INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The spinal cord contains a neural circuitry capable of generating rhythmic locomotor pattern in the absence of any peripheral or descending inputs (Grillner 1981). Such neural circuitry, commonly known as central pattern generators (CPG), has been studied in detail in lower vertebrates such as the lamprey (Brodin and Grillner 1986a,b). It comprises a few components including excitatory interneurons and mutually inhibitory interneurons. While N-methyl-D-aspartate (NMDA) is primarily responsible for mediating the excitatory component of the CPG (Brodin et al. 1985; Buchanan and Grillner 1987), glycine is found to be primarily responsible for mediating the reciprocal inhibition.

In normal cats, the CPG has been shown to be "activated" by descending inputs (such as the reticulospinal tract) and acts as a command center to generate goal-oriented and anticipatory adaptive changes to locomotion (for review, see Rossignol 1996). In lamprey, it has been shown that the reticulospinal tract that "drives" the CPG activates NMDA and non-NMDA receptors (Buchanan et al. 1987). Peripheral afferents that also utilize excitatory amino acids (EAA) as neurotransmitter also interact with the CPG to adjust the locomotor pattern for corrective changes. Thus the CPG is in constant interaction with both the sensory and supraspinal input to adapt the locomotor pattern to changes in the environment.

Exogenous applications of pharmacological agents that mimic the neurotransmitter of the descending pathways and act on receptors of intraspinal neurons have also been shown in different preparations to be effective in either triggering and/or modulating locomotion. EAA has been reported to induce locomotion in different in vitro preparations mainly the lamprey (Brodin and Grillner 1985a,b; Cohen and Wallen 1980; Grillner et al. 1981; Poon 1980). The receptor subtypes for mediating the action of the EAA were found to be NMDA and non-NMDA receptor, kainate but not quisqualate. It appears, however, that NMDA and kainate plays a different role in locomotion. For example, in reduced preparations in which locomotor activity was recorded from motor nerves, activation of NMDA receptor induced a locomotor pattern (fictive locomotion) which resembles the electromyographic (EMG) pattern during actual swimming in intact lamprey (Brodin and Grillner 1985a) and tadpole (Dale and Roberts 1984). Activation of kainate receptors elicited a much faster swimming in both lamprey and tadpole, and the induced swimming activity was also less regular than NMDA-induced swimming (for review, Grillner 1986).

In an isolated brain stem-spinal cord preparation of neonatal rats, dopamine alone, EAA (L-aspartic or L-glutamic acid), or NMDA was effective in evoking locomotor activity recorded as rhythmic ventral root discharges (Smith and Feldman 1987). NMDA receptors were most effective in inducing locomotor activity; other non-NMDA receptors (kainate and quisqualate) were weakly effective or ineffective (Smith et al. 1988). In another spinal cord-hindlimb neonatal rat preparation, bath application of N-methyl-D,L-aspartate (NMA) elicited alternating rhythmic activity in extensor and flexor muscles that was completely blocked by NMDA receptor antagonist, 2-amino-5-phosphonovaleric acid (AP5) (Kudo and Yamada 1987). NMDA alone or together with serotonin was also effective in producing a robust locomotor rhythm in neonatal rats (Cazalets et al. 1992; Cowley and Schmidt 1994; Kiehn et al. 1996). Recent study shows that even in the absence of action potential, by removing extracellular Ca²⁺ or adding tetrodoxin (TTX) to the perfusion medium, a combined NMDA and 5HT was still effective in triggering stable locomotor activity measured by ventral root activity in neonatal spinal cord (Tresch and Kiehn 2000).

In Xenopus embryos, bath application of NMDA or kainate caused a sustained motor output pattern similar to swimming evoked by natural stimulation of the intact animals (Dale and Roberts 1984). Recent studies show that excitatory premotor interneurons play a major role in determining the frequency of the swimimng CPG through glutamatergic synaptic drive (Sillar and Roberts 1993; Zhao and Roberts 1998). In in vitro spinal cord of adult spinal frogs, only NMA was effective in eliciting locomotor activity (McClellan and Farel 1985). In decerebrate and spinal rabbits, systemic injection of MK-801, an NMDA-channel blocker, dose-dependently blocked the spontaneous and L-DOPA-induced fictive locomotion (Fenaux et al. 1991). In a brain stem-spinal cord preparation from an adult urodele, fictive rhythmic motor patterns were evoked by bath application of NMDA with D-serine (Delvolve et al. 1999).

In acute in vivo preparation such as the decerebrate cats, NMDA was also found to produce a well-coordinated fictive locomotor pattern (Douglas et al. 1993). In these decerebrate cats, intrathecal administration of NMDA was found to elicit hindlimb fictive locomotion similar to that evoked by the mesencephalic locomotor region (MLR). Kainate and quisqualate were ineffective in producing fictive locomotion in these decerebrate cats (Douglas et al. 1993). In acutely spinalized marmoset monkey, clonidine evoked alternating fictive locomotor activity in all monkeys, whereas NMDA or NMA were found to evoke rhythmic alternating nerve activity in 7 of 10 monkeys (Fedirchuk et al. 1998).

In awake chronic preparations, the ability of EAA to initiate locomotion is not known. Previous work on spinal cats have mainly dealt with the role of monoaminergic drugs (Barbeau and Rossignol 1991; Chau et al. 1998a,b). Noradrenergic drugs (L-DOPA and clonidine), but not serotonergic or dopaminergic drugs, were found to trigger treadmill locomotion in chronic spinal cats during the first week after spinalization (Barbeau and Rossignol 1991). Typically, at 2-3 days postspinalization, before any drug injection, the spinal cat has very little hindlimb movement when placed on a moving treadmill belt. In these preparation, injection of noradrenergic drugs triggered well-coordinated locomotion with weight support and plantar paw placement in cat (Barbeau and Rossignol 1991; Barbeau et al. 1987; Chau et al. 1998a,b). This marked locomotor effect was seen within minutes when clonidine (an 2 agonist) was given intrathecally in a 3d-spinal cat (Chau et al. 1998b).

To date, NMDA receptors appears to be the primary EAA receptors necessary for locomotion. There is, however, no report on the effect of NMDA in triggering locomotion in awake spinal cats. The present study was undertaken to examine the ability of EAA to trigger locomotion in adult chronic spinal cat shortly after spinalization (within 10-12 days). Another paper will deal with the modulation of locomotion in the intact cat as well as the chronic spinal cat by the same agonist and antagonists of NMDA receptors (N. Giroux, C. Chau, H. Barbeau, T. A. Reader, and S. Rossignol, unpublished data). NMDA was found to have limited effects on locomotion in both intact and long-term spinal cats. AP5, an NMDA receptor antagonist, somewhat disturbed locomotion in intact cats by reducing weight support but altogether abolished locomotion in spinal cats, suggesting the importance of glutamatergic system in mediating locomotion in chronic cats. Some parts of this work have been published in abstract form (Chau et al. 1994).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Five adult cats, previously trained to walk on a treadmill were used for this study. We examined the effects of NMDA in these cats during the early postspinalization period (shortly after spinalization where there was no locomotion, usually within 6 days postspinalization) and intermediate postspinalization period (period where there were few rudimentary steps with inconsistent foot placement, usually around 7-8 days postspinalization). Thus in the intermediate period, there were only limited locomotor movements such as the attempts of foot placement, and an increase in the ability to generate consecutive rhythmic steps even though the steps were very small compared with normal. During the experiments, treadmill locomotion and responses to cutaneous stimulation were recorded before drug injection and at different times intervals after each intrathecal drug injection.

The effect of NMDA was tested in four cats (CC6, CC7, NG3, and NG5) during the early-postspinalization period and was tested in three cats (NG2, NG3, and NG5) during the intermediate-postspinalization period. All cats were chronically implanted with EMG electrodes and an intrathecal cannula. In four cats (CC6, CC7, NG2, and NG5), nerve cuff electrodes were also chronically implanted on the superficial peroneal nerve for reflex testing. The pre- and postdrug trials were carried out on the same day and could be compared with normal treadmill locomotion of the same cat (a within subject design where each animal has its own baseline for comparison).

Surgeries

A detailed description of the surgeries has been made elsewhere (Chau et al. 1998a,b). Briefly, all surgeries were carried out in aseptic conditions, and all procedures followed a protocol approved by the Ethics Committee of Université de Montréal.

INTRATHECAL CATHETERIZATION. The intrathecal cannulation technique (Chau et al. 1998b). was adapted from the procedure of Espey and Downie (1995), A midline incision from the cranium to C_2-3 level was made and a Teflon tubing (24LW) was inserted through the cisterna magna down to approximately L₄-L₅ spinous processes. The other end of the Teflon tube, previously connected to a cannula connector pedestal (Plastic One) was secured on the skull with acrylic cement as a port of entry. The location of the cannula tip on postmortem examination are as follows: L₅ (right side, ventrolateral) for CC6; L_4-5 (right side, ventrolateral) for CC7; L_3-4 (central, dorsolateral)for NG2; L₂ (dorsolateral) for NG3; and L₄ (ventrolateral to dorsal rootlets) for NG5.

IMPLANTATION OF EMG ELECTRODES. Four cats were implanted with two 15-pin-head connectors (TRW Electronic Components Group) while one cat was implanted with only one connector secured to the cranium using acrylic cement. A pair of stainless steel wires were sewn into the belly of selected muscles with the Teflon coating removed at the point of insertion. One unpaired wire was placed deeply under the skin of the neck and served as an electrical ground. Bilaterally implanted muscles include iliopsoas (Ip), a hip flexor; sartorius anterior (Srt), a hip flexor and knee extensor; semitendinosus (St), a knee flexor and hip extensor; vastus lateralis (VL), a knee extensor; gastrocnemius lateralis (GL), an ankle extensor and knee flexor; tibialis anterior (TA), an ankle flexor; and gastrocnemius medialis (GM), mainly an ankle extensor.

IMPLANTATION OF NERVE CUFF ELECTRODES. In four cats (CC6, CC7, NG2, and NG5), custom-made bipolar cuff electrodes made from polymer (Julien and Rossignol 1982) (~1 cm length) were used to stimulate the superficial peroneal nerve (~6 mm between electrodes leads) through a 2-pin-head connector.

SPINALIZATION. A laminectomy was performed at the Th13 vertebra. At the level of transection, local anesthetics [lidocaine hydrochloride (Xylocaine), 2%] was placed on the spinal cord (a few drops) and injected into the spinal cord directly (~200 µl injections on each side). The intrathecal cannula was identified, and care was taken to avoid any damage during the complete severance of the spinal cord with a micro scissor. Once the bottom of the vertebral spinal canal could be clearly visualized, an absorbable hemostat (Surgicel, oxidized regenerated cellulose) was used to fill the space between the rostral and caudal ends of the spinal cord. The completeness of the spinal transection was later confirmed with histological analysis (10-µm serial sections using the Kluver-Barrera method).

Animal cares

Animals were placed in an incubator until they regained consciousness after surgery. They were then housed in spacious individual cages (104 × 76 × 94 cm) with food and water. In the first postop day, torbugesic (Butorphanol tartrate, 0.05 mg/kg sc, every 6 h) was given. After spinalization, cats were placed in cages lined with foam mattresses and were attended to a few times daily to check and clean the head connectors, to flush the intrathecal cannula with sterile saline, to express the bladder manually, and to generally inspect and clean the hindquarters.

Recording procedures and protocol

LOCOMOTION. All cats were placed on the treadmill and locomotion was recorded a few days after the intrathecal catheterization and the implantation of EMG electrodes and/or nerve cuff electrodes. Locomotion at different speeds was recorded while the cats walked freely on the treadmill belt. This served as the baseline controls (the intact trials) for each cat and was recorded for 6-7 days for CC6 and CC7 and for 7, 11, and 4 days for NG2, NG3, and NG5, respectively.

ELECTRICAL STIMULATION. Single pulse of 250 µs duration was delivered (Grass S88 stimulator) at 0.4-0.5 Hz through the cuff electrodes. The stimulation was given either at rest, standing, or sitting. The stimulus signal was displayed on an oscilloscope together with selected EMGs. The threshold (T) of the stimulation was determined by observing a just detectable response in St at rest. The EMG responses to the electrical stimulation was also digitized at 1 kHz, and computer averaged.

Drug applications

Three drugs were used in these experiments. They are NMDA, an EAA receptor agonist; dihydrokanic acid (DHK), an EAA uptake blocker, and clonidine (2,6,-dichloro-N-2 imidazolidinylid-enebenzenamine), 2-noradrenergic agonist (all drugs are from Sigma). All drugs were dissolved in sterile saline solution and were injected as a bolus into the spinal cord through the intrathecal cannula. All bolus injections were of 100 µl per dose followed by a subsequent bolus injection of saline (approximately 100-150 µl) to fill the dead space of the cannula and ensure the expulsion of the drug into the intrathecal space of the spinal cord. NMDA (0.1-10 mM it) were given to CC6 and CC7 in multiple injections in attempt to trigger locomotion in the early postspinalization phase. NMDA (1 mM it) was given to NG2, NG3, and NG5 at different days postspinalization. NMDA was injected at 3 days for CC6 and at 3, 4, 5, and 6 days for CC7. DHK (5 mM it) was also given to CC7. NMDA was injected at 7 and 11 days for NG2; at 6 and 8 days for NG3; and at 5, 6, 7, and 8 days for NG5. The upper limit of liquid volume given in one session was ~800 µl over 4-5 h. It should be noted that the dose range that we injected in the spinal cat was much smaller than that reported to cause spinal cord damage in chick embryo (Llado et al. 1999). Also, the chronic spinal cats were kept for a long time, and we could not observe any obvious deterioration of the locomotor performance of the cat that would have resulted from a mechanical or toxic damage to the cord.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results are from experiments performed in five chronic spinal cats. To test the ability of NMDA in initiating locomotion soon after spinalization, different doses of NMDA (100 µl of 0.1-10 mM) were injected in four cats soon after spinalization (starting at 3 days postspinalization in cats CC6 and CC7, at 6 days in cat NG3, and at 5 days in cat NG5). In three cats, NG2, NG3, and NG5, a fixed dose of NMDA (1 mM it) was also injected at an intermediate stage (~6-7 days postspinalization) to test the effect of NMDA on a preexisting but very rudimentary locomotor pattern. Video records of each trial were carefully reviewed, and representative trials were selected for detailed quantitative kinematic and EMG analysis.

Effects of intrathecal NMDA on locomotion in the early postspinalization phase

NMDA alone or potentiated with DHK did not trigger locomotion in four spinal cats (total of 7 injections) during the early postspinal days (3-6 days). Figure 1A shows a typical example of a 3d-spinal cat (CC7) before NMDA injection. Thirty minutes after the NMDA injection (1 mM it, Fig. 1B), some limited rhythmic movements of the hindlimbs were seen but with no EMG activity from knee extensors. Because there were very little movements from the hip, the hindlimb remained in an extended position. There was therefore neither plantar foot placement nor weight acceptance. After an additional dose of NMDA (5 mM it; results not shown), the already faint rhythmic movements of the hindlimbs actually decreased. This limited effect of NMDA was sharply contrasted by the effects of a noradrenergic agonist, clonidine. Indeed a subsequent clonidine injection (Fig. 1C) elicited a robust and well-coordinated locomotor pattern with alternating flexor and extensor burst activities. The cat was also capable of plantar foot placement, good weight support during stance as can be seen from a marked increase in knee extensor EMG activities (Fig. 1C), and could walk 0.5 m/s. Therefore the inability of the previous NMDA injection in triggering locomotion could not be attributed to the inability of the cat to generate spinal locomotion.

View larger version (37K): [in this window] [in a new window]   Fig. 1. The effects of N-methyl-D-aspartate (NMDA) on the initiation of locomotion in an early spinal cat (3 days, CC7). A: stick diagram (1 step cycle) and raw electromyographic (EMG) traces of hindlimb flexors (Srt, St) and extensor (VL) of the left (L) and right (R) sides during treadmill locomotion before any drug injection. B: locomotion recorded at 30 min after NMDA (1 mM it). C: locomotion recorded at 16 min after clonidine (4 mM it) injection. Clonidine was injected 2 h after the NMDA injection. An organized locomotor motor pattern, with plantar foot placement and weight support of the hindquarters, was observed.

A higher concentration of NMDA (2 consecutive doses of NMDA, 5 mM it each given 30 min apart) given in another 3 days spinal cat (CC6) still did not initiate any organized locomotion but did produce sporadic spontaneous high-frequency rhythmic movements (10 Hz) which lasted for bouts of of 0.5 to 1 s as shown in Fig. 2. It consisted of a high-frequency burst of flexor and extensor activity. The spontaneous high-frequency movements were also recorded in another spinal cat (CC7) where a higher concentration of NMDA was given.

View larger version (37K): [in this window] [in a new window]   Fig. 2. The effects of high concentration of NMDA in a 3 days spinal cat (CC6). A: raw EMG traces of hindlimb muscles during treadmill locomotion (0.2 m/s) before drug injection. No locomotion can be elicited even with strong perineal stimulation. B: recordings taken at 9 min after a 2nd dose of NMDA (5 mM it) given at 30 min after the 1st dose (5 mM it), multiple episodes of spontaneous paw-shake activity () was observed when the cat was at rest (lying on a foam mattress). C: expanded portions of one paw-shake episode indicated by  in B.

In one cat, (CC7) NMDA was capable of initiating rhythmic stepping movements but with very little weight support at 4 days postspinalization as shown in Fig. 3. While there was no foot placement before NMDA (Fig. 3A), 7 min after the first NMDA injection (1 mM), there was a marked increase in stepping with plantar foot placement on the left hindlimb (Fig. 3C). On the right hindlimb (results not shown), only the tip of the toes touches the treadmill belt with no dorsiflexion at the ankle (toe placement). These stepping movements were very different from the locomotor pattern triggered after clonidine injection on the previous day. First, the stepping was only triggered with maximal perineal stimulation and was asymmetrical. While the left limb showed consistent plantar foot placement and alternate EMG burst activities (Fig. 3D), the right limb had inconsistent toe placement, and much weaker EMG activities (Fig. 3B). Second, the locomotor pattern also fatigued easily. Typically, the pattern became disorganized after only a few cycles. Third, there was also very little weight support from the cat and the experimenter had to support most of the weight of the cat, as could be seen in the low activities on extensor muscles, especially the left ankle extensor (LGL). Last, the cat could only cope with treadmill speed 0.2m/s. With an additional dose of NMDA (1 mM) the maximal treadmill speed that the cat could cope with was 0.3 m/s.

View larger version (31K): [in this window] [in a new window]   Fig. 3. NMDA elicited stepping movements in a 4-day spinal cat (CC7). A: raw EMG of hindlimb flexor (St, Srt) and extensor muscles (VL, GL), duty cycles, and stick diagram of a 4-day spinal cat before NMDA injection. B: at 7 min post-NMDA injection, (1 mM it), stepping movements was elicited and are characterized by some asymmetry, little weight support even with proper foot placement on the left side. The cat was also unable to follow treadmill speed beyond 0.2 m/s.

In summary, NMDA injection did not initiate robust locomotion as did the noradrenergic agonist, clonidine. In most cases, the maximal effect after NMDA injection was an increase in some alternating movement of the hindlimbs but without any plantar foot placement nor any weight acceptance during stepping. In one case (CC7), NMDA elicited stepping with foot placements, but these steps were asymmetrical, transient, limited to low treadmill speed, and with very little weight acceptance during stance.

Effects of intrathecal NMDA on locomotion in the intermediate phase postspinalization

While NMDA has been shown repeatedly not to be able, a few days after spinalization, to trigger robust locomotion in spinal cats contrary to a noradrenergic agonist such as clonidine, NMDA significantly ameliorated the locomotor pattern in all three spinal cats (5 injections) that had already regained the ability to generate rudimentary stepping with occasional foot placement but little weight support, usually around 6-8 days postspinalization. Figure 4 shows that before NMDA injection, a 7d-spinal cat (NG2) was only capable of making very small but regular steps (stance length = 43 ± 18 mm, 13.7% of intact values) with a toe contact (not plantar) at the onset of stance as noted by the horizontal bar in the stick diagram (Fig. 4A). The EMG traces show disorganized but occasional burst activity of knee flexors (St) and very weak hip flexor activity (Srt). At 17 min after NMDA injection (1 mM it), significant changes in the locomotor pattern was observed as shown in Fig. 4B. There was a marked increase in EMG activity such as the hip flexor (Srt), knee flexor (St), and ankle extensors (GL and GM). The foot placement also improved as seen in the stick diagram at the beginning of the stance phase (indicated by a shaded bar). The locomotor pattern was robust and regular with marked increase in step length (almost 4 times the pre-NMDA value), step cycle duration, joint angular excursion, and weight support. In addition, the cat was able to walk 0.6-0.7 m/s (see Fig. 5B), whereas it could only cope at 0.2-0.3 m/s before NMDA injection. The effect of NMDA persisted for 24 and 72 h after NMDA injection as shown in Fig. 4, C and D, respectively. Despite some deterioration in the locomotor pattern over time, namely a decrease in the activity of flexor muscle (St, Srt) and especially the extensor muscle (GL, GM), the spinal cat was still able to generate some stepping on the treadmill.

View larger version (32K): [in this window] [in a new window]   Fig. 4. The effects of NMDA on locomotion during the intermediate phase (7 days postspinalization). A: stick diagram (1 step cycle) and averaged EMG of hindlimb flexors (Srt, St) and extensors (GL, GM) synchronized to LSt during treadmill locomotion (0.4 m/s) before any drug injection. A shaded bar was placed under the stick diagram representing the onset of the stance phase; cat NG2. B: locomotion recorded at 17 min after NMDA (1 mM it). Note the marked increase in step cycle duration as well as the marked increase in flexor and extensor muscle activities. C: locomotion at 24 h after NMDA injection. Note the decrease in the muscle activities in RSrt and RVL. D: locomotion at 72 h after NMDA injection, a further deterioration in the muscle activities was seen.

View larger version (31K): [in this window] [in a new window]   Fig. 5. Changes in EMG activities and cycle duration in an intermediate spinal cat (NG2) after NMDA injection (1 mM it). A: histogram of flexor and extensor burst duration during intact state and at different times after NMDA injection. Note that the burst duration of St was increased 2- to 3-fold as compared with during the intact locomotion. B: changes in the averaged ankle extensor burst (LGL) duration and step cycle duration as a function of treadmill speed at 24 h after NMDA injection. The number in parentheses indicate the number of cycles measured.

A further examination of the EMG activities after NMDA injection in the intermediate phase revealed a marked increase in the flexor (St) burst duration as shown in Fig. 5. Because there was little locomotion in the spinal cat before giving the drug, quantitative analysis of the EMG pattern was limited to locomotion after NMDA injection where a clear EMG bursting pattern can be observed. These were then compared with the readings of trials in the same cat in the intact state. Typically, the burst duration of the flexor muscle (St) during locomotion was much shorter than that of the extensor muscle (GL) in normal cats. After spinalization, the locomotor pattern observe after NMDA injection is characterized by a marked increase in the flexor burst duration while the extensor burst increase to a lesser extent, thus the ratio between flexor and extensor muscle activation was changed. The numeric values of the muscle burst duration of the intermediate-spinal cat studied was shown in Table 1. Note that the increase in the flexor (St) burst duration is more marked than that of the extensor (VL, GL).

In addition, at 24 h after NMDA injection, the ankle extensor (GL) burst duration and the step cycle duration were found to decrease as the speed of the treadmill increases, indicating the improved ability of the cat (NG2) to adapt the locomotor pattern to increasing treadmill speeds (0.6-0.7 m/s). This is in contrast to the observation in the same cat before NMDA injection where it can only cope with a maximum treadmill speed of 0.2-0.3 m/s.

Long-term effects of NMDA

The effects of NMDA lasted for a long time ranging from 2-4 days after injection. A detailed evolution of the locomotor performance in 1 spinal cat (NG2) is shown in Fig. 6. The cycle duration was measured as an index of the effects of the drugs as well as an indirect indication of the quality of the locomotion. As shown in the figure, the spinal cat had received NMDA at 7 days and later at 11 days postspinalization (indicated by the arrows). After NMDA injection at 7 days, a marked increase in cycle duration was seen at 17 min after injection () as compared with cycle duration before NMDA injection (). The cycle duration after NMDA injection also approached the values previously obtained in the intact state as represented by the shaded band. Note that even after 24 h after NMDA injection, there was still a marked increase in the cycle duration (771 ± 64 ms) as compared with before NMDA injection (533 ± 80 ms). By 72 h (10 days), the locomotor pattern began to deteriorate and was characterized by a decrease in weight support, less consistent plantar foot placement during stance, and decreased step cycle duration. A second NMDA injection was given to the cat on the following day (11 days). Again, a marked increase in cycle duration was still observed 24 h after the injection (12 days). The effect of NMDA diminished with time as previously observed and by 14 days, the cycle duration was comparable to 11 days, before NMDA injection. The locomotion improved by 17 days, which by then some extent of locomotor recovery is likely to have occurred and probably enhanced from the training received during experimentation.

View larger version (20K): [in this window] [in a new window]   Fig. 6. The prolonged effects of NMDA on locomotion of an intermediate spinal cat NG2. The step cycle duration during treadmill locomotion (0.4 m/s) shown as a function of time postspinalization after NMDA injection () in 1 spinal cat. , cycle duration during spinal locomotion before NMDA injection; , cycle duration after NMDA injection. For example, on 7 days postspinalization, cycle duration before and 17 min after NMDA was shown by  and , respectively. The cycle duration during intact locomotion (means ± SD) was shown by .

The prolonged locomotor effects of NMDA was consistently observed in all three intermediate-spinal cats (NG2, NG3, NG5) as shown in Fig. 7. The step cycle duration recorded at 24 h after NMDA injection remain high as compared with the predrug level. In two of three cats, the postdrug level was significantly higher than the predrug level.

View larger version (24K): [in this window] [in a new window]   Fig. 7. Effects of NMDA in 3 intermediate spinal cats at 24 h after injection. Histogram showing the cycle duration in 3 intermediate spinal cats ranging from 6 to 8 days postspinalization. For each cat, the step cycle duration was obtained during intact state (), pre-NMDA injection (), and at 24 h after NMDA injection (). *, significant changes compared with the pre-NMDA values (P = 0.05).

Latency of effect

While the three intermediate spinal cats markedly improved locomotion after NMDA injection, the time it takes to reach maximal locomotor performance (latency), however, differed among them. In one cat (NG2, 7 days), marked locomotor effects were apparent as early as 17 min after drug delivery; in another cat (NG5, 6 days) while bouts of organized stepping with foot placement and weight support were also observed at 2.5 h post-NMDA injection, these effects were extremely transient. In fact, maximal effects of NMDA were not seen until the following day (24 h after injection) where a robust locomotor pattern (0.6-0.7 m/s) with weight support and foot placement was observed. Similar findings were seen in NG3 (7 days) where some locomotor effects could be observed at 1 h post-NMDA (1 mM it) such as more regular alternating stepping movements and a more organized EMG pattern. However, there was no foot placement during stepping until the following day.

To summarize, our results show that in the early postspinalization period (when there was no locomotor ability), NMDA alone or with DHK at best increased some rhythmic stepping but did not trigger locomotion with foot placement or weight support in chronic spinal cats. In the intermediate postspinalization period, where the spinal cats began to show some locomotor ability (very small steps with occasional foot/toes placements), NMDA significantly enhanced the locomotor pattern with marked increase in step length, cycle duration, regular plantar foot placement and weight support, muscles activities and an organized EMG pattern. While the onset of the aforementioned effects differed among the three spinal cats tested, the duration of these effects were similar and were long-lasting.

Effects of intrathecal NMDA on cutaneous reflex excitability

In the early phase postspinalization, NMDA significantly increased the cutaneous excitability in the spinal cats (n = 2). In one spinal cat (CC7), the increase in cutaneous reflex was also found to be dose-dependent. After the first NMDA injection (1 mM it), the threshold of electrical stimulation decreased from 640 to 540 µA. After a subsequent NMDA injection (5 mM it), while there was no changes in the locomotor pattern, the threshold of electrical stimulation decreased to 375 µA (40% of control). For the same stimulation intensity, the cutaneous reflex response was also bigger after NMDA injection as shown in Fig. 8A. In this 3d-spinal cat (CC7), there was a small LSrt response to electrical stimulation before NMDA injection. After NMDA (1 mM it), with the same stimulating parameters, there was a marked increase in LSrt, LSt, and RVL response.

View larger version (25K): [in this window] [in a new window]   Fig. 8. The cutaneous reflex response during the early and intermediate phase postspinalization. A: averaged responses to electrical stimulation of superficial peroneal nerve while standing in a 3 days spinal cat (CC7) before (13 stimulations) and 30 min after NMDA (1 mM it) injection (25 stimulations). The current of stimulation was 640 µA before and after NMDA injection. B: averaged responses to electrical stimulation of superficial peroneal nerve while lying on the right side in a 11-day spinal cat (NG2) before and 35 min after NMDA (1 mM it) injection. The current of stimulation was 120 µA before and after NMDA injection.

In the intermediate phase postspinalization where NMDA (1 mM it) significantly improved the locomotor pattern, it did not change the cutaneous reflex response (n = 2, 4 injections). For example, the threshold of electrical stimulation at rest (lying on the side) did not change after NMDA injection. With the same stimulating intensity, the cutaneous reflex response was also found to be similar as shown in Fig. 8B. It shows that in a 11-day spinal cat (NG2), the cutaneous reflex response, stimulated at same intensity (120 µA) was comparable before and at 31 min post-NMDA (1 mM it).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Summary of results

First, during the early postspinalization period (ranging from 3-6 days postspinalization), NMDA (1 mM it) did not trigger robust locomotion (with weight support and foot placement) in contrast to the noradrenergic agonist, clonidine. A higher dose (5 mM it) of NMDA triggered spontaneous fast paw shaking without effects on locomotion. Second, in the intermediate phase, ~7-8 days postspinalization, when the cats (n = 3) were capable of initiating some rudimentary stepping, NMDA markedly enhanced the preexisting rudimentary locomotor pattern in all three cats with preferential enhancement of the flexor muscle activity. Third, in contrast to a 2 noradrenergic agonist, which markedly decreased the cutaneous excitability (Chau et al. 1998a,b), NMDA dose-dependently increased the cutaneous excitability. Fourth, NMDA exerted a prolonged effect on locomotor activities (>24 h).

Initiation of locomotion during the early phase postspinalization

Previous work from our laboratories have shown that 2 noradrenergic agonists (clonidine, oxymetazoline, and tizanidine) triggered coordinated locomotion with weight support in chronic spinal cats within the first week postspinalization (Barbeau et al. 1987; Chau et al. 1998a,b). In light of the important role of NMDA in locomotion and the success of NMDA in triggering locomotion in a variety of preparations (see introduction), it is surprising that we were unable to initiate robust locomotion in adult cats early on after spinalization as did the noradrenergic agonists. It is possible that in an awake behaving animal, a bolus injection of NMDA may activate receptors that are not involved in the generation of locomotion, such as pain receptors (Raigorodsky and Urca 1987). The increase in cutaneous excitability and the spontaneous "fast paw shake" (FPS) movements that we observed (with higher dose of NMDA) is also likely the result of excitation of sensory pathways. That is, the NMDA receptors are diffusely distributed in the CNS, the widespread activation of NMDA receptors in awake animal might lead to interference with the expression of locomotion. This is in accordance with recent studies that while NMDA triggered highly regular and long-lasting locomotor rhythm in the isolated turtle spinal cord-hindlimb preparation (in vitro), NMDA was far less effective in the immobilized spinal turtle preparation (in vivo) where it triggered only weak and irregular bursting from hip nerves or no knee extensor discharges (Currie 1999). The relative inefficiency of NMDA in in vivo turtle preparation but not in vitro preparation supports the notion that excessive sensory stimulation (by NMDA) in the in vivo state may have interfered with the expression of locomotor rhythm.

A higher dose of NMDA (10 mM it) did not elicit any locomotor activities but spontaneous hindlimb movements resembling FPS movements (Smith et al. 1980). FPS is characterized by high-frequency muscles activation (alternating flexor and extensor) with short cycle duration (55-110 ms) (Smith et al. 1980). Our findings show that the frequency of the FPS rhythmic motor pattern in chronic spinal cats elicited by NMDA is 8-10 Hz, which is out of the range for locomotion and is similar to the FPS previously recorded in fictive cat preparation (Pearson and Rossignol 1991). In the study by Douglas et al. (1993), they also reported that during fictive locomotion evoked by NMDA, high-frequency rhythmic activity (7.5 Hz) was observed and occasionally interrupted the locomotion.

In addition to paw shake response, EAA have also been reported to produce other types of motor rhythm such as spontaneous wiping, hindlimb kicking, and jumping movements in adult spinal frogs (McClellan and Farel 1985).

Thus it appears that a mere increase in NMDA concentration was not sufficient to initiate locomotion. Indeed, it is known that there is a posttraumatic release of EAA that acts on NMDA receptors contributing to the secondary damage (Haghighi et al. 1996; Young 1988) after acute spinal cord injury, and as such this does not appear to trigger locomotion.

Role of NMDA on mediating cutaneous reflex

An NMDA-antagonist, AP5, when applied to the hindlimb enlargement of the turtle spinal cord, was found to reduce the response amplitude of the flexion reflex (Stein and Schild 1989). This evidence suggests that NMDA receptors on sensory interneuron play a role in the spinal cord processing of cutaneous information.

Similar to our findings, intrathecal NMDA (1 mM) was also found to increase persistent hindlimb flexion (induced by electrical stimulation) in anesthetized intact and spinal rat (3 days), and pretreatment of MK-801 prevented the enhancement of flexion (Moore et al. 1992). The spontaneous FPS we observed after a high dose of NMDA injection (>5 mM it) may be related to an increased excitability of the cutaneous afferent innervating the paw (Smith et al. 1980).

Effects of NMDA in the intermediate phase postspinalization

At ~6-8 days postspinalization, the cat was able to make a few rudimentary steps that may be related to a decrease in endogenous release of EAA or perhaps reflects that some degree of functional neuronal connectivity required for stepping was reestablished. Perhaps once this connectivity is in place, NMDA enhanced the excitation of the CPG through mechanisms implicating motoneurons and interneurons.

The injection of NMDA may also mimick the effect of primary afferent stimulation. Electrophysiological recordings of feline spinal neurons showed that L-glutamate was released by terminals of low-threshold primary afferent on stimulation that acts primarily on quisqualate receptor, and the excitatory interneuron, in response to the stimulation, releases L-aspartate, which acts on the NMDA receptors (Davies and Watkins 1983). Thus L-glutamate and L-aspartate were suggested to be the transmitter involved in mono- and polysynaptic responses, respectively. This notion was supported in another study where recording were made from the spinal dorsal horn neurons in the isolated neonatal rat hemisected spinal cord preparation. It was also found that the excitatory amino acids released at the myelinated primary afferent terminal on stimulation act on kainate/quisqualate receptors (which was responsible for the short-latency response) and that coactivation of NMDA receptors occurs at a slower time course (the long latency response) (Morris 1989). Therefore in our study, the marked locomotor effect of NMDA observed in the intermediate spinal cats may indeed be partly attributed to the NMDA activation of sensory pathway resembling that of peripheral stimulation. This may partly explain the marked increased in the flexor burst activity observed. It may also be related to the perineal stimulation given during locomotion in early or intermediate spinal cats, which often produced a somewhat exaggerated hindlimb flexion as shown in the stick diagrams in Fig. 4B. In fact, differential modulation of EMG burst duration and amplitude of flexors and extensors in chronic spinal cats has already been documented using various drugs (Barbeau and Rossignol 1991).

Intrinsic neuronal mechanisms

NMDA receptors have been shown to play a key role in affecting the intrinsic neuronal properties of the CPG. For example, NMDA receptors have been implicated in the rhythmic bursting activity of spinal neurons during fictive locomotion (Grillner and Wallen 1985; Wallen and Grillner 1987). NMDA-induced intrinsic motoneuronal membrane potential oscillations (TTX-resistant) was also reported in tadpole (Sillar and Simmers 1994) and neonatal rat (Hochman et al. 1994; McClean et al. 1997). These NMDA-induced oscillatory changes in membrane potential may play a role in mediating various rhythmic motor pattern such as locomotion or perhaps the FPS (McClean et al. 1997). NMDA-induced intrinsic membrane properties were also found in interneurons (Grillner and Wallen 1985; Sigvardt et al. 1985; Wallen and Grillner 1987). The voltage-dependent properties of NMDA receptor-ion channel appears to play a key role in the generation of low-frequency, steady burst pattern (Brodin and Grillner 1985a,b; Brodin et al. 1985). NMDA, along with other substance such as serotonin, noradrenaline, has also been found to induce plateau potentials in the motoneurons and interneurons in the spinal cord (see Kiehn et al. 1997).

Plateau potentials are intrinsic properties of the neurons that maintain a prolonged state of depolarization in the absence of synaptic inputs, thus significantly amplifying the level of excitatory output (Kiehn 1991). Plateau potentials have been reported in the motoneurons of spinal cats during fictive locomotion induced by L-DOPA, clonidine, or a serotoninergic precursor, 5-hydroxytryptophan (5-HTP) (Conway et al. 1988; Hounsgaard et al. 1988; Schomburg and Steffens 1996), and interneuron of spinal rats (Kiehn et al. 1996) induced by NMDA and 5-HTP. The main role of the NMDA-induced intrinsic membrane properties (both motoneurons and interneurons) and plateau potential is to shape the duration and enhance the amplitude of the final motoneuronal output (Kiehn 1991) thereby modifying the characteristics of the induced motor pattern (Kiehn et al. 2000). NMDA-induced membrane oscillation may not be needed for rhythm generation as transmitter-induced locomotor activity could still be possible in tadpole, lamprey, and neonatal rats in Mg²⁺-free saline, which abolishes the NMDA-induced membrane oscillations but not the EAA-mediated excitatory postsynaptic potentials (EPSPs) (for review, see Kiehn et al. 1997). It is therefore suggested that the role of NMDA in regulating pattern generation by the spinal CPG may be through amplification and enhancement of the synaptic inputs between the CPG elements or transmission from sensory pathways (Kiehn et al. 1997).

NMDA-mediated prolonged effects

Prolonged effects (24-72 h) on locomotion exerted by a single dose of NMDA was observed during the intermediate phase of locomotion. How could these prolonged effects be explained from other studies? The actions of spinal NMDA receptors was extensively studied in the pain control system where NMDA receptors have also been reported to have prolonged effects as documented from behaviors in animal models. The plasticity of pain control system was proposed to involve mechanism similar to those involved in learning (Svendsen et al. 1998). Both NMDA and AMPA receptors have been shown to contribute to the long-term potentiation (LTP) of the C fibers after tetanic stimulation of sciatic nerve, lasting for 6 h. In another study, while activation of NMDA receptors (superficial infusion) was shown to induce LTP of the spinal C-fiber-evoked potentials in spinalized rats, it failed to induce LTP in intact rats, suggesting a critical role by the tonic descending inhibition.

Potentiation of neuronal NMDA responses by substance P is thought to play a role in nociception and the persistent pain (Rusin et al. 1993). It has also been suggested that NMDA receptor activation releases substance P from nociceptive terminals (Liu et al. 1997). Recent study reports a co-localization of NMDA receptors and substance P receptors in rat spinal cord at all levels (Benoliel et al. 2000).

It has been shown in the lamprey that bath application of tachykinin substance P for 10 min resulted in a long-lasting (>24 h) modulation of the NMDA-evoked locomotor network activity (Parker and Grillner 1998). Tachykinin acts on sensory inputs in the lamprey, modulating excitability of the primary afferent and sensory interneuron (Parker and Grillner 1996). The tachykinin-mediated long-lasting effects (>2 h) was found to involve protein synthesis (Parker and Grillner 1999; Parker et al. 1998).

Thus the long-term effects exerted by NMDA receptors activation in our study may involve the aforementioned mechanisms, such as induction of LTP as seen in pain and reflex studies. Also, given the close interactions between NMDA receptors and neuropeptides such as tachykinin, we cannot exclude the involvement of neuropeptides (e.g., substance P), which may cascade a sequence of cellular changes contributing to the long-term effects on locomotor network activity.