INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In vertebrates, primary afferents that normally convey sensory information (orthodromic spikes) from the periphery to the CNS, can also convey rhythmic bursts of antidromic spikes either during fictive locomotion in vitro (in cat, Dubuc et al. 1985, 1988; Gossard et al. 1989, 1991; in hamster, Bagust et al. 1985; Chen et al. 1993; and in rat, Bagust et al. 1997; Vinay and Clarac 1999) or during treadmill walking in vivo (in cat, Beloozerova and Rossignol 1994, 1995 and in rat, Bulgakova et al. 1985; Piliavskii et al. 1988). Such antidromic discharges have been correlated both in vertebrates and invertebrates with primary afferent depolarizations (PADs) that accompany presynaptic inhibition (Eccles et al. 1962, 1963; Jiménez et al. 1988; Kennedy et al. 1974; Kirk and Wine 1984; Sillar and Skorupski 1986). Centrally, PADs reduce the amplitudes of the orthodromic sensory spikes and thereby the amount of the transmitter release they induce, thus reducing the excitatory postsynaptic potentials in the postsynaptic neurons (Cattaert and El Manira 1999; Cattaert et al. 1992, 1994; Hedwig and Burrows 1996; Kirk 1985; Pearson and Goodman 1981).

In the crayfish Procambarus clarkii, during fictive locomotion, 90% of the primary afferents of a leg proprioceptor (the coxo-basal chordotonal organ, CBCO) display phasic bursts of PADs (4-20 mV in amplitude) locked in phase with the locomotor rhythm (El Manira et al. 1991b). The large-amplitude PADs (15-20 mV) seen in 40% of these afferents are able to trigger spikes that travel antidromically in the sensory axons toward the periphery (El Manira et al. 1991b). In addition, these antidromic spikes have no postsynaptic effect centrally (Cattaert et al. 1994; El Manira et al. 1991b; Gossard 1995), but recent data show that antidromic spikes interfere with the incoming sensory discharge (Bévengut et al. 1997, Gossard et al. 1999). In this paper, we looked at how the characteristics of bursts of antidromic spikes modify the sensory activity using the in vitro preparation of the crayfish locomotor system, which offers the possibility of simultaneously stimulating and recording intracellularly from identified CBCO afferents.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Experimental animals

Experiments were performed on male and female crayfish (P. clarkii, n = 21) weighing 25-30 g. The animals were purchased from a commercial supplier (Château Garreau, France), kept in circulating fresh water at 18-20°C and fed once a week.

In vitro preparation

An in vitro preparation of the thoracic nervous system was used (El Manira et al. 1991a; Sillar and Skorupski 1986). It consists of the last three thoracic and the first abdominal ganglia dissected together with all the nerves of the two proximal segments of the left fifth pereiopod (Fig. 1A). The strand of the CBCO was also dissected and kept intact, and its distal end was attached to an electromagnetic puller VT101 (Ling Dynamic Systems, Meudon-la-Forêt, France) controlled by a home-made function generator. The preparation was pinned dorsal side up on a silicone elastomer (Sylgard)-lined petri dish (Dow Corning, Wiesbaden, Germany). The nervous system was continuously superfused with oxygenated control saline (in mM): 195 NaCl, 5.36 KCl, 13.6 CaCl₂, 2.6 MgCl₂, and 2.98 HEPES (Sigma Chemical, St Louis, MO) at pH = 7.6. The fourth and fifth ganglia were desheathed to improve the superfusion of the central neurons and to allow intracellular recordings from CBCO axon terminals.

View larger version (30K): [in this window] [in a new window]   Fig. 1. Experimental preparation and protocol. A: the isolated preparation of the crayfish comprises the last 3 thoracic ganglia (T3-T5) and the 1st abdominal one (A1), the proximal part of the left 5th leg nerves, the coxo-basopodite chordotonal organ (CBCO) and its sensory nerve (Cbn). Axon terminals from the sensory afferents from the CBCO (CBT) were recorded intracellularly within the left 5th hemiganglion. B: intracellular recording of a tonic firing afferent (CBT) spontaneously active while a bursts of antidromic spikes was delivered (Antidromic stimulation). C: in an expanded time scale of the CBT recording, the different cellular events can be discriminated: the antidromic and orthodromic spikes, and the postsynaptic depolarizations of the afferent (PAD). A single antidromic spike was elicited by each square pulse of depolarizing current (i) injected.

Recordings

Intracellular recordings from CBCO afferent terminals (CBTs, Fig. 1B) were performed with glass micropipettes (Clark Electromedical Instruments, Reading, UK) filled with 3 M KCl (resistance, 10-20 M) connected to an Axoclamp 2A amplifier (Axon Instruments, Foster City, CA) used in the current-clamp mode. Data were displayed and printed on a eight-channel digital oscilloscope Hioki 8825 (Hioki E. E., Nagano, Japan), recorded on a digital tape recorder Biologic-1800 (Biologic, Claix, France), and stored onto a personal computer through appropriate interface (Micro 1401) and software (Spike2) from Cambridge Electronic Design (Cambridge, UK). Interface and software were also used to synchronize the imposed movements of the CBCO strand with the injection of current pulses into the recorded afferent.

Antidromic stimulation and data analysis

Each antidromic spike in the CBTs (Fig. 1, B and C) was triggered by injecting a square pulse of depolarizing current. The amount of injected current and the duration of each pulse were set to trigger only one antidromic spike per current pulse. As shown in Fig. 1C, the analysis of the intracellular recording of the CBT enabled us to discriminate easily between the antidromic and the orthodromic spikes.

In most of the experiments, antidromic spikes were delivered in bursts whose frequency and duration were controlled. Otherwise stated, each experimental trial was composed by N bursts of antidromic spikes separated by 40-s time intervals. To find out whether the antidromic spikes in a given sensory neuron modified its orthodromic spiking activity, a peristimulus histogram was calculated (bins of 20-30 ms) and normalized by dividing the bin values by N. Thus in the results, the normalized histograms express the averaged number of occurrences of the orthodromic spikes per bin against time.

Statistical analysis

The results are given as mean values ± SE. The statistical significance of the effects of a parameter (frequency, duration, number of spikes) of antidromic burst onto the orthodromic activity was assessed by a Newman-Keuls multiple comparison test following a one-way ANOVA. In other cases, a Student's t-test was used to assess statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this work, we have studied the effects of both the frequency and the duration of antidromic spike trains on the firing activity of the CBCO sensory neurons. The characteristics of the antidromic bursts were chosen in the range of values observed in CBCO neurons during episodes of fictive locomotion in vitro. In such intracellular recordings from CBCO sensory neurons, antidromic spikes occur in bursts (duration, 1-4 s; frequency, 20-100 Hz). The effects of trains of antidromic spikes were tested on the activity of the different CBCO cell types (Le Ray et al. 1997; Mill 1976). We have analyzed 34 sensory neurons (11 position-sensitive, 19 phasotonic, and for 4 phasic). In these experiments, 82% of the sensory neurons (28/34) showed a modification of their orthodromic activity in response to the antidromic bursts. However, the six that did not respond belonged to the three types of CBCO neurons: three presented a tonic, two a phasotonic, and one a phasic firing activity.

Effects of antidromic spikes on the activity of tonic firing neurons

EFFECTS OF ANTIDROMIC SPIKES DELIVERED AT LOW FREQUENCY ONTO TONIC SENSORY ACTIVITY. Bursts of antidromic spikes of 20-s duration were delivered at low frequency (1-10 Hz) separated by 20-s intervals (Fig. 2A). For each interval of 20 s, with and without the antidromic stimulations, the mean frequency of firing of the orthodromic spikes was calculated. The results were then expressed as percentage of the mean frequency without stimulation (100%; Fig. 2B). The mean frequency of the orthodromic spikes decreased linearly as the frequency of the antidromic bursts increased (r² = 0.989), showing that <5% of the orthodromic activity was present at 10 Hz. Furthermore a significant reduction of the orthodromic frequency was already seen for antidromic bursts at 1 Hz (Student's t-test: P < 0.01).

View larger version (32K): [in this window] [in a new window]   Fig. 2. Effects of low-frequency antidromic discharge on the orthodromic activity of tonic firing afferent. A: intracellular recording from tonic afferents (CBT) are shown for before and during antidromic stimulations at 1 Hz (top) and 5 Hz (bottom). B: the reduction of the orthodromic firing activity of sensory afferents is given as percentage of the control mean frequency (values calculated >20 s before the antidromic stimulations and 20 s during the antidromic stimulations) against the frequency of the antidromic stimulations (from 1 to 10 Hz). Data in B and C are from 2 different experiments.

EFFECTS OF BURST OF ANTIDROMIC SPIKES ONTO TONIC SENSORY ACTIVITY. Antidromic bursts of fixed duration (from 100 ms to 1 s) were delivered with increasing frequency. In Fig. 3A, the intracellular recording of a CBT is shown when an antidromic discharge of 1-s duration at 30 Hz was applied (top) and the averaged number of orthodromic spikes per 20 ms are given for n = 21 stimulations (bottom). For all of the tonic firing neurons that responded when the antidromic bursts were delivered (8 of 11), three results were obtained. First, the averaged number of orthodromic spikes during the stimulations was reduced for burst frequency <20 Hz and suppressed for frequency >20 Hz. Second, as the frequency of the antidromic bursts increased, a silent period (see  in Fig. 3A) in the orthodromic activity appeared. For antidromic burst durations of 200 ms, the delay before the orthodromic firing resumed increased significantly with increasing antidromic burst frequency (1-way ANOVA: P < 0.0001 for each given burst duration: 200 ms, 500 ms, and 1 s; linear regression r²: 0.122, 0.187, and 0.679 for burst durations of 200 ms, 500 ms, and 1 s, respectively; Fig. 3B). However, when the duration of the antidromic bursts was decreased, silent periods of same duration were obtained for higher antidromic burst frequency. Besides, it is worth noting that whatever the antidromic burst frequency for short burst durations such as 100 ms, the silent periods remained around 300 ms without significant variations (1-way ANOVA: P = 0.3454). Third, when the orthodromic activity resumed, the averaged number of orthodromic spikes increased more slowly to control values as the antidromic burst frequency increased. The comparison of this progressive recovery was analyzed by calculating the mean frequency of the orthodromic firing activity during the first 2 s after the orthodromic activity resumed (see F in Fig. 3A). On the one hand, for a given duration of antidromic burst (>100 ms), the mean orthodromic frequency decreased significantly with increasing antidromic burst frequency (Fig. 3C; 1-way ANOVA: P < 0.0001; linear regression r²: 0.212 and 0.265 for burst durations of 200 ms and 1 s, respectively). Thus orthodromic firing activity took longer to resume for higher than lower burst frequency. On the other hand, for short antidromic burst durations (e.g., 100 ms), the averaged frequency of the resumed sensory activity was independent from the frequency of the antidromic bursts (1-way ANOVA: P = 0.9409). Similar effects were observed in 8 of 11 position-coding CBCO neurons. Three neurons did not display any silent period outlasting the antidromic burst; however, their activity was decreased and blocked during the antidromic burst when delivered over a threshold frequency.

View larger version (46K): [in this window] [in a new window]   Fig. 3. Effects of bursts of antidromic spikes on the orthodromic activity of tonic firing afferents. A: an intracellular recording of a CBT (top) during a 30-Hz antidromic burst of 1-s duration is shown with an histogram of the averaged number of orthodromic spikes per bin of 20 ms against time for n = 21 antidromic stimulations. To analyze the effects of the antidromic bursts, the frequency of the orthodromic spikes was calculated for an interval of 1 s starting 1 s after the cessation of the stimulation (F) and by measuring the duration of the silence () between the end of the stimulation and the 1st orthodromic spike occurring for N trials at a given antidromic frequency. B: histograms of the mean duration (±SE) of the silence () is given against the frequency of bursts of antidromic spikes for a constant duration (100 ms, 200 ms, 500 ms, 1 s). C: the mean frequency (±SE) of the orthodromic activity (F) is given against the frequency of bursts of antidromic spikes for a constant duration (100 ms, 200 ms, 1 s). Data in A-C are from the same experiment. The number of trials for each condition is indicated on top of the histogram bars.

View larger version (18K): [in this window] [in a new window]   Fig. 4. Effects of bursts of a constant number of antidromic spikes at increasing frequency on the orthodromic activity of tonic firing afferents. The mean (±SE) silence duration (A and B, top) and the mean frequency (±SE) of the orthodromic activity (A and B, bottom) against the frequency of the antidromic stimulations are given for bursts of 10 antidromic spikes in A and 20 antidromic spikes in B. Data in A and B are from the same experiment. The number of trials for each condition is indicated on top of the filled histogram bars.

Cumulative effects of successive antidromic trains

To analyze a possible summation of the effects produced by the antidromic bursts onto the orthodromic activity of tonic firing neurons, antidromic bursts (500-ms duration, 50 Hz) were regularly delivered at 10- and at 2.5-s inter-burst intervals. When inter-burst intervals of 10 s were used (Fig. 5A, top), a slight cumulative effect was observed. The mean instantaneous frequency between the antidromic bursts (Fig. 5A, bottom) decreased progressively after each of the successive antidromic burst (e.g., 7.38 ± 0.68 Hz in control situation, 4.81 ± 0.91 Hz after the 4th antidromic stimulation). When inter-burst intervals of 2.5 s were used (Fig. 5B), the mean frequency of the sensory discharge strongly decreased after the first antidromic burst and then progressively decreased after each subsequent antidromic burst up to the 15th antidromic burst, after which no further decrease of the orthodromic firing frequency occurred. The initial activity could be restored, however, after a 20-s rest (Fig. 5C). After this rest, a new series of antidromic bursts resulted in a decrease in sensory activity very similar to the effects of the first antidromic series. If the inter-burst intervals were further decreased (not shown), the orthodromic activity was silenced. Cumulative effects of antidromic bursts were observed in six of eight other tonic CBCO neurons. The two neurons that did not display any cumulative effect were the same that did not show any silent period outlasting the antidromic discharge.

View larger version (26K): [in this window] [in a new window]   Fig. 5. Effects of varying the interburst intervals of antidromic bursts on the orthodromic activity of tonic firing afferents. Antidromic spikes were delivered in bursts of 500-ms duration at 50 Hz at fixed interburst intervals (10 s in A, 2.5 s in B and C). A: the successive bursts of antidromic spikes produced a decrease of the instantaneous firing frequency (Inst. Freq.) more pronounced after each consecutive burst. B: in the same neuron, the mean frequency of the orthodromic activity decreases progressively after each antidromic stimulation (rank number). C: a raster display of the orthodromic activity is shown for 2 series of antidromic stimulations separated by a 20-s resting period. Data in A-C are from the same experiment.

Effects of antidromic spikes on the activity of phasotonic firing neurons

Phasotonic neurons produce a phasic burst of orthodromic spikes during either stretch or release movements imposed to the CBCO stranddepending on their movement sensitivityand a tonic orthodromic discharge more or less pronounced depending on the maintained lengths of the strand (Figs. 6 and 7). To study the effects of the bursts of antidromic spikes on the orthodromic activity, we activated these sensory neurons by applying ramp movements to the CBCO strand, i.e., from a maintained position of the strand, a ramp movement was applied, followed by a maintained position before a ramp movement in opposite direction resumed the first maintained position. Bursts of antidromic spikes were elicited either during or before the first ramp movement.

View larger version (31K): [in this window] [in a new window]   Fig. 6. Effects of bursts of antidromic spikes delivered during the movement of the CBCO strand on the orthodromic activity of phasicotonic firing afferents. A: the orthodromic activity of a phasicotonic afferent (CBT) is displayed during the movement of the strand (mvt, bottom) in control condition (top) and during the application of an 10-Hz antidromic burst lasting the whole duration of the 1st ramp movement (middle). The number (1) represents The phasic response of the CBT to the applied movement (1) and the 1st second of maintained position of the strand (2) show the durations for which the mean firing frequency of the orthodromic activity are calculated. B: histogram of the mean frequency (±SE) of the phasic response (1) against control (Cont) and increasing frequency of the antidromic bursts for N stimulations. C: histogram of the mean frequency (±SE) during the maintained position (2) against control (Cont) and increasing frequency of the antidromic bursts for N stimulations. Data in A-C are from the same experiment. The number of trials for each condition is indicated on top of the histogram bars.

View larger version (55K): [in this window] [in a new window]   Fig. 7. Effects of bursts of antidromic spikes delivered prior to the movement of the CBCO strand on the orthodromic activity of phasicotonic firing afferents. The averaged number of the orthodromic spike activity per bin of 10 ms is plotted against time for N antidromic stimulations during control (CONT) and during increasing frequency of the antidromic bursts: in A1, the antidromic bursts have a 1-s duration, in B1, they have a constant frequency of 100 Hz. The mean frequency (±SE) of the phasic response (1) and during the maintained position (2) are plotted against control (CONT) and increasing frequency of the antidromic bursts for N stimulations, respectively, in A, 2 and 3, and against control (CONT) and increasing duration of the antidromic bursts for N stimulations, respectively in B, 2 and 3. Data in A and B are from 2 different experiments. The number of trials for each condition is indicated on top of the histogram bars.

EFFECTS OF ANTIDROMIC BURSTS DURING THE RAMP MOVEMENT. When antidromic bursts were triggered during the movement (Fig. 6A; control, top; stimulation, middle; movement, bottom), the mean frequency of the orthodromic phasic discharge [see (1) in Fig. 6A] decreased as the frequency of the antidromic bursts increased (Fig. 6B). A one-way ANOVA was performed (P < 0.0001), and a Newman-Keuls multiple comparison test assessed a significant decrease for all antidromic bursts >10 Hz versus control (P < 0.001) and for 10-Hz bursts versus control (P < 0.01). Moreover, the phasic discharge was totally abolished when the antidromic burst frequency reached a certain value between 50 and 80 Hz depending on the neuron tested. In addition, the antidromic discharges applied during the ramp movements were able to modify the firing frequency during the maintained positions (Fig. 6C). This effect was characterized by calculating the mean frequency of the orthodromic tonic discharge during the first 2 s [see (2) in Fig. 6A] of the maintained position. A one-way ANOVA was performed (P = 0.0066), and a Newman-Keuls multiple comparison test assessed that the reduction of the mean frequency of the tonic discharges during the maintained positions after antidromic bursts at 10 and 15 Hz was not significantly different for control (P > 0.05 in both cases). However, a significant reduction of the tonic discharges versus control was observed for antidromic burst frequency at 20 Hz (P < 0.05 for 20 Hz compared with control; P < 0.05 for 25 Hz compared with control; P < 0.01 for 40 Hz compared with control; P < 0.001 for 50 Hz compared with control). Furthermore, increasing the frequency of antidromic burst >20 Hz did not significantly increase the inhibition of the orthodromic activity during the maintained position (Newman-Keuls multiple comparison test: P > 0.05 for all pairs of frequency between 20 and 50 Hz). Therefore antidromic bursts delivered during the ramp movements have stronger effect in reducing the orthodromic activity during the phasic responses than during the maintained positions. Similar results were obtained from 18 of 19 phasotonic CBCO neurons. One CBCO neuron was much less sensitive to antidromic trains (no significant decrease of its movement-related response was observed for antidromic trains <15 Hz, and no significant change in its tonic discharge during the maintained position was observed whatever the antidromic frequency used).

EFFECTS OF ANTIDROMIC BURSTS DELIVERED BEFORE THE RAMP MOVEMENT. The antidromic bursts were delivered prior to the movement in order that their terminations matched the beginnings of the movement. The effects of the antidromic bursts were analyzed by calculating the mean frequency of the orthodromic activity during the applied movement and during the first second of the maintained position after the movement termination [see, respectively, (1) and (2) in Fig. 7A1].

View larger version (43K): [in this window] [in a new window]   Fig. 8. Effects of bursts of 40 antidromic spikes when delivered prior to the movement of the CBCO strand on the orthodromic activity of phasicotonic firing afferents. The mean frequency (±SE) of the phasic response (1) and during the maintained position (2) are plotted against control (CONT) and increasing frequency of the antidromic bursts for N stimulations, respectively in A and B. The data in A and B are from the same experiment. The number of trials for each condition is indicated on top of the histogram bars in A.

Effects of antidromic spikes on the activity of phasic firing neurons

Phasic neurons produce a phasic burst of orthodromic spikes during either stretch or release movements imposed on the CBCO strand and are silent during maintained positions. In these neurons, when antidromic bursts with either a constant duration or a constant frequency were delivered during or before the movements of the strand, similar results to those with the phasotonic neurons were obtained. In a phasic sensory neuron, the effects of antidromic bursts consisting in fixed numbers of spikes delivered prior to the movements imposed to the proprioceptor, are illustrated in Fig. 9. As for the phasotonic sensory neurons, the decrease in the orthodromic mean frequency was correlated with an increase in antidromic frequency. For example antidromic burst durations of 1 s (Fig. 9, A-C, ) or 400 ms (Fig. 9, A-C, ) progressively decreased the response to movement as the antidromic frequency was increased. Antidromic bursts exerted a highly significant decrease of the response to movement (1-way ANOVA: P < 0.0001 for values in A-C; Newman-Keuls multiple comparison test P < 0.001 for each of the different frequency against control). Similar results were observed in 3 other experiments. In addition, when a fixed number of antidromic spikes were triggered, the higher the antidromic burst frequency, the larger the decrease of the sensory activity. However, this effect increased up to a limit value over which increasing the antidromic burst frequency did not decrease further the sensory response to movement. Therefore as was the case in phasotonic neurons, the number of antidromic spikes, per se, does not account for the amount of inhibition observed on the sensory response to movement. The amount of inhibition is rather related to the frequency at which antidromic spikes are delivered, and to the duration of the antidromic burst. When a fixed number of antidromic spikes is used, the antidromic burst duration became a limiting factor to induce a further inhibition of the orthodromic activity during the phasic responses to movements even though the antidromic burst frequency was increased. Similar results were observed in three other experiments.

View larger version (29K): [in this window] [in a new window]   Fig. 9. Effects of bursts of fixed number of antidromic spikes delivered prior to the movement of the CBCO strand on the orthodromic activity of phasic firing afferents. The mean frequency (±SE) of the phasic response (1) is plotted against control (CONT) and increasing frequency of the antidromic bursts for N stimulations, when the bursts comprise 20 (A), 30 (B), and 40 (C) antidromic spikes. , bursts of antidromic spikes at a same frequency; , bursts of same duration. Data in A-C are from the same experiment. The number of trials for each condition is indicated on top of the histogram bars.

Effect of antidromic bursts compared with adaptation

A possible explanation of the effects of the antidromic bursts on the sensory activity of the CBCO neurons could be that antidromic spikes contribute to adaptation of the sensory discharge. Adaptation could account for the decrease of the instantaneous firing frequency of the orthodromic spikes during the response to ramp movement imposed to the CBCO strand (Fig. 10A). At the beginning of the ramp movement, the sensory discharge could reach an instantaneous firing frequency of 250 Hz (see Fig. 10A, ). However, at the end of the sensory response the instantaneous frequency of this sensory discharge was lower and ~100 Hz (see Fig. 10A, ). This decrease in the sensory response was consistently observed when the ramp movement was repeatedly applied. In Fig. 10A, the instantaneous frequency of 12 movement responses have been averaged (see ; bin size = 50 ms). If adaptation was responsible for the decrease in frequency of the sensory discharge along the imposed ramp movement, then an antidromic stimulation delivered prior to the imposed movement would probably result in a decrease of the sensory response to movement. To test for this hypothesis, we have compared the time course of the phasic response to movement in two conditions: in the absence (Fig. 10A) and after an antidromic burst (stim, Fig. 10B) delivered prior to ramp movement imposed to the strand. The frequency of the antidromic burst (75 Hz) was chosen as to be less than the mean instantaneous frequency of the control response (82.5 Hz). The antidromic bursts comprised 20 antidromic spikes (Fig. 10B) and its duration was 255 ms. After the antidromic bursts, the mean frequency of the phasic responses (number of spikes divided by the response duration) significantly decreased (Student's t-test: P < 0.001; Fig. 10C). However, this comparison does not allow concluding if the inhibition produced by antidromic discharge results from the adaptation of the sensory discharge that would have been "naturally" observed after 255 ms of sensory activity. Therefore we have superimposed the two responses with a delay of 255 ms, corresponding to the duration of the antidromic burst (Fig. 10D). If adaptation was the only phenomenon responsible for the decrease of the orthodromic activity, the three overlapping bins (the 3 last bins of the mean control phasic burst and the 3 first of the mean phasic response after the antidromic burst) should be more or less superimposable. This was not the case (Fig. 10, D and E). The mean frequency of the sensory activity after the 75-Hz antidromic burst (1st 3 , 44.4 ± 2.4 Hz) was significantly smaller than the mean frequency of the sensory activity in control (last 3 , 59.9 ± 3.3 Hz; Student's t-test P = 0.0009). This result, drawn from the analysis of averaged instantaneous frequency over several cycles, was confirmed by the analysis instantaneous frequency in single assays. In control conditions, the maximal instantaneous firing frequency of the sensory discharge measured 255 ms after the movement onset was 93 Hz (mean of 12 trials). By contrast, after 255 ms of antidromic stimulation, it was only 55 Hz (mean of 14 trials), which was 38 Hz lower than the expected value if adaptation was the only mechanism involved. Similar results were observed in three of three other experiments. Although the kinetics of the response (frequency against time) during imposed ramp movement was different among the various CBCO neurons studied, the same effect was observed: an antidromic train delivered prior to the movement onset always induced a larger decrease of the sensory activity than an orthodromic train (of the same frequency). Therefore adaptation is not the only mechanism involved in the inhibitory effect produced by antidromic spikes on sensory activity.

View larger version (28K): [in this window] [in a new window]   Fig. 10. Antidromic burst and adaptation of sensory discharge. A: control situation: response of a movement-coding sensory neuron to ramp stretch of the CBCO. The histogram represents the average of 12 responses expressed in number of spikes per second. , instantaneous frequency of 1 response. B: effect of a train of 20 antidromic spikes delivered at the frequency of 75 Hz on the response to the same ramp stretch as in A. Same representation as in A. C: bar diagram of the averaged (n = 12 trials, ) control response and the averaged (n = 14 trials, ) test response obtained after a 75-Hz antidromic burst. D: superimposition of the histograms presented in A and B. The  has been delayed by 267 ms (duration of the antidromic burst) to compare the decrease of the test response with the adaptation of the control response. Note that the last 3 bins of the control response overlap with the 1st 3 bins of the test response. E: bar diagram of the averaged (n = 12 trials, ) control response and the averaged (n = 14 trials, ) test response which overlapped in D. Error bars represent SE.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Antidromic discharges in primary afferents

In both vertebrates and invertebrates, a common physiological mechanism used to regulate the afferent sensory message is the PAD. PADs occurred in many different system and can be induced by sensory inputs (Gossard 1996; Marchand et al. 1997; Newland et al. 1996), by supraspinal structure discharges (Vinay and Clarac 1999), and by the activity of the locomotor networks (El Manira et al. 1991b; Gossard 1996). In many of these systems, antidromic spikes are evoked when the PADs reach the afferent firing threshold. They were first studied in decerebrate cats by Barron and Matthews in the 1930s and identified as antidromic spikes by Toenis (1938) and then by Barron and Matthews (1938). They do not result from temperature changes (Dubuc et al. 1988; Repkin et al. 1976); they are seen during inflammatory processes (Cervero and Laird 1996; Willis 1999) and hypoxia (Petho et al. 1999) and during rhythmic movements such as scratching (Baev and Kostyuk 1982) or walking. During fictive locomotion in cats (Beloozerova and Rossignol 1994; Dubuc et al. 1985, 1988; Duenas et al. 1990; Gossard et al. 1989, 1991) rhythmic antidromic discharges occur in many cutaneous and muscle primary afferents and are phase-locked with various part of the locomotor cycle. In group I afferents, antidromic bursts are locked in phase with the flexor phase of locomotion (Duenas et al. 1990; Gossard et al. 1991) and are only present in the afferents of the flexor and the bifunctionnal muscles (Gossard et al. 1991).

In the crayfish, P. clarkii, recordings from sensory afferents have not been performed yet in freely walking animals; however, spontaneous antidromic discharges are recorded in primary afferents of the CBCO proprioceptor during episodes of fictive locomotion in the in vitro preparation (El Manira et al. 1991b). During fictive locomotion, spontaneous rhythmic bursts of antidromic spikes occur in identified sensory fibers. Within bursts, their instantaneous frequency, which can reach 150 Hz, is, however, irregular and generally ranges between 20 and 100 Hz (unpublished data). The duration of antidromic discharges occurring during fictive locomotion are generally in the range of 1-3 s (Cattaert et al. 1992; El Manira et al. 1991b), and they occur mainly at the end of the depressor activity and the onset of levator activity. However, during fictive locomotion induced by oxotremorine (10⁵ M), a muscarinic agonist of acetylcholine, the period of the locomotor rhythm is very slow (10-40 s) when compared with the values seen during walking in vivo (800-1,200 ms) (Jamon and Clarac 1995). Therefore during locomotion in vivo, the characteristics of antidromic discharges may be very different from the in vitro conditions.

Antidromic burst effects on sensory activity

During fictive locomotion, the sensory discharge of most CBCO afferents (90%) is centrally controlled by PADs, and it is further modulated in the 44% of the afferents in which antidromic bursts occur. In this work, we were able to elicit antidromic bursts in all recorded afferents even though the sensory discharge was only modulated in 82% of them (28/34). Therefore the first question which arise is how many of the CBCO afferents present antidromic bursts during real locomotion, and the second is whether antidromic discharges will affect >44% of the afferents during a faster locomotor rhythm as during in vivo walking.

Apart from the six CBCO afferents that did not respond to antidromic bursts, all the others showed a clear decrease of their orthodromic activity in response to antidromic bursts. However, in the quiescent preparations we used, the antidromic burst effects were heterogeneous within the same types of CBCO afferents. In tonic afferents, the orthodromic discharge can be silenced; antidromic spikes elicited at frequency >20 Hz suppress the orthodromic activity and slow its recovery to control values, and short antidromic bursts rhythmically elicited at short time intervals can suppress the orthodromic activity. In addition antidromic spikes occurring at low frequency can also decrease the sensory discharges. In phasotonic and phasic afferents, antidromic bursts occurring during movements imposed to the CBCO strand can silence the phasic responses to those movements when the antidromic burst frequency reaches a value between 50 and 70 Hz depending on the afferent tested. When the antidromic bursts occur prior to movement onset, short antidromic bursts at high frequency produce a strong decrease of the phasic responses. This inhibitory effect is larger in phasotonic than in phasic afferents. Thus the orthodromic phasic activity of the phasic afferents are less sensitive to antidromic bursts than that of the phasotonic afferents. However, the antidromic burst duration is a limiting factor to produce a decrease of the phasic responses even when the burst frequency is high. In phasotonic afferents, antidromic bursts occurring during or before the movements are also able to decrease the orthodromic activity during the maintained positions of the strand. However, this orthodromic activity is never silenced even for long duration antidromic bursts at high frequency.

In the rat, Lin and Fu (1991) have shown that 36% of the stretch afferents from the gastrocnimius-soleus muscle presented a decreased orthodromic activity when antidromic stimulations were delivered to the nerve. However, among those, two types of afferents were differentiated on the basis of their sensitivity to antidromic bursts. In 52% of the afferents (type I), 10-s antidromic stimulations did not produce an inhibitory effect when delivered at a frequency <300 Hz, while in 47% of them (type II) burst frequency of 50 Hz was enough. In addition, 10-s antidromic busts at 300 Hz produced a decrease of the orthodromic activity and a recovery to control value in <4 s in type I and a cessation of the orthodromic activity an a recovery to control value in 15 s in type II. The silence duration in type II increased with increasing antidromic bursts frequency. In the cat, Gossard et al. (1999) have shown that antidromic bursts were able to decrease or stop the orthodromic activity in cutaneous and muscle afferents, depending on the frequency and the type of afferents. Moreover, as in our experiments, these authors have shown that antidromic bursts induced, in some cases, silent periods and slow recovery of the orthodromic activity. In addition, when antidromic bursts were rhythmically elicited, accumulated decreases of the orthodromic activity were seen in some of the afferents. However, in stretch-muscle afferents (groups I and II), single burst or repetitive ones were able to only increase the orthodromic spike intervals in which they occur. In conclusion, antidromic bursts exert inhibitory effect on sensory activity in primary afferents of crayfish, rat and cat. These inhibitory effects share similar characteristics. 1) The strength of inhibition varies among the sensory afferent population (some sensory neurons being very sensitive to antidromic bursts while others are almost not affected). 2) Marked inhibitory effects may lead to a complete silencing of the sensory activity outlasting the antidromic burst for hundreds of milliseconds. 3) Inhibitory effects of antidromic bursts may accumulate when repeatedly delivered. Such similarities in the characteristics of the inhibition produced by antidromic bursts onto sensory activity in rat, cat, and crayfish may indicate that common mechanisms are involved in these different animal species.

Functional hypotheses during locomotion

In the crayfish during walking, the CBCO provides the locomotor network with two types of information: about the angular aperture of the joint and about the velocity and the direction of the angular movement (Le Ray et al. 1997). In addition, during fictive locomotion, the locomotor network rhythmically modulates the sensory inflow using PADs (Bévengut et al. 1997; Cattaert et al. 1999). During locomotion, PAD amplitude may reach the threshold for spike initiation and antidromic bursts are elicited (Cattaert et al. 2001), which further modify the sensory messages. However, so far only functional hypotheses can be discussed as we do not know yet which types of CBCO afferents can display antidromic bursts in vivo.

In CBCO afferents during fictive locomotion in vitro, antidromic bursts generally occur at the transition between the stance and the swing phase (Bévengut et al. 1997). The role of such antidromic bursts is therefore likely related to the control of MN activity. The motoneuron activity is controlled by network interneurons and by the sensory information they receive. Proprioceptive inputs exert a particularly effective control onto motoneurons because CBCO afferent make monosynaptic connections onto them (El Manira et al. 1991a). However, these monosynaptic connections are always designed to support negative feedback termed resistance reflex in arthropods and stretch reflex in vertebrates (Clarac et al. 2000). During active movements, such negative feedback connections would be inappropriate because they would oppose the movement. The occurrence of powerful presynaptic inhibition and of antidromic bursts in primary afferents during such active movements would contribute to block the resistance reflex and thereby facilitate the movement. On the other hand, during fictive locomotion, the resistance reflex is not only modulated in amplitude but also in sign. Reflex reversal has been observed in crustacea (DiCaprio and Clarac 1981; El Manira et al. 1990; Skorupski and Sillar 1986; Vedel 1982), in insects (Bässler 1986), in the cat (Pearson 1995), and in human (De Serres et al. 1995; Duysens et al. 1990), during active movements such as locomotion. In the crayfish, reversal reflex involves nonspiking interneurons (assistance reflex interneuronsARIN (Le Ray and Cattaert 1997). Therefore two parallel processes occur during reflex reversal: sensory inflow is blocked in some CBCO afferents involved in resistance reflex, while other afferents convey the same type of sensory information to assistance interneurons. This observation is compatible with our observation that bursts of large PADs and antidromic discharges were present in a fraction of the CBCO terminals. We hypothesize that CBCO afferents that do not present large PADs would likely be the one that are involved in the disynaptic assistance reflex pathway, whereas CBCO afferents in which large PADs and antidromic discharges were observed are likely to be involved in resistance the monosynaptic reflex pathway.

Mechanisms of action of the antidromic spikes

One possible physiological role of antidromic discharges might be to create a barrage to the centripetal sensory inflow in specific phases of the locomotor cycle (Dubuc et al. 1988). In crayfish, the collision between orthodromic and antidromic spikes does not seem the case (Bévengut et al. 1997). This conclusion was drawn from the measurement of conduction times of sensory spikes (3-10 ms) in the CBCO sensory nerve. To block all sensory spikes, a collision mechanism should then require a frequency >300 Hz, whereas in active preparations the frequency of antidromic spikes is generally in the range 20-100 Hz. This does not rule out the occurrence of collisions. Near simultaneous occurrence of an orthodromic and an antidromic spike in the vicinity of the spike-initiating zone may induce an undetected collision by the method we used. Such collision would, however, block only a small fraction of the sensory spikes. The situation seems to be different in the cat, where mean frequency of antidromic discharges can reach 275 Hz (Dubuc et al. 1988).

Other postulated mechanisms are those which will modify the sensitivity of the sensory neurons (Bévengut et al. 1997; Gossard et al. 1999; Lin and Fu 1998). A possibility would be that antidromic discharges induce an adaptation of the sensory discharge. During a continuous stimulus many sensory neurons present such a decrease in their activity. This possibility was explored in the case of the CBCO neurons. Two time scales for adaptation need to be considered: a long-term adaptation (over minutes) and a short-term adaptation (over tens of milliseconds). In the absence of antidromic discharges, tonic CBCO neurons generally show very little long-term adaptation if any after their steady-state activity is reached (i.e., after a fixed position is set). By contrast, movement-sensitive CBCO neurons present more pronounced long-term adaptation of their response to movement. However, this adaptation is very slow (after 10 min of sinewave movement, the global response may decrease by 10-20%). In such a case, a rest of 5 min was enough for restoring the initial sensitivity of the neurons. In the experiments presented in this report, series of recordings in control situation alternated with series of recordings in antidromic situation. At the end of each series, we waited 5-10 min rest before recording a new series. Then control series were averaged and antidromic series were averaged independently for each condition. This procedure avoided any long-term adaptation bias on the effect of antidromic discharges. However, during imposed movement, short-term adaptation occurs. For example, after a position is reached an maintained, the continuous discharge of tonic sensory neurons rapidly adapt (Le Ray et al. 1997). Similarly, during a ramp movement, the induced sensory generally adapt from the beginning to the end of the imposed ramp (Fig. 10A). The effect of antidromic discharge is compared with this adaptation process in the following paragraph.

The present results demonstrate that antidromic discharges modify the sensitivity of the sensory neurons. The involved mechanisms may act at the site of spike generation or at the primary transductory zone. For example, at the site of spike generation, antidromic spikes could increase the refractory period following each orthodromic spike by opening voltage-dependent conductances and therefore produce a decrease in the probability that the spike-initiating zone elicits an orthodromic spike. In this hypothesis, no difference would exist between orthodromic and antidromic spikes. This phenomenon would be responsible for the total blocking of the sensory activity during antidromic stimulation delivered at a higher rate as the sensory activity (Fig. 3). However, several lines of evidence indicated that other mechanisms were to be involved. 1) Antidromic bursts induced an inhibition of the sensory activity that outlasted the antidromic stimulation by several seconds (Figs. 3 and 4). 2) The inhibitory effect induced by antidromic bursts increased when antidromic bursts were repeatedly delivered (Fig. 5). 3) Antidromic bursts elicit a more pronounced decrease of sensory activity than adaptation would have produced (Fig. 10). What difference may exist between orthodromic and antidromic spikes except the fact that they are conveyed in different directions? In sensory neurons displaying a tonic activity, simultaneous occurrence of an orthodromic and an antidromic spike in the vicinity of the receptor site could be a possible mechanism by which antidromic spikes produce an inhibition of the sensory activity. In such occurrences, a collision very close to the trigger zone of the sensory neuron would induce a larger afterhyperpolarization than in the absence of collision. Such collisions would not be detected because they occur in the CBCO itself. Such mechanisms would explain the inhibitory effect of antidromic spikes delivered at low frequency. However, this explanation does not hold for antidromic trains delivered in silent phasic neurons prior to imposed movement (Fig. 7). It is also possible that the effect of antidromic trains involve mechanisms acting at the primary transductory zone. Such mechanisms that may exist alone or in conjunction with the previous one could modify the mechano-transducing properties of the sensory neurons by decreasing the receptor potential induced by a given mechanical stimulus. The analysis of the mechanisms by which antidromic spikes modify the orthodromic firing activity will require recordings to be performed from the mechano-sensory dendrites and soma of the sensory neurons.