see Chapter 5), aswell as the neuromuscular synapse between the vagus nerve and cardiacChapter 6129Neurotransmittersand TheirReceptorsPurves06 5/13/04 2:53 PM Page 129
    • muscle fibers, ACh serves as a transmitter at synapses in the ganglia of thevisceral motor system, and at a variety of sites within the central nervoussystem. Whereas a great deal is known about the function of cholinergictransmission at neuromuscular junctions and ganglionic synapses, theactions of ACh in the central nervous system are not as well understood.Acetylcholine is synthesized in nerve terminals from the precursorsacetyl coenzyme A (acetyl CoA, which is synthesized from glucose) andcholine, in a reaction catalyzed by choline acetyltransferase (CAT; Figure6.2). Choline is present in plasma at a high concentration (about 10 mM)and is taken up into cholinergic neurons by a high-affinity Na+/cholinetransporter. After synthesis in the cytoplasm of the neuron, a vesicular AChNeurotransmitters and Their Receptors 131Figure 6.1 Examples of small-molecule and peptide neurotransmitters. Small-mol-ecule transmitters can be subdivided into acetylcholine, the amino acids, purines,and biogenic amines. The catcholamines, so named because they all share the cate-chol moiety (i.e., a hydroxylated benzene ring), make up a distinctive subgroupwithin the biogenic amines. Serotonin and histamine contain an indole ring and animidazole ring, respectively. Size differences between the small-molecule neuro-transmitters and the peptide neurotransmitters are indicated by the space-fillingmodels for glycine, norepinephrine, and methionine enkephalin. (Carbon atoms areblack, nitrogen atoms blue, and oxygen atoms red.)▲TABLE 6.1Functional Features of the Major NeurotransmittersPostsynaptic Rate-limiting Removal Type ofNeurotransmitter effectaPrecursor(s) step in synthesis mechanism vesicleACh Excitatory Choline + CAT AChEase Small, clearacetyl CoAGlutamate Excitatory Glutamine Glutaminase Transporters Small, clearGABA Inhibitory Glutamate GAD Transporters Small, clearGlycine Inhibitory Serine Phosphoserine Transporters Small, clearCatecholamines Excitatory Tyrosine Tyrosine Transporters, Small dense-(epinephrine, hydroxylase MAO, COMT core,norepinephrine, or largedopamine) irregulardense-coreSerotonin (5-HT) Excitatory Tryptophan Tryptophan Transporters, Large,hydroxylase MAO dense-coreHistamine Excitatory Histidine Histidine Transporters Large,decarboxylase dense-coreATP Excitatory ADP Mitochondrial Hydrolysis to Small, clearoxidative phosphor- AMP andylation; glycolysis adenosineNeuropeptides Excitatory Amino acids Synthesis and Proteases Large,and inhibitory (protein synthesis) transport dense-coreEndocannabinoids Inhibits Membrane lipids Enzymatic Hydrolasis Noneinhibition modification of lipids by FAAHNitric oxide Excitatory and Arginine Nitric oxide synthase Spontaneous Noneinhibitory oxidationaThe most common postsynaptic effect is indicated; the same transmitter can elicit postsynaptic excitation or inhibition depending on the nature of the ion channelsaffected by transmitter binding (see Chapter 7).Purves06 5/13/04 2:53 PM Page 131
    • 132 Chapter Sixtransporter loads approximately 10,000 molecules of ACh into each cholin-ergic vesicle.In contrast to most other small-molecule neurotransmitters, the postsynap-tic actions of ACh at many cholinergic synapses (the neuromuscular junctionin particular) is not terminated by reuptake but by a powerful hydrolyticenzyme, acetylcholinesterase (AChE). This enzyme is concentrated in thesynaptic cleft, ensuring a rapid decrease in ACh concentration after its releasefrom the presynaptic terminal. AChE has a very high catalytic activity (about5000 molecules of ACh per AChE molecule per second) and hydrolyzes AChinto acetate and choline. The choline produced by ACh hydrolysis is trans-ported back into nerve terminals and used to resynthesize ACh.Among the many interesting drugs that interact with cholinergic enzymesare the organophosphates. This group includes some potent chemical warfareagents. One such compound is the nerve gas “Sarin,” which was made notori-ous after a group of terrorists released this gas in Tokyo’s underground rail sys-tem. Organophosphates can be lethal because they inhibit AChE, causing AChto accumulate at cholinergic synapses. This build-up of ACh depolarizes thepostsynaptic cell and renders it refractory to subsequent ACh release, causingneuromuscular paralysis and other effects. The high sensitivity of insects tothese AChE inhibitors has made organophosphates popular insecticides.Many of the postsynaptic actions of ACh are mediated by the nicotinicACh receptor (nAChR), so named because the CNS stimulant, nicotine, alsoAcetyl CoAAcetylcholineAcetylcholineAcetylcholinereceptorsNa+/cholinetransporterCholine acetyl-transferaseO+O+Choline++Acetate CholineCholineAcetylcholinesterasePresynapticterminalPostsynapticcellCoA (CH3)3NCH2CH2HOCH3CS +(CH3)3NCH2 CH2CH3 OC(CH3)3NCH2CH2HOGlucosePyruvateCH3OO−CVesicularAChtransporterFigure 6.2 Acetylcholine metabolismin cholinergic nerve terminals. The syn-thesis of acetylcholine from choline andacetyl CoA requires choline acetyltrans-ferase. Acetyl CoA is derived from pyru-vate generated by glycolysis, whilecholine is transported into the terminalsvia a Na+-dependent transporter.Acetylcholine is loaded into synapticvesicles via a vesicular transporter. Afterrelease, acetylcholine is rapidly metabo-lized by acetylcholinesterase, andcholine is transported back into theterminal.Purves06 5/13/04 2:53 PM Page 132
    • Neurotransmitters and Their Receptors 133ReceptorMembrane2 nm3 nm6.5 nmα αγ δ(A) (B) (C)(D)NCOutsidecellInsidecell3 nmAChAChαβγαδFigure 6.3 The structure of the nACh receptor/channel. (A) Each receptor sub-unit crosses the membrane four times. The membrane-spanning domain that linesthe pore is shown in blue. (B) Five such subunits come together to form a complexstructure containing 20 transmembrane domains that surround a central pore. (C)The openings at either end of the channel are very large—approximately 3 nm indiameter; even the narrowest region of the pore is approximately 0.6 nm in diame-ter. By comparison, the diameter of Na+or K+is less than 0.3 nm. (D) An electronmicrograph of the nACh receptor, showing the actual position and size of the pro-tein with respect to the membrane. (D from Toyoshima and Unwin, 1990.)binds to these receptors. Nicotine consumption produces some degree ofeuphoria, relaxation, and eventually addiction (Box A), effects believedto be mediated in this case by nAChRs. Nicotinic receptors are the best-studied type of ionotropic neurotransmitter receptor. As described inChapter 5, nAChRs are nonselective cation channels that generate exci-tatory postsynaptic responses. A number of biological toxins specificallybind to and block nicotinic receptors (Box B). The availability of thesehighly specific ligands—particularly a component of snake venom calledα-bungarotoxin—has provided a valuable way to isolate and purifynAChRs. This pioneering work paved the way to cloning and sequenc-ing the genes encoding the various subunits of the nAChR.Based on these molecular studies, the nAChR is now known to be alarge protein complex consisting of five subunits arranged around a cen-tral membrane-spanning pore (Figure 6.3). In the case of skeletal muscleAChRs, the receptor pentamer contains two α subunits, each of whichbinds one molecule of ACh. Because both ACh binding sites must beoccupied for the channel to open, only relatively high concentrations ofthis neurotransmitter lead to channel activation. These subunits also bindother ligands, such as nicotine and α-bungarotoxin. At the neuromuscu-lar junction, the two α subunits are combined with up to four other typesof subunit—β, γ, δ, ε—in the ratio 2α:β:ε:δ. Neuronal nAChRs typicallydiffer from those of muscle in that they lack sensitivity to α-bungaro-Purves06 5/13/04 2:53 PM Page 133
    • 134 Chapter Sixtoxin, and comprise only two receptor subunit types (α and β), which arepresent in a ratio of 3α:2β. In all cases, however, five individual subunitsassemble to form a functional, cation-selective nACh receptor.Each subunit of the nAChR molecule contains four transmembranedomains that make up the ion channel portion of the receptor, and a longextracellular region that makes up the ACh-binding domain (Figure 6.3A).Unraveling the molecular structure of this region of the nACh receptor hasprovided insight into the mechanisms that allow ligand-gated ion channelsto respond rapidly to neurotransmitters: The intimate association of theACh binding sites with the pore of the channel presumably accounts forthe rapid response to ACh (Figure 6.3B–D). Indeed, this general arrange-ment is characteristic of all of the ligand-gated ion channels at fast-actingsynapses, as summarized in Figure 6.4. Thus, the nicotinic receptor hasserved as a paradigm for studies of other ligand-gated ion channels, at thesame time leading to a much deeper appreciation of several neuromusculardiseases (Box C).Box AAddictionDrug addiction is a chronic, relapsingdisease with obvious medical, social,and political consequences. Addiction(also called substance dependence) is apersistent disorder of brain function inwhich compulsive drug use occursdespite serious negative consequencesfor the afflicted individual. The diag-nostic manual of the American Psychi-atric Association defines addiction interms of both physical dependence andpsychological dependence (in which anindividual continues the drug-takingbehavior despite obviously maladaptiveconsequences).The range of substances that cangenerate this sort of dependence iswide; the primary agents of abuse atpresent are opioids, cocaine, ampheta-mines, marijuana, alcohol, and nicotine.Addiction to more “socially acceptable”agents such as alcohol and nicotine aresometimes regarded as less problem-atic, but in fact involve medical andbehavioral consequences that are atleast as great as for drugs of abuse thatare considered more dangerous. Impor-tantly, the phenomenon of addiction isnot limited to human behavior, but isdemonstrable in laboratory animals.Most of these same agents are self-administered if primates, rodents, orother species are provided with theopportunity to do so.In addition to a compulsion to takethe agent of abuse, a major feature ofaddiction for many drugs is a constella-tion of negative physiological and emo-tional features, loosely referred to as“withdrawal syndrome,” that occurwhen the drug is not taken. The signsand symptoms of withdrawal are differ-ent for each agent of abuse, but in gen-eral are characterized by effects oppo-site those of the positive experienceinduced by the drug itself. Consider, asan example, cocaine, a drug that wasestimated to be in regular use by 5 to 6million Americans during the decade ofthe 1990s, with about 600,000 regularusers either addicted or at high risk foraddiction. The positive effects of thedrug smoked or inhaled as a powder inthe form of the alkaloidal free base is a“high” that is nearly immediate butgenerally lasts only a few minutes, typi-cally leading to a desire for additionaldrug in as little as 10 minutes to half anhour. The “high” is described as a feel-ing of well-being, self-confidence, andsatisfaction. Conversely, when the drugis not available, frequent users experi-ence depression, sleepiness, fatigue,drug-craving, and a general sense ofmalaise.Another aspect of addiction tococaine or other agents is tolerance,defined as a reduction in the responseto the drug upon repeated administra-tion. Tolerance occurs as a consequenceof persistent use of a number of drugsbut is particularly significant in drugaddiction, since it progressivelyincreases the dose needed to experiencethe desired effects.Although it is fair to say that the neu-robiology of addiction is incompletelyunderstood, for cocaine and many otheragents of abuse the addictive effectsinvolve activation of dopamine receptorsin critical brain regions involved in moti-vation and emotional reinforcement (seeChapter 28). The most important of theseareas is the midbrain dopamine system,Purves06 5/13/04 2:53 PM Page 134
    • A second class of ACh receptors is activated by muscarine, a poisonousalkaloid found in some mushrooms (see Box B), and thus they are referred toas muscarinic ACh receptors (mAChRs). mAChRs are metabotropic andmediate most of the effects of ACh in brain. Several subtypes of mAChR areknown (Figure 6.5). Muscarinic ACh receptors are highly expressed in thestriatum and various other forebrain regions, where they exert an inhibitoryinfluence on dopamine-mediated motor effects. These receptors are alsofound in the ganglia of the peripheral nervous system. Finally, they mediateperipheral cholinergic responses of autonomic effector organs—such asheart, smooth muscle, and exocrine glands—and are responsible for theinhibition of heart rate by the vagus nerve. Numerous drugs act as mAChreceptor agonists or antagonists, but most of these do not discriminatebetween different types of muscarinic receptors and often produce sideeffects. Nevertheless, mACh blockers that are therapeutically useful includeatropine (used to dilate the pupil), scopolamine (effective in preventingmotion sickness), and ipratropium (useful in the treatment of asthma).Neurotransmitters and Their Receptors 135especially its projections from the ven-tral-tegmental area to the nucleus acum-bens. Agents such as cocaine appear toact by raising dopamine levels in theseareas, making this transmitter moreavailable to receptors by interfering withre-uptake of synaptically releaseddopamine by the dopamine transporter.The reinforcement and motivation ofdrug-taking behaviors is thought to berelated to the projections to the nucleusacumbens.The most common opioid drug ofabuse is heroin. Heroin is a derivativeof the opium poppy and is not legallyavailable for clinical purposes in theUnited States. The number of heroinaddicts in the United States is estimatedto be between 750,000 and a millionindividuals. The positive feelings pro-duced by heroin, generally described asthe “rush,” are often compared to thefeeling of sexual orgasm and begin inless than a minute after intravenousinjection. There is then a feeling of gen-eral well-being (referred to as “on thenod”) that lasts about an hour. Thesymptoms of withdrawal can beintense; these are restlessness, irritabil-ity, nausea, muscle pain, depression,sleeplessness, and a sense of anxietyand malaise. The reinforcing aspects ofthe drug entail the same dopaminergiccircuitry in the ventral tegmental areaand nucleus acumbens as does cocaine,although additional areas are certainlyinvolved, particularly the sites of opioidreceptors described in Chapter 9.Interestingly, addiction to heroin orany other agent is not an inevitable con-sequence of drug use, but depends criti-cally on the environment. For instance,returning veterans who were heroinaddicts in Vietnam typically lost theiraddiction upon returning to the UnitedStates. Likewise, patients given otheropioids (e.g., morphine) for painful con-ditions rarely become addicts.The treatment of any form of addic-tion is difficult and must be tailored tothe circumstances of the individual. Inaddition to treating acute problems ofwithdrawal and “detoxification,” pat-terns of behavior must be changed thatmay take months or years. Addiction isthus a chronic disease state that requirescontinual monitoring during the life-time of susceptible individuals.ReferencesAMERICAN PSYCHIATRIC ASSOCIATION (1994)Diagnostic and Statistical Manual of MentalDisorders, 4thEdition (DSM IV). Washington,D.C.HYMAN, S. E. AND R. C. MALENKA (2001)Addiction and the brain: The neurobiologyof compulsion and its persistence. NatureRev. Neurosci. 2: 695–703.LAAKSO, A., A. R. MOHN, R. R. GAINETDINOVAND M. G. CARON (2002) Experimentalgenetic approaches to addiction. Neuron 36:213–228.O’BRIEN, C. P. (2001) Goodman and Gilman’sThe Pharmaceutical Basis of Therapeutics, 10thEdition. New York: McGraw-Hill, Chapter24, pp. 621–642..Purves06 5/13/04 2:53 PM Page 135
    • 136 Chapter SixBox BNeurotoxins that Act on Postsynaptic ReceptorsPoisonous plants and venomous animalsare widespread in nature. The toxinsthey produce have been used for a vari-ety of purposes, including hunting, heal-ing, mind-altering, and, more recently,research. Many of these toxins havepotent actions on the nervous system,often interfering with synaptic transmis-sion by targeting neurotransmitter recep-tors. The poisons found in some organ-isms contain a single type of toxin,whereas others contain a mixture of tensor even hundreds of toxins.Given the central role of ACh recep-tors in mediating muscle contraction atneuromuscular junctions in numerousspecies, it is not surprising that a largenumber of natural toxins interfere withtransmission at this synapse. In fact, theclassification of nicotinic and muscarinicACh receptors is based on the sensitivityof these receptors to the toxic plant alka-loids nicotine and muscarine, which acti-vate nicotinic and muscarinic ACh recep-tors, respectively. Nicotine is derivedfrom the dried leaves of the tobacco plantNicotinia tabacum, and muscarine is fromthe poisonous red mushroom Amanitamuscaria. Both toxins are stimulants thatproduce nausea, vomiting, mental confu-sion, and convulsions. Muscarine poi-soning can also lead to circulatory col-lapse, coma, and death.The poison α-bungarotoxin, one ofmany peptides that together make upthe venom of the banded krait, Bungarusmulticinctus (Figure A), blocks transmis-sion at neuromuscular junctions and isused by the snake to paralyze its prey.This 74-amino-acid toxin blocks neuro-muscular transmission by irreversiblybinding to nicotinic ACh receptors, thuspreventing ACh from opening postsyn-aptic ion channels. Paralysis ensuesbecause skeletal muscles can no longerbe activated by motor neurons. As aresult of its specificity and its high affin-ity for nicotinic ACh receptors, α-bun-garotoxin has contributed greatly tounderstanding the ACh receptor mole-cule. Other snake toxins that block nico-tinic ACh receptors are cobra α-neuro-toxin and the sea snake peptide erabu-toxin. The same strategy used by thesesnakes to paralyze prey was adopted bySouth American Indians who usedcurare, a mixture of plant toxins fromChondodendron tomentosum, as an arrow-head poison to immobilize their quarry.Curare also blocks nicotinic ACh recep-tors; the active agent is the alkaloid δ-tubocurarine.Another interesting class of animaltoxins that selectively block nicotinicACh and other receptors includes thepeptides produced by fish-huntingmarine cone snails (Figure B). These col-orful snails kill small fish by “shooting”venomous darts into them. The venomcontains hundreds of peptides, known asthe conotoxins, many of which targetproteins that are important in synaptictransmission. There are conotoxin pep-tides that block Ca2+channels, Na+chan-nels, glutamate receptors, and ACh(B) (C)(A)(A) The banded krait Bungarus multicinctus.(B) A marine cone snail (Conus sp.) uses ven-omous darts to kill a small fish. (C) Betelnuts, Areca catechu, growing in Malaysia. (A,Robert Zappalorti/Photo Researchers, Inc.; B,Zoya Maslak and Baldomera Olivera, Uni-versity of Utah; C, Fletcher Baylis/PhotoResearchers, Inc.)Purves06 5/13/04 2:53 PM Page 136
    • Neurotransmitters and Their Receptors 137GlutamateGlutamate is the most important transmitter in normal brain function.Nearly all excitatory neurons in the central nervous system are glutamater-gic, and it is estimated that over half of all brain synapses release this agent.Glutamate plays an especially important role in clinical neurology becauseelevated concentrations of extracellular glutamate, released as a result ofneural injury, are toxic to neurons (Box D).Glutamate is a nonessential amino acid that does not cross the blood-brainbarrier and therefore must be synthesized in neurons from local precursors.The most prevalent precursor for glutamate synthesis is glutamine, which isreleased by glial cells. Once released, glutamine is taken up into presynapticreceptors. The array of physiologicalresponses produced by these peptides allserve to immobilize any prey unfortu-nate enough to encounter the cone snail.Many other organisms, including othermollusks, corals, worms and frogs, alsoutilize toxins containing specific blockersof ACh receptors.Other natural toxins have mind- orbehavior-altering effects and in somecases have been used for thousands ofyears by shamans and, more recently,physicians. Two examples are plant alka-loid toxins that block muscarinic AChreceptors: atropine from deadly night-shade (belladonna), and scopolaminefrom henbane. Because these plantsgrow wild in many parts of the world,exposure is not unusual, and poisoningby either toxin can also be fatal.Another postsynaptic neurotoxin that,like nicotine, is used as a social drug isfound in the seeds from the betel nut,Areca catechu (Figure C). Betel nut chew-ing, although unknown in the UnitedStates, is practiced by up to 25% of thepopulation in India, Bangladesh, Ceylon,Malaysia, and the Philippines. Chewingthese nuts produces a euphoria causedby arecoline, an alkaloid agonist of nico-tinic ACh receptors. Like nicotine, arecol-ine is an addictive central nervous sys-tem stimulant.Many other neurotoxins alter trans-mission at noncholinergic synapses. Forexample, amino acids found in certainmushrooms, algae, and seeds are potentglutamate receptor agonists. The excito-toxic amino acids kainate, from the redalga Digenea simplex, and quisqualate,from the seed of Quisqualis indica, areused to distinguish two families of non-NMDA glutamate receptors (see text).Other neurotoxic amino acid activatorsof glutamate receptors include ibotenicacid and acromelic acid, both found inmushrooms, and domoate, which occursin algae, seaweed, and mussels. Anotherlarge group of peptide neurotoxinsblocks glutamate receptors. Theseinclude the α-agatoxins from the funnelweb spider, NSTX-3 from the orb weaverspider, jorotoxin from the Joro spider,and β-philanthotoxin from wasp venom,as well as many cone snail toxins.All the toxins discussed so far targetexcitatory synapses. The inhibitoryGABA and glycine receptors, however,have not been overlooked by the exigen-cies of survival. Strychnine, an alkaloidextracted from the seeds of Strychnosnux-vomica, is the only drug known tohave specific actions on transmission atglycinergic synapses. Because the toxinblocks glycine receptors, strychnine poi-soning causes overactivity in the spinalcord and brainstem, leading to seizures.Strychnine has long been used commer-cially as a poison for rodents, althoughalternatives such as the anticoagulantcoumadin are now more popularbecause they are safer for humans. Neu-rotoxins that block GABAA receptorsinclude plant alkaloids such as bicu-culline from Dutchman’s breeches andpicrotoxin from Anamerta cocculus. Dield-rin, a commercial insecticide, also blocksthese receptors. These agents are, likestrychnine, powerful central nervoussystem stimulants. Muscimol, a mush-room toxin that is a powerful depressantas well as a hallucinogen, activatesGABAA receptors. A synthetic analogueof GABA, baclofen, is a GABAB agonistthat reduces EPSPs in some brainstemneurons and is used clinically to reducethe frequency and severity of musclespasms.Chemical warfare between specieshas thus given rise to a staggering arrayof molecules that target synapsesthroughout the nervous system.Although these toxins are designed todefeat normal synaptic transmission,they have also provided a set of power-ful tools to understand postsynapticmechanisms.ReferencesADAMS, M. E. AND B. M. OLIVERA (1994) Neu-rotoxins: Overview of an emerging researchtechnology. TINS 17: 151–155.HUCHO, F. AND Y. OVCHINNIKOV (1990) Toxinsas Tools in Neurochemistry. Berlin: Walter deGruyer.MYERS, R. A., L. J. CRUZ, J. E. RIVIER AND B. M.OLIVERA (1993) Conus peptides as chemicalprobes for receptors and ion channels. Chem.Rev. 93: 1923–1926.Purves06 5/13/04 2:53 PM Page 137
    • 138 Chapter Sixterminals and metabolized to glutamate by the mitochondrial enzyme gluta-minase (Figure 6.6). Glutamate can also be synthesized by transamination of2-oxoglutarate, an intermediate of the tricarboxylic acid cycle. Hence, some ofthe glucose metabolized by neurons can also be used for glutamate synthesis.The glutamate synthesized in the presynaptic cytoplasm is packaged intosynaptic vesicles by transporters, termed VGLUT. At least three differentVGLUT genes have been identified. Once released, glutamate is removedfrom the synaptic cleft by the excitatory amino acid transporters (EAATs).There are five different types of high-affinity glutamate transporters exist,some of which are present in glial cells and others in presynaptic terminals.Glutamate taken up by glial cells is converted into glutamine by the enzyme(A)NCNPore loopOutsideInsideCFour transmembrane helices Three transmembrane helicesplus pore loopAssembled subunits(B)TransmitterbindingsiteTransmitterSubunits(combi-nation of4 or 5requiredfor eachreceptortype)AMPA(C)ReceptorGlu R1Glu R2Glu R3Glu R4NMDANR1NR2ANR2BNR2CNR2DGABA Glycineα1−7β1−4γ1−4ερ1−3δα1α2α3α4βnAChKainateGlu R5Glu R6Glu R7KA1KA2α2−9β1−4γδPurinesP2X1P2X2P2X3P2X4P2X5P2X6P2X7Serotonin5-HT3Figure 6.4 The general architecture of ligand-gated receptors. (A) One of the sub-units of a complete receptor. The long N-terminal region forms the ligand-bindingsite, while the remainder of the protein spans the membrane either four times (left)or three times (right). (B) Assembly of either four or five subunits into a completereceptor. (C) A diversity of subunits come together to form functional ionotropicneurotransmitter receptors.Purves06 5/13/04 2:53 PM Page 138
    • Neurotransmitters and Their Receptors 139glutamine synthetase; glutamine is then transported out of the glial cells andinto nerve terminals. In this way, synaptic terminals cooperate with glialcells to maintain an adequate supply of the neurotransmitter. This overallsequence of events is referred to as the glutamate-glutamine cycle (see Fig-ure 6.6).Several types of glutamate receptors have been identified. Three of theseare ionotropic receptors called, respectively, NMDA receptors, AMPA recep-tors, and kainate receptors (Figure 6.4C). These glutamate receptors arenamed after the agonists that activate them: NMDA (N-methyl-D-aspartate),AMPA (α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate), and kainicacid. All of the ionotropic glutamate receptors are nonselective cation chan-nels similar to the nAChR, allowing the passage of Na+and K+, and in somecases small amounts of Ca2+. Hence AMPA, kainate, and NMDA receptoractivation always produces excitatory postsynaptic responses. Like otherionotropic receptors, AMPA/kainate and NMDA receptors are also formed(A)NCG-proteinbinding siteIIVVIIIIIIINeuro-transmitterbinding siteVVIReceptorsubtypeReceptorclassmGlu R1mGlu R5mGlu R2mGlu R3GlutamateClass IClass IImGlu R4mGlu R6mGlu R7mGlu R8Class IIIDopamine NE, Epi Histamine Serotonin Purines MuscarinicGABABGABAB R1GABAB R2D1AD1BD2H1 5-HT 1 A typeP typeA1A2aA2bA3P2xP2yP2zP2tP2u5-HT 35-HT 25-HT 45-HT 55-HT 65-HT 7H2H3M1M2M3M4M5β1β2β3α2α1D3D4(B)Figure 6.5 Structure and function of metabotropic receptors. (A) Thetransmembrane architecture of metabotropic receptors. These mono-meric proteins contain seven transmembrane domains. Portions ofdomains II, III, VI, and VII make up the neurotransmitter-bindingregion. G-proteins bind to both the loop between domains V and VIand to portions of the C-terminal region. (B) Varieties of metabotropicneurotransmitter receptors.Purves06 5/13/04 2:53 PM Page 139
    • 140 Chapter SixBox CMyasthenia Gravis: An Autoimmune Disease of Neuromuscular SynapsesMyasthenia gravis is a disease thatinterferes with transmission betweenmotor neurons and skeletal musclefibers and afflicts approximately 1 ofevery 200,000 people. Originallydescribed by the British physicianThomas Willis in 1685, the hallmark ofthe disorder is muscle weakness, partic-ularly during sustained activity.Although the course is variable, myas-thenia commonly affects muscles con-trolling the eyelids (resulting in droop-ing of the eyelids, or ptosis) and eyemovements (resulting in double vision,or diplopia). Muscles controlling facialexpression, chewing, swallowing, andspeaking are other common targets.An important indication of the causeof myasthenia gravis came from theclinical observation that the muscleweakness improves following treatmentwith inhibitors of acetylcholinesterase,the enzyme that normally degradesacetylcholine at the neuromuscularjunction. Studies of muscle obtained bybiopsy from myasthenic patientsshowed that both end plate potentials(EPPs) and miniature end plate poten-tials (MEPPs) are much smaller thannormal (see figure; also see Chapter 5).Because both the frequency of MEPPsand the quantal content of EPPs are nor-mal, it seemed likely that myastheniagravis entails a disorder of the postsyn-aptic muscle cells. Indeed, electronmicroscopy shows that the structure ofneuromuscular junctions is altered,obvious changes being a widening ofthe synaptic cleft and an apparentreduction in the number of acetyl-choline receptors in the postsynapticmembrane.A chance observation in the early1970s led to the discovery of the under-lying cause of these changes. Jim Patrickand Jon Lindstrom, then working at theSalk Institute, were attempting to raiseantibodies to nicotinic acetylcholinereceptors by immunizing rabbits withthe receptors. Unexpectedly, the immu-nized rabbits developed muscle weak-ness that improved after treatment withacetylcholinesterase inhibitors. Subse-quent work showed that the blood ofmyasthenic patients contains antibodiesdirected against the acetylcholine recep-tor, and that these antibodies are pre-sent at neuromuscular synapses.Removal of antibodies by plasmaexchange improves the weakness.Finally, injecting the serum of myas-thenic patients into mice producesmyasthenic effects (because the serumcarries circulating antibodies).These findings indicate that myas-thenia gravis is an autoimmune diseasethat targets nicotinic acetylcholinereceptors. The immune responsereduces the number of functional recep-tors at the neuromuscular junction andcan eventually destroys them altogether,diminishing the efficiency of synaptictransmission; muscle weakness thusoccurs because motor neurons are lesscapable of exciting the postsynapticmuscle cells. This causal sequence alsoexplains why cholinesterase inhibitorsalleviate the signs and symptoms ofmyasthenia: The inhibitors increase theconcentration of acetylcholine in thesynaptic cleft, allowing more effectiveactivation of those postsynaptic recep-tors not yet destroyed by the immunesystem.Despite all these insights, it is stillnot clear what triggers the immune sys-tem to produce an autoimmune(A)(B)–4510150480 200.05 0.10 0.20 0.50 1 2 340 60 80 0 20 40 60 80 0 20 40 60 80NormalEMGrecordingofmuscleactionpotentials(mV)NumberofMEPPsTime (ms)MEPP amplitude (mV)Myasthenia gravis afterneostigmine treatmentMyasthenia gravisNormalMyastheniagravis(A) Myasthenia gravis reduces the efficiency of neuromuscular transmission. Electromyographsshow muscle responses elicited by stimulating motor nerves. In normal individuals, each stimu-lus in a train evokes the same contractile response. In contrast, transmission rapidly fatigues inmyasthenic patients, although it can be partially restored by administration of acetyl-cholinesterase inhibitors such as neostigmine. (B) Distribution of MEPP amplitudes in musclefibers from myasthenic patients (solid line) and controls (dashed line). The smaller size of MEPPsin myasthenics is due to a diminished number of postsynaptic receptors. (A after Harvey et al.,1941; B after Elmqvist et al., 1964.)Purves06 5/13/04 2:53 PM Page 140
    • from the association of several protein subunits that can combine in manyways to produce a large number of receptor isoforms (see Figure 6.4C).NMDA receptors have especially interesting properties (Figure 6.7A). Per-haps most significant is the fact that NMDA receptor ion channels allow theentry of Ca2+in addition to monovalent cations such as Na+and K+. As aresult, EPSPs produced by NMDA receptors can increase the concentrationof Ca2+within the postsynaptic neuron; the Ca2+concentration change canthen act as a second messenger to activate intracellular signaling cascades(see Chapter 7). Another key property is that they bind extracellular Mg2+.At hyperpolarized membrane potentials, this ion blocks the pore of theNMDA receptor channel. Depolarization, however, pushes Mg2+out of thepore, allowing other cations to flow. This property provides the basis for avoltage-dependence to current flow through the receptor (dashed line in Fig-ure 6.7B) and means that NMDA receptors pass cations (most notably Ca2+)Neurotransmitters and Their Receptors 141response to acetylcholine receptors. Sur-gical removal of the thymus is beneficialin young patients with hyperplasia ofthe thymus, though precisely how thethymus contributes to myastheniagravis is incompletely understood.Many patients are treated with combi-nations of immunosuppression andcholinesterase inhibitors.ReferencesELMQVIST, D., W. W. HOFMANN, J. KUGELBERGAND D. M. J. QUASTEL (1964) An electrophysi-ological investigation of neuromusculartransmission in myasthenia gravis. J. Physiol.(Lond.) 174: 417–434.PATRICK, J. AND J. LINDSTROM (1973) Autoim-mune response to acetylcholine receptor. Sci-ence 180: 871–872.VINCENT, A. (2002) Unravelling the pathogen-esis of myasthenia gravis. Nature Rev.Immunol. 2: 797–804.Postsynaptic cell+H3NCOO−O+GlutamateGlutamineGlutaminaseGlutamate receptorsVGLUTEATTPresynapticterminalGlutamateGlutamineGlutaminesynthetaseGlial cellNH2CCH2CH2CHCOO−+GlutamateNH3COO−CH2CH2CHEATTFigure 6.6 Glutamate synthesis andcycling between neurons and glia. Theaction of glutamate released into thesynaptic cleft is terminated by uptakeinto neurons and surrounding glial cellsvia specific transporters. Within thenerve terminal, the glutamine releasedby glial cells and taken up by neurons isconverted back to glutamate. Glutamateis transported into cells via excitatoryamino acid transporters (EATTs) andloaded into synaptic vesicles via vesicu-lar glutamate transporters (VGLUT).Purves06 5/13/04 2:53 PM Page 141
    • 142 Chapter Sixonly during depolarization of the postsynaptic cell, due to either activationof a large number of excitatory inputs and/or by repetitive firing of actionpotentials in the presynaptic cell. These properties are widely thought to bethe basis for some forms of information storage at synapses, such as mem-ory, as described in Chapter 24. Another unusual property of NMRA recep-tors is that opening the channel of this receptor requires the presence of a co-agonist, the amino acid glycine (Figure 6.7A,B). There are at least five formsof NMDA receptor subunits (NMDA-R1, and NMDA-R2A through NMDA-R2D); different synapses have distinct combinations of these subunits, pro-ducing a variety of NMDA receptor-mediated postsynaptic responses.Whereas some glutamatergic synapses have only AMPA or NMDA recep-tors, most possess both AMPA and NMDA receptors. An antagonist ofMembrane potential (mV)EPSC(pA)(B)EPSC(pA)505075 1002525−25050 75 10025050 75 1002500EPSC(pA)5025−250EPSC(pA)5025−25000−50−505050−100−100−150150100100(C)(A)AMPA onlyNMDA onlyAMPA and NMDATime (ms)Time (ms)Time (ms)Mg2+,glycineGlycine,no Mg2+No glycine,no Mg2+GlutamateGlycineMg2+bindingsiteMg2+ChannelporeK+Na+Ca2+Figure 6.7 NMDA and AMPA/kainate receptors. (A) NMDA receptorscontain binding sites for glutamate and the co-activator glycine, as well asan Mg2+-binding site in the pore of the channel. At hyperpolarized poten-tials, the electrical driving force on Mg2+drives this ion into the pore of thereceptor and blocks it. (B) Current flow across NMDA receptors at a range ofpostsynaptic voltages, showing the requirement for glycine, and Mg2+blockat hyperpolarized potentials (dotted line). (C) The differential effects of glu-tamate receptor antagonists indicate