Peripheral targets of centrally located putative accessory neurons of MRO in the isopod Ligia exotica

Muscle receptor organs (MROs) are generally found in both the thorax and abdomen of Malacostracan crustaceans (Alexandrowicz, 1951, 1967; Pilgrim, 1960; Rydqvist, 1992; Macmillan and Field, 1994; Wallis et al., 1995). The abdominal MROs of the decapod crustaceans Homarus vulgaris and Palinurus vulgaris are innervated by two efferent axons called accessory nerves (thick and thin nerves), and by an X-fiber(Alexandrowicz, 1951). Kuffler and Eyzaguirre (1955) reported the inhibitory action of accessory neurons in crayfish. The reflex inhibitory action of accessory neurons activated by MROs of the same and neighboring segments was then reported (Eckert,1961; Fields et al.,1967; Jansen et al., 1970a,b, 1971). Based on these studies,Wine and Hagiwara (1977)explored the origin of such accessory neurons in the CNS of the crayfish Procambarus clarkii, and in the next-posterior abdominal ganglion found four putative accessory neurons: Acc-1 (accessory-1) to Acc-4. These neurons have not, however, been directly demonstrated to inhibit the stretch-activated response of MRO by intracellular studies combined with dye labeling.

In the isopod Ligia oceanica, the presence of the MRO was first described by Alexandrowicz(1967) and its stretch-activated response was recorded by Alexander(1971) only from the rapidly adapting stretch receptor. A subsequent study on the stretch receptor in Ligia exotica reported that MROs with rapidly and slowly adapting stretch receptors were located in the third to the eighth thoracic segments,and also occurred in the reduced abdominal segments(Niida et al., 1995a). The presence of inhibitory accessory neurons in isopods has been reported in Armadillidium vulgare (Niida et al., 1991, 1998) and Bathynomus doederleini (Niida et al.,1995b; Iwasaki et al.,2001). The location and morphology profiles of the putative accessory neurons of isopods are similar to those in crayfish, being located just below the surface on the dorsal side in the thoracic ganglia, so it is not difficult to impale them with electrodes for intracellular recording. In particular, one of the centrally located neurons in the thoracic ganglion,possibly equivalent to the putative thick accessory neuron of crayfish, is visible under a dissecting microscope. We were thus able to investigate the accessory neurons of Ligia exotica intracellularly.

We first examined the putative accessory neurons in the next-caudal ganglion of the thorax by backfilling with Lucifer Yellow to the segmental nerve, as described by Wine and Hagiwara(1977). The inhibitory effects of these neurons were then examined and their morphology identified by intracellular dye marking. Part of the present study has already been reported in abstract form (Hama and Niida,2000).

Specimens of the isopod Ligia exotica Roux were collected along the coast of the Seto Inland Sea near Ushimado Marine Laboratory, Faculty of Science, Okayama University, Japan. They were kept at 20°C in a container filled with seawater to a depth of 0.1-0.2 cm, and provided with timber and stones to recreate their habitat. By this simple method we could maintain animals reliably for a long period. Both males and females (overall length 1-3 cm) were used in this study.

After anesthetizing the animals in cooled Ligia physiological saline (Yamagishi and Ebara,1985), the legs, sternites and viscera were removed. The second and seventh tergites of the thoracic segments were then fixed with small pins on cork lining the bottom of an acrylic plastic chamber. The proximal cut end of the third segmental nerve (N3), equivalent to N2 of the abdominal ganglion in decapods (Iwasaki et al.,2001), was placed on the tip of a polyethylene tube (1.5 mm in diameter) filled with 5% Lucifer Yellow (w/v in aqueous solution), and subjected to 0.5 s pulses of negative current (0.5μA) for 1.5 h. Lucifer Yellow-filled preparations were kept for 12 h at 4°C to allow diffusion of the dye, then fixed in 10% formalin for 1 h and cleared in glycerin containing an anti-fade reagent (FluoroGuard; Bio-Rad Laboratories, Hercules, CA,USA).

The inhibitory effect of a putative accessory neuron on the stretch-activated response of a thoracic stretch receptor (TSR) was determined between the fifth thoracic ganglia and the TSRs of the fifth somites (TSR-4s)(Fig. 1A) except for one experiment, which tested between the fourth thoracic ganglion and TSR-3. The fifth thoracic ganglia and TSR-4s are easily accessible for performing electrical stimulation, stretch stimulation, extra- and intracellular recordings. The sternites were removed from the ventral surface of the animal,exposing the thoracic and abdominal ganglia with connectives. The nerve cords and ganglia posterior to the second thoracic ganglion were left intact together with tergites of the fifth and sixth thoracic segments, and all the segmental nerves except N3 of the fourth thoracic ganglion were cut off. Accordingly, the TSR of the fifth somite has a central connection through N3 of the fourth thoracic ganglion (Fig. 1A). To access accessory neurons lying on the dorsal side of the thoracic ganglion, the dorsal side of the ganglia with connectives, with only one N3 segmental nerve as mentioned above, was exposed(Fig. 1A), and fixed with small pins to cork lining the chamber. The chamber was filled with saline and kept at 20°C.

All of the segments except the fifth and sixth thoracic segments were cut off and their hemisegments were left by the bisection of tergites along the midline. The tergum of the fifth thoracic segment was fixed with small pins,and stretch stimuli (ST) were then delivered by manually pulling the free tergite of the sixth thoracic segment with forceps mounted on a micromanipulator (Fig. 1A).

A neuron impaled with a glass microelectrode and tentatively identified as a putative accessory neuron, based on antidromic stimulation to N3 of the fourth thoracic ganglion, was injected with depolarizing currents (5-10 nA). The inhibitory effect on the stretch-activated response of the thoracic stretch receptor was then ascertained extracellularly in N3 through an en passant suction electrode (Fig. 1A). Simultaneously, in order to label the impaled neurons in the CNS, a hyperpolarizing intermittent current (–4 nA; frequency, 1 Hz;duration, 500 ms) was passed through an intracellular electrode filled with Lucifer Yellow for more than 60 min. To ensure diffusion of Lucifer Yellow,these preparations were occasionally stored overnight in a refrigerator(4°C).

The stretch-activated response of a rapidly adapting stretch receptor of Ligia exotica ceases within a very short period during stretch stimulation. Accordingly, unlike with slowly adapting stretch receptors, it is difficult to examine inhibition of the stretch-activated response by the extracellular recording method above. Instead, we directly recorded inhibitory synaptic potentials from the receptor cells of rapidly adapting stretch receptors by injecting depolarizing currents into the putative accessory neurons in the CNS. Three electrodes were used for this: the first for impaling the rapidly adapting stretch receptor neuron, the second en passant at N3 and the third for current injection into neuron #3. Furthermore, to show that hyperpolarizing potentials in receptor cells of stretch receptors were synaptically mediated, the membrane potential of rapidly adapting stretch receptor neurons was changed at three levels by clamping currents. Subsequently, the impaled neurons were labeled with Lucifer Yellow and identified.

When a putative accessory neuron such as neuron #1 failed to inhibit a stretch-activated response, or when we were unable to trace its Lucifer Yellow-filled terminal site, its peripheral target was intracellularly determined as shown in Fig. 1B. The first intracellular electrode (R1) was inserted into neuron #1 in TG5 and the second (R2) and third (R3) electrodes into the DEM1 and DEM2,respectively. Electrical stimulation was made at the nerve to deep extensor muscles (npm) through an en passant electrode (S). Recordings from impaled neurons and muscles, or intracellular stimulation were performed using conventional intracellular techniques, with glass microelectrodes filled with either 5% Lucifer Yellow in 1 mol l^-1 LiCl (resistance 15-30 MΩ) or 2 mol l^-1 potassium citrate (5-10 MΩ).

Lucifer Yellow was applied to N3 in 30 specimens, using the third to fifth thoracic ganglia. Motor neurons, unidentified primary afferents, putative accessory neurons and axons of thoracic stretch receptors (TSRs) were labeled. The following description is limited to a given segment, but it also applies to the third to seventh thoracic segments.

The labeling profiles of TSR axons are shown in Fig. 2A,B, where the axon of TSR-4 (also see Fig. 1A)bifurcates near the root of N3 in TG4; one ascends (AXas) while the other descends (AXde). Each axon runs through the central part of the ganglia,medially along the connective and ending in the tritocerebrum or in the fourth abdominal ganglion (not shown), as in Armadillidium vulgare(Niida et al., 1998).

Lucifer Yellow-backfilling regularly stained three putative accessory neurons #1, #2 and #3, in order of decreasing cell size, whose location and morphology profiles were consistent in each thoracic ganglion(Fig. 2A,B). The detailed profiles of these putative accessory neurons are described below.

During a single or repetitive electrical stimulation of N3, a glass microelectrode was impaled on the perikaryon of the supposed neuron #1,producing a small antidromic response with a resting membrane potential of 40-60mV (not systematically studied). Immediately thereafter, depolarizing currents (4-10 nA) were passed into the impaled neuron, but we never found spike initiation in itself, and an ongoing stretch-activated response,monitored through an en passant suction electrode, was not inhibited at all. Six such neurons were intracellularly studied and stained with Lucifer Yellow. All the Lucifer Yellow-filled neurons displayed the morphology shown in Fig. 3A, and the profile matched neuron #1 (Fig. 2B). The cell body (approximately 60μm) is located in the central area of TG4,and gives off a short neurite that bifurcates anteriorly and posteriorly. The posterior short branch further divides into two branches (arrows in Fig. 3A). The anterior neurite,of large diameter, runs through the connective, reaching the caudal region of the next-anterior ganglion (TG3), and issues numerous branches in the area where the extensor and flexor motor neurons of TG3 lie. The neurite then sharply curves, descends the connective and enters N3. Its subsequent pathway was traced near the deep extensor muscle (DEM), but we could not follow it any further due to weak Lucifer Yellow-labeling. Based on this, we applied backfilling with Lucifer Yellow to the more proximal cut end of the nerve(npm) innervating the deep extensor muscle, and this labeled neuron #1 in TG4 and motor neurons in TG3 (small arrowheads in Fig. 3Bi). We concluded that neuron #1 is not an inhibitory accessory neuron, but rather a motor neuron of DEM. The deep extensor muscle consists of the medial deep extensor muscle(DEM1) and lateral deep extensor muscle (DEM2)(Fig. 1B, and also see Fig. 7). To determine which muscle is innervated by neuron #1, we recorded junctional potentials from DEM1 and DEM2 in separate animals (ten specimens in total, five of each), as shown in Fig. 1B. When the npm was weakly (2 V) electrically stimulated, no responses were seen in neuron #1,DEM1 or N3 (Fig. 4A). A more intense electrical stimulus was then delivered, which resulted in both an antidromic response in neuron #1 and compound action potentials at N3(Fig. 4B), but there was still no response in DEM1. Electrical stimulation at an even higher intensity (3 V)finally recruited a junctional potential in DEM1(Fig. 4C). Similarly,junctional potentials of DEM2 were then recorded in a separate specimen. Electrical stimulation at a low intensity (3 V) was made at the npm, with no response (Fig. 4D). A slightly more intense stimulus produced only compound action potentials at N3. An even more intense stimulus was then delivered at the npm, and produced a junctional potential in DEM2, an antidromic response in neuron #1, and compound action potentials. If neuron #1 sends its axon to DEM2, the simultaneous occurrence of an antidromic response in neuron #1 and a junctional potential in DEM2 with electrical stimulation at a given intensity would indicate that they have similar excitation thresholds (Fig. 4E). In contrast, an antidromic response in neuron #1 and a junctional potential in DEM1 are evoked by electrical stimulation of their axons with different thresholds (Fig. 4B,C). This indicates that DEM1 is not innervated by neuron #1,but rather by motor neurons in the next-anterior ganglion (TG3).

Following electrical stimulation of N3 in the fourth thoracic ganglion, an antidromic response of shorter duration than that of neuron #1 was intracellularly recorded from the neuron in the central area of the fifth thoracic ganglion. This shorter duration of the antidromic response was useful as a landmark to distinguish neurons #2 and #3 from neuron #1. When TSR-4 was activated by stretch stimuli, its response was recorded from N3 through an en passant suction electrode (Fig. 1A). Approx. 5 s after beginning stretch stimulation, depolarizing currents were intracellularly passed from the impaled neurons. The impaled neurons then initiated impulse discharges(Fig. 5A); they conducted along the connective anterior to the fifth thoracic ganglion and appeared in N3 as small impulses (single and double asterisks in Fig. 5B). However, there was no effect on the stretch-activated response of TSR-4. A hyperpolarizing current was then injected into the impaled neuron for Lucifer Yellow staining (profile shown in Fig. 6), which was identified as neuron #2 in Fig. 2B. The cell body of approximately 30μm lies in the anterior hemiganglion contralateral to the side where the axon exits at N3(Fig. 6A). The neurite from the cell body goes down obliquely and forms a deformed Ω-like dendritic structure in the central area of the fifth thoracic ganglion. The neurite issuing from the Ω-like structure then goes upward and bifurcates near the root of N3. One branch reaches the posterior region of the fourth thoracic ganglion and gives off small dendritic branches (double arrows in Fig. 6A). The other runs toward the periphery via N3; it courses npm and terminates in the superficial extensor muscle (SEM1 and SEM2) within the area-1 in Fig. 6, shown enlarged in area-2, in which the Lucifer Yellow-filling nerve toward the SEM1 and SEM2 passes through the stretch receptor without giving off axon branches. The spatial arrangement of the superficial extensor muscle and thoracic stretch receptors is shown in Fig. 7. The example in Fig. 6A thus enabled us to determine its destination by overnight diffusion of Lucifer Yellow. The staining quality was insufficient due to movement of the dye, but another sample provided a detailed profile of neuron #2 with the Ω-like dendritic structure (Fig. 6B). Seven samples were thus stained intracellularly. They showed almost the same morphological characteristics and had no inhibitory effect on the stretch-activated response.

The response of impaled neurons to antidromic stimulation at N3 was ascertained, and a stretch stimulus was then delivered to TSR-4 (see Fig. 1A). The response of stretch-activated TSR-4 was extracellularly recorded from N3 through an en passant electrode (Fig. 8A). At the beginning of stretch stimulation, indicated as an upward deflection in the lower trace in Fig. 8A, rapidly adapting large impulse discharges (arrows) occurred together with slowly adapting impulse discharges, which were maintained during stretch stimulation. Concomitantly with the stretch-activation of TSR-4, the impaled neuron produced a few sporadic impulses and a series of EPSPs(Fig. 8B,C); the two parts of the trace in rectangles in Fig. 8Bi,ii were enlarged, and show stretch-activated EPSPs in Fig. 8Ci,ii. This indicates that there is synaptic contact between TSR-4 and the impaled neurons. Therefore, depolarizing currents were passed into the impaled neuron(asterisks in Fig. 8B), which produced regenerative impulses in its own cell. Simultaneously, the preceding stretch-activated response of TSR-4 with slowly adapting impulse discharges was apparently suppressed (asterisks in Fig. 8A) along with the presence of small impulses at a high frequency in the part indicated by the asterisks. After completing the physiological experiments, the impaled neurons(N=6) were stained intracellularly with Lucifer Yellow to reveal the morphology of neuron #3 (Fig. 2B). As seen in Fig. 9, a cell body (about 20 μm) lies in the fifth posterior hemiganglion. The neurite from the cell body gives off dendritic branches,turns sharply to the left and forms an H-like dendritic structure in the central area of the thoracic ganglion. The neurite courses obliquely, reaches the root of N3 and descends to the periphery. We were unable to trace its axon to the stretch receptor, however, due to weak Lucifer Yellow-filling.

To find the possible synaptic sites of the axon of TSR on neuron #3, we tried intracellular recording from neuron #3 in TG-5 and TSR axons passing through TG-5 along with their intracellular labelings. As a result, the descending axon (Axde) of TSR-4 (Fig. 10A) completely overlapped one component of the H-like dendritic structure of neuron #3 in TG-5 (thin arrow in Fig. 10A). If the TSR axon via N3 on the opposite side was intracellularly labeled, the other component of the H-like dendritic structure (arrowhead in Fig. 10A) would also be overlapped by the descending TSR axon. This presumption is shown in another example (Fig. 10B), where the distal component (to the cell body of neuron #3) of the H-like dendritic structure was overlapped by the contralateral (to the cell body of neuron #3)descending TSR axon via N3 in TG3.

We examined whether rapidly adapting stretch receptors (RC1 in Fig. 1A) are innervated by neuron #3 (Fig. 11). A receptor cell (RC1) was penetrated with a glass microelectrode filled with potassium citrate and depolarizing currents were then injected into neuron #3(Fig. 11A). Since the resistance of the intracellular electrode became too high, the bridge was unbalanced and could not be compensated. The effect of the depolarizing currents was manifested as impulse discharges through the en passantelectrode at N3 (asterisks in Fig. 11A) and hyperpolarizing potentials were produced in the rapidly adapting stretch receptor neuron (RC1 in Fig. 11A). This potential is considered to be synaptically mediated, since the amplitude of the hyperpolarizing potential changed with the injection of hyperpolarizing or depolarizing currents into the stretch receptor neuron(Fig. 11B).

We assumed that putative accessory neurons in L. exotica are morphologically similar to some accessory neurons in crayfish(Wine and Hagiwara, 1977). To support this assumption, the locations and morphology of neurons #1-3 in L. exotica were compared to Acc-1 through Acc-4, which have been identified in decapods. The somata of both neuron #1 and Acc-1, the largest of their respective groups, usually lie near the center of the ganglion, and neuron #1 gives off small dendritic branches latero-posteriorly (arrowhead in Fig. 3A). Such branches can also be seen in Acc-1 of crayfish (fig. 21a in Leise et al., 1987) and in the accessory neuron of a squat lobster (figs 3 and 4 in Wallis et al., 1995). `As a surprising feature', Wine and Hagiwara(1977) noted that Acc-1 in crayfish had major branches in the next-anterior ganglion. This structure can also be clearly seen in the intracellularly stained neuron #1 of L. exotica (double arrows in Fig. 3A). Neuron #2 and accessory-2 also have similar positional and morphological characteristics; each soma is located contralaterally in the anterior hemiganglion, and neuron #2 has an Ω-like dendritic structure in the center of the ganglion (Fig. 6A,B), while this structure is deformed and deviates laterally from the center of ganglion in Acc-2 of the crayfish,(fig. 10 in Wine and Hagiwara, 1977). Neuron #3 and Acc-4, which are both small, are located contralaterally in the caudo-lateral quadrant. Their courses toward the periphery are similar: the axon from each soma turns sharply and goes up obliquely. The accessory neuron corresponding to Acc-3 in crayfish is consistently lacking in this animal. This was the same in isopods such as Armadillidium vulgare(Niida et al., 1998), Bathynomus doederleini (A. Niida and A. Namba, unpublished data) and Idotea balthica, in which all the neurons highly similar to neurons#1-3 were considered to be motoneurons(Kutsch and Heckmann,1995).

At the early stages of this experiment, we tentatively regarded neurons#1-3 as putative accessory neurons, based on the results of Wine and Hagiwara(1977). Our intracellular experiments combined with Lucifer Yellow-labeling demonstrated, however, that neurons #1 and #2 are actually motor neurons, and only neuron #3 is an inhibitory accessory neuron. Neuron #1 is morphologically equivalent to Acc-1,which has been considered to be a thick accessory neuron with inhibitory action. Leise et al. (1987)proposed that in the crayfish Pacifastacus leniusculus `Acc-1 might be a common inhibitor, inhibiting the fast extensor muscle as well as MRO, since backfillings from npm routinely stained them'. When Acc-1 is an inhibitory accessory neuron, we expect it is an authentic GABA-ergic neuron based on studies to date (for example, Elekes and Florey, 1987). Such GABA-ergic neurons were immunocytochemically studied in the CNS of the crayfish (Mulloney and Hall,1990), and there were at most four neurons showing GABA-like immunoreactivity (GLI) as accessory neurons in one abdominal ganglion. In addition, no relationship between thick or thin accessory neurons and neurons with GLI was mentioned. In an immunocytochemical study in isopods,GABA-antibody did not label neuron #1. Instead, neuron #1 was labeled with a tyrosine hydroxylase antibody (Tsumura et al., 2000). Similar results were also obtained in the isopod Armadillidium vulgare(Pollák et al., 1999). Accordingly, we suspect that Acc-1 is GABA-ergic.

As shown in Fig. 8, neuron#3 receives synaptic input from the thoracic stretch receptor; however, the post-synaptic response of neuron #3 (Fig. 8B,C) to a stretch stimulus seems to be weak, since its stretch-activated impulses are sporadic. Such a weak synaptically mediated response might be due to the recording site, but in this case recordings were made in the vicinity of the H-like dendritic structure of neuron #3 at the central region of TG5 (arrows in Fig. 9). In the present experiment, the amplitude of the stretch stimulus was not large enough to induce noticeable impulse discharges of the rapidly adapting stretch receptor (arrows in Fig. 8A; see the stretch-activated response in Fig. 1A). Accordingly, this weak synaptic activity may be partly related to the strength of stretch stimulation. In addition, we note the reported synaptic efficiency of ipsilateral and contralateral inputs to accessory neurons of crayfish (Jansen et al., 1970a,b). Considering these synaptic inputs, we assume that the convergence of at least two sensory inputs may be required for the initiation of impulse discharges in neuron #3; i.e. the input from stretch receptors on the right and left sides of the same segment or from two successive segments. This interpretation may be supported by the unique neural structure of neuron #3; in the central region of the thoracic ganglion, there are two parallel vertical dendrites of neuron #3, with which the axons of the stretch receptor on each side may make synaptic contact. These assumptions are further supported by the circumstantial evidence shown in Fig. 10. Thus, neuron #3 may be fully activated when it receives simultaneous synaptic inputs from the two vertical dendrites. This could be ascertained in future by a detailed study of the relationship between reflex inputs and outputs in a given segment or adjacent segments.

This work was supported in part by a grant from the Hungarian National Research Fund (0TCA T026652) to L. Molnár. We thank M. Tsumura for assisting with Lucifer Yellow backfilling and drawings.