Sequential Synaptic Excitation and Inhibition Shape Readiness Discharge for Voluntary Behavior

See allHide authors and affiliations

Science  15 Apr 2011:
Vol. 332, Issue 6027, pp. 365-368
DOI: 10.1126/science.1202244


How do animals initiate voluntary behavior? A key phenomenon in neuroscience is the readiness or preparatory neural activity in specific regions of the animal brain. The neurons and synaptic mechanisms mediating this activity are unknown. We found that the readiness discharge is shaped by sequential synaptic excitation and inhibition in the brain of crayfish (Procambarus clarkii). The readiness discharge neurons extended axon collaterals that appeared to activate recurring local interneurons. Therefore, we propose that the readiness discharge is formed by sequential synaptic events within the brain without feedback signals from downstream ganglia. The circuit involved is suited for signal processing for self-generated voluntary initiation of behavior.

Before the initiation of voluntary behavior, neuronal activity shows a remarkable and characteristic change in specific brain regions (18). The spike activity called readiness discharge that transiently precedes the voluntary initiation of behavior has been reported extensively in primates (24) and rodents (57). However, the membrane potential dynamics underlying the readiness discharge are yet to be seen on the dendritic processes of relevant neurons in the brain. The readiness discharge in the brain of crayfish Procambarus clarkii (8) might offer a clue to understanding the cellular mechanisms subserving the readiness discharge in relation to voluntary initiation of behavior. In crayfish, the readiness discharge can be recorded in the circumesophageal commissure (8), which contains axons of brain neurons descending to thoracic ganglia where the central pattern generator for walking resides (9). They increase their spike discharge rate a few seconds before the behavioral onset and decrease it once the walking behavior is initiated (8). The transient spike activity before the voluntary behavioral initiation is a common key feature of the readiness discharge. However, it is unknown what the synaptic mechanisms of the readiness discharge are and what types of neurons are involved in shaping the spike activity.

To investigate the neurophysiological mechanisms underlying the voluntary initiation of walking behavior, we performed intracellular recordings from single cells in the brains of behaving crayfish on a spherical treadmill (8). No external stimulus was applied so that they initiated, continued, and terminated walking behavior at their own disposal. We obtained physiological data from 69 neurons, which were successfully stained with the fluorescent dye Lucifer yellow. Identified neurons were categorized into three groups functionally and into two groups morphologically. Functional categorization was based on the timing of their spike activity in relation to the onset and offset of behavior that were statistically estimated on the basis of electromyographic (EMG) recording from the mero-carpopodite flexor muscle of the second walking leg on either side (10). Because the muscles of all eight walking legs were activated almost simultaneously within the time range of 65.89 ± 6.40 msec before the onset of leg movement (10), we categorized the synaptic and spike activities as “preceding” when the change occurred more than 500 msec before the behavioral onset. Morphologically, we categorized the neurons into descending neurons with their axons in the commissure and local interneurons whose neuritic projection was confined to within the brain. Because the initiation and motor control of locomotion have been related to the central complex in insects (11, 12), we scrutinized the homologous neuropils (13) in the crayfish brain and characterized the distribution of dendritic processes of behaviorally relevant interneurons to identify the main region of synaptic activation involved in the voluntary behavior.

We successfully penetrated the dendrite of one readiness discharge neuron and recorded the membrane potential change associated with voluntary initiation of walking 18 times (Fig. 1A). The animal showed a variety of walking including forward and backward walking (Fig. 1A). Irrespective of walking direction, the membrane potential showed a transient depolarization before the behavioral onset (Fig. 1A). It then showed a long-lasting hyperpolarization once the animal started walking (Fig. 1A). This sequential change of membrane potential controlled the transient discharge of action potentials before the behavioral onset. Eighteen superimposed traces of membrane potential and smoothed electromyographic recordings indicate the consistency of the sequential change (Fig. 1B). The peri-onset time spike histogram (Fig. 1B) shows that the spike discharge transiently increased before the onset and then decreased sustainedly (Fig. 1B). This change in the spike discharge rate is similar to the readiness spike activity recorded in the circumesophageal commissure (8).

Fig. 1

Membrane potential dynamics of a readiness discharge neuron. (A) Membrane potential (Vm) changes of a readiness discharge neuron aligned to the behavioral onset with walking trajectories shown on the right. (B) (Top) An overlay of the membrane potential traces and of the smoothed muscle activity traces. (Bottom) Raster display and peri-onset time histogram illustrating the spike activity. (C) (Top) Effects of current injection into the cell on the responses to synaptic input. (Bottom) Voltage responses of the cell to constant-current pulse injection. (D and E) Expanded voltage changes shown with the bars in (C). (E) (Right) Comparison of standard deviation (SD) of Vm between at-rest and pre-onset.

Membrane potential change that would cause transient spike discharges preceding the onset of walking can be based on synaptic input from other upstream cells or endogenous conductance change as in pacemaker cells. When a slight positive (+0.1, +0.2, +0.3 nA) current was injected into the neuron, the number of action potentials increased before the behavioral onset, whereas after the onset the hyperpolarization was remarkably enhanced with no spike discharge superimposed (Fig. 1C). Injection of a stronger depolarizing current caused spike discharges of the cell (25 to 65 spikes/sec), but no noticeable motor effect was observed in the EMG activity. When a negative current (–1 nA) was injected into the neuron, the number of spikes decreased and a slight depolarization was observed before the behavioral onset (Fig. 1C). When a stronger negative current (–2 nA) was injected, the transient spike discharge before the behavioral onset completely disappeared and the membrane potential showed a noticeable depolarization after the onset (Fig. 1C). These findings suggest that the transient spike discharge before the behavioral onset of walking is caused by excitatory synaptic input followed by inhibitory synaptic input. To investigate the membrane conductance change during these synaptic inputs, we injected constant current pulses into the neuron (–1 nA, 2 Hz) with the bridge balance circuit kept optimized (Fig. 1C). The voltage response was transiently decreased before the onset corresponding to the period of readiness discharge, subsequently recovered the magnitude observed at rest, and then began to decrease gradually about 5 s after the behavioral onset (Fig. 1, C and D). The first transient decrease in the membrane resistance suggests that the readiness discharge before the behavioral onset of walking could be caused by an excitatory synaptic input, whereas the late sustained decrease in the membrane resistance suggests that the later phase of postonset hyperpolarization is caused by an inhibitory synaptic input. The early phase of hyperpolarization that immediately follows the behavioral onset is thought to be caused by another inhibitory synaptic input that is located on the dendrite electrotonically more distant than the late inhibitory input because no membrane conductance change was detected for the early hyperpolarization. Before the behavioral onset (from –5 s to 0 s) compared with at rest (from –25 s to –20 s), we observed a significant rise (paired t test, P < 0.01) in the variance of membrane potential fluctuation from which action potentials were removed (Fig. 1E, gray filled circles). Furthermore, a notable rise in the variance of membrane potential fluctuation was observed when the cell was hyperpolarized (Fig. 1E, black filled circles), when most voltage-gated ion channels are considered to be closed. These observations suggest that the increased membrane conductance before the behavioral onset is due not to endogenous but to synaptic mechanisms.

The morphology of the readiness discharge neurons shown in Fig. 1 was investigated by iontophoretically injecting Lucifer yellow into the cells. The neurons extended their axons through the medial and ventral region of the esophageal neuropil where they branched into many collaterals, down to the circumesophageal commissure in its ventromedial quarter (fig. S1) corresponding to the Wiersma’s area 70 (14). This area contained the fibers that evoked walking behavior upon electrical stimulation (8, 15, 16). The somata were located in the ventral paired anterior cluster (VPAC) of the protocerebrum. The neurons projected their dendrites contralaterally in the anterior and posterior medial protocerebral neuropil (AMPN and PMPN), as well as in the medial antennal neuropil (MAN) of the deutocerebrum (Fig. 2, A to D). These characteristics were mostly shared with candidate readiness discharge units obtained by backfill staining from the cut end of a small bundle including the unit (8) (Fig. 2C). It should be noted here that the branches in the esophageal neuropil (OEN) and the antennal neuropil (AnN) showed varicose ramifications (Fig. 2, A, B, and E), suggesting that these were presynaptic arborizations. Thus, they were not dendritic branches but axonal collaterals for sending the corollary signal of readiness discharge within the brain.

Fig. 2

Morphology of readiness discharge neurons. (A) Horizontal (left) and sagittal (right) projection images of the neuron stained after recording (Fig. 1). (B) Dendritic structure of the neuron on the horizontal (left) and sagittal (right) planes. (C) Horizontal (left) and sagittal (right) projection images of the neurons obtained by backfill staining. The left figure was adapted for the present comparison from our previous publication (8). (D) A schematic drawing of the crayfish brain. PB, protocerebral bridge; AMPN, anterior medial protocerebral neuropil; CB, central body; PMPN, posterior medial protocerebral neuropil; ON, olfactory neuropil; DCN, deutocerebral commissure neuropil; AcN, accessory lobe; LAN, lateral antenna I neuropil; MAN, medial antenna I neuropil; TN, tegumentary neuropil; AnN, antenna II neuropil; OEN, esophageal neuropil. (E) Varicose arborization of the neuron. The region shown by a dashed rectangle in (A) was enlarged.

In a previous study based on extracellular recordings, we proposed a descending parallel scheme for voluntary initiation, continuation, and termination of walking behavior (8). In this study, we also identified descending interneurons that represent the continuation and termination pathways, respectively (Fig. 3). The continuation neurons increased their spike discharge rate before the behavioral onset and a high rate was maintained during walking (Fig. 3A). The termination neurons were activated before the voluntary behavioral offset and became silent after the offset (Fig. 3B). Thus, the continuation and termination neurons were sequentially activated to complete voluntary walking.

Fig. 3

Physiology and morphology of continuation and termination neurons. (A) (Top) An overlay of two membrane potential traces of a continuation neuron. (Middle) An overlay of smoothed traces of muscle activity. (Bottom) A peri-onset time histogram illustrating the spike activity during walking. (B) (Top) An overlay of four membrane potential traces of a termination neuron. (Middle) An overlay of smoothed traces of muscle activity. (Bottom) Peri-offset time histogram illustrating spike activity at the time of behavioral offset. (C) Dendritic projection of a continuation neuron on the horizontal (left) and the sagittal (right) planes. (D) Dendritic projection of a termination neuron on the horizontal (left) and sagittal (right) planes.

In either type of descending neuron, intracellular recordings revealed that synaptic activity changed preceding the initiation of walking. However, in the termination neuron, the subthreshold change did not produce consistent action potentials before onset (Fig. 3B). The continuation neurons mainly projected their dendrites to the neuropils [AMPN, PMPN, and central body (CB)] and additionally to the MAN and OEN (Fig. 3C). The termination neurons mainly projected to the PMPN and AMPN (Fig. 3D).

We identified 45 additional descending neurons, including those that showed changes in synaptic activity without spike activity changes preceding, following, and delayed to the behavioral onset. The preceding and following types of neurons tended to project their dendrites to the medial protocerebral neuropil rather than the neuropils in the deutocerebrum or the tritocerebrum, whereas delayed-type neurons tended to project to the deutocerebrum and tritocerebrum (fig. S6A). Therefore, the confined arborization in the medial protocerebral neuropils and those synaptic activities suggest that the main synaptic activation for voluntarily initiated walking of crayfish takes place in the medial protocerebrum.

We demonstrated that the readiness discharge was regulated by synaptic excitation, followed by subsequent synaptic inhibition. A possibility to be considered here in relation to the inhibition is that it is caused by continuation neurons. The varicose arborization of the readiness discharge neurons (Fig. 2) and continuation neurons within the brain (Fig. 3C and fig. S2) could send signals on their own activity to other neurons within the brain. Because the dendritic regions of the readiness discharge neurons did not overlap with the output regions of the continuation neurons, their synaptic and spike activities are not directly but polysynaptically transmitted to the readiness discharge neuron. For voluntary locomotion, the parallel and sequential recruitment of descending neurons thus has to be organized locally within the brain. In this study, we encountered local neurons that showed activity changes associated with spontaneous initiation of walking (figs. S3 to S5). Although it remains unknown whether the descending neurons are organized by direct synaptic contacts of the local neurons, those spiking and nonspiking activities showed remarkable changes in association with the self-generated voluntary walking (figs. S3 and S4). Further experiments are necessary to define their connections and causal relationships. However, our data clarified the physiological and morphological characteristics of the brain neurons relevant for self-initiated voluntary walking in crayfish and showed that they are organized and activated by presynaptic brain neurons, not by endogenous mechanisms.

Supporting Online Material

Materials and Methods

Figs. S1 to S6


References and Notes

  1. Acknowledgments: This work was supported by Grants-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan (18657025 and 20370028).
View Abstract

Stay Connected to Science

Navigate This Article