INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The isolated spinal cord of the neonatal rat is a frequently used model of mammalian central pattern generation. A regular locomotor-like rhythm is produced in this preparation by bath application of various neurochemicals (Cazalets et al. 1992; Cowley and Schmidt 1994; Kjaerulff and Kiehn 1996; Kremer and Lev-Tov 1997; Kudo and Yamada 1987; Smith et al. 1988). Studies of the neonatal rat spinal cord showed that the rhythmogenic capacity associated with hindlimb locomotion is maximal at the caudal-thoracic and rostral-lumbar segments (Cazalets et al. 1995; Cowley and Schmidt 1997; Kjaerulff and Kiehn 1996; Kremer and Lev-Tov 1997; Tresch and Kiehn 1999) and that it decreases gradually in the rostrocaudal direction (Cowley and Schmidt 1997; Kjaerulff and Kiehn 1996; Kremer and Lev-Tov 1997; Tresch and Kiehn 1999). In our recent studies, we showed that the rhythmogenic capacity of the mammalian spinal cord is not restricted to its caudal-thoracic and the limb moving segments and that activation of sacrocaudal afferents (SCA) produced alternating left-right efferent bursts in lumbosacral efferents and rhythmic tail movements. The rhythm persisted in the sacral cord after surgical removal of the lumbo-thoracic and cervical segments of the cord (Lev-Tov et al. 2000b). More recent studies showed that the rhythmogenic capacity of the sacral cord could be activated also by bath-applied N-methyl-D-aspartate (NMDA) and serotonin (5HT) (Cazalets and Bertrand 2000; Lev-Tov and Delvolvé 2001). This rhythm, however, is not as regular as the SCA-induced rhythm, and additional or different neurochemicals are often required to sustain it (Lev-Tov and Delvolvé 2001). The present study was aimed at establishing the optimal conditions required for inducing drug-free activation of the sacrocaudal rhythm and at testing the dynamic range of the rhythm as a function of the stimulation parameters. The study was also aimed at understanding the activity pattern of the principal motoneuron pools of the sacrocaudal region during the rhythm and its relevance to the organization of the sacrocaudal circuitry. This aim was obtained by studying the movements produced by SCA stimulation and the pattern of muscular activity underlying these movements. Some of the preliminary findings appeared in an abstract (Lev-Tov et al. 2000a) and in a mini-review (Lev-Tov and Delvolvé 2001).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Preparations

Spinal cord preparations were isolated from P4-P6 ether-anesthetized rats with or without an intact tail (Lev-Tov et al. 2000b). The cord was transferred to a recording chamber and superfused continuously with an oxygenated Krebs saline (e.g. Kremer and Lev-Tov 1997; Lev-Tov et al. 2000b).

Stimulation and recordings

Slow ventral root potentials (VRPs) were recorded by suction electrodes from pairs of sacral ventral roots at 0.1 Hz to 10 kHz using a high-gain AC amplifier. Sharp electrode intracellular recordings were obtained from S₂-S₃ motoneurons impaled from the ventral or ventrolateral aspect of the cord and identified by the presence of antidromic spikes. Microwire electromyographic (EMG) recordings (100 Hz to 10 kHz) were obtained from the flexor caudae longus (FCL), extensor caudae lateralis (ECL), and abductor caudae dorsalis (ACD). Muscle stimulation was obtained by constant-current square-pulses applied to the EMG microwires. Rhythmic activity was obtained by repetitive stimulation of SCA. The threshold (T) was measured at the beginning of each experiment from short-latency (monosynaptic) responses induced by single-pulse stimulation in the homologous or one of the adjacent ventral roots.

Video recordings and analyses

Tail movements in the mediolateral direction were monitored by a frontally positioned video camera. Movements in the dorsoventral plane were reflected by a mirror and monitored concurrently. EMGs produced in two different pairs of tail muscles (left and right flexors and abductors or flexors and extensors) during the video-monitored movements were recorded using a pulse code modulated (PCM) recorder. The activity of the left flexor was recorded also on the audio track of the video camera and used as a timing mark to match the video-recorded movements and the PCM-recorded EMGs. Video clips of tail movements were stored on a computer hard disk at 25 fps using a frame grabber, and the movements were analyzed from the respective video frames. Consecutive stick diagrams (see Figs. 5 and 6) constructed from four digitized reference points along the tail (base, proximal, distal, and tip) describe movements in the mediolateral and dorsoventral planes [after correcting the distortions (angle, direction) of the reflected images]. To allow better understanding of the temporal relation between the activity of the tail musculature and corresponding movements in the mediolateral plane, the data digitized from the video frames were displayed as a function of time and expressed as displacements from the midline.

Statistical analysis

Three characteristics of the rhythm were measured and tested: the cycle time, burst duration, and phase (with respect to a desired reference). The frequency of motoneuron firing during bursts was measured in some of the intracellular studies. The cycle time, burst duration, and firing frequency were analyzed by linear statistics while the phase data were analyzed by circular statistics. Because the duration of regular rhythmic activity induced by afferent stimulation is limited, there was a need to repeat identical stimulus trains a number of times during each experiment. The resultant data were pooled only if one-way ANOVA (linear data) or Watson and Williams test (circular data) revealed no significant differences between the data samples. Similar approach was used to pool data samples from different experiments in a given series.

ANOVA followed by Tukey method for multiple comparisons or by Tumhane method (when nonequal variance was detected by Bartlett's test) was used to compare means of cycle time, burst duration, and firing frequency of different stimulation frequencies or intensities.

The mean phase and the vector r that describes the concentration of phase values around the mean were calculated from the raw phase values. Rayleigh's test was used to determine whether the phase values are uniformly distributed around the circle. One-sample test for the mean angle (Zar 1984) was applied to determine whether the onset of activity of ipsilateral abductors and extensors during the rhythm is different from that of the flexors (used as the 0-cycle reference). Multi-sample testing was performed to compare the mean phase values of any pair of tested factors (the Watson-Williams test) (Zar 1984).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Induction and maintenance of sacrocaudal rhythmicity by afferent stimulation

Although the rhythmogenic sacrocaudal networks could be activated by drug-applied neurochemicals (Cazalets and Bertrand 2000; Lev-Tov and Delvolvé 2001), controlled drug-free activation of these networks is obtained by mechanical or electrical stimulation of SCA (Lev-Tov et al. 2000b). Figure 1A shows rhythmic activity recorded from S₂ ventral roots during a 50-pulse 1-Hz stimulus train applied to the right S₄ dorsal root. The cycle time of the rhythm was fairly constant, while the burst duration increased with time and then reached a plateau (Fig. 1A, bottom). Analysis of the phase difference between the left and right efferent bursts revealed a mean phase shift of 0.51 ± 0.06 (mean ± circular SD), n = 22 cycles. The phase concentration vector (r) was 0.94. 

View larger version (46K): [in this window] [in a new window]   Fig. 1. The regularity of the sacrocaudal rhythm. Recordings (50 Hz to 10 kHz) of rhythmic activity from the left and right ventral roots of S2 in 2 different preparations (A and B) during a 50-pulse, 1-Hz stimulus train applied to the right S4 dorsal root at 1.6 T (A) and at 2 T (B). Raw data of the cycle time and burst duration during this train are plotted as a function of time below (A and B, bottom left). The circular distribution of the raw phase values measured during the train (single train in A and 2 identical trains in B) are shown superimposed with the r vector describing the concentration of the phase values around the mean (A and B, bottom right).

In some preparations, the cycle time of the rhythm was slowed down during the train and the firing became less intense toward the end of the train (Fig. 1B). The phase relation remained relatively stable under these conditions (0.5 ± 0.08 cycle, r-vector = 0.87, n = 33 cycles in 2 identical stimulus trains). To determine whether the changes in cycle time and burst duration reflect the beginning of a possible breakdown of the rhythm, we prolonged the stimulus train and monitored the changes in the sacral efferent activity. Figure 2A shows that stimulation of the right S₄/Ca₁ dorsal roots at 1 Hz and 1.9 T produced a regular left-right alternating rhythm (left). Stimulation of the left S₄/Ca₁ dorsal roots at 1 Hz and 2T produced a similar rhythm (right). When the stimulus train applied to right S₄/Ca₁ was prolonged (Fig. 2B), the rhythm was replaced by bilateral tonic activity, then long-duration intermittent rhythmic bursts could be detected (B1), and finally the efferent activity was virtually blocked (B2). Stimulation of the left S₄/Ca₁ dorsal roots at this stage (Fig. 2B3, ) restored the rhythm. Thus the suppression of the rhythm during the train does not reflect an impairment of the rhythmogenic circuitry but rather a failure of the stimulated afferents to sustain a steady rhythmic drive.

View larger version (25K): [in this window] [in a new window]   Fig. 2. Prolonged activation of sacrocaudal networks. A: rhythmic activity elicited by 50-pulse 1-Hz stimulus trains of the right S4-Ca1 dorsal roots at 1.9 T (Right SCA, left) and the left S4-Ca1 at 2 T (Left SCA, right) was recorded from the left and right (L and R) S2 ventral roots at 0.1 Hz to 10 kHz. B: sample recordings from the same preparation taken during a prolonged repetitive stimulus train of the right S4-Ca1 dorsal root at 1 Hz, 113, 211, and 297 s after the beginning of the train (1, 2, and 3, respectively).  in 3, the time at which the stimulation of the ipsilateral dorsal root was stopped and stimulation of the contralateral S4-Ca1 dorsal root (left SCA) started.

Modulation of sacrocaudal rhythmicity by the intensity of afferent stimulation

Our previous intracellular studies of sacral motoneurons showed that stimulation of sacral afferents induced slow phasic membrane potential oscillations superimposed on a progressive depolarizing drive (Lev-Tov et al. 2000b). Figure 3, A and B, shows intracellular recordings from a sacral motoneuron and VRP recordings from the same segment during rhythmic activity induced by 50-pulse, 4-Hz trains at different stimulation intensities. Stimulation of the afferents at 1 T failed to produce the rhythm (not shown). Stimulation at 1.2 T (Fig. 3A) produced eight activity cycles superimposed on a 15-mV depolarization. Bursts of low-frequency spikes (mean firing frequency per burst: 11.6-16.8 Hz) were elicited at the peaks of the last six cycles. The concurrently recorded VRPs exhibited rhythmic cycles with marginal spiking activity. Changing the stimulation intensity from 1.2 to 6 T (Fig. 3B) increased the depolarizing drive to 30 mV, elevated the firing rate during the oscillatory peaks to means of 28-56 Hz per burst, and significantly decreased (t-test, P < 0.01) the cycle duration from 1.58 ± 0.26, n = 8, to 1.08 ± 0.22, n = 11. The rhythmic efferent firing recorded from the ventral root was augmented substantially under these conditions.

View larger version (38K): [in this window] [in a new window]   Fig. 3. Effects of the intensity of afferent stimulation on the sacrocaudal rhythm. Intracellular recordings from a motoneuron in the right S2 segment (R-S2 MN) are superimposed with extracellular recordings of ventral root potentials (VRPs, 0.1 Hz to 10 kHz) from the left and right (L, R) S2 ventral roots. Rhythmic activity was induced by 50-pulse, 4-Hz stimulus trains at 1.2 and 6 T in A and B, respectively. The resting membrane potential before the train ( · · · ) was 77 mV. Effects of stimulation intensity on the cycle time of the rhythm are shown in C. The rhythmic threshold of the afferents (RT) was determined for each preparation during 50-pulse 5-Hz trains, and the tested stimulation intensities were normalized by these values. Data obtained on stimulation intensities of 1-1.25, 1.5, and 2-2.5 RT in 8 different experiments were pooled (METHODS), and the respective means of cycle duration ± SD were plotted as a function of intensity. The mean cycle duration was: 1.57 ± 0.45, 1.42 ± 0.34, and 1.16 ± 0.4 s, for 1-1.25, 1.5, and 2-2.5 RT, respectively.

In this and other experiments, a limited range of stimulation intensities could induce regular rhythmicity. The minimal stimulus intensity required to produce the rhythm at any given stimulation frequency was found to be higher than the measured threshold (T, see METHODS). This intensity, the "rhythmic threshold" (RT), was inversely related to the stimulation frequency. Above a given intensity level, dorsal root stimulation induced tonic excitation or inhibition and thereby perturbed the rhythmicity (not shown).

Quantitative analyses of the effects of stimulation intensity on the rhythm were performed on data from eight different experiments. The stimulation-intensity range for evoking rhythmic activity during 50-pulse, 5-Hz stimulus trains was determined for each experiment. These intensities were expressed with respect to the 5-Hz RT. Figure 3C shows that the cycle duration was inversely related to the stimulation intensity. The decrease in the cycle time was statistically significant: 1.5 RT < 1-1.25 RT > 2-2.5 RT (ANOVA followed by Tamhane method for multiple comparisons, P < 0.01).

Modulation of sacrocaudal rhythmicity by the frequency of afferent stimulation

Rhythmic activity could be produced by single stimuli and by stimulus trains applied at various frequencies. Because the rhythm produced by stimulation frequencies >10 Hz was often perturbed by tonic excitation or inhibition, the effects of frequency on the rhythm were tested over a frequency range of 1-10 Hz, and the stimulus intensity was adjusted to produce regular rhythmic activity over this range. Figure 4A shows intracellular recordings from an S2 motoneuron and VRP recordings from the same segment during and following 30-pulse stimulus trains applied to S₄-Ca₁ dorsal roots at 1 and 5 Hz. Regular rhythmic activity developed gradually during the 1-Hz train. The rhythm was superimposed on a slow rising (mean slope = 1.01 mV/s) depolarization, reaching 19 mV at the end of the train. The mean firing frequency per burst varied from 9 to 24 Hz. During the 5-Hz train, the depolarizing drive developed much faster (4.2 mV/s), reaching 14.1 mV at the end of the train, and the regular rhythmic activity was obtained from the beginning of the train. The mean firing frequency per burst was 11-21 Hz. The mean cycle time was 1.51 ± 0.3, n = 15 and 1.31 ± 0.21, n = 5 at 1 and 5 Hz, respectively.

View larger version (48K): [in this window] [in a new window]   Fig. 4. Effects of the frequency of afferent stimulation on the sacrocaudal rhythm. A: intracellular recordings from a right S2 motoneuron (R-S2 MN) and VRPs recorded from the left and right S2 ventral roots are shown during and following 30-pulse, 2.4-T stimulus trains applied to the right S4-Ca1 dorsal root at 1 and 5 Hz. The membrane potential before the train ( · · · ) was 62 mV in both cases. Note initial time locked bursts (at 1 and 5 Hz) followed by slow appearance of the rhythm at 1 Hz and a much faster development of cyclic activity at 5 Hz. Also note the progressive reduction in spike amplitude with the increasing depolarizing drive and the gradual recovery with the posttetanic decay of the drive. B: samples of efferent bursts recorded at 50 Hz to 10 kHz, from the left and right (L and R) S2 ventral roots during 50-pulse trains applied to the left and right Ca2-Ca3 dorsal root at 1, 5, and 10 Hz at 1.2 T. The superimposed frames denote a single activity cycle. C: the mean normalized cycle duration obtained in 5 different experiments is plotted as a function of the stimulation frequency. The cycle times obtained during 50-pulse stimulus trains applied at 1, 2, 4, 5, and 10 Hz in each experiments were normalized by the mean cycle time obtained at 1 Hz (that way, the effect of variations in stimulation intensity in different experiments on the frequency response, was factored out). Data were compared by 1-way ANOVA and consequently pooled into 3 groups: 1-2, 4-5, and 10 Hz. Cases in which the cycle time was slowed down substantially during the train were omitted. The mean cycle time was: 1 ± 0.21, 0.81 ± 0.25, and 0.64 ± 0.53 for 1- to 2-, 4- to 5-, and 10-Hz trains, respectively. The mean normalized burst duration in these experiments was: 1.03 ± 0.36, n = 155; 0.88 ± 0.27, n = 117; and 0.7 ± 0.26, n = 64 cycles for 1-2, 4-5, and 10 Hz, respectively.

The VRPs recorded from the S₂ ventral roots during 50-pulse stimulus trains applied at 1, 5, and 10 Hz to the right S₄ dorsal root in another experiment are shown in Fig. 4B. The cycle duration (superimposed rectangles) decreased as the stimulus frequency was increased from 1 to 5 or 10 Hz. Figure 4C shows the results of quantitative analyses of five experiments performed in this series. The mean normalized cycle periods decreased significantly with the increased frequency of stimulation (10 Hz < 1-2 Hz > 4-5 Hz; 4-5 Hz > 10 Hz, ANOVA followed by Tukey method for multiple comparisons, P < 0.01).

The same was true for the duration of the efferent bursts (Fig. 4 legend). Significant shortening of the normalized burst duration was observed with increasing stimulation frequency: 10 < 1-2 > 4-5 Hz; 4-5 > 10 Hz (1-way ANOVA followed by Tukey method for multiple comparisons, P < 0.01). The phase lag between the left and right efferent bursts was found to be invariable on the different intensities and frequencies of afferent stimulation tested in our experiments.

Activity of the tail musculature and the movements produced following stimulation of SCA

Figure 5A shows stick diagrams describing the mediolateral tail movements during one of the activity cycles produced by mechanical stimulation of the mid-tail region. The tail (ventral side up) moved first from midline to left (I), then all the way to right (II) and then back to midline (III). Figure 5B shows the mediolateral rhythmic movement of the tip of the tail as a function of time. Five constant-amplitude cycles are clearly detectable. The stimulus also produced a prominent ventroflexion of the tail. The concurrent positions of the tail in the dorsoventral plane are displayed as stick diagrams on top (Fig. 5B, arrows denote the time of occurrence). After the stimulus (the 1st 360 ms), the tail was elevated gradually, and curled ventrally to reach a steady ventroflexed position (1st sequence of stick diagrams). This ventroflexion was sustained nearly throughout the rhythmic episodes, and the tail was lowered back to its prestimulus position just before cessation of the rhythm (right-most stick diagram).

View larger version (27K): [in this window] [in a new window]   Fig. 5. Tail movements and the activity of the tail musculature following sacrocaudal afferents (SCA) activation. A: stick diagrams describing 1 of 5 activity cycles of mediolateral movements produced by tail pinching (see METHODS). The stick diagrams are divided into 3 consecutive parts (I, II, and III) for clarity. The changes in the x-y coordinates are expressed as percentage of tail length (26.5 mm). B: stick diagrams describing the tail movements in the dorsoventral plane following the tail pinch described in A are superimposed with the time domain display of the concurrently monitored lateral movements of the tip of the tail. A bar and 2 arrows denote the time of occurrence of the movements described by the 1st 9 consecutive stick diagrams. The other stick diagrams represent the dorsoventral position of the tail at different poststimulus times (arrows). The heavy dotted segment of the time domain display denotes the cycle described in A. The peaks and troughs in the time domain display of the abductions correspond to maximal right and left abduction, respectively. C: electromyographic (EMG) recordings of right (R) and left (L) tail flexors and abductors are superimposed with 2 identical traces describing the resultant abductions of the proximal tail. Based on simultaneous video and EMG recordings. The dotted reference lines in B and C denote the midline (x = 0). D: bar diagrams of the activity of the principal tail muscles following 10-pulse 10-Hz stimulus trains of the S4 dorsal roots. The bars denote the mean burst duration (from the mean onset to offset of the EMG bursts) of each muscle normalized by the cycle period of the flexor caudae longus (FCL). The circular standard deviations of the onset and offset of the EMG bursts are superimposed. Data were pooled from 6 different experiments. Ipsi- and contralateral refer to the side of stimulated afferents (either left or right S4 dorsal roots in this case). The r vector and the number of cycles were ipsilateral-onset: flexors (n = 134), extensors (r = 0.89, n = 24) and abductors (r = 0.94, n = 84); ipsilateral offset: flexors (r = 0.84, n = 131), extensors (r = 0.84, n = 18), abductors (r = 0.88, n = 82); contralateral onset: flexors (r = 0.88, n = 131), extensors (r = 0.85, n = 16), and abductors (r = 0.85, n = 85); contralateral offset: flexors (r = 0.8, n = 126); extensors (r = 0.63, n = 13) and abductors (r = 0.84, n = 85).

Movements induced after short stimulus trains (5-10 pulse, 10 Hz) were similar to those described in the preceding text (ventroflexion accompanied by alternating abductions). The relation between the activity of the tail musculature and the movements following mechanical stimulation of the tail is shown in Fig. 5C. Coactivation of the left flexors and abductors was observed during left abduction, and coactivation of the right flexors and abductors was evident as the tail moved all the way to the right.

The temporal relation between the activities of tail flexors extensors and abductors following short stimulus trains (data were pooled from 6 different experiments) is demonstrated in Fig. 5D. The bars represent the average burst duration normalized by the cycle time of the tail flexors. This diagram shows that in addition to the longer duration of the flexor bursts, the flexors were recruited significantly (P < 0.01) before the abductors, and nonsignificantly (P > 0.05) after the extensors (Rayleigh's 1-sample test for the mean angle).

Activity of the tail musculature and the movements produced during SCA stimulation

Figure 6A shows that during the stimulus train (rectangle), the central part of the tail moved laterally to the left and right while the tip and base of the tail remained virtually immobile (stick diagrams on the left). These low-amplitude excursions were gradually attenuated toward the end of the stimulus train (see the time domain display). High-amplitude abductions appeared immediately after the stimulus train, (stick diagrams on the right, and the superimposed time domain display). These latter movements resembled those produced following mechanical and short-train activation of SCA (e.g., Fig. 5). The rhythmic movements during and following the long stimulus train were superimposed on a sustained ventroflexion (not shown). This ventroflexion developed gradually during the train, reaching a maximal level toward the end of the train and declining in amplitude after the train as the high-amplitude abductions began. The tail returned to its initial horizontal position on termination of the rhythmic movements (not shown).

View larger version (45K): [in this window] [in a new window]   Fig. 6. Tail movements produced during long stimulus trains of SCA and the muscle activities associated with them. A: lateral movements of the 4 digitized regions of the tail are shown as a function of time during (frame) and following 50-pulse, 10-Hz train applied to the right S4 dorsal root of a tail spinal cord preparation. Two sets of stick diagrams corresponding to activity cycles sampled during and after the train (heavy dots) are superimposed above. Each of these sets is also shown subdivided into 3 consecutive groups for clarity. Peaks and troughs in the time domain display are maximal right and left abduction, respectively. B: EMG recordings of the left (L) and right (R) tail flexors and abductors are shown superimposed with 2 identical traces describing the concurrently monitored abductions of the "proximal" tail. The superimposed frame denotes the stimulus train.

Simultaneous EMG and video recordings showed left-right activation of flexors and a suppressed activity of abductors during the train, followed by a strong coactivation of flexors and abductors at a given side of the tail, after the stimulus train (Fig. 6B). Thus the low-amplitude abductions during the train were produced by alternating left-right activation of flexors, while the high-amplitude abductions after the train involved coactivation of flexors and abductors on a given side of the tail, and alternating left-right activation of these muscles (Fig. 6B). Interestingly, the suppression of the activity of abductors/extensors during the train was enhanced with the intensity of afferent stimulation. The left-right alternating flexor activity persisted and was replaced by bilateral tonic bursts of flexors only at very high stimulation intensities (not shown).

Analyses of EMGs recorded in six experiments during the stimulus trains (Fig. 7A) indeed showed that the majority of the flexor bursts during the stimulus trains (58%) was not accompanied by abductor bursts. Extensor bursts were totally abolished in 43% of the cases where flexor bursts were clearly detected. The duration of abductor and extensor bursts during the train was significantly shorter than that of the flexor bursts (ANOVA/Tukey, P < 0.001). The pronounced differences between the duration of the flexor bursts and that of the abductor and extensor are also evident in the bar diagrams in Fig. 7B. The bar diagrams also show that the onset of flexor activity was followed by a delayed onset of extensor and abductor activity. These delays were statistically significant (Rayleigh's 1-sample test for the mean angle, P < 0.01). These findings suggest that flexors were recruited before the extensors and abductors during the stimulus trains and that some of the stimulated afferents are capable of diminishing the abductor and extensor activity under these conditions.

View larger version (24K): [in this window] [in a new window]   Fig. 7. The activity patterns of the tail muscles during SCA stimulation. A: frequency histograms of the duration of flexor, abductor, and extensor bursts produced during 50-pulse 10-Hz stimulus trains in 6 different experiments. F, the number of failures to induce EMG bursts (burst duration = 0). The mean duration of the activity bursts was 0.51 ± 0.26, 0.25 ± 0.15, and 0.23 ± 0.09 s for flexor, abductor, and extensor, respectively. B: bar diagrams of the activity of the principal tail muscles during 50-pulse 10-Hz stimulus trains of the S4 dorsal roots. Means shown with circular SD were pooled from 6 different experiments (METHODS). The phase values were normalized by the cycle of the flexor caudae longus. Ipsi- and contralateral refer to the side of stimulated afferents (either left or right S4 dorsal roots in this case). The r vector and the number of cycles were ipsilateral-onset: flexors (n = 155), extensors (r = 0.75, n = 29), and abductors (r = 0.81, n = 44); ipsilateral-offset: flexors (r = 0.66, n = 150), extensors (r = 0.76, n = 26), and abductors (r = 0.88, n = 44); contralateral-onset: flexors (r = 0.81, n = 113), extensors (r = 0.78, n = 16), and abductors (r = 0.71, n = 41); contralateral-offset: flexors (r = 0.73, n = 109), extensors (r = 0.79, n = 12), and abductors (r = 0.84, n = 41). C: contribution of the principal tail muscles to the mediolateral tail movements. Abductions of the tip of the tail produced by alternating stimulus trains (10 pulse, 10 Hz) applied to the left (light bars) and right (heavy bars) abductor caudae dorsalis (abductors), extensor caudae lateralis (extensors), and flexor caudae brevis (flexors). Data were digitized from video frames sampled at 12.5 fps. The stimulus train to the right muscle (r-stim. bar) was applied as the stimulus train to the left muscle (l-stim. bar) was ended. The repetition rate was 0.5 Hz. The 2 sets of bars superimposed on the abductor data demonstrate the stimulation paradigm (L-stim. and R-stim. denote stimulation of the left and right muscle respectively). The activity of the unstimulated muscles was monitored during the experiments and the stimulation intensity was adjusted to exclude nonspecific muscle activation by current spread from the stimulation electrodes.

Direct unilateral stimulation of the tail muscles produces abduction of the tail

Because flexor activity was sufficient of producing rhythmic abduction of the tail, we assessed the relative contribution of each pair of the principal tail muscles to the abduction of the tail by application of alternating sequences of stimulus trains through the EMG microwires to the muscles. We first stimulated the left and right abductors, then the extensors, and finally the left and right flexors. The lateral movements of the mid-tail during these stimulus trains (see METHODS) are plotted as a function of time in Fig. 7C. These data show that both flexors and extensors are capable of producing significant abductions of the tail.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In the first part of this work, we studied the dynamic range of SCA-induced rhythms. We found that an alternating left-right rhythm could be induced in sacral efferents by activation of SCA and that increasing the stimulation intensity or frequency accelerated the rhythm by 30-40%. High stimulation intensities and rates perturbed the rhythm. The rhythm was blocked during prolonged (usually >50 s) stimulus trains of SCA but could be restored by stimulation of contralateral afferents.

In the second part of the study, we studied the tail movements produced by SCA and the underlying activity pattern of the tail muscles. Our main findings were that SCA activation produced ventroflexion followed by rhythmic abductions of the tail. Low-amplitude abductions appeared during stimulus trains and high-amplitude abductions appeared after the trains. During stimulus trains the activity of abductors/extensors was suppressed and the activity of flexors was not changed. After stimulus trains, we observed an alternating left-right activation of the tail muscles, and coactivation of the principal muscles on each side of the tail.

The findings described in the preceding text and their relevance to neural control of automatic movements are discussed in the following text.

Drug-free rhythmicity in the mammalian spinal cord, boundary conditions, and dynamic range

Sensory-induced rhythmogenesis has been reported in a number of preparations including lampreys (McClellan 1984), amphibian tadpoles (Soffe 1991), cats (Grillner and Zangger 1979), and neonatal rats (Smith et al. 1988). Moreover, stimulation of the skin of the perineal region (including the base of the tail) has also reported to initiate treadmill locomotion in spinal cats (Pearson and Rossignol 1991). The SCA-induced rhythm described in the present study has several advantages. The induction of the rhythm is obtained in drug-free media, the duration of the rhythmic activity can be set by the experimenter (between 1-2 and 30-60 s), the stimulus trains can be repeated many times in each experiment, the resultant rhythm has a substantial drive, and it is rather reproducible. The rhythm induced by dorsal root stimulation is different from the rhythm induced by bath-applied drugs. First, dorsal root stimulation activates afferent projections to central pattern generator (CPG) neurons (reviewed in Rossignol 1996) as well as excitatory and inhibitory afferent projections to motoneurons (e.g., Pinco and Lev-Tov 1993a for the neonatal rat; review by Baldissera et al. 1981). Therefore the rhythm is superimposed on tonic excitation and/or inhibition of motoneurons that is capable of perturbing the rhythm during high-intensity or -frequency stimulus trains. Second, the rhythmic drive produced by dorsal root stimulation is not constant. It increases at the beginning of the stimulation, reaching a plateau-like level at a later stage, and decreasing gradually after the trains. Therefore the cycle time and burst duration of the rhythm varies during parts of the stimulus trains and following these trains. Third, long stimulus train of dorsal root afferents is capable of producing regular rhythmic activity for tens of seconds (30-60 s). During longer stimulus trains, the rhythmic drive decreases, the cycle time increases, and the rhythm is finally blocked. Because normal rhythmicity could be restored at that stage by stimulation of contralateral afferents, it is suggested that the decline in the rhythmic drive and its block reflect an inability of the stimulated afferents to sustain a steady rhythmic drive. This inability may result from a prolonged synaptic depression similar to the depression described for neonatal afferent pathways during high- and low-frequency stimulus trains (Lev-Tov and Pinco 1992; Pinco and Lev-Tov 1993a,b; Seebach and Mendell 1996). Failure of action potentials to invade afferent terminals, branch point blockade of action potentials, and depletion of neurotransmitter stores or inactivation of presynaptic release sites (reviewed in Lev-Tov 1995, 1998) are only few of the possible mechanisms that should be considered in this regard.

Finally, it is possible to modulate the rhythm within given ranges of intensity, frequency and duration of sacral dorsal root stimulation. In this way, a 2.5-fold increase in intensity and a 10-fold increase in frequency of afferent stimulation were capable of speeding the rhythm by 30-40% and of changing the firing rate during each cycle. This dynamic range is rather narrow comparing to that observed for the neurochemically induced locomotor rhythm at different NMDA concentrations (Cazalets et al. 1992; Kudo and Yamada 1987; Smith et al. 1988).

In summary, stimulation of sacral dorsal roots is an effective and reliable mean to induce drug-free rhythmicity in the sacrocaudal cord of the neonatal rat. The SCA-induced rhythm can be modulated by controlled variations in the stimulation parameters, and its dynamic range is narrower than that of the neurochemically induced locomotor rhythm.

Movements induced by SCA stimulation and the muscles involved in their generation

What are the movements produced by SCA stimulation? How are these movements produced? The sacrocaudal rat spinal cord controls the movements executed by the tail muscles (Brink and Pfaff 1980; Grossman et al. 1982; Masson et al. 1991). Our studies revealed that SCA stimulation produced ventroflexion of the tail accompanied by low-amplitude rhythmic abduction during stimulus trains and by high-amplitude rhythmic abduction after the trains.

We showed that the ventroflexion of the tail during the stimulus trains was produced by left-right alternating activation of the flexors. The action of extensors and the abductors was suppressed under these conditions by a presumed tonic inhibition (see following text). The ventroflexion, however, was maintained with a slightly reduced angle after the train when the extensors and abductors were coactivated with the flexors at a given side of the tail. This latter finding indicates a strong bias of the tail flexors on SCA stimulation. This flexor bias may be ascribed to the absence of supraspinal control in the isolated spinal cord preparation. Interestingly, flexor spasms and prolonged tonic activity of tail flexors that could be augmented by cutaneous stimulation have also been described at different stages following spinalization of rats at the S₂ level (Bennett et al. 1999).

What is the basis for the rhythmic abduction induced by SCA stimulation? The high-amplitude abductions are accounted for mainly by alternating left-right activation of the abductors. Our studies revealed, however, that the coactivated flexors and extensors could also contribute to the rhythmic abduction (Fig. 7C) and that during the trains when the abductors and extensors are suppressed, the alternating activity of the flexors was sufficient to produce low-amplitude rhythmic abduction of the mid-tail (Fig. 6).

These findings imply that the alternating left-right efferent bursts observed during high-intensity stimulus train are generated mainly by activation of flexors, while those produced during low-intensity trains or after stimulus trains involve coactivation of flexor, extensor, and abductor motoneurons within a given hemicord, and alternating activation of the left and right pools of these motoneurons. The relevance of these patterns is discussed in the following text.

Rhythmogenesis and motor output in the mammalian spinal cord

What can we learn about the sacrocaudal circuitry from the activity pattern of the tail muscles? Locomotion (reviewed by Rossignol 1996); FRA-induced rhythmicity (Jankowska et al. 1967a,b; Lundberg 1979, also see: Baldissera et al. 1981; Hultborn et al. 1998) paw shaking (Pearson and Rossignol 1991), and fictive scratching in the cat (reviewed in Gelfand et al. 1988), and turtle (reviewed in Stein et al. 1998) are characterized by flexor-extensor alternations within a limb during the rhythm. In contrast, as mentioned in the preceding text, the rhythmic pattern of the sacrocaudal network exhibits mainly an alternating left-right activation of the tail muscles and coactivation of the three principal muscles on each side of the tail during each cycle. This pattern might imply that there is crossed inhibition between the two sides of the sacral cord during the rhythm and that there is no reciprocal inhibition between ipsilateral flexors and extensor/abductor motoneurons during the rhythm.

The notion of crossed inhibition between the sacral hemicords during the rhythm is supported by our intracellular studies of sacral motoneurons during the rhythm (Lev-Tov et al. 2000b). The issue of reciprocal inhibition between CPGs at a given side of the sacrocaudal cord is more complicated. Inhibitory postsynaptic potentials (IPSPs) could be produced in sacral motoneurons by either ipsi- or contralateral stimulation of SCA (Lev-Tov et al. 2000b). In the present study, stimulus trains of the ipsi or contralateral SCA produced partial or complete suppression of the rhythmic bursts of extensors and abductors. The rhythmic activity of flexors was not perturbed under these conditions. The degree of extensor/abductor suppression increased with the intensity of afferent stimulation. Therefore it is suggested that stimulation of SCA, activated strong reciprocal and crossed inhibitory projections to tail extensor/abductor motoneurons and thereby produced tonic inhibition of these motoneurons. This tonic inhibition is alleviated after the stimulus trains. Further clarification of the extent of reciprocal inhibitory projections from SCA to flexor and extensor/abductor motoneuron pools, their tonic and possible phasic activation during the rhythm, must await additional intracellular recordings from identified flexor and extensor/abductor motoneurons. Functional identification of motoneurons in theses recordings can be now based on the differential suppression of abductor/extensor motoneurons during stimulus trains (e.g., Figs. 6 and 7).

In summary, the present work provided basic understanding of the organization of the recently described sacrocaudal rhythmogenic network and the behavior produced by it. The simplicity of this network and its accessibility makes it a potential model for studies of neural control of automatic movements in the isolated spinal cord.