Search

The travelling wave of curvature which propels a fish forward arises from the interaction of the patterns of motoneurone activity generated by the spinal cord with the mechanical properties of (1) the muscle, (2) the skin, bone and connective tissues of the body, and (3) the water in which it is swimming. Furthermore, in the lamprey, a powerful feedback system has been demonstrated which allows local body curvature to influence the timing of the activity pattern generated by the spinal cord. The relative timing between activation and curvature are illustrated for both closed- and open-loop conditions, using data from intact swimming lampreys and from an in vitro preparation of lamprey spinal cord and notochord. The mechanical behaviour of a lamprey has been simulated with a mathematical model based on springs, dashpots, light rods, point masses and power units incorporating properties of lamprey muscle. Results are presented which illustrate the behaviour of a lamprey out of water. To anticipate the inclusion of the lamprey body model in the computation of the fluid dynamics, a hydrodynamical model has been developed in which the body motion and the forward swimming have been prescribed by mathematical functions. Results are presented to illustrate the hydrodynamic vortex structure as predicted by a two-dimensional, time-dependent numerical solution of the Navier-Stokes equations, including both viscous and inertial terms.

A two dimensional continuum model for the body mechanics of the lamprey is derived from a simple discrete rod and pivot structure. Each element in the discrete structure consists of two smoothly jointed light rods with perpendicular extensions at each of the midpoints between which is fixed a quasi muscle segment. The muscle segment is attributed with the viscous and elastic properties of all the animal tissue plus the ability to produce force. The travelling wave of muscle activation in the real animal is modelled by a corresponding time dependent forcing term at each segment. A linearisation of the ensuing continuum model, corresponding to low curvature dynamics, is investigated. The profiles obtained compare favourably with those of a lamprey moving out of water on a smooth surface. In addition the phase difference at each point on the body between the wave of muscle activation and the mechanical wave observed on the body indicates that the mechanical wave progresses slower than, but at the same frequency as, the wave of activation; this is a property that is also observed in the freely swimming lamprey.

Rhythmic motor activity requires coordination of different muscles or muscle groups so that they are all active with the same cycle duration and appropriate phase relationships. The neural mechanisms for such phase coupling in vertebrate locomotion are not known. Swimming in the lamprey is accomplished by the generation of a travelling wave of body curvature in which the phase coupling between segments is so controlled as to give approximately one full wavelength on the body at any swimming speed. This article reviews work that has combined mathematical analysis, biological experimentation and computer simulation to provide a conceptual framework within which intersegmental coordination can be investigated. Evidence is provided to suggest that in the lamprey, ascending coupling is dominant over descending coupling and controls the intersegmental phase lag during locomotion. The significance of long-range intersegmental coupling is also discussed.

1. This paper reports the results of an investigation of the basic mechanisms underlying intersegmental coordination in lamprey locomotion, by the use of a combined mathematical and biological approach. 2. Mathematically, the lamprey central pattern generator (CPG) is described as a chain of coupled nonlinear oscillators; experimentally, entrainment of fictive locomotion by imposed movement has been investigated. Interpretation of the results in the context of the theory has allowed conclusions to be drawn about the nature of ascending and descending coupling in the lamprey spinal CPG. 3. Theory predicts and data show that 1) the greater the number of oscillators in the chain, the smaller is the entrainment frequency range and 2) it is possible to entrain both above and below the rest frequency at one end but only above or below at the other end. 4. In the context of the experimental results, the theory indicates the following: 1) ascending coupling sets the intersegmental phase lags, whereas descending coupling changes the frequency of the coupled oscillators; 2) there are differences in the ascending and descending coupling other than strength; and it also suggests that 3) coupling slows down the oscillators.

Edge cells are mechanoreceptive neurones located in the lateral tracts of the lamprey spinal cord. Phasic activation of these cells by lateral bending can entrain the activity of the locomotor central pattern generator. During fictive locomotion induced by bath-applied NMDLA (N-methyl-D,L-aspartate) or sensory stimulation, edge cells receive synaptic input. In this paper we provide evidence that the tonic inhibition received during sensory-induced fictive locomotion is glycinergic. We also show that during fictive locomotion induced by application of NMDLA to the spinal cord, bicuculline unveils phasic synaptic activity in edge cells. During sensory-evoked fictive locomotion, by contrast, only tonic synaptic activity is apparent in the presence of bicuculline.

Rhythmic ventral root activity has been induced in the spinal cord of the lamprey in vitro in the presence of strychnine. Bicuculline (a GABAA blocker) increases the initial frequency and decreases the episode duration of such activity, whereas diazepam (which potentiates GABA action) has the opposite effect. In addition, bicuculline slows the depolarisation of ventral horn neurones. However, voltage clamp at potentials positive to the interburst potential does not reveal net outward current, even in the presence of diazepam, indicating that the effects of GABA on the rhythmic activity must be presynaptic to the impaled cells.

A comparison has been made of the patterns of muscle activity during swimming in the intact and spinal lamprey, and the patterns of ventral root activity in the in vitro preparation of the lamprey spinal cord. Electromyographic (e.m.g.) activity was recorded with intramuscular bipolar electrodes from three segmental levels in intact lampreys swimming in a swim-mill at a range of swimming speeds. The patterns of activity obtained were similar to those seen in elasmobranch and teleost fish. After high spinal transection, lampreys could be induced to swim continuously for a period of several minutes in the swim-mill by a light initial mechanical stimulation of the tail or dorsal fin. The patterns of e.m.g. activity obtained from spinal animals at a range of swimming speeds were similar to those obtained in the intact state. Portions of spinal cord were isolated encompassing those segments from which e.m.g. recordings had been made and ventral root recordings were made in vitro of the rhythmic activity induced by bath application of D-glutamate. In all experiments the mean duration of the bursts of activity at any segmental level was directly proportional to the mean cycle duration, and the constant of proportionality (about 0.36) was similar for all three types of preparation. In all preparations the mean time delay for the activation of segments in the rostral-caudal direction was proportional to the cycle duration and to the number of segments between recording positions. The proportionality constant, or phase lag per segment, was approximately equal to 0.01 in all three types of preparation.

Fictive locomotion in the lamprey, elicited by N-methyl-DL-aspartate or D-glutamate, was compared in two buffer solutions, one using bicarbonate, the other using HEPES. No differences were observed in burst proportion, cycle length, or phase lag per segment between these buffers solutions. These data indicate that HEPES-buffered artificial cerebrospinal fluid, which does not require gassing with carbon dioxide to maintain pH, may be a suitable alternative to bicarbonate-buffered artificial cerebrospinal fluid.

Edge cells are stretch receptors with receptive fields in the lateral margin of the lamprey spinal cord. Their activity may entrain the spinal central pattern generator (CPG). We have recorded intracellularly from edge cells during short episodes of fictive locomotion. Every cell tested exhibited numerous chloride-sensitive IPSPs. We have therefore provided evidence that edge cell sensitivity is modulated by the activity of the spinal CPG.

Strychnine is shown to abolish left-right alternation in fictive locomotion induced by sensory stimulation. Robust rhythmic activity, characterized by left-right coactivation at each segmental level, is seen in the presence of strychnine at all doses used (0.5-20 microM). The proportion of the cycle occupied by the ventral root bursts and the rostral-caudal coordination is similar to that seen in the absence of strychnine. Furthermore, the rhythm is abolished by cis-2,3-piperidine dicarboxylic acid (PDA), 2-amino-5-phosphonovalerate (APV), or the removal of Mg2+ from the perfusate, as in the absence of strychnine. Voltage clamp was applied to ventral horn neurons during stimulation in the presence of strychnine, with holding potentials negative to the plateau potential associated with a ventral root burst but positive to the potential in the interburst. Inward current was seen during the ventral root burst, but no outward current was seen at burst termination or during the interburst. The results indicate that in fictive locomotion induced by endogenous release of NMDA receptor agonists, left-right alternation is dependent on glycinergic transmission. Furthermore, evidence is provided that in the absence of glycinergic transmission, burst termination may depend on NMDA receptor-linked voltage-sensitive processes.