INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sensory neurons are not mere passive conduits of afferent information. They also perform important computational operations on that information. The role of the integrative functions of sensory neurons can be advantageously studied in certain invertebrates (e.g., Baccus 1998; Burrows and Matheson 1994; Chiel et al. 1990; Mar and Drapeau 1996; Van Essen 1973). In some cases both the sensory cells, and the circuitry to which they connect, are readily accessible for study. For example, in the feeding system of the marine mollusk Aplysia, some of the relevant sensory neurons have large cell bodies that are located in the CNS (Chiel et al. 1986; Miller et al. 1994; Rosen et al. 1979, 1982; Weiss et al. 1986), and there is considerable information about the identified neurons that comprise the central circuitry that controls feeding responses (Hurwitz et al. 1997; Hurwitz and Susswein 1996; Plummer and Kirk 1990; Rosen et al. 1991; Susswein and Byrne 1988; Teyke et al. 1993). One set of sensory neurons in the feeding system of Aplysia, the radula mechanoafferent (RM) neurons, are particularly interesting because they have been found to contain neuroactive peptides. Furthermore, the activity and excitability of the RM cells can be affected by excitatory and inhibitory synaptic inputs (Miller et al. 1994), and, in addition to responding to external mechanical stimulation of the radula, the cells respond to stretch and contraction of a subradula tissue (Cropper et al. 1996). A broad aim of this, and a companion study (Rosen et al. 2000), was to investigate the integrative actions of sensory neurons and the mechanisms by which sensory neurons interact with pattern-generating elements and higher order neurons that generate behavior.

Initial attempts to analyze the function of RM sensory cells were complicated by indications that the RM neuron population was heterogeneous. The neurons differed greatly in size and in morphological characteristics. Some had a monopolar shape, whereas others had a bipolar appearance. Furthermore, many, but not all of the cells, were found to be immunopositive for small cardioactive peptide (SCP). In the present study, the two largest RM neurons were investigated and shown to be distinctly identifiable based on morphological and electrophysiological criteria. The cells, B21 and B22, were both found to be immunopositive for SCP, but had different types of synaptic interconnections to specific motor neurons, interneurons, and other sensory cells. This paper focuses on the nature of the diverse synaptic outputs and inputs of B21 and B22. We also describe different forms of homosynaptic plasticity that occur in the connections between these cells and certain identified follower neurons. Elsewhere (Rosen et al. 2000), we report the results of the study of a variety of sources of inputs that can produce phase-specific heterosynaptic modulation of the outputs of a RM neuron during ongoing feeding motor programs.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

The subjects were Aplysia californica weighing 300-450 g. They were maintained in 14°C, aerated, artificial seawater (ASW). Two types of preparations were used: an odontophore preparation and a dissected odontophore preparation. The odontophore is the tongue-like structure arising from the floor of the cavity of the buccal mass (feeding apparatus). It is capped by the radula, the chitinous, rasp-like, grasping surface that is divided into two halves by a medial longitudinal groove about which the halves fold and unfold during closing and opening movements (Howells 1942). Internally, the odontophore consists of a semirigid core made up of continuations of the radula, the radula stalk, and the radula sac (including the collistylar cap) to which attach the: I4 muscles; I5 accessory radula closer (ARC) muscles; and the I7-I10 complex of radula opener muscles (Evans et al. 1996). The side walls of the odontophore consist of a spongy support tissue (the rotella) that contains bolsters (fluid-filled channels) that contribute to the strength, shape, and flexibility of the structure (Drushel et al. 1997; Eales 1921).

The odontophore preparation, with innervation provided by the cerebral and buccal ganglia, was used chiefly to determine the identity of the RM neurons (Fig. 1A). The general properties of RM neurons, including their receptive fields, response characteristics, and synaptic connections to identified higher order neurons were determined with this preparation. The preparation included the structures comprising the core of the odontophore (see above), the rotella, the muscles known to be involved in radula opening and closing, and some of the muscles involved in forward and backward rotations of the radula. A number of the larger muscles that lead into the odontophore were cut to free it from the floor of the buccal cavity. Particular care was taken to preserve innervation provided by the buccal ganglion via the radula nerve. In selected experiments, innervation provided by buccal nerves 2 and 3 was also preserved. The nomenclature of the nerves follows previous convention (Gardner 1971). The cerebral-buccal connectives and cerebral ganglion were retained in this preparation, while the remainder of the nervous system was discarded.

View larger version (74K): [in this window] [in a new window]   Fig. 1. Photographs of the odontophore and dissected odontophore preparations. A: the odontophore preparation consisted of the radula with its support structures (e.g., rotella) and the innervation provided by the buccal (buccal g.) and cerebral (cerebral g.) ganglia via the radula nerve (radula n.) and the cerebral-buccal connective (C-B conn.). The external (dorsal) surface of the radula is viewed from above (anterior toward top, posterior toward bottom). It appears as it would in its rest position in the buccal mass of a quiescent animal with its foot on the substrate below. During a bite, the grasping surface (grasp surf.) would be rotated in the anterior direction, and the radula halves would open and close around the medial longitudinal groove (groove). B: the dissected odontophore preparation had a series of cuts made to expose internal structures of the odontophore while leaving much of the innervation intact. The 1st was a partial longitudinal cut starting at the anterior face of the odontophore. When combined with 2 additional dorsal cuts through the radula (top), it was possible to reflect the front and lateral surfaces of the odontophore to the sides and view internal tissues. The tissues of the semirigid radula are continuous with the centrally placed radula sac. The region where these structures merge is the site of attachments of the I4 (I4 leaflets) and accessory radula closer (ARC) muscles. The ventral surface of the radula sac is the site of attachment of the I7-I10 complex of radula opener muscles, of which the I7 muscle can be seen (rad. opener). The rotella composes much of the sides and front of the odontophore, and its internal face is visible. The buccal and cerebral ganglia, the C-B connectives, and the radula nerve, which are normally found along the posterior surface of the odontophore, have been rotated 90° to the bottom of the preparation. Calibration bars, 1.0 cm.

The dissected odontophore preparation was used in experiments in which radula motor neurons needed to be identified, and for the anatomic and physiological experiments on the subradula tissue (SRT), a thin, flat muscle sheet (Cropper et al. 1996) that lies immediately below the full extent of the chitinous radula. The SRT contains the terminal fields and terminal specializations of the RM neurons (Miller et al. 1994). For the dissected odontophore preparation a partial cut was first made through the rotella muscles forming the anterior wall of the odontophore. Additional small cuts through the dorsal part of the radula permitted the rotella halves to be splayed open to visualize the I4 and I5 muscles (ARC) muscles (see Fig. 1B and legend). In some experiments additional dissection isolated the SRT. The bottom of the radula sac (collistylar cap) was sectioned and removed. The subradula tissue covering the interior surface of the radula stalk-radula sac complex was then peeled off, starting from the anterior ventral region and moving into the posterior dorsal region. Continued peeling of the tissue separated it from the internal face of the radula proper. The insertions (attachment sites) of the I4 and I5 muscles to the subradula tissue were preserved, as were the muscles and their nerves. The subradula tissue was separated from the edges of the rotella by cutting across the borders of these tissues. The dissected tissues of the odontophore and attached ganglia were pinned to the silicone elastomer (Sylgard) floor of a recording chamber.

Electrophysiology

Neurons were impaled with double-barreled microelectrodes for intracellular recording and stimulation. The electrodes were made of thin-walled glass that contained 2 M potassium acetate. They were beveled so that their impedances ranged from 10 to 15 M. To identify cells and examine their morphologies, the potassium acetate in the stimulating electrode was replaced by a 3% solution of 5(6)-carboxyfluorescein dye in 0.1 M potassium citrate, titrated to pH 8.0 with KOH (Rao et al. 1986). The impedance of the dye-containing electrode was 30-40 M. In other experiments, aimed at obtaining accurate measurements of the resting potentials of identified cells, the electrodes were filled with 2 M potassium chloride and beveled so that their impedances were again between 10 and 15 M. Up to four simultaneous intracellular recordings were obtained using conventional electrometers. The recording chamber (Lucite) was divided into two compartments that were interconnected by several petroleum jelly (Vaseline)-filled notches, allowing for the passage of nerves or connectives and selective infusions of ASW solutions containing controlled concentrations of relevant ions (e.g., K⁺, Mg²⁺, and Ca²⁺) and chemostimulants to the central ganglia or the peripheral tissues. Unless otherwise indicated, all multiple cell recordings were from cells located in the same buccal hemiganglion.

Mechanical and chemical stimulation

Tactile stimuli were provided manually by a fire-polished Pasteur pipette or a set of flexible von Frey hairs consisting of plastic fibers that were heated and pulled to various diameters. The "hairs" were calibrated for delivery of punctate stimuli with forces ranging between 0.1 and 10 g. Automatically controlled stimuli were provided by a custom made tapper consisting of a wooden rod, 1 mm diam, attached to the membrane of a minispeaker that was driven by a Grass S4 stimulator (Cropper et al. 1996). Chemical stimuli consisted of a piece of moistened, commercial dried seaweed (Laver, Roland Foods), presented by a hand-held forceps, or graded concentrations of seaweed extract gently perfused over the radula (Susswein et al. 1976). Other chemical stimuli consisted of 4 M NaCl and solutions of amino acids found in seaweed, e.g., glutamic acid (Jahan-Parwar 1972).

Morphology

Neurons were filled with 5(6)-carboxyfluorescein dye (Kodak) by iontophoretic ejection from microelectrodes. To reduce the active transport of the dye out of the cells (Steinberg et al. 1987), probenecid (10 mM final concentration) was added to the ASW bathing medium, and the preparation was kept for 24-48 h at 4°C (Rosen et al. 1991). The unfixed tissues were viewed with a Nikon fluorescence microscope. Confirmation of cell morphology was made with Lucifer yellow ejection, followed by fixation in paraformaldehyde, and clearing in methyl salicylate.

Immunocytochemistry

Previously described methods for indirect immunofluorescence, whole-mount mapping of Aplysia ganglia were followed (Longley and Longley 1986; Miller et al. 1991). In brief, individual neurons were identified by electrophysiological criteria and/or their responses to mechanosensory stimuli. Cells were filled with Lucifer yellow (4%) by iontophoretic ejection from microelectrodes (5 to 10 nA; pulses 0.5 s on, 0.5 s off; 30 min), and ganglia were fixed for 2-4 h (room temperature) in 4% paraformaldehyde. Nonspecific antibody binding was blocked by preincubating tissues in a phosphate buffer solution containing 0.8% normal goat serum (NGS; Miles Sci., Naperville, IL) and 2% Triton X-100 for 2 h at room temperature. Primary rabbit antiserum against small cardioactive peptide B (SCP_B) (provided by A. Mahon) (see Lloyd et al. 1985) was then applied (1:50 to 1:200). After incubation and rinsing (1-2 days each), secondary antibody (goat anti-rabbit immunoglobulin G rhodamine conjugated Fab fragment; Cappel, Malvern, PA) was applied at a 1:50 or 1:100 dilution. After incubation and washing, tissues were mounted on depression slides in phosphate buffer: glycerol (1:6), viewed under a Leitz microscope equipped with epifluorescence (filterpack N-2 for rhodamine, D for Lucifer yellow), and photographed with Tri-X film.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

SCP-containing radula mechanoafferent neurons can be identified as individuals

The cluster of SCP-containing RM neurons, located on the rostral surface of each buccal hemiganglion, was found to consist of a heterogeneous population of ~40 cells of varying position, size, shape, and axon distribution pattern (Miller et al. 1994). Typically the SCP-positive cluster of cells contained three or four neurons that had a distinctive bipolar morphology, whereas the remainder were monopolar or multipolar cells. Other nearby sensory neurons were found to have similar response properties, but were not SCP-immunoreactive. To expedite analysis, unique individuals were sought that consistently showed SCP immunoreactivity and that could also be unambiguously identified by morphological and electrophysiological criteria (Fig. 2). We found that it was possible to identify the two largest, bipolar, SCP-containing RM neurons, that are found at the medial aspect of the rostral RM cluster. In each of four preparations, in which the cells were filled with Lucifer yellow dye, the identified cells, B21 and B22, tested positive for SCP immunoreactivity (Fig. 3). In 48 of 50 preparations only a single neuron with the distinctive properties of B21 was found. B22 on the other hand shared a number of properties with two to three other bipolar, although smaller, SCP-immunoreactive cells.

View larger version (117K): [in this window] [in a new window]   Fig. 2. Photomicrographs of an identified radula mechanoafferent (RM) neuron (B21) in the rostral cluster of small cardioactive peptide (SCP)-immunoreactive cells and 2 cells that connected to it in the ipsilateral buccal hemiganglion. A: 3 neurons on the right rostral surface of the buccal hemiganglion were impaled with double-barreled microelectrodes and identified by electrophysiological criteria. The cells were then filled with Lucifer yellow dye. The lateral cell (left, ) was buccal-to-cerebral interneuron, B19 (Rosen et al. 1991). The middle cell (dark arrow) was a monopolar RM neuron. The medial cell (*) was RM neuron B21, identified by its characteristic size, position, radula mechanosensory receptive field, and the inhibitory postsynaptic potential (IPSP) it received from identified multifunction neuron B4. The axon of B21 in the radula nerve (light arrow) is visible. B21 is connected to B19 and the monopolar RM cell by a nonrectifying electrical synapse. B: whole mount exhibiting SCP immunoreactivity in the right buccal hemiganglion of the same field of the specimen shown in A. Note that B21, but not B19, was SCP immunoreactive. The small monopolar RM neuron was also weakly immunoreactive. The largest cells containing immunoreactive material belonged to the ventral cluster of motor neurons (see Church and Lloyd 1991; Lloyd et al. 1985). bn1, buccal nerve 1; cbc, cerebral-buccal connective; esoph n., esophageal nerve; rad n., radula nerve. Calibration bar, 400 µm.

View larger version (44K): [in this window] [in a new window]   Fig. 3. SCP-immunoreactivity of identified RM neurons B21 and B22. A: photomicrograph of neurons B21 (bottom) and B22 (top) after the cells had been positively identified by electrophysiological criteria and filled with Lucifer yellow dye. B: same field of view after the buccal ganglion was processed for whole-mount SCP immunoreactivty with a rhodamine conjugated second antibody. The reference arrows indicate that the cells filled with Lucifer were also positive for SCP. Calibration bars, 60 µm.

Although B21 and B22 share a number of common electrophysiological properties (e.g., resting potential, spike amplitude, spike duration, sensory threshold, and receptive fields), the cells differed in morphological appearance, axon distribution, and pattern of synaptic connectivity. A summary of some of their similarities and differences is shown in Table l. As detailed below, key differences used to routinely distinguish the cells include the findings that B21 receives a conventional, monosynaptic inhibitory postsynaptic potential (IPSP) from identified neurons B4 and B5. B4 and B5 are virtually identical cells that are located next to one another, and have combined sensory (Jahan-Parwar et al. 1983), motor (Evans et al. 1996; Rosen et al. 1982), and interneuronal functions. By convention we refer to the more medially located cell as B4, but either cell individually can be termed B4/5. B22 does not receive an input from B4/5. Furthermore, B21 produces a slow excitatory postsynaptic potential (EPSP) in neuron B15, but B22 does not. B22 is invariably electrically coupled to B4/5 and B16, but B21 is not. Routine dye-fills showed consistent morphological differences correlated with the physiological findings.

Morphological properties of B21 and B22

B21 is the largest cell (long dimension, 80-120 µm diam in 400-g animals, n = 16) within the rostral cluster of SCP-containing RM neurons (Figs. 2 and 3). It is usually found at the ventromedial edge of the cluster. B21 is a bipolar cell with a rounded soma that projects a lateral process into the ipsilateral buccal hemiganglion and a thick (10-20 µm diam) medial process into the contralateral buccal hemiganglion (Fig. 4, top). In the buccal commissure the medial axon bifurcates and sends another main branch into the radula nerve. The radula axonal branch itself divides at the first major bifurcation of the radula nerve, thereby innervating both the left and right sides of the radula. Both the lateral and medial B21 axons periodically give rise to tufts of short processes that terminate in the neuropil of the buccal ganglion.

View larger version (29K): [in this window] [in a new window]   Fig. 4. Comparison of the morphologies of B21 and B22. A: drawings of 3 B21 neurons that were filled with 5(6)-carboxyfluorescein dye and incubated with probenecid for 24 h B: drawing of 3 B22 carboxyfluorescein dye-filled neurons. B21 was identified by its size, position, receptive field, and the IPSP it received from neuron B4. B22 was identified by similar criteria except that it did not receive an IPSP from B4. It was connected to B4 by a weak electrical synapse. Calibration as shown.

B22 is an oblong shaped, bipolar cell, that is considerably smaller than B21 (Fig. 3B). Its soma size measured along its long dimension is 60-90 µm. It is found adjacent to B21 in a more dorsal position. Like B21, B22 has a lateral and a medial main axon. The medial axon sends a branch into the radula nerve that innervates both the left and right sides of the radula. The main axons give rise to long filamentous processes that extensively ramify in the neuropil of the ipsilateral buccal hemiganglion (Fig. 4, bottom).

Fields of terminal specializations

When either a B21 or a B22 neuron was filled with 5(6)-carboxyflourescein and incubated with probenicid for 72 h (n = 6), it was possible to follow the peripheral axons of the cells in the main branches of the radula nerve (Fig. 5A) to the radula surface. Each axon entered a membranous sheath (the SRT) that lies just below the chitinous radula. The subradula tissue adheres to the matrix of support material that holds the teeth of the radula. Along the surface of the SRT, which is apposed to the support material (external surface), the finest axonal branches of the RM neurons terminated in varicose specializations (Fig. 5, B and C). The opposite (internal) surface of the SRT is the site of attachment of the buccal muscles of the odontophore implicated in control of radula closure, in particular the I4 muscle leaflets (Eales 1921; Howells 1942; Scott et al. 1991), and I5 (accessory radula closer, ARC) muscle (Cohen et al. 1978). No varicosities of B21 or B22 processes were observed along the internal surface of the SRT.

View larger version (66K): [in this window] [in a new window]   Fig. 5. Peripheral terminals of a B22 neuron filled with 5(6)-carboxyfluorescein and incubated with probenecid for 72 h. A: low power view of a buccal hemiganglion showing the soma (curved arrow) of the dye-filled B22 cell and its central processes. Straight arrow indicates the B22 axon that enters the radula nerve. B: medium power view of a portion of the SRT showing the peripheral processes of the B22 neuron and terminal varicosities on the face of the SRT that is apposed to the chitinous radula. The opposite face is the site of insertion of the I4 and I5 (ARC) muscles. Arrow indicates a region that was viewed at higher magnification and shown in C. C: high power view of the field of terminal varicosities in the subradula tissue (SRT) opposite the region of insertion of the I4 and ARC muscles onto the tissue. Arrow points to a region with a high-density of endings. Calibration bars: A, 50 µm; B, 2 µm; C, 1 µm.

Receptive fields

B21 and B22 were found to have bilateral mechanosensory receptive fields that generally coincided with their fields of terminal specializations that lie just below the radula surface. The receptive fields were confined chiefly to the chitinous radula surface and to the posterior surface of the odontophore immediately adjacent to the radula (Fig. 6, left). The fields generally included invaginations of the radula that form the median groove and the posterior (trailing) edges of the radula halves that contact food that is grasped by the radula. The sensitivity to tactile stimuli within the receptive field was variable. Loci of maximal sensitivity were found along, and within, the median groove and also along the trailing edge of the radula. The receptive fields of B21 and B22 were mapped in 14 preparations in which both cells were found. The overall shape and size of the fields of individual B21 and B22 neurons were similar, although B21 neurons typically exhibited larger areas of maximal sensitivity.

View larger version (32K): [in this window] [in a new window]   Fig. 6. Receptive fields (left) and response properties (right) of identified RM neurons B21 and B22. Top: the receptive field of a typical B21 neuron was found on the grasping surface of the radula and extended to the adjacent (posterior) surface of the odontophore. The radula surface was probed with hand-applied von Frey hairs of different diameters that provided calibrated punctate forces that ranged from 0.1 to 5.0 g. The thresholds for a response at given loci are indicated by the symbols shown in the bottom of the figure, with the large symbols representing the lower thresholds. Brief application (~0.1 s at arrow) of the lowest threshold stimulus (0.1 g) to the center of the receptive field, evoked a burst of spikes that was recorded from the soma of the B21 neuron (right, top trace). When the von Frey hair was applied continuously for 1 s (horizontal line), a rapidly adapting response was evoked (right, 2nd trace from top). Bottom: the typical B22 neuron had a receptive field similar to that of the B21 cell. The responses to punctate and maintained mechanical stimuli were similar. The response thresholds, however, measured within the receptive field of the B22 neuron showed that the lowest threshold responses were evoked at fewer loci compared with B21.

Response properties

B21 and B22 are low-threshold, rapidly adapting, mechanoafferent neurons that respond vigorously to punctate pressure stimuli applied to the surface of the radula (Fig. 6, right arrows). The responses recorded from each cell's soma consisted of bursts of A-spikes (axon spikes, which reflect action potentials that fail to invade the soma) having instantaneous spike rates ranging between 8 and 35 Hz. The number of evoked spikes and the spike frequency of the response bursts generally increased with the force of the stimulus, but repeated applications of the same stimulus often evoked variable responses. The cells exhibited sensory adaptation when the stimulus was maintained for 1 s (Fig. 6, right horizontal lines). There was never an "off-response" when maintained stimuli of 1 s were terminated (n = 100; 45 preparations). Moving stimuli, pulling stimuli, or mechanical stimuli that exerted shearing forces on the radula were ineffective in eliciting sensory responses, as were a variety of chemical stimuli including seaweed extracts and 4 M NaCl (Miller et al. 1994). Sensory responses of B21 could also be observed in the dissected odontophore preparation when mechanical stimuli were applied directly to the SRT that was stripped from the radula support matrix. Under these conditions, recordings from the somata of the cells revealed full spikes, as well as the smaller potentials that appeared to be blocked spikes (A-spikes) (Weiss et al. 1986; see Van Essen 1973). To study this in greater detail we used an electromechanical device to present brief, reproducible, suprathreshold mechanical stimuli. The stimuli were presented at 6 Hz, which is below the rate at which the cells typically discharge in response to a maintained stimulus. At this low rate, the stimuli evoked two types of A-spikes: small A-spikes with an amplitude of <10 mV or large A-spikes with an amplitude in excess of 20 mV (Fig. 7A). Moreover, when a succession of large A-spikes was evoked in B21, they invariably showed a further gradual increase in amplitude. In contrast to what is seen when full spikes are evoked by a maintained stimulus, the A-spikes failed to evoke discernible EPSPs in postsynaptic (follower) cells such as motor neuron B8a. When, however, RM neuron B21 was depolarized 20 mV from its resting membrane potential by injection of intracellular current, all the threshold mechanical stimuli applied at the periphery evoked full soma spikes (evident by the high spike amplitudes, 8-10 mV larger than large A-spikes, and the presence of hyperpolarizing afterpotentials), and EPSPs were recorded from follower neuron B8a. Axon spikes rapidly returned when the current injection was terminated. When the strength of the brief mechanical stimulus was increased well above threshold (Fig. 7B), the repeated stimuli at 6 Hz had an increased likelihood of evoking the large-type A-spike in B21, but the A-spikes still failed to evoke postsynaptic responses in B8a. These results suggest that action potentials generated in the periphery of the RM neurons may be vulnerable to branch block at peripheral, as well as at central branch points of the processes of the RM neurons. Moreover, the membrane potential at, or near, the soma may be important in regulating the blockade. The mechanism by which strong or maintained stimuli promote the generation of large-amplitude spikes is not known but may involve enhanced depolarizations at branch-block points, due to summation of depolarizing afterpotentials or electrical interactions (see SENSORY NEURONS) between mechanosensory neurons with overlapping receptive fields. In addition to responses evoked by peripheral mechanical stimuli contacting the radula surface, the cells are known to fire in response to contractions of the subradula tissue (Cropper et al. 1996).

View larger version (47K): [in this window] [in a new window]   Fig. 7. Amplitude of B21 A-spikes and the synaptic output of B21 evoked by mechanical stimulation of the subradula tissue were enhanced by depolarization of the cell's soma. Brief tactile stimuli applied to the SRT at threshold levels (generated by movement of a modified loudspeaker cone) (see Cropper et al. 1996) evoked responses in RM neuron B21 and excitatory synaptic potentials in a follower motor neuron B8a. The responses were enhanced by intrasomatic injection of depolarizing current into B21. A: when B21 was at resting potential (left), repetitive (6 Hz) punctate taps of the SRT at threshold levels produced either small (<10 mV) rapid depolarizations that represent blocked axon spikes (A-spikes) in B21, or larger (>20 mV), nearly full, action potentials. The 1st 6 responses are axon spikes. No discernible excitatory postsynaptic potentials (EPSPs) were recorded in motor neuron B8a. When B21 was depolarized 20 mV (middle, left of center), each tactile stimulus applied to the SRT evoked a large A-spike or full-blown action potential in B21 and one-for-one facilitating EPSPs in neuron B8a. When the membrane potential of B21 was returned to resting level (middle, right of center), each tactile stimulus evoked either a small A-spike or a large A-spike, but no EPSPs were evoked in neuron B8a. Depolarization of B21 again (right) resulted in large A-spikes or full spikes in B21 and facilitating EPSPs in neuron B8a. B: repetitive (6 Hz) punctate stimulation of the SRT at suprathreshold levels evoked large A-spikes in B21, but no discernible EPSPs in neuron B8a.

Synaptic connectivity

In addition to their activation by mechanostimulation of the radula, B21 and B22 also receive synaptic input from at least four types of neuronal sources (Fig. 20 is a summary diagram of the neurons found interconnected with B21 and B22). Many of the connections of B21 and B22 consist of nonrectifying electrical synapses. Depending on the context, these connections can be considered as either input or output pathways.

SENSORY NEURONS. One type of synaptic interconnection to B21 and B22 includes electrical connections with other SCP-immunoreactive RM neurons. As described in a previous report (Miller et al. 1994), SCP-immunoreactive neurons are part of an electrically coupled network of sensory cells. Neurons B21 and B22 were found to make strong electrical synaptic connections to each other (Fig. 8, A and B), and to their homologues in the contralateral buccal hemiganglion (Fig. 8, A-C). Moreover, many other monopolar and bipolar neurons comprising the SCP-immunoreactive RM cell clusters make weak electrical synaptic connections to B21 and B22. For the majority of cases of pairs of cells tested (14 of 25 or 56%), the electrical coupling ratio was similar in both directions of current flow.

View larger version (13K): [in this window] [in a new window]   Fig. 8. Electrical synaptic connections between B21, B22, and interganglionic interneuron, B19. A: injection of depolarizing and hyperpolarizing current pulses into the soma of a left B21 neuron (LB21) produced depolarizing or hyperpolarizing electrotonic potentials in the ipsilateral B22 cell (LB22), in the contralateral B21 neuron (RB21), and in the contralateral B19 cell (RB19). B: similarly, injection of depolarizing and hyperpolarizing current pulses into the soma of a left B22 RM neuron (LB22) produced depolarizing or hyperpolarizing electrotonic potentials in the ipsilateral B21 neuron (LB21), and also in the contralateral B21 (RB21) and contralateral B19 (RB19) neurons, respectively. Injection of depolarizing and hyperpolarizing current pulses into the soma of the contralateral right B21 RM neuron in C and right B19 neuron in D produced corresponding electrotonic potentials in the other cells that were monitored simultaneously. The values of the calibration bars for the pairs of traces are given at the right of the traces, with the thicker bars indicating the cell in which current was injected.

INTERNEURONS. A second source of synaptic interconnection to identified neurons B21 and B22 is the bilateral pair of interganglionic interneurons, B19s. Each B19 sends its main axon from one buccal hemiganglion to the cerebral ganglion via the ipsilateral cerebral-buccal connective (Fig. 2A). B19 provides bilateral chemical synaptic input to cerebral-to-buccal interneurons (e.g., CBI-2), and to bilateral groups of cerebral motor neurons that control the lips and the extrinsic muscles of the buccal mass (Chiel et al. 1986; Rosen et al. 1991). Each B19 makes a strong electrical synaptic connection to RM neuron B21 in the ipsilateral buccal hemiganglion (Fig. 8, C and D). Injection of hyperpolarizing or depolarizing current pulses into the soma of a B19 neuron produced hyperpolarizing or depolarizing electrotonic potentials in the ipsilateral and contralateral B21 and B22 neurons, respectively (Fig. 8, A-D). The responses in the ipsilateral B21 and B22 cells were always greater than those in the contralateral cells (n = 20 pairs).

View larger version (8K): [in this window] [in a new window]   Fig. 9. Electrical synaptic connection between identified RM neuron B21 and buccal interneuron B64. In normal ASW, injection of hyperpolarizing current pulses into the B21 soma produced hyperpolarizing electrotonic potentials in the B64 cell (top left trace). Similarly, injection of depolarizing constant current pulses into the soma of B21 produced a depolarizing electrotonic potential in neuron B64 and often an active response that caused B64 to fire a burst of action potentials that outlasted the electrotonic potential (top right trace). The burst of action potentials failed to produce significant depolarizing electrotonic potentials in B21, although a weak electrotonic potential could be detected. Calibrations as shown.

MULTIFUNCTION NEURONS. Another source of synaptic input to B21 and B22, and the only known source of inhibitory chemical synaptic input to one of the RM neurons, is provided by multifunction neurons B4 and B5. B4 and B5 make inhibitory, chemical, synaptic connections to B21, but not B22 (Fig. 10). Repetitive spikes elicited in these cells produce nondecrementing, fast IPSPs in RM neuron B21 (Fig. 10, A1, B1, and B4) and small, depolarizing electrotonic potentials in neuron B22 (Fig. 10A). Indicative of monosynaptic connections, the synaptic potentials follow the presynaptic spikes, one for one, with constant delay, even at high frequencies of firing of B4 or B5. In addition, the IPSPs in B21 were eliminated in ASW solutions containing a high concentration of Mg²⁺ and a low concentration of Ca²⁺ (Fig. 10B2) but were not blocked by raising the concentration of Mg²⁺ and Ca²⁺, which would increase the threshold of interposed interneurons (Fig. 10B3).

View larger version (20K): [in this window] [in a new window]   Fig. 10. Putative monosynaptic connections of multifunction neuron B4 to RM neurons B21 and B22, and to radula closer motor neuron B8a. A1: in normal artificial seawater (ASW), intracellular injection of depolarizing current into B4, sufficient to elicit a train of action potentials, evoked one-for-one IPSPs in identified RM neuron B21, and one-for-one electrotonic potentials in RM neuron B22. A2: injection of hyperpolarizing current into B4 produced a hyperpolarizing potential in B22, and a barely detectable shift of membrane potential in B21. B: the monosynapticity and chemical nature of the inhibitory connections of B4 to B8a and B21. B1: in normal ASW, a fast IPSP in B21 and B8a was evoked by each spike in B4. B2: in the presence of an ASW solution containing an increased concentration of Mg2+ (4 times) and decreased concentration of Ca2+ (0.5 times), the fast IPSPs were no longer detectable, and only a small tonic depolarization in B21 was evident, probably due to weak electrical coupling between B4 and B21. B3: in the presence of an ASW solution containing an increased concentration of Mg2+ (2 times) and of Ca2+ (5 times), which blocks the actions of interposed neurons, one-for-one fast IPSPs in B8 and B21 were still seen, without any obvious change in synaptic delay. B4: on return of the preparation to normal ASW, there was recovery of the fast IPSPs evoked by B4 in B21 and B8a. All recordings were made from cells located in the same buccal hemiganglion.

MOTOR NEURONS. B21 and B22 were found to have synaptic interconnections with many motor neurons found in the ventral motor neuron cluster of each buccal hemiganglion. The investigation focused on those identified motor neurons that innervated muscles of the odontophore, with the exception of a cell that was an example of a motor neuron with a strong electrical connection to the RM sensory neurons. The motor neurons on the ventral surface that receive RM connections appear to be involved in one or another aspect of the radula closure phase of buccal motor programs. One pair of motor neurons consists of previously identified neurons B8a and B8b (Church and Lloyd 1991; Gardner 1977; Morton and Chiel 1993) that innervate portions of the I4 muscle. A second pair consists of motor neurons B15 and B16 that innervate the I5 or ARC muscle (Cohen et al. 1978). We have identified a third type of cell that appears to be involved in radula closure. This cell we provisionally identify as B82. B82 is a motor neuron that innervates the anterodorsal I1/I3 buccal muscles that are located over the jaws. Elsewhere (Rosen et al. 2000) we show that during a buccal motor program driven by the firing of CBI-2, B82 fires in a phase similar to that of radula closer motor neurons such as B16 and B8a/b. The type of synaptic potential that each B21 and B22 evoked in each of their follower motor cells was different (see summary in Fig. 20). Some of the synapses are electrical, others are chemical, and some are mixed electrical-chemical. As described in detail in the following sections, some of the sensory cell connections exhibit very marked homosynaptic plasticity at particular identified motor neurons.

RADULA CLOSURE MOTOR NEURONS B8A AND B8B. Two motor neurons with axons in the radula nerve have been implicated in the control of radula closure (Church and Lloyd 1994; Morton and Chiel 1993). Consistent with earlier descriptions, we refer to the cells as B8 neurons, namely B8a and B8b. B8a is the most lateral, and B8b the next most lateral, ventral cluster motor neuron that receives an IPSP from multifunction neuron B4/5 (Fig. 10B). In our experiments, the sister cell B8b was found in the vicinity of B8a in 30% of 12 preparations examined. We have found that B8a innervates portions of the leaflets of the I4 muscle, that insert into the ventral regions of the odontophore and attach to the SRT in the vicinity of the grasping surface of the radula. B8a/b was identified by observing that it produced closure of the radula in the odontophore preparation. B8a/b was also identified by the characteristic EPSP it receives from identified RM neuron B21 (Figs. 1113). Both B8a and B8b receive a nondecrementing EPSP from neuron B21, which exhibits marked synaptic facilitation and summation (Figs. 11-13), such that brief bursts of B21 spikes, within normal physiological rates, are sufficient to trigger action potentials in the B8a and B8b motor neurons. The synaptic facilitation is most clearly evident when B21 is fired between 8 and 12 Hz (Fig. 12). An individual B21 neuron is capable of producing facilitating EPSPs in both the ipsilateral and contralateral B8a motor neurons, although the response of the ipsilateral follower cell is consistently larger than that of the contralateral follower. The latter finding was evident when simultaneous recordings were made from the left and right B21 neurons, as well as from the left and right B8a motor neurons (Fig. 13, A and B). The result is consistent with the morphological findings indicating that the medial axon of each B21 neuron projects to the contralateral buccal hemiganglion and that the number and density of putative terminal branches of the B21 cells are greater in the region closest to the soma of B21 (Fig. 4).

View larger version (29K): [in this window] [in a new window]   Fig. 11. EPSPs evoked by RM neuron B21 in I4 motor neurons B8a and B8b. Motor neurons were identified by their characteristic positions and the IPSPs they receive from multifunction neuron B4. B8a and B8b were also identified by visualization of their main axons in the radula nerve, after dye-filling at the end of the experiment. Repetitive, brief (10 ms), intracellular stimulation (16 Hz) of B21, evoked EPSPs in B8a and B8b that summated and were sufficient to evoke motor neuron spikes.

View larger version (17K): [in this window] [in a new window]   Fig. 12. Homosynaptic modulation of the putative chemical synaptic connections of RM neuron B21 to motor neurons B8a and B15. Control recordings were made of the activity of motor neuron B82. A: B21 was stimulated with repetitive intracellular current pulses so that it fired at 4 Hz. Electrotonic potentials were evoked in neurons B15 and B82, but not in B8a. B: firing B21 at 6 Hz produced low-amplitude chemical EPSPs in B8a and electrotonic potentials in B15 and B82. C: firing B21 at 8 Hz produced facilitating EPSPs in B8a and a slow, chemical EPSP that accompanied the one-for-one electrotonic potentials evoked in B15. Only electrotonic potentials were recorded in B82. D: firing B21 at 10 Hz produced facilitating EPSPs that evoked 3 spikes in B8a (the rising phase of the spikes were filtered by the pen recorder) and a marked slow EPSP in neuron B15. Responses were recorded in a preparation that was bathed in an ASW solution containing high divalent cations (3 times normal Mg2+ concentration, 3 times normal Ca2+ concentration). Calibrations as shown.

View larger version (20K): [in this window] [in a new window]   Fig. 13. Bilateral synaptic connections of RM neuron B21. A: intrasomatic current injection into the right B21 neuron (RB21), sufficient to elicit a train of RB21 spikes, evoked facilitating EPSPs in the ipsilateral right, as well as the contralateral left, B8a neuron (RB8a and LB8a neurons, respectively). Evoked electrotonic potentials were also recorded in the left B21 neuron (LB21). B: conversely, current injection into the left B21 neuron (LB21), sufficient to elicit a train of LB21 spikes, evoked facilitating EPSPs in the ipsilateral left, as well as the contralateral right, B8a neuron. Again, evoked electrotonic potentials were also recorded in the right B21 neuron. In both A and B, the amplitude of the EPSPs evoked in the ipsilateral B8a neuron was greater than the amplitude of the EPSPs evoked in the contralateral B8a neuron.

ARC MOTOR NEURONS B15 AND B16. B15 and B16 are the two motor neurons that innervate the ARC (or I5) muscle (Brezina et al. 1994; Cohen et al. 1978; Probst et al. 1994; Vilim et al. 1996, Weiss et al. 1992; Whim and Lloyd 1990). When a single B21 RM neuron was intracellularly stimulated so that it fired a burst of action potentials, one for one EPSPs were evoked in neuron B16 (Fig. 14, A and C). B15 exhibited a multiphasic EPSP consisting of rapid potentials that arose on a slowly rising tonic depolarization that slowly decayed when the firing of B21 was terminated (Fig. 14A). Responses in B15 and B16 were observed in the absence of any responses in B4 (Fig. 14A), suggesting that B21 was probably not evoking a buccal motor program that, in turn, evoked synaptic inputs to B15 and B16. To determine which, if any, of the evoked potentials was due to chemical synaptic transmission, the preparations were bathed in a high Mg²⁺, low Ca²⁺ ASW solution. When B21 was fired under these conditions, the synaptic potentials in B16 were completely eliminated (Fig. 14B). The gradually increasing slow synaptic potential in B15 was eliminated, but the cell still exhibited fast potentials and a tonic depolarization, reflecting electrical coupling. The existence of electrical coupling was supported by injecting hyperpolarizing current pulses into B21 and B22, which resulted in hyperpolarizing electrotonic potentials in neuron B15 (see Fig. 15, C and D). Using the above protocol, B22 was found to produce a weak electrical EPSP in neuron B16. B22 was also found to be electrically coupled to neurons B4 (Fig. 10A), B15 (Fig. 15, B and D) and B19 (Fig. 8, B and D).

View larger version (22K): [in this window] [in a new window]   Fig. 14. Putative monosynaptic connections of RM neuron B21 to ARC motor neurons B15 and B16. A: in normal ASW, constant current intracellular stimulation of B21 (horizontal line), sufficient to elicit a train of spikes, evoked fast EPSPs in motor neuron B16, combined fast and slow EPSPs in neuron B15, and no response in neuron B4. B: when the preparation was bathed in a high Mg2+ (4 times normal concentration), low Ca2+ (0.5 times normal concentration) ASW solution and B21 was again fired, the fast EPSP evoked in B16, and the slow EPSP in B15, were abolished, whereas an evoked depolarizing electrotonic potential in B15 remained. C: when the preparation was next bathed in a high divalent cation ASW solution (2 times normal Mg2+ concentration, 5 times normal Ca2+ concentration) an elicited burst of B21 spikes evoked constant latency, fast EPSPs in B16 and a slow EPSP in B15 that added to the electrotonic potentials evoked in that cell. Calibrations as shown.

View larger version (11K): [in this window] [in a new window]   Fig. 15. Outputs of RM neurons B21 and B22 converge on motor neuron B15. A: constant current, depolarizing, intracellular stimulation of B21 (horizontal line), sufficient to elicit a train of spikes, evoked a complex EPSP in neuron B15 and an electrotonic potential in neuron B22. No response was observed in multifunction neuron B4. Note that at this gain and sweep speed, small, fast electronic potentials in B21 and B22 cannot be resolved in the records. B: constant current, depolarizing intracellular stimulation of B22 (horizontal line), sufficient to elicit a train of spikes, evoked electrotonic potentials in neurons B15, B21, and B4. C: injection of a hyperpolarizing current pulse into B21 (horizontal line) produced a hyperpolarizing electrotonic potential in neurons B15 and B22, but not in B4. D: injection of a hyperpolarizing current pulse into B22 (horizontal line) produced hyperpolarizing electrotonic potentials in neurons B4, B15, and B21.

B82 NEURONS. Motor neuron B82 is one of a pair of bilateral, medium-large neurons found among cells comprising the ventral cluster of motor neurons in the buccal ganglion (see Fig. 2F of Rosen et al. 2000). It is accessible from the rostral surface and is found near the lateral edge of the cluster in the vicinity of B8a and B8b. B82 sends an axon out ipsilateral buccal nerve 2, and its firing produces contraction of middorsal I1 muscles of the buccal mass. The contraction results in a shortening of the muscles above the jaws and in a forward movement of the pharyngeal tissue comprising the dorsal wall of the buccal cavity. Buccal nerve 2 trifurcates as it enters the buccal mass. The B82 axon passes in the dorsal branch of the trifurcation (termed BNc) (Warman and Chiel 1995). B82 does not receive any input from B4, but it does receive substantial input from unidentified interneurons, which cause it to fire rhythmically during buccal motor programs. B21 makes a nonrectifying electrical synapse with B82 (Fig. 16) as does B22. A neuron, previously identified as B45 produces jaw shortening (Church and Lloyd 1991, 1994) and is in a similar position to B82. B45, however, has a relatively small cell body (P. J. Church, personal communication) and sends its axon into buccal nerve 3 rather than 2. 

View larger version (13K): [in this window] [in a new window]   Fig. 16. Electrical synaptic connection of identified RM neuron B21 to motor neuron B82 in the buccal ganglion. In normal ASW, injection of hyperpolarizing and depolarizing current pulses into the soma of neuron B82 produced hyperpolarizing and depolarizing electrotonic potentials, respectively, in the ipsilateral B21 neuron. Similarly, injection of hyperpolarizing and depolarizing current pulses into the soma of the B21 cell produced hyperpolarizing or depolarizing electrotonic potentials in the ipsilateral B82 neuron. These responses persisted in a seawater solution (4 times normal Mg2+; 0.5 times normal Ca2+) that blocks chemical synaptic transmission. Calibration bars are shown to the right of each trace.

Homosynaptic neuromodulation of the synaptic connections of B21

Two of the chemical synaptic connections of B21 show intrinsic (homosynaptic) modulation, that results in larger EPSPs (facilitation) when B21 is fired at increasing rates of firing. As shown in Fig. 12, when B21 was stimulated with repetitive intracellular current pulses so that it fired at 4 Hz (Fig. 12A), fast electrotonic potentials were evoked simultaneously in neurons B15 and B82 (2 cells that make electrical synapses with B21), but not in B8a (a cell that receives chemical synaptic input from B21). Firing B21 at 6 Hz (Fig. 12B) produced small, chemical EPSPs in B8a and electrotonic potentials again in B15 and B82. However, when B21 was fired at 8 Hz (Fig. 12C), it produced strongly facilitating EPSPs in B8a and a slowly rising EPSP that accompanied the one for one electrotonic potentials evoked in B15. Only electrotonic potentials were seen in B82. Firing B21 at 10 Hz (within its normal frequency range) produced facilitating EPSPs in B8a that exceeded B8a's spike threshold and also produced a marked slow EPSP in neuron B15 that was particularly evident during the latter part of the EPSP and during the decay phase (Fig. 12D). Elsewhere we present evidence that the transmitter release from B21 is dependent on its baseline membrane potential (Rosen et al. 2000).

Although the chemical synaptic connection of B21 to B8a can exhibit very marked facilitation, the duration of the facilitation appeared to be remarkably brief, persisting for <1 s (Fig. 17) as studied in three preparations.

View larger version (39K): [in this window] [in a new window]   Fig. 17. Facilitation and very short-duration posttetanic potentiation of the EPSP B21 evoked in B8a. For each run, single spikes were initially evoked in B21 by brief (8 ms) intracellular current pulses delivered at 2 Hz. At this rate, no evoked EPSP could be detected in B8a. A series of high-frequency (tetanizing) pulses was then used to evoke spikes in B21 at 20 Hz for 1 s. These parameters were selected because they are within the range of firing of B21 observed when the receptive field of B21 was mechanically stimulated. The high-frequency spikes resulted in facilitating EPSPs in B8a. Following the high-frequency spikes, B21 was returned to firing at 2 Hz. If a test spike in B21 occurred within 200 ms after the tetanus (arrow, A) an evoked EPSP could be seen, indicating that the EPSP was potentiated relative to the pretetanization condition, in which B21 firing did not evoke an observable EPSP. If, however, the test pulse occurred at an interval of 400 ms (B), an evoked EPSP was no longer detectable, indicating that the potentiation was no longer present or was minimal.

B21 activity affects the excitability of ARC motor neuron B15

Previous work has shown that the excitability of B15 is increased following exogenous applications of the peptide SCP, which modulates several ionic currents and elevates levels of second messengers in the cell, including cAMP (Sossin et al. 1987; Taussig et al. 1989). Because B21 contains SCP, we tested whether the slow EPSP produced by B21 in neuron B15 might be associated with an alteration of the excitability of B15. B15 was repeatedly depolarized by intracellular current pulses of fixed duration and intensity so that a constant number of action potentials were elicited. Firing of B21 resulted in an increase in the number of B15 spikes that were elicited by the current pulses (Fig. 18, A and C). The effect was strongest when current pulses in B15 were presented during the tail of the evoked slow EPSP in B15 (Fig. 18A). At least part of the effect may be due to membrane depolarization of B15 because depolarization of B15 alone can increase its excitability (Fig. 18B). Nevertheless, even when stimulation of B21 did not produce any obvious long-lasting membrane depolarization in the B15, a small excitability increase in the cell could still be observed (Fig. 18C).

View larger version (26K): [in this window] [in a new window]   Fig. 18. B21 modulates the excitability of motor neuron B15. A: at regular intervals, B15 was intracellularly stimulated with a depolarizing, constant current pulse sufficient to elicit a train of action potential spikes (controls, left and right). When B21 was fired before B15 stimulation (top middle), so that the slow EPSP B21 evoked in B15 coincided with the intracellular stimulus, there was an increase in the number of spikes elicited in B15. B: to test whether the change in B15 excitability was due in part to a shift in membrane potential, B15 was depolarized by several millivolts during the intracellular test pulse. An increase in the number of action potentials was observed (middle) compared with controls (left and right). C: when B21 was fired before B15 stimulation (bottom middle), so that the slow EPSP that B21 evoked in B15 decayed and no longer coincided with the intracellular stimulus (middle), the number of spikes elicited in B15 still increased in number. Calibrations as shown.

Outputs to pattern-generating neurons

In a previous section, B21 was shown to make a nonrectifying electrical synapse with interganglionic interneuron B19 (Fig. 8). The synapse allows for the possibility of bidirectional communication between the cells. In a companion paper we show that activity in B19 can affect the chemical synaptic output of B21 to motor neurons B8a and B15 (Rosen et al. 2000). B19 is a powerful regulator of the biting command-like interneuron CBI-2 (Rosen et al. 1991) and produces a complex inhibitory-excitatory monosynaptic EPSP that contributes to the timing of the firing of CBI-2, and thereby can contribute to pattern generation. Because B21 affects B19, we predicted that firing of B21 might have an indirect effect on CBI-2 and consequently might have an effect on pattern generation. When CBI-2 was intracellularly stimulated with a maintained, constant current pulse, such that the rate of the spiking that was elicited was insufficient to drive a motor program, the additional firing of a burst of B21 spikes evoked several cycles of a motor program (Fig. 19). The effect was seen in four of four preparations that were tested and suggests the B21 might provide important input to the feeding pattern generating circuitry.

View larger version (29K): [in this window] [in a new window]   Fig. 19. B21 activity can affect the feeding pattern-generating circuitry. In an odontophore preparation bathed in normal ASW, a constant, intracellular current pulse (horizontal line, top) was used to elicit a rate of spiking in command-like interneuron CBI-2 that was insufficient to drive a motor program. When, during the course of the CBI-2 stimulation, B21 was also stimulated (horizontal line, bottom), several cycles of a motor program were initiated that incorporated the phasic activity of multifunction neuron B4 and buccal-to-cerebral interneuron B19. Note that each cycle of the motor program that is elicited produced periodic inhibitory and excitatory synaptic input to B21.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Identification of unique SCP-containing RM neurons in the buccal ganglion

This study establishes the identity of two neurons (B21 and B22) that are constituents of the cluster of SCP-containing RM neurons found on the rostral surface of each buccal hemiganglion (Lloyd et al. 1985; Miller et al. 1994). It appears as if there is only one, or at most two, B21 neurons in each hemiganglion. However, B22 shares some properties with other smaller, bipolar, SCP-immunoreactive cells.

Comparisons with other mechanoafferents in Aplysia

B21 and B22 exhibit rapidly adapting responses to mechanical stimuli that are similar to other types of mechanoreceptor neurons that have been studied in the cerebral (Chiel et al. 1990; Rosen et al. 1979, 1982; Weiss et al. 1986), abdominal (Byrne et al. 1974; Dubuc and Castellucci 1991), and pleural ganglia (Walters et al. 1983) of Aplysia. Two types contain a peptide that may function as a cotransmitter. One contains the neuropeptide sensorin A (Brunet et al. 1991; Gapon and Kupfermann 1996), whereas RM neurons contain the neuropeptide SCP and do not appear to contain sensorin A (Miller et al. 1994).

The two types of peptidergic mechanoafferent cells differ in several respects. Elsewhere we provide evidence that diverse synaptic inputs received by some RM cells (Rosen et al. 2000) provide rapid modulation of the magnitude of the synaptic outputs of the cells. The output of the sensorin mechanoafferents also can be modulated by heterosynaptic input, but this modulation occurs over a relatively long time course and does not appear to involve fast, conventional synaptic inputs (Byrne and Kandel 1996; Rosen et al. 1989). Furthermore, all the known synaptic outputs of the RM cells are nondecrementing or facilitating, whereas the sensorin sensory cells all produce decrementing synaptic potentials, although high-frequency firing of these cells can result in posttetanic potentiation (Walters and Byrne 1984). The excitatory chemical synapses between B21 and motor neurons B8a and B8b exhibit a remarkably prominent synaptic facilitation (homosynaptic modulation) that strongly depends on the frequency of firing of B21, but exhibits little posttetanic potentiation. The facilitation might provide a mechanism by which very weak, or short-duration, sensory information may be prevented from having an effect on the CNS. The short duration of posttetanic potentiation may function to diminish cycle to cycle variation of the synaptic potential.

Possible functions of B21 and other RM neurons

The RM peripheral endings terminate in the subradula tissue (Miller et al. 1994) that underlies the tooth-lined chitinous radula, the structure that grasps and releases food during feeding movements. The RM receptive fields are located chiefly on the grasping surfaces of the radula. The relatively low firing thresholds for tactile stimuli contacting the radula make RM neurons suitable for detecting the position or movement of the radula halves and/or detecting the presence of food or other objects contacting the radula. During feeding, the grasping surfaces of the radula open and close, and in addition, they protract and retract between the jaws and the esophagus. Mechanosensory information arising from the strategically located receptive fields of B21 and B22 can be conveyed monosynaptically to various identified motor neurons and interneurons that control the muscles producing radula closure (Fig. 20). The motor neurons include B8a and B8b, which control the leaflets of the I4 muscle; B15 and B16, which control the I5 (ARC) muscle; and B82, which controls dorsal muscles of the buccal mass that act conjointly with radula movements to move food between the jaws and the esophagus. These muscles are involved in both ingestive behaviors (e.g., biting, swallowing), and egestive behaviors (rejection), which can be elicited by nonfood stimuli contacting the radula.

View larger version (36K): [in this window] [in a new window]   Fig. 20. Summary diagram of the synaptic connections of RM neuron B21 (A) and B22 (B) to identified neurons in the feeding pattern-generating circuitry. Mechanical stimulation of the radula (bottom) generates action pote