Physiology and Structure of Three New Uropod Proprioceptors in the Crayfish Procambarus Clarkii

The uropods are the terminal appendages of the tailfan of crustaceans and lie on either side of the medial telson. Each uropod is biramous, consisting of a basal protopodite (arising from the abdomen) bearing a blade-like endopodite and exopodite. The uropods are used in many types of behaviour, including equilibrium reflexes, such as righting and steering, as well as in a variety of postures and active behaviours, including defence and escape swimming (Larimer and Kennedy, 1969; Bowerman and Larimer, 1974a,b; Takahata et al. 1981; Newland and Neil, 1987; Newland, 1989).

The neuronal control of the uropods has been studied in detail, since it represents a relatively simple motor system that is, however, complex enough to produce asymmetrical as well as symmetrical postures. Larimer and Kennedy (1969) and Bowerman and Larimer (1974a,b) showed that the uropods participate in a variety of behaviours that can be driven by the stimulation of command fibres in the connectives. Some of these behaviours include the reflexive steering and righting responses that are controlled by descending interneurones with inputs from the vestibular organs, the statocysts (Takahata and Hisada, 1982). These interneurones make direct connections with uropod motor neurones and, in addition, make connections, with the uropod motor neurones via non-spiking interneurones (Nagayama et al. 1984); Takahata and Hisada, 1986). Other studies have shown that non-spiking and spiking interneurones in the sixth abdominal ganglion act as premotor and sensory integration sites in uropod motor pathways, including some for behaviour other than the steering and righting reflexes (Reichert et al. 1982; Sigvart et al. 1982; Nagayama and Hisada, 1985).

Experiments in which the nerve roots of the sixth abdominal ganglion were cut have shown that the neuronal control systems for steering and righting behaviour do not require proprioceptive feedback from the uropods (Larimer and Kennedy, 1969; Takahata et al. 1985). Perhaps for this reason investigations of uropod proprioceptor structure and function (in a range of behaviours) have largely been neglected. A single type of stretch receptor has been reported in the anomuran decapods. This consists of an elastic strand and accessory muscle complex spanning the telson-uropod articulation at the base of the uropods. The primary sensory neurones have centrally located somata in the terminal abdominal ganglion. In Emerita analoga the neurones signal stretch of the telson-uropod joint with graded electrotonic potentials (Paul, 1972), while in Galathea strigosa they fire spikes, as well as producing electronic potentials (Maitland et al. 1982). In Homarus gammarus the receptor fires spikes when stretched (Laverack, 1989). The treading-water behaviour of Emerita analoga was shown by Paul (1976) to be dependent on feedback from this receptor; in Galathea strigosa the receptor mediates efferent reflex activity of two motor neurones and of the receptor muscle itself (Maitland et al. 1982). One further undefined phasic-tonic receptor that detects movement of the telson is apparently located in the region of the crayfish anus (Barth, 1964).

In this report we show that each of the three joints of the crayfish uropod possesses previously undescribed proprioceptors, and that these sense organs are capable of signalling position, velocity and direction of movement of each uropod segment.

The uropods were initially surveyed by immersing partially dissected preparations in van Harreveld’s saline to which 6 drops of 1 % acidified Methylene Blue (Pantin, 1964) were added per 100 ml. After 4–7 h (at 5°C) the axons, somata and scolopidia in the stretch receptor organs were stained. The organs were further dissected as required. However, for receptors that had bipolar neurones within the peripheral nerve trunks (see Results), the best stains developed after 10–24h in a solution of 3 drops of Methylene Blue per 100ml of saline. All preparations were illuminated with fibre optic lighting and drawn using a Wild M5 binocular microscope with camera lucida. Thirty-four preparations were stained.

Physiological studies were carried out on 29 additional preparations. An isolated abdomen-tailfan preparation was pinned out on a wax-bottomed dish and covered with van Harreveld’s saline. Individual uropod segments were moved in the plane of their articulation by a solenoid connected to a function generator which delivered sinusoidal, triangular or trapezoidal waveforms. The solenoid probe terminated in a sharpened pin which was inserted through the cuticle of the segment under study; this provided a direct-drive, artefact-free movement. The remainder of the uropod was immobilized with pins. Angles of normal joint movement were measured visually against a small protractor fixed behind the moving segment.

Extracellular nerve recordings were made with suction or oil-hook electrodes (Wilkins and Wolfe, 1974) leading to conventional a.c. preamplifiers, while uropod movements were monitored with a selenium photocell. Both signals were stored on a TEACR-410 tape recorder for subsequent photography and analysis. The nerve roots were always cut proximal to the receptor to eliminate efferent signals and, in most cases, cut distal to the receptor to eliminate non-receptor distal afferent signals.

Stretch receptor organs were found in close proximity to each of the three major joints of the uropod: the abdomen-protopodite (AP) joint, the protopodite-en-dopodite (PEn) joint and the protopodite-exopodite (PEx) joint (Fig. 1). All three organs appear to be chordotonal organs (COs) because, in at least one Methylene Blue stain of each, scolopidia were observed at the dendritic endings of the sensory neurones. The organs are named according to the joint that they span: APCO, PEnCO and PExCO.

PEnCO is a strand CO, in which an elastic strand of connective tissue containing sensory neurones stretches across the joint. APCO and PExCO each consists of a web of connective tissue extending from a major nerve root to the hypodermis of the appropriate uropod segment. This latter type is unusual since bipolar sensory neurones occur along its nerve root as well as in the web itself. All the COs are innervated by nerve roots 2 or 3 of the sixth abdominal ganglion, as described below.

This organ consists of two branched webs of hyaline connective tissue which emerge from the sheath of root 3, proximal to the entry of the nerve into the protopodite (Figs 1 and 2A). The webs spread and attach to the hypodermis on the medio-ventral aspect of the protopodite (arrows, Fig. 2A). A minor web branch often attaches to root 2 and/or the productor muscle (pro in Fig. 2A).

A maximum of 19 bipolar neurones per organ (mean±s.D. = 13±3, N=16) were stained, of which 5±2 occurred on root 3, just beneath the nerve sheath. These latter neurones may not necessarily be homologous with those of the web, since, in at least one preparation, a bipolar root cell (asterisk in Fig. 2B) had a distal process (presumably a dendrite) that bifurcated, sending one branch out to a root 3 motor nerve and the other distally past the APCO. It was clear, however, that the neurones within the web possessed scolopidia and formed the main chordotonal organ innervation.

This webbed organ arises from root 2 and sends a major, tough hyaline strand of elastic tissue to the ventro-medial corner of the exopodite, adjacent to the insertion of the exopodite productor muscle (Figs 1, 3A). Other minor branches of elastic tissue extend more laterally to the ventral hypodermis.

A maximum of eight sensory neurones per organ (mean±s.E.=5±l) was observed in nine preparations. Of these 3±1 were root cells. An additional cluster of 9–16 very small neurones was often associated with the distal attachment of PExCO (arrowhead, Fig. 3B). These cells sent dendrites along the hypodermis to the region of arthrodial membrane at the protopodite-exopodite joint, but did not appear to possess scolopidia. Therefore, they do not appear to be joint stretch receptors.

This strand chordotonal organ consists of an elastic strand of connective tissue which spans the protopodite-endopodite joint and receives innervation from a small branch of root 3 in the protopodite (see Fig. 1). The strand arises from the arthrodial membrane on the distal rim of the protopodite and connects to the proximal, medial corner of the endopodite (Fig. 4A). The sensory neurones, somata and scolopidia lie either in the web-like expansion of tissue, where the sensory nerve joins the elastic strand, or on the strand itself (Fig. 4B).

In the nine organs examined, the maximum number of bipolar neurones recorded was 12±2 (mean±s.D.); usually these included 10 large neurones, plus two very small ones embedded in the medial (distal) end of the band (arrowhead in Fig. 4B). Two additional axons (asterisk in Fig. 4B) were often seen passing through the PEnCO near the lateral (proximal) edge of the web, and continuing on to the protopodite hypodermis.

Since the abdominal-protopodite joint can be moved in both the vertical and horizontal planes (see Fig. 5G), sensitivity to each plane of movement was tested separately. Whole-nerve responses to sinusoidal movement of the protopodite at 10,1.0 and 0.1 Hz in the vertical plane are shown in Fig. 5A-C, while responses to horizontal movements at the same frequencies are shown in Fig. 5D-F. Some units responded to movements in both planes, while others were sensitive to either vertical or horizontal movements, but not to both. The following two examples demonstrate units sensitive to both planes of movement. At 10Hz the largest units (dotted) fired on both vertical extension and horizontal opening (upward movements of the lower trace, Fig. 5A,D). At 1 Hz a medium unit (dotted) fired on vertical flexion (Fig. 5B) and on horizontal opening and closing (Fig. 5E).

The rows of dots (rasters) in Fig. 6 represent firing patterns of neurones during horizontal and vertical protopodite movement. Only medium-and large-sized units were directionally sensitive in the APCO, as illustrated by the upper two rasters in Fig. 6 (two neurones that fire on closing only) and in the large spikes of Fig. 5D (two neurones that fire both on opening and on closing). Most of the smaller units did not show directional sensitivity (Fig. 6, lower rasters, and small spikes in Fig. 5C,F), but fired only in a certain range of angles, regardless of the direction of imposed movement. This is characteristic of phasic-tonic units, and is discussed later. The plots of Fig. 6 show responses at 1 and 10 Hz.

When tested for velocity-sensitivity with horizontal ramps, many units showed a peak response at 20° s⁻¹, although three (which were large units) had peak sensitivities at 200°s⁻¹ (Fig. 7A). The APCO showed very little response to the lowest-velocity horizontal ramp (0.2°s⁻¹). Vertical movement velocity receptors responded at all velocities tested. Eight units showed increased firing with increased velocity, up to the maximum tested (200°s⁻¹, Fig. 7B). These units showed little response at 2°s⁻¹ and the greatest increase between 20 and 200°s⁻¹ Four additional units showed low velocity-sensitivity (Fig. 7C), with peaks at 0.2, 2..0 or 20°s⁻¹ and no response above 2.0 or 20°s⁻¹. Therefore, range fractionation occurred over the full range of velocities tested for vertical movement, but only for the higher two velocities of horizontal movements.

Adaptation of phasic units to continuous sinusoidal movement in the horizontal plane was tested at 1 and 10 Hz and is plotted as the number of spikes per cycle versus cycle number (Fig. 8). No adaptation occurred over 13 cycles at 1 Hz, while only slight adaptation occurred at 10Hz over 16cycles.

Position-sensitive units were tested by giving ramp-and-hold stimuli in vertical and horizontal directions. Only one or two vertical-position-sensitive units were ever recorded in any preparation. These were phasic-tonic units (e.g. smallest spike, Fig. 9) which fired a burst of spikes with each flexion or extension ramp (upwards on bottom trace), as well as firing more strongly at each succeeding position approaching full extension. An almost linear relationship between spike frequency and position was demonstrated (Fig. 10A, different preparation) as the protopodite was moved from full flexion to full extension and back. Horizontal position-sensitive units showed range fractionation over the entire 14–16° range of movement (Fig. 10B-F). Three of nine units analyzed were phasic-tonic and directionally sensitive (two fired during opening ramps while one fired during closing). The remaining six showed no movement sensitivity. These purely tonic horizontal position units showed marked hysteresis, the nature of which depended on the particular neurone: those with maximum firing frequencies at closed angles (Fig. 10B,C) fired most strongly at any particular angle if that angle was approached from a more closed angle. The opposite was true for neurones with maximum firing frequencies at open angles. At some angles neurones either failed to fire or fired at up to 1Hz depending on the direction of approach (Fig. 10E).

Although this organ had the fewest sensory neurones, it nevertheless showed velocity and position sensitivity. Ramp-and-hold 12° movements revealed: (a) that the largest units have high velocity thresholds (Fig. HA); (b) that the medium units are position-sensitive and fire tonically (Fig. 11C); and (c) that the smallest units are direction-and velocity-sensitive (Fig. 11B,C). A detailed analysis of single-unit directional sensitivity is shown in Fig. 12A, in which each raster represents the firing pattern of a single unit during trapezoidal 20° movement at 20°s⁻¹. The three smallest units (lower three rasters) respond to closing (Rl), opening (R2) and both directions (R3), while the larger tonic units (upper rasters) fire mostly in the static open position.

The velocity sensitivity of six single units is plotted in Fig. 12B for velocities from 2 to 2000°s^−l. Three high-frequency units are plotted together in Fig. 12Bi, while the remaining three low-frequency units are plotted together in Fig. 12Bii. The peak responses for all units except one occur at 200–2000°s⁻¹ (approximately 1–10Hz of sinusoidal movement). The single exception had a peak response at 20°s⁻¹ (Fig. 12Bii) and fired at a low frequency.

Adaptation to continuous sinusoidal stimulation was analyzed in two neurones: a large unit sensitive to high velocity, and a small unit that responded to closing only (Fig. 12C). The small (a) unit adapted to 66% of its initial firing rate within 15 cycles at 1 Hz, while at 10 Hz it fired weakly and was completely adapted by the seventh cycle. The high-threshold unit (b) did not fire at 1Hz, and at 10 Hz it dropped from an initial firing rate of 4 spikes cycle⁻¹ to a’ plateau of 2 spikes cycle⁻¹ by the fifth cycle, with occasional spike failure thereafter. It continued to fire at this low rate for many cycles.

PExCO neurones showed maximum position sensitivity at or nearly at the fully opened position of the exopodite (i.e. 14, 17 and 20°), and were silent or fired weakly over the rest of the range (Fig. 12D). These units also showed a strong hysteresis in response, determined by the direction from which any given setting was approached. Two of the units fired most strongly during the opening sequence of position settings (the opposite hysteresis pattern was noted for APCO neurones).

The 12 neurones of the strand chordotonal organ, PEnCO, monitor position, velocity, direction and probably acceleration of endopodite movement. Whole nerve recordings during sinusoidal movements at 1.0 and 0.1 Hz (Fig. 13B,C) showed that the strongest firing occurred while the endopodite passed through the open end of its displacement range (between approximately 9 and 18°). During stimulation at 10 Hz, the PEnCO fired most strongly during closing movements (Fig. 13A). This record also shows that a very sudden closing movement superimposed on the sinusoidal pattern evoked a brief discharge of at least one large, high-threshold unit.

When stimulated at 1 Hz, the four smallest neurones (Fig. 14, lower rasters) are phasic and directionally sensitive (primarily to closing), while the larger neurons do not show directionality and represent a combination of phasic and phasic-tonic cells which fire during both opening and closing (Fig. 14).

Phasic units were analyzed for velocity-sensitivity with ramp movements varying from 2.4 to 720°s⁻¹ (Fig. 15A, arbitrarily divided into two plots for clarity). As with the other chordotonal organs, the majority of phasic units responded most strongly to high velocities and showed little response at low velocities. All units except three had peak responses at 240-720° s⁻¹. The three exceptions (Fig. 15Ai) showed peak responses at 24°s⁻¹ (equivalent to 1Hz triangular movement). The units showing peaks at the highest velocities were large, high-threshold neurones, and may represent acceleration receptors. They gave no responses below 240° s⁻¹.

Thus, the PEnCO has rather limited range fractionation and is primarily a high-velocity stretch receptor.

The adaptation to continuous sinusoidal stimulation varied among the three phasic units analysed (Fig. 15B). Curve a was from a medium-sized unit which adapted to 78 % of its initial firing rate within 8 cycles at a stimulation frequency of 1Hz. At the same frequency a small unit, curve b, adapted to 64% of its initial firing rate over the same period. However, the rate of adaption was faster for the medium unit, which had a generally higher firing rate. Curve c was from a large, high-threshold unit stimulated at 30 Hz. Its response dropped rapidly from 3 to 2 spikes cycle⁻¹ and remained steady over many cycles in this and other experiments.

The position-sensitive units showed range fractionation over most of the range of endopodite displacement (Fig. 15C). For five units, which fired tonically at the positions tested, peak responses occurred at 4°, 4–6°, 5–8° and 9–12°. Hysteresis in the response curves was marked for two of the units (a and e): the sequence of steps towards the fully opened position gave stronger responses than the closing sequence, as noted for position-sensitive tonic units in PExCO. The tonic firing frequencies of the other three neurones were low (usually less than 5 Hz).

Three previously undescribed proprioceptors in the tailfan of P. clarkii are shown to monitor position, movement velocity and direction of movement of the three joints in the uropods of the crayfish. Although the strand chordotonal organ, PEnCO, is typical of those found in crustacean legs, antennae and maxillipeds (Mill, 1976), the other two organs are highly unusual because of their attachment to nerve roots of the terminal abdominal ganglion, and because of the presence of sensory somata in these roots. The presence of bipolar sensory neurone somata in peripheral nerves is unusual in Crustacea. The centrally associated somata of a spiking stretch receptor in the telson of Galathea strigosa occur at the base of root 1 of the terminal ganglion (Maitland et al. 1982). A spiking stretch receptor in the abdominal nerve cord of the crayfish (Kennedy et al. 1966) is sensitive to cord distension during extreme extension of the abdomen and uropods. However, it is not known whether the receptor soma is in the cord or in a ganglion. The best-documented case is the presence of several cuticular mechanoreceptor somata in nerve roots 1 and 2 of crayfish abdominal ganglia (Pabst and Kennedy, 1967). These bipolar and tripolar neurones occur within a few millimetres of each ganglion and apparently are not associated with any chordotonal organs. The long, branched dendrites have broad receptive fields in the soft ventral cuticle of the abdomen, where the dendritic terminals bear distal expansions. The uropod root cells of the present study generally appear to innervate the CO connective tissue webs, but in one case the distal process of a root cell of APCO extended past the CO and bifurcated into a motor branch of root 3. Since this cell was clearly not part of the CO, it, and possibly others in the uropods, could have been a cuticular mechanoreceptor (although bipolar somata have not been reported previously in root 3 of the abdominal ganglia). In insect peripheral nerves, a few neurosecretory and mechanosensory somata are found in two dipteran genera and one phasmid species (Finlayson and Osbourne, 1968; Finlayson, 1976).

The results from this study and those on the telson stretch receptor (Paul, 1972; Maitland et al. 1982; Laverack, 1989) indicate that the uropods are almost as richly innervated by proprioceptors as are the other major appendages of crustaceans, which have one or two enteroreceptors (chordotonal organs, myochordotonal organs or non-spiking coxal receptors) per joint, as well as apodeme tension receptors in certain muscles. It is clear from previous deafferentation studies, however, that the uropod proprioceptors are not involved in the equilibrium righting and steering reflexes mediated by statocysts and leg proprioceptors (Hisada and Neil, 1985). They must, therefore, be involved in proprioceptive feedback control of the uropods during voluntary behaviour. Bowerman and Larimer (1974a,b) described a number of command interneurones in the crayfish which activated the uropods as part of specific behavioural drives. They demonstrated 14 command units that produced phasic behaviour such as swimming, escape, forward walking and backward walking. At least 20 other units activated tonic postures involving the uropods. These behaviours included abdominal extension, turning, abdominal flexion, general promotion, bilateral cheliped lifting plus abdominal extension (as in the defence posture) and bilateral cheliped lifting plus abdominal flexion. It may be postulated that these descending drives provide the neuronal frames of reference in which uropod proprioceptive feedback operates.

The physiology of the chordotonal organs indicates that, although they can pionitor position and movement parameters of all uropod joints, there are constraints on how effectively such parameters are measured. This is shown in Table 1, which summarizes the properties of the three COs.

Position monitoring is limited to open angles of the exopodite and extended angles of the protopodite, but includes the full range of endopodite positions. The maximum discharge of the position-sensitive units is obtained if open positions have been reached by opening movements in both the exopodite and the endopodite. Thus, there is enhanced sensitivity to flared positions of the crayfish tailfan. It is of interest to note that many of the command-driven postures reported by Bowerman and Larimer (1974a,b) involved flaring the exopodites and endopodites to open positions. These movements are often accompanied by abdominal extension, as seen in the lateral merus defence posture. Forward walking is another behaviour that is accompanied by uropod flaring and abdominal extension and, in this case, the degree of exopodite-endopodite opening is often variable and finely tuned, depending upon the walking rate and substratum conditions (L. H. Field personal observation). We have noted that, in general, crayfish in an aquarium tend to maintain the uropods in an open or semi-open position most of the time, whether resting or moving. However, in a habitat with water currents, the uropods would become very important in maintaining a hydrodynamically streamlined profile. Newland et al. (1988b) and Maude and Williams (1983) have studied orientation behaviour of the Norwegian lobster Nephrops norvegicus and eight species of freshwater crayfish, respectively. Although the behaviour of these crustaceans differs in some respects (crayfish exhibit both upstream and downstream orientation, while lobsters only orient downstream), it is clear that in both genera the uropods are deployed in a flared, partially flexed posture which decreases hydrodynamic drag from either direction. All the above observations fit well with the tonic position sensitivities found for the uropod COs.

Table 1 indicates that sensitivity to movement of the exopodite and endopodite joints, but not of the protopodite joint, is restricted. Thus, the joint that controls the entire uropod attitude relative to the abdomen is monitored through four orders of magnitude of velocity in the vertical plane, but the exopodites and endopodites, which adjust the degrees of tailfan flare, are primarily monitored at high velocities. In addition, it should be noted that the large phasic units of all three COs are capable of monitoring continuous cyclic uropod movement at 10–30Hz with little adaptation. Two important points arise from these observations. First, the utilisation of the uropods in the vertical plane of movement is important in behaviour such as water-current orientation (Newland, 1988b) and one would expect proprioceptive feedback to monitor perturbations in any set posture, particularly in the dynamic resistance to water flow of the above example. Second, the rapid tail flips in escape and swimming behaviour could be effectively monitored by phasic units in all three uropod joints, because the frequency responses for adaptation exceed measured tail-flip frequencies [e.g. Nephrops (2.6±0.1 Hz, Newland et al. 1988a) and Procambarus (7–14 Hz, from data in Wine and Krasne, 1972)]. In addition, position proprioceptive feedback from the COs could enhance the rapid opening movements of the uropods which occur during the powerful flexion stroke of the tail flip. Resistance reflexes might utilise negative feedback to prevent excessive opening of the tailfan during abdominal flexion. Similarly, during re-extension of the abdomen following the power-stroke of the swimming cycle, the uropods are held in a closed position to minimise drag (Webb, 1979). The properties of the chordotonal organs are such that they could detect drag-induced opening of the uropods towards the end of extension and generate appropriate resistance reflexes to streamline posture. The above ideas could provide guidelines for investigations of central projections and reflex pathways mediated by uropod chordotonal organs.

PLN was supported by a fellowship from the Japan Society for the Promotion of Science.