Spinal and supraspinal functions of noradrenaline in the frog embryo: consequences for motor behaviour
- School of Biology, Bute Medical Buildings, University of St Andrews St Andrews, Fife KY16 9TS, Scotland, UK
- Corresponding author K. T. Sillar: School of Biology, Bute Medical Buildings, University of St Andrews, St Andrews, Fife KY16 9TS, Scotland, UK. Email: kts1{at}st-andrews.ac.uk
Abstract
The monoamine noradrenaline (NA) can initiate and/or modulate locomotion in a variety of vertebrates. Here we report that exogenous NA application can facilitate two completely different fictive behaviours in embryos of the common frog Rana temporaria, depending on whether spinal networks are connected to supraspinal centres. When the nervous system is intact, NA elicits a non-rhythmic coiling motor response, reminiscent of a spontaneous behaviour appropriate to drive hatching movements, but has only minor effects on evoked swimming activity. After the spinal cord has been severed from the brain, spontaneous coiling is no longer observed, nor can NA elicit it, but the amine can ‘release’ swimming rhythm generation in response to electrical skin stimulation. The rhythm is similar, but relatively inflexible when compared to fictive swimming recorded from intact animals. Our pharmacological tests indicate that α1-adrenoreceptors are involved in the permissive role of NA during spinalised rhythmic swimming and that the fictive coiling response to NA in intact animals involves descending inputs and the activation of β1-adrenoreceptors. Furthermore, the subtle effects of NA on evoked swimming in intact animals were mimicked by either α1- or α2-adrenoreceptor activation, reversibly decreasing motor burst durations and increasing their frequency. We discuss our results with reference to the known synergistic actions of NA with another aminergic neuromodulator, serotonin, and raise the possibility that these amines may actively regulate the release of one another during locomotion, in addition to their respective post-synaptic targets in the spinal cord.
The rhythmic movements responsible for vertebrate locomotion are driven by central pattern generators located in the spinal cord (Delcomyn, 1980; Grillner, 1985). One major goal in the pursuit of the spinal mechanisms underpinning locomotion is the recovery of function after injury (Schouenborg & Kiehn, 2002). Therefore, many studies are conducted in which injury is inflicted by either complete or partial spinal cord lesion (Barbeau et al. 1999). Historically, ‘fictive’ rhythmic activity appropriate to drive locomotor movements has been most successfully restored after complete lesion using pharmacological agents that replace the normal source of descending aminergic inputs (Viala & Buser, 1969; Forssberg & Grillner, 1973; Barbeau & Rossignol, 1990, 1991; Pearson & Rossignol, 1991; Kiehn et al. 1992). However, the same agents can actually have deleterious effects on locomotor recovery in animals with only a partially lesioned spinal cord, abolishing the weight support and stepping movements that can been attained solely through treadmill training (Rossignol et al. 2001). The differential effects of amines during rhythmic movements in intact versus spinalised animals will be important to resolve, because similar anomalies are apparent during pharmacotherapy in human patients (Dietz et al. 1995; Remis-Neris et al. 1999).
Frog embryos can provide useful insights into the location-specific effects of descending aminergic inputs. First, their nervous system is remarkably simple, yet exhibits many of the basic features of mammalian rhythmogenesis (Roberts, 1990). Second, the spinal circuitry responsible for rhythm generation in frog embryos is readily accessible to physiological and pharmacological manipulations. Third, the behavioural repertoire that emerges from this circuitry is limited and easily identified. Finally, there is a considerable body of evidence that aminergic modulation of spinal circuitry occurs in frog embryos (Sillar et al. 1998, 2002). We have chosen to investigate the location-specific effects of noradrenaline (NA) in embryos of the common frog Rana temporaria, primarily because, in contrast to a closely related species, embryos of the clawed frog Xenopus laevis (McDearmid et al. 1997; Fischer et al. 2001; Merrywest et al. 2002), very little is known of the noradrenergic modulation of locomotion in Rana (McDearmid & Sillar, 1997). In addition, there is reason to believe that descending aminergic inputs are more advanced in this species as compared to Xenopus at equivalent developmental stages (van Mier et al. 1986; Woolston et al. 1994). This is based on a comparison of the respective states of spinal serotonergic innervation, which are thought to be instrumental in transforming the relatively inflexible embryonic motor pattern in Xenopus (Kahn et al. 1982) into a more flexible larval one (Sillar et al. 1991, 1992a,b, 1995). Therefore any investigations using Rana are likely to shed light on more mature mechanisms of aminergic interaction with spinal networks. Here, we report that in Rana, NA can also have differing roles depending on the presence or absence of descending inputs. Moreover, these discrepancies can be explained on the basis of pharmacology. Aspects of this work have been published in abstract form (McLean & Sillar, 2001).
METHODS
Animals
Rana temporaria embryos were obtained during the spring breeding season (ca February to April) from a commercial source (Blades Biological Ltd, Cowden, UK) or collected from local duck ponds and kept in dechlorinated tap water between 4 and 22 °C until at the correct developmental stage (Fig. 1A; Gosner, 1960). All experiments were performed on pre-feeding embryos and are not therefore subject to UK Home Office regulations regarding animal experimentation as described in the Animals (Scientific Procedures) Act 1986.
A, experiments were conducted using Rana embryos staged according to Gosner (1960). Ba, fictive motor responses were recorded from embryos immobilised in α-bungarotoxin by using glass suction electrodes placed between the myotomes (as numbered from the otic capsule) on either the left (L) or right (R) sides. For spinalisation experiments, the dorsal portion of the first few myotomes was removed to facilitate access to the spinal cord. Bb, a schematic diagram of the brain and rostral spinal cord illustrating the approximate site of lesion (between vertical hatched lines). Eye, grey circle; otic capsule, grey oval; myotomal clefts, numbered diagonal lines. For details see Methods.
Electrophysiological recordings
Rana embryos were first anaesthetised in tricaine methanesulphonate (MS-222, 0.1–1 %; Sigma-RBI, Poole, UK) and then scored along the dorsal fin to aid immobilisation by immersion (ca 60–90 min) in the neuromuscular blocker α-bungarotoxin (12.5 μm; Sigma-RBI). Once removed from the toxin, the embryos were pinned down through the notochord on their right side onto a rotatable platform covered in silicone elastomer (Sylgard-184, Dow-Corning, Midland, MI, USA) within a 5 ml chamber, through which 100 ml of saline was re-circulating at 5 ml min−1 (composition (mm): 115 NaCl; 2.5 KCl; 2.5 NaHCO3; 10 Hepes; 1 MgCl2; and 2 CaCl2; pH 7.4). The flank skin was then removed from approximately the otic capsule to the anus using fine forceps and finely etched tungsten dissecting needles. For contralateral recordings, the embryo was positioned dorsal side up, secured with etched tungsten pins, and a portion of the flank skin on the right side was also removed (Fig. 1Ba). For spinalisation experiments, the dorsal portion of the second to the fifth myotome (as numbered from the otic capsule) was removed to facilitate access to the spinal cord (Fig. 1Bb), which was then severed from the brain using dissecting needles at the level of the third to the fourth myotome, and the embryos were left to recover for 10–20 min. Suction electrodes (ca 50–70 μm tip diameter), fashioned out of non-filamented borosilicate glass capillaries (1 mm outer diameter; Clarke Electromedical Instruments, Reading, UK) and fine-bore polythene tubing (2 mm outer diameter; Smith Industries Medical Systems Portex Ltd, Keene, NH, USA), were either placed over the intermyotomal clefts to record extracellular ventral root activity or placed on the tail skin to evoke it, with 1 ms electrical current pulses delivered via fine copper wire attached to a DS2 isolated stimulator (Digitimer, Welwyn Garden City, UK; Fig. 1Ba).
Data acquisition and analysis
Electrophysiological data were amplified × 10 000 via a differential AC amplifier (A-M Systems, Sequim, WA, USA) and stored on videotape using a PCM-4/8 adapter (Medical Systems Corporation, Greenvale, NY, USA). Data were either printed as hard copy off-line with an ORP oscillographic recorder (Yokogawa Ltd, Runcorn, UK) or digitised off-line with an A/D interface (Cambridge Electronic Design, Cambridge, UK), and were measured using the Spike 2 data analysis software (Cambridge Electronic Design) and DataView analysis software (courtesy of W. J. Heitler, University of St Andrews, St Andrews, Scotland, UK). Measurable parameters for non-rhythmic coiling activity included the duration of motor bursts, their longitudinal delay along the body and their spontaneous frequency. The burst durations and longitudinal delays were measured during 30 s of light dimming, when fictive coiling activity is most likely to occur (Soffe, 1991), while the spontaneous frequency was measured 5–10 min after drug application. Measurable parameters for rhythmic swimming activity included burst durations and longitudinal delay, but also the time between successive motor bursts (cycle period), the phase of contralateral motor bursts (calculated as the contralateral delay divided by the cycle period) and the total amount of time swimming activity lasted (episode duration). During experiments the amplitude of motor bursts could fluctuate according to the relative seal of the suction electrode, so this parameter could not be reliably analysed. Swimming was elicited by electrical stimulation and three consecutive episodes measured in control, drug and wash conditions with a 60 s rest interval (omitting the first episode due to the confounding effects of differing rest intervals). Single averages were subsequently determined for the measurable parameters per animal, the data were pooled and Student's paired t test was used to test significance from control values at P < 0.05. All statistical analysis was conducted using Microsoft Excel for Windows. Data are reported in the text as means ± s.e.m. A total of 86 Rana embryos were used in this study.
Drugs
The drugs used in this series of experiments were arterenol (noradrenaline, NA; 10–20 μm), clonidine (150–200 μm), isoproterenol (50–100 μm), metaproterenol (50–150 μm), phentolamine (50–100 μm), phenylephrine (150–200 μm) and propranolol (50–100 μm). All drugs were purchased from Sigma-RBI and dissolved fresh daily in distilled water. Drug concentrations were determined from those used effectively in Xenopus embryos and larvae (Fischer et al. 2001; Merrywest et al. 2002) and were washed off by replacement with fresh saline for 15–20 min.
RESULTS
Induction of a non-rhythmic motor pattern by NA in intact preparations
Rana embryos have a very limited behavioural repertoire, which includes rhythmic swimming movements that generate forward propulsion and non-rhythmic coiling movements that do not constitute locomotion (Soffe, 1991). It is believed that the non-rhythmic behaviour could be termed ‘hatching’, as it occurs spontaneously during the hatching period, precedes rhythmic locomotory swimming behaviour and would be appropriate to drive the strong body flexions needed to free embryos from their substantial, gelatinous egg membranes (Soffe, 1991). Spontaneous fictive motor activity can also be recorded in immobilised Rana embryos, which, like that observed in the freely behaving animal, consists of both swimming activity (29.2 ± 11.0 % of total, n = 52; for total value, see Table 1) and coiling-like activity (70.8 ± 11.0 % of total, n = 52). The fictive coiling activity was therefore more prevalent, particularly during dimming of the experimental bath illumination (Fig. 2Aa). The basic effects of NA have already been described in preliminary form (McDearmid & Sillar, 1997; McDearmid, 1998), but we have repeated them here to allow for a quantitative comparison with the effects of various pharmacological agents. Shortly after the bath application of NA, the frequency of fictive motor activity increased (Fig. 2Ab) for a transient period of time (ca 5–10 min) after which a wash (20 min or more) in fresh saline was required before the response could be repeated (data not shown). The motor activity induced by NA was predominantly coiling activity (90.1 ± 3.9 % of total, n = 9) and was qualitatively indistinguishable from that which occurs spontaneously or in response to light dimming, as it was characterised by prolonged bursts of motor activity, which propagated in a rostrocaudal direction and never appeared to occur synchronously on the opposite side of the body during contralateral recordings (Fig. 2B).
Summary of behavioural responses to NA and their pharmacology
Aa, recordings from the fourth (L4) and eighth (L8) intermyotomal clefts in a particularly quiescent preparation illustrate how dimming of the bath illumination (between black arrowheads) can trigger bouts of non-rhythmic motor activity (or fictive coiling), which can also occur spontaneously (grey arrowhead). Ab, shortly after the bath application of NA (at black arrow), bouts of predominantly fictive coiling behaviour are elicited for a transient period of time. Dashed lines are to illustrate trace continuity. B, fictive coiling responses elicited by NA on a faster time scale. C, excerpts of evoked swimming taken approximately 500 ms into the episode in both control conditions (Ca) and the presence of NA (Cb) illustrate some of the measurable parameters, including burst durations (BD), cycle periods (CP), longitudinal delay (LD) and the phase of contralateral bursts (Φ), calculated as the contralateral delay (CD) over the cycle period. L, left side; R, right side. Shaded grey boxes in the inter-burst intervals are present for illustrative purposes. D and E, further quantification of the measurable parameters with representative data from one animal illustrates a comparison between control and NA values. Note that there are more cycles in an episode in the presence of NA, illustrating its effect on episode durations.
Given the known modulatory effects of NA in tadpoles of Xenopus laevis (McDearmid et al. 1997), we were surprised to find no significant effects of NA on the parameters of the swimming rhythm in Rana (Fig. 2Ca and b). While there was a clear trend for burst durations, cycle periods and longitudinal delays to decrease, these differences were not statistically significant (Table 1). There was, however, a noticeable increase in the duration of swimming episodes in the presence of NA (Fig. 2D), similar to that described for Xenopus embryos (Fischer et al. 2001). Despite this increase in episode durations, burst durations and longitudinal delays in the presence of NA remained correlated with cycle periods (Fig. 2Ea and b), as they have been described in detail during control conditions (Soffe, 1991).
Facilitation of swimming by NA in spinalised preparations
Thus, in the immobilised but intact animal, NA selectively increased the probability of occurrence of the non-rhythmic coiling behaviour. We next severed the spinal cord from the brain to assess whether the spinal circuitry alone could account for this response to bath-applied NA. When deprived of inputs from descending interneurons following acute spinalisation, Rana embryos do not produce spontaneous motor activity as they normally do, there is no response to dimming of the bath illumination (Fig. 3Aa) and NA did not elicit the coiling activity described above. In a minority of cases (n = 3/10), NA instead initiated an episode of rhythmic swimming activity (Fig. 3Ab). However, by far the most consistent response to NA in spinalised preparations was the initiation of swimming by stimulation of the flank skin following trains of electrical current pulses (n = 10/10; Fig. 3C). Prior to bath application of NA, electrical skin stimulation elicited sustained bursts of motor activity that appeared uncoordinated. Shortly after application of NA, episodes of swimming could be elicited by skin stimulation (Fig. 3B and C), but only for a transient period of time (ca 5–10 min), after which a rest (ca 20–30 min) was needed before swimming could be elicited again (Fig. 3C). Interestingly, this time window appears to mirror that witnessed for the NA response in intact preparations. The swimming behaviour elicited by electrical stimulation was similar to that in spontaneous episodes of swimming and was characterised by a brief acceleration in swimming frequency followed by a gradual deceleration (Fig. 3D). The cycle periods gradually increased as the episode of swimming progressed, excepting a brief period at the beginning of spontaneous episodes, which corresponded to poorly co-ordinated activity (Fig. 3D) that often propagated in a caudorostral direction (see negative longitudinal delay values in Fig. 3Eb). Similar to swimming in intact Rana embryos, spinalised swimming was characterised by a contralateral alternation of motor bursts, as evidenced by a phase value at or near 0.5 (Table 1; cf. Fig. 2 and Fig. 3). However, unlike fictive swimming recorded in intact embryos (cf. Fig. 2), motor burst durations (Fig. 3Ea) and longitudinal delays (Fig. 3Eb) did not correlate with cycle period. Similarly, the cycle periods rarely exceeded 100 ms and burst durations never exceeded 50 ms, in contrast to the case in intact animals (Table 1).
Aa, recordings from the ninth (L9) and eleventh (L11) intermyotomal clefts in an acutely spinalised preparation illustrate that no spontaneous fictive coiling is present, nor can dimming of the bath illumination elicit it (between black arrowheads). Ab, in a minority of cases, shortly after the bath application of NA (at black arrow) a bout of fictive swimming was elicited. Dashed lines illustrate trace continuity. Note that due to the time scale fictive swimming looks like one continuous motor burst. Within this episode are discrete motor bursts as illustrated in B. B, electrically evoked swimming in the presence of NA in spinalised preparations. Shaded grey boxes in the inter-burst intervals are present for illustrative purposes. C, swimming was more reliably elicited by persistent electrical stimulation of the flank skin (artefacts at asterisks) for a transient period of time. Breaks in subsequent episodes represent 60 s rest intervals, except for the last, which represents a 25 min rest interval (at grey arrowhead). L, left side; R, right side. D and E, further quantification of measurable parameters of NA-evoked swimming and electrical stimulus-evoked swimming in the presence of NA with representative data from different animals. Note the differences as compared to intact animals in Fig. 2D and E on identical scales.
Pharmacology of the spinal NA response
The data thus far suggest that NA can broadly affect network excitability, but with two different behavioural outcomes depending on the presence or absence of inputs from descending interneurons in the brain. We next adopted a pharmacological approach to ascertain the receptor subtypes involved in each response. First, we examined the pharmacology of the response to NA when embryos were spinalised, as this was likely to be the simplest scenario. Prior to NA application, electrical stimuli can only elicit sustained, poorly coordinated bursts of motor activity as described above (Fig. 4Aa). When the broad-spectrum adrenergic α-receptor antagonist phentolamine was applied (n = 5), the subsequent application of NA could not facilitate the generation of rhythmic swimming movements (Fig. 4Ab). Phentolamine alone had no observable effects (data not shown). However, in the presence of the broad-spectrum adrenergic β-receptor antagonist propranolol, and NA (n = 8), rhythmic swimming movements could be generated in response to electrical stimuli (Fig. 4Ac). Again, propranolol alone had no observable effects (data not shown). The properties of fictive swimming facilitated by NA in the presence of propranolol were similar to those in the presence of NA alone; namely, cycle periods did not correlate with either burst durations or longitudinal delays (Fig. 4C). Also, burst durations and cycle periods rarely exceeded 50 and 100 ms (Fig. 4C), respectively, and the frequency of swimming gradually declined during the course of an episode (Fig. 4B). In addition, after approximately 5–10 min, swimming activity could not be elicited (data not shown). We did not, however, attempt to see whether swimming could again be elicited after a rest.
Aa, recordings from the fifth (L5) and eighth (L8) intermyotomal clefts in an acutely spinalised preparation illustrate that in control conditions, persistent electrical stimulation of the flank skin (truncated artefacts at asterisks) results in only poorly coordinated motor activity, which in general lasts no more than a few seconds. Dashed lines represent a break in the trace of approximately 500 ms, to illustrate this point. Ab, pre-application of the α-adrenoreceptor antagonist phentolamine occludes the normal facilitive effects of NA. Ac, after a wash in fresh saline, pre-application of the β-adrenoreceptor antagonist propranolol cannot occlude the facilitive effects of NA. Shaded grey boxes in the inter-burst intervals are for illustrative purposes. B and C, further quantification of measurable parameters during evoked swimming in the presence of NA and in the presence of propranolol (Prop) and NA from two different animals illustrates how similar the parameters are. Note, however, that episodes in the presence of both propranolol and NA were not as long as those in NA alone (NA data in B first illustrated as black triangles in Fig. 3D).
These results suggest the involvement of an α-adrenoreceptor subtype, but not a β-adrenoreceptor, in the facilitation of rhythm generation in the spinal cord by NA. To test this, the α2-adrenoreceptor agonist clonidine and the α1-adrenoreceptor agonist phenylephrine were each bath-applied separately. In the presence of clonidine (n = 4), no amount of electrical stimulation could facilitate swimming (Fig. 5Aa and b). To ensure that this was not a failure to respond on the part of the animal, NA was always applied after washout of clonidine. The subsequent application of NA consistently facilitated swimming after electrical stimuli (Fig. 5Ac). However, bath application of phenylephrine could mimic the noradrenergic response (n = 7; Fig. 5Ba and b). Again, the properties of fictive swimming resembled those elicited by NA (Fig. 5C and D). Notably, in contrast to NA alone (cf. Fig. 3Eb), the longitudinal delay during fictive spinalised swimming in the presence of either propranolol and NA (Fig. 4C) or phenylephrine (Fig. 5D) rarely propagated in a caudorostral direction.
A, recordings from the ninth (L9) and eleventh (L11) intermyotomal clefts in an acutely spinalised preparation illustrate that before (a) and after (b) bath application of the α2-adrenoreceptor agonist clonidine, electrical stimulation (artefacts at asterisks) cannot elicit rhythmic swimming movements. However, after a wash in fresh saline and the subsequent bath application of NA in the same animal, rhythmic swimming can be evoked (c). B, recordings from a different animal also from the ninth (L9) and eleventh (L11) intermyotomal clefts illustrate that bath application of the α1-adrenoreceptor agonist phenylephrine results in the electrical stimulation of swimming activity (b) where previously there was none (a). Shaded grey boxes in the inter-burst interval are present for illustrative purposes. C and D, further quantification of measurable parameters during evoked swimming in the presence of NA and in the presence of phenylephrine (Phee) from two different animals illustrates how similar the parameters are. Note, however, that episodes in the presence of phenylephrine were not as long as those in the presence of NA alone (NA data in C first illustrated as black triangles in Fig. 3D).
Pharmacology of the NA response in intact animals
In contrast to the clear α-adrenoreceptor mimicry of the effects of NA in spinal preparations, pre-application of phentolamine (n = 6) could not occlude the facilitation of non-rhythmic, coiling activity by NA in intact preparations (Fig. 6Aa, Ba and b). Neither could clonidine (n = 5) nor phenylephrine (n = 6) mimic the effect of NA on the frequency of spontaneous activity (Table 1; Fig. 6Bc). This lends support to the idea that intact descending systems are essential for the proper expression of the fictive coiling response, given the presumed localisation of α1-adrenoreceptors in the spinal cord. Pre-application of propranolol, on the other hand, could occlude the effects of NA on spontaneous activity (Fig. 7Ab and Bb). This therefore implies a role for β-adrenoreceptors in the NA coiling response in intact preparations, via a descending interneuronal pathway. In support of this, bath application of the non-specific β1,2-adrenoreceptor agonist isoproterenol could potently mimic the intact NA response (Table 1), increasing the occurrence of non-rhythmic coiling activity (Fig. 6Bd), for a transient period of time (Fig. 6Ac). Furthermore, the β2-adrenoreceptor agonist metaproterenol could not mimic the response to NA in intact animals (Fig. 6Ad and Bd). Taken together, these data suggest an important role for the β1-adrenoreceptor subtype in the selective facilitation of non-rhythmic coiling behaviour by NA.
Aa, recordings from the ninth (L9) and tenth (L10) intermyotomal clefts in an intact animal illustrate an excerpt of fictive swimming taken 500 ms from the beginning of the episode in control conditions. Ab, after 5–10 min in the presence of clonidine, fictive swimming frequency noticeably increases. Ac, this can be readily reversed after a 15–20 min wash in fresh control saline. Ad, in the same animal, subsequent application of phenylephrine for 5–10 min also increases swimming frequency. Ae, again, this is a readily reversible effect. Note that the two drugs were applied to the same animal on this occasion to illustrate the reversibility of these effects. Shaded grey boxes in the inter-burst intervals are present for illustrative purposes. B and C, further quantification of these parameters with representative data from two different animals illustrates that NA in the presence of propranolol (Prop/NA), but not in the presence of phentolamine (Phent/NA), consistently decreases burst durations and cycle periods throughout the course of an episode (downward and leftward shift) without dramatically altering their correlation. Note that phentolamine also occludes the increase in cycles per episode normally witnessed in the presence of NA.
Aa, recordings from the seventh (L7) and tenth (L10) intermyotomal clefts in an intact animal illustrate that pre-application of the general α-adrenoreceptor antagonist phentolamine cannot occlude the facilitation of the non-rhythmic, fictive coiling response to bath-applied NA (at black arrow). Ab, similarly placed electrodes in a different animal illustrate that pre-application of the general β-adrenoreceptor antagonist propranolol occludes the NA effect (at black arrow). Ac, bath application of the general β1,2-receptor agonist isoproterenol (at black arrow) to a different animal potently mimics the effect of NA on spontaneous fictive coiling activity. Ad, however, in yet another animal, application of the β2-adrenoreceptor agonist metaproterenol (at black arrow) cannot mimic the effect of NA on spontaneous fictive coiling activity. B, the percentage of total spontaneous activity (as reported in Table 1) taken up by either fictive swimming (□) or fictive coiling (▪) is graphically represented in bar charts. Normally, spontaneous fictive activity is dominated by fictive coiling activity. Phent, phentolamine; Prop, propranolol; Clon, clonidine; Phee, phenylephrine; Isop, isoproterenol; Meta, metaproterenol.
In addition to monitoring the effect of pharmacological agents on spontaneous activity, we also measured their effects on evoked swimming parameters. This was due to the subtle effects of NA on swimming, which could possibly have been clearer in response to selective pharmacological agents. While agonists and antagonists to β-adrenoreceptors did not appear to have any significant effects on the swimming rhythm (Table 1), activation of α-adrenoreceptor subtypes did (Fig. 7). For example, both clonidine (Fig. 7Ab) and phenylephrine (Fig. 7Ad) significantly (and reversibly, Fig. 7Aa, c and e) decreased burst durations and cycle periods, particularly with respect to one another (i.e. the percentage of cycle period taken up by burst duration or duty cycle; see Table 1). In addition, pre-antagonism of β-adrenoreceptors by propranolol unmasked similar effects with the subsequent bath application of NA (Fig. 7Ba–c), effects that were completely occluded by phentolamine (Fig. 7Ca–c). In contrast to these clear effects, those on the longitudinal delay of motor bursts were not significant (Table 1). However, since the separation distance of recording electrodes rarely exceeded four myotomes (ca 1 mm), the relatively short distance could have obscured effects that would have been more apparent if electrodes were placed further apart. Regardless, selective α-adrenoreceptor activation appears to modulate the swimming rhythm, in parallel with the inductive effects of NA on spontaneous coiling behaviour via β-adrenoreceptors.
DISCUSSION
Our results illustrate that: (1) the non-rhythmic, coiling motor pattern facilitated by NA via β1-adrenoreceptor activation requires intact descending systems and therefore cannot be explained by the effects of NA on the spinal circuitry alone; (2) NA can ‘release’ swimming rhythm generation in the transected spinal cord via α1-adrenoreceptors; and (3) fictive swimming in the presence of NA and the absence of descending inputs is less variable, suggesting that a descending influence, susceptible to both α1- and α2-adrenoreceptor modulation, is required to express a more flexible swimming repertoire. We will now discuss possible mechanisms that could account for the observed results.
Descending inputs and motor control
Descending inputs from the brain innervate the spinal cord during development in a spatiotemporal fashion that closely matches the acquisition of mature locomotor capabilities, such as postural control and walking (Vinay et al. 2000). This has led to the proposal that descending inputs are causally linked to the maturation of spinal networks (Sillar, 1994; Nishimaru & Kudo, 2000; Norreel et al. 2003; see however, Branchereau et al. 2002). However, there is evidence that immature neural networks can perform mature motor tasks, and that it is the descending information that changes instead (Le Feuvre et al. 1999). Regardless, supraspinal centres can exert a powerful influence over spinal rhythm generation, initiating, modulating and terminating locomotion (Grillner & Wallen, 2002; Li et al. 2003). Following spinal damage, the extent of spinal lesion will obviously determine the relative contribution of supraspinal centres to locomotion. Any pharmacological agent will therefore activate not only post-synaptic receptors on spinal neurons but also pre-synaptic receptors on the terminals of descending neurons. Such a situation has been proposed to account for the conflicting effects of clonidine on locomotor recovery in partially lesioned versus totally lesioned cats. For example, in chronically spinalised animals, intrathecal injection of clonidine immediately induces hindlimb stepping movements on a treadmill (Chau et al. 1998). When only the ventral or ventrolateral portions of the spinal cord are lesioned, intrathecal clonidine now interferes with the recovery of postural and motor control obtained through prior treadmill training (Brustein & Rossignol, 1999). It has been proposed that the latter is due to autoreceptor activation on the terminals of intact descending noradrenergic neurons and a consequent decrease in endogenous NA release. However, the sheer complexity of the mammalian nervous system makes it difficult to separate central effects from, for instance, proprioceptive feedback, which contributes to the normal motor pattern in cats (Pearson et al. 1998).
It is therefore encouraging that in the frog embryo, where the cellular and synaptic machinery underlying locomotion is far less complicated, different motor responses to NA are also observed, depending on the relative integrity of descending inputs. In intact animals, NA elicits a non-rhythmic motor response via β1-adrenoreceptor activation (Fig. 8A), which requires descending interneurons and corresponds to the earliest behaviour observed during the hatching period (Soffe, 1991). This ‘hatching’ response appears to be particularly light sensitive, as dimming of the experimental bath illumination can reliably elicit it (see Fig. 2Aa; Soffe, 1991; McLean et al. 2001). One possible source of descending excitation is the pineal eye. The light-sensing ability of the pineal eye and its role in behaviour are well described in Xenopus tadpoles (Roberts, 1978; Foster & Roberts, 1982; Jamieson & Roberts, 2000), where dimming of the illumination elicits swimming via a descending excitatory glutamatergic pathway (Jamieson & Roberts, 1999). Preliminary results suggest the dimming response is more efficacious in the presence of NA (McDearmid, 1998), indicating that NA may facilitate pineal signalling to reticulospinal centres. The facilitation of glutamatergic signalling by NA via β-adrenoreceptors is not unprecedented (Gereau & Conn, 1994). It is therefore likely that NA is facilitating glutamate release in the spinal cord, in such a way as to favour the expression of strong, ipsilateral body flexions. This β-adrenoreceptor-mediated response dominates any secondary effects on evoked swimming, which are only unmasked during selective α-adrenoreceptor activation. Our spinalisation experiments indicate that α1-adrenoreceptors are located on spinal neurons where they increase network excitability, facilitating swimming in the presence of NA but the absence of descending inputs (Fig. 8B). It would appear in the intact animal that this effect is manifested as an acceleration in swimming frequency and an increase in episode duration. However, we cannot completely rule out the involvement of additional α1-adrenoreceptors located on descending neurons. The fact that α2-adrenoreceptor activation accelerated swimming frequency, but that these receptors did not appear to be present on spinal neurons, certainly suggests such interactions with supraspinal centres.
A, with the central nervous system intact, NA facilitates hatching behaviour via β1-adrenoreceptor activation. B, by severing the spinal cord from the brain, NA can no longer induce β1-adrenoreceptor-mediated hatching behaviour, but does activate α1-adrenoreceptors that ‘release’ swimming behaviour. C, NA probably facilitates hatching via β1-adrenoreceptors located on descending interneurons (DIN), whilst it facilitates swimming via α1-adrenoreceptors located on neurons of the spinal central pattern generator (sCPG). It is possible that NA and serotonin (5-HT) interact at the brainstem level (dotted lines indicate possible synaptic interactions). Studies in higher vertebrates suggest that NA can inhibit its own release directly via α2-adrenoreceptors and more indirectly via α1-adrenoreceptors, which increase 5-HT release, which subsequently inhibits NA release (Svensson et al. 1975; Baraban & Aghajanian, 1980; Charlety et al. 1991).
An important caveat when conducting any pharmacological investigation is the specificity of the agents used and their affinity for the receptors they bind to, which may differ between species and according to developmental stage (Schmidt & Jordan, 2000). The concentration of drugs used in our study was determined from those used effectively in a closely related amphibian species at an equivalent stage of development (Fischer et al. 2001; Merrywest et al. 2002). While we cannot completely rule out the possibility that agonists were not entirely selective for a particular receptor, the clean separation of behavioural actions seen by the drugs at our concentration range makes this unlikely.
Duelling amines: NA versus serotonin
The often synergistic and complementary actions of different amines on fictive locomotion are well described, perhaps no more so than in tadpoles of the clawed frog Xenopus laevis, in which bath application of serotonin (5-HT; Sillar et al. 1992b; Wedderburn & Sillar, 1994) or NA (McDearmid et al. 1997; Fischer et al. 2001; Merrywest et al. 2002) leads to a strengthening or weakening of swimming, respectively. Both 5-HT and NA primarily accomplish this via opposing modulatory effects on a common pre-synaptic target, namely inhibitory glycine release from commissural interneurons onto motor neurons (McDearmid et al. 1997). The modulatory effects of 5-HT in Rana embryos are also known, where it increases burst durations and slows swimming frequency (Woolston et al. 1994). Our findings further support a role for 5-HT in the expression of a variable and flexible swimming motor pattern in Rana, since spinalised swimming in the presence of NA, but the absence of raphespinal innervation, is considerably more stereotyped. More importantly, however, until now there has been no evidence that NA modulates fictive swimming in Rana. Critically, the effect on fictive swimming mediated by NA is the opposite of that described for 5-HT, which suggests that as in Xenopus, 5-HT and NA have synergistic roles in Rana.
One intriguing possibility suggested by our results is that NA and 5-HT may interact upstream of the spinal cord, in the brainstem. It is known in mammals that the locus coeruleus and the raphe nuclei modulate the levels of activity of one another (Mongeau et al. 1998; Szabo & Blier, 2001; Pudovkina et al. 2002). For instance, NA neurons inhibit their own activity via α2-adrenergic autoreceptors (Svensson et al. 1975) and modulate the activity of 5-HT neurons via excitatory α1-adrenoreceptors (Baraban & Aghajanian, 1980). Similarly, 5-HT neurons can inhibit their own activity (Adell et al. 2002) and that of NA neurons (Charlety et al. 1991) via 5-HT1A receptors. Interactions between the aminergic nuclei would have consequential repercussions on rhythm generation in the spinal cord (Fig. 8C). For instance, if NA was increasing spinal 5-HT release, then autoinhibition of NA release could result in a decrease in 5-HT release. This effect on endogenous 5-HT release could explain why clonidine decreased burst durations and cycle periods. Clearly, detailed anatomical descriptions of the expression patterns of aminergic receptor subtypes are now needed to supplement the present data by clarifying at what levels the amines may interact and thus further elucidate the mechanisms of their location-specific effects.
Conclusions
Our pharmacological tests of the spinalised response to NA in Rana indicate an excitatory role for the α1-adrenoreceptor subtype in the spinal cord, as has been described in the cat (Chau et al. 1998; Lee & Heckman, 1999), neonatal rat (Sqalli-Houssaini & Cazalets, 2000) and Xenopus tadpole (Merrywest et al. 2003). Similarly, they implicate α2-adrenoreceptors in the modulation of locomotion and β1-adrenoreceptors in the initiation of motor activity, two effects that have also been described for NA in these species (Barbeau & Rossignol 1990, 1991; Kiehn et al. 1999; Fischer et al. 2001). It could be that these pharmacologically distinct roles are determined by the timing during development of descending NA axons, or alternatively the amine could naturally either elicit a coiling response or modulate swimming in parallel throughout larval life, the selection of which would be determined by its release site and the receptors activated. Regardless, our results are potentially very important as they suggest that the amines may actively regulate the release of each other during locomotion, in addition to their respective post-synaptic targets in the spinal cord. These interactions would not be apparent in reduced preparations with the spinal cord isolated, except of course where there were intraspinal sources of amines, such as in the lamprey (Svensson et al. 2001). Therefore, any pharmacological evaluation utilising the intact nervous system will have to consider secondary effects on other aminergic populations (or other descending systems). The simplicity of the tadpole preparation makes this possibility relatively easy to confirm and could shed light on similar mechanisms that may be at work in higher vertebrates.
Acknowledgments
We are very grateful to J. R. McDearmid and H. Fischer for useful discussions. We also thank S. D. Merrywest for critically reading earlier versions of this manuscript and post hoc data retrieval. This work was supported by The Wellcome Trust.
Footnotes
-
Author's present address D. L. McLean: Department of Neurobiology and Behavior, Life Sciences Building, SUNY Stony Brook, Stony Brook, NY 11794-5230, USA.
-
- Received April 14, 2003.
- Accepted June 11, 2003.
- © The Physiological Society 2003
