Interneurones Co-Ordinating the Ventilatory Movements of the Thoracic Spiracles in the Locust

Whenever several segments are involved in the production of a rhythmical movement it is necessary that their activity be co-ordinated. Typically the neural elements in one segment act as the dominant pacemaker for a rhythm, which is then imposed apon the others, although these may also be capable of some innate rhythmicity. So-called ‘co-ordinating’ interneurones must therefore connect the various segments to carry information about the rhythm. Interneurones with this property can be recognized, and often characterized as individuals, in the central nervous system of vertebrates. For example, in locusts ventilation requires the co-ordinated activity of muscles in abdominal and thoracic segments. The abdominal muscles contract rhythmically to provide the propulsive forces that drive air around the complex network of tracheae. Muscles also control the aperture of the paired openings, the spiracles, on each abdominal segment. They contract rhythmically in time with ventilation opening during expiration and closing during inspiration. The thoracic muscles involved in normal ventilation are those which control the spiracles, causing them to open during inspiration and to close during expiration. A one-way circulation of air in the tracheae thereby results. When ventilation is stressed other thoracic muscles may be recruited that move the head rhythmically.

The dominant pacemaker for both the abdominal and thoracic movements is within the metathoracic ganglion (Miller, 1960). In this ganglion are interneurones with a putative co-ordinating function whose axons either descend to the abdominal ganglia or ascend to the other thoracic ganglia. If the metathoracic ganglion is destroyed, individual abdominal ganglia can sustain a ventilatory rhythm, but the thoracic ganglia cannot. Some abdominal interneurones involved in ventilation are known by extracellular recordings from the connectives that link the abdominal ganglia (Lewis, Miller & Mills, 1973), and from intracellular recordings made in the metathoracic ganglion from one of them (Pearson, 1980). The output connexions of these interneurones are not known. By contrast the output connexions of two sets of thoracic interneurones have been described in some detail from intracellular recordings made from pairs of spiracular and flight motor neurones (Burrows, 1975 a, b, 1981). One set, probably consisting of two interneurones, causes inhibition of spiracular closer motor neurones during inspiration (Burrows, 1982). A second set probably of two interneurones is active during expiration exciting closer motor neurones and some flight motor neurones, as well as inhibiting spiracular opener motor neurones (Burrows 1975 a, b). In this paper the first set of thoracic co-ordinating interneurones has been identified by intracellular recording and staining. It is shown that there are two interneurones which make inhibitory synaptic connections with thoracic spiracular closer motor neurones. Manipulation of the pattern of spikes in one interneurone alters the pattern of the motor output but not the frequency of the ventilatory rhythm itself.

Adult locusts, Schistocerca americana gregaria (Dirsh) (=S. gregaria (Forskål)) were obtained from our crowded culture. Intracellular recordings were made from the somata of interneurones in the metathoracic ganglion and motor neurones in the mesothoracic ganglion in a locust mounted with its ventral surface uppermost. The ganglia were exposed, stabilised upon a steel platform and the thorax perfused with saline according to procedures described in detail elsewhere (Hoyle & Burrows, 1973). The glass microelectrodes had d.c. resistances of 30–40 MΩ in saline when filled 2 M potassium acetate, and higher resistances when filled with 0·4 M cobalt chloride. Prior to recording, the sheath covering the ventral surface of the ganglia was treated for 2 min with a 1 % (w/v) solution of protease (Sigma Type VI) in saline. The interneurones were stained by the intracellular injection of cobalt (Pitman, Tweedie & Cohen, 1972) and the subsequent intensification of the sulphide preciral wate with silver in a whole ganglion (Bacon & Altman, 1977). Drawings of a stained interneurone were made with a camera lucida. Transverse wax sections 10 μm thick were made of the ganglia containing the stained neurones. The nomenclature of the longitudinal tracts within a ganglion is taken from the scheme devised by Gregory (1974) for the cockroach. All electrical signals were stored on magnetic tape using an FM tape recorder for later photography, or analysis with a DL 4000 signal processor (Data Laboratories Ltd., London). The description of the morphology is based upon three stains of the interneurone, and the physiology upon simultaneous recordings from an interneurone and a spiracular motor neurone in five locusts.

An electrode probing within the metathoracic ganglion posterior to the emergence of lateral nerves 3 encounters many neurones whose patterns of spikes or synaptic potentials are linked to the ventilatory rhythm. The majority subsequently prove to be motor neurones but some are interneurones. One particular type of interneurone, encountered on both the left and the right sides of the ganglion, produces bursts of spikes in time with the ventilatory movements of the thoracic and abdominal muscles. The activity of these interneurones can be related more precisely to the particular phase of ventilation by correlating their spikes with those recorded simultaneously in a motor neurone whose activity is already well known. A suitable motor neurone is a closer of the mesothoracic spiracles which produces bursts of spikes that start abruptly at the beginning of expiration and end before the onset of the next inspiration (Fig. 1 a). This burst of spikes is therefore a convenient indicator of the duration of expiration.

The spikes in one of these interneurones straddle both the expiratory and inspiratory phases of ventilation (Fig. 1 a, b). The spikes begin in the second half of expiration and continue throughout inspiration at a steady frequency (Fig. 1 c). At the start of the next expiration there is an abrupt repolarization so that only an occasional spike in the interneurone overlaps the next burst of spikes in the closer motor neurone (Fig. 1 c).

When one of these interneurones is spiking during inspiration, hyperpolarizing potentials also occur in spiracular closer motor neurones (Fig. 2 a). These potentials reverse in polarity at a membrane potential more negative than the resting potential and are therefore inhibitory synaptic potentials (IPSPs) (Burrows, 1982). By triggering the sweep of an oscilloscope with a spike of the interneurone and superimposing several sweeps, an IPSP is seen to follow each spike with a consistent latency (Fig. 2 b, c). The association between spikes and IPSPs is made particularly clear when 256 spikes are used to trigger a signal averager and events in the motor neurone before and after the spike are examined (Fig. 3). Both sets of observations indicate a latency of 3·8 ms in one particular locust, from the peak of the spike to the onset of the IPSP. Taken together, it would seem appropriate to interpret these observations indicate that an interneurone synapses upon the closer motor neurone and there evokes IPSPs. There must always remain some uncertainty from physiological observations alone as to whether a connexion is direct, particularly in this example where the recording sites are 2 mm apart and separated by processes whose conduction velocities are not known. Nevertheless, there can be little doubt that the connexion is chemically mediated as the IPSPs have a reversal potential (Burrows, 1982). Further evidence that the spikes in an interneurone evoke IPSPs in the motor neurone is derived by passing pulses of current into an interneurone to increase its frequency of spikes during inspiration (Fig. 2 d, e). The result is that the frequency of IPSPs in the motor neurone is increased accordingly, and the overall level of hyperpolarization is accentuated.

Other output connexions of this interneurone can be inferred from previous recordings from spiracular motor neurones. The two mesothoracic spiracular closer motor neurones and the two closer motor neurones of the first median nerve of the metathoracic ganglion have common IPSPs during inspiration (Burrows, 1982). The two prothoracic closer motor neurones also have very similar patterns of IPSPs during inspiration to those in the homologous meso-or metathoracic motor neurones. It is to be expected therefore, given similar uncertainties to those described above that this interneurone also synapses upon these motor neurones. No other thoracic motor neurones or interneurones have been sought as possible targets for the output of this interneurone. The morphology of an interneurone (Fig. 4) suggests that it would not synapse upon abdominal neurones so that no search has been made for connexions with these.

The number of interneurones evoking IPSPs in the closer motor neurones can be deduced from the following observations. The IPSPs during inspiration wax and wane in amplitude forming a sequence of ‘beats’ (Fig. 2). This suggests that there are at least two interneurones responsible for the IPSPs whose frequency of spikes is slightly different. Recordings from any of the thoracic spiracular closer motor neurones show this phenomenon. A convenient example to consider further is a prothoracic closer motor neurone as the long pro-mesothoracic connectives allow the next experiment. Cutting one of these connectives abolishes the beats and reduces the number of IPSPs distinguishable during inspiration. The inference is that cutting the connective has abolished one source of the IPSPs. A similar result can be obtained by manipulating the frequency of spikes in a single interneurone by passing depolarizing or hyperpolarizing current through the intracellular microelectrode. The Result is to alter the frequency of beats of the IPSPs, or to abolish them completely depending upon the extent to which the frequency of spikes has been changed (Fig. 9 a, b). The simplest interpretation of these results is that there is a pair of symmetrically arranged interneurones which evokes the IPSPs; one with an axon in the left and the other with an axon in the right connective of the thoracic central nervous system.

Following physiological characterization using microelectrodes containing cobaltous chloride, the morphology of an interneurone was revealed by the injection of cobaltous ions (Fig. 4). Interneurones that have the physiological properties described above have the same morphology. Two such interneurones could be recognized morphologically, one with its cell body on the left of the ganglion, the other with its cell body on the right. The cell body of one of these interneurones is approximately 30 μm diameter and lies within the ventral cortex of cell bodies. Its co-ordinates are given by a line drawn across the ganglion level with the anterior edge of the main nerve to the hind leg (nerve 5) and an antero-posterior line from the middle of the ipsilateral anterior connective. The primary neurite runs medially and dorsally from the cell body to form a short expanded segment on the midline of the ganglion and level with the anterior edge of nerve 3 (Figs. 4, 5). The majority of neuropilar branches arise from this expanded segment. They are profuse and extend from the expanded segment to form a roughly circular area with a radius of 300μm, when view’ed dorsally. The branches which have a narrow and even diameter extend to both the left and right sides of the ganglion.

The majority of the branches in this region are restricted to the neuropile of the first abdominal ganglion which is fused to the metathoracic ganglion (Fig. 5 a). A few extend into the neuropile of the metathoracic segment at the level of the expanded segment (Fig. 5 c, d), whilst the more anterior branches are clearly within the dorsal neuropile. Within the first abdominal segment there are branches around both the dorsal and ventral edges of the neuropile (Fig. 5 e). The axon also arises from the expanded segment approximately at the level of the section shown in Fig. 5 c. It runs diagonally to the side of the ganglion opposite to the cell body and then turns anteriorly to enter the Lateral Dorsal Tract (Fig. 5 a, b). The axon continues to run anteriorly in this tract and eventually enters the anterior connective where it occupies a position in the dorsal-lateral quadrant. Before entering the connective however, the axon gives rise to some medially and some laterally directed branches in the most dorsal region of the neuropile (Figs. 4, 5 a). These have an appearance quite distinct from the branches arising more posteriorly. They are sparsely branched and with an uneven diameter along their length which lends them a beaded appearance. There is almost no overlap between these sparse anterior branches and the more profuse posterior ones.

Within the mesothoracic ganglion the branches of an interneurone are more sparse. The majority are restricted to the same side of the ganglion as the axon, but some project across the midline particularly in the more posterior regions. Many branches project to the same regions of the neuropile as do the branches of the spiracular closer motor neurones (see Burrows, 1982). The axon enters the pro-mesothoracic connective but has, as yet, not been traced to the prothoracic ganglion.

Currents of increasing intensity applied to an interneurone progressively reduce the number of spikes and reveal a persistent and rhythmical fluctuation of the membrane potential in time with ventilation (Fig. 6a-c). The membrane is gradually depolarized during one expiratory phase, reaches a peak when the excitatory synaptic input to a spiracular closer motor neurone stops and is usually maintained at this level during inspiration (Fig. 6 b, c). Throughout the depolarizing phase there are small fluctuations probably caused by synaptic potentials. However, no single sequence of synaptic potentials can be identified in these intrasomatic recordings as the cause of the overall depolarization.

At the start of the next expiration the membrane is repolarized rapidly at the same time as the depolarization in a closer motor neurone (Fig. 6 d, e). The inputs to the interneurone now occur in waves similar to the excitatory waves in a spiracular closer motor neurone (Fig. 6 J, e). When the waves are particularly prominent in the two neurones, it appears that they are common and evoked by the same presynaptic intet neurones. When the interneurone is hyperpolarized the waves decrease in amplitude and reach an apparent reversal potential, indicating that they are inhibitory synaptic potentials.

Each burst of spikes in an interneurone begins with a few at a low frequency, the number of these being correlated with the amount of overlap with the preceding expiration (Fig. 7 a). Likewise at the end of a burst there may also be spikes at a low frequency, the number of which is correlated with the amount of overlap with the next expiration. The spikes that do not overlap with expiration occur at a steac frequency. For example, at a ventilatory rate of 1–2 Hz the spikes in one interneuroi occurred at 66 Hz (Fig. 7 a). The small amount of deviation from this basic frequem is revealed by a histogram of the intervals between 5000 spikes in approximately 15 bursts (Fig. 7 b). The histogram is narrow with most intervals between 12 and 20 n (corresponding to instantaneous frequencies of 83 and 50 Hz) and with a peak 15 ms (66 Hz). There is a tail of longer intervals caused by the spikes at the beginnir and end of each burst.

The number of spikes in a burst is, however, variable as is the duration of tl ventilatory cycle itself. For example, in one locust examined, the number of spik ranged from 25 to 58, although the basic frequency of spikes in either the short or 1 the long bursts remained the same. The number of spikes is correlated (correlatic coefficient = 0·92) with the duration of the ventilatory cycle (Fig. 8 a): the short the ventilatory cycle the fewer spikes in each burst of the interneurone.

Injection of current into an interneurone through an intracellular microelectroc alters the frequency of its spikes but has no effect upon the frequency of the ventila: tory rhythm. There is no correlation between the number of spikes allowed in a interneurone and the duration of the ventilatory cycle (Fig. 8b). Occasionally manipulation of the frequency of spikes in an interneurone is correlated with a change in the ventilatory rhythm. These occasional correlations can be attributed to unpredicted changes in the ventilatory rhythm and are therefore not causal correlations. Likewise, pulses of current injected into an interneurone have no detectable effect upon the timing of the next phase of ventilation whether they occur during expiration and increase or decrease thé level of polarization, or during expiration and increase (Fig. 2d) or decrease the frequency of spikes. In other words, the ventilatory rhythm is not reset by these manipulations of an interneurone.

Nevertheless an interneurone can have quite pronounced effects on the form of the motor output of the spiracular closer motor neurones. When an interneurone spikes in its normal pattern there are usually few if any motor spikes during inspiration (Fig. 9 a). The expiratory burst of spikes begins at high frequency, continues with the spikes in groups and then declines slowly in frequency (Fig. 10). When the interneurone is depolarized, the level of hyperpolarization in a motor neurone during inspiration is increased (Fig. 96) and the frequency of spikes at the start of expiration is increased (Fig. 10). The duration of the burst is reduced because it is terminated earlier by the higher frequency of spikes in the interneurone (Fig. 9b, 10). There is no effect upon the pattern of spikes in the middle of the expiratory burst.

When the interneurone is hyperpolarized, spikes occur in the motor neurone throughout inspiration at a frequency inversely related to the number of spikes in the interneurone (Figs. 9c, d; 10). The initial frequency of spikes during expiration is now lower and the burst itself is prolonged. The transition from expiration to inspiration is also more blurred.

The interneurones described here co-ordinate the movements of the thoracic spiracles in the ventilatory rhythm. They would appear from physiological evidence to synapse upon spiracular closer motor neurones in each segment evoking synaptic potentials that are chemically mediated and inhibitory. The result is to inhibit spikes in these motor neurones during inspiration. The spiracular closer motor neurones are unusual in that their axons divide within an unpaired median nerve to innervate muscles on both the left and the right sides of the body. Both anatomical and physiological evidence indicates that the interneurones themselves are also symmetrically arranged in the central nervous system. The most parsimonious interpretation of the common IPSPs in the two closer motor neurones of one segment is that they are evoked by just two interneurones. Thus one interneurone has its soma on the left side of the metathoracic ganglion and its axon in the right anterior connective, whilst the other interneurone has its soma on the right and its axon on the left. Further symmetry is added by the fact that the two interneurones each synapse upon the two closer motor neurones in a particular segment. The co-ordination of the ventilatory rhythm in the thoracic spiracles, particularly in the mesothorax where there are only closer motor neurones would thus appear to be quite simple. The two interneurones described here are responsible for the inhibition during inspiration and probably two, yet to be identified but inferred from the synaptic potentials they evoke (Burrows, 1975b), are responsible for the excitation leading to spikes during expiration.

The interneurones which convey the ventilatory rhythm to the motor neurones would appear, in turn, to be driven by other interneurones. During expiration they receive an inhibitory synaptic input which usually abolishes their spikes. This input begins before the end of the preceding inspiratory phase and persists at the start of the next inspiratory phase. The effect is to lengthen the intervals between the spikes of an interneurone on both occasions. The inhibitory input appears to be derived from the same source as that which excites spiracular closer motor neurones and some flight motor neurones and which inhibits some inspiratory motor neurones (Burrows 1975 a, b). The underlying cause of the burst of spikes at a steady frequency during inspiration is less clear. An applied hyperpolarization fails to give clear evidence of a causal synaptic input. Thus it cannot be said whether an interneurone is actively depolarized by a synaptic input, or simply returns, following the cessation of the inhibitory input, to a membrane potential that is lower than the threshold for spike initiation. It was not possible to examine the interneurone in the absence of a ventilatory rhythm to determine whether it would then produce spikes at the frequency observed during the plateau phase of inspiration.

The two interneurones do not appear to be directly connected to each other. Manipulation of the output of one interneurone does not affect the output of the other as judged by the IPSPs it produces in the closer motor neurones. Moreover, the IPSPs in the motor neurones show ‘beats’ indicating that the spikes in the two interneurones are not closely linked.

The tests so far employed do not reveal any inherent rhythmicity in the interneurones. Pulses of depolarizing or hyperpolarizing current applied to an interneurone are without effect upon the ventilatory rhythm. This limited test also indicates that an interneurone is unable to affect directly, or by altering the motor output and hence the sensory feedback, other interneurones involved in producing the ventilatory rhythm. By contrast, pulses of current injected into an interneurone which conveys the ventilatory rhythm to the abdominal part of the nervous system, can reset this rhythm (Pearson, 1980). This abdominal interneurone can also alter the frequency of the ventilatory rhythm if it is depolarized or hyperpolarized (Pearson, 1980), but similar tests applied to a thoracic interneurone revealed no consistent effects. Nevertheless the thoracic interneurones do control the pattern of the motor output by influencing both the duration of the expiratory bursts of spikes in the closer motor neurones, and their initial frequency.

The emerging picture of ventilation in the locust can be summarized as in Fig. 11Despite good descriptions of the pattern of activity in abdominal motor neurones (Lewis et al. 1973; Hustert, 1975), of thoracic spiracular motor neurones (Miller, 1960; Burrows, 1978; 1981), of a presumed abdominal co-ordinating interneurone (Pearson, 1980) and here of thoracic interneurones, no speculation is worthwhile as to the mechanism by which the rhythm is itself generated. Clearly the task now at hand is to identify more interneurones and describe their interactions.

This work was supported by NIH grant 1RO1 NS16058-01.