Long-Term Adaptation of a Phasic Extensor Motoneurone in Crayfish

Experiments with the fast excitor neurone (FCE) supplying the claw closer muscle in juvenile crayfish have identified a distinct form of plasticity that persists for at least 1 week (Lnenicka & Atwood, 1985a,bf This phenomenon, known as long-term adaptation (LTA), is evoked by conditioning this relatively inactive neurone for 2h per day for 3–14 days. Such conditioning causes the cell to release less transmitter at the beginning of a series of test stimuli and to display less synaptic depression when stimulated at 5 Hz. These changes make the FCE physiologically more similar to tonic crustacean neurones, which typically are more active, release less transmitter at low frequencies, and display marked facilitation but little depression (e.g. Atwood & Bittner, 1971). Conditioning also causes the synaptic terminals to become more varicose in appearance, and they can be seen to contain enlarged, branched mitochondria, making the phasic axon’s terminals morphologically similar to those of tonic neurones (Lnenicka et al. 1986).

LTA appears to involve the neuronal cell body, since the synaptic changes can be prevented by severing the nerve containing the FCE axon (Lnenicka & Atwood, 1985a). In addition, LTA can be induced using procedures that preferentially depolarize the cell body and dendrites (Lnenicka & Atwood, 1985a,b). This suggests that the synaptic changes may involve macromolecules that are synthesized in the soma and transported down the axon. Further studies to identify such compounds are limited because the FCE axon and soma are not very accessible for amino acid incorporation studies and protein analysis. In addition, the synaptic changes exhibited by the FCE are restricted to juvenile crayfish, which would provide very little material for biochemical study.

The primary goal of the present study was to determine whether neurones other than the FCE exhibit LTA. We also hoped to identify other preparations that might be more suitable for examining the molecular mechanisms. We report here that conditioning stimuli supplied to an identifiable phasic abdominal extensor motoneurone elicit the electrophysiological changes characteristic of LTA: a long-lasting reduction in the initial amplitude of EPSPs and resistance to synaptic depression. In addition, these effects occur in both juvenile and adult crayfish, indicating that the ability of the extensor motoneurone to exhibit LTA is not lost with ageing.

Crayfish (Procambarus clarkii), obtained commercially, were maintained in aerated freshwater tanks at 16 °C on a diet of carrots and lentils. Most experiments were performed on ‘intermediate-sized’ juvenile crayfish (Lnenicka & Atwood, 1985b), which were 5 –6cm in total length (rostrum to telson), had carapace lengths ranging from 2·3 to 2·7 cm, and weighed approximately 3 g. In experiments involving adult crayfish, the animals were 10 –13 cm long, had carapace lengths of 4·0 –5·5cm, and weighed between 14 and 45 g.

Extensor motoneurones were conditioned in situ by stimulating the second root of the third abdominal segment while monitoring the effects of stimulation by recording electromyographic potentials (EMGs) from the extensor muscles (Fig. 1). Stimulation was effected through two stainless-steel wires implanted through the dorsal surface of the abdomen on opposite sides of the second root (Fig. 1A). For EMG recordings, two wires were implanted near the posterior edge of the fourth abdominal segment, one lying near the midline and the other closer to the lateral edge. The wires were held in place with ‘Five-minute Epoxy’ and cyanoacrylate glues.

The second root of the third abdominal segment carries six motoneurones that innervate phasic extensor muscles in the third and fourth abdominal segments (Parnas & Atwood, 1966). The EMG electrodes recorded electrical activity from muscles in both segments. Thus, stimulating the extensor motoneurones produced a complex EMG waveform (Fig. 1B). The simplest procedure was to stimulate all the motoneurones during conditioning and, subsequently, to determine the effects of conditioning on a selected, identifiable neurone. Therefore, the intensity of the conditioning stimulus was kept just high enough to maintain the maximal EMG waveform. Subsequently, the synaptic properties of axon 3 (Parnas & Atwood, 1966) were studied (see below). A comparison of EMG recordings and intracellularly recorded EPSPs was made in three preparations and confirmed that the EMG wave resulted from activation of the phasic extensor neurones, including axon 3..

Conditioning stimuli were applied on the left side of the animal; homologous neurones on the right side served as controls. Responses to continuous stimulation at 5 Hz could not be maintained for long periods owing to depression, which caused the EMGs to diminish. Our aim was to observe responses throughout the conditioning period to ensure that the motoneurones had been activated and that the stimulating electrodes had not polarized. Depression was minimized by using 1 s bursts of 5 Hz stimulation, delivered every 2 s, to give an average conditioning frequency of 2·5 Hz. Conditioning stimuli were applied for 4 h per day, to obtain a stimulus paradigm comparable to that employed for the FCE (Lnenicka & Atwood, 1985a, b).

The dorsal half of the abdomen was dissected as described by Parnas & Atwood (1966) and superfused with aerated saline. The temperature ranged from 13·5 to 15·5°C but did not vary by more than 0·5°C for any given preparation. It was necessary to lower the Ca²⁺ concentration and raise the level of Mg²⁺ in the saline to prevent rapid twitches that could dislodge the recording electrodes and damage the muscle cell under study. Two different salines were used (Table 1). One, which contained five times the normal concentration of Mg²⁺ (5 × Mg saline), greatly reduced the EPSP size and was used for most experiments. The other (2·5 × Mg saline) did not reduce the EPSPs as severely and was used mainly for studying synaptic depression. In both salines, the level of NaCl was adjusted to compensate for changes in osmolality. Each segment of the abdomen contains three pairs of phasic extensor muscles, designated M, Li and L₂ (Parnas & Atwood, 1966). The present study took advantage of the fact that, of the six motor axons originating in the third abdominal segment, only one excitor (axon 3) and the common inhibitor (axon 5) innervate muscle L₁ of the fourth abdominal segment. EPSPs and synaptic currents were recorded from this muscle (Fig. 2) to simplify identification of the axon. In most experiments, axon 3 was selectively excited by placing a suction electrode with a very small tip diameter (20 μm) directly over it. With stimulation at 5 Hz, however, this procedure was not practical since the EPSPs initially facilitated, producing large contractions that dislodged the nerve from the stimulating electrode. In such cases it was necessary to stimulate the second root with a large suction electrode and to block the IPSPs from axon 5 using 1 mmol 1⁻¹ picrotoxin (Atwood et al. 1967). Although the common inhibitor and all the excitatory neurones were probably driven during conditioning, we confined our analysis of synaptic properties to axon 3 and did not attempt to determine whether the other neurones were affected as well.

Intracellular recordings were obtained using conventional microelectrodes filled with 3 mol 1⁻¹ KC1 and having resistances of 5 –15 M Ω. All EPSP amplitudes were corrected for non-linear summation (Martin, 1955), assuming a reversal potential of +11·5 mV (Onodera & Takeuchi, 1975). Recordings were made from selected sites across the width of muscle L₁ with the aid of an ocular micrometer.

Quantal synaptic currents were recorded with ‘macropatch’ recording electrodes (Dudel, 1981) that had openings of 10 μm diameter at their tips. These electrodes were carefully positioned over synaptic spots, and the seal around the tip was improved by pressing gently against the surface of the muscle fibre. If there was any sign of damage, such as a high rate of spontaneous release, the recording site was not used. Recordings were stored on FM tape for subsequent analysis. Quantal content (m) was measured directly by counting the total number of quanta released and dividing by the total number of stimuli (Johnson & Wernig, 1971). A large number of stimuli (310 –700) was supplied to the axon at a low frequency (0·5 Hz) which produced neither facilitation nor depression.

Binomial parameters p and n were estimated using a computerized method (B. Smith, J. M. Wojtowicz, and H. L. Atwood, in preparation; Wojtowicz et al. 1988), which: (a) generates various possible combinations of p and n, (b) calculates the ‘maximum likelihood estimate’ of n and p based on comparisons with the observed frequency distribution of quantal releases, and (c) selects the binomial values that provide the best fit to the observed distribution. A simple binomial model in which the probability of release is independent, stationary and uniform at all release sites was assumed. Since the model assumes stationariness of p, segments of the recorded responses (50 or 100 responses at a time) were analysed in series and, if the data showed a systematic change in p over time, the data were not used. Goodness of fit was determined by comparing the observed quantal distributions with those predicted from fitted values of p and n using a x² test with N—3 degrees of freedom and employing Yates’ correction factor (Johnson & Wernig, 1971). Only distributions having at least one degree of freedom with the χ²-test (i.e. those with at least one release of three quanta) were considered for estimating the binomial parameters.

Statistical comparisons between the conditioned and contralateral (control) sides of the experimental animals were made using a t-test for matched pairs (Ferguson, 1971).

Juvenile crayfish were conditioned for 3 days and were allowed to rest for 1 day before EPSPs were recorded in isolated preparations. EPSPs from axon 3 were recorded in 5 × Mg saline (Table 1) at a low stimulus frequency (0·1 Hz), which produced neither facilitation nor depression. Recordings were made at five positions, evenly spaced across the width of the muscle (Fig. 3). EPSP amplitudes from the conditioned side of the animal were significantly lower (P < 0·005) at all five recording sites. Combining the data from all sites gave mean values (±S.E.M.) of 1·60 ± 0·47 mV for the conditioned side and 8·55 ± 1·44 mV for the control side (t = 4·529; P < 0·001), for an average reduction of 81 %.

To investigate whether the reduction in EPSP amplitude was long-lasting, the time between the end of the 3-day conditioning procedure and EPSP measurement was varied from 1 to 7 days. EPSP amplitudes were measured at three sites in each muscle, and the results were averaged and expressed as a percentage of the mean EPSP amplitude on the unconditioned side (Fig. 4). During the first 5 days after conditioning, the EPSPs increased to about 50 % of control values, and there was no significant change thereafter. A linear regression of the data from which Fig. 4 was obtained gave a straight line with a slope significantly different from zero (t = 3·38; N = 31; P <0·01). This indicates a partial recovery of EPSP amplitude during the first 5 days after conditioning. The mean amplitude of EPSPs on day 7 was still significantly lower on the conditioned side (4·03 ± 0·48 mV) than on the control side (9·49 ± 1·71 mV; N = 7; t = 3·18; P < 0·01). Thus, as with the FCE, the reduction in EPSP amplitude persisted for at least 1 week.

The effect of conditioning on neuromuscular depression was examined by comparing initial values of the EPSP amplitude (measured at 0·1 Hz) with values obtained at the end of 20 min of continuous stimulation at 5 Hz. The 5 × Mg saline caused a very large reduction in the amplitude of unfacilitated EPSPs and significant facilitation at 5 Hz, which masked any depression for 20 min or longer. To study depression, a saline representing a more modest reduction in the ratio of Ca²⁺ to Mg²⁺ was used (2·5× Mg saline). This did not reduce the amplitude of unfacilitated EPSPs as severely (compare Figs 3 and 5) and allowed depression to occur within 20 min at 5 Hz. Facilitation was observed at 5 Hz and usually led to regenerative responses and rapid twitches, which were particularly prominent in unconditioned muscles. To avoid muscle damage, the microelectrode was removed after responses had been recorded at low frequency (0·1 Hz) and subsequently was re-inserted into the same cell after contractions at 5 Hz had subsided. EPSPs were obtained only from the muscle cell at the medial edge of L₁, since this cell was clearly visible in all preparations and could be reliably identified and penetrated. Responses at 5 Hz were monitored throughout using an extracellular electrode to record field potentials.

The amplitudes of EPSPs recorded before and at the end of 20 min of stimulation at 5 Hz are shown in Fig. 5A, which was obtained from animals that had been conditioned for 3 days and allowed 3 –4 days to rest. At the end of the 5 Hz train, EPSPs on the control side had decreased by 94% of their initial, unfacilitated amplitude (measured at 0·1 Hz), whereas only a 35 % reduction had occurred on the conditioned side (Fig. 5). This resulted in a mean EPSP amplitude that was significantly higher on the conditioned side (t = 2·64; N = 8 preparations; P < 0·025). Thus, as reported for the FCE (Lnenicka & Atwood, 1985b), conditioning produced a smaller initial EPSP that was more resistant to depression at 5 Hz.

Adult crayfish, weighing up to 45 g, were conditioned in the same manner as juvenile crayfish, except that the conditioning period was extended from 3 to 7 days. EPSPs were recorded 3-4 days after the end of conditoning. EPSP amplitudes, averaged from three sites on each muscle, were lower for conditioned neurones (3·69 ± 0·71 mV) than for controls (14·2 ± 2·2 mV), and the differences were statistically significant (t = 4·48; N=9; P < 0·005). Overall, a 74% reduction in EPSP amplitude was observed. Synaptic depression, which was examined in six preparations, showed the same trend as for juvenile crayfish (Fig. 5B). Stimulation at 5 Hz produced a small (21 %) reduction in EPSP amplitude on the conditioned side and a large (86 %) reduction on the control side. At the end of the 5 Hz train, EPSP amplitudes were higher on the conditioned side than on the control side (t = 2·71; P <0·025).

Synaptic currents were recorded in juvenile crayfish (in 5xMg saline) to determine whether the reduction in EPSP amplitude reflected a change in transmitter release. Individual quanta, indicated by distinct steps in the current trace (Fig. 6A), were clearly distinguishable. The quantal currents were approximately 100 –500 pA in amplitude and had rise times of less than 0·5 ms. Quantal content (m) was estimated at each recording site by applying up to 700 stimuli to the axon, counting the number of occurrences of 0,1, 2, 3 or more quantal releases (e.g. Fig. 6B), and dividing the total number of releases by the number of stimuli (Johnson & Wernig, 1971). Recordings were made from one, two or three sites on a given muscle, and the quantal contents were combined to give an average value of m for that preparation. In all, 11 sites from conditioned extensors and 10 sites from their contralateral controls were examined. Mean values of m (Fig. 6C) were significantly smaller on the conditioned side (t = 3·818; N =7 preparations; P < 0·005), and indicated an average reduction in quantal content of 65 %.

Binomial parameters n and p were estimated for recording sites that met the following criteria: (a) the rate of release did not vary consistently over time and (b) the number of releases was sufficient to compare the observed and predicted frequency distributions using a x²-test with N-3 degrees of freedom (see Materials and methods). Ten recording sites (five conditioned and five controls) from four animals met both criteria (Table 2). These sites showed a decrease in the estimated value of p in each preparation. In most cases, the estimated value of p was very low on the conditioned side (<0·10), making estimates of n highly unreliable (McLachlan, 1978). Therefore, no conclusions are drawn regarding the question of whether changes in n also occurred.

The present study demonstrates LTA in one of the phasic motoneurones innervating the crayfish abdominal extensor muscles. The electrophysiological changes induced by daily conditioning are similar in most respects to those previously reported for the crayfish FCE. Such similarities include a marked reduction in transmitter release, a decrease in synaptic depression at 5 Hz, and a long persistence (at least 7 days). Unlike the FCE, however, the extensor neurone investigated in the present study does not lose the ability to exhibit LTA during the transition to adulthood.

The synaptic current recordings provide direct evidence that conditioning causes a reduction in transmitter release. A quantal analysis of initial EPSPs was not possible with the FCE, which displays marked synaptic depression at low stimulus frequencies (Pahapill et al. 1985). However, differences in quantal content between conditioned and unconditioned FCE axons were apparent at the end of a 5 Hz train (Lnenicka & Atwood, 1985b). The cumulative evidence from both systems demonstrates directly that conditioning causes the nerve terminals to release less transmitter initially, whereas at the end of a high-frequency train the amount of transmitter released is actually greater for conditioned axons than for controls.

Possible changes in input resistance of the postsynaptic muscle cells were not investigated in the present study. The observed decrease in quantal content (65 %), however, was large enough to account for most of the observed reduction in EPSP amplitudes (81 %). We did not attempt to determine whether conditioning caused a change in mean quantal size, since the amplitude of individual quantal events at each recording site depends on the seal resistance around the rim of the macropatch electrode, which might not be the same in each case. Conditioning the FCE causes no change in input resistance of its postsynaptic muscle fibres (Lnenicka & Atwood, 19856).

The binomial analysis performed in the present study assumed a simple binomial model in which the probability of release at each site is equal and independent. Although this may not always be the case at crustacean neuromuscular synapses (Atwood & Tse, 1988; B. Smith, J. M. Wojtowicz & H. L. Atwood, in preparation), there was not sufficient evidence to reject the simple binomial model for any of the quantal distributions reported in the present study, as indicated by the x²-test.

A reduction in transmitter release can result from a decrease in the pool of transmitter immediately available for release or from changes in the ability of the nerve terminal to activate transmitter release. Cumulative evidence favours the latter process as a mechanism underlying LTA. Conditioning does not appear to deplete the stores of available transmitter, since conditioned neurones still display facilitation and even release more transmitter than controls when stimulated continuously at 5 Hz (Lnenicka & Atwood, 1985b). A similar argument has been made for the frog cutaneous pectoris preparation (Hinz & Wernig, 1988), where the degree of facilitation is greater for conditioned axons than for controls. The decrease in the binomial parameter p observed in the present study suggests a decrease in the probability of transmitter release at individual release sites and, thus, provides some evidence for a direct effect on excitation-secretion coupling.

The reduction in quantal content during LTA is likely to involve mechanisms that decrease calcium influx into the nerve terminals. Such mechanisms could include decreased depolarization in the synaptic regions (Parnas et al. 1982), possibly through a loss of electrical excitability of the nerve terminals (Dudel et al. 1984). Alternatively, the calcium channels could be affected directly.

The ability of the conditioned nerve to sustain transmitter release at 5 Hz is thought to be related to observed increases in mitochondrial volume within the synaptic terminals of the FCE (Lnenicka et al. 1986). It remains to be shown whether such changes also accompany LTA in the abdominal extensors. Further studies of the time course of such changes are also warranted.

The appearance of LTA in adult crayfish is somewhat surprising. Previous experiments with the FCE showed that conditioning for up to 14 days caused no change in initial EPSP amplitude and only small changes in fatigue resistance compared with pronounced changes in these factors in juvenile crayfish (Lnenicka & Atwood, 1985b). Such age-dependence was thought to reflect a greater potential for synaptic plasticity in younger animals, presumably because greater synaptic growth may be required to compensate for changes in postsynaptic input resistance as the muscle fibres grow (Lnenicka & Mellon, 1983). In the present study, a longer conditioning period was used for adult animals than for juveniles (7 days instead of 3 days). Thus, it is not clear whether older animals display LTA as readily as do juveniles. Nevertheless, the data indicate that at least some neurones do not lose the ability to exhibit LTA during ageing.

To date, experiments aimed at studying LTA have been limited to crayfish motoneurones. However, a recent study involving the motoneurones that innervate frog cutaneous pectoris muscles (Hinz & Wernig, 1988) demonstrates a phenomenon very similar to LTA. Stimulation of the motoneurones for several hours per day for 5 –8 days caused reductions in evoked endplate potentials and in quantal content similar in magnitude to the changes reported here. In addition, the degree of synaptic depression during brief, high-frequency trains was reduced (Hinz & Wernig, 1988).

Evidence from a number of systems indicates that chronic decreases in activity also produce synaptic changes consistent with LTA. For example, tetrodotoxin blockade increases the amplitude of EPSPs generated by guinea-pig sympathetic neurones (Gallego & Geijo, 1987), by la sensory afferent neurones (Gallego et al. 1979) and by mammalian motoneurones (Snider & Harris, 1979). Immobilization causes increased transmitter release and increased synaptic fatigue in vertebrate motoneurones (Robbins & Fischbach, 1971) and in the crayfish FCE (Pahapill et al. 1985). Sensory deprivation causes similar changes in the developing sensory systems of insects (Bloom & Atwood, 1980; Murphey & Matsumoto, 1976). Thus, a growing body of evidence suggests that LTA reflects a general property of neurones, allowing them to adjust their release characteristics to suit the demands of their chronic activity patterns.

We thank Dr Gregory Lnenicka for valuable discussions and advice and Ms Marianne Hegström-Wojtowicz for assistance in preparing the figures. This work was supported by a fellowship (to AJM) from the Alberta Heritage Foundation for Medical Research and by a grant (to HL A) from the Natural Sciences and Engineering Research Council of Canada.