The Neuronal Basis of Feeding in the Snail, Helisoma, with Comparisons to Selected Gastropods

Progress in Neurobiology 63 (2001) 383–408 www.elsevier.com/locate/pneurobio

The neuronal basis of feeding in the snail, Helisoma, with comparisons to selected gastropods

A. Don Murphy

Department of Biological Sciences and Laboratory of Integrati6e Neuroscience, Uni6ersity of Illinois at Chicago, Chicago, IL 60607, USA

Abstract

Research on identified neurons during the last quarter century was forecast at a conference in 1973 that discussed ‘neuronal mechanisms of coordination in simple systems.’ The focus of the conference was on the neuronal control of simple stereotyped behavioral acts. Participants discussing the future of such research called for a comparative approach; emphasis on structure– function interactions; attention to environmental and behavioral context; and the development of new techniques. Significantly, in some cases amazing progress has been made in these areas. Major conclusions of the last quarter century are that so-called simple behaviors and the neural circuitry underlying them tend to be less simple, more flexible, and more highly modulated than originally imagined. However, the comparative approach has, as yet, failed to reach its potential. Molluscan preparations, along with arthropods and annelids, have always been at the forefront of neuroethological studies. Circuitry underlying feeding has been studied in a handful of species of gastropod molluscs. These studies have contributed substantially to our understanding of sensorimotor organization, the hierarchical control of behavior and coordination of multiple behaviors, and the organization and modulation of central pattern generators. However, direct interspecific comparisons of feeding circuitry and potentially homologous neurons have been lacking. This is unfortunate because much of the vast radiation of the class Gastropoda is associated with variations in feeding behaviors and feeding apparatuses, providing ample substrates for comparative studies including the evolution of defined circuitry. Here, the neural organization of feeding in the snail, Helisoma, is examined critically. Possible direct interspecific comparisons of neural circuitry and potentially homologous neurons are made. A universal model for central pattern generators underlying rasping feeding is proposed. Future comparative studies can be expected to combine behavioral, morphological, electrophysiological, molecular and genetic techniques to identify neurons and define neural circuitry. Digital resources will undoubtedly be exploited to organize and interface databases allowing illumination of the evolution of homologous identified neurons and defined neural circuitry in the context of behavioral change. © 2001 Elsevier Science Ltd. All rights reserved.

Keywords: Aplysia; Heterobuccal nerve; Ventrobuccal nerve; Postsynaptic nerve

Contents

1. Introduction ...... 384

2. Investigations of oral behaviors and underlying neural mechanisms in gastropod molluscs .... 385

Abbre6iations: A, Aplysia; aj, anterior jugalis; B, buccal; BC, buccal commisure; BCN1, buccocerebral neuron, type 1; C, Clione; CBC, cerebrobuccal connective; con, Contralateral; CPG, central pattern generator; DBN, dorsobuccal nerve; E, excitatory; EPSP, excitatory postsynaptic potential; ET, esophageal trunk; GN, gastric nerve; H, Helisoma; HBN, Heterobuccal nerve; I, inhibitory; in, inﬂuential; Ip, ipsilateral; IPSP, inhibitory postsynaptic potential; L, Lymnaea; LBN, laterobuccal nerve; MN, motor neuron; nor, nitric oxide responsive; P, Planorbis; PBN, posterobuccal nerve; pj, posterior jugalis; PSP, postsynaptic potential; rad tens, radular tensor; RN, radular nerve; S1, subunit 1; S2, subunit 2; S3, subunit 3; SN, salivary nerve; slrt, supralateral radular tensor; SRT, subradular tissue; VBN, ventrobuccal nerve. E-mail address: [email protected] (A.D. Murphy).

2.1. Helisoma tri6ol6is as a neuroethological model system ...... 385 2.2. Feeding behavior and its anatomical basis in basommatophoran snails...... 385

3. Neurophysiological correlates of feeding in Helisoma...... 386 3.1. The original model of the Helisoma buccal CPG ...... 388 3.2. A triphasic buccal motor pattern underlies feeding behavior ...... 388 3.3. The current model of the multifunctional buccal CPG ...... 390 3.4. Plasticity of organization in a multifunctional CPG ...... 391 3.5. Identiﬁed CPG interneurons and motor neurons...... 391 3.6. Evidence that the triphasic standard buccal motor pattern mediates the typical feeding behavior ...... 392 3.7. Regurgitation ...... 393 3.8. Transmitters of CPG interneurons ...... 394 3.9. Sensory and modulatory pathways regulating the buccal CPG...... 395 3.10. Higher order modulation of the CPG...... 395

4. Comparisons of gastropod neurons and feeding circuitry ...... 396 4.1. Biological and methodological complications for interspecific comparisons of gastropod feeding ...... 396 4.1.1. Incomplete data ...... 396 4.1.2. Interspecific diversity ...... 397 4.1.3. Intraspecific diversity ...... 397 4.1.4. Non-congruence of interneuronal and motor neuronal patterns ...... 398 4.2. Specific buccal motor patterns ...... 398 4.3. Interspecific comparisons of CPG interneurons and feeding motor neurons ...... 399 4.4. Protraction phase interneurons ...... 401 4.5. Motor and influential neurons ...... 402 4.6. Retraction phase interneurons ...... 402 4.7. Hyper-retraction phase interneurons ...... 403

5. Concluding remarks ...... 404

Acknowledgements ...... 405

References ...... 405

1. Introduction attention to the environmental and behavioral context in which the experimental subject normally functions; There was a remarkable symposium, titled ‘Neuroid and the development of new techniques. In addition, and Neuronal mechanisms of Coordination in Simple the marriage of genetics and neurobiology was presaged Systems’, held at the annual meeting of the American at the conference (Hoy, 1974; Ikeda and Kaplan, 1974; Society of Zoologists in December, 1973 (Kammer and Levinthal, 1974), though the recent explosion of molec- Westfall, 1974). The symposium summarized the state ular biology and its effects on neurobiology could not of research, a quarter century ago, on mechanisms of be anticipated. coordination in a number of systems of identifiable In the ensuing quarter century, significant progress neurons as well as in non-neural systems. The focus was has been, and continues, to be made in each of these on the neuronal (and non-neuronal) control of simple areas. A consensus has arisen that the behavioral out- ‘stereotyped’ behavioral acts, or, in the ethological lan- puts of such simple systems is not so stereotyped as first guage of the time, the neural bases underlying ‘fixed thought and that even simple neural circuitry is highly action patterns’ (e.g. Kater, 1974; Kristan, 1974; Lent, modulated (e.g. Harris-Warrick and Marder, 1991; 1974; Mayeri et al., 1974; Naitoh, 1974; Selverston, Marder and Calabrese, 1996). Progress in neurogenetics 1974; Spencer, 1974; Stein, 1974). Hierarchical interac- has exploded. New staining and recording techniques tions among potential behaviors and the modulation of and application of molecular biology techniques have behavior by learning were addressed (Davis et al., illuminated structure–function relationships at levels 1974). Participants discussed the future of such re- unimagined in 1973. Perhaps the area of inquiry pro- search. They called for, a comparative approach; em- posed at the symposium 25 years ago that has fulfilled phasis on structural and functional interactions; its potential least is the emphasis on comparative stud- A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 385 ies and, in particular, what such studies might tell us applicable neural organization underlying gastropod about the evolution of neural systems and behaviors. I feeding behaviors has been proposed previously. The will suggest below that the potential for such studies Helisoma model will provide a reference point for exam- has not decreased and that the comparative study of the ining the adaptive radiation of buccal circuitry. These neurobiology of molluscan feeding is pregnant with gastropod buccal CPGs appear to be comprised of possibilities. multiple interneuronal subunits, much like the unit CPGs hypothesized to control joint movements during verte- brate locomotion (cf. Grillner and Wallen, 1985). The 2. Investigations of oral behaviors and underlying individual CPG subunits are conditional neuronal oscil- neural mechanisms in gastropod molluscs lators, which can be independently activated. The activities of CPG subunits too can be linked functionally in Investigations of the neuronal control of feeding in different combinations and in different temporal se- several different opisthobranch and pulmonate gas- quences to produce different behaviors. Both physiolog- tropod molluscs began in earnest in the early 1970s (e.g. ical state and acute sensory stimuli can modulate the Levitan et al., 1970; Gardner, 1971; Rose, 1971; Berry, organization of the CPG to produce the appropriate 1972; Kater, 1974; Kupfermann, 1974; Siegler et al., behavior. 1974). Opisthobranchs and pulmonates have a number In an effort to facilitate further comparative and of large neurons that are identifiable uniquely in each evolutionary studies, similarities in structure, function individual of a species. These uniquely identifiable neu- and pharmacology of certain potentially homologous rons have facilitated greatly the physiological and neurons will be addressed in those species, in which the anatomical circuit analyses. buccal CPG organization has been analyzed most (He- In addition, the presence of uniquely identifiable lisoma, Planorbis, Lymnaea and Aplysia). Some of the neurons indicates heritable developmental programs that difficulties involved in interspecific comparisons will be generate these neurons, thus offering the possibility for addressed and a plea made for more comparative studies. analyses of the adaptive radiation of feeding circuitry. Though the potential for comparative studies of gas- 2.1. Helisoma tri6ol6is as a neuroethological model tropod feeding circuitry to illuminate aspects of the system evolution of homologous neurons and neural circuits was recognized in the early 1970s (cf. last paragraph of Kater, About 1970, an extensive survey of the gross anatomy 1974), that potential remains largely unrealized. Feeding of the buccal ganglia and associated structures involved behaviors and their neural bases have now been more or in feeding in various gastropods was made by S.B. Kater less studied intensively in several gastropod species such (personal communication). The purpose of this survey as Aplysia, Clione, Helix, Helisoma, Incilaria, Limax, was to find the preparation offering the best combination Na6anax, Planorbis, Planorbarius, Pleurobranchea and of visual identifiability of buccal neurons and robustness Tritonia (Bulloch and Dorsett, 1979; Brace and Quicke, of feeding behavior in semi-intact preparations. An 1981; Cohan and Mpitsos, 1983; Peters and Altrup, 1984; albino strain of Helisoma tri6ol6is was selected and Arshavsky et al., 1988a,b,c, 1989; Cappell et al., 1989a,b; developed as a neuroethological preparation for examin- Delaney and Gelperin, 1990; Kobatake et al., 1992; ing feeding behavior. Hemoglobin inside glial cells pro- Church and Lloyd, 1994; Quinlan et al., 1995, 1997). vided a red background contrasting with the white ‘polka Surprisingly, however, little effort has been made to dots’ of neuronal somata (cf. Pentreath et al., 1985). examine the similarities and differences in the neural Simultaneous intracellular recordings from identifiable organization of feeding in the various gastropods. neurons and extracellular recordings from muscles could This report will focus on the neural bases of feeding be made while the animal performed oral behaviors (e.g. and regurgitation in the snail, Helisoma tri6ol6is (Mol- Kater, 1974). lusca, Gastropoda, Pulmonata, Basommatophora, Planorbidae). It also proposes the working hypothesis 2.2. Feeding beha6ior and its anatomical basis in that the similarities outweigh greatly the differences in basommatophoran snails the neural organization of feeding in those well-studied grazing gastropods that feed with rasping or grasping Helisoma and several other basommatophoran snails radulae (e.g. the opisthobranchs Aplysia, and Tritonia, are omnivorous with a variety of feeding mechanisms the basommatophoran snails, Helisoma, Planorbis, although, as a group, the basommatophora are more Planorbarius, and Lymnaea and stylomatophoran snails stereotyped in their feeding behaviors than are the and slugs, Helix, Limax,). A model developed to repre- opisthobranchs (cf. Audesirk and Audesirk, 1985; sent the organization of a multifunctional buccal central Thomas et al., 1985). The anatomy and/or dynamics of pattern generator (CPG) in Helisoma is offered as a the feeding apparatuses have been studied in a number general model for gastropod buccal CPGs. No generally of species (e.g. Hubendick, 1957; Demian, 1962; Hem- 386 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 brow, 1973; Dawkins, 1974; Kater, 1974; Goldschmed- retraction. In the third phase, the odontophore is hy- ing and DeVlieger, 1975; Brace and Quicke, 1981; per-retracted, bringing the tip of the radula near the Smith, 1988, 1990; Arnett, 1996). All of these studies opening of the esophagus. If the snail is feeding slowly, drew heavily on the work of Carriker (1946) on the odontophore returns to the rest position during the Lymnaea stagnalis.InHelisoma feeding behaviors and inter-bite interval. During the protraction stroke, the their anatomical bases were initially examined by Kater fulcrum, about which the odontophore rotates, is (1974), who also tried to correlate neural motor pat- moved posteriorly such that the tip of the odontophore terns generated by the buccal ganglia with feeding is not in close apposition to the food groove. One behavior. Smith (1988, 1990) extended Kater’s observa- variation of the typical feeding behavior is a sequence tions on the anatomy and dynamics of the feeding of repetitive swallows, during which the dorsal lip of apparatus. Later Arnett (1996) combined videomi- the odontophore is thrust or hyper-retracted repeatedly croscopy of buccal mass movements with simultaneous toward the esophagus without intervening protractions extracellular myograms and intracellular recordings of the odontophore (Arnett, 1996). from motor neurons during oral behaviors in semi-intact preparations. She focused on the roles of three prominent muscles of the feeding apparatus and their 3. Neurophysiological correlates of feeding in Helisoma activity patterns, and identified motor neurons innervating these muscles. Specific neural patterns mediating Neurophysiological studies have focused on the neu- typical feeding, regurgitation, and swallowing behaviors ral basis of the rhythmic scraping or rasping move- were determined. ments of the odontophore described above. The The mouth of Helisoma opens into a buccal cavity pioneering study of Kater (1974) influenced greatly not within a muscular pharyngeal buccal mass. The buccal only the subsequent studies of feeding in Helisoma, but mass consists of approximately 30 muscles, most occur- also comparative studies of feeding in other gastropods ring as bilateral pairs (Smith, 1990). The odontophore (e.g. Benjamin and Rose, 1979; Bulloch and Dorsett, (tooth carrier) arises from the floor of the buccal cavity. 1979). This influence had both positive and negative The odontophore consists of a tongue-like ‘cartilage’ and the attached file-like dentated radula. The chitinous radular teeth are secreted by the subradular membrane, which is continuous with the epithelial lining of the oral cavity. The odontophore cartilage is not true cartilage but is a complex tissue including intrinsic muscle fibers and mechanical support structures. These snails typically eat microflora by protracting the odontophore forward and downward through the mouth and rasping the radula over the substrate. Adja- cent to the mouth are one dorsal and two lateral chitonous mandibles. Helisoma cuts small pieces from macrophytes by employing the dorsal lip of the odontophore and the dorsal mandible as ‘scissors’. Using the odontophore as a scoop with a moving radular con- veyor belt, these snails will also feed on loose material (e.g. in the surface film of ponds) and they also eat carrion routinely. We have not observed Helisoma to use the buccal mass as a pump for ingesting semi-de- cayed matter without use of the odontophore, but this method was reported for the closely related snail Biom- phalaria (Thomas et al., 1985). Fig. 1. A schematic diagram of a feeding cycle in Helisoma. The At rest, the odontophore sits at about a 45° angle. anterior of a snail is depicted from a lateral perspective as if there were an optical sagital section through the mouth, odontophore During a typical feeding cycle (Fig. 1), the tip (or dorsal (odontopore cartilage, OC, plus the attached radula, R), buccal mass lip) of the odontophore is ‘protracted’ or rotated for- (BM) and esophagus (E). In position 1, the odontophore rests at ward and downward and protruded through the mouth. about a 45° angle. The odontophore is protracted through position 2 During the second phase, the ondotophore is rasped to the vertical position 3 with the lip of the odontophore protruding over the substrate and ‘retracted’ back to the rest through the mouth. The second odontophoral image in ‘3’ depicts the odontophore at the end of the rasp portion of the retraction phase. position within the oral cavity. The tip of the odon- Position 4 depicts the odontophore near the end of the retraction tophore is kept in close apposition to the food groove phase. Position 5 shows the hyper-retracted odontophore, which has along the midline of the roof of the buccal mass during brought food particles to the opening of the esophagus. A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 387

myograms attributed to the posterior jugalis (pj), a major odontophoral protractor muscle, were associated on a one-to-one basis with action potentials in motor neuron B21. However, the pj muscle is very thin and overlies the relatively massive supralateral radular tensor (slrt) muscle. It is difficult to distinguish pj activity from underlying tensor muscle activity when recording from the intact buccal mass (cf. Rose and Benjamin, 1979). Recently, feeding in Helisoma has been demon- Fig. 2. A schematic diagram of the original model of the Helisoma strated to arise from a triphasic motor pattern (see buccal pattern generator and the putative feeding motor pattern. The below, Fig. 3; Arnett, 1996; Quinlan and Murphy, motor pattern consisted of alternating bursts in retractor (R) motor neurons and ‘protractor’ (P) motor neurons (now known to be 1996; Quinlan et al., 1995, 1997) corresponding to the hyper-retractor neurons). This pattern was thought to be generated protraction, retraction, and hyper-retraction phases of by a single neuronal oscillator comprised of approximately 20 pairs of the feeding cycle. The slrt muscle contracts during electrotonically coupled interneurons, collectively called the ‘cyber- retraction and even more vigorously during hyper-re- chron’ (C). The cyberchron was thought to drive and time the action traction. The muscle potentials ascribed to the pj pro- potential bursts in motor neurons by exciting simultaneously the retractor neurons and inhibiting protractor neurons. The protractor tractor muscle and associated with neuron B21 action motor neurons would then fire bursts of action potentials from potentials by Kater (1974) were undoubtedly recorded postinhibitory rebound. (After Kater, 1974). from the underlying slrt. No morphological or other distinctive characteristics of neuron B21 were pub- aspects. A series of ‘mistakes’ by Kater (1974) illus- lished, so it is not uniquely identifiable. However the trates generally applicable pitfalls in correlating neu- large neurons B18 and B19 displayed similar patterns of ronal activity with corresponding evoked muscle synaptic and action potentials to the neuron designated activity in relatively complex neuromuscular systems. A B21 by Kater (1974). Neurons B18 and B19 innervate visual interpretation of the feeding behavior, combined the slrt, and during feeding motor activity, they fire with the misidentification of the muscular target of a neuron, led to a miscorrelation of activity in a set of neurons with the protraction phase of the feeding cycle and further led to the conclusion that the feeding cycle and its underlying neural pattern were biphasic rather than triphasic. Therefore, I will explain the reason for and the consequences of, the mistakes made in consid- erable detail. During typical rapid rhythmic feeding in Helisoma, there is no pause at the ‘rest’ position between the retraction and hyper-retraction phases, and protraction for the subsequent bite cycle commences immediately upon termination of the hyper-retraction phase. There- fore, Kater (1974) separated the feeding cycle into only two phases — protraction, and retraction. Kater recorded a biphasic buccal motor pattern with alternating bursts of action potentials in two sets of motor neurons (e.g. Fig. 2). The bursts of action potentials in one set of motor neurons (represented by neuron B27, see Fig. 3) were correctly correlated visually with the retraction phases of the feeding cycles, and hence, this set of motor neurons was designated ‘retractor motor neurons’ (Kater, 1974). Since there was thought to be Fig. 3. The triphasic feeding pattern of buccal motor neuron activity only a biphasic pattern, the remaining phase of neural in Helisoma. Simultaneous intracellular recordings from identified activity recorded was taken to represent protraction, buccal motor neurons demonstrated three discrete phases (1, 2, 3) of and neurons generating action potentials during this excitation in each of the four cycles of feeding activity depicted. part of the cycle were designated ‘protractor motor Protractor motor neuron B6 was depolarized during phase 1, and neurons’ (e.g. neurons B17, B18, B19, B20, and B21; inhibited during phase 2. Retractor motor neuron B27 was depolarized during phase 2, and Hyper-retractor motor neuron B19 was Kater, 1974). inhibited during phase 2 and depolarized during phase 3. Note, that The correlation of the neural pattern and feeding the pattern depicted in Fig. 2 corresponds to phases 2 and 3 of the behavior appeared to be confirmed when extracellular feeding pattern. 388 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 bursts of action potentials during the hyper-retraction during any given feeding cycle and, by chance, the phase (Arnett, 1996). Careful dissection of the buccal active neurons were not recorded. mass, allowing myograms to be recorded from the pj Electrotonic coupling and its plasticity were studied protractor muscle independently of the slrt, revealed extensively in this network (Kaneko et al., 1978; Mer- activity in pj myograms occurring after the hyper-re- ickel and Gray, 1980). The potential roles of electro- tractor activity (e.g. B19 bursts) of the previous feeding tonic coupling during rhythmic action potential burst cycle and prior to retractor activity in, for instance generation in a reverberating coupled system were ex- neuron B27. Subsequently, it was demonstrated that amined by computer modeling (Merickel et al., 1977, protraction phase neurons B6 and B8 innervate the pj 1978). Some members of the cyberchron network were (see below; Arnett, 1996). Thus the bi-phasic electro- also shown to have ‘bursty’ membrane properties (i.e. a physiological pattern described by Kater represents the region of negative slope resistance in their I–V curve). second and third phases (i.e. retraction and hyper-re- Later, however, it was shown that the IPSPs evoked in traction) of the triphasic feeding pattern (compare Figs. ‘protractor’ (i.e. hyper-retractor) motor neurons by ac- 2 and 3). The activity that Kater associated with pro- tion potentials in cyberchron neurons reversed at mem- traction actually mediates hyper-retraction of the odon- brane potentials much more depolarized than the tophore to facilitate swallowing. reversal potential of the IPSPs evoked by the buccal CPG during phase 2 (i.e. retraction) of the feeding cycle 3.1. The original model of the Helisoma buccal CPG (Murphy, 1991). More recently, it has been shown that activity evoked in some ‘cyberchron neurons’ triggers The retractor motor neurons described by Kater regurgitation. Thus, this group of neurons has been (1974) displayed excitatory postsynaptic potentials (EP- redesignated the Buccal A Cluster (BAC) neurons (Ar- SPs) underlying their action potential bursts. Simulta- nett, 1996) since they are not directly the components neously, the ‘protractor’ (actually hyper-retractor and of the buccal CPG and they neither time nor drive the radular tensor) motor neurons displayed prominent activity of motor neurons during feeding. inhibitory postsynaptic potentials (IPSPs). Electrotonic 3.2. A triphasic buccal motor pattern underlies feeding coupling among synergistic motor neurons was demon- beha6ior strated. A model of the buccal CPG (Fig. 2) was developed whereby only a single neural oscillator would A triphasic buccal motor pattern (Fig. 3) was de- simultaneously evoke EPSPs in the retractor motor scribed in Helisoma (Quinlan and Murphy, 1991, 1996; neurons and IPSPs in the ‘protractor’ motor neurons. Quinlan et al., 1995, 1997). It is quite similar to pat- The protractor motor neurons would subsequently gen- terns reported to underly feeding in the basomma- erate a burst of action potentials from post-inhibitory tophoran snail, Lymnaea (Elliott and Benjamin, 1985a) rebound. Indeed, the ability of these motor neurons to and the opisthobranch, Tritonia (Bulloch and Dorsett, generate a vigorous burst of action potentials following 1979). Since the Helisoma buccal ganglia can produce a termination of a hyperpolarizing pulse was number of different motor patterns, this triphasic pat- demonstrated. tern was designated the ‘standard pattern’ of buccal A set of an estimated 20 or so pairs of electrotoni- motor activity and was hypothesized first (Quinlan and cally coupled neurons was thought to drive and time Murphy, 1991) and later, demonstrated to underly the the activity of the motor neurons, and thus, were called typical feeding (Arnett, 1996). This pattern allowed ‘cyberchron neurons’. These neurons seemed to meet neurons to be classified as active in one or more specific the criteria for comprising the feeding CPG. These phase(s) of the feeding pattern. A number of neurons neurons evoked PSPs of the appropriate sign for the active in each of the three phases has been identified retraction phase of the feeding cycle in a number of (see Figs. 4 and 14 and Table 1). identified motor neurons. A brief stimulation of a cy- Protraction phase motor neurons (or presumptive berchron neuron could sometimes trigger a number of motor neurons) include neurons B3, B6, B7, and B8 cycles of rhythmic activity in feeding motor neurons. (Fig. 5). These neurons are excited during phase 1 and However, action potential bursts typically were not inhibited during phase 2. They recover slowly from the recorded from cyberchron neurons during sustained phase 2 hyperpolarization, remaining inhibited through rhythmic activity (e.g. Kater, 1974). Cyberchron neu- phase 3. The physiology of neuron B3 differs from that rons did, however, display typically the depolarizations of neurons B6, B7, and B8. In rapid feeding rhythms, it coinciding with the synaptic activity in motor neurons fires bursts of action potentials only during phase 1. during the retraction phase of the cycle. Therefore, it However, if there is a slow feeding rhythm, neuron B3 was hypothesized that a few neurons of the electrotoni- fires a burst of action potentials not only during phase cally-coupled network of approximately 20 pairs of 1, but also in the interphase following phase 3. If there cyberchron neurons were generating action potentials is rhythmic retraction and hyper-retraction phase activ- A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 389

Fig. 4. A map of the somata of identified Helisoma buccal neurons. Both the caudal surface and the rostral surface of the animal’s left buccal ganglion are depicted. Each identified neuron would have a mirror image homolog in the right ganglion (not shown). Note, that some terminology is altered from previous publications. The rostral and caudal surfaces were previously called ventral and dorsal, respectively; the VBN and LBN were previously designated the HBN and VBN, respectively (Kater, 1974). Neurons on the rostral surface were named formerly VB1-VB10 (to indicate ventral buccal neurons; Lukowiak and Murphy, 1987) and they have been renamed 101–110. The changes were made to conform to the in situ morphology of the structures, and to prior terminology of basommatophoran molluscan morphologists (e.g. Carriker, 1946; Hembrow, 1973), and to facilitate comparisons with other gastropods. BC, buccal commisure; CBC, cerebrobuccal connective; ET, esophageal nerve trunk; LBN, laterobuccal nerve; PBN, posterobuccal nerve; VBN, ventrobuccal nerve. ity but no protraction phase activity, neuron B3 fires a burst of action potentials following the hyper-retraction phase. The specific muscular target(s) of neuron B3 have not been identified, but they may help return the odontophore to the rest position. Neuron B3 projects an axon to the buccal mass via the ipsilateral esophageal nerve trunk (ET) and one of its branches, the dorsobuccal nerve (DBN). Neurons B6 and B8 innervate the pj muscle bilaterally and ipsilaterally,

Table 1 The sign of PSPs evoked in identified Helisoma effector neurons by each CPG subunita Fig. 5. Morphology of identified Helisoma phase 1 protractor motor NeuronS1 S2 S3 Axonal projections neurons. Photographs of neurons injected with the fluorescent dye, Lucifer Yellow. All of the neuronal somata are on the caudal surfaces B5I I I Ip and con Ets (GNs, DBNs) of the buccal ganglia. Orientation of ganglia is the same as in Fig. 4. B4 E E Ip and con Ets (SNs, DBNs) Influential motor neuron B6 (top) has axons in both PBNs and B34EE Ip LBN innervates the pj muscle bilaterally. Neuron B7 (middle) has an B3EI Ip ET (DBN) extensive neuritic arbor in the ipsilateral buccal ganglion and a single B6E I Ip and con. PBNs axon that traverses the ipsilateral VBN. Neuron B3 (lower left) was B7 E I Ip VBN photographed in the right buccal ganglion. It sends a proximal axon B8 E I Ip PBN segment toward the CBC. The axon makes a sharp turn to traverse B17 I E Ip LBN the esophageal trunk to the dorsobuccal nerve branch (not shown), B19I E Ip and con VBNs, ip LBN which it then follows into the buccal mass. Neuron B8 was pho- B18 I/E E Ip and con VBNs tographed in the left buccal ganglion and has an axon only in the B26 EE Ip and con VBNs ipsilateral PBN. It innervates the ipsilateral pj. Calibration bar, 100 m B27 EE Ip and con LBNs M. B29 E Ip and con PBNs B110 E Ip and con VBNs and LBNs respectively, via the posterobuccal nerves (PBNs). Ax- ons of neuron B7 enter the buccal mass via the ipsilat- a I, inhibitory; E, excitatory; Ip, ipsilateral; con, contralateral; ET, eral ventrobuccal nerve (VBN; previously known as the esophageal trunk; GN, gastric nerve; DBN, dorsobuccal nerve; SN, salivary nerve; VBN, ventrobuccal nerve; PBN, posterobuccal nerve; heterobuccal nerve, see the legend of Fig. 4 for explana- LBN laterobuccal nerve. tion of nomenclature). Neuron B34 is excited in phases 390 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408

1 and 2 and can have its highest spike frequency in either phase, depending on conditions. Neuron B34 has an axon in the ipsilateral LBN. Retraction (phase 2) motor neurons and presumptive motor neurons include neurons B110 (aka VB10), B26, B27, and B29 (Fig. 6). Neurons B29 and B110 generate bursts of action potentials only during phase 2. Neuron B29 has axon branches in both PBNs that pass through the pj muscle (but do not innervate it) and continue to unidentiﬁed targets deeper in the buccal mass. Neuron B110 has bilateral axon branches in both VBNs and LBNs. It innervates the slrt muscle (M. Zoran, personal communication). Neurons B26 and B27 are excited Fig. 7. Current model of the Helisoma buccal CPG. Squares represent strongly by S2 interneurons and more weakly by S3 the three interneuronal subunits (S1, S2, and S3) of the CPG. Each interneurons. During slow motor patterns, neuron B27 subunit is a conditional oscillator that serves as a unit pattern generator and provides the primary excitation to a corresponding subset of buccal motor neurons (S1MN, S2MN, and S3MN). Lines ending in perpendicular bars indicate excitatory connections, and those ending in closed circles indicate inhibitory connections. Arrows indicate that both the oscillatory properties of each subunit, and the synaptic connections between subunits, are potential targets of modulation.

displays a rapid burst of action potentials in phase 2 and fires more slowly in phase 3 (Quinlan and Murphy, 1996). During a vigorous feeding pattern, neuron B27 generates bursts of action potentials confined almost entirely to phase 2 of the pattern because it displays a postburst hyperpolarization that inhibits its firing during phase 3 (e.g. Fig. 3). Neuron B26 has axons in ipsilateral and contralateral VBNs and innervates the slrt muscle and Neuron B27 has axons in the ipsilateral and contralateral LBNs and innervates the aj muscle. Hyper-retraction (phase 3) motor neurons include neurons B17, B18, and B19. Neuron B17 has an axon in the ipsilateral LBN but specific muscular targets have not been identified. Neurons B18 and B19 have axons projecting bilaterally to innervate the slrt muscle via the VBNs (aka HBNs). Neuron B19 also innervates the lateral portions of the ipsilateral aj muscle via the ipsilateral LBN, and in some preparations, there is an axon in the contralateral LBN and a bilateral innervation of the aj (Arnett, 1996).

3.3. The current model of the multifunctional buccal CPG

Fig. 6. Morphology of phase 2 retractor motor neurons. Neuron The CPG consists of three semi-independent in- B101 (top) was viewed and photographed on the rostral surfaces of the buccal ganglia. It has an extensive neuritic arbor in both ganglia terneuronal subunits, S1, S2, and S3, each of which is and in the buccal commissure. It is bipolar and has axons in ipsi- and an oscillator capable of generating rhythmic bursts of contralateral VBNs and LBNs. Neurons B27 (middle) and B29 action potentials or, in the case of S2, rhythmic plateau (bottom) were viewed and photographed on the caudal surfaces of potentials (Fig. 7 and see Quinlan et al., 1995). Each the buccal ganglia, with orientation as in Fig. 4. Neuron B27 has an subunit provides excitation or inhibition to subsets of extensive ipsilateral neuritic arbor and has axons that traverse the ipsi- and contralateral LBNs. Influential motor neuron B29 (bottom) buccal motor neurons. Some motor neurons are excited has axons in both PBNs that pass through the pj muscle to unknown (or inhibited) by more than one subunit (Table 1) and, targets. Calibration bar, 100 mM. thus, can display action potential bursts, which span all A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 391 or part of the duration of more than one phase of the larizations of interneurons could reset the phase of cycle (Fig. 3 and cf. Quinlan and Murphy, 1996). motor patterns; and (5) that activity of the putative Therefore, (and for additional reasons) a number of interneurons during non-feeding motor patterns elicits motor neurons have action potential bursts that overlap PSPs similar to the phase-characteristic PSPs evoked by in time and do not fit discretely into only one of the the interneurons during the feeding motor pattern. A three phases of the feeding cycle. In the feeding mode, number of CPG interneurons have been identified and S1 interneurons stimulate protraction (phase 1) motor their activity patterns confirmed the flexible organiza- neurons and S2 interneurons. This excitation brings S2 tion of the multifunctional pattern generator (cf. Quin- interneurons to threshold for plateau potential genera- lan and Murphy, 1996). tion. S2 interneurons provide feedback inhibition onto S1 interneurons, terminating their activity. Simulta- 3.5. Identified CPG interneurons and motor neurons neously, S2 plateau potentials inhibit S3 interneurons. When the inhibition subsides, S3 interneurons fire from Commonly for invertebrates, it is impossible to pi- post-inhibitory rebound. Thus, feeding results from an geon-hole neurons as definitively ‘motor neurons’, ‘sen- S1-S2-S3 sequence of activity. As long as there is tonic sory neurons’ or ‘interneurons’ since a given neuron excitation to S1 interneurons or an excitatory milieu of may function in multiple roles (e.g. Miller and Selver- neuromodulators (e.g. high dopamine concentrations), ston, 1985; Evans et al., 1996; Hurwitz et al., 1996). rhythmic feeding activity will continue (Quinlan et al., With that caveat, in Helisoma, identified neurons active 1997). in each of the three phases of feeding could be catego- rized as pure interneurons, influential neurons, or fol- 3.4. Plasticity of organization in a multifunctional CPG lower neurons on the basis of function and morphology. Central pattern generator (CPG) elements A flexible organization of the multifunctional buccal include pure interneurons and influential neurons (cf. CPG was suggested by the analyses of many simulta- Arshavsky et al., 1988a). Pure interneurons either pro- neous intracellular recordings of two to four neurons ject axons to the cerebral ganglia (BCN1s) or have their from a set of 20 or so frequently monitored identified processes confined to the buccal ganglia and the proxi- neurons. Sets of PSPs, characteristic in sign and ampli- mal areas of buccal nerve roots (e.g. N1a, N1b, N1c). tude of PSPs occurring during a given phase of the They do not project axons to peripheral targets. Influ- triphasic feeding motor pattern, were recorded from ential neurons are multifunctional neurons that have multiple neurons and often occurred either sporadi- peripheral axons with either sensory or motor functions cally, or rhythmically, but in the absence of a full but which also participate in the generation and/or triphasic feeding pattern. These phase-characteristic modulation of buccal motor patterns. Influential neu- PSPs were correlated consistently across buccal neurons can play roles in the CPG that are as critical and rons, suggesting that CPG interneurons could be acti- prominent as those of pure interneurons. vated in functional motor patterns other than the Phase 1 pure interneurons include three pairs of triphasic pattern, or perhaps in non-functional degener- BCN1 neurons, with axons crossing the buccal commis- ate motor patterns. sure and traversing the contralateral cerebrobuccal con- However, the existence of correlated PSPs, that ap- nective with axonal arbors both in the buccal and pear to be the characteristic of those of a particular cerebral ganglia. In addition, neurons N1a, N1b, and phase of the standard feeding pattern, does not indicate N1c have processes confined to the buccal ganglia or necessarily that these PSPs arise from feeding CPG proximal areas of buccal nerves (Fig. 8). Neuron B6 is interneurons. For example, the BAC neurons (aka cy- a phase-1 influential neuron that also serves as a motor berchron neurons, Kater, 1974) are not a part of the neuron innervating the pj muscle (Arnett, 1996). Stimu- buccal CPG but can evoke widespread PSPs of the sign lation or hyperpolarization, respectively, of neuron B6 appropriate for phase 2 PSPs. Therefore, to identify can evoke or inhibit the feeding motor pattern. interneurons definitively as components of a multifunc- Central pattern generator (CPG) interneurons evok- tional buccal CPG and to show that modulation of ing retraction phase (i.e. phase 2) PSPs during feeding their activities could result in multiple motor patterns, are the pair of glutamatergic interneurons, B2 (Quinlan it was necessary to show, (1) that they were active et al., 1995). Interneuron B2 is a phase 2 influential during a particular phase of the triphasic feeding pat- neuron, activity in which appears necessary and suffi- tern; (2) that stimulation of putative interneurons in cient to account for S2-evoked PSPs. Neuron B2 has an quiescent preparations evoked phase-characteristic axon, which traverses the buccal commissure and forms postsynaptic potentials (PSPs) in a set of follower neu- a loop and returns to the ipsilateral neuropile. Neuritic rons; (3) that hyperpolarization of the putative in- processes arborize in both buccal ganglia and axon terneurons during a feeding pattern eliminated branches also project to the buccal mass via the poste- appropriate phase-specific PSPs; (4) that brief hyperpo- rior buccal nerves. Functions of the peripheral axons of 392 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 neuron B2 are unclear and could include sensory or The major elements of subunit 3 of the Helisoma motor roles (Quinlan et al., 1995). Neuron B29 is a buccal CPG are the pair of pure buccal interneurons phase-2 influential neuron, stimulation of which can N3a (Quinlan and Murphy, 1996). Activity in neurons evoke phase 2-like PSPs in other buccal motor neurons N3a is necessary and sufficient for phase-3 PSPs during (cf. Figs. 3, 6 and 9). However, hyperpolarization of the triphasic feeding pattern and also accounts for neuron B29 does not eliminate phase-2 activity in other phase 3-like PSPs occurring in non-feeding patterns. A neurons. No ‘pure interneurons’ with axons confined to cluster of S3 influential neurons (B101–104) that are the CNS and active in phase 2 have been described in immunoreactive to small cardioactive peptide B (Fig. Helisoma. However, Arshavsky and colleagues reported 10) have somata on the rostral (aka ventral) surfaces of ‘group 2’ interneurons in the snail, Planorbis, (like the buccal ganglia and axons in the posterior buccal nerves (Lukowiak and Murphy, 1987). In addition to Helisoma in the family Planorbidae) that were active in their roles as S3 CPG elements, these neurons function phase 2 of the feeding cycle and had no processes as radular mechanoafferents (unpublished data). They projecting from the buccal ganglia (Arshavsky et al., compensate apparently for load on the radula by acti- 1988a). vating S3 interneurons, and thus, the hyperretraction motor neurons.

3.6. E6idence that the triphasic standard buccal motor pattern mediates the typical feeding beha6ior

Since several different rhythmic motor patterns were recorded from identified buccal neurons in Helisoma,it was necessary to determine which neural pattern mediated the typical rasping behavior described above. This required semi-intact preparations, in which an incision was made along the dorsal midline of the head region, the body wall was reflected outward, and the esophagus and extrinsic buccal retractor muscles were severed to allow the buccal mass to be tilted forward. This brought the buccal ganglia, situated on the caudal surface of the buccal mass, into a dorsal position. A microplatform placed underneath the buccal ganglia stabilized the preparation for intracellular recordings while the buccal mass was free to perform feeding movements (Arnett, 1996; Arnett and Murphy, 1991, 1992). A reasonably good facsimile of feeding behavior could be evoked by feeding stimulants, with the caveat that incomplete protraction of the odontophore was seen, possibly due to a loss of hydrostatic pressure within the body cavity and within the buccal mass. The triphasic motor pattern (Fig. 3) was simultaneously recorded intracellularly from identified neurons and extracellularly from specific muscles of semi-intact adult snails performing feeding-like movements of the odontophore. However, the position of the odontophore Fig. 8. Morphology of protraction phase S1 interneurons. Interneu- could only be seen during part of the feeding cycle in rons were stained by intracellular injection with Lucifer Yellow. semi-intact adults. It was, therefore, necessary to define Tracings were made from sets of Ektachrome slides of each neuron external features of the buccal mass that were the taken at multiple focal planes and superimposed. There are six pairs of identified S1 interneurons. The BCN1 group consists of three pairs diagnostic for the position of the odontophore during of similar neurons that are not readily distinguishable from each each phase of the feeding cycle. Newly hatched snails other. These neurons send axons across the buccal commissure and (about 1 mm in shell diameter) were videotaped while traverse the contralateral CBC (upper left) to form a terminal arbor feeding on glass slides or transparent plastic sheets. in the cerebral ganglion. Often, a short neuritic process extends into Such small snails are semi-transparent and the positions one or more of the esophageal trunks (as shown here) or posterobuccal nerves. Neurons N1a, N1b and N1c do not have axons leaving the of the odontophore were correlated readily with exter- buccal ganglia though N1b has processes extending a few hundred nal features of the buccal mass. For any given phase of micrometers into the CBCs. the feeding cycle, the positions and overall contours of A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 393

Fig. 9. Neuron B29 is a retraction phase influential neuron. Simultaneous intracellular recordings were made from neuron B19, which is inhibited during phase 2 and excited during phase 3 of the feeding pattern. Activity in neuron B29 was recorded with a high resistance unbalanced Lucifer Yellow electrode. At the beginning of the record neuron B29 was held hyperpolarized (0.8 nA) and neuron B19 was firing tonically. Neuron B29 was released periodically from hyperpolarization, which evoked anode break bursts of action potentials in neuron B29 that triggered phase 2-like IPSPs in neuron B19 (labeled ‘2’ in the second series of anode break bursts). Spontaneous bursts in B29 coincided with IPSPs in B19, and a depolarizing pulse ‘D’ injected into neuron B29 evoked a phase 2-like IPSP in neuron B19. Calibrations, 10 s, 40 mV. the buccal mass, as well as the degrees of protrusion of Microscope slides were coated with agar containing the radular sac were distinctive. Since similar positions watermelon extract, a potent feeding stimulant. Lister- and contours of the buccal mass and radular sac were ine was applied via Pasteur pipettes near the mouths of seen during feeding behaviors of semi-intact adults, it newly hatched snails feeding vigorously on the water- was possible to correlate the stages of the feeding cycle melon-coated slides. These snails would perform imme- with the appropriate phases of the triphasic neural diately one or several odontophore reversals (i.e. pattern. The aj, pj, and slrt muscles are prominent regurgitation strokes). A number of features distin- muscles of the buccal mass that are visible when the guished regurgitation cycles from feeding cycles. The mass is viewed from an external perspective. Their states of contraction or relaxation are major determi- nants of the positions of the odontophore and of the contours of the buccal mass during the three phases of the feeding cycle (Arnett, 1996). The pj muscle contracts during protraction of the odontophore and different regions of the aj and slrt muscles contract both during retraction and hyper-retraction. Neurons B6 and B8 innervate the pj, bilaterally and ipsilaterally, respectively, and generate bursts of action potentials during phase 1 of the feeding pattern. Neuron B27 generates action potential bursts primarily during phase 2 and innervates the medial portion of the aj muscle, which contracts during retraction. Neuron B19 displays action potential bursts during phase 3 of the standard buccal motor pattern and innervates the lateral regions of both the aj and slrt muscles, which contract vigorously during hyper-retraction of the odontophore. Thus, identification of motor neurons for the aj, pj, and slrt muscles confirmed both the roles of these muscles during feeding and that the standard triphasic buccal motor pattern mediates the most typical feeding behavior (Arnett, 1996).

3.7. Regurgitation

Regurgitation can be triggered in basommatophoran snails by feeding them noxious materials (Bovbjerg, Fig. 10. Morphology of phase 3 interneuron N3a and influential 1968). We observed serendipitously that Listerine is an neuron B101. Tracings of Lucifer Yellow stained neurons made from effective emetic. This observation was exploited to trig- Ektachrome slides taken at multiple focal planes. Neuron N3a (top) ger regurgitation in intact newly hatched snails, in is located on the caudal surface of the buccal ganglia. Neuron (B101) lower is the largest of a cluster of small cardioactive B immunoreac- which the dynamics of the buccal mass and movements tive neurons on the rostral surface of the buccal ganglia. It has short of the odontophore could be observed (Arnett, 1996; neuritic processes in the proximal ipsilateral CBC and ET. It has Arnett and Murphy, 1994). axons in both PBNs and functions as a radular mechanoreceptor. 394 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 power stroke of the odontophore switched from retraction during feeding to protraction during regurgitation. During feeding, the fulcrum of the odontophore (i.e. the point about which the odontophore rotates) is moved forward at the beginning of the retraction stroke. Thus, the dorsal lip of the odontophore is apposed closely to the lining of the oral groove during retraction and food particles are swept posteriorly. During regurgitation, the fulcrum of the odontophore moved forward at the beginning of protraction so that the lip of the odontophore was apposed to the oral groove during protraction, thus sweeping food particles out the mouth. The fulcrum, then, moved backward at the beginning of retraction so that the lip of the odon- Fig. 11. GABA excites protraction phase motor neurons and evokes rhythmic activity in CPG subunit 2 by pharmacologically distinct tophore was not apposed to the oral groove during mechanisms. GABA triggered rhythmic action potential bursts inter- retraction. Additionally, the protraction phase (i.e. rupted by phase 2 IPSPs in the influential protractor motor neuron phase 1) was prolonged during regurgitation. Finally, B6. Neuron B6 fired tonically in normal physiological saline (top there was, typically, no hyper-retraction of the odon- trace). At the first arrowhead, 100 mM was superfused over the tophore during regurgitation cycles. Only the protrac- preparation. At the second arrowhead, the GABA was washed out with normal saline. The GABAB receptor agonist, baclofen, triggered tion/retraction phases of the trajectory were displayed. rhythmic S2 activity as indicated by the phase 2 IPSPs in neuron B6,

The physiological correlates of regurgitation included but it had no depolarizing effect on neuron B6. Muscimol, a GABAA/ prolonged phase 1 bursts of action potentials (underly- C receptor agonist depolarized phase 1 motor neuron B6 but did not ing the protraction phase), relative to those seen during trigger rhythmic activity in subunit 2. Calibrations 40 mV, 10 s. feeding, and a suppression of S3 activity, therefore, eliminating hyper-retraction. The neural pattern under- of neuron N1a during a feeding pattern can reset the lying regurgitation was examined in semi-intact adult phase. The dopamine antagonist, sulpiride, blocks the snails, in which the esophagus was canulated with effects of both dopamine application and of neuron polyethylene tubing connected via a two-way valve to N1a stimulation. Interneuron N1a also stains with the separate syringes containing watermelon extract or 25% formaldehyde/gluteraldehyde histochemical stain that is listerine (Arnett, 1996; Arnett and Murphy, 1994). indicative of catecholamines and is highly specific for Feeding behavior and the triphasic feeding motor pat- dopamine in molluscan nerve cells. Application of feed- tern could be triggered in quiescent snails by perfusing ing stimulants (e.g. watermelon extract) triggers activity the oral cavity with watermelon extract. Switching the in neuron N1a and consequent feeding behavior. valve to admit 25% listerine into the oral cavity, trig- GABAergic interneurons are also elements of S1 of gered a number of strong regurgitation cycles, with the buccal CPG. Two of the three pairs of BCN1 vigorous protraction of the odontophore, interspersed neurons are GABA immunoreactive and application of with occasional weak regurgitation cycles with only GABA has a number of effects on identified buccal partial protraction of the odontophore. The strong neurons (Richmond et al., 1991, 1994; Murphy, 1993). protraction cycles were correlated with the prolonged GABA depolarizes a number of identified protraction action potential bursts in S1 interneuron N1a, relative phase neurons that are excited by S1 interneurons (Fig. to those seen during feeding. Neither hyper-retraction 11). This effect is mimicked by muscimol and blocked of the odontophore, nor phase 3 action potential bursts by picrotoxin as one would expect if GABA or in motor neuron B19 were observed. A GABAC-like receptors were activated. This excitatory effect of GABA is enhanced by chloride injection into 3.8. Transmitters of CPG interneurons the phase-1 motor neurons suggesting that they nor- Here, we will focus on the neurotransmitters, do- mally have a high concentration of chloride ions pamine, GABA and glutamate because their roles internally. within the Helisoma buccal CPG are best characterized. Either stimulation of BCN1 interneurons or superfu- Exogenous application of 5–10 mM dopamine elicits sion of GABA also triggers rhythmic S2 activity. This the triphasic feeding motor pattern in Helisoma reli- activation of rhythmic S2 activity is mimicked by ba- ably. An endogenous source of dopamine is phase 1 clofen and is not affected by picrotoxin, suggesting that interneuron N1a of the buccal CPG (Quinlan et al., it is mediated by a GABAB-like mechanism. Finally, 1997). Stimulation of interneuron N1a by current injec- bath application of GABA (500 mM–1 mM) hyperpo- tion evokes the triphasic feeding pattern. N1a is an larizes phase 3 interneuron N3a as well as a number of intrinsic element of the CPG. A brief hyperpolarization phase 3 motor neurons (Richmond et al., 1994). This A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 395 inhibitory effect is also mimicked by muscimol and Perfusion of the oral cavity with a feeding stimu- blocked by picrotoxin. Thus, at these concentrations, lant (e.g. watermelon extract) via canulation of the GABA evokes an S1–S2 pattern of activity similar to proesophagus evoked the triphasic feeding pattern the pattern seen during regurgitation. and feeding movements (Arnett, 1996; Quinlan et al., The primary phase 2 PSPs arise from glutamatergic 1997; see above). Feeding was evoked even if there interneuron B2 (Quinlan et al., 1995). Interneuron B2 were no neural connections between the buccal gan- has been co-stained by injection of the dye Lucifer glia and the rest of the central nervous system. Yellow and by immunocytochemical staining with glu- Hence, afferent information must have reached the tamate antibodies. The phase 2 EPSPs evoked by stim- buccal ganglia via the buccal nerve roots. A number ulation of neuron B2 are mimicked by Kainate/AMPA of possibilities could account for the differences in the and are blocked by CNQX (Quinlan and Murphy, nature of the motor patterns observed by Horwitz 1991). The phase 2 IPSPs are mimicked by quisqualate and Senseman (1981), Quinlan et al. (1997). Perhaps and have no known antagonist. Since glutamate is a cannula in the mouth triggers mechanoafferents considered generally an excitatory transmitter, neuron that interfere with the generation of the triphasic B19 was placed in isolation in acute cell culture to feeding pattern. Alternatively, differences in the determine if the inhibitory effect of glutamate was chemostimuli may have contributed to the differences direct or indirect. Glutamate or quisqualate applied to in observed motor patterns. the isolated neurons B19 directly evoked hyperpolarizing responses. 3.10. Higher order modulation of the CPG

3.9. Sensory and modulatory pathways regulating the Two descending modulatory elements have been buccal CPG characterized in Helisoma — the pair of giant serotonergic interneurons C1 (Granzow and Kater, 1977; Sensory or modulatory afferents important for oral Granzow and Rowell, 1981; Murphy et al., 1985a), and behaviors impinge upon the buccal ganglia from one of the inhibitory pleurobuccal interneurons Pl1 (Murphy, two general pathways. Information from the 1990). Neuron C1 is excited by feeding stimulants ap- supraesophageal ring of ganglia (i.e. the ‘brain’) reaches plied to the lips and oral veil. Stimulation of neuron the buccal ganglia via the cerebrobuccal connectives C1, or bath application of 1 mM serotonin, evokes (CBCs). Information from the buccal mass, oral cavity, consistently the rhythmic phase-locked activity in S2 odontophore, salivary glands and esophagus can reach and S3 of the buccal CPG. Serotonin can, but usually the buccal ganglia via buccal nerve roots that innervate does not, activate the full triphasic feeding pattern. The the buccal mass, salivary glands and esophagus. effects of serotonin on S1 activity are variable and a Mechano- and chemosensory afferents from the lips, number of qualitatively different motor patterns (all oral veil and tentacles affect mainly the buccal ganglia via axons to the cerebral ganglia in the lip and tentacu- displaying linked S2–S3 activity) can be evoked by lar nerves. serotonin application (Quinlan and Murphy, 1996). An Patterned motor activity evoked by chemosensory S2–S3 pattern without activity in S1 mediates rhythmic stimulation of the Helisoma oral cavity with feeding swallowing (Arnett, 1996). stimulants (10% wheat germ or 20% fish meal) was Patterned activity of the buccal CPG is inhibited demonstrated by Horwitz and Senseman (1981). Can- rapidly and completely either by exogenous application nulae were placed in the mouth and in the esophagus of the neuropeptide FMRFamide (Murphy et al., and substances were perfused from the mouth through 1985b), or by stimulation of the putatively FMR- the oral cavity and esophagus. Examination of the Famidergic interneuron Pl1 (Murphy, 1990). The so- figures shows that rhythmic activity occurred in S2 and mata of paired interneurons Pl1 are on the ventral S3, indicated by IPSPs in neuron B19, similar to S2- surfaces of the pleural ganglia. The main axon of evoked IPSPs, and action potential bursts in neuron neuron Pl1 traverses the pleuropedal commissure, the B19, similar to those evoked by S3. However, the pedocerebral commisure, and the cerebrobuccal com- activity in S2 and S3 was not phase-locked into the missure to reach the buccal ganglia. It has neuritic S2–S3 sequence, as it is during the typical feeding arbors in the parietal, pleural, pedal, cerebral and buc- pattern and S1 activity was not monitored. Senseman cal ganglia. Thus, it is perfectly situated to integrate a noted (personal communication) that there were rhyth- variety of stimuli and coordinate activity in the feeding mic contractions of the buccal mass but it was difficult and locomotory CPGs and to activate the whole body to tell if they were true feeding movements and the withdrawal response. Whether it performs all of these experimental protocol forced the feeding stimulants integrative roles remains to be determined (but see through the oral cavity from mouth to esophagus. below). 396 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408

4. Comparisons of gastropod neurons and feeding cir- Ridgeway, 1995). More recently, neurons similar to cuitry Helisoma neuron Pl1, which inhibits buccal motor patterns, have been identified in a number of basomma- Two key related questions need to be addressed. tophoran and stylomatophoran pulmonates and in the 1. Is the model of the buccal CPG, developed to carnivorous pteropod opisthobranch, Clione (Alania, account for data from Helisoma and discussed 1995; Alania and Sakharov, 1996 and M. Alania, per- above (or, for that matter, any other model), gener- sonal communication). It was suggested above that alizable to provide a universal model for gastropod neuron Pl1 was ideally situated to coordinate feeding buccal neuronal organization? locomotion and defensive withdrawal. The effects of 2. Are buccal neurons and their interconnections con- Pl1 on locomotion or withdrawal have yet to be exam- served sufficiently to allow homologies of identified ined in Helisoma. However, increased activity in FMRF- neurons to be ascertained and the evolution of amidergic pleurobuccal neurons in Clione, similar to circuitry at the cellular and subcellular levels to be Helisoma neurons Pl1, not only inhibited feeding but analyzed? was also correlated positively with both spontaneous The tentative answer to both questions is yes, for and induced accelerations of the locomotor CPG (M. those gastropods with rasping or grasping radular feed- Alania, personal communication). ing modes. However, more comparative data are needed for a definitive answer. 4.1. Biological and methodological complications for The key feature of the Helisoma CPG organization is interspecific comparisons of gastropod feeding that it is composed of multiple semi-independent unit pattern generators (or CPG subunits) that can be acti- There are several real or potential factors that may vated or inactivated independently and functionally complicate interspecific comparisons of gastropod feed- linked in different temporal sequences. Thus, a number ing motor patterns and underlying neuronal circuitries. of different motor patterns can be produced. In He- (1) Incomplete and non-congruent experimental data lisoma, there are three such subunits. During evolution, sets are available for each of the species that have been other gastropod species may have added an additional examined. (2) Interspecific diversity of feeding behav- interneuronal subunit(s), or lost one, to accommodate iors in gastropds. (3) Intraspecific diversity of oral different behavioral niches. behaviors and the underlying multiplicity of motor Confirmation of cellular homologies is problematic patterns in any given species. (4) Incomplete congru- (e.g. Croll, 1987; Breidbach and Kutsch, 1995). Here, ence of action potential firing patterns in motor neu- we will not absorb ourselves with attempted ‘proofs’ of rons with respect to those seen in the CPG homology in specific cases. None-the-less, the Dar- interneurons. Motor neuron firing patterns are more winian paradigm of relation of species by common diverse and complex than are those in interneurons, due descent is generally accepted at present. Analyses of cell to multiple interneuronal synaptic inputs, multiple ef- lineages and patterns of neuronal differentiation indi- fects of transmitters, and diverse intrinsic membrane cate clear homologies of identified neurons in some properties among motor neurons, and the modulation cases (cf. Comer and Robertson, this volume and in- of all of these. Thus, similar CPG patterns could ap- cluded references). Many gastropod neurons have dis- pear different purely as a function of the motor neurons tinguishable unique identities. Comparative monitored. All of the above factors can be addressed, morphological, physiological, and neurochemical data as indicated below. provide a strong case for cellular homology for a few gastropod neurons. Below, I will suggest candidate 4.1.1. Incomplete data homologies, especially for a number of CPG interneu- One major difficulty in revealing the generalities of rons (cf. Fig. 14). gastropod feeding has been that studies on different Perhaps the best evidence for homologies to date in species have focused on many different aspects of feed- molluscan feeding circuitries relates to higher order ing. For instance, studies in the stylomatophora, and in modulatory interneurons and to giant buccal neurons the carnivorous opisthobranch, Pleurobranchaea, have that innervate the gut (e.g. Altrup, 1987; Lloyd et al., focused primarily on sensory processing, behavioral 1988; Bulloch and Ridgeway, 1995), or the salivary choice, higher order control of feeding, and its modula- glands (e.g. Kater et al., 1978; Bahls et al., 1980; tion by behavioral state or learning (e.g. London and Barber, 1983; Bahls et al., 1995). A pair of giant Gillette, 1984; Mpitsos and Cohan, 1986; Delaney and serotonergic neurons has been identified in the cerebral Gelperin, 1990; Kemenes, 1994; Jing and Gillette, ganglia of numerous gastropods and these neurons 1995). Thus, relatively little is known about the organi- display great morphological, physiological, and func- zation of the buccal CPGs in these species. Even in tional similarities (e.g. Weiss and Kupfermann, 1976; those species in which the buccal CPGs have been Granzow and Rowell, 1981; Croll, 1987; Bulloch and studied extensively (e.g. Aplysia, Helisoma, Lymnaea, A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 397

Planorbis), different aspects of the CPGs have been even be non-functional. Over the years, numerous buc- studied. Concerted efforts at a comparative synthesis of cal motor patterns have been asserted or implied in the the neuronal organization of feeding behaviors has literature to represent fictive feeding, often on the basis been lacking. of an electrophysiological recording from a single neuron and without concurrent behavioral data. Buccal 4.1.2. Interspecific di6ersity motor patterns often change qualitatively on a cycle by Mechanisms of feeding in gastropods are diverse (e.g. cycle basis and multiple qualitatively different motor Audesirk and Audesirk, 1985; Thomas et al., 1985), patterns from identified neurons have been depicted as leaving the question as to what extent neuronal cir- ‘fictive feeding’ in the same figure. Frequently, the cuitry underlying these behaviors has been conserved original author is conservative, referring to a recording phylogenetically. Perhaps the most disconcerting sce- as a buccal motor pattern, only to have it referenced in nario in relation to a comparative synthesis could be subsequent reports as a ‘feeding motor pattern’. How that gastropod feeding structures and behaviors may then, amongst such confusion, can meaningful interspe- have diverged to such an extent that few commonalities cific comparisons of feeding motor patterns and their remain in their buccal neuronal organization. This situ- underlying neuronal circuitry be made? ation may hold for some species with highly specialized Fortunately, feeding behaviors and the dynamics of feeding, such as Melibe (Hurst, 1968) or Na6anax, buccal masses and odontophores can be analyzed in (Susswein et al., 1987; Cappell et al., 1989a,b) where the intact freely behaving juvenile gastropods by videomi- need for an odontophore has been eliminated largely or croscopy. Reasonable facsimiles of feeding behavior wholly. However, most gastropods that have been sub- can be evoked in semi-intact preparations while electro- jected to neurophysiological analyses use rhythmic physiological recordings are made from identifiable movements of a toothed radula during feeding. These neurons and/or muscles (e.g. Arnett, 1996; Drushel et include herbivorous or omnivorous pulmonates (e.g. al., 1997). A ‘typical feeding behavior’ can be selected the basommatophorans Helisoma, Planorbis and and the standard feeding motor pattern underlying this Lymnaea and the stylomatophorans Achatina, Helix, feeding behavior determined. With effort and luck mo- Incilaria, Limax) and opisthobranchs that graze on tor neurons displaying common sets of PSPs and the algae or sessile invertebrates (e.g. Aplysia, Tritonia). interneurons mediating these sets of PSPs can be iden- Even aggressive carnivorous opisthobranchs that use tified uniquely. Sets of phase-linked PSPs in multiple specialized structures for capturing prey typically ingest identified motor neurons, suggestive of a common pre- or swallow the prey with rhythmic buccal movements synaptic interneuron or set of interneurons, may be (e.g. Clione, Pleurobranchaea). seen to occur in linked fashion in temporal sequences Close phylogenetic relatedness, combined with simi- different from that seen in the standard feeding pattern. larities of ecological niches and feeding behaviors, Thus, plasticity in the organization of the buccal CPG should facilitate the analyses of potentially homologous can be characterized, and sometimes controlled, phar- neuronal organization. Although opisthobranchia and macologically or by modulatory interneurons. Finally, pulmonata represent different subclasses of gastropoda, the interactions among identified modulatory interneu- a consensus suggests that they are much more closely rons, CPG interneurons and motor neurons can be related than are the different subgroups of the highly studied. diverse ‘prosobranch subclass’, a paraphyletic grouping Key features of the organization of the Helisoma no longer considered systematically useful (cf. Ponder buccal CPG, and of patterns of activity recorded from and Lindberg, 1996). Here, a significant degree of ho- buccal motorneurons, help illustrate difficulties encoun- mology of identifiable neurons of the feeding circuits of tered when comparing different motor patterns. Each opisthobranchs and pulmonates is hypothesized. Simi- of the three CPG subunits is a semi-independent condi- larities in identified neurons and neuronal organization tional oscillator that can generate rhythmic activity that are indicated below. may or may not be linked to activity in the other subunits. Furthermore, the order of activity of particu- 4.1.3. Intraspecific di6ersity lar subunits is variable. For instance, subunit 3 may A major biological factor obfuscating interspecific generate action potential bursts that are linked to S2 comparisons of ‘feeding motor patterns’ and their neu- activity but the S3 activity can precede that of S2, ral bases is that each species examined has a plastic instead of following it as it does in the typical feeding multifunctional CPG that generates multiple buccal behavior (cf. Arnett, 1996; Quinlan and Murphy, 1996). motor patterns. Some of these patterns (usually moni- A subunit also may be active more than once per cycle. tored with the nervous system in vitro or in a semi-in- For example, subunit 3 may generate bursts of activity tact preparation) mediate variations of feeding both preceding and following S2 activity (Fig. 12). behavior, some mediate regurgitation or other non- Another complication relative to CPG analysis is that feeding oral behaviors and some of the patterns may there are strong gradations in the intensity of S1 and S3 398 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 activity (cf. Quinlan and Murphy, 1996; Quinlan et al., rons have feedback inhibition to terminate the S1 1997). Different interneurons active within the same bursts (Quinlan et al., 1995). In spite of this, many phase of the feeding pattern can have different onsets motoneurons receive excitation both from S1 and from for firing and not all interneurons of a given phase need S2 interneurons, and thus, may fire action potentials to be active in all motor patterns. across phases 1 and 2 (or similarly 2 and 3, or even 1, 2, and 3) of the feeding pattern (cf. Table 1). An 4.1.4. Non-congruence of interneuronal and motor additional complicating factor is that biphasic synaptic neuronal patterns responses may occur in follower neurons. In particular, Perhaps the most prominent complication in feeding several neurons that are postsynaptic to S2 interneu- motor pattern analysis across species arises from the rons (e.g. neuron B18, which is primarily excited during fact that action potential burst patterns of many mo- phase 3) show an initial hyperpolarization followed by toneurons will not show an abrupt onset or termination a depolarization. A consequence of these neural organi- of the bursts at the ‘borders’ between subunit activities, zational features is that while each ‘feeding’ motor even if a rhythmic S1–S2–S3 standard feeding pattern neuron reflects the activity of one or more subunits of is being generated (see below). Therefore, some iden- the buccal CPG, no single motor neuron provides a tified motor neurons provide a much clearer reflection direct mirror image of the activity of the CPG. of CPG activity than other identified motor neurons. One can overcome the above complications in motor To know for certain which motor neurons provide a pattern analysis by having unequivocally identifiable reliable monitor of CPG activity, relevant interneurons neurons that can serve as reference monitors of activity must be identified, and recorded with specific motor in CPG subunits. Helisoma neuron B19, for example, is such a reference neuron that reflects reliably the activity neurons under a variety of conditions (e.g. with differ- of CPG subunits 2 and 3 (Quinlan and Murphy, 1996). ent modulators in the bath). In general, more than one One should monitor synaptic potentials and not just neuron will need to be monitored simultaneously to extracellular spiking patterns, because similar spiking obtain a complete reflection of CPG activity. patterns can arise from different patterns of synaptic In Helisoma, since the onset and duration of bursts activity. Finally, interneurons must be identified and of action potentials in different S1 interneurons varies, their activity confirmed to be coincident with CPG there will be differences in firing onset in motor neurons subunit activity during feeding patterns. They must be innervated by the S1 interneurons. There is, however, shown to evoke phase-characteristic PSPs in appropri- an abrupt transition from S1 to S2 interneuronal activate identified follower neurons during feeding patterns. ity because S1 interneurons excite S2 interneurons to Hyperpolarization of the identified interneuron should threshold for plateau potential generation and S2 neu- eliminate at least some of the PSPs evoked by the CPG subunit. Similarly, the identified interneuron should be able to evoke phase-characteristic PSPs in identified follower neurons when stimulated in quiescent preparations or in preparations generating non-feeding patterns. These PSPs need not be identical to the PSPs generated during feeding patterns since synaptic responses may be modulated during feeding.

4.2. Speciﬁc buccal motor patterns

Triphasic patterns of motor neuron activity, similar to that shown above for Helisoma (Fig. 3) have been reported to underlie feeding in the closely related pulmonate, Lymnaea (Elliott and Benjamin, 1985a) and in the opisthobranch, Tritonia hombergi (Bulloch and Fig. 12. Multiple patterns of inhibition and excitation in Helisoma Dorsett, 1979). Somata of groups of neurons that are neuron B19 indicate the ﬂexibility of organization of the buccal CPG. During the typical S1–S2–S3 feeding motor pattern neuron B19 is active in equivalent phases of the feeding cycles in inhibited by S2 neuron B2 and is excited by S3 neuron N3a. However several different species have been localized to similar S2 and S3 are semi-independent interneuronal oscillators that can be areas of their respective buccal ganglia (Bulloch and independently rhythmically active or linked in different temporal Ridgeway, 1995). sequences. Upper left — rhythmic activity with S3 excitation preced- Putative feeding cycles with two, three, or more ing S2 inhibition. Upper right — rhythmic activity with S3 excitation both preceding and following S2 inhibition. Lower left — rhythmic phases have been reported in other gastropods. Biphas- S2 inhibition with no S3 excitation. Lower right — rhythmic S3 ic cycles reported for the planorbid basommatophoran excitation with no S2 inhibition. snails appear to represent partial characterizations of a A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 399

from phase 1 (protraction) innterneurons and from Phase 2 (retraction) interneurons. The onset of the phase 2 depolarization corresponds to a decrease in action potential amplitude and a resultant butterfly appearance of the recording (Fig. 13A cf. Hurwitz and Susswein, 1996). In the instance shown an unidentified motor neuron of the ventral cluster displayed phase 1 inhibition, phase 2 excitation, and inhibition in a third phase. This suggests that neurons B4/B5 may receive excitation from three different sets of interneurons. Similarly, neuron B8 is known to receive excitation during both phase 1 and 2 (Morton and Chiel, 1993). Another unidentified ventral cluster motor neuron displayed distinctive phase 1 and 2 IPSPs followed by a phase 3 burst of action potentials. These ‘phase 3 effects’ could plausibly be explained by mechanisms not requiring phase 3 interneurons, but they are consistent certainly with and suggestive of a tripartite CPG. It remains to be determined whether this third phase of activity in Aplysia arises from interneurons homologous to Helisoma interneuron N3a. Biphasic motor patterns in other gastropods such as Clione (Arshavsky et al., 1989), Pleurobranchea (Davis et al., 1973), and Helix (Peters and Altrup, 1984) too may represent partial Fig. 13. A triphasic buccal motor pattern in Aplysia. A, Simultaneous characterizations of a triphasic pattern. intracellular recordings from a neuron B4/B5 displayed the character- Buccal rhythms consisting of four or five phases have istic ‘butterfly appearance’ that results from the transition from phase been reported and may result from the designation of 1 excitation to phase 2 excitation. An unidentified motorneuron (MN-1) displayed phase 1 inhibition, phase 2 excitation and phase 3 the quiescent interval between cycles as a ‘phase’ of the inhibition. B, A neuron B8 displayed excitation in phases 1 and 2 of cycle. For example, the triphasic pattern of feeding a cycle while an unidentified motor neuron (MN-2) displayed distinc- motor neuron activity in Lymnaea (Elliott and Ben- tive inhibitory inputs during phases 1 and 2 followed by a phase 3 jamin, 1985a) was earlier reported to be a four-phase excitation. cycle (cf. Fig. 14 in Benjamin and Rose, 1979). Alterna- tively, the retraction of the radula from its most pro- triphasic feeding cycle. The biphasic pattern of alternat- tracted position to a ‘rest’ position is sometimes ing bursting in motor neurons described in Helisoma by subdivided into separate rasp and retraction phases. Kater (1974), as discussed above, corresponded clearly However, four or five phase cycles (cf. Fig. 2 in Sossin to phases 2 and 3 of the standard feeding motor et al., 1987) are within the capabilities of a tripartite pattern. Similarly, Arshavsky et al. (1988a) described a pattern generator with flexible linkages among indepen- biphasic pattern of protraction and retraction, sepa- dent conditional oscillators. For example, in Helisoma, rated by a quiescent interval, in Planorbis. The pattern a rhythmic four-phase S1–S3–S2–S3 pattern can be described in Planorbis corresponded to phases 1 and 2 generated (unpublished observations). of the Helisoma triphasic feeding pattern. Phase 3 motor neurons that mediate the hyper-retraction of the 4.3. Interspecific comparisons of CPG interneurons and radula inexplicably were not characterized in Planorbis, feeding motor neurons even though unlabeled somata of what appear to be neurons homologous to the large neurons B18 and B19 Few direct comparative studies of similar motor neu- of Helisoma were included in the Planorbis buccal map. rons and interneurons involved in feeding in different In Aplysia, a biphasic pattern subserves protraction gastropod species have been attempted. Though the (phase 1) and retraction (phase 2) of the radula (Hur- buccal ganglia of pulmonates and opisthobranchs have witz et al., 1996). A third phase may be involved in a plethora of ‘identifiable’ neurons, there are relatively swallowing. Perrins and Weiss (1998) presented a few buccal neurons in any species that are truly triphasic pattern in Aplysia but suggested that the third ‘uniquely identified.’ Insufficient physiological and mor- phase may actually represent a pause between biphasic phological data is published to allow re-identification of cycles. However, triphasic buccal motor patterns resem- most buccal neurons that at first glance would seem to bling those of Helisoma are recorded readily in Aplysia be ‘identified’. For instance, on the map of the caudal neurons (Fig. 13). Neurons B4/B5 receive excitation (aka dorsal) surface of the buccal ganglia of Helisoma 400 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 provided by Kater (1974) 23 pairs of neurons are and biochemical characterizations of ‘identified’ neu- depicted and named. However, this study preceded the rons are needed for each of the species investigated. availability of the dye, Lucifer Yellow and these neu- We focused on pure interneurons (i.e. no peripheral rons were ‘identified’ on the basis of relative soma size processes) and influential neurons (i.e. they participate and position and patterns of electrophysiological activ- in motor pattern expression but also have peripheral ity, with minimal morphological data from Procion processes that may have sensory or motor functions). stains. Only five of these 23 pairs of neurons are While many phase 1, 2, and 3 motor neurons have been uniquely re-identifiable today (B4, B5, B17, B18, B19). identified in several species (e.g. Church and Lloyd, With the exception of those five neurons plus B27, B101 1994), many of these have great similarities to other and B110, (Fig. 4) a combination of physiological motor neurons in the same species and relatively few recordings and morphological stains is necessary for have been characterized sufficiently to allow ready in- unique identification of Helisoma buccal neurons. terspecific comparisons. Neurons were characterized be- Simlarly, Benjamin and colleagues (Benjamin et al., low on the basis of in which phase of the feeding cycle 1979; Rose and Benjamin, 1979, 1981a,b) identified ten they are most prominently active. Interneurons and ‘types’ of buccal follower neurons in Lymnaea with influential neurons with major roles in the buccal CPGs criteria similar to those used by Kater (1974). Only four were placed in a ‘composite universal tripartite CPG of these ten types (neurons B1–B4) represent uniquely model’ (Fig. 14). Probable or potentially homologous identifiable neurons without using morphological data, interneurons were boxed together. which has been provided only sporadically (cf. Staras et al., 1998). Arshavsky et al. (1988a,b,c) identified nine ‘groups’ of buccal neurons in Planorbis, again using similar physiological criteria to those of Kater and of Benjamin, with only sporadic concurrent morphological data. The current author is also not blameless. While Lucifer Yellow is used routinely in microelectrodes, yielding morphological data concurrently with physio- Fig. 14. A composite model for a universal gastropod tripartite logical data in this author’s laboratory, much of that buccal central pattern generator. The model displays three subunits data sits in reams of chart recordings and photographic (S1, S2, S3) of the CPG, each of which represents a semi-independent negatives on laboratory shelves. I used these examples neuronal oscillator that provides the primary excitation to sets of not to castigate my colleagues but to illuminate the past motor neurons active during corresponding phases of a feeding cycle. Neurons in each of the large boxes are active during one of the three and current situations, and to call for more published phases. Neurons within the smaller rectangles inside the large boxes evolutionarily significant data on identifiable neurons, represent putatively homologous interneurons based on morphologi- both in print and on the world wide web (cf. Comer cal and physiological characteristics. In some cases, these putative and Robertson, this volume). homologs have also been shown to have the same neurotransmitter. Note that the BCN1 group is a heterogenous set of phase 1 interneu- In spite of the paucity of direct comparative studies, rons with apparent variability in number among the different gas- several analogous and potentially homologous neurons tropods. They have similar morphologies and are all phase 1 have been identified in the basommatophoran snails, interneurons but some fire early during phase 1 and some later during Helisoma, Planorbis and Lymnaea, and in two different phase. With the exception of Planorbis group 2, the other designa- opisthobranchs, the sea hare, Aplysia, and the tions represent individual identified neurons. The arrows indicate that there are excitatory interactions among interneurons of different pteropod, Clione. The estimated number of neurons in types within each subunit but should not suggest that every interneu- the buccal ganglia of these species varies by about a ron within a subunit excites every other interneuron. The bar with a factor of two (Boyle et al., 1983). Thus, not every perpendicular bar indicates that S1 interneurons excite S2 interneu- neuron in one species will have a unique homolog in rons. The bars with filled circles represent inhibition. The darker rectangles within each subunit represent the neurons that seem to be other species. For instance, a common developmental necessary and sufficient to generate a triphasic motor pattern. The cell lineage in two species might result in a lack of a question marks indicate that, (1) a phase 1 interneuron with a one-to-one relationship between neurons if there is a morphology similar to Helisoma interneuron N1a was described in difference in the number of cell divisions or in pro- Clione but not given a name; (2) the effects of Lymnaea interneurons grammed cell death in the two species. None-the-less it N3p and N3t are inhibitory to those motor neurons that are excited by S3 so in that sense they are not a principal part of subunit 3; and seems certain that these related gastropods have ho- (3) it is not yet known whether Aplysia B21 and B22 activate mologous neurons in the sense that particular neurons hypothetical phase 3 interneurons. However, like the Helisoma cluster result from similar developmental pathways resulting in of influential neurons B101–104, they are small cardioactive peptide similar form and function (cf. Breidbach and Kutsch, immunoreactive radular mechanoafferent neurons on the rostral surfaces of the buccal ganglia. They have similar morphologies and they 1995). The comparisons below are made in an attempt fire action potentials during the latter part of spontaneous buccal to spur comparative research. More studies combining motor patterns (i.e. during phase 3). H=Helisoma;P=Planorbis; morphological, electrophysiological, pharmacological L=Lymnaea;A=Aplysia;C=Clione. A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 401

Table 2 Analogous, possibly homologous motor neurons and inﬂuential neurons in Helisoma (H.), Lymnaea (L.), and Aplysia (A.)a

Helisoma Lymnaea Aplysia Axonal projections Muscle targets

Phase 1 (protraction) neurons In, B6 In, B31/32 H. PBNs=A.RN H.pjm=A.I2m B8 B7nor B61/62 H. PBN, A. RN, L. pjN H. and L.pj=A.I2 m B7 B7a A. SRT mn H. VBN=A. N1; H.?, A. SRT, L. pj? L. VBN or LBN? B3 B5 H. and L. DBN ? Phase 2 (retraction) neurons In, B29 In, B10 H. and L. PBNs ? Perhaps rad tens m Phase 3 (hyper-retraction) neurons B19 B4 H. and L. VBNs and LBNs H. and L. slrt and aj

a in, Inﬂuential; PBN, posterobuccal nerve; RN, radular nerve, PJ, posterior jugalis; nor, nitric oxide responsive; SRT, subradular tissue; VBN, ventrobuccal nerve; LBN, laterobuccal nerve; DBN, dorsobuccal nerve; rad tens, radular tensor; slrt, supralateral radular tensor; aj, anterior jugalis.

The distinction between influential and motor neu- morphology and physiological effects on buccal motor rons is somewhat arbitrary. There are gradations in the patterns (Yeoman et al., 1995). It has been reported to abilities of motor neurons to influence the buccal be non-dopaminergic but the negative evidence is not CPGs, in large part because of extensive electrotonic compelling (Vehovsky and Elliott, 1995). The dopamin- coupling not only within but across these groups. It has ergic antagonists fluphenazine and ergometrine failed to long been known that molluscan motor neurons can block the effects of stimulation of Lymnaea interneuron influence interneurons. For instance, it was shown by N1L. However, the effectiveness of neither antagonist Granzow (1979) that motor neuron B19 could, in some against exogenous dopamine was tested in this system. preparations, influence the buccal CPG. Some poten- Furthermore, somata in the vicinity of that of Lymnaea tially homologous motor neurons and influential neu- neuron N1L displayed dopamine immunoreactivity rons that were not placed in the composite CPG model (Elekes et al., 1991). Another possible homolog of these are indicated in Table 2. neurons is a protraction phase interneuron, which has a similar morphology and soma position in the carnivo- 4.4. Protraction phase interneurons rous opisthobranch, Clione (Arshavsky et al., 1989). Two other types of phase 1 pure buccal interneurons Very similar, perhaps homologous, phase 1 (protrac- in Helisoma, neurons N1b and N1c, were described tion) catecholaminergic (almost certainly dopaminergic) above (Fig. 8). Neuron N1b is morphologically, at least interneurons have been characterized in Helisoma superficially, similar to the dopaminergic Aplysia in- (Quinlan et al., 1997) and Aplysia (Kabotyanski et al., terneuron B20 (Teyke et al., 1993). There are somata 1998). Dopamine may be a universal initiator of gas- that display catecholaminergic histochemistry in the tropod feeding motor patterns (e.g. Wieland and vicinity of the somata of Helisoma interneurons N1b, Gelperin, 1983; Kyriakides and McCrohan, 1989; but whether these interneurons are dopaminergic re- Kabotyanski et al., 1994; Quinlan et al., 1997). Neurons mains to be determined. Interneurons N1c have somata N1a in Helisoma and B65 in Aplysia are similar strik- located laterally on the dorsal surfaces of the buccal ingly. Their somata are in similar locations near the ganglia. Interneuron N1c projects an axon to the con- lateral edges of the caudal surfaces of the buccal gan- tralateral buccal ganglion that forms a loop and returns glia. The main axons cross the buccal commissure but to the ipsilateral buccal ganglion (Fig. 8). This protrac- do not exit the buccal ganglia. They have neuritic tion phase interneuron is putatively a homolog of the arbors in each of the paired buccal ganglia. Stimulation SO interneuron of Lymnaea studied extensively (Elliott of these neurons mimics the effects of bath application and Benjamin, 1985b; Yeoman et al., 1993). A similar of dopamine or DOPA. Both neurons N1a and B65 phase 1 interneuron, also designated the SO, was also displayed catecholamine histochemical staining. In He- reported in Planorbis (Arshavsky et al., 1988a,c). In- lisoma, the dopamine antagonist, sulpiride, blocked the terneurons of this type have, thus, far been described effects of bath application of dopamine and of neuron only in basommatophoran snails. N1a stimulation. A widespread class of protraction phase interneurons Lymnaea interneuron N1L is also a candidate ho- are represented by the type 1 bucco-cerebral interneu- molog of Helisoma neuron N1a and Aplysia neuron rons (BCN1 neurons) in Helisoma. Depolarization of B65 and is similar to them in relative soma position, these neurons can drive feeding motor patterns and 402 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 their action potentials can evoke one-for-one EPSPs in 4.5. Motor and influential neurons some protraction phase motor neurons. Neurons of this class have somata near the margins of, or sometimes Helisoma phase 1 motor neurons B3, B6, B7, and B8 underneath, the somata of the medial giant buccal were described above (Fig. 5). Motor neurons B31, neurons (e.g. Helisoma neuron B5, Lymnaea neuron B32, B61 and B62 of Aplysia have physiological and B2) that innervate the esophagi of Helisoma, Planorbis, morphological similarities to neurons B6 and B8 of Lymnaea, Aplysia, and Clione (Arshavsky et al., 1988a, Helisoma. Aplysia motor neurons B31 and B32 and 1989; Richmond et al., 1991). The BCN1 interneurons B61/B62 innervate the intrinsic-2 (I2) muscle, a protrac- project an axon across the buccal commissure to tra- tor muscle that is analogous/homologous to the pj verse the contralateral cerebrobuccal connective and muscle innervated by B6 and B8 of Helisoma (Hurwitz enter the cerebral ganglion. Each has a neuritic arbor in et al., 1994; Arnett, 1996). Neurons B31/B32 of Aplysia both buccal ganglia and a terminal arbor in the cerebral and B6 of Helisoma are strongly influential and can ganglion contralateral to the BCN1 soma. Retrograde evoke buccal motor patterns. An SRT (subradular tissue) motor neuron in Aplysia appears morphologically stains of severed CBCs in Helisoma with either CoCl2 or Lucifer Yellow show consistently the three neurons and physiologically similar to Helisoma neuron B7 (Fig. of this morphological type in each buccal ganglion. The 5 and Borovikov et al., 1997), but it is not yet known if three pairs of BCN1 neurons are not yet readily distin- Helisoma’s B7 innervates the SRT. These neurons have guishable from each other, but they are heterogenous. axons in equivalent buccal nerves (Aplysia N3 and Two of the three pairs of BCN1 neurons in Helisoma Helisoma VBN, formerly called the HBN). Two phase 1 are GABA-immunoreactive (Richmond et al., 1991). motor neurons were uniquely identified recently in The numbers of this class of phase 1 interneurons Lymnaea, B7a and B7nor. Neuron B7a is somewhat similar morphologically to Helisoma B7 and Aplysia’s appear to be different in different gastropods. Planor- SRT motor neuron in that it has an axon in either the bis, (family Planorbidae, like Helisoma) was reported to ipslateral VBN or the LBN (these nerves were not have four pairs of these neurons, based on retrograde distinguished). Neuron B7a is also an influential neuron backfills of the CBC (Arshavsky et al., 1988a). Origi- that is electrotonically coupled to neuron N1M (Staras nally, based upon physiological recordings and a few et al., 1998). Lymnaea neuron B7nor is responsive to Lucifer Yellow injections, there were thought to be nitric oxide and innervates the pj muscle via the pj about ten pairs of interneurons of this type in Lymnaea nerve (Park et al., 1998), and may be related to the (Rose and Benjamin, 1981b; Elliott and Benjamin, Helisoma B6, B8 and Aplysia B31/B32, B61/B62 1985a). They were referred to at that time as N1 neurons. interneurons. It now appears, based upon retrograde staining of CBCs, that there is only one pair of neurons 4.6. Retraction phase interneurons of this type in Lymnaea (Kemenes and Elliott, 1994). It is called the medial N1 neuron or neuron N1M Pure phase 2 buccal interneurons with no axons (Yeoman et al., 1995). traversing buccal nerve roots include the group 2 neu- Two protraction phase interneurons, B34 and B63, rons of Planorbis, and neuron B64 in Aplysia (Ar- similar to the Helisoma BCN1 morphological type, shavsky et al., 1988b; Hurwitz and Susswein, 1996). have been identified in the opisthobranch Aplysia (Hur- The Planorbis group 2 neurons displayed endogenous witz et al., 1997). Interestingly, interneuron B63 ap- rhythmic activity when isolated from the buccal ganglia pears to be active during the protraction phase of all (Arshavsky et al., 1988b). Aplysia interneuron B64 was buccal motor patterns that have a protraction phase, found to be active consistently during Aplysia buccal whereas B34 appears to be devoted to egestion patterns. motor patterns. However, hyperpolarization of left and The carnivorous opisthobranch, Clione, based upon right neurons B64 failed to eliminate phase 2 PSPs and retrograde staining of the CBC, is thought to have only failed to disrupt buccal motor patterns. Such hyperpo- a single pair of protraction phase interneurons (PIN1) larization revealed large phase 2 depolarizations resem- of this morphological type (Arshavsky et al., 1989). bling the plateau potentials seen in the glutamatergic Thus, currently, there appears to be a greater number influential interneurons B2 of Helisoma and N2v of of BCN1 type phase 1 interneurons in the pulmonate Lymnaea (see below). It appears that Aplysia neuron basommatophoran planorbid snails than in either the B64 and the Planorbis group 2 neurons may contribute closely related pulmonate basommatophoran lymnaeid to CPG subunit 2 rhythmicity and have synaptic effects snails or the opisthobranchs. However, these relative on subsets of buccal interneurons, but activity in the numbers are based on examinations of very few species, influential glutamatergic interneurons like B2 of He- and are based largely upon retrograde staining tech- lisoma and N2v of Lymnaea may be necessary and niques, which can be quite capricious. sufficient to provide the primary output of subunit 2. A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 403

Glutamate is the primary phase 2 (retraction phase) quiescent preparations, and inhibition of interneurons neurotransmitter, evoking pharmacologically distinct N3a eliminates phase 3 PSPs in buccal motor patterns EPSPs and IPSPs in subsets of buccal motor neurons in (Quinlan and Murphy, 1996). The ‘phase 3 interneu- Helisoma (Quinlan and Murphy, 1991). Stimulation of rons’ of Lymnaea clearly are neither morphologically neuron B2 by current injection evokes phase 2-like nor functionally equivalent to interneuron N3a of He- PSPs in all buccal neurons examined. Hyperpolariza- lisoma. Two types of phase 3 interneurons, N3 tonic tion of interneuron B2 during ongoing buccal motor (N3t), and N3 phasic (N3p) have been identified in patterns eliminates phase 2 PSPs in all buccal neurons Lymnaea. If one hypothesizes that both Helisoma and examined (Quinlan et al., 1995). Interneuron B2 ex- Lymnaea have homologs of each of these phase 3 tends an axon to the contralateral buccal ganglion, interneuronal types the relationships among the real where it makes a loop and extends back to the ipsilat- and hypothetical neuronal types remain unclear. Effects eral buccal ganglion. It forms a neuritic arbor in both of eliminating activity in either N3t or N3p during a buccal ganglia and it extends at least one, sometimes feeding buccal motor pattern in Lymnaea have not been two axonal branches out each of the posterior buccal demonstrated. Both neurons evoke PSPs on buccal nerves. interneurons and motor neurons. However, it is partic- As with Helisoma, glutamate is the primary phase 2 ularly unclear why neuron N3t is considered part of the neurotransmitter in Lymnaea. Recently, Brierly and buccal CPG. Interneuron N3t fires tonic action poten- colleagues (Brierley et al., 1997a,b,c) described a virtu- tials essentially continuously except when it is inhibited ally identical and presumably homologous retraction by phase 1 or 2 interneurons, hence the appellation phase, presumptively glutamatergic, interneuron (N2v) ‘tonic’. It is active in quiescent ganglia, and it is active in Lymnaea. The originally identified Lymnaea retrac- in the ‘inactive interval’ between cycles during slow tion phase N2 interneurons (now called N2d) were buccal rhythmic activity. Thus, the PSPs evoked by N3t shown by photoablation to be unnecessary for the very much resemble the tonic chloride-dependent PSPs production of normal buccal motor patterns (Rose and recorded in many Helisoma buccal neurons (e.g. B19) Benjamin, 1981b; Kemenes and Elliott, 1994). The N2d either during, or in the absence of, buccal motor pat- neurons have peripheral axons and may be primarily terns (Kater, 1974). The Lymnaea N3p neuronal activ- sensory or motor neurons that are coupled electrotoni- ity appears restricted to phase 3 of the Lymnaea feeding cally to interneuron N2v. Neurons B29 in Helisoma pattern. However, its synaptic effects are opposite in (Figs. 6 and 10) and B10 in Lymnaea (Staras et al., sign of those of Helisoma phase 3 interneuron N3a on 1998) are morphologically similar, relatively weakly, equivalent classes of motor neurons. For instance, there influential phase 2 neurons. Putative glutamatergic re- is a set of similar hyper-retraction (aka swallowing) and traction phase interneurons in Aplysia remain to be radular tensor motor neurons in Helisoma and identified but glutamate mimics some of the retraction Lymnaea. The largest of these motor neurons are B19 phase PSPs in Aplysia (Fig. 15). in Helisoma and B4 in Lymnaea. These neurons have similar positions of their somata, similar morphologies, 4.7. Hyper-retraction phase interneurons similar (though not identical) physiological activity (see section on comparisons of motor neurons below), and Comparisons of phase 3 interneurons among gas- similar muscular targets. Both neurons innervate the tropods are more difficult than comparisons among slrt and aj muscles. Helisoma neuron B19 is excited by phase 1 and phase 2 interneurons because of ‘pre- phase 3 interneuron N3a. In fact, the firing patterns of sumably missing data’. In Helisoma, interneurons N3a B19 and of interneuron N3a are almost identical to (Fig. 10) account for the primary excitatory and in- each other during feeding and other motor patterns hibitory phase 3 PSPs in buccal motor neurons. Stimu- (Quinlan and Murphy, 1996). The Lymnaea N3p in- lation of interneuron N3a evokes phase 3-like PSPs in terneuron, however, evokes IPSPs on Lymnaea radular

Fig. 15. Glutamate depolarizes at least some Aplysia buccal neurons that are excited during the retraction phase of the feeding cycle. An intracellular recording from an Aplysia neuron B4/B5 showed no spiking activity in Aplysia saline. Glutamate (1 mM) was superfused over the preparation (first arrowhead) and it caused a depolarization of about 10 mV, which resulted in continuous firing of action potentials. Note that the glutamate also triggered some EPSPs onto neuron B4/B5 as indicated by the upward deflection of the baseline and a concomitant decrease in action potential amplitude. At the second arrowhead, normal physiological saline (NS) was superfused over the preparation. 404 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 tensor neuron B4 during the action potential burst of neuron B4. Thus, for these tensor motor neurons, interneuron N3p (and N3t) appears to act as a brake on the system. Clearly, they do not drive the action potential bursts that occur in phase 3 motor neurons (e.g. B4). To quote Rose and Benjamin (1981a)), ‘‘The input which we have called ‘N3’ is more interesting. Although it is difficult to see what function this input performs in inhibiting cells which are already hyperpolarized (5 cells, 7 cells) it is clear that the N3 inhibitory input fractionates 4-group and 8 cell bursts into a series of short duration bursts …. It is difficult to provide a functional explanation of the excitatory N3 input to the 3 cell, since this occurs after the 3 cell burst, and all the e.p.s.p.s are subthreshold’’. The phase 3 bursts in Lymnaea B4 (and four-cluster cells) have been attributed to post-inhibitory rebound (PIR) in the motor neurons, just as phase 3 activity in Helisoma B19 was attributed originally to PIR (Kater, 1974). However, Lymnaea’s neuron B4 shows activity patterns similar to Helisoma’s B19 including rhythmic and sporadic excitation independent of phase 2 inhibition. This is consistent with the presence of an unidentified excitatory interneuron similar to interneuron N3a, which accounts for the primary phase 3 excitation in Helisoma motor neurons (compare Figs. 12 and 16). There is evidence for phase 3 excitation and inhibition in Aplysia (Fig. 13). However, no Aplysia interneurons similar either to Helisoma neuron N3a or Lymnaea neurons N3t and N3p have been reported. However, Fig. 16. Multiple patterns of activity in Lymnaea mimic the motor Aplysia has a population of SCP-immunoreactive radu- plasticity seen in Helisoma. During a typical feeding pattern (upper lar mechanoafferents on the rostral surface of the buc- trace) Lymnaea neuron B4 receives inhibition during phase 2 and cal ganglia. These neurons are morphologically similar excitation during phase 3. However, phase 3-like excitation and phase 2-like inhibition can be linked to produce patterns other than the to the B101–104 (akaVB1–VB4) cluster of Helisoma standard feeding pattern such as the S3–S2–S3 pattern (second SCPB-ir radular mechanoafferents. These Helisoma trace). Each subunit can also be active independently yielding either neurons are effectively phase 3 interneurons (see a rhythmic S2 pattern (third trace) or a rhythmic S3 pattern (lower above). In addition, at least some of the Aplysia radular trace). All four of these patterns were recorded from the same neuron mechanoafferents fire action potentials in the latter part B4. The similarity of motor patterns suggests a similar CPG organization to that seen in Helisoma. (swallowing phase?) of spontaneous buccal motor patterns (Miller et al., 1994). It remains to be seen whether many invertebrate systems have the great attribute of stimulation of these Aplysia neurons can evoke PSPs, experimental tractability. This experimental tractability similar to swallowing phase PSPs in buccal motor of systems with relatively small numbers of identifiable neurons. Direct comparative studies of hyperretraction/ swallowing phase interneurons are especially needed at neurons led to another major finding — in general, this time. hard-wired circuit diagrams do not underly particular behaviors. There has been a demise of the concept of ‘stereotypy’ with respect to behaviors and the neural 5. Concluding remarks outputs of even ‘relatively simple’ circuits. The well- modulated motor system emerged triumphant. Without We have entered the fourth decade of research on the identifiability of unique neurons, it would be virtu- identified neurons and the circuits which they comprise. ally impossible to ascertain the degree, to which a single It is appropriate to consider what they have taught us neuron’s membrane characteristics and functional inter- and what they might teach us in the near future. actions with other neurons could be modulated to fit Perhaps the single greatest point of consensus to arise different situations at different times. during this period is that what were initially considered It has also become clear that similar rhythmic motor ‘simple systems’ are not simple. Though not simple, programs — that is, activity in two or more sets of A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 405 motorneurons with bursts of action potentials offset by Fred Bahls, Kelly Karas-Hedrick, Irene Yashina and phase lags — could arise from a wide variety of neural Vicky Leung for reading earlier versions of the circuits. However, the characterized circuits were all manuscript. composed using a circumscribed number of discrete ‘building blocks’, intrinsic membrane properties and types of synapses (cf. Getting, 1989). Both intrinsic References membrane properties and specific synaptic connections can be modulated on a moment to moment basis so Alania, M., 1995. Pleuro-buccal projections in pulmonate molluscs. that multiple functional motor patterns can arise from Acta Biol. Hung. 46, 267–270. a given group of identified neurons. As indicated Alania, M., Sakharov, D.A., 1996. FMRF-amidergic nature of pleuro-buccal projecting neurons common to diverse pulmonate throughout this volume, future progress based upon molluscs. Soc. Neurosci. Abstr. 22, 1405. identified neurons will encompass further reductionist Altrup, U., 1987. Inputs and outputs of giant neurons B1 and B2 in and molecular analyses of CPG building blocks and the buccal ganglia of Helix pomatia: an electrophysiological and computational analyses of circuit function. morphological study. Brain Res. 414, 271–284. During the past 25 years, some of the experimental Arnett, B.C., 1996. Feeding and regurgitation: two modes of operation of the buccal central pattern generator in Helisoma. Ph.D. disadvantages of small mammalian neurons versus thesis, University of Illinois at Chicago, Chicago, IL. large identifiable invertebrate neurons have been miti- Arnett, B.C., Murphy, A.D., 1991. Correlations of specific neural gated by technical advances — notably the widespread patterns and buccal muscle contractions with dynamic stages of development of patch clamp techniques and molecular the feeding cycle of Helisoma tri6ol6is. Soc. Neurosci. Abstr. 17, expression systems. Many laboratories working initially 1391. with the systems of identified neurons in invertebrates Arnett, B.C., Murphy, A.D., 1992. Organization and function of a multifunctional neuromuscular system during feeding in Helisoma have switched to mammalian systems. In addition, the tri6ol6is. Soc. Neurosci. Abstr. 18, 1414. diversity of circuits with similar outputs, the common- Arnett, B.C., Murphy, A.D., 1994. Neural basis of regurgitation ality of circuit components, and the functional modula- elicited by an emetic in Helisoma. Soc. Neurosci. Abstr. 20, 22. tion of a given circuit to produce diverse outputs, have Arshavsky, Y.I., Deliagina, T.G., Meizerov, E.S., Orlovsky, G.N., led to the suggestion that what we have already learned Panchin, Y.V., 1988a. Control of feeding movements in the freshwater snail Planorbis corneus I. Rhythmic movements of the from several invertebrate systems may have effectively buccal ganglia. Exp. Brain Res. 70, 310–322. brought to an end ‘the search for new CPG circuits’. Arshavsky, Y.I., Deliagina, T.G., Orlovsky, G.N., Panchin, Y.V., We feel that suggestion to represent a short-sighted 1988b. Control of feeding movements in the freshwater snail view. Though the technical advantages offered by large Planorbis corneus II. Activity of isolated neurons of buccal gan- identifiable neurons over smaller neurons lacking a glia. Exp. Brain Res. 70, 323–331. Arshavsky, Y.I., Deliagina, T.G., Orlovsky, G.N., Panchin, Y.V., unique identity may have decreased, the greatest at- 1988c. Control of feeding movements in the freshwater snail tributes of identifiable neurons for neurobiologists are Planorbis corneus III. Organization of the feeding rhythm genera- their unique identities per se! These unique identities tor. Exp. Brain Res. 70, 332–341. allow short- and long-term analyses of circuit functions Arshavsky, Y.I., Deliagina, T.G., Orlovsky, G.N., Panchin, Y.V., and their modulations. Furthermore, ‘‘With this com- 1989. Control of feeding in pteropod mollusc, Clione limacina. pendium of data, we will have the unique opportunity Exp. Brain Res. 78, 387–397. Audesirk, T., Audesirk, G., 1985. Behavior of gastropod molluscs. In: to examine the interspecific diversity of the neuronal Willows, A.O.D. (Ed.), The Mollusca, Neurobiology and Behav- mechanisms underlying a specific behavior. This ap- ior. Part 1, vol. 8. Academic Press, Orlando, pp. 2–94. proach will be the reductionistic counterpart to the Bahls, F., Kater, S.B., Joyner, R.W., 1980. Neuronal mechanisms for ethological analysis of phylogenetic ritualization bilateral coordination of salivary gland activity in Helisoma.J. (Lorenz, 1966) and should provide significant insight Neurobiol. 11, 365–379. Bahls, F.H., Emery, D.G., Haydon, P.G., 1995. Glutamate-mediated into the evolution of neuronal circuitry.’’ This opti- synaptic transmission between neuron B4 and salivary cells of mistic quotation marks the closing remarks of Kater Helisoma tri6ol6is. Invert. Neurosci. 1, 123–131. (1974). We are now poised to realize this opportunity. Barber, A., 1983. Nervous control of the salivary glands of the carnivorous mollusc Philine aperta. J. Exp. Biol. 107, 331–348. Benjamin, P.R., Rose, R.M., 1979. Central generation of bursting in Acknowledgements the feeding system of the snail, Lymnaea stagnalis. J. Exp. Biol. 80, 93–118. Benjamin, P.R., Rose, R.M., Slade, C.T., Lacy, M.G., 1979. Mor- Work reported here was supported in part by NSF phology of identified neurones in the buccal ganglia of Lymnaea grants BNS-9121374 and IBN-9728769 and by the Uni- stagnalis. J. Exp. Biol. 80, 119–135. versity of Illinois Campus Research Board. I thank Berry, M.S., 1972. Em electrically coupled small cells in the buccal Bridgette Arnett for the use of figures from her Ph.D. ganglia of the pond snail, Planorbis corneus. J. Exp. Biol. 56, 621–638. thesis and Danielle Turley for the data in Fig. 16. I Borovikov, D., Rosen, S.C., Cropper, E.C., 1997. SCP-containing thank Jack Gibbons, Andy Carol, John Dowd and sensory neurons in Aplysia are peripherally activated as a result of Mike Hollinstat for helping in figure preparations and motor neuron activity. Soc. Neurosci. Abstr. 23, 1045. 406 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408

Bovbjerg, R.V., 1968. Responses to food in lymnaeid snails. Physiol. Drushel, R.F., Neustadter, D.M., Shallenberger, L.L., Crago, P.E., Zool. 41, 412–423. Chiel, H.J., 1997. The kinematics of swallowing in the buccal Boyle, M.B., Cohen, L.B., Macagno, E.R., Orbach, H., 1983. The mass of Aplysia californica. J. Exp. Biol. 200, 735–752. number and size of neurons in the CNS of gastropod molluscs Elekes, K., Kemenes, G., Hiripi, L., Geffard, M., Benjamin, P.R., and their suitability for optical recording of activity. Brain Res. 1991. Dopamine-immunoreactive neurones in the central nervous 266, 305–317. system of the pond snail Lymnaea stagnalis. J. Comp. Neurol. Brace, R.C., Quicke, D.L.J., 1981. Activity patterns in radular mo- 307, 214–224. toneurons of the snail, Planorbarius. J. Comp. Physiol. 142, Elliott, C.J.H., Benjamin, P.R., 1985a. Interactions of pattern gener- 259–270. ating interneurons controlling feeding in Lymnaea stagnalis.J. Breidbach, O., Kutsch, W., 1995. The Nervous Systems of Inverte- Neurophysiol. 54, 1396–1411. brates: An Evolutionary and Comparative Approach. Birkhauser, Elliott, C.J., Benjamin, P.R., 1985b. Interactions of the slow oscilla- Boston. tor interneuron with feeding pattern-generating interneurons in Brierley, M.J., Staras, K., Benjamin, P.R., 1997a. Behavioral function Lymnaea stagnalis. J. Neurophysiol. 54, 1412–1421. of glutamatergic interneurons in the feeding system of Lymnaea: Evans, C.G., Rosen, S., Kupfermann, I., Weiss, K.R., Cropper, E.C., plateauing properties and synaptic connections with motor neu- 1996. Characterization of a radular opener neuromuscular system rons. J. Neurophysiol. 78, 3386–3395. in Aplysia. J. Neurophysiol. 76, 1267–1281. Brierley, M.J., Yeoman, M.S., Benjamin, P.R., 1997b. Glutamate is Gardner, D., 1971. Bilateral symmetry and interneuronal organiza- the transmitter for N2v retraction phase interneurons of the tion in the buccal ganglia of Aplysia. Science 173, 550–553. Lymnaea feeding system. J. Neurophysiol. 78, 3408–3414. Getting, P.A., 1989. Emerging principles governing the operation of Brierley, M.J., Yeoman, M.S., Benjamin, P.R., 1997c. Glutamatergic neural networks. Annu. Rev. Neurosci. 12, 185–204. N2v cells are central pattern generator interneurons of the Goldschmeding, J.T., DeVlieger, T.A., 1975. Functional anatomy and Lymnaea feeding system: new model for rhythm generation. J. innervation of the buccal complex of the freshwater snail Neurophysiol. 78, 3396–3407. Lymnaea stagnalis. Proc. K. Ned. Akad. Wet. 78, 468–476. Bulloch, A.G.M., Dorsett, D.A., 1979. The functional morphology Granzow, B., 1979. Identified higher-order neurons which regulate and motor innervation of the buccal mass of Tritonia hombergi.J. the feeding motor program of Helisoma tri6ol6is. Ph.D. thesis, Exp. Biol. 79, 7–22. University of Iowa. Bulloch, A.G.M., Ridgeway, R.L., 1995. Comparative aspects of Granzow, B., Kater, S.B., 1977. Identified higher-order neurons gastropod neurobiology. In: Breidback, O., Kutsch, W. (Eds.), controlling the feeding motor program of Helisoma. Neuroscience The Nervous Systems of Invertebrates: An Evolutionary and 2, 1049–1063. Comparative Approach. Birkhauser Verlag, Boston, pp. 89–113. Granzow, B., Rowell, C.H.F., 1981. Further observations on the Cappell, M.S., Spray, D.C., Susswein, A.J., Bennett, M.V., 1989a. serotonergic cerebral neuronses of Helisoma (Mollusca, Gas- Neuronal analysis of pharyngeal peristalsis in the gastropod Na- tropoda): the case for homology with the metacerebral giant cells. 6anax in terms of identified motoneurons innervating identified J. Exp. Biol. 90, 283–305. muscle bands. I. Muscle band identifiability. Brain Res. 502, Grillner, S., Wallen, P., 1985. Central pattern generators for locomo- 258–265. tion, with special reference to vertebrates. Annu. Rev. Neurosci. Cappell, M.S., Spray, D.C., Susswein, A.J., Bennett, M.V., 1989b. 8, 233–261. Neuronal analysis of pharyngeal peristalsis in the gastropod Na- Harris-Warrick, R.M., Marder, E., 1991. Modulation of neural net- 6anax in terms of identified motoneurons innervating identified works for behavior. Annu. Rev. Neurosci. 14, 39–57. muscle bands. II. Radial and circumferential motor fields. Brain Hembrow, D., 1973. Observations on the structure and function of Res. 502, 266–279. the buccal mass of Planorbarius corneus (L.). Proc. Malac. Soc. Carriker, M.R., 1946. Morphology of the alimentary system of the snail Lymnaea stagnalis appressa Say. Trans. Wis. Acad. Sci. Arts London 40, 505–521. Lett. 38, 1–88. Horwitz, I.S., Senseman, D.M., 1981. Chemosensory stimulation of Church, P.J., Lloyd, P.E., 1994. Activity of multiple identified motor the oral cavity of the snail Helisoma. Chem. Sens. 6, 1–12. neurons recorded intracellularly during evoked feedinglike motor Hoy, R.R., 1974. Genetic control of acoustic behavior in crickets. programs in Aplysia. J. Neurophysiol. 72, 1794–1809. Am. Zool. 14, 1067–1080. Cohan, C.S., Mpitsos, G.J., 1983. The generation of rhythmic activity Hubendick, B., 1957. The eating function in Lymnaea stagnalis. in a distributed motor system. J. Exp. Biol. 102, 25–42. Arkiv. Zool. 10, 511–521. Croll, R.P., 1987. Identified neurons and cellular homologies. In: Ali, Hurst, A., 1968. The feeding mechanism and behavior of the opistho- M.A. (Ed.), Nervous Systems in Invertebrates. Plenum Press, New branch Melibe leonina. Symp. Zool. Soc. London 22, 151–166. York, pp. 41–59. Hurwitz, I., Susswein, A.J., 1996. B64, a newly identified central Davis, W.J., Mpitsos, G.J., Siegler, M.V.S., Pinneo, J.M., Davis, pattern generator element producing a phase switch from protrac- K.B., 1974. Neuronal substrates of behavioral hierarchies and tion to retraction in buccal motor programs of Aplysia californica. associative learning in Pleurobranchaea. Am. Zool. 14, 1037– J. Neurophysiol. 75, 1327–1344. 1050. Hurwitz, I., Goldstein, R.S., Susswein, A.J., 1994. Compartmental- Davis, W.J., Siegler, M.V.S., Mpitsos, G.J., 1973. Distributed neu- ization of pattern-initiating and motor functions in the B31 and ronal oscillators and efference copy in the feeding system of B32 neurons of the buccal ganglia of Aplysia californica.J. Pleurobranchea. J. Neurophysiol. 36, 258–274. Neurophysiol. 71, 1524–1527. Dawkins, M., 1974. Behavioral analysis of coordinated feeding move- Hurwitz, I., Neustadter, D., Morton, D.W., Chiel, H.J., Susswein, ments in the gastropod Lymnaea stagnalis (L.). J. Comp. Physiol. A.J., 1996. Activity patterns of the B31/B32 pattern initiators 92, 255–271. innervating the I2 muscle of the buccal mass during normal Delaney, K., Gelperin, A., 1990. Cerebral interneurons controlling feeding movements in Aplysia californica. J. Neurophysiol. 75, fictive feeding in Limax maximus II. Initiation and modulation of 1309–1326. fictive feeding. J. Comp. Physiol. 166, 311–326. Hurwitz, I., Kupfermann, I., Susswein, A.J., 1997. Different roles of Demian, E.S., 1962. Comparative morphological studies on the buc- neurons B63 and B34 that are active during the protraction phase cal mass of the Lymnaeacea. Gotesborgs K. Vetensk-Vitterh.- of buccal motor programs in Aplysia californica. J. Neurophysiol. Samh. Handl. Ser. B9, 1–30. 78, 1305–1319. A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408 407

Ikeda, K., Kaplan, W.D., 1974. Neurophysiological genetics in Merickel, M., Gray, R., 1980. Investigation of burst generation by Drosophila melanogaster. Am. Zool. 14, 1055–1066. the electrically coupled cyberchron network in the snail Helisoma Jing, J., Gillette, R., 1995. Neuronal elements that mediate escape using a single-electrode voltage clamp. J. Neurobiol. 11, 73–102. swimming and suppress feeding behavior in the predatory sea slug Merickel, M.B., Eyman, E.D., Kater, S.B., 1977. Analysis of a Pleurobranchaea. J. Neurophysiol. 74, 1900–1910. network of electrically coupled neurons producing rhythmic activ- 6 6 Kabotyanski, E.A., Ziv, I., Baxter, D.A., Byrne, J.H., 1994. Experi- ity in the snail Helisoma tri ol is. IEEE Trans. Biomed. Eng. 24, mental and computational analyses of a central pattern generator 277–287. underlying aspects of feeding behavior of Aplysia. Neth. J. Zool Merickel, M., Kater, S.B., Eyman, E.D., 1978. Burst generation by electrically coupled network in the snail Helisoma: analysis using 44, 357–373. computer simulation. Brain Res. 159, 331–349. Kabotyanski, E.A., Baxter, D.A., Byrne, J.H., 1998. Identification Miller, J.P., Selverston, A.I., 1985. Neural mechanisms for the pro- and characterization of catecholaminergic neuron B65, which duction of the lobster pyloric motor pattern. In: Selverston, A.I. initiates and modifies patterned activity in the buccal ganglia of (Ed.), Model Neural Networks and Behavior. Plenum Press, New Aplysia. J. Neurophysiol. 79, 605–621. York, pp. 37–48. Kammer, A.E., Westfall, J.A., 1974. Neuroid and neuronal mecha- Miller, M.W., Rosen, S.C., Schissel, S.L., Cropper, E.C., Kupfer- nisms of coordination in simple systems. Am. Zool. 14, 881–882. mann, I., Weiss, K.R., 1994. A population of SCP-containing Kaneko, C.R., Merickel, M., Kater, S.B., 1978. Centrally pro- neurons in the buccal ganglion of Aplysia are radula mechanoaf- grammed feeding in Helisoma: identification and characteristics of ferents and receive excitation of central origin. J. Neurosci. 14, an electrically coupled premotor neuron network. Brain Res. 146, 7008–7023. 1–21. Morton, D.W., Chiel, H.J., 1993. The timing and activity in motor Kater, S.B., 1974. Feeding in Helisoma tri6ol6is. The morphological neurons that produce radula movements distinguishes ingestion and physiological bases of a fixed action pattern. Am. Zool. 14, from rejection in Aplysia. J. Comp. Physiol. A173, 519–536. 1017–1036. Mpitsos, G.J., Cohan, C.S., 1986. Comparison of differential Pavlo- Kater, S.B., Murphy, A.D., Rued, J.R., 1978. Control of the salivary vian conditioning in whole animals and physiological prepara- glands of Helisoma by identified neurons. J. Exp. Biol. 72, 91– tions of Pleurobranchaea: implications of motor pattern 106. variability. J. Neurobiol. 17, 499–516. Kemenes, G., 1994. Processing of mechano- and chemosensory infor- Murphy, A.D., 1990. An identified pleural ganglion interneuron mation in the lip nerve and cerebral ganglia of the snail Helix inhibits patterned motor activity in the buccal ganglia of the snail, pomatia (L.). Neurosci. Behav. Physiol. 24, 77–87. Helisoma. Brain Res. 525, 300–303. Kemenes, G., Elliott, C.J.H., 1994. Analysis of the feeding motor Murphy, A.D., 1991. The Helisoma cyberchron network revisited: pattern in the pond snail, Lymnaea stagnalis: photoinactivation of what is its role in feeding. In: Kitts, K.S., Boer, H.H., Joosse, J. (Eds.), Molluscan Neurobiology. North Holland, Amsterdam, pp. axonally stained pattern-generating interneurons. J. Neurosci. 14, 25–29. 153–166. Murphy, A.D., 1993. GABA is an excitatory amino acid acting on Kobatake, E., Kawahara, S., Yano, M., Shimizu, H., 1992. Control specific identified feeding motoneurons in the snail, Helisoma. of feeding rhythm in the terrestrial slug Incilaria bilineata. Neu- Soc. Neurosci. Abstr. 19, 1274. rosci. Res. 13, 257–265. Murphy, A.D., Barker, D.L., Loring, J.F., Kater, S.B., 1985a. Kristan, W.B., Jr, 1974. Neural control of swimming in the leech. Sprouting and functional regeneration of an identified serotoner- Am. Zool. 14, 991–1002. gic neuron following axotomy. J. Neurobiol. 16, 137–151. Kupfermann, I., 1974. Feeding behavior in Aplysia: a simple system Murphy, A.D., Lukowiak, K., Stell, W.K., 1985b. Peptidergic modu- for the study of motivation. Behav. Biol. 10, 1–26. lation of patterned motor activity in identified neurons of He- Kyriakides, M.A., McCrohan, C.R., 1989. Effect of putative neuro- lisoma. Proc. Natl. Acad. Sci. USA 82, 7140–7144. modulators on rhythmic buccal motor output in Lymnaea stag- Naitoh, Y., 1974. Bioelectric basis of behavior in Protozoa. Am. nalis. J. Neurobiol. 20, 635–650. Zool. 14, 883–893. Lent, C.M., 1974. Neuronal control of mucus secretion by leeches: Park, J.H., Straub, V.A., O’Shea, M., 1998. Anterograde signaling by toward a general theory of serotonin. Am. Zool. 14, 931–942. nitric oxide: characterization and in vitro reconstitution of an Levinthal, C., 1974. Neural development in isogenic organisms, a identified nitrergic synapse. J. Neurosci. 18, 5463–5476. summary report. Am. Zool. 14, 1051–1054. Pentreath, V.W., Radojcic, T., Seal, L.H., Winstanley, E.K., 1985. Levitan, H., Tauc, L., Segundo, J.P., 1970. Electrical transmission The glial cells and glia-neuron relations in the buccal ganglia of among neurons in the buccal ganglion of a mollusc, Na6anax Planorbis corneus (L.): cytological, qualitative and quantitative inermis. J. Gen. Physiol. 55, 484–496. changes during growth and ageing. Philos. Trans. R. Soc. London Lloyd, P.E., Kupfermann, I., Weiss, K.R., 1988. Central peptidergic B Biol. Sci. 307, 399–455. neurons regulate gut motility in Aplysia. J. Neurophysiol. 59, Perrins, R., Weiss, K.R., 1998. Compartmentalization of information 1613–1626. processing in an Aplysia feeding circuit interneuron through membrane properties and synaptic interactions. J. Neurosci. 18, 3977– London, J.A., Gillette, R., 1984. Functional roles and circuitry in an 3989. inhibitory pathway to feeding command neurones in Pleurobran- Peters, M., Altrup, U., 1984. Motor organization in pharynx of Helix chaea. J. Exp. Biol. 113, 423–446. pomatia. J. Neurophysiol. 52, 389–399. Lorenz, K., 1966. Agression. Harcourt, Brace and World, New York. Ponder, W.F., Lindberg, D.R., 1996. Gastropod phylogeny — chal- Lukowiak, K., Murphy, A.D., 1987. Molluscan model systems for lenges for the 90s. In: Taylor, J.T. (Ed.), Origin and Evolutionary the study of neural peptides. In: Boulton, A.A., Baker, G.B., Radiation of the Mollusca. Oxford, New York, pp. 135–154. Pittman, Q.J. (Eds.), Neuromethods Peptides, vol. 6. The Hu- Quinlan, E.M., Murphy, A.D., 1991. Glutamate as a putative neuro- mana Press, Clifton, NJ, pp. 439–476. transmitter in the buccal central pattern generator of Helisoma Marder, E., Calabrese, R.L., 1996. Principles of rhythmic motor tri6ol6is. J. Neurophysiol. 66, 1264–1271. pattern generation. Physiol. Rev. 76, 687–717. Quinlan, E.M., Murphy, A.D., 1996. Plasticity in the multifunctional Mayeri, E., Koester, J., Liebeswar, G., 1974. Functional organization buccal central pattern generator of Helisoma illuminated by the of the neural control of circulation in Aplysia. Am. Zool. 14, identification of phase 3 interneurons. J. Neurophysiol. 75, 561– 943–956. 574. 408 A.D. Murphy / Progress in Neurobiology 63 (2001) 383–408

Quinlan, E.M., Gregory, K., Murphy, A.D., 1995. An identiﬁed Sossin, W.S., Kirk, M.D., Scheller, R.H., 1987. Peptidergic modula- glutamatergic interneuron patterns feeding motor activity via both tion of neuronal circuitry controlling feeding in Aplysia. J. Neu- excitation and inhibition. J. Neurophysiol. 73, 945–956. rosci. 7, 671–681. Quinlan, E.M., Arnett, B.C., Murphy, A.D., 1997. Feeding stimu- Spencer, A.N., 1974. Non-nervous conduction in invertebrates and lants activate an identiﬁed dopaminergic interneuron that induces embryos. Am. Zool. 14, 917–929. the feeding motor program in Helisoma. J. Neurophysiol. 78, Staras, K., Kemenes, G., Benjamin, P.R., 1998. Pattern-generating 812–824. role for motoneurons in a rhythmically active neuronal network. Richmond, J.E., Bulloch, A.G., Bauce, L., Lukowiak, K., 1991. J. Neurosci. 18, 3669–3688. Evidence for the presence, synthesis, immunoreactivity and up- Stein, P.S.G., 1974. Neural control of interappendage phase during take of GABA in the nervous system of the snail Helisoma locomotion. Am. Zool. 14, 1003–1116. 6 6 tri ol is. J. Comp. Neurol. 307, 131–143. Susswein, A.J., Achituv, Y., Cappell, M.S., Bennett, M.V., 1987. Richmond, J.E., Murphy, A.D., Lukowiak, K., Bulloch, A.G.M., Pharyngeal movements during feeding sequences of Na6anax iner- 1994. GABA regulates the buccal motor output of Helisoma by mis (Gastropoda: Opisthobranchia) in successive stages of dissec- two pharmacologically distinct actions. J. Comp. Physiol. A174, tion. J. Exp. Biol. 128, 323–333. 593–600. Teyke, T., Rosen, S.C., Weiss, K.R., Kupfermann, I., 1993. Do- Rose, R.M., 1971. Patterned activity of the buccal ganglion of the paminergic neuron B20 generates rhythmic neuronal activity in nudibranch mollusc Archidoris pseudoargus. J. Exp. Biol. 55, the feeding motor circuitry of Aplysia. Brain Res. 630, 226–237. 185–204. Thomas, J.D., Nwanko, D.I., Sterry, P.R., 1985. The feeding strate- Rose, R.M., Benjamin, P.R., 1979. The relationship of the central gies of juvenile and adult Biomphalaria glabrata (Say) under motor pattern to the feeding cycle of Lymnaea stagnalis. J. Exp. simulated natural conditions and their relevance to ecological Biol. 80, 137–163. theory and snail control. Proc. R. Soc. London B 226, 177–209. Rose, R.M., Benjamin, P.R., 1981a. Interneuronal control of feeding Vehovsky, A., Elliott, C.J.H., 1995. The hybrid modulatory pattern in the pond snail Lymnaea stagnalis. I. Initiation of feeding cycles / by a single buccal interneurone. J. Exp. Biol. 92, 187–201. generating N1L interneuron in the buccal feeding system of Rose, R.M., Benjamin, P.R., 1981b. Interneuronal control of feeding Lymnaea is cholinergic. Invert. Neuro. 1, 67–74. in the pond snail Lymnaea stagnalis. II. The interneuronal mecha- Weiss, K.R., Kupfermann, I., 1976. Homology of the giant serotoner- nism generating feeding cycles. J. Exp. Biol. 92, 203–228. gic neurons (metacerebral cells) in Aplysia and pulmonate mol- Selverston, A.I., 1974. Structural and functional basis of motor luscs. Brain Res. 117, 33–49. pattern generation in the stomatogastric ganglion of the lobster. Wieland, S.J., Gelperin, A., 1983. Dopamine elicits feeding motor Am. Zool. 14, 957–972. program in Limax maximus. J. Neurosci. 3, 1735–1745. Siegler, M.V.S., Mpitsos, G.J., Davis, W.J., 1974. Motor organiza- Yeoman, M.S., Parish, D.C., Benjamin, P.R., 1993. A cholinergic tion and generation of rhythmic feeding output in the buccal modulatory interneuron in the feeding system of the snail, ganglia of Pleurobranchaea. J. Neurophysiol. 37, 1173–1196. Lymnaea. J. Neurophysiol. 70, 37–50. Smith, D.A., 1988. Radular kinetics during grazing in Helisoma Yeoman, M.S., Vehovszky, A., Kemenes, G., Elliott, C.J.H., Ben- tri6ol6is (Gastropoda: Pulmonata). J. Exp. Biol. 136, 89–102. jamin, P.R., 1995. Novel interneuron having hybrid modulatory- Smith, D.A., 1990. Comparative buccal anatomy in Helisoma (Mol- central pattern generator properties in the feeding system of the lusca, Pulmonata, Basommatophora). J. Morphol. 203, 107–116. snail, Lymnaea stagnalis. J. Neurophysiol. 73, 112–124.