A discussion meeting: Mechanisms of integration in the nervous system

R. K. MURPHEY, W. W. WALTHALL, G. A. JACOBS

Department of Biology, State University of New York, Albany, NY 12222, U.S.A.

Studies of neurospecificity in the cricket cereal sensory system are reviewed and a decade of experimentation is examined in the light of recently obtained anatomical data. The nearly complete description of the anatomy of the afferent projection and the dendritic fields of the interneurones is a very powerful tool in the interpretation of previously published physiological experiments and the design of new ones. The results of the physiology available indicate that the excitatory receptive fields of directionally-selective interneurones are a joint function of an orderly afferent projection and the dendritic structure of the first order interneurones.

The mechanisms which shape the orderly afferent projection have been assessed and are compared with the work on vertebrate sensory systems. It is concluded that both positional interactions of the type conceived by Sperry (1963) and competitive interactions of the type conceived by Hubel, Wiesel and Levay (1977) are involved in producing the cereal afferent projection. The limits of a purely anatomical approach to the study of neurospecificity are considered in light of the work on this cricket sensory system.

R. B. LEVINE

Department of Biology Rice University Houston, Texas, 77001, U.S.A.

During metamorphosis insects undergo dramatic changes in both form and behaviour. Cell birth and death, as well as neurone respecification all contribute to the overall reorganization of the nervous system. Within the visual and chemosensory processing areas of the insect brain large numbers of newly-generated adult neurones are incorporated into the larval nervous system during metamorphosis. In the abdominal motor systems, however, identified larval neurones are retained to assume a new adult role. This respecification of motorneurone function involves not only the acquisition of a new target muscle, but also the reorganization of dendritic morphology, and alterations in the interconnexions between neurones. In one example of the latter, an identified abdominal motorneurone in the hawkmoth Manduca sexta grows new dendritic processes during metamorphosis, and changes its synaptic relationship with an abdominal stretch receptor such that an interaction that was inhibitory during larval life, becomes excitatory in the adult. In another example, identified sensory neurones that evoke a larval flexion behaviour, later participate in the defensive gin trap reflex that is characteristic of the pupal stage. In both instances the formation of new pathways is a two-step process in that the new circuits do not become behaviourally relevant as they are formed, but instead are activated abruptly at the appropriate time. In the case of the gin trap reflex an identified peptide hormone is responsible for the activation step.

MICHAEL BASTIANI¹, KEIR G. PEARSON², AND COREY S. GOODMAN¹

¹Department of Biological Sciences Stanford University Stanford, CA 94305, U.S.A.

²Department of Physiology University of Alberta Edmonton, Alberta, Canada

We will discuss ideas emerging from our studies on selective axonal fasciculation in the grasshopper embryo that have implications for the organization of the adult neuropil in insects and perhaps other animals. While one of our laboratories has been studying the embryonic development of the sibling G and C interneurones (in the mesothoracic segment) and their lineal homologues (in other segments), the other laboratory has been studying the morphology and physiology of these same two neurones and their segmental homologues in the adult nervous system. Our embryonic studies show that the growth cone of the G neurone selectively fasciculates with the A/P fascicle in preference to all other longitudinal axon fascicles as it turns anteriorly. The C neurone also fasciculates in this same bundle as it turns posteriorly. The homologues of G and C in other thoracic and abdominal segments fasciculate in this same bundle. However, early in their morphological differentiation, they reveal interesting segmental differences. Our studies on the adult nervous system show that the segmental homologues of the G and C neurone share many properties in common (e.g. axons in the LDT: lateral dorsal tract) while other features are quite different. The notion emerging from these studies is that a basic segmentally-repeated pattern arises during embryogenesis: a stereotyped axonal scaffold upon which growth cones faithfully fasciculate. Evolutionary plasticity allows the specialization of lineally equivalent neurones in different segments within the context of the presence or absence of particular axons which they prefer, and the neuropilar neighbourhood that they find themselves in as a consequence of their selective fasciculation.

K.F. FISCHBACH AND M. HEISENBERG

Institut für Genetik und Mikrobiologie; Röntgenring 11, 8700 Würzburg, W.-Germany.

The importance of the genome for behaviour in vertebrates and invertebrates has been amply demonstrated in behavioural genetics. A causal understanding of the relationship, however, requires an understanding of how genes influence brain development and brain function. Behaviour is under the direct control of the brain, genes do not control behaviour directly. The obvious gap is being filled by neurogenetics.

Certain brain structures in insects (e.g. identifiable neurones) seem to be as rigidly controlled by the genome as the nervous system of the nematode Caenor-habditis elegans. Other structures (e.g. mushroom bodies) are influenced by experience, similar to some vertebrate systems.

Information on the rigidity of the genetic control can be derived from the comparison of isogenic grasshoppers. The role of individual genes can be studied in Drosophila melanogaster. In the latter system many single gene mutations affecting the brain and behaviour have been isolated. They alter either the development of neural circuits or modify cellular functions of neurones.

Mutations of both categories often influence only certain functional subsystems, leaving others unaffected. Therefore, functional subsystems are to a certain degree ontogenetic units under independent genetic control. Telling examples are the sexual dimorphisms of brain structure and behaviour found in many insects. Accordingly, brain functions are organized in a parallel, not in a hierarchical fashion. This seems to be an outcome of the evolutionary need for independent genetic modifiability of functions.

DANIEL L. ALKON

NINCDS-NIH Marine Biological Laboratory, Woods Hole, Massachusetts.

A system of ionic membrane currents interacts with the repeated responses during conditioning of a system of neurones to effect long-term change of specific ionic channels and thereby encode a learned stimulus association. Prolonged elevation of intracellular calcium during conditioning causes subsequent persistence of increased excitability by reducing the number of open K⁺ channels (I_A and probably I_c) within identified soma membranes. This Ca^{+ +}–mediated reduction of K⁺ channels is thought to involve changes of Ca⁺⁺ – calmodulin-dependent phosphorylation of distinct membrane proteins and is contrasted with short-term neurohumoral channel regulation which is thought to involve cyclic-AMP-dependent phosphorylation.

R. D. HAWKINS

Center for Neurobiology & Behavior, College of Physicians and Surgeons, Columbia University and New York State Psychiatric Institute, New York, New York 10032, U.S.A.

The defensive siphon and gill withdrawal response of Aplysia is a simple reflex, mediated by a well-defined neural circuit, that exhibits sensitization in response to strong stimulation of the tail. The siphon withdrawal reflex also exhibits classical conditioning when a weak stimulus to the siphon or mantle shelf (the conditioned stimulus or CS) is paired with a shock to the tail (the unconditioned stimulus or US). Cellular studies indicate that the mechanism of this conditioning shares aspects of the mechanism of sensitization of the reflex: presynaptic facilitation due to a cAMP mediated decrease in K⁺ current and consequent broadening of action potentials in the sensory neurones. Thus, tail shock (the US) produces greater facilitation of the monosynaptic EPSP from a sensory to a motor neurone if the tail shock is immediately preceded by spike activity in the sensory neurone than if the shock occurs without spike activity (sensitization) or if the shock and spike activity are presented in a specifically unpaired pattern. This activity-dependent amplification of facilitation involves a greater broadening of action potentials in paired than in unpaired sensory neurones and appears to be due to a greater depression of the same serotonin -and cAMP-sensitive K⁺ current involved in sensitization. Preliminary experiments suggest that this amplification of facilitation may be due to priming of the adenylate cyclase by Ca⁺⁺ which enters the sensory neurones during the paired spike activity.

These results indicate that a mechanism of classical conditioning of the withdrawal reflex is an elaboration of the mechanism underlying sensitization. By analogy, the mechanisms of yet more complex forms of learning may in turn be built from combinations of the cellular mechanisms of these simple forms of learning.

JOHN P. MILLER AND GWEN A. JACOBS

Department of Zoology, University of California, Berkeley, Ca. 94720, U.S.A.

The geometry and electrical properties of a neurone determines how synaptic inputs and endogenously generated currents are integrated and transformed into the signals it transmits to other cells. The dependence of neuronal integration upon dendritic geometry has been studied extensively over the last three decades, both by experimentalists and by theoreticians. We review some of the general principles that have emerged from this work, and summarize recent studies that serve to illustrate these. The discussion is organized around the analysis of neuronal structure at three different levels. At the “macroscopic” level, we show how the dendritic branching structure of an identified interneurone in the cricket cereal afferent system determines the directional sensitivity within its receptive field. At the “microscopic” level, we illustrate the dependence of synaptic efficacy upon dendritic length, and demonstrate a very surprising result: that the extension (or “growth”) of a dendrite out beyond the point of a synaptic contact can increase the efficacy of that synapse. At the “ultrastructural” level, we show how the structural and electrical properties of dendritic spines might have profound effects upon synaptic integration.

EVE MARDER

Biology Brandeis University, Waltham, MA 02254, U.S.A.

Understanding fully the operation of a neural circuit requires both a description of the properties of the individual neurones within the circuit as well as the characterization of their synaptic interactions. These aims are often particularly difficult to achieve in neural circuits containing electrically-coupled neurones. In recent years two new methods, photoinactivation after Lucifer Yellow injection and intracellular injection of pronase, have been employed to delete selectively single neurones or small groups of neurones from neural circuits. These techniques have been successfully used in the analysis of circuits containing electrically-coupled neurones and in several systems new roles for electrical synapses for the integrative function of neural circuits have been proposed.

In the nervous systems of both the leech and lobster it is now known that synaptic interactions previously thought direct are mediated through an interposed, electrically coupled neurone. In the pyloric system of the stomatogastric ganglion of the lobster, Panulirus interruptus, the Lucifer Yellow photoinactivation technique has permitted a separate analysis of the properties of several electrically coupled neurones previously thought quite similar. We now know that Anterior Burster (AB) intemeurone and the Pyloric Dilator (PD) motor neurones which together act as the pacemaker ensemble for the pyloric network differ in many regards including: a) their intrinsic ability to generate bursting pacemaker potentials, b) their neurotransmitters, c) their sensitivity to some neurotransmitters and hormones, d) the neural inputs they receive, e) their pattern of synaptic connectivity.

The implications of these results for the functioning of this and more complex neuronal circuits will be discussed.

ROBERT K. S. WONG, RICHARD MILES AND ROGERTRAUB

Department of Physiology and Biophysics, The University of Texas Medical Branch, Galveston, Texas 77550, U.S.A.

1. Synchronized rhythmic discharge of large neuronal populations is a fundamental property of the mammalian central nervous system. We have used the in vitro slice preparation together with computer simulation to carry out a detailed study of this phenomenon.

2. Our data derived from the hippocampal slice preparation show that rhythmic oscillations of a neuronal population can arise as a result of local integration processes in the absence of phasic afferent input. The generation of synchronized activity critically depends on the intrinsic bursting capability of the neurones and the presence of recurrent excitatory connections between these cells. Furthermore, the occurrence of this activity is regulated by inhibitory synaptic events.

3. We have examined the contributions of both cellular and synaptic processes to the generation of synchronized discharge by carrying out computer simulations of the behaviour of networks of 100–500 neurones. The properties of the model neurone resemble those determined from slice experiments and we have explored patterns of neuronal connectivity which result in synchronized activity.

4. We consider that the information on neuronal synchronization is fundamental to an understanding of the mammalian CNS. The data are relevant to studies of central programs underlying cyclic motor acts such as walking and respiration. Similar processes may also be involved in the generation of prominent electrical activities recorded from the cortex such as the σ and α rhythms.

M1CHA E. SPIRA

Department of Neurobiology, Institute of Life Sciences, The Hebrew University of Jerusalem, Israel.

Neuronal interactions mediated by alteraction of the extracellular K⁺ concentration have been demonstrated to occur between populations of neurones as well as among single neurones in very restricted regions. The interactions mediated by K⁺ ions may range from low efficacy interactions, in which the effects of increased [K⁺]_o around the non active cells can be recorded only after massive stimulation of a large population of neurones or high frequency stimulation of single neurones, to very effective interaction in which a single action potential in one neurone is sufficient to produce a depolarization of several mV of a second neurone. Such efficient K⁺ mediated interactions can not be distinguished from postsynaptic responses induced by chemical or electrotonic synapses by shape, amplitude or time course.

The efficacy of K⁺ mediated interaction depends on a large number of variables. These include: (a) the passive and active membrane properties of the neurone or a particular neuronal segment which serves as a source for K⁺ efflux, (b) the dimensions, shape and properties of the extracellular space, (c) the activity of glial and neuronal cells in regulating the K⁺ concentration; and (d) the membrane properties of the affected neurone especially of sensitivity to the concentration gradient of K⁺ across the neuronal membrane.

The present paper analyzes the various factors which influence the efficacy of neuronal interactions mediated by potassium. Approaches and experimental manipulations which enable us to differentiate between the PSPs initiated by chemical and electrotonic synapses and PSPs initiated by transient increases of [K⁺]_o in a restricted region are described. Properties such as synaptic delay, rise time, decay time, reversal potential, summation, facilitation and directionality of K⁺ mediated interactions are compared with chemical and electrotonic synapse.

The effect of increased K⁺ on the integration of information by single neurones as well as at the level of networks will also be described. Finally, the possible role of K⁺-mediated interactions among neurones during developmental processes and in disease are discussed.

ROBERT M. PITMAN

Department of Physiology & Pharmacology, St. Andrews University, Fife, Scotland.

‘Typically’ chemical synaptic transmission takes place when an influx of calcium ions during a presynaptic nerve impulse triggers exocytosis of neurotransmitter substance from synaptic vesicles. The neurotransmitter diffuses across the synaptic cleft and occupies receptors embedded in the subsynaptic membrane. This interaction (directly or via a second messenger) operates characteristic ion channels and produces an increase in the postsynaptic membrane permeability to particular ions. Depending on the ionic species to which the postsynaptic membrane becomes more permeable, the physiological response will be an excitatory or an inhibitory postsynaptic potential. The action of neurotransmitters may be terminated either by enzymic inactivation or by cellular uptake mechanisms.

Over the last decade it has become clear that a neurotransmitter substance may exert a number of different actions on a single postsynaptic neurone. These may involve opening or closure of either voltage-independent or voltage-dependent ion channels. It is also possible that in some instances transmitters may act on neuronal biochemical systems to modify the physiology of postsynaptic cells without altering their electrical characteristics directly.

Analysis of the postsynaptic actions of neurotransmitter substances has become further complicated by the increasing body of evidence which indicates that more than one transmitter substance (one of which may be a peptide) can be released from a single presynaptic neurone. The significance of such dual transmitter systems has yet to be fully elucidated.

The efficacy of transmission across many synapses may be modified by either presynaptic or postsynaptic mechanisms; both transmitter release and postsynaptic responsiveness may depend on the recent history of a single synapse, on synaptic inputs from other neurones or on circulating neuro-active substances.

S. R. SHAW

Department of Physiology, Dalhousie University, Halifax, Nova Scotia, Canada B3H 4JI.

The early stages of visual processing are particularly concerned with amplifying and increasing the reliability of low level visual signals and with accentuating the spatial and temporal resolution of the input, using local and lateral interactions. The insect retina is particularly convenient for study of these processes because the neural connections are known in detail in several species in the columns (cartridges) of the first synaptic zone (lamina), which is accessible to physiological recording. The status of the scheme of connections is reviewed in the light of recent quantitative counts in the lamina of diptera. Individual synapses are stereotyped units incorporating up to four postsynaptic elements. There is a lack of significant lateral connections between cartridges that might mediate contrast enhancement by lateral inhibition. Conversely, two of the strongest connections are reciprocal synapses that feed directly back to photoreceptor terminals. A peculiar feature is the synaptic input to a laterally extending network of coupled glial cells, forming the second most abundant connection in the lamina. Visual adaptation first involves transient inhibitory feedback (self inhibition) on to the photoreceptor terminals from within the same cartridge, with the reciprocal synaptic circuits implicated. A slower, maintained phase of adaptation is non-synaptic, extends laterally, and is mediated by large field potentials set up across local extracellular barriers. The nature of these barriers and the possible role of the glial cells is discussed. Finally, the lamina exposes particularly clearly an intriguing general question of how a stereotyped neural system manages to evolve new functions, schackled as it is with an invariant cell complement. Our preliminary comparative results suggest that despite this, a flexible response to evolutionary pressures has been possible.

M. V. S. SIEGLER

Department of Zoology, Downing Street, University of Cambridge, Cambridge, England CB2 3EJ

As their name implies, local neurones arborize within anatomically restricted regions of a nervous system, and the connections that they make establish local circuits. In arthropods, local neurones may arborize wholly within a segmental ganglion, or within a specialized region of the brain. Recent findings indicate that their local interactions may be mediated either by spiking, or by nonspiking interneurones. In the thoracic ganglia of insects, for example, spiking local interneurones are largely responsible for the local processing of primary sensory inputs, whereas nonspiking, local ones play a major role in the control and coordination of motor neurone activity at the segmental level. The latter effects are mediated by the graded release of transmitter, onto inter-or motor neurones. Evidence from other arthropods indicates that restricted regions of long neurones may also be involved in local circuits and local interactions, with both spikes, and sustained changes in membrane potential effecting the release of transmitter Sustained inputs may, in addition, alter the local efficacy of spike-mediated postsynaptic potentials.

J. J. WINE

Department of Psychology, Stanford University, Stanford, Ca. 94305, U.S.A.

The abdominal nervous system of the crayfish contains six serially homologous ganglia, each containing approximately 600 neurones. No two ganglia are identical and the ganglia interact extensively. Therefore, studies confined to intraganglionic interactions yield limited and sometimes misleading information.

Each ganglion contains intrinsic (local) interneurones, motor neurones and projecting interneurones in roughly equal numbers, except in the specialized terminal ganglion where the ratio of these cells is approximately 3:2:1. Although the number of nerve cell bodies in a ganglion is small enough to be tractable, integration occurs in the neuropile, which contains terminals from interneurones and afferents that outnumber the neurones originating in the ganglion by at least ten to one.

The abdominal nervous system responds almost exclusively to a variety of mechanosensory stimuli. It also has very limited light sensitivity; but other modalities, notably chemosensitivity, are undescribed and may be lacking.

The effectors of the abdomen consist of fast axial muscles used for tailfiip-powered escape, slow axial muscles used for abdominal posture, appendage muscles used for swimmeret beating, and slow muscles of the intestine and rectum that control gut emptying. The fast and slow muscles of the tailfan are specialized homologs of the axial and swimmeret muscles.

The abdominal nervous system represents only 3–4% of the 100,000 neurone crayfish CNS. Most sensory information gathered in the abdomen is sent to the rostral CNS for processing, and many abdominal motor programs are activated by descending commands. Nevertheless, a surprising degree of autonomy is present, and at least some motor programs for every motor system can be activated in isolated abdomens. A recent finding is the gradual development (over weeks) of an autonomous motor program in an isolated abdomen. It is instructive to compare the operation of this system with that of the stomatogastric nervous system, which is smaller by two orders of magnitude.

A. ROBERTS, M. DALE, S. R. SOFFE

Department of Zoology University of Bristol Woodland Park Road, Bristol, U. K.

It is generally unclear how brief stimuli lead to sustained changes in posture or to prolonged rhythmic activity like locomotion. Theories for central nervous rhythm generation, with a few exceptions, ignore this problem and propose an external excitatory drive. We will examine current ideas about this drive, describe results from Xenopus embryos showing fictive swimming in intact and spinal preparations, and present a proposal for how this activity is sustained. The proposal will be evaluated using simulation on a neural network model.

S. GRILLNER

Department of Physiology III, Karolinska Institutet, Stockholm, Sweden.

The lamprey spinal cord, in isolation or with the brainstem, can be used in vitro. The motor patterns underlying the swimming movements can be elicited by: (1) a pharmacological activation of a specific type of neuronal receptor (NMDA-receptor), that may in other systems give rise to an unstable membrane potential, (2) by stimulation of the brainstem or (3) by tactile activation of skin regions left innervated. In the latter case the initiation of “fictive” swimming is partially caused by a release of a transmitter-activating NMDA-receptor, as judged by the effect of NMDA receptor-blockers. The central pattern generator (CPG) is strongly influenced by feedback from mechano-sensitive elements, which at least partially reside within the spinal cord. The edge cell in the lamprey spinal cord serves as an intra-spinal mechanoreceptor.

The ability to generate a coordinated motor output is distributed, since spinal cord sections down to 1.5–2 segments can be made to generate alternating activity. Motoneurones receive an approximately synchronous alternating excitatory and inhibitory drive in each swim cycle and do not appear to be part of the CPG. Motoneurones supplying different parts of the body wall on the same side of a body segment have different morphology with ramifications around different descending axons. During fictive locomotion, the input drive signal to motoneurones (located close to each other but with different morphological characteristics) may differ substantially with regard to the y-relationship (± 25%) and the shape of the oscillation. This implies that even at a segmental level motoneurones may be further subdivided, and furthermore that the ipsilateral net-work generating the drive signal to ipsilateral motoneurones generates a more complex and individualized output than previously assumed. Motoneurones are not part of the rhythm-generating circuit. The large identifiable interneurones are not required for rhythmic activity to occur although they may be phasically active in the swim cycle. The small segmental interneurones have not yet been completely characterized. Many are phasically active during “fictive locomotion” and lack an apparent axon. Their phase relationship in relation to the burst patterns vary over the entire swim cycle.