| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Nobel Institute for Neurophysiology, Department of Neuroscience, Karolinska institutet, SE 17177 Stockholm, Sweden
Submitted 8 March 2004; accepted in final form 7 June 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
50%, presumably due to presynaptic and soma-dendritic effects documented previously. It is not required for burst generation, but acts as a powerful modulator. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Kiehn and Butt 2003; Marder and Bucher 2001; Nusbaum and Beenhakker 2002). The lamprey brain stem–spinal cord provides a well-studied model system for vertebrate locomotion (Grillner 2003). In the lamprey locomotor CPG, there are, in addition to the excitatory glutamatergic burst generating circuitry (Cangiano and Grillner 2003), inhibitory glycinergic mechanisms for production of right-left alternation (Buchanan and Grillner 1987; Cohen and Harris-Warrick 1984; Grillner and Wallén 1980). The core of the bilateral CPG is thus a glutamate–glycine network, but in addition, there are several modulatory systems using GABA, dopamine, 5-HT, and peptides for fine tuning of the locomotor network (see Grillner 2003).
This study is concerned with the contribution of the GABAergic system to the activity in the lamprey locomotor CPG. GABAergic neurons acting through GABAB receptors are known to reduce the locomotor burst frequency, whereas the role of GABAA receptors remains unclear (Tegnér et al. 1993; see below). Three types of GABAergic spinal neurons have been identified (Brodin et al. 1990): 1) multipolar cells in the lateral gray matter, 2) small bipolar cells located at the base of the dorsal horn and just under the dorsal columns, and 3) cells with ciliated processes extending into the central canal (Schotland et al. 1996) and with axons extending to the lateral edge of the spinal cord forming a plexus around the dendrites of stretch-sensitive edge cells (Grillner et al. 1984). GABA-immunoreactive fibers are generally ubiquitous within the gray and white matter (Christenson et al. 1991). A GABAergic locomotor-driven modulation of presynaptic terminals of primary afferent fibers and interneuronal axons in the spinal network provides presynaptic inhibition (Alford and Grillner 1991; Alford et al. 1991; El Manira et al. 1996), phase-locked to the locomotor pattern. It is mediated through both GABAA and GABAB receptors. GABAB receptors, additionally, act at the soma-dendritic level of interneurons and motoneurons and modify the intrinsic membrane properties (El Manira and Bussieres 1997; Matsushima et al. 1993; Wikstrom and El Manira 1998).
In a previous study, Tegnér et al. (1993) reported that agonists of GABAA (muscimol and diazepam) and GABAB (baclofen) receptors reduced the frequency of bursting, whereas conversely, antagonists of GABAA (bicuculline methiodide) and GABAB receptors (phaclofen and saclofen) increased the frequency of bursting. GABA uptake blockers (nipecotic acid) and a benzodiazepine receptor agonist induced a pronounced slowing of the locomotor burst rate. These results, taken together, strongly indicate an endogenous role for the GABAergic synaptic transmission during locomotion.
Tegnér et al. (1993) showed that bicuculline methiodide caused an irregular motor pattern in addition to its effects on locomotor frequency. Subsequently, it was shown that bicuculline methiodide, in addition to its effects on GABAA receptors, also reduces the slow afterhyperpolarization (Debarbieux et al. 1998; Johnson and Seutin 1997; Pflieger et al. 2002) due to a direct effect on calcium-dependent K+ channels (KCa). Since a blockade of KCa channels profoundly affects the burst generation itself (El Manira et al. 1994; Hill et al. 1992), the conclusions reached, when using bicuculline methiodide as a GABAA receptor antagonist, were thus invalidated. The main purpose of this study is therefore to elucidate the contribution of GABAA receptors during fictive locomotion using selective GABAA antagonists (gabazine or picrotoxin). Concentrations of picrotoxin or gabazine 100 µM do not affect the afterhyperpolarization (Seutin et al. 1997). Gabazine is a competitive (Chambon et al. 1985) and selective antagonist of the GABAA receptor and has no known affects on the glycine channel (Heaulme et al. 1986). A low affinity site exists to which gabazine binds noncompetitively (Heaulme et al. 1986). Picrotoxin is a noncompetitive antagonist of GABAA receptors, interacting with the chloride channel of the GABAA receptor. On reticulospinal neurons in lamprey, picrotoxin selectively blocked synaptic responses at concentrations <20 µM (Matthews and Wickelgren 1979), and it had little effect on glycine responses in giant interneurons and Müller cells (Homma 1983; Homma and Rovainen 1978; Martin 1978). Above 20 µM, it lowered glycine responses and inhibitory postsynaptic potentials (IPSPs) in Müller cells. Picrotoxin has been reported to act nonselectively at higher doses in many systems and affect glycine receptors (Davidoff and Aprison 1969; Yoon et al. 1998).
We also explored the possible role of GABAC receptors, which have not previously been studied in the context of locomotion. GABAC receptors have a widespread distribution, including spinal cord, retina, and the optic tectum (see Johnston et al. 2003). Like glycine receptors and other members of the ligand-gated chloride channel superfamily, some GABAC receptors exhibit picrotoxin sensitivity (Dibas et al. 2002; Goutman and Calvo 2004). The highly specific antagonist of GABAC receptors, (1,2,5,6-tetrohydropyridine-4-yl)methylphosphosphinic acid (TPMPA) (Chebib and Johnston 1999), however, had no effect on the burst frequency.
We show here that the spinal GABAergic system is active during fictive locomotion and provides a marked depression of the output frequency of the locomotor CPG. This effect is mediated partially via GABAA receptors, since the burst frequency increased significantly up to nearly twofold when selective GABAA antagonists were administered, and partly via GABAB receptors that further enhance these effects.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Adult lampreys (Lampetra fluviatilis) were handled according to Karolinska Institutet's guidelines, and experiments were performed with permission from "Stockholms Norra försöksdjursetiska nämnd" (local ethical committee). The lampreys were held in aquaria at 5°C. Isolated notochord-spinal cord preparations were dissected from lampreys under anesthesia produced by 100–150 mg/l MS-222 (tricaine methane sulfonate). These preparations included pieces of spinal cord (10–25 spinal segments in length, 6 cm) taken from the region caudal to the gills and rostral to the dorsal fin. Pieces of spinal cord were mounted in Sylgard-lined chambers. The bath temperature was maintained between 1.3 and 7.7°C. For any given experiment, variations in temperature were controlled to within 2°C. Preparations were superfused with chilled, oxygenated HEPES-buffered saline consisting of (in mM) 138 NaCl, 2.1 KCl, 1.8 CaCl2, 1.2 MgCl2, 4 glucose, 2 HEPES, and 0.5 L-glutamine [pH adjusted to 7.4 with 1 M NaOH to which N-methyl-D-aspartate (NMDA; 75, 100, or 150 µM) was added to induce locomotion]. Superfusion rates used were 2.0 ml/min for picrotoxin and gabazine, 1.0–1.2 ml/min for TPMPA, and 2.5 ml/min for the experiments in which two concentrations of NMDA were used and the effect of gabazine was tested. In some experiments (2-level NMDA-excitation experiments), recordings were made from two 10-segment pieces simultaneously. GABAA receptor antagonists were added to the superperfusate separately in the different experiments. Final solutions were adjusted to a pH of 7.4.
Drugs
The antagonists include picrotoxin (0.5–100 µM; Sigma Aldrich, St. Louis, MO), gabazine/SR95531 [6-imino-3-(4-methoxyphenyl)-1(6H)-pyridazinebutanoic acid hydrobromide; 10–3–103 µM; Tocris, Bristol, UK], TPMPA [(1,2,5,6-tetrahydropyridin-4-yl)methylphosphinic acid; 0.6–320 µM; Tocris], the GABAB antagonist CGP55845(20 µM; Tocris), and NMDA (50–150 µM; Tocris).
Electrophysiology
Recordings were made from ventral roots at least four segments from the cut edge of the spinal cord. Bursts of action potentials recorded from the ventral roots (fictive locomotion) were recorded using glass extracellular suction electrodes. The amplified signals (Differential AC Amplifier, model 1700, A-M Systems, Carlsborg, WA) were band-pass filtered from 300 to 500 Hz or 100 Hz to 1 kHz, depending on the experiment and digitized at 2–5 kHz (Digidata 1200A and Axoscope 1.0 software, Axon Instruments, Union City, CA).
Experimental protocols and sampling
In all experiments, 3-min records (5 min in some early experiments) of burst frequencies and temperature of the bath were recorded every 5 min. Burst frequencies were measured by hand from 30-s samples of the above records (very 1st 30 s of a 5-min sampling period). The reliability of this burst frequency measurement was double checked by fast Fourier transform analysis and from autocorrelograms in gabazine and picrotoxin concentration-response experiments involving burst quality, alternation quality, and phase lag analyses. Burst frequencies were stable after 2–5 h of exposure to 100 µM NMDA. The five 5-min periods (viz., 25 min) immediately before the first drug administration were selected for the assessment, as described above, of stability of the pretreatment control burst frequency. Of 22 experiments, the SD of the stabilized burst frequencies before the first treatment was below 0.05 Hz (n = 22), except for one gabazine experiment, where the SD was 0.076 Hz. For the concentration-response experiments with picrotoxin, the 25-min treatment periods were interposed with 25-min washes with 100 µM NMDA in HEPES (control) superfusate. In the concentration-response experiments with gabazine, 25-min treatment intervals were interposed with 50-min washes as above. In the case of the latter two experiments, the highest burst frequency from a 30-s record of a 5-min sample period at, or near, the change from drug to washout was generally chosen as representative. In approximately one-half the experiments, the highest burst frequency occurred during the last part of the drug application, whereas in the remainder, it had a lag into the wash period. In 80 different cases of presentations of different drugs at different concentrations, the rhythm broke down six times (3 picrotoxin experiments and 3 experiments involving CGP44845. In these cases, the highest burst frequency during the treatment period was used even if it was not within the last 5 min. Since the TPMPA burst frequencies showed little variation, the average frequencies for the entire application periods were used. In the experiments involving antagonism of GABAB receptors with CGP55845 the treatment periods with either 20 µM CGP55845or 20 µM CGP55845plus 10 µM gabazine were 120, 45, 90, and 60 min long for experiments 1–4, respectively (see Fig. 8), and mean burst frequencies were calculated from the 30-s records from five 5-min sample periods at the end of the treatment period.
|
Analysis of fictive locomotion employed Datapac (Run Technologies, Mission Viejo, CA) for burst proportions, cycle durations, CV of cycle duration, and fast Fourier transform analysis of burst frequency. Origin software (Northhampton, MA) was used for time series analysis (cross- and autocorrelative), determinations of burst and alternation quality, as well as rostro-caudal phase lags, phase of alternation, and burst frequency. Statistical analyses were performed with GraphPad Prism (San Diego, CA).
Burst quality was assessed both by CV of cycle duration and by autocorrelative processing of 3- to 5-min-long ventral root recordings. When correlated against itself, a single ventral root recording yields a maximum value of the autocorrelative correlation coefficient (arbitrary units) when initially there is no lag between the record and its identical copy (see Fig. 4A, inset). Sequential calculation of the correlation coefficient after repeated lag-shifts of the record with its copy yields a function of the value of the correlation coefficient versus the lag, or the autocorrelogram. The height of the first peak, 
, of the correlogram reflects the temporal periodicity, or rhythmicity, of the bursts as well as the combined power of the signals (compound action potentials) making up the burst. Spikes occurring between bursts degrade the distinctness of the burst pattern and result in a diminished peak-to-trough difference ( – 
) in the correlogram (see Fig. 4A, inset). Dividing by ( + ) to adjust for relative differences in signal power, we employ a measure of burst quality (Cangiano and Grillner 2003) defined as
 |
|
Cross-correlative analyses were made of contralateral ventral root recordings from the same spinal level (±1 segment) producing cross-correlograms of the form depicted in the inset of Fig. 5A. In these cases, the correlation function tends to approach a local minimum at zero lag when there is the least overlap between bursts. Ideal alternation would be represented as maxima of the cross-correlogram symmetrically distributed on either side of the lag axis at –0.5 and +0.5. Similar to burst quality, an alternation quality (AQ) can be operationally defined and quantified using the mean, 
, of the two peak values, 1 and 2, of the correlogram to either side of the x = 0 axis. Thus
 |
|
representing the summed delay of bursts in successive segments. Dividing by the sum of both the average cycle period determined from the correlogram, 
, and the number of segments between the recording electrodes, s, yields the intersegmental phase lag (expressed as a percentage). The convention of assigning positive or negative values to phase lag was made consistent with those used previously, whereby bursts occurring later in more caudal segments, corresponding to forward swimming, are described with positive phase lags (Wallén and Williams 1984). Typical delays (converted from correlative lag values) are in the order of 0.5–1% of the cycle period per segment. Our preparations consisted of about 16 segments between ventral root electrodes. The phase lag is then expressed as the per-segment delay (ms) divided by the cycle period (ms) times 100%. Statistics
Mean ± SE is reported after the mean value. Repeated measures ANOVAs were performed for analyses of burst characteristics for all three drugs along with Dunnett's a posteriori test for multiple comparisons with the control value. Tukey's and Newman-Keuls a posteriori tests were employed in comparisons involving GABAB blockade with CGP55845 One- and two-tailed paired t-tests were employed as indicated in the text in comparisons involving treatment combinations with gabazine and/or CGP 55845. Two-tailed, paired t-tests were used for the comparison of all mean values of burst frequency in experiments involving two levels of NMDA with and without gabazine treatment. One- and two-tailed paired t-tests were employed as indicated in the text in comparisons involving treatment conditions with gabazine and/or CGP 55845. Statistical tests were performed on cycle duration (inverse) transformations and reported as frequency values. Mean ± SE is reported as a value after percentages. All bars represent SE in relevant graphs.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
To test the potential effect of GABAA antagonists on the pattern of fictive locomotion, we utilized the isolated spinal cord-notochord preparation superfused with NMDA (see METHODS) and recorded changes in the ventral root burst activity (Fig. 1, A–C). We subsequently applied GABAA antagonists at different concentrations in the NMDA solution. Both gabazine, a competitive antagonist, and picrotoxin, a noncompetitive, chloride channel blocker, were administered.
|
2 h), the burst frequency recovered to the control level. From each experiment (as detailed in METHODS), the burst frequency from near the end of each 25-min treatment period was used to produce the summary histogram in Fig. 1E. This figure summarizes four experiments and shows that a significant increase (Dunnett's a posteriori test, P < 0.05) occurred already with a low level of gabazine (0.01 µM). At concentrations from 0.1 to 100 µM, the increase was highly significant (Dunnett's, P < 0.001), and at the highest concentration, the mean increase in burst frequency reached 90 ± 19% (n = 4). The prominent effects observed with gabazine suggest an important GABAA inhibitory influence in the locomotor network.
PICROTOXIN INCREASES BURST FREQUENCY.
Picrotoxin, a noncompetitive blocker of the chloride channel associated with the GABAA receptor, elicited a marked concentration-dependent increase in burst frequency similar to that of gabazine (Fig. 2, A–D). As for the gabazine experiments described above, the burst frequencies from a 5-min period near the drug change to washout was used to produce the summary histogram in Fig. 2E. If at the highest concentrations, the rhythm deteriorated, the last, and thus in all cases the highest, frequency before deterioration was measured. At 1 µM, the increase in frequency was highly significant (Dunnett's, P < 0.001, n = 4; Fig. 2E). The average increase in burst frequency with 7.5 µM was 100 ± 9% (n = 4). Higher concentrations (7.5–10 µM) resulted in break down of the burst pattern in three of four preparations, with long periods of synchronous activity in the ventral roots, similar to that observed with glycine receptor antagonists (Grillner and Wallén 1980; McPherson et al. 1994) (Fig. 6B, bottom). These effects of picrotoxin are presumably nonspecific and can most likely be explained by an action on glycinergic chloride channels (see INTRODUCTION).
|
|
30 min long, and burst frequencies were calculated, as previously described, from the last period or the highest frequency achieved before drug change. To investigate if the action of the GABAA receptor is dependent on the control burst frequency, locomotor activity was induced by 75 and 150 µM NMDA. The control burst frequencies were, on average, 1.3 ± 0.09 and 1.9 ± 0.12 Hz, respectively (n = 6; Fig. 3). When adding 10 µM gabazine, the average burst frequency increased significantly (26%; P < 0.01) from 1.3 to 1.7 ± 0.07 Hz (n = 6) with the lower concentration of NMDA, and at the higher level, a greater increase from 1.9 to 2.9 ± 0.17 Hz (50%, P < 0.005, n = 6) occurred. The GABAA receptor antagonist thus enhances burst frequency at both levels of locomotor activity, but the effect is greater at higher burst frequencies (see DISCUSSION).
|
Not only is the frequency of ventral root bursting important to consider but also the overall pattern of activity. This can be estimated by an index for burst quality (see METHODS; Buchanan 1999a, b; Cangiano and Grillner 2003), the proportion of the locomotor cycle taken up by the burst, and the regularity of bursting.
EFFECTS ON BURST QUALITY. The burst quality (Fig. 4A; METHODS) was measured using an autocorrelation of 3 min of rectified and fivefold decimated ventral root burst activity and comparing the height of the first peak and first trough of the correlogram (see Fig. 4A, inset). The sample periods chosen were identical to those used for the burst frequency analysis described above for gabazine and picrotoxin. Burst quality for all preparations was normalized to the control level (equal to a value of 1.0) and averaged to produce the mean values in Fig. 4. Neither gabazine nor picrotoxin (acting on chloride channels directly) produced a measurable effect on burst quality (ANOVA, P >> 0.05). The strong suggestion of a deterioration of rhythm with picrotoxin treatment, however, prompted us to examine the issue further by analyzing the rhythm for alternation quality.
BURST PROPORTION REMAINED GENERALLY CONSTANT.
Burst proportion is the duration of the ventral root burst taken as a fraction of the cycle period. It was measured from the same sample records used to measure burst frequency above. Twenty to 48 cycles were used to determine a mean burst proportion. It generally remains between 25 and 40% of the cycle duration during locomotion (Grillner 1974; Wallén and Williams 1984), regardless of the actual cycle duration. Figure 4B shows variation of the normalized burst proportion at different levels of the two GABAA antagonists used here (left ordinate). No statistically detectable differences in absolute or normalized mean were found with gabazine or picrotoxin (ANOVA, P > 0.05).
Superimposed on the fast burst pattern, a slow modulation sometimes occurred (around 0.1 Hz or below, cf. Aoki et al. 2001) at higher concentrations of GABAA antagonists. The slow modulation tended to show both patterns of right-left alternation and synchrony, as did the fast burst rate of normal fictive locomotion. In this superimposed slow rhythm, the burst proportion could vary outside the 25–40% range.
EFFECTS ON THE CV OF CYCLE DURATION. A second factor affecting the burst quality is the regularity of the burst pattern. The CV of the cycle duration (SD/mean cycle duration) can be used as a measure of the regularity. Like burst proportion, CV of cycle duration was determined from the same 5-min sample periods used to determine burst frequency in a treatment period. Twenty to 48 cycles were used to determine the mean value. In Fig. 4B, the CV of the cycle duration is normalized to the control value and averaged to produce mean CV of cycle durations. These values are plotted at different concentrations of gabazine and picrotoxin (right ordinate). No changes in CV were observed (ANOVA, P > 0.05).
Effects on the quality of segmental coordination
The pattern of right-left alternation in pairs of ventral roots at the same segmental level was analyzed, using cross-correlation of 3-min records, as with burst quality, to arrive at a measure of the average quality of alternation (see METHODS; Fig. 5A, inset). Again, the sample periods were the same as used for determining burst frequency above.
The effects on the AQ for three gabazine and four picrotoxin experiments are shown in Fig. 5A1 along with the methods of analysis (Fig. 5A2). With gabazine, the quality of alternation (Fig. 5A1, bottom) remains near the control level at all concentrations (ANOVA, P = 0.062). The response profile for picrotoxin was different from that of gabazine: picrotoxin significantly depressed alternation quality (P = 0.0018). The significant effects were at 5 and 7.5 µM picrotoxin (Dunnett's, P < 0.001), the latter showing a decline of 58% (Fig. 5A1).
For both gabazine and picrotoxin, the average phase value (Fig. 5A1, Phase) of alternation remained unaltered from its control value of 0.50 (±SE for both drugs <0.01, n = 4) across the various concentrations, ranging from 0.4 to 0.6. Thus neither gabazine nor picrotoxin had any effect on phase of alternation (P >> 0.05).
Effects on intersegmental phase lag
During locomotion, there is an intersegmental rostro-caudal phase lag of around 1% of the cycle duration per segment, regardless of the actual cycle duration (Cohen and Wallén 1980; Grillner 1974; Lansner et al. 1998; Zelenin et al. 2001). This phase lag remains in the isolated spinal cord (Wallén and Williams 1984). As described in METHODS, the cross-correlation of a rostral and an ipsilateral caudal ventral root bursting pattern yielded a correlogram function whose principal peak is shifted from the lag = 0 (ms) axis by the amount of delay between the bursts. When divided by the number of segments, this value yields a per-segment phase delay. The sample periods chosen were identical to those used for the burst and alternation qualities described above, and the method of measurement is shown in Fig. 5B2. In this study, intersegmental phase lag remained essentially unresponsive to GABAergic antagonism (Fig. 5B1) for all picrotoxin concentrations (ANOVA, P = 0.2222) and concentrations of gabazine under 100 µM (Dunnett's, P > 0.05). There was a moderate effect of doubling the average phase lag from 0.35 to 0.7% at the very highest concentration of gabazine (100 µM, Dunnett's, P < 0.001, ANOVA, P = 0.009). However, the general constancy of the phase lag remained for other concentrations of the drugs even though the frequency of bursting increased markedly with both gabazine and picrotoxin.
Effects of very high concentrations of gabazine and picrotoxin on the locomotor activity
With gabazine, even at a very high concentration of 100 µM, the fast rhythmic alternating bursting was retained, as evident from Fig. 6A1–A3 and the cross-correlogram in Fig. 6A4. In contrast, with a high concentration of picrotoxin, the alternating motor pattern broke down and changed to simultaneous bursts on the left and right ventral roots at a very slow rate (0.12 Hz; Fig. 6, B2 and B4). In three of four picrotoxin experiments, high concentrations at either 7.5 or 10 µM produced this pattern of bursting with a rhythm of about 0.1 Hz (Cangiano and Grillner 2003; Cohen and Harris-Warrick 1984). Such slow, synchronous rhythms are characteristic of a glycinergic blockade, which presumably occurs with high doses of picrotoxin, as has been described previously for strychnine (Aoki et al. 2001; McPherson et al. 1994). In contrast, a very weak, slow modulation superimposed on the fast rhythm can sometimes be detected under control conditions (Brodin and Grillner 1985), and it could also be discerned after gabazine (Fig. 6A, middle traces).
GABAC receptors are not involved in the regulation of burst frequency
To explore whether GABAC receptors become activated during swimming, a selective GABAC antagonist, TPMPA, was administered in concentrations ranging from 0.6 to 320 µM, which encompasses the 10–50 µM used in rat brain slice and spinal cord (Kirischuk et al. 2003; Rozzo et al. 1999). The concentrations of TPMPA (Fig. 7) were presented in successive, 30-min treatment periods without intervening washouts. Since the burst frequencies obtained on exposure to TPMPA showed no clear change, the frequencies measured from each of the five periods of a treatment period were averaged to produce a single mean frequency. Summary of the burst frequencies averaged over the whole period are shown in Fig. 7D. No effect on burst frequency was observed in four preparations, even at the highest level of TPMPA (P > 0.05). This finding is important also in relation to the effects of picrotoxin, which has been reported to act nonspecifically on GABAC receptors (Wang et al. 1995). The results thus suggest that GABAC receptors do not contribute to the spinal pattern generation and that the divergent effects of picrotoxin cannot be accounted for by an action on this receptor subtype.
|
Tegnér et al. (1993) showed that administration of GABAB antagonists resulted in a marked increase of burst frequency. It was now important to establish whether the effects of GABAA antagonists documented here would add further to the GABAB-induced effects. Using the protocol described in METHODS, control (100 µM NMDA + HEPES) was superfused before, between, and after the applications of 20 µM CGP55845 a GABAB antagonist, and/or 10 µM gabazine (Fig. 8B). Furthermore, HEPES-only washes were superfused for 0.5 h after either 20 µM CGP55845alone or 20 µM CGP55845+ 10 µM gabazine. These HEPES-only washes were followed by control NMDA/HEPES with burst frequencies very close to the control NMDA values (mean change compared with pretreatment control = 1.8%, SD = 1.8%, n = 8). Figure 8, A1–A4, shows records of bursting during the treatment. Administration of 20 µM CGP55845caused an acceleration of the locomotor burst pattern from 1.8 ± 0.2 to 2.4 ± 0.3 Hz (n = 4, P < 0.01, Tukey's and Newman-Keuls; Fig. 8B), in confirmation of Tegnér et al. (1993). Adding 10 µM gabazine plus 20 µM CGP 55845 resulted in an additional increase to 2.8 ± 0.2 Hz (n = 4, P < 0.05, Newman-Keuls; Fig. 8B). With regard to the locomotor system, the GABAA and GABAB receptor subtypes thus act in a synergistic fashion, which could not be assumed a priori, since the two receptor types could be expressed in different cellular compartments. In one of four experiments, the locomotor rhythm became very irregular with the addition of 20 µM CGP55845 and in two of four experiments, the same effect occurred with the application of 10 µM gabazine plus 20 µM CGP55845 A regular burst activity was resumed after washout.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
; Tegnér et al. 1993) had been compromised by the later finding that this GABAA antagonist also affected KCa channels (Debarbieux et al. 1998; Johnson and Seutin 1997; Pflieger and Dubuc 2000), which importantly contribute to the locomotor pattern generation. This study shows that gabazine, so far known to be a selective antagonist, causes a pronounced concentration-dependent increase of burst frequency up to a twofold increase (Fig. 1E), without a concomitant change in burst quality, burst proportion, or CV (Fig. 4) and essentially no effect on intersegmental coordination (Fig. 5B). The gabazine blockade can thus, in contrast to bicuculline, selectively affect burst rate without changes in the regularity of locomotor activity.
Picrotoxin, the other GABAA antagonist tested, is known to affect glycine-activated Cl– channels at higher doses. With picrotoxin, we could confirm that the burst rate increased, but with higher doses, the alternation quality decreased progressively, and at the highest dose, the pattern converted to slow, synchronous, bilateral bursts (Fig. 6, B2 and B4), which is similar to the effects of the glycine receptor antagonist strychnine (Aoki et al. 2001; McPherson et al. 1994). Our data suggest that, with picrotoxin, a progressive concentration-dependent blockade of glycine receptors occurs, as in other systems (Pflieger et al. 2002).
GABAC receptors activate a related Cl– channel with different and slow kinetics. This receptor subtype is expressed in a variety of tissues including the rodent retina, hippocampus, and spinal cord (see Johnston et al. 2003). There is evidence that it contributes to the spinal network in neonatal rats (Rozzo et al. 1999). To investigate if GABAC receptors are involved in the control of the locomotor network, we administered a selective antagonist, TPMPA. It did not, however, exert any effect, even at a very high dosage. Thus this receptor subtype does not appear to affect the locomotor network, provided of course that the antagonist is effective in the lamprey CNS. Furthermore, with interference of picrotoxin on GABAC receptors ruled out, our results support evidence presented above that, at high concentrations, picrotoxin exerts effects on glycine receptors that contribute to the deterioration of rhythmic coordination. Having clarified this problem with picrotoxin, we base our conclusions preferentially on the results obtained with gabazine.
Contribution of the GABAA component of the GABA system
The endogenous GABAA receptor activation thus contributes to a prominent slowing of the locomotor burst rate, both at low and high control burst rates (NMDA drive), but the effect is more prominent at higher burst rates (Fig. 3). Conversely, a potentiation of the endogenous GABAA activation through activation of the benzodiazepine receptor with diazepam led to a further substantial slowing of the burst rate (20%) with maintained regular activity (Tegnér et al. 1993). The possible working range of the GABAA component is thus substantial.
Endogenous GABA action via both GABAA and GABAB receptors
A potentiation of the endogenous action of the GABA system can be produced by administration of GABA uptake blockers (nipecotic acid), which results in a pronounced overall slowing of the burst rate by around 50% (Tegnér et al. 1993). This effect is partially reversed by GABAB antagonists, with the remaining part being due to GABAA receptors. In these experiments, we show that both GABAB and GABAA receptors contribute to the endogenous slowing of the locomotor burst activity (Fig. 8).
Mode of action of the GABAergic system
From previous studies (Alford et al. 1991), it is known that GABA interneurons are active in phase with the ipsilateral burst activity, as inferred from the phasic GABAergic presynaptic depolarization that occurs in the terminals of network interneurons. There are two types of GABAergic interneurons that are prime candidates for the effects observed here: the large multipolar subtype in the gray matter and the small dorsal bipolar subtype (Brodin et al. 1990). The former would appear the most likely, since it is large with numerous ramifications in the gray matter. The small bipolar subtype projects to the terminals of sensory afferents and co-expresses neuropeptide Y (NPY) (Bongianni et al. 1990; Parker et al. 1998). This subtype is most likely responsible for mediating presynaptic phasic inhibition to sensory afferents (El Manira et al. 1997), but probably not to network interneurons, since they appear to have limited axonal ramifications.
The GABA system clearly exerts part of its burst rate effects through the presynaptic modulation via GABAA and GABAB receptors on network interneurons, which will decrease the excitatory drive. However, most likely, there are also direct effects on the soma-dendritic level of network neurons. Administration of the GABAA agonist muscimol can stop network activity altogether, while the GABAB agonist baclofen slows the locomotor network activity profoundly (Grillner and Wallén 1980; Tegnér et al. 1993). Part of this latter action is through a depression of low- and high-voltage–activated Ca2+ channels and a concomitant decrease of the activation of KCa channels, and thereby the afterhyperpolarization, and in part via reduction of Ca2+-dependent postinhibitory rebound (Matsushima et al. 1993; Tegnér and Grillner 2000; Tegnér et al. 1998). These presumed postsynaptic effects would be expected to slow network activity by reducing the frequency adaptation, due to the afterhyperpolarization, thus prolonging burst periods, and also by reducing the postinhibitory rebound excitation, following each inhibitory phase, leading to less activation of action potentials.
What is the role of GABAergic modulation of the locomotor system?
The locomotor system can, on one hand, generate locomotor activity when the GABA system is largely blocked, but the activity is then at a higher burst rate. The GABA system is thus not required for burst generation to occur, but it will extend the working range of the network in the lower end of the frequency span, without affecting the intersegmental coordination. The GABA action is larger at higher rates of locomotor activity (Fig. 3). Whether the degree of GABAergic modulation is entirely intrinsic and driven by the spinal network or is subject to additional external modulation from the brain stem is so far unknown. The latter possibility could provide an additional way to modulate burst frequency without affecting intersegmental coordination. A similar GABA modulation of the locomotor network appears to occur in both mammals and amphibians and may thus be a general vertebrate trait (Cazalets et al. 1998; Sillar et al. 2002).
It is interesting to compare this system with the spinal 5-HT system, another powerful intrinsic modulator, which is also activated from the locomotor network and contributes to a slowing of the locomotor burst rate due to an indirect action on KCa channels and via its effects on synaptic transmission (Harris-Warrick and Cohen 1985; Hill et al. 2003; Kozlov et al. 2001; Wallén et al. 1989; Zhang and Grillner 2000; but see Svensson et al. 2003). 5-HT, in lamprey and in other vertebrate locomotor networks, also promotes the occurrence of steady regular locomotor activity. In contrast to the GABA system, the 5-HT system generates an increased intersegmental phase lag with increasing levels of 5-HT (Matsushima and Grillner 1992), which would result, not only in a lower locomotor frequency, but also in more than one undulatory locomotor wave along the body.
In conclusion, it is important to realize that these two modulator systems are both turned on when the locomotor network becomes active, both contribute to a slowing of the burst pattern, and both modify the network properties to generate a more stable locomotor rhythm.
|
GRANTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: S. Grillner, Nobel Inst. for Neurophysiology, Dept. of Neuroscience, Karolinska Institutet, SE 17177 Stockholm, Sweden (E-mail: Sten.Grillner{at}neuro.ki.se).
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONGRANTSACKNOWLEDGMENTSREFERENCES |
|---|
Alford S and Grillner S. The involvement of GABAB receptors and coupled G-proteins in spinal GABAergic presynaptic inhibition. J Neurosci 11: 3718–3726, 1991.[Abstract]
Aoki F, Wannier T, and Grillner S. Slow dorsal-ventral rhythm generator in the lamprey spinal cord. J Neurophysiol 85: 211–218, 2001.
Bongianni F, Christenson J, Hokfelt T, and Grillner S. Neuropeptide Y-immunoreactive spinal neurons make close appositions on axons of primary sensory afferents. Brain Res 523: 337–341, 1990.[CrossRef][ISI][Medline]
Brodin L, Dale N, Christenson J, Storm-Mathisen J, Hokfelt T, and Grillner S. Three types of GABA-immunoreactive cells in the lamprey spinal cord. Brain Res 508: 172–175, 1990.[CrossRef][ISI][Medline]
Brodin L and Grillner S. The role of putative excitatory amino acid neurotransmitters in the initiation of locomotion in the lamprey spinal cord. II. The effects of amino acid uptake inhibitors. Brain Res 360: 149–158, 1985.[CrossRef][ISI][Medline]
Buchanan JT. Commissural interneurons in rhythm generation and intersegmental coupling in the lamprey spinal cord. J Neurophysiol 81: 2037–2045, 1999a.
Buchanan JT. The roles of spinal interneurons and motoneurons in the lamprey locomotor network. Prog Brain Res 123: 311–321, 1999b.[ISI][Medline]
Buchanan JT and Grillner S. Newly identified ‘glutamate interneurons’ and their role in locomotion in the lamprey spinal cord. Science 236: 312–314, 1987.
Cangiano L and Grillner S. Fast and slow locomotor burst generation in the hemispinal cord of the lamprey. J Neurophysiol 89: 2931–2942, 2003.
Cazalets JR, Bertrand S, Sqalli-Houssaini Y, and Clarac F. GABAergic control of spinal locomotor networks in the neonatal rat. Ann NY Acad Sci 860: 168–180, 1998.
Chambon JP, Feltz P, Heaulme M, Restle S, Schlichter R, Biziere K, and Wermuth CG. An arylaminopyridazine derivative of gamma-aminobutyric acid (GABA) is a selective and competitive antagonist at the GABAA receptor site. Proc Natl Acad Sci USA 82: 1832–1836, 1985.
Chebib M and Johnston GA. The "ABC" of GABA receptors: a brief review. Clin Exp Pharmacol Physiol 26: 937–940, 1999.[CrossRef][ISI][Medline]
Christenson J, Bongianni F, Grillner S, and Hokfelt T. Putative GABAergic input to axons of spinal interneurons and primary sensory neurons in the lamprey spinal cord as shown by intracellular Lucifer yellow and GABA immunohistochemistry. Brain Res 538: 313–318, 1991.[CrossRef][ISI][Medline]
Cohen AH and Harris-Warrick RM. Strychnine eliminates alternating motor output during fictive locomotion in the lamprey. Brain Res 293: 164–167, 1984.[CrossRef][ISI][Medline]
Cohen AH and Wallén P. The neuronal correlate of locomotion in fish. "Fictive swimming" induced in an in vitro preparation of the lamprey spinal cord. Exp Brain Res 41: 11–18, 1980.[ISI][Medline]
Davidoff RA and Aprison MH. Picrotoxin antagonism of the inhibition of interneurons by glycine. Life Sci 8: 107–112, 1969.[ISI][Medline]
Debarbieux F, Brunton J, and Charpak S. Effect of bicuculline on thalamic activity: a direct blockade of IAHP in reticularis neurons. J Neurophysiol 79: 2911–2918, 1998.
Dibas MI, Gonzales EB, Das P, Bell-Horner CL, and Dillon GH. Identification of a novel residue within the second transmembrane domain that confers use-facilitated block by picrotoxin in glycine alpha 1 receptors. J Biol Chem 277: 9112–9117, 2002.
El Manira A and Bussieres N. Calcium channel subtypes in lamprey sensory and motor neurons. J Neurophysiol 78: 1334–1340, 1997.
El Manira A, Shupliakov O, Fagerstedt P, and Grillner S. Monosynaptic input from cutaneous sensory afferents to fin motoneurons in lamprey. J Comp Neurol 369: 533–542, 1996.[CrossRef][ISI][Medline]
El Manira A, Tegnér J, and Grillner S. Calcium-dependent potassium channels play a critical role for burst termination in the locomotor network in lamprey. J Neurophysiol 72: 1852–1861, 1994.
El Manira A, Tegnér J, and Grillner S. Locomotor-related presynaptic modulation of primary afferents in the lamprey. Eur J Neurosci 9: 696–705, 1997.[CrossRef][ISI][Medline]
Goutman JD and Calvo DJ. Studies on the mechanisms of action of picrotoxin, quercetin and pregnanolone at the GABA
1 receptor. Br J Pharmacol 141: 717–727, 2004.[CrossRef][ISI][Medline]
Grillner S. On the generation of locomotion in the spinal dogfish. Exp Brain Res 20: 459–470, 1974.[ISI][Medline]
Grillner S. Control of locomotion in bipeds, tetrapods and fish. In: Handbook of Physiology. The Nervous System II. Motor Control. Bethesda, MD: American Physiol Soc, 1981, sect. 1, p. 1179–1236.
Grillner S. The motor infrastructure: from ion channels to neuronal networks. Nat Rev Neurosci 4: 573–586, 2003.[ISI][Medline]
Grillner S and Wallén P. Does the central pattern generation for locomotion in lamprey depend on glycine inhibition? Acta Physiol Scand 110: 103–105, 1980.[ISI][Medline]
Grillner S, Williams T, and Lagerback PA. The edge cell, a possible intraspinal mechanoreceptor. Science 223: 500–503, 1984.
Harris-Warrick RM and Cohen AH. Serotonin modulates the central pattern generator for locomotion in the isolated lamprey spinal cord. J Exp Biol 116: 27–46, 1985.
Heaulme M, Chambon JP, Leyris R, Molimard JC, Wermuth CG, and Biziere K. Biochemical characterization of the interaction of three pyridazinyl-GABA derivatives with the GABAA receptor site. Brain Res 384: 224–231, 1986.[CrossRef][ISI][Medline]
Hill R, Matsushima T, Schotland J, and Grillner S. Apamin blocks the slow AHP in lamprey and delays termination of locomotor bursts. Neuroreport 3: 943–945, 1992.[ISI][Medline]
Hill RH, Svensson E, Dewael Y, and Grillner S. 5-HT inhibits N-type but not L-type Ca(2+) channels via 5-HT1A receptors in lamprey spinal neurons. Eur J Neurosci 18: 2919–2924, 2003.[CrossRef][ISI][Medline]
Homma S. Physiology and pharmacology of putative transmitters in lamprey central nervous system. Prog Neurobiol 20: 287–311, 1983.[CrossRef][ISI][Medline]
Homma S and Rovainen CM. Conductance increases produced by glycine and gamma-aminobutyric acid in lamprey interneurones. J Physiol 279: 231–252, 1978.
) and picrotoxin (
). Inset: method of calculation. BQ is calculated using the heights of the 1st peak 
