Innervation of the Anterior and Posterior Levator Muscles of the Fifth Leg of the Crab Carcinus Maenas

Coxal and basal muscles provide the power for walking and swimming of decapod crustaceans (Ayers & Davis, 1977; Hoyle & Burrows, 1973). We are particularly interested in the basi-ischiopodite levator muscles of crabs because of their involvement in an additional motor function, limb autotomy (McVean, 1973; McVean & Findlay, 1976; Moffett, 1975). Fredericq (1892) demonstrated that the large anterior levator (AL) provides the force that fractures the limb at the preformed fracture plane in the basi-ischiopodite. The two radically different functions performed by this muscle, limb elevation and limb fracture, might be regulated by precise control over a wide range of force development. Another possibility is that other muscles could assist or prevent limb fracture. Interest has centred on the posterior levator (PL) because its contraction can alter the AL tendon angle (Moffett, 1975 ; McVean & Findlay, 1976). A third possibility is that the different heads of AL, which pull the tendon in different directions, are differentially employed in locomotion and autotomy, thus determining how force is directed on the tendon insertion. The cuticular stress detector 1 (CSD₁: Clarac & Wales, 1970) monitors stress in the cuticle distal to the AL tendon insertion, and is thought to play a role in regulating the motor patterns either to prevent accidental autotomy or to promote limb fracture upon injury (Moffett, 1975; Clarac, 1976; Findlay, 1978). Our goal in the work reported here was to acquire a more detailed knowledge of the innervation of the two levator muscles so that the many unresolved questions concerning the neuromuscular mechanism of autotomy can be approached.

Bévengut, Simmers & Clarac (1983) recently described the coxal and basi-ischial muscles and their innervation in the fifth leg of the crab Carcinus maenas, and we have adopted some of their terminology. However, they did not study the distribution of motoneurones among the thoracic heads of AL, nor did they deal with the innervation of the small coxal head of the anterior levator, (ALc), which is present only in the first and fifth pereiopods.

We have included the coxal head innervation in this study not because it is crucial to production of autotomy (which occurs perfectly well in the second, third and fourth pereiopods which lack this head) but because most of the investigations of autotomy have utilized recordings from fifth leg nerves or muscles. In this pair of legs, the proximity of ALc to the fibres of PL could result in ALc cross-talk in the PL myograms. The presence of ALc is acknowledged in some descriptions of the levator muscles and not others (McVean, 1973, 1974; McVean & Findlay, 1976). An early erroneous description of the innervation of ALc relative to that of PL (McVean, 1973) has further blurred what we consider to be the important distinction between these two coxal muscles.

Green shore crabs (Carcinus maenas) were obtained from Northeast Marine Environmental Institute, Monument Beach, MA. For physiological recordings, we removed the carapace and viscera and cannulated the sternal artery to perfuse the thoracic ganglia with oxygenated saline (in mmol I⁻¹): NaCl, 500; KC1, 12; MgCl₂, 20; CaCl₂, 12; buffered with Tris maleate to pH 7·2 (DiCaprio & Clarac, 1981).

Nerves from three crabs were used for nerve cross-sections. Fixation was as described by Peracchia & Mittler (1972) with two modifications: (1) both the primary fixative (5 % glutaraldehyde in 0·1 moll⁻¹ sodium cacodylate) and the rins buffer contained 0·35 mol I⁻¹ sucrose and 11 mmol I⁻¹ CaCl₂, and (2) before OsO₄ fixation the nerves were individually pre-embedded in a drop of 2 % agar to maintain orientation during subsequent processing (Ridgway & Chestnut, 1984). After ethanol dehydration and overnight infiltration with propylene oxide:resin (1:1), specimens were embedded in Epon-Araldite. Semithin (1 fim) sections from the proximal ends of the nerves were stained with Stevenel’s Blue (Ridgway, 1986). Ultrathin sections from the same region were stained with 2 % aqueous uranyl acetate followed by lead citrate (Reynolds, 1963). Montages were taken at 1000×, printed at 3000× and analysed with an Apple II image analysis system (Bioquant II software).

For cobalt and nickel backfilling, a 0·1–0·3 mol I⁻¹ solution of either CoCI₂ or NiCl₂ was used. When two nerves were backfilled, the ends of the nerves were placed in separate wells, one filled with NiCb and the other with CoCl₂. Specimens were backfilled for 12–19 h at 4°C, rinsed with fresh saline, developed with rubeanic acid, dehydrated, and cleared in methyl salicylate. Cobalt-filled cells were yellow, nickel-filled cells were dark blue, and cells with axons in both nerves were brick red (Quicke & Brace, 1979).

Polyethylene suction electrodes were used for recording from and stimulating nerves. The smallest possible length of nerve was picked up in the electrode tip to reduce the likelihood of time shifts in motoneurone recruitment during electrical stimulation. Signals were amplified, displayed and recorded by standard methods.

Muscle fibre recordings were made using glass capillary electrodes filled with 2·5 mol I⁻¹ potassium acetate. Resistances ranged from 10 to 40 MΩ. Electrodes were mounted so as to dangle from a chlorided silver wire to minimize damage to the fibres and minimize artifacts attributable to muscle contraction. 185 AL fibre penetrations were made in 53 animals, with 102 recordings from 25 animals yielding data on the postsynaptic potential (PSP) categories recruited by nerve stimulation. 25 PL fibre penetrations were made in nine animals, of which 12 penetrations were combined with nerve stimulation to give information on PSPs.

The levator muscles and their innervation in the fifth leg of Carcinus are shown in Fig. 1. As described by Bévengut et al. (1983), the three thoracic heads (the two Ientrai bundles 7 and 8 and the more dorsal bundle 9) are innervated by nerve trunks arising from the anterior thoracico-coxal root (ATh-Cx). The ventral bundles are both innervated by the more proximal levator nerve bundle (ALvN) and the dorsal head receives its innervation via a nerve bundle (ALdN) that arises more distally (Fig. 1C). Additional details on the nerves that branch from ATh-Cx are given by Bévengut et al. (1983).

The coxal head of the anterior levator, ALc (or bundle 10: Bévengut et al. 1983), consists of 11 or fewer muscle fibres (Schmiege & Moffett, 1985). It receives its innervation from a small nerve (ALcN) that extends from the body of the dorsal head of the muscle, ALd (Fig. IB).

The two PL heads (bundles 11 and 12: Bévengut et al. 1983; PPLM and RPLM: McVean & Findlay, 1976) are innervated by the most distal branch of the posterior thoracic bundle that also innervates the remotors and depressor muscle (posterior thoracicocoxal root, P Th-Cx: Bévengut et al. 1983). A link between the two thoracic roots is provided by an anastomosis (ANAS: Fig. 1C) which includes an axon of the common inhibitor (described below). The PL nerve (PLN) parallels ALcN, but it runs more deeply and passes under the ALc fibres to reach PL (Fig. IB). Even more ventral in this region of the thorax is a branch of the depressor nerve (DcN) that innervates coxal depressor fibres (Fig. IB).

The nerves innervating ALv, ALd, ALc, PL and a central anastomosis that connects the anterior and posterior thoracico-coxal trunks were sectioned for electron microscopy and counts of axon numbers were compared with counts made on adjacent 1 μm sections stained for light microscopy. This procedure allowed us to compare the resolution of axons at the light and electron microscopic level. Consistent results were obtained from the three crabs; data for the comparison of light and electron micrographs from a single crab are given in Table 1. The axons had diameters ranging from 0·5 to 50μm. We were able to count all axons with a diameter above 3 μm in thick (lμm) plastic sections observed and photographed using a light microscope. An example of light microscopic resolution of ALdN (Fig. 2A) shows at least 16 unambiguous axons. When the region included in the box in Fig. 2A was viewed under electron microscopy (Fig. 2B), four smaller (<3μm) axons that could not be detected in the light micrograph became apparent. We could detect eight of the nine small axons in ALcN at the light microscope level, due to the heavy investment of loose glial wrappings which surround each axon (Fig. 2C). Electrophysiological recordings indicated that at least six of the nine neurones in ALcN are motoneurones (see below).

Nine cell bodies stained repeatedly in backfills of ALdN (Fig. 3) and as many as 12 cells were stained in some preparations. We typically found 5–8 somata on the dorsal surface of the ganglion, two relatively small somata midway through the ganglion, and a single, large soma located ventrally near the midline. The contralateral position of the cell body of this last neurone was most clearly demonstrated in bilateral backfills, in which the initial segments of the homologous cells crossed at the midline (Fig. 3). Neither unilateral backfills nor bilateral backfills employing differential staining indicated dye coupling of the contralateral-midline neurones, the common inhibitor neurones (Moffett & Yox, 1986). The major dendrites of these cells were ipsilateral to their axons and overlapped only slightly the distinctive horseshoe-shaped dendritic projections formed by other AL neurones.

In backfills of ALvN, 6–7 cell bodies stained repeatedly (Fig. 3). Five somata on the dorsal surface of the ganglion and the common inhibitor soma were stained brick red when ALvN and ALdN were filled with cobalt and nickel, respectively, indicating that these cells are shared by the two heads. The two midlevel cells and at least one of the dorsal cells consistently stained dark blue, indicating that their projections are restricted to ALd. One large soma located within the horseshoe shaped dendritic projections always stained yellow, indicating that its axonal projections are restricted to ALv. These details are summarized on the right side of Fig. 3.

We found it difficult to backfill PLN, possibly due to its length. The two cells illustrated (Fig. 3, left side) were stained in two ganglion preparations and bilaterally in another preparation. Physiological evidence suggests that the large midline neurone also innervates PL but we never filled it via this nerve. Backfills of ALcN were not obtained because this nerve does not branch directly from ATh-Cx but instead projects from the dorsal head of the muscle. Attempts to fill ALcN in situ were unsuccessful, possibly due to the length or small diameter of the axons.

Backfills of the anastomosis (ANAS) between ATh-Cx and PTh-Cx were undertaken because of physiological evidence (given below) that one of the PL motoneurones crosses in this junction. Backfills of ANAS consistently revealed three cell bodies in the ganglion (Fig. 3, left side); one was a large, ventral midline cell. By pairing backfills of ANAS with backfills of ALvN or ALdN on the same side, we showed that the same contralaterally positioned midline cell was stained via both the levator nerves and the anastomosis. This neurone, the common inhibitor, also has projections in nerves to the basi-ischial depressor, the coxal promotor and anterior and posterior remotor muscles (Moffett & Yox, 1986).

We used various combinations of nerve trunk stimulation and recording to determine numbers of shared units in 16 crabs (Table 2). Stimulation of the distal end of ATh-Cx allowed simultaneous recordings from both ALdN and ALvN. Up to six units could be distinguished in ALdN recordings, and in ALvN there were five. At least four shared units were recorded from these two nerves in three of eight recordings. Detection of five spike heights in ALvN was supported by recordings in which this nerve was stimulated and five spikes were propagated antidromically to A Th-Cx (Table 2). Similarly, six units were activated in A Th-Cx when ALdN was stimulated. Stimulation of the small nerve to the coxal head (ALcN) allowed the greatest resolution of spikes because of the conduction distance. An example of unit recruitment in ATh-Cx and ALdN by increasing stimulus voltage applied en passant to ALcN is shown in Fig. 4A. The smallest and last of the six axons to be recruited in ATh-Cx and ALdN was also present in ANAS, which connects ATh-Cx with P Th-Cx (Fig. 4B,C).

When ALdN was stimulated, one of the four units shared with the recorded nerve, ALvN, and the last one to be recruited, was found to be the common unit present in ANAS (Fig. 5).

When we stimulated ANAS we activated one axon in PLN (Fig. 6). If we also stimulated P Th-Cx while continuing to stimulate ANAS, we could recruit two additional units in PLN (Fig. 6). Thus at least two of the axons that we observed in cross-sections of PLN have their origin in P Th-Cx and the third axon in the PLN reaches it via ANAS. In addition to the PLN axon recorded in ANAS, there are two other motoneurones in ANAS that innervate the coxal anterior remotor. At least one of these also innervates the posterior remotor.

When all the nerve trunks had been cut centrally, we were still able to record spontaneous activity in ANAS and to elicit bursts of activity in response to probing the ventral ridge at the junction between the coxa and thorax.

Our main purpose in recording from levator muscle fibres was to determine the types of responses that the motoneurone population could evoke in the various heads of the muscles and therefore whether the different heads of the levator muscles were specialized for different functions. To this end, we gathered data on muscle fibre resting potentials and numbers and types of postsynaptic potentials (PSPs) supplying the fibres. We did not feel that we could reliably characterize individual AL motoneurones from preparation to preparation by spike height or other criteria, and therefore did not attempt to split the nerve and to stimulate axons individually to provide a detailed map of the distribution of individual motoneurones supplying the muscle heads.

The mean of the resting potentials for all penetrations of muscle fibres in the three heads of AL was —67 mV (Table 3). In 102 penetrations of AL muscle fibres in 25 animals, PSPs were evoked by electrical stimulation of the levator nerves with increasing stimulus strength. The majority, 81 %, were innervated by more than one axon. Many fibres (43 %) have double innervation, 30% have triple innervation, but fibres with four, five or possibly even six PSPs were found; although with increasing stimulus strength, shifts in the time of activation of units already responding made the detection of additional PSPs difficult.

The types of PSP responses we recorded from fibres in the subdivisions of the muscles and their incidence were very similar for the different muscle heads (Table 3). Depolarizing muscle fibre responses to AL nerve stimulation were grouped into three size categories on the basis of their responses to single shocks applied to the nerve. PSPs in the first category (‘small’) ranged from <1 to 5 mV and were found in most AL fibres penetrated. Small PSPs ranged from weakly to strongly facilitating types. These PSPs produced no action potentials in the muscle fibres and indeed could produce so small a response to a single electrical shock that some of them may have been overlooked. The second (‘intermediate’) category of PSPs was found in 72% of the fibres. These had peak amplitudes ranging from 5 to 15 mV and responses that facilitated and summated at frequencies greater than 4 Hz. At high stimulus frequencies (about 100 Hz) graded spikes could be observed. Although it is possible that the intermediate PSPs summated to threshold, another possibility is that the intermediate PSPs acting in consort with simultaneously active small PSPs produced the active response. This is in contrast to the ‘large’ (15–25 mV peak) PSPs observed in 29 % of AL fibres. When single pulses were applied to activate the large PSPs, the fibres would often reach threshold and fire an action potential. The ability of the large PSPs to generate an action potential diminished as the preparation aged, and overshooting action potentials had usually disappeared within 20 min after the crab was prepared for recording.

An example of a dual recording from an ALd and an ALv fibre that indicates sharing of three of the six PSPs that were evoked upon stimulation of the cut distal portion of ATh-Cx is shown in Fig. 7. Our estimates of the number of axons innervating a given fibre are conservative, because some of the smallest excitatory potentials often did not emerge from the baseline unless activated repeatedly (unit 1, Fig. 7). In addition, the PSPs of the common inhibitor may have been undetectable or reversed at the membrane potential of many of the fibres, as indicated below.

Our simultaneous recordings from muscle fibres in the different heads confirmed the data for shared units obtained from dual nerve recordings. For instance, a dual penetration of fibres in the coxal head and another in the dorsal head revealed six shared motoneurones. Evidence for four shared PSPs in the dorsal and ventral heads of AL was provided by two dual-fibre penetrations.

Inhibitory postsynaptic responses were recorded in five penetrations of AL muscle fibres. Fig. 8A shows simultaneous recordings from two AL fibres in which a PSP was recruited at the same threshold in both fibres, but the fibre with a membrane potential of only —55 mV exhibited a hyperpolarizing potential, while the fibre with a membrane potential of —71 mV had a corresponding depolarizing potential. The mean resting potential for the muscle fibres in AL was —66·8 mV (Table 3), so detection of the common inhibitor input was complicated by the fact that its PSP reversal potential must have been near to or more positive than the resting potential of most fibres penetrated. The smallest PSPs recorded in many fibres (Table 3) may have been attributable to the inhibitor rather than to an excitatory input, and in many cases the recruitment of such an input during recordings of compound PSPs may have gone undetected.

In several instances, evidence that we had recruited an inhibitory motoneurone was provided by membrane potential hyperpolarization that developed during high-frequency stimulation. Fig. 8B shows a dual penetration of ALd fibres in which the first, very small PSP (the inhibitor?) was present in the lower trace but was not detected in the fibre recorded in the upper trace; the fibres shared the second PSP that was evoked in the fibre recorded in the lower trace. When the nerve was stimulated at 70 Hz, the fibre shown in the lower trace hyperpolarized despite the activation of the depolarizing potential. The fibre in the upper trace, which lacked the first PSP, did not show the hyperpolarization. A greater hyperpolarizing effect was apparent in the lower trace at 100 Hz. Similar hyperpolarization at stimulus strengths that were below the threshold of any excitatory inputs was observed upon repetitive stimulation in two recordings from PL fibres, and in one additional case.the hyperpolarization could be driven by stimulation of ANAS, which carries the common inhibitor input for PL.

The average membrane potential of PLN fibres was –67·8 mV (N = 25, S.D. = 11·5). Recordings of PL muscle fibre responses to stimulation of the PL nerve revealed one triply innervated fibre. All three of the postsynaptic potentials encountered in that one instance were depolarizing, but one was undoubtedly attributable to activation of the common inhibitor. In most cases, we stimulated P Th-Cx proximal to ANAS, so only the two excitatory PL motoneurones originating in that trunk would have been activated, but in one instance in which PLN was stimulated, very small hyperpolarizing PSPs were observed, and the membrane potential hyperpolarized upon repetitive stimulation, despite the presence of an additional depolarizing PSP. Six of the 12 PL fibres that we studied were doubly innervated and the remaining five fibres were singly innervated. Fibres from all regions of the muscle were sampled. The range of PSP amplitude was similar to that recorded in the AL fibres, with the exception that action potentials were elicited only upon repeated stimulation.

In the work reported here, we have shown that in Carcinus it is theoretically possible for the major thoracic heads of AL to be independently active in different motor programmes due to the segregation of at least some of the motoneurones. The possibility that the distinct heads of AL might have physiologically distinguishable roles in routine elevation and autotomy was suggested by McVean & Findlay (1976) on the basis of different insertions of the fibres on the AL tendon, and the observation that, in Carcinus, the angle between the axes of the two major thoracic subdivisions of the muscle is 30° at mid-arc (McVean, 1973). Complete segregation of motoneurones among the thoracic heads of AL is documented in the swimming crab Portunus, where Hoyle & Burrows (1973) found that in the fifth legs the 10 motoneurones that supply AL are completely segregated to provide separate innervation of the three thoracic heads. This segregation is reflected by different contractile and electrical responses in the fibres of the different heads which are related to their use in swimming, and there is no evidence that it facilitates production of autotomy. Indeed, the fifth legs of swimming crabs are autotomized with more difficulty than the other limbs (White & Spirito, 1973; S. Moffett, personal observations). In contrast to the segregated condition of innervation of the heads of AL in the fifth legs of swimming crabs, Moffett (1975) found 8–10 axons in light microscope sections of both ALdN and ALvN in the land crab Cardisoma, and provided recordings that showed that several motoneurones are shared by the two nerves, and that at least two additional large motoneurones not seen in resistance reflexes were activated synchronously in both nerves during the autotomy motor pattern. The innervation of AL muscle heads of the fifth legs in Carcinus (Fig. 9) appears to be intermediate between the specialized situation in swimming crabs and the relatively unsegregated condition of the muscle heads found in land crabs; this intermediate condition is reflected by the observation that Carcinus exhibits swimming-like motor patterns with the fifth legs (Bévengut et al. 1983).

As indicated in Fig. 9, we conclude that AL in Carcinus is innervated by 7–10 motoneurones, depending on whether the data are drawn from electrophysiological (Fig. 9A) or anatomical (dye projection) methods (Fig. 9B). Our data agree fairly well with the nine motoneurones reported by Bévengut et al. (1983) on the basis of cobalt dye backfills. The difference between the number of neurones detected in nerve backfills and in electrophysiological recordings may reflect limitations of spike detection in whole nerve recordings. Recruitment of additional small action potentials after large units were active would be difficult to detect in compound recordings. Even the nerve backfills could have failed to demonstrate cell bodies belonging to cells that had very small axons, but nonspecific staining could have provided for an overestimate of the number of neurones having axons in a filled nerve. Identification of one of the shared motoneurones as the common inhibitor (Yox & Moffett, 1986; Rathmayer & Bévengut, 1986), has already been suggested by the single nerve backfills of Bévengut et al. (1983).

The electron micrographs revealed more axons than could be accounted for by any of the techniques for identifying efferent axons. A similar discrepancy was reported by Parsons (1982) for the nerve innervating the flexor muscle of Carcinus. The number we report for the nerve supplying ALd is the same as that reported by McVean (1974), except that we included the three axons under 3μm, and McVean did not count any axons under 3μm. In Callinectes sapidus, F. W. Tse & H. L. Atwood (personal communication) have found five large-diameter axons in electron micrographs of ALd and have electrophysiological evidence for five excitatory axons as well as an inhibitor. It is likely that the smallest axons that we observe in the electron micrographs belong to sensory neurones (Fig. 2D). We obtained evidence for afferent activity in ANAS, and Tse, Govind & Atwood (1983) have described what appears to be a sensory structure within ALd of Callinectes.

We have demonstrated that the two heads of PL in Carcinus are innervated by two excitatory motoneurones that are not shared with AL and, additionally, by the common inhibitor (Figs 6, 9). Most earlier reports have indicated two motoneurones, one tonically active and the other phasically active (Clarac & Wales, 1970; McVean, 1974; Moffett, 1975). The input of the common inhibitor would have been overlooked in myograms. Both excitatory motoneurones of PL are known to be active during locomotion (Clarac & Wales, 1970) and Findlay & McVean (1977) used the coxa-basal (CB) chordotonal reflex to evoke activation to the tonic unit and cuticular stress detector (CSD₁) stimulation to evoke activity in the phasic unit in their mapping of PL fibres.

Although it is accepted that AL produces the force for autotomy, this does not rule out the hypothesis that PL normally assists in the autotomy reflex (McVean & Findlay, 1976). The contribution of PL cannot be very important, however, as Moffett (1975) confirmed Fredericq’s (1892) observation that the PL tendon can be cut without interfering with autotomy in Carcinus, and that in Cardisoma autotomy occurs with the PL tendon and muscle intact but PLN cut.

Evidence that PL has a role in producing the fracture comes from myograms recorded from the two heads of PL during autotomy in Carcinus (McVean & Findlay, 1976, p. 372). In these recordings, activity in the posterior head (PPL) ceases prior to production of the fracture by AL. In contrast, McVean & Findlay’s myogram of the rotatory portion of PL (RPLM) shows little activity before the injury but a burst of activity coincident with autotomy and the activation of AL. This recording is difficult to reconcile with the subsequent demonstration (Findlay, 1978) that both heads of PL share the two excitatory motoneurones. The hypothesis that the two PL heads have different functions could have been based on supposed rotatory head recordings that were contaminated by cross-talk from ALc. We have demonstrated that this small head, which is adjacent to the rotatory part of PL in the fifth legs, is innervated by most or all of the motoneurones that activate ALd, and therefore would have been active in the contraction that led to the fracture. The recordings made by McVean & Findlay (1976) from the levatory part of PL, which is farther from ALc, are more likely to have reflected the response of the entire PL muscle.

The role of PL in locomotion and autotomy is still not resolved, since the peculiar orientation of its major tendon is clearly designed to act against the tendon of AL rather than to levate the limb (McVean, 1973, 1982; Moffett, 1975). Findlay & McVean (1977) confirmed Moffett’s (1975) observation that activity in PL excitatory units is suppressed in response to limb injury, and there is general agreement (Moffett, 1975; Findlay, 1978) that CSD₁ plays a role in preventing accidental autotomy, but Moffett feels that its reflex activation of the phasic PL unit is over ridden in the autotomy response, whereas Findlay suggests that CSD₁ inhibition of AL motoneurones must be overridden for the fracture to occur and that CSD₁ drive to the PL phasic motoneurone promotes autotomy. Only intracellular recordings wil] reveal how sensory activity activated in CSD₁ by stress in the cuticle is integrated with the injury stimulus during the autotomy motor programme. Our characterization of the motoneurones that innervate PL and those that are shared by ALd and ALv or unique to one or the other head will facilitate identification of individual neurones and interpretation of sensory inputs received by the intracellularly penetrated cells.

We are grateful to R. A. DiCaprio, C. R. Fourtner, J. R. King and to two anonymous reviewers for their assistance and helpful comments. Part of the work reported here was applied to the Master’s degree awarded to DPY under the direction of C. R. Fourtner at SUNY, Buffalo. We acknowledge with gratitude the use of the WSU Electron Microscope Center and support provided by the National Science Foundation, Grant BNS 8022762 to SM.