Control of motor activity in crayfish by the steroid hormone 20-hydroxyecdysone via motoneuron excitability and sensory-motor integration

Arthropods are characterized by a rigid cuticle exoskeleton that is replaced regularly in order to allow growth or metamorphosis (Wigglesworth, 1972). ‘Molt’ (or ‘molting’) typically defines a very short and stereotyped sequence, i.e. the shedding of the old cuticle, also known as ecdysis. However, it also frequently designates the whole series of events that prepare and follow ecdysis, more precisely corresponding to ‘premolt’ and ‘postmolt’, respectively. Molting hormones, named ecdysteroids, trigger and control this crucial developmental period. Indeed, hormonal titers fluctuate and reach a large peak during premolt, declining before ecdysis. These fluctuations, during the rising as well as the falling phases of the peak, are able to orchestrate the whole cascade of events leading not only to the synthesis of a new cuticle (reviewed in Charles, 2010), but also to many other related changes, including changes in behavior. For example, in insects, it has been demonstrated that a precocious ecdysteroid rise is involved in the initiation of wandering behavior before metamorphosis (Riddiford, 1976), whereas the drop of ecdysteroid concentrations after the hormonal peak is generally associated with an arrest of locomotion and the onset of a stereotyped behavioral sequence leading rapidly to ecdysis (reviewed in Truman, 2005; Žitňan et al., 2007). Similarly, in crustaceans, animals become quiescent in late premolt (Lipcius and Herrnkind, 1982; Chang, 1995), and ecdysis depends on a drop in ecdysteroid concentrations (Maissiat and Graf, 1973). However, ecdysteroids may also have transient stimulating effects on behavior during early premolt, as observed for agonistic interactions in lobsters (Bolingbroke and Kass-Simon, 2001).

Numerous studies have emphasized that, in addition to the epidermis, the nervous system is one of the main targets of ecdysteroids, which are able to control neuron physiology at various levels, including neurogenesis and neurodegeneration, dendritic remodeling, synaptic connectivity or efficacy, as well as neuron intrinsic biophysical properties and ionic conductances (reviewed in Weeks and Levine, 1990; Levine et al., 1995; Consoulas et al., 2000; Truman, 2005). In insects, the effects of molting hormones are particularly intense during metamorphosis and have thus been mainly described during this crucial period. By contrast, in crustaceans, in which metamorphosis is absent or limited to the very first postembryonic molts, the effects of ecdysteroids on the nervous system appear more discreet (e.g. Cooper and Ruffner, 1998; Cromarty and Kass-Simon, 1998) and have not been extensively studied.

In our previous studies on our model species, the red swamp crayfish, we analyzed the electrophysiological properties of leg motoneurons (MNs) controlling posture and locomotion, using an in vitro preparation of the nervous chain (Cattaert et al., 1990; Cattaert et al., 2010). In the present study, we were interested in analyzing the effects of the main ecdysteroid, 20-hydroxyecdysone (20E), on this in vitro preparation. However, in order to combine an electrophysiological study with behavioral analysis, we injected the hormone into animals; some of them had their ventral nervous chain dissected in order to analyze nerve activities and MN excitability, whereas the others had their behavior analyzed. The results presented below show that 20E, under our experimental conditions, leads to a decrease of MN excitability and premolt locomotion in the red swamp crayfish.

Experiments were performed on male form I adult red swamp crayfish [Procambarus clarkii (Girard 1832)] weighing 20–30 g. The animals were collected locally, in the Réserve Naturelle de Bruges, after prefectural authorization, and were maintained in isolation for at least 3 weeks before any experiment, in order to avoid interference from past social experience. They were kept at 18–20°C on a 12 h:12 h light:dark cycle, and fed once a week with shrimp pellets and carrots.

The natural occurrence of ecdysis was noted for each animal. Premolt, also known as D-stage in crustaceans (Drach, 1939), typically corresponds to the period extending from apolysis (i.e. separation of epidermis from cuticle) to ecdysis. Postmolt is the period of consolidation of the new cuticle, following ecdysis (A and B stages in crustaceans). In order to avoid disturbing their behavior, we did not manipulate the animals to examine the appendages or swimmerets (Aiken, 1973) to determine precisely the beginning of premolt and the further progression of the molt cycle. However, we considered that, in relation to the size of our animals, premolt extended at least over the 2 weeks before ecdysis and postmolt over the 3 weeks after ecdysis. The day of ecdysis and the day after were considered to be early postmolt. Intermolt was then considered to begin after postmolt (at least 3 weeks after a molt) and to end before premolt (a minimum of 2 weeks before the next molt).

Before any injection or dissection, animals were anesthetized by chilling in ice-cold water for 15–20 min.

The steroid hormone 20E and saline components were purchased from Sigma-Aldrich (St Louis, MO, USA). 20E was dissolved at a concentration of 10 μg μl⁻¹ in a saline solution composed of (in mmol l⁻¹) 195 NaCl, 5 KCl, 13 CaCl₂, 2 MgCl₂ and 3 HEPES, at pH 7.65. Treated animals received a single dose of 5 μg 20E per gram animal fresh mass, injected dorsally between two tergites into the abdominal muscles. Experiments made with lower doses were consistent with the observations of Krishnakumaran and Schneiderman (Krishnakumaran and Schneiderman, 1970), who noted that 20E concentrations higher than 3 μg g⁻¹ fresh mass were necessary for premolt induction. For control conditions, animals were sham injected with 0.5 μl of saline solution per gram fresh mass. After injection, each animal was returned to isolation in its aquarium.

An in vitro preparation of the thoracic nervous system was made [following Sillar and Skorupski (Sillar and Skorupski, 1986) and El Manira et al. (El Manira et al., 1991)] from control animals or from 20E-injected animals, 3 days after injection. After anesthesia, each animal was decapitated. A careful dissection of the ventral nerve cord was made, containing the last three thoracic (T3–T5) and the two first abdominal (A1, A2) ganglia, and was removed with all the nerves of the two proximal segments of the left fifth leg (Fig. 1A). The preparation, pinned dorsal side up on a Sylgard-lined Petri dish, was continuously superfused with oxygenated saline solution (same composition as for injections, see above). The fourth and fifth thoracic ganglia, as well as the first abdominal ganglion, were desheathed to improve the perfusion of the central neurons and to allow for intracellular recordings (Fig. 1B).

The coxo-basipodite chordotonal organ (CBCO), which monitors the movements of the second joint (coxo-basipodite; Fig. 1A), was also dissected out and kept intact. The distal end of its elastic strand was attached to an electromagnetic puller (VT101, Ling Dynamic Systems, Meudon-la-Forêt, France) controlled by a homemade function generator that allowed the application of sine wave movements to the CBCO strand (Fig. 1A) to mimic upward (during stretch) and downward (during release) movements of the leg.

More detailed information on the study of sensory-motor circuit has been already published (El Manira et al., 1991; Le Bon-Jego and Cattaert, 2002), and is described in the  Appendix.

Crayfish were transferred individually in a dark container (10 cm diameter) to the center of an aquatic arena (50×25 cm; Fig. 2A), where they were maintained for 1 min. The container was removed after this period, and each animal was released into the aquatic arena where its exploratory behavior was recorded for 10 min with a video camera (Handycam, DCR-SR78, Sony, Tokyo, Japan). After recording, each animal was returned to its aquarium. The first test was performed before 20E or sham injection, the others at the same time (ca. 10:00 h) during each of the five following days.

We used EthoVision XT8 software (Noldus, Wageningen, The Netherlands) to analyze video records (dark line, Fig. 2B), more particularly by measuring total distance covered and instantaneous velocity during the last 8 min of recording; the first 2 min of recording, during which animals may have unusual reactions after their release from the dark container, were not taken into account.

Results are given as means ± s.e.m. Statistical analyses were performed using Prism (GraphPad Software, La Jolla, CA, USA).

Three days after the injection of 20E in treated animals or of saline in control animals, in vitro motor nerve activity was recorded from isolated preparations of the ventral nerve cord using extracellular electrodes placed close to the motor nerves that control movements of the leg. Fig. 3A shows examples of depressor nerve activity and corresponding mean frequency in a control (Fig. 3Ai) and in a treated preparation (Fig. 3Aii). Although the depressor nerve presented a tonic discharge in each condition, the mean frequency dropped after 20E treatment, both in terms of the number of active units and the unit discharge frequency. Such an analysis was repeated on the other motor nerves from 22 treated preparations and 27 control preparations (Fig. 3B). Very significant differences were confirmed for the tonic activity of the depressor nerve (23.30±3.59 Hz in controls versus 5.35±1.92 Hz in treated preparations; Mann–Whitney P <0.0001) and the promotor nerve (21.73±3.27 Hz in controls versus 12.79±3.39 Hz in treated preparations; Mann–Whitney P<0.05). By contrast, the levator and the remotor nerves did not display significant differences. However, their activities were not increased after 20E treatment (as might be expected in response to the decreases in their antagonist nerves), but rather a slight decrease occurred. Indeed, the pooled total activity of all motor nerves (depressor, levator, remotor and promotor) dropped significantly after 20E treatment (Fig. 3B, all nerves), from 17.86±1.80 Hz in control preparations to 9.19±1.56 Hz in treated preparations (Mann–Whitney, P<0.0001). Interestingly, 20E exerted maximal effects on depressor nerve activity, which is responsible for downward movements of the leg (i.e. rising posture) and for the power stroke phase of walking.

Because the effect of 20E on the activity of the depressor nerve was particularly apparent, we analyzed the electrophysiological properties of depressor MNs with intracellular electrodes (Fig. 4). We measured their firing thresholds and the input resistances (R_in) in control and treated preparations (Fig. 4A) from their responses to different intensities of square current pulses. The current necessary to trigger a depressor MN discharge was lower in control preparations (example in gray trace, Fig. 4Ai) than in treated preparations (example in black trace). This discharge threshold was measured in 10 depressor MNs from 10 control preparations and in 12 depressor MNs from 10 treated preparations. The mean current threshold for spikes of depressor MNs from 20E-treated preparations was 7.42±1.10 nA, whereas it was only 4.2±0.79 nA in depressor MNs from control preparations (Mann–Whitney, P<0.05; Fig. 4Aii). This change in threshold current was not due to a difference in the resting potential of depressor MNs, which showed no significant difference between control and treated preparations (Fig. 4B, Mann–Whitney test, P=0.29)

A difference in MN R_in may help explain this difference in discharge threshold. Injection of a −1 nA square current induced a larger hyperpolarization in depressor MNs of control preparations than in MNs of treated preparations (see examples in Fig. 4Ci). Over all experiments, mean R_in dropped from 8.24±1.93 MΩ in control preparations (N=14) to 2.4±0.44 MΩ in 20E-treated preparations (N=5), which was a significant difference (Mann–Whitney, P<0.05; Fig. 4Cii). Similarly, depressor MN time constants (τ, measured by fitting, with one exponential decay time function, the recovery after −1 nA hyperpolarizing current pulses) were shorter in depressor MNs from treated preparations (black trace, Fig. 4Di) compared with control preparations (gray trace). Differences in depressor MN time constants measured in all experiments were consistent with differences in R_in. The mean time constant significantly decreased after 20E treatment (Mann–Whitney, P<0.01), from 16.13±2.69 ms in control preparations to 2.40±0.44 ms in treated preparations (Fig. 4Dii).

The activity of MNs also depends on the drives they receive from afferent inputs, including sensory inputs. Here we analyzed the effect of 20E on a sensory-motor loop that mediates a resistance reflex. This reflex consists of activation of the depressor nerve in response to an elevation of the leg, and it participates in coordination of locomotion (El Manira et al., 1991). We applied sine wave movements to the sensory receptor (CBCO) that controls vertical position of the leg (Fig. 5A), and recorded the resistance reflex response from the depressor nerve. Extracellular spikes were observed during CBCO release that mimics an elevation of the leg, as previously described (El Manira et al., 1991). The reflex discharge was regular in control preparations (Fig. 5A, top), but weak and irregular in treated preparations (Fig. 5A, below).

In order to quantify these changes over all experiments, we divided cyclic motor activity into several bins (Fig. 5B). The discharge frequency of bins at the peak of the release was significantly higher (see asterisks above the nine bars centered around the discharge peak, between 3.2 and 4.8 s) than the minimum discharge frequency bin recorded during the stretch phase (see gray bar at 1.8 s in Fig. 5B, left). By contrast, in treated preparations, there was no significant difference between mean discharge frequency of stretch and release phases (Wilcoxon test, P=0.11, data not shown). The weakness of the reflex was confirmed by the lack of a significant difference between the peak response and the minimum frequency of the stretch phase (gray bar at 1.8 s, Fig. 5B, right). Similar results were found for the levator motor nerve resistance reflex discharge (data not shown).

The smaller resistance reflex response observed in depressor nerves might be a consequence of a decrease in sensory inputs. Therefore, we analyzed the response of CBCO to sine wave movements in control and in treated preparations (Fig. 6) by examining the discharge frequency and the number of active CBCO units.

CBCO discharge frequency was much higher in control preparations (example in Fig. 6Ai) than in the treated preparations (example in Fig. 6Aii). The decrease of CBCO sensory activity after 20E treatment was consistently observed over 16 treated preparations. The mean CBCO discharge frequency, recorded during 300 s, significantly decreased (Mann–Whitney, P<0.05) from 94.22±7.70 Hz in control preparations to 67.61±10.47 Hz in treated preparations. However, the number of active CBCO units was not significantly different in control preparations (16.80±1.34 units) and in treated preparations (13.36±1.34 units, Fig. 6B, right). Therefore, the decrease of the resistance reflex was not due to silencing of CBCO sensory units.

The resistance reflex can also be observed intracellularly in MNs kept hyperpolarized by injection of a continuous current in order to prevent spiking. Indeed, in a silent MN, application of sine wave movements to the CBCO strand elicits alternately depolarization and hyperpolarization of the membrane potential in response to synaptic inputs from CBCO units (El Manira et al., 1991; Le Ray and Cattaert, 1997). Two examples of intracellular responses of two depressor MNs in control (gray) and treated (black) preparations are presented in Fig. 7A. The amplitude of the depolarization during the release phase of CBCO movement is larger in the control preparation than in the treated preparation. This difference was confirmed (Mann–Whitney, P<0.01) by measuring reflex response amplitudes in intracellularly recorded depressor MNs from 18 control (1.11±0.15 mV) and eight treated (0.33±0.12 mV) preparations (Fig. 7B).

Intracellular recordings allowed us access to synaptic inputs. Depressor MNs exhibit post-synaptic potentials (PSPs), both excitatory (EPSPs) and inhibitory (IPSPs). Synaptic inputs triggered by CBCO units during imposed sine wave movements were analyzed in control and treated preparations (see Materials and methods). Examples of superimposed unitary EPSPs are presented in Fig. 8Ai and 8Aii (left), respectively, for control and treated preparations, together with the corresponding averages from all EPSPs (Fig. 8Ai,ii, right). Note that after the spike in CBCO units, the peak of a monosynaptic EPSP is rapidly reached (black arrows). In addition, various polysynaptic components (gray arrows) combine during the decay phase of monosynaptic EPSPs. However, the shapes of these EPSPs were different in control and treated preparations. Indeed, the very slow decaying phase of control EPSPs (Fig. 8Ai, right) is due to the presence of polysynaptic components (Fig. 8Ai, left). By contrast, in depressor MNs of treated preparations, EPSPs rarely present polysynaptic components (Fig. 8Aii, left) and the average trace of EPSPs presents a faster decay phase (Fig. 8Aii, right).

The mean peak amplitude of these EPSPs was similar in control and treated preparations (Fig. 8B). However, EPSP decay measured 15 ms after peak (see gray arrows in Fig. 8A), which represents the polysynaptic components, significantly decreased (Mann–Whitney, P<0.05) from 0.14±0.02 mV (in 17 control preparations) to 0.07±0.01 mV over all unitary EPSPs (in eight treated preparations) (Fig. 8B). This 50% decrease has important functional consequences by limiting the temporal summation of CBCO EPSPs, and thereby the amplitude of the resistance reflex. Note that the mean number of CBCO units contacting a single depressor MN was not significantly different between control (5.12±0.73) and treated preparations (3.5±0.76).

IPSPs recorded on depressor MNs were also analyzed (Fig. 8C). No IPSPs were found in depressor MNs recorded from the eight treated preparations (Table 1). In contrast, eight IPSPs were recorded from depressor MNs in 17 control preparations. The mean peak amplitude of these IPSPs was −0.17±0.03 mV and the amplitude measured 15 ms after IPSP peak was −0.05±0.03 mV.

These results demonstrate that 20E reduces sensory motor integration. However, the effects on sensory inputs are less important than those observed on polysynaptic components via interneuron silencing. Nevertheless, these effects were also weaker than those observed above on intrinsic properties of MNs. All these effects lead to MN decreased activity, which might lead to decrease of locomotion. Therefore, we analyzed the functional consequences of 20E treatment on crayfish locomotion.

After sham or 20E injection, the total distance walked and the mean instantaneous velocity were measured every day for 5 days. Values were expressed as a relative change from day 0 value (i.e. before treatment). The relative distance walked of sham controls remained unchanged during all test days (Kruskal–Wallis test, n.s., N=9; Fig. 9A, left), indicating that crayfish did not become accustomed to the arena, and that the injection procedure itself had no effect on crayfish behavior. In contrast, 20E treatment significantly changed the total distance walked (two-way ANOVA, N=9, P<0.05; Fig. 9A, left). At the third day after 20E treatment, 94% of crayfish decreased their global activity (Tukey's post hoc test, P<0.05). This difference was even more significant (P<0.01) on the fifth day. Similarly, 20E treatment significantly changed the relative mean instantaneous velocity (two-way ANOVA, P<0.01; Fig. 9A, right).

In order to further study the effect of 20E on walking velocity, we compared the distribution of instantaneous velocities before and 5 days after 20E treatment. These two distributions were significantly different (Chi-square test for trend, P<0.0001). The median of the distribution shifted from 1.55 cm s⁻¹ before treatment (Fig. 9B, left) to 0.87 cm s⁻¹ 5 days after 20E injection (Fig. 9B, right). The shape of the distribution clearly shows that before treatment crayfish walked similarly at a variety of speeds between 0.5 and 2.5 cm s⁻¹, whereas after treatment the histogram presents a peak of walking speed at 0.5 cm s⁻¹, and animals very rarely walked faster than 1.5 cm s⁻¹.

We also analyzed the evolution of the total distance walked during natural molts (Fig. 10A). Animals were regularly observed (at least twice a week, over ca. 3 months) during the course of their natural molting cycle. Among the 11 animals studied, nine ecdysed and were recorded during the whole intermolt–molt sequence. Observations were averaged after being synchronized by the day of ecdysis; premolt was considered to correspond to the 2 weeks before ecdysis (gathering observations made from 6 to 1 days before ecdysis for the first week, and from 12 to 7 days before ecdysis for the second week). During premolt and postmolt phases, the total distance walked decreased (Fig. 10B) in all but one animal (Fig. 10C). As a result, the mean total distance walked recorded during premolt phases was significantly different (Wilcoxon matched-pairs signed rank test, P<0.05), switching from 341.6±65.9 to 199.4±69.0 cm during molt (42% decrease) (Fig. 10C). All these data show that premolt, either natural or induced by 20E, is characterized by a decrease in locomotion, which is consistent with the decrease of MN excitability observed in vitro.

Injection of a sufficient amount of 20E into crayfish is known to induce a rapid premolt, but this accelerated development is frequently fatal (Krishnakumaran and Schneiderman, 1970). Indeed, our treated animals generally failed to ecdyse and died ca. 7 days after treatment under our experimental conditions. However, neurons studied in our isolated preparations 3 days after injection (or even 5 days after, data not shown) had no observable characteristics of dying cells; they maintained a constant membrane potential for at least 3h in vitro at a voltage similar to that of control preparations. Moreover, dying neurons, if any, would be characterized by an increase of spike frequency due to uncontrolled depolarization, whereas MNs of our treated preparations were almost silent due to a maintained hyperpolarization; thus, the changes described in this study after 20E injection were clearly not due to neuronal death.

More precisely, we observed a modification of neuronal plasticity in crayfish locomotor networks following 20E treatment, mainly affecting motor outputs and, to a lesser extent, sensory inputs to MNs. Indeed, hormone injections induced changes in depressor MN excitability because of the alteration of cell membrane properties, more particularly the decrease of input resistance. It is important to note that this decrease in input resistance was not due to a change in membrane resting potential, which remained constant before and after 20E treatment under our experimental conditions. Previous studies in insects have already shown that input resistance may strongly differ after metamorphosis in leg MNs that persist from larvae to adults, suggesting a possible role for developmental hormones (Duch and Levine, 2000; Rose and Levine, 2000). To our knowledge, however, our observations on crayfish represent the first report showing that 20E is able to control this parameter in the absence of any structural change of the locomotor apparatus. Yet it must be noted that 20E may also control cell excitability of some other neurons, in particular insect neurosecretory cells, without changes in input resistance (Hewes and Truman, 1994).

Together with the decrease of MN excitability, 20E treatment also modified sensory inputs, as observed with the strong alteration of MN synaptic potentials in our model. Indeed, the depolarisation of depressor MNs induced by sensory inputs from the CBCO was also decreased after 20E treatment. A hormonally mediated weakening of synaptic connections between the sensory and motor neurons has already been observed during insect metamorphosis, involving MN dendrites, breaking contacts with sensory neurons (Weeks et al., 1997). However, our results indicate that monosynaptic sensory–motor connections were not modified in our model after 20E treatment. Indeed, we observed that the decrease of reflex response after 20E treatment was mainly related to a change in the decay phase of EPSPs (Fig. 9), indicating a lower activity of excitatory interneurons. Moreover, our data show that the decrease of reflex response was not due to an increase of the activity of inhibitory interneurons, because they became silent. In our model, sensory-motor activity involves at least three types of interneurons (Le Ray et al., 1997; Le Bon-Jego and Cattaert, 2002; Le Bon-Jego et al., 2004). Our results show that the recruitment of two of them, i.e. the excitatory and inhibitory interneurons of resistance reflex, was decreased or suppressed by 20E treatment (our experimental conditions did not allow us to analyze the effects of 20E on the third type, the assistance reflex interneurons). The question remains open whether 20E effects involves specific targets in both excitatory and inhibitory interneurons or are rather the consequence of a similar decrease in interneuron excitability as observed in MNs.

In summary, 20E controls both MN excitability and sensory-motor integration in crayfish leg networks. However, between these two types of cooperative effects, the drop in input resistance of MNs (70%) seems to be mostly responsible for the reduction of motor activity.

Neuron excitability is generally thought to involve endogenous conductance, as a result of ion channels, i.e. membrane proteins. Numerous ion channels have been identified or described in vertebrates and invertebrates (reviewed in Armstrong and Hille, 1998). Though we have not yet identified the types of conductance involved in the membrane properties analyzed in our experiments in crayfish, it seems likely that ecdysteroids may act, either directly or indirectly, on such membrane protein(s). Preliminarily, however, it remains to establish whether the effects of 20E were direct on leg MNs. Of course, ecdysteroids are able to produce direct changes in neurons, as shown by cell culture approaches in insects (Levine and Weeks, 1996). In crayfish, direct and fast effects of 20E have already been described on neuromuscular junctions by local applications of 20E (Cooper and Ruffner, 1998). However, our results were obtained only after 20E injections into animals and our attempts to reproduce them by applying 20E directly on our in vitro preparation did not yield any noticeable change, neither in network activity nor in MN properties, even after 1 day of application (data not shown). This suggests that the 20E effects observed in this study were possibly indirect, either involving another organ or, more probably, some relay in superior nervous centers absent from our in vitro preparation, likely the brain. Indeed, such a role for the brain as the primary site of 20E action has already been documented in the control of insect wandering behavior (Dominick and Truman, 1986; Miller and Levine, 2006).

Another possible explanation could be that the effects of 20E observed in our study might nevertheless be direct on MNs, but after a longer cascade, extending for more than a day. Indeed, though steroid hormones, including ecdysteroids, may act through short-term pathways, interfering directly with membrane receptors that they may share with neurotransmitters [e.g. GABA (Okada et al., 1998); dopamine (Srivastava et al., 2005)], they more frequently have long-term effects, involving nuclear receptors that control gene expression (reviewed in Spindler, 1997; Spindler et al., 2009). The 20E effects observed in the present study (recorded 3 days after hormone injection) were obviously long-term effects, suggesting some possible control(s) on gene expression. Ecdysteroids are indeed able to trigger a cascade of events, involving a complex interplay of nuclear receptors, that allow coordinated and time-delayed expression of numerous proteins, as illustrated by the control of dopa-decarboxylase gene expression (reviewed in Hiruma and Riddiford, 2009), cuticular protein secretion (Charles, 2010) or neuroendocrine cascade regulating ecdysis (reviewed in Žitňan et al., 2007). Importantly, similar long-term effects of ecdysteroids on identified ion channels have been described in insects (see Grünewald and Levine, 1998; Börner et al., 2006; Garrison and Witten, 2010).

Finally, it is also very probable that the response of crayfish neural networks to 20E might be a complex summation of rapid and direct effects, such as those described on neuromuscular junctions (Cooper and Ruffner, 1998), together with long-term and possibly indirect effects, as suggested by our results.

Our behavioral observations on the effect of natural or induced premolt in crayfish are consistent with previous studies on insects (reviewed in Truman, 2005) and crustaceans (Lipcius and Herrnkind, 1982; Chang, 1995) reporting a reduction of locomotor behavior during premolt, particularly at the approach of ecdysis. This decreased motor activity is generally related to the weakness of the new exoskeleton on which muscle attachments are progressively transferred in late premolt. However, such a decrease of locomotion is more surprising when considering that, at the onset of premolt, animals have to find an appropriate place to molt quietly. This is particularly true in our crayfish model, which is a very aggressive and omnivorous species; if animals are maintained in high-density groups in the laboratory, every individual undergoing a molt is rapidly devoured by its congeners (authors' unpublished observations). It is thus obvious that, in natural conditions, animals should leave the plants where they generally feed, sometimes in high density, to find an isolated place. If so, this behavior would be comparable to the wandering behavior described before insect metamorphosis. Insect wandering behavior is in part controlled by a small rise in ecdysteroid titers (Riddiford, 1976), whereas the arrest of locomotion accompanying ecdysis behavior is dependent on the drop in molting hormone (Truman, 2005). Thus, one should consider the possibility that our injections of high concentrations of 20E into crayfish only highlighted the inhibition of locomotion by high and subsequently declining 20E concentrations, but not a possible excitation by a discrete hormonal rise. However, our attempts to verify this possibility using injections of lower quantities of 20E only gave intermediate inhibitory results, but no excitatory response in our test conditions (data not shown). It remains possible, however, that our behavioral test (isolated animals in an open field) was not adequate to reveal a behavior that may depend on more complex environmental interactions (e.g. grouped animals in presence of shelters), or that single 20E injections did not sufficiently mimic the subtle fluctuations of 20E occurring during the normal induction of premolt. An alternate possibility is that locomotion might be inhibited by any concentration of ecdysteroids in crayfish leg networks, but is activated at the level of tail-flip networks, as this kind of escape swimming varies in premolt animals (Tamm and Cobb, 1978; Cromarty and Kass-Simon, 1998; authors' unpublished observations). Indeed, ecdysteroids may also have opposite effects in different MNs (Weeks and Truman, 1985; Cromarty and Kass-Simon, 1998).

In conclusion, our experiments demonstrate that 20E is able to durably reduce the excitability of crayfish leg MNs and, more generally, the activity of the whole postural and locomotor network. From the electrophysiological MN properties after 20E treatment, it appears likely that this network would be less capable to respond to descending commands during the molting period than during intermolt.

We thank Prof. D. H. Edwards for insightful discussions, critical reading and correcting the English of the manuscript. We also thank Stéphane Builles and Denis Cheyrou from the Réserve Naturelle de Bruges for providing crayfish.

The depressor MNs receive sensory input from the CBCO. This organ consists of an elastic strand of connective tissue that is attached proximally to the dorsal edge of the coxopodite and distally to an apodeme at the proximal-dorsal edge of the basipodite. Embedded within this strand are approximately 40 sensory neurons that project to the ipsilateral neuropile (the region in which neurons form their synaptic contacts) and make monosynaptic and polysynaptic connections with the depressor MNs (El Manira et al., 1991; Le Bon-Jego and Cattaert, 2002). Half of the CBCO sensory neurons are activated when the CBCO strand is stretched, and the other half are activated when the band is released (El Manira et al., 1991). Thus, this proprioceptive organ monitors movements of the limb in the vertical plane. In the in vitro preparation, a pin holds the proximal end of the CBCO while a mechanical puller imposes movements to its distal end (Fig. 1A).

To ensure that the CBCO was not damaged during the dissection, we only used preparations with robust sensory neuron activity in response to movements imposed on the CBCO strand. In order not to damage the CBCO during the experiment, stretch movements were applied starting from the most released position of the CBCO strand, and the total amplitude of the movement was one-third of the released CBCO strand length (1–1.8 mm). The movement control voltage traces were visualized on an oscilloscope and stored on a computer.

Extracellular recordings from the motor nerves innervating the four leg muscles and from the sensory nerve of the CBCO were made using stainless steel pin electrodes contacting the nerves and insulated with Vaseline (VWR BDH Prolabo, Strasbourg, France). The differential extracellular signals were amplified (×5000–10,000) and filtered (high-pass 30 Hz, low-pass 30 kHz, 50 Hz notch filter) using Grass Technologies AC preamplifiers (Warwick, RI, USA). The bath solution was grounded using a small silver plate that was previously chlorided using chlorine bleach. Stimulation of nerves was carried out with a programmable pulse generator (Master-8, A.M.P.I., Jerusalem, Israel) and a stimulus isolation unit (Isoflex, A.M.P.I.). Intracellular recordings from depressor MNs (Fig. 1B) were made using glass micropipettes (Clark Electromedical Instruments, Reading, UK) filled with 3 mol l⁻¹ KCl (resistance 10–30 MΩ). The intracellular electrodes were connected to an Axoclamp 2B amplifier (Axon Instruments, Foster City, CA, USA) used in current-clamp mode. In Crustacea, the somata of MNs lie outside the neuropile (Fig. 1B) and are linked to the arbor of the neuron via a thin process, and so do not participate in the electrical activity of the neuron. For these reasons, intracellular recordings were made from the main neurite, where EPSPs could be recorded (Fig. 1B).

Depressor MNs were identified following the procedure used in Hill and Cattaert (Hill and Cattaert, 2008). The resting membrane potential of MNs was usually in the range of −80 to −65 mV. Stability of resting membrane potential over a long period of time (>30 min) was used as a criterion for evaluation of cell health during recordings. In crustacean MNs, soma and neurites do not actively convey spikes. Therefore, spike amplitude was generally small (<20 mV) at the recording site. Data were digitized and stored on a computer hard disk through an appropriate interface (Power1401) and software (Spike2) from Cambridge Electronic Design (Cambridge, UK).

Data were analyzed using Spike2 analysis software. Spikes recorded from the CBCO nerve were identified according to their waveform based on a template matching protocol (wavemark). Templates were built automatically with a duration corresponding to the mean duration of sensory spikes (ca. 1.5 ms in duration). The sampling rate for CBCO nerve recording was set to 20 kHz, which resulted in templates containing ca. 30 points. The procedure used two criteria to identify a spike: (1) more than 90% of the points should be in the confidence limits of the template; and (2) the maximum amplitude change for a match was less than 5%. This procedure was applied off-line. After the completion of this protocol, each identified CBCO unit (spike shape) was assigned an arbitrary number. Subsequently, a spike-triggered average was performed for each CBCO unit, allowing us to observe in a given MN the occurrence of any postsynaptic events related to this unit.

 
• 20E

  20-hydroxyecdysone

 
• CBCO

  coxo-basipodite chordotonal organ

 
• EPSP

  excitatory post-synaptic potential

 
• I

  depolarizing current

 
• IPSP

  inhibitory post-synaptic potential

 
• MN

  motoneuron

 
• PSP

  post-synaptic potential

 
• R_in

  input resistance

 
• τ

  time constant