Control of abdominal extension in the freely moving intact crayfish Cherax destructor: II. Activity of the superficial extensor motor neurones

An important aspect of motor control is load compensation. How does an animal adjust its motor output to generate the same movement despite changing environmental load? The neuromuscular system that controls postural movements of the crayfish abdomen has attracted interest as a system well-suited for the study of load compensation. The motor neurones and superficial muscles that control the movement of each abdominal joint are accessible to experimental investigation and have been well studied (Kennedy and Takeda, 1965a,b; Kennedy et al., 1966).

The crayfish abdomen consists of six articulated segments, and the position of the whole abdomen can be considered as the sum of the individual joint positions. Postural movements at each joint are effected by the dorsal superficial extensor muscles and the ventral superficial flexor muscles. The superficial extensor muscle and superficial flexor muscle groups are innervated by a set of motor neurones in each hemisegment. Each set contains five excitatory motor neurones and one peripheral inhibitor numbered in order of increasing fibre diameter (Kennedy and Takeda, 1965b; Fields et al., 1967). Each motor neurone possesses its own characteristic conduction velocity, level of spontaneous activity and discharge pattern. Each has a unique source of central activation and pattern of muscle innervation, and the different-sized junctional potentials produced by each motor neurone vary in their effectiveness to facilitate and summate to activate muscle contraction (Kennedy and Takeda, 1965b; Evoy et al., 1967; Kennedy et al., 1966; Drummond and Macmillan, 1998). In the postural neuromuscular system of the abdomen, the myotome is displaced with respect to the neurotome (Hughes and Wiersma, 1960; Fields et al., 1967). The superficial extensor motor neurones (SEMNs) originating from abdominal ganglion 3, for example, travel in the second nerve to innervate the superficial extensor muscle in the fourth abdominal segment (A4), and this superficial extensor muscle attaches to the anterior edge of A5 to draw it forward and produce extension (Fields et al., 1967).

A characteristic feature of the crayfish abdominal posture system is the observed reciprocity both between the motor neurones innervating the antagonistic superficial extensor muscle and superficial flexor muscle groups and between the excitatory and inhibitory motor neurones innervating a single muscle group (Kennedy et al., 1966). During an extension movement, the extensor excitors and the flexor inhibitor are centrally activated while the antagonistic motor neurones, the flexor excitors and the extensor inhibitor, are centrally inhibited. The reciprocity is primarily attributed to connections at the premotor neurone level (Sokolove and Tatton, 1975; Miall and Larimer, 1982), although inhibitory and excitatory cross connections between some of the larger efferents also contribute to the observed reciprocity (Evoy et al., 1967; Tatton and Sokolove, 1975; Toga and Hisada, 1993).

Closely associated with the superficial extensor muscle of each abdominal segment is a tonic muscle receptor organ (MRO). Each MRO consists of a sensory neurone with its dendrites embedded in a thin receptor muscle. The sensory neurone is excited by an increase in receptor muscle tension (Wiersma et al., 1953) and is involved in two postural reflexes. First, the tonic sensory neurone activates the inhibitory thick accessory neurone, which synapses onto the sensory neurone and inhibits receptor discharge (Eckert, 1961). Studies suggest that during active extension the accessory neurone is inhibited (Sokolove, 1973; Sokolove and Tatton, 1975). Second, the tonic sensory neurone reflexly excites a single excitatory motor neurone, SEMN2. It was hypothesised that if the joint bridged by the receptor muscle was flexed, the sensory neurone would respond and excite SEMN2 so that the superficial extensor muscle would contract until the receptor muscle unloaded, thus turning off this load-compensating feedback loop (Fields, 1966; Fields et al., 1967). SEMN4, a motor neurone shared by the receptor muscle and superficial extensor muscle, is capable of adjusting the tension of the receptor muscle (Fields and Kennedy, 1965; Fields et al., 1967).

Fields et al. (1967) proposed an elegant model in which the MRO could compensate for load throughout the extension of the crayfish abdomen. During unloaded extension, the receptor muscle would contract along with the superficial extensor muscle at the same rate, and the sensory neurone would remain silent. If, however, a load was encountered that resisted the extension, the receptor muscle spanning the lagging joint would develop tension faster than the superficial extensor muscle could unload it, and the sensory neurone would fire. The sensory neurone would then increase the motor output to the superficial extensor muscle via SEMN2 to ensure that the working muscle kept pace with the centrally determined tension development in the receptor muscle.

This model was supported by recordings from the MRO in the intact, freely behaving crayfish Procambarus clarkii (Sokolove, 1973), although it is unclear whether the sensory neurone–SEMN2 reflex is the only load-compensating mechanism operating during the abdominal extension. When the abdomen was physically blocked from reaching full extension in these experiments, several large SEMNs increased their firing rates along with the sensory neurone. As the sensory neurone reflexly excites only the small SEMN2 (Fields and Kennedy, 1965; Fields, 1966; Fields et al., 1967; Drummond and Macmillan, 1998), it was suggested that higher centres may be detecting the sensory neurone signal and increasing the drive onto some of the SEMNs (Sokolove, 1973), a logical proposition given that the sensory neurones bifurcate in their ganglion of entry and send a process to the brain (Bastiani and Mulloney, 1988). In a separate experiment, however, Page (1978) also identified an increase in the activity of some excitatory SEMNs in response to load, but in the absence of sensory neurone activity. The extensions in those experiments were evoked by stimulating extension-producing interneurones (for a review, see Larimer, 1988). Page (1978) suggested that a load-compensating mechanism not involving the MRO must also operate in some circumstances.

The experiments described above were conducted in the crayfish P. clarkii. Recent experiments on the Australian crayfish Cherax destructor suggested that the sensory neurone–SEMN2 reflex does contribute to load compensation in that species (McCarthy and Macmillan, 1995). Subsequent recordings from the MRO in intact animals (McCarthy and Macmillan, 1999), however, demonstrate that this could only occur very early in the extension movement as the sensory neurone is not active for most of the extension. Here, we report an examination of the activity patterns of some of the SEMNs in C. destructor during postural extensions of the abdomen to test whether load compensation occurs in the absence of sensory neurone input.

Details of the methods of selecting crayfish (Cherax destructor Clark) for the experiments and recording from the dorsal nerve during postural extension of the abdomen were identical to those reported by McCarthy and Macmillan (1999). At the completion of each behavioural experiment, however, the animal was immobilised and killed by decapitation, and an additional experiment was conducted to confirm the identity of the neurones recorded.

The abdomen was severed from the cephalothorax with at least the last two thoracic ganglia still attached to the abdominal nerve cord to increase the level of spontaneous activity in the abdominal motor centres (Cattaert et al., 1992; Burdohan and Larimer, 1995). The abdomen was pinned dorsal side up and immersed in crayfish saline (composition in mmol l⁻¹: 205 NaCl, 5.4 KCl, 13.5 CaCl₂, 2.6 MgCl₂, 10 Tris, 5 maleic acid, pH 7.4 at 17–18°C). A small piece of cuticle was removed from above the MRO using a dental drill, and a modified oil–hook electrode (Wilkens and Wolfe, 1974), mounted on a micromanipulator, was attached to the dorsal nerve at a point distal to the first electrode. Although a flexion motor program was dominant, there were periods when the SEMNs fired spontaneously in most animals. Furthermore, the motor neurones under investigation could all be activated by stroking regions of the cuticle and tailfan with a small brush. The action potentials were recorded at both electrodes, and the relative axonal conduction velocities of the individual nerve fibres were calculated. At the conclusion of this experiment, a post mortem analysis was conducted as described in McCarthy and Macmillan (1999), and in no case was visual evidence found for damage to the dorsal nerve or its branches.

The overall firing rate of a neurone during an extension, whether blocked or unblocked, was measured by calculating the mean firing rate over 2 s from the start of the extension. Comparisons of mean rates were made using a Student’s t-test (all data sets Shapiro–Wilk tested for normality). Comparisons of neuronal firing rates at the start and end of the extension between animals and between the treatments of loading and blocking of the abdomen were made by counting the firing rate of the neurone under examination for the 200 ms at the start of the extension movement and for the first 200 ms upon reaching full extension or encountering the block. A count was made for at least four different extensions for each animal and each treatment. Comparisons between animals were analysed using analysis of variance (ANOVA), and the firing rates before and after treatment were compared using a repeated-measures ANOVA for paired data. Data were processed using the statistical analysis package Prophet (sponsored by National Center for Research Resources, NIH, Bethesda, MD, USA; ©1997 BBN Systems and Technologies).

The electrode used in these experiments permitted the action potential of a single neurone to be recorded at a consistent size and shape throughout a single experiment. Between experiments, however, the spike amplitude of a neurone differed relative to that of other neurones, presumably because of electrode placement. As a result, relative spike amplitude alone was an inadequate criterion for discriminating between the efferent units when recording en passant. Each neurone was confidently identified in these experiments by measuring its relative conduction velocity during the post-behaviour experiment (Sokolove and Tatton, 1975). Since the attachment of the recording electrode remained unchanged, the neurones continued to be recorded in the same manner as when the animal was fully intact, and the identity of each could be unambiguously established.

The sensory neurone of the tonic MRO was easily identified by its tonic activity when the abdomen was flexed (Fig. 1A) and was confirmed in the post-behaviour experiment by the direction of propagation of its action potential. SEMN6 had the greatest conduction velocity of the postural efferent neurones, and its conduction velocity was similar to that of the tonic sensory neurone but opposite in direction. SEMN5, the peripheral inhibitor, had the second fastest conduction velocity of the postural efferents. SEMNs 1–4 were difficult to identify because their spikes were always relatively small in the recording, and details of their firing patterns could never be followed during active extension because of the high firing rate of SEMN6. As a result, only SEMN6, SEMN5 and the sensory neurone could be examined in detail. These neurones fired the largest action potentials in almost all the recordings, and their firing patterns could be followed accurately. In most experiments, they could be readily distinguished from one another, and the data presented here are based only on recordings where the identity of the neurones is unambiguous.

SEMN6 typically fired the largest action potentials in the recording and always discharged when the abdomen actively extended. It was the first excitatory SEMN to fire during most platform-drop-generated extensions and typically fired at a very high rate early in the extension, declining to a lower rate as the extension proceeded. After the abdomen had achieved full extension, SEMN6 continued to fire at a relatively constant rate for long periods (Fig. 1Ai,B). This contrasted with its activity during some ‘self-generated’ extensions that the animal produced while remaining on the platform. During these extensions, SEMN6 was active only when the abdomen actively extended and ceased firing when the abdomen became stationary in an extended position (Fig. 1Aii). In two animals, SEMN6 was monitored as the abdominal extension was evoked, and the water was then drained from the tank using a siphon (Fig. 2). As the water level was lowered around the suspended crayfish, SEMN6, which had started to adapt to full extension, increased its firing rate. The sensory neurone was not active when the abdomen was in this position, demonstrating that another receptor is detecting the load and increasing drive onto SEMN6.

To investigate the nature of sensory input on the activity of SEMN6 in more detail, we blocked the movement of a specific segment of the abdomen using a blocking rod. This method limits the contractions of the muscles in only one segment and was used because we had found previously that placing weights on a segment can load more anterior segments as well (McCarthy and Macmillan, 1999). A comparison was made of the mean firing rate in the 200 ms before extension and the 200 ms after completion of four extensions in four different animals before and after blockage of the posterior segment at 50 % of full extension. We detected no difference in the initial firing rates of SEMN6 in this group of animals (F_1,15=1.92, P=0.18), but the difference between the initial rate and the rate at the end of the extension was significant when the movement was blocked at 50 % extension (F_1,30=33.94, P=0.001). The mean firing rate of SEMN6 was always greater during extension when the joint posterior to the segment from which the recording was being made was blocked at approximately 50 % of the extension movement (Fig. 3C). This increase in SEMN6 firing rate was consistent for the three segments examined (A2, N=5; A3, N=2; A4, N=3) and occurred without any MRO activity (Fig. 3A) in almost all experiments. Examples of the firing patterns of SEMN6 in specific animals are presented in Fig. 3B. The results demonstrate that a receptor other than the MRO is detecting the load and increasing drive onto SEMN6 to compensate for the load.

A similar pattern of SEMN6 firing was seen when the joint being examined was allowed to extend normally and a neighbouring joint was blocked. For example, the firing rate of SEMN6 in A4, which produces an extension at the A4–A5 joint, increased when the A3–A4 joint was blocked (Fig. 4) in three of four experiments. This finding demonstrates that the receptor detecting the block is increasing drive onto the SEMN6 of more than one segment.

SEMN5, the peripheral inhibitor to the superficial extensor muscle, was generally a large unit in the recording and was silent when the animal was stationary with its abdomen fully flexed. When a postural extension of the abdomen was evoked by a platform-drop, SEMN5 was commonly the first SEMN to fire, and it continued firing throughout the extension (Figs 1B, 5A) even though it is considered to fire reciprocally with the excitatory SEMNs. In some animals, SEMN5 fired at a higher rate early in the extension, while in others it fired consistently throughout. Although SEMN5 was always active during extensions evoked by a platform-drop, this was not typical of all extensions; when the animal extended its abdomen while remaining on the platform, SEMN5 often fired at a very low rate (Fig. 5B). This suggests that there are differences in the central mechanisms generating the motor output for platform-drop and ‘self-generated’ extensions. Fig. 5C shows SEMN5 activity in the same animal as it actively pushes its abdomen down on the platform. This movement is characterised by a large flexion of the A2–A3 joint and provides a reliable and easy identification of SEMN5.

In two of three experiments in which SEMN5 activity was monitored in A2 as the A2–A3 joint was blocked from fully extending, its mean firing rate decreased during the extension [animal 1, N=10, 100 % extension (22.6±3.1 Hz) versus 0 % extension (17.0±1.7 Hz), P=0.0023; animal 2, N=6, 100 % extension (29.7±4.3 Hz) versus 50 % extension (28.7±5.0 Hz), P=0.7171; animal 3, N=6, 100 % extension (48.6±5.3 Hz) versus 50 % extension (41.7±4.0 Hz), P=0.0043].

The high firing rate of SEMN6 in Cherax destructor during extension reinforces its description as the ‘power’ neurone (Drummond and Macmillan, 1998). SEMN6 was always active whenever the crayfish extended its abdomen, and during extensions evoked by a platform-drop it continued firing even after the abdomen had reached its fully extended position. This differs from platform-drop extensions in Procambarus clarkii, where SEMN6 ceased firing when the abdomen reached full extension (Sokolove, 1973). SEMN6 is capable of rapid tension development in the muscle (Kennedy et al., 1966; Drummond and Macmillan, 1998), and its high firing rate would contribute significantly to tension development in the superficial extensor muscle during extension. One disadvantage of the dominance of this motor neurone in the recording is that the activity of the medium-sized and small SEMNs could not be seen clearly. This study does not, therefore, shed light on whether SEMN4 is active during the extensions, nor does it provide evidence that sensory neurone activity before the extension movement reflexly excites SEMN2.

The finding that SEMN6 increased its firing rate when the joint it extends was physically blocked from fully extending, in the absence of sensory neurone activity, is a clear indication that a receptor other than the MRO is mediating load compensation. This is not consistent with the prediction that higher centres may be detecting the MRO input and compensating accordingly (Sokolove, 1973) and supports the suggestion that a receptor other than the MRO is the primary element involved in load compensation in the crayfish abdomen (Page, 1978). The increased SEMN6 activity in segments either side of the joint being blocked demonstrates that the load-compensating output is intersegmental. This is a surprising result given that these segments were unhindered and fully extended.

Both Sokolove (1973) and Page (1978) noted that several excitatory SEMNs increased their firing rates when the abdomen of P. clarkii was blocked/loaded, and a general drive onto the extensor system may also be occurring in C. destructor. This idea gains limited support from the finding in the present study that the peripheral inhibitor to the extensor muscle, SEMN5, decreased its firing rate in two of three animals when the abdomen was blocked.

One of the more likely receptors to be mediating load compensation is the cord stretch receptor. In C. destructor, these extension-sensitive proprioceptors are located within the ventral nerve cord (Savati and Macmillan, 1992) and are active over the entire range of abdominal extension, which corresponds to an increase in nerve cord length of approximately 40 % in P. clarkii (Grobstein, 1973a). Excitation of the cord stretch receptors through stretch of the nerve cord excites the extensor system and suppresses the flexors (Kennedy et al., 1966; Grobstein, 1973b). Recent experiments on the reflex pathways of the cord stretch receptor in C. destructor support the suggestion that they could be involved in this type of load compensation (Drummond and Macmillan, 1997).

Ventral mechanoreceptors (Pabst and Kennedy, 1967) and tactile afferents (Fields and Kennedy, 1965; Fields, 1966; Hausknecht, 1996) also influence postural motor neurone output. These receptor systems may play a role in load compensation, although reflex pathways to achieve this have not been investigated. Macmillan and Dando (1972) described tension receptors associated with the apodemes of both the extensor and flexor muscles in the walking legs of brachyurans, and it is possible that there are receptors of this type associated with other crustacean muscles. While a receptor of this type appears to be ideal for mediating load compensation, none has been described in association with the superficial extensor muscle.

The finding that load compensation in SEMN6 is mediated by a receptor other than the MRO raises the question of the role of the MRO in natural behaviour. The extensions examined in these experiments were evoked by lowering a platform from below the suspended crayfish. During this type of extension in C. destructor, the sensory neurone was active during full flexion and fell silent as the extension movement commenced, so it could contribute to the extension at this point (McCarthy and Macmillan, 1995, 1999). Platform-drop extensions are triggered primarily by the excitation of receptors of the walking legs when the substratum is removed (Page, 1981) and, in nature, the response serves to maintain an animal’s equilibrium as it falls through the water column. It is important to remember that other types of extension are possible. There were occasions when the animal produced its own extensions while remaining on the platform. In these cases, SEMN6 ceased firing when the abdomen was fully extended and the extension was maintained through the firing of smaller SEMNs. It is possible that the MRO plays a more significant role in some of the other types of extension.

A load-compensating role analogous to that described for the MRO was proposed for the myochordotonal organ in the meropodite-carpopodite (M-C) joint of the walking legs of the crab (Evoy and Cohen, 1971). The receptor muscle of the myochordotonal organ is co-innervated with the main flexor muscle of the M-C joint to make it sensitive to imposed opposition to an ongoing movement. The ablation of these organs failed to eliminate load-compensatory responses; the firing rate of the motor neurones increased to the same level as for intact animals upon the addition of a load (Fourtner and Evoy, 1973). The authors concluded that the function of these organs is not to mediate load compensation, but to determine the end point of the flexion stroke. In C. destructor at least, the abdominal MRO may be signalling the achievement of full flexion in a similar way (McCarthy and Macmillan, 1999).

It is somewhat surprising that SEMN5 was consistently active throughout the extension, given that it is classified as an extension ‘antagonist’ and hyperpolarises the muscle (Kennedy et al., 1966). Weak central reciprocity between SEMN5 and the excitatory SEMNs is, however, also a feature of platform-drop extensions in intact P. clarkii (Sokolove, 1973). This contrasts with the strong reciprocity observed during extensions generated by the stimulation of extension-producing interneurones (Evoy and Kennedy, 1967; Page, 1978). These observations provide further support for the suggestion that postural movements are generated by a ‘constellation’ of synaptically interacting premotor interneurones (Larimer, 1988). They also raise questions as to what role SEMN5 plays during extension. SEMN5 innervates only approximately 30 % of the superficial extensor muscle and its innervation predominates in the ‘slower’ muscle fibres (Fields and Kennedy, 1965; Fields, 1966; Fields et al., 1967; Hausknecht, 1996). As these slow muscle fibres contribute more to the maintenance of abdominal position rather than to active extension (Kennedy et al., 1966), one role for SEMN5 may be to nullify the excitatory input to these muscle fibres during active extension.