Neural control Editorial overview

Neural control Ed ito ria l Ste n G rillne r Ka ro linska Institute , Sto c kho lm , o ve rvie w a nd A p o sto lo s Swe d e n, a nd Ve te ra ns Minne so ta , C urre nt O p inio n The old concept of a neuron as a Iixed entity with constant properties is no longer appropriate. A given cell may in different functional contexts take on very different roles. Synaptic modulatoty input may modify a cell from a faithful relay neuron in a collective to an individualistic pacemaker taking over the role of the prima donna and performing intricate burst configurations. Some synaptic inputs may gate the different types of Ca2+ or K+ channels, and others may modulate the ion channels determining the afterhyperpolarization, which controls instantaneous spike interval and frequency adaptation. Most rapid synaptic interactions in the different networks depend on transmitters that act directly on ligand-gated ion channels, e.g. glutamate, glycine and GABA,. As important are the modulators acting via G-protein-mediated synaptic transmission, as a result of activation of monoamine, metabotropic glutamate, GABAu, and peptidergic receptors. These inputs can modify a cell’s performance dramatically, and therefore modify the performance of a network in a way that makes it change its overall pattern generation, This type of modulation is reviewed by Hultbom and Kiehn (pp 770-775) at the single cell level, and by Selverston (pp 776780) and Kristan (pp 781-787), who deal with behavior at the network level. Presynaptic modulation may be as important as synaptic modulation at the soma-dendritic level (see Clarac et al, pp 764769). At the alferent level it provides gating of sensory transmission in specific phases of a movement. In different patterns of behavior, the same alferent axon may be subject to presynaptic inhibition from different classes of interneurons, which may use different transmitters, e.g. GABAn, histamine or 5-HT. In two networks it has now become evident that the synaptic transmission from both excitatory and inhibitory network interneurons is modulated at a presynaptic level, thereby introducing a new powerful site for network modulation, The neural network that has provided perhaps the most important general insights over the past two decades is the stomatogastric system of crustacea, both with regard to basic Affa irs Me d ic a l C e nte r, Minne a p o lis, USA in Ne uro b io lo g y The brain has evolved to control and coordinate the different subsystems of the body, whether concerned with feeding or respiration, or achieving perfection in playing the violin or tight-rope walking. The nervous system is essentially a large number of neuronal networks each specialized to solve a specific task of sensoty, motor or cognitive nature. An important goal of current neuroscience is to understand how these different networks operate and interact, thereby generating our complex and varied behavioral repertoire. The interaction is at all levels from general conceptual issues on a gross organizational scale to molecular and cellular levels. In this overview we will begin at a cellular and molecular level and proceed to network and systems levels. @J Curre nt P. G e o rg o p o ulo s 1992, 2757-758 network mechanisms at the cellular level, and mechanisms for recombination of groups of neurons within one network or several subsystems, to provide the required behavioral flexibility. In parallel, a number of other invertebrate models from Aplysia and several molluscs, annelids and insect systems have contributed greatly. In vertebrates the situation is more complex because of the vast number of cells and possible interactions. Only very few lower vertebrate models provide sufficient information at the cellular level to account for the operation of behaviorally relevant networks. Rapidly accumulating evidence from mammalian and avian nervous systems means we can be optimistic about the possibilities of reaching an understanding on the cellular mechanisms used in mammalian networks controlling basic patterns of behavior like chewing, feeding, respiration, and locomotion, reviewed by Jordan et al. (pp 794-801) and Richter et al. (pp 788793). Although the detailed circuits are far from complete, basic features of these subsystems for initiation and activity level control, pattern generation, and sensory interaction are now being analyzed in detail. At another level it is interesting to consider Billington and Levine’s review (pp 847-851) discussing the role of neuropeptide Y (NPY), a well conserved neuropeptide, which appears to be involved in the regulation of feeding. Increased levels of NPY in the paraventricular nucleus of the hypothalamus promote feeding, via a combination of different motor acts that need to be appropriately sequenced. The converse behavioral state, satiety, also appears to be affected by a peptide, this time cholecystokinin. The brainstem is critical not only for controlling the large variety of basic motor patterns considered above, but also for other types of control mechanisms such as the degree of alertness, or different stages of sleep. The level of activity in noradrenergic neurons of the locus coeruleus is correlated to the level of attention. These neurons project to most parts of the nervous system from the sacral level to neocortex and cerebellum, and may cause a general potentiation of neural activity in these areas. Hobson (pp 75~763) focuses on REM sleep and the role of the pedunculopontine nucleus (PPN), which is located close to the locus coeruleus. This choline@ nucleus is thought to trigger REM sleep and maintain this type of behavioral state. It is thought to elicit the cortical EEG pattern, the characteristic decrease of muscle tone, and the rapid eye movements that give rise to the designation of REM sleep. The cerebellum is still enigmatic with its stereotypic neuronal circuitry. It receives a great variety of inputs, e.g. fro m cardiac and respiratory receptors, proprioceptors, and the retina, as well as efferent copies of the motor commands. The question of whether or not the cerebellum plays a role in motor learning still divides researchers in different camps. Biology Lt d I SSN 0 9 5 9 -4 3 8 8 757 758 N e ura l c ont rol The expression ‘motor learning’ is in itself so imprecise that it should probably be abandoned for more accurate expressions that relate to, for instance, the modulation of the am plitude of a reflex, learning of new patterns of coordination, or a recombination of previously acquired or innate motor components. On the other hand, it has been clearly demonstrated by Ito and others that a climbing fiber activation can produce a long-term depression of parallel fiber synapses, but the question is not yet settled as to whether the pattern of activation of climbing fibers in these experiments is close enough to their physiological mode of operation. Glickstein (pp 802-806) summarizes the extensive data on the eye blink reflex model in an insightful way and favors the role of the cerebellum as a learning machine. There is little doubt that a key to the understanding of the cerebellum is in the climbing fiber effects on Purkinje cells and their pattern of activation by parallel fibers. Bloedel is critical of a role of climbing fibers as ‘teachers’ of parallel fibers, active in a time window around that of the complex spikes produced by the climbing fibers. He has instead proposed that the climbing fibers, still acting as ‘error detectors’, produce a facilitation of the parallel fiber activity during the tenth of a second or so when the complex spike influences the Purkinje cell excitability. The climbing fibers could thereby contribute to an effective reflex adaptation. The actual learning would in his view take place outside the cerebellum. Several important questions have been actively pursued concerning the ongoing control of arm and hand movements. Three major aspects of this problem are reviewed in this issue. The Erst deals with the role of proprioceptors in motor control. Hasan (pp 824-829) discusses convincingly recent evidence pointing to an important role of proprioceptive input in contributing to the steering of multi-joint movements and for temporal coordination among joints. The notion that a.Eerent signals from muscles serve only to resist external perturbations is untenable. Instead, an integrated picture emerges in which the ongoing proprioceptive signals provide the necessary background information for generation of the movement, and its on-going correction from deviations produced, for example, by interaction forces. Moreover, the effects of local proprioceptive perturbations spread widely to other joints thus redefining the functional unit to the whole limb. The specific case of sensorymotor coordination in grasping and object manipulation is reviewed by Johansson and Cole (pp 815-823), and in particular those mechanisms that integrate visual and somatosensory signals with internal models of manipulation and its anticipated somatosensory effects. A similar problem concerns the interplay between geometrical factors of limb configuration and kinetic variables of posture and movement. Iacquaniti (pp 807-814) reviews the results of recent studies indicating that these aspects interact appreciably in their effects on behavior, and their requirements for neural control, Moreover, the anticipatory muscle activity preceding impact of moving objects with the hand points strongly to the existence of internal models of kinetic variables, such as momentum and limb impedance. The elucidation of the neural substrates of these internal representations will be of great importance. A central question in the cortical control of movement concerns the cortical areas involved. It was thought for a long time that the motor cortex was the ‘Enal common path’ through which the cortical processing in various areas was transmitted to subcortical structures, including the spinal cord. This idea assumed that premotor areas exert their motor influence by converging projections on the motor cortex, and that direct influence of the spinal motor apparatus, if present at all, was minimal. This picture has now changed, however. It has been shown that the premotor cortical areas, although interconnected with the motor cortex, have themselves substantial projections to the spinal cord. Their electrical microstimulation evokes movements, and the activity of their cells, recorded in behaving monkeys, relates to motor behavior. This Ending has refocused the issue on the role and cooperation among these areas, rather than putting all the onus on the motor cortex itself. The review by Dum and Strick (pp 836839) is centered on the premotor areas located in the medial wall of the cerebral hemisphere. In addition to the supplementary motor area (WA), three more areas have been identified, all in the banks of the cingulate sulcus. The identification of these new areas raises very interesting issues of interpretation of results obtained previously using positron emission tomography. For example, the cortical activation during imagined movements was taken to be localized in the SMA (P Roland et al., J Neuropkysioll980, 43:118-136). In view of the current anatomical findings, it is possible that this activation could reflect activation of the cingulate premotor areas alone, or in combination with the SMA. This is an important issue that could be partly resolved by using the high spatial resolution afforded by high strength (4 Tesla) magnetic resonance functional imaging of motor areas (S-G Kim et al., J Neuropbysiol1992, in press). The coordination of motor function among various brain areas is reviewed by Goldman-Rakic et al. (pp 83W335). Path ways through which information can be transmitted from more anterior prefrontal areas to premotor areas and the motor cortex have now been identified. Moreover, projections from prefrontal cortex to basal ganglia structures and the superior colliculus have identified cortical-subcortical circuits involved in the control of eye movements. A specific case of eye movement control concerns holding in memory signals for memorized saccades. Such signals have been identified in prefrontal and posterior parietal areas. The parietal area involved in this function is the lateral intraparietal. The role of this recently identified area in saccadic eye movements (‘parietal eye field’) is reviewed by Andersen et al. (pp 840-846). Results of several ingenious experiments indicate that this area performs truly visuomotor integration and is not simply concerned with attention or with processing of purely movement-related or sensory information. The projection of this area to the intermediate layers of the superior colliculus, and its interconnection with the frontal eye Eeld identify another circuit for saccadic control. It is clear that the networks of the brain are not only fascinating, but also understandable, given patience and dedicated researchers. S Grillner, Karolinska Institute, Nobel lnstitute of Neurophysiology, 60400, S-10401, Stockholm, Sweden. Box AP Georgopoulos, Brain Sciences Center, Veterans Affairs Medical Center, One Veterans Drive, Minneapolis, Minnesota 55417, USA.