Multifaceted Expression of Peptidergic Modulation in the Feeding System of Aplysia

Abstract

Neuropeptides are present in species throughout the animal kingdom and generally exert actions that are distinct from those of small molecule transmitters. It has, therefore, been of interest to define the unique behavioral role of this class of substances. Progress in this regard has been made in experimentally advantageous invertebrate preparations. We focus on one such system, the feeding circuit in the mollusc Aplysia. We review research conducted over several decades that played an important role in establishing that peptide cotransmitters are released under behaviorally relevant conditions. We describe how this was accomplished. For example, we describe techniques developed to purify novel peptides, localize them to identified neurons, and detect endogenous peptide release. We also describe physiological experiments that demonstrated that peptides are bioactive under behaviorally relevant conditions. The feeding system is like others in that peptides exert effects that are both convergent and divergent. Work in the feeding system clearly illustrates how this creates potential for behavioral flexibility. Finally, we discuss experiments that determined physiological consequences of one of the hallmark features of peptidergic modulation, its persistence. Research in the feeding system demonstrated that this persistence can change network state and play an important role in determining network output.

Graphical Abstract

INTRODUCTION

All behaviors are regulated by neuromodulators, such as neuropeptides. Peptidergic neuromodulation has been extensively studied in a number of experimentally advantageous model systems (e.g., refs 1–4). This Review summarizes ~30 years of research that has studied peptidergic modulation of the feeding circuit of the marine mollusc Aplysia. This effort has been highly successful and resulted in the characterization of a number of novel peptide families. To accomplish this, techniques were developed that were used to target peptides that were of particular interest, e.g., all the bioactive peptides present in a particular neuromuscular system. Since these approaches are applicable to other systems, they are described in brief, along with the methods subsequently used to localize novel peptides to identified neurons.

Further, research in the Aplysia feeding circuit has made important advances in defining the unique role of peptides functioning as cotransmitters. Thus, techniques have been developed to measure peptides released both peripherally (e.g., at the neuromuscular junction) and in the CNS. We describe these techniques and include brief accounts of the results of physiological experiments designed to characterize peptidergic mechanisms of action at both the circuit and cellular level. Finally, we describe current thinking regarding the importance of peptidergic modulation for normal feeding behavior.

CARDIOACTIVE PEPTIDES ARE PRESENT IN THE FEEDING CIRCUIT

The first peptides studied in the feeding system of Aplysia were purified from extracts of molluscan ganglia based on their bioactivity, e.g., their effects on the molluscan heart. Peptides characterized in this manner are FMRFamide⁵ and SCP_B,⁶ which is a member of a group of peptides explicitly referred to as small cardioactive peptides (SCPs).⁷ Subsequently, a second SCP (SCP_A) was identified. SCP_A was originally “predicted” as a result of the cloning of its precursor protein.⁸ Cloning experiments were “comparative”; they took advantage of the fact that peptides are expressed in different amounts in different neurons. Thus, in Aplysia, a particular precursor can be a significant proportion of the total protein in a neuron or can be virtually absent. To clone the SCP precursor, SCP_B immunocytochemistry was used to identify neurons that fell into each category. Two identified cells that contain SCP are the giant neurons B1/B2, which innervate the gut.^8,9 The bag cells (which are not part of the feeding circuit) are not SCP immunoreactive.⁸ A cDNA library was constructed from buccal ganglia, which innervate muscles involved in feeding. The cDNA library was screened with radiolabeled cDNA generated from B1/B2 and from the bag cells. Results were compared to identify mRNA specifically expressed in B1/B2. Subsequently, the predicted structure of SCP_A was confirmed when it was purified from material extracted from the gut of Aplysia.¹⁰

B1/B2 are not the only neurons in the feeding circuit that contain cardioactive peptides. This was originally demonstrated using immunocytochemistry (e.g., refs 9 and 11–14). Importantly, techniques were also developed to verify results biochemically using reverse phase high-pressure liquid chromatography (RP-HPLC) (e.g., refs 9, 11, and 15) (Figure 1A). RP-HPLC experiments were designed to compensate for the fact that a single neuron contains very little peptide. Ganglia were incubated in a radiolabeled amino acid (e.g., ³⁵S methionine) so that neurons would synthesize radiolabeled peptides. To determine whether one of these peptides was the peptide of interest (e.g., an SCP), individual neurons were removed from ganglia and subjected to RP-HPLC in the presence of easily detectable quantities of synthetic material (e.g., synthetic SCP). If radiolabeled and synthetic peptides coeluted, material was subjected to at least one more stage of chromatography under different conditions.

Figure 1.

(A) Procedure used to localize peptides to identified neurons. Neurons were physiologically identified and injected with an inert dye, Fast Green. Ganglia were incubated in a radiolabeled amino acid present in the peptide of interest (e.g., ³⁵S methionine). Dye marked cells were individually dissected and placed in tubes containing nanomolar (easily detectable) quantities of synthetic peptides of interest (e.g., the small cardioactive peptides (SCPs)). The mixture was subjected to RP-HPLC. Positive results were obtained when a radiolabeled peptide had chromatographic properties identical to those of the synthetic material. After Figure 3A in ref 85. (B) The accessory radula closer (ARC) muscle is innervated by two motor neurons, B15 and B16. B15 and B16 both contain acetylcholine (ACh) in small clear vesicles (small open circles). Additionally, members of two peptide families, the small cardioactive peptides (SCPs) and buccalins (BUCs) are colocalized in dense core vesicles (large shaded circles) in B15. Members of a third peptide family, the myomodulins (MMs) are colocalized with the BUCs in B16. (C) Procedure used to purify peptides from the ARC neuromuscular system. An ARC muscle extract was prepurified to remove large proteins and salts, and then was subjected to RP-HPLC. All of the resulting fractions were bioassayed, which yielded a number of peaks of bioactivity. For simplicity, only two peaks are shown (red and blue). All bioactive fractions were rechromatographed, until optically pure material was obtained. Optically pure peaks were subjected to gas phase sequence analysis. After Figure 2A in ref 85. (D) Convergent and divergent postsynaptic effects of the SCPs and MM_A. The SCPs and MMA all increase cAMP levels in the ARC, activate PKA, and increase muscle relaxation rate and contraction size. The latter effect is mediated by an enhancement of a calcium current. Additionally, at high doses MM_A decrease contraction size, by inducing a potassium current. After Figure 1A from ref 86. (E) Preparation used to measure peptide release from the accessory radula closer (ARC) motor neurons. Motor neurons were impaled with two electrodes (one to inject current (I), the other to monitor changes in membrane potential (V)). The ARC muscle was suspended outside the recording chamber and was encased in grease and parafilm to prevent dehydration (not shown). The muscle was perfused via an artery and the perfusate was collected as drops, which fell directly into radioimmunoassay (RIA) tubes. After Figure 1 from ref 81.

Once peptides had been localized to specific neurons in the feeding circuit, an obvious next question was, are these peptides acting as neurotransmitters? This issue was initially addressed in experiments that took advantage of the large size of the SCP-positive neurons B1/B2.^8,9 In one study, the subcellular SCP distribution was determined using electron microscopy and immunocytochemistry.¹⁶ These experiments demonstrated that the SCPs are present in large dense-core vesicles in the somatic cytoplasm of B1/B2, in neurites within the neuropil, and in an axon in a peripheral nerve. Other experiments conducted in cultured neurons demonstrated that the SCPs are released from single cells in a calcium and activity dependent manner.¹⁷ These data taken together with previous demonstrations of bioactivity strongly supported the idea that peptides were acting as neurotransmitters in the feeding circuit.

In summary, early studies of cardioactive peptides laid a foundation for much later research in the feeding system. They were the first to establish that neuropeptides function as transmitters in this circuitry, and techniques and tools were developed that facilitated subsequent research.

THE ARC NEUROMUSCULAR SYSTEM

To determine the physiological role of peptide cotransmission, experiments were initially conducted in experimentally advantageous neuromuscular systems including a B1/B2-gut preparation,¹⁸ the radula opener neuromuscular system,^19,20 the I2 neuromuscular system,^21,22 and the I3 neuromuscular system.^23–28 For brevity, we will focus on the system in which the most progress was ultimately made, the accessory radula closer (ARC) muscle and its two cholinergic motor neurons B15 and B16²⁹ (Figure 1B).

The ARC neuromuscular system was of particular interest because it had been used to study peripheral aspects of a behavioral state referred to as food induced arousal. This state is manifested as progressive increases in the speed and strength of biting as animals begin to feed.^30,31 Food-induced arousal is in part mediated by modulatory, potentiating effects of serotonin released from the giant metacerebral cells (MCCs).^32–37

Early experiments demonstrated that one of the ARC motor neurons (B15) contains the SCPs³⁸ (Figure 1B). Further, early work demonstrated that the SCPs exert modulatory effects at the neuromuscular junction that are similar to those of serotonin.³⁹ This led to the suggestion that parallel peptidergic and serotonergic pathways might mediate similar aspects of arousal.³⁹ Subsequent research described below sought to determine whether this was the case, or whether peptidergic modulation played a distinct role in behavior.

Characterization of Novel Neuropeptides.

Initially, however, a question addressed was, how many different peptides are actually present in the ARC neuromuscular system? To answer this question, material was extracted from ARC muscles and peptides were purified using sequential RP-HPLC (Figure 1C). Fractions of interest were identified by their bioactivity in the ARC neuromuscular system itself. Namely, muscle contractions were generated by stimulating a motor neuron, and aliquots of HPLC fractions were exogenously applied to determine whether there was an effect on contraction parameters. This was a highly successful approach that resulted in the purification and sequencing of a number of novel neuropeptides. Six peptides are myomodulins (MMs): MM_A, MM_B, MM_C, MM_D, MM_E, and MM_F.^40–42 E MM_F Using biochemical techniques, two of these peptides were localized to the non-SCP containing ARC motor neuron (B16)^40,41 (Figure 1B). Three peptides are buccalins (BUCs): BUCa,⁴³ BUCb,⁴⁴ and BUCc.⁴⁵ These peptides are present in both ARC motor neurons B15 and B16^43,44,46 (Figure 1B). Finally, the sequences of three peptides terminated in Phe-Arg-Phe-amide. Consequently, the peptides were named FRF_A, FRF_B, and FRF_C.⁴⁷ The FRFamide peptides are present in sensory neurons that innervate the ARC muscle.⁴⁸

These results suggested that members of three novel “families” of peptides had been purified: the MMs, BUCs, and the FRFamides. To determine whether this was the case, relevant peptide precursor proteins were characterized. This was initially accomplished using oligonucleotides designed from the sequence of known peptides to isolate clones from a buccal ganglion cDNA library.^45,49 Using these techniques, a precursor was characterized that contains 19 peptides with a buccalin-like sequence.⁴⁵ Three of the sequences are identical to the peptides purified from the ARC neuromuscular system. The other 16 are present as a single copy and were designated D through S in order of their appearance on the precursor protein.

Also characterized was a precursor that contained seven different MMs: MM_A (10 copies) MM_B (1 copy), MM_D (1 copy), MM_F (1 copy), MM_G (1 copy), MM_H (1 copy), and MM_I (1 copy).^49,50 Three are single copy peptides that were not originally purified from the ARC neuromuscular system (MM_G, MM_H, and MM_I). What was initially more puzzling, however, was that two peptides purified from the ARC neuromuscular system were not present on the cloned precursor (MM_C and MM_E). These peptides were referred to as the orphan MMs until a second gene (MM gene 2 (MMG2)) was identified.⁵¹ This gene encodes both MM_C and MM_E as well as four novel peptides (MMG2-DPa, MMG2-DPb, MMG2-DPd, and MMG2-DPf). The distribution of MMG2 containing neurons was determined using both in situ hybridization and immunocytochemistry. Somewhat surprisingly, MMG2 is mostly expressed in the pedal ganglion,⁵¹ which is involved in the control of locomotion,^52,53 and not generally associated with feeding. Physiological experiments demonstrated, however, that MM containing pedal neurons modulate ARC neuromuscular activity.⁵¹

The FRFamide peptides are similar to a well characterized invertebrate neuropeptide, FMRFamide, in that both terminate in Arg-Phe amide. A question of interest therefore was, are these peptides all members of one peptide family? In part, this question was addressed by cloning the FRFamide precursor using degenerate primers designed based on the amino acid sequences of the FRFamides.⁴⁸ The cloned precursor does not contain FMRFamide, but does contain the three FRFamide peptides purified from the ARC neuromuscular system. It also encodes two single copy FRFamides (FRF_D and FRF_E) that were not originally purified, but which are expressed in the buccal ganglion.⁴⁸ These data indicate that the FRFamides and FMRFamide belong to different families. This idea was supported by localization experiments that demonstrated that the FRFamide and FMRFamide precursors are expressed in different populations of feeding sensory neurons.⁴⁸

In general, cloning experiments made an important contribution to the characterization of ARC peptides in that they demonstrated they there are members of four (not three) families of novel peptides (two families of MMs, BUCs, and FRFamides). Further, it was this work that led to the identification of the pedal input to the ARC.

The novel peptides that were identified as a result of their bioactivity on the ARC muscle are not specific to that system. They are present in other parts of the Aplysia nervous system (e.g., refs 54 and 55), and they (or related peptides) are present in other species. To give some examples, MMs or MM-like peptides have been reported in worms (e.g., leeches⁵⁶ and Caenorhabditis elegans^64,65), insects (i.e., locusts⁶⁶ and fruit flies⁶⁷), crustaceans (e.g., crabs^57,58 and lobsters⁵⁹), snails (e.g., Helix⁶⁰ and Lymnaea^61,62), and other sea slugs (e.g., Clione⁶³). BUCs or buccalin-like peptides have been reported in the crab,^57,58 Clione,⁶³ Lymnaea,⁶¹ and C. elegans.^64,65 Further, two recent bioinformatic studies of neuropeptide precursors and receptors have suggested that the MMs⁶⁸ and BUCs⁶⁹ are present in other species as well.

Physiological Role of Peptide Cotransmitters.

Some of the ARC peptides exert effects that are similar to those of serotonin. Serotonin acts postsynaptically and increases the amplitude and relaxation rate of motor neuron induced contractions via a cAMP dependent mechanism.^32–34 Generally speaking, the SCPs³⁹ and some of the MMs^40−42,70 do the same. The increase in amplitude is at least in part mediated by an enhancement of a dihydropyridine-sensitive, high threshold calcium current⁷¹ (Figure 1D). The effect on relaxation rate is mediated by activation of PKA,^70,72 and the subsequent phosphorylation of an Aplysia version of a molecule that regulates muscle contractions in other species (twitchin).^73,74 This similarity in bioactivity led to the suggestion that serotonergic and peptidergic pathways might play parallel roles in behavior, i.e., that they might simultaneously exert similar actions.

To determine whether this is the case, one set of experiments sought to determine when the ARC peptides are released during normal feeding. The prevailing view at the time held that peptides release only occurs when neurons fire at relatively high frequencies, e.g., during times of stress.⁷⁵ This type of model contrasted with what had been demonstrated for serotonergic modulation. The MCCs begin to fire when animals are initially exposed to food (i.e., before consummatory behavior is initiated), and their firing frequency decreases as animals begin to feed.³⁵ This suggests that serotonergic modulation plays a preparatory role in normal behavior.

To determine when the ARC peptides are released, the firing patterns of the ARC motor neurons were characterized in intact animals.^76,77 These firing patterns were simulated in vitro, and peptide release was initially detected indirectly (e.g., by determining whether a particular pattern of motor neuron activity produced an increase in cAMP levels in the ARC muscle^77–79). Later experiments monitored peptide release directly using peptide radioimmunoassays (RIAs)^80–83 (Figure 1E). These experiments were important because they established that peptides are released even when the ARC motor neurons fire at the low end of their physiologically relevant range. Further, it was possible to determine release dynamics.⁸³ Thus, low levels of peptide were detected after only a few minutes of stimulation, but with maintained stimulation, release facilitated dramatically over time, peaked at about 40 min, and then declined to some extent. Depletion was not observed. Even after hours of motor neuron activity, peptide levels in terminals were only reduced by ~20%.⁷⁷ Taken together, this suggests that peptidergic modulation begins slowly as consummatory feeding is initiated, and is maintained throughout the behavior. If so, modulation is to some extent “serial” rather than “parallel” with serotonergic modulation preceding peptidergic modulation.

Also arguing against a “parallel” type of model is the fact that effects of some of the ARC peptides are distinctly different from those of serotonin. Namely, the BUCs, the FRFamide peptides, and MM_A at high doses decrease rather than increase, the amplitude of motor neuron induced contractions.^{41,43,44,46–48} Inhibitory effects of the BUCs are presynaptically mediated, i.e., there is a decrease in the amount of acetylcholine released from the ARC motor neurons.^43,44,46 Effects of the FRFamides and inhibitory effects of MM_A are postsynaptically mediated, i.e., these peptides activate a 4-aminopyridine (4-AP) sensitive potassium current in the ARC muscle⁸⁴ (e.g., Figure 1D).

Some of the peptides that exert inhibitory effects are in sensory neurons that innervate the ARC muscle. This is the case for the FRFamides and FMRFamide.⁴⁸ The physiological significance of this input is not currently understood. Other peptides, the BUCs, are in the ARC motor neurons themselves.⁴³ They are present in the dense core vesicles that contain potentiating peptides, and potentiating and depressing peptides are coreleased^81,82 (Figure 1B). At first this seems counterintuitive. It becomes less so when the fact that effects of the two types of peptides are actually only “partially” antagonistic is taken into account (i.e., there is an antagonistic effect on contraction amplitude, but depressing peptides are unlike potentiating peptides in that they do not alter muscle relaxation rate) (Figure 1D). Consequently, even if equipotent amounts of potentiating and depressing peptides were to be released, parameters of muscle contraction would be altered in that relaxation rate would be increased. Further, the inhibitory BUCs are present and are released in smaller amounts than peptides with primarily potentiating effects (the SCPs).^81,82 This suggests that at least under some conditions effects of potentiating peptides on contraction amplitude may not be completely opposed by effects of depressing peptides.

Figure 2 illustrates the current thinking concerning why corelease of potentiating and depressing peptides may be beneficial.⁸⁵ When animals begin to feed, bite magnitude and frequency increase (at least in part due to effects of the serotonergic MCCs) (Figure 2B1, B2). Without peptidergic modulation, the ARC might not have time to fully relax between bites, i.e., a certain level of baseline tension might develop (Figure 2B1). This would be detrimental because contractions of the ARC close the food-grasping organ, the radula. Consequently, if there is tension in the ARC, it will be relatively difficult to open the radula. In fact, under some circumstances, opening might not occur at all (as shown in Figure 2B2). Peptidergic modulation potentially alleviates this situation. Thus, the SCPs and MMs increase the relaxation rate of the ARC, which will produce a decrease in its baseline tension (Figure 2C1). Although this will tend to facilitate opening there may be circumstances where opening still does not occur (as shown in Figure 2C2). In this situation, a partial decrease in contraction amplitude may be necessary (as would occur if the BUCs are released with the SCPs and MMs) (Figure 2D1, D2). This model suggests that peptidergic modulation will be particularly important when animals feed quickly. Consistent with this idea, peptide release per action potential is increased under these conditions.^80,82

Figure 2.

Schematic representation of effects of modulatory peptides on the accessory radula closer (ARC) neuromuscular system. A1–D1 show contractions of the radula closer (magenta) and opener (cyan) muscles separately. A2–D2 show the integrated output of the system. In A2–D2, the dashed line marks the neutral position. When plots are above the dashed line, the radula is closed. When plots are below the line, the radula is open. In A1 and A2, the cycle period is long enough to allow both the closer and opener muscles to fully relax before the next contraction. In B1 and B2, the frequency and amplitude of the contractions have been increased. The muscles no longer have time to fully relax and radula opening no longer occurs because the opener is a weaker muscle. In C1 and C2, the contraction frequency and amplitude are the same as in B1 and B2, but the relaxation rate of the contractions has been increased (as would be the case with release of the SCPs and/or MMs). The contractions now relax to a greater degree, but functional opening may not occur. In D1 and D2, inhibitory effects of peptides on amplitude have been added (e.g., BUC effects). In this situation, functional opening is more apt to occur.

A final difference between peptidergic and serotonergic modulation is that peptidergic potentiation is mediated by more than one neurotransmitter (e.g., by the SCPs which are in one motor neuron and the MMs which are in another neuron). At first glance, this might also seem counterintuitive, since the actions of these two classes of peptides are partially convergent (e.g., they both increase relaxation rate) (Figure 1D). It has been suggested that a key to understanding the physiological significance of this apparent redundancy stems from the divergent actions of the two classes of peptides. MM_A and the SCPs differentially activate the 4-AP sensitive K current in the ARC, i.e., MM_A effects are larger than SCPs effects. (The 4-AP sensitive K current depresses muscle contractions.)⁸⁶ (Figure 1D). Divergence on its own couples the effects of a transmitter.⁸⁶ For example, a particular concentration of MM_A on its own will produce a contraction that has a certain amplitude:relaxation ratio. If the MM_A concentration is altered, the amplitude/relaxation ratio may change, but in a MM_A specific manner, e.g., if more MM_A is released, relaxation rate will increase but eventually amplitude will decrease. With MM_A alone, it becomes impossible to produce a contraction that has a very fast relaxation rate but does not have a depressed amplitude. A contraction of this nature is more readily produced if the total peptide concentration is increased by adding SCP to MM_A, (instead of increasing the concentration of MM_A). SCP is like MM_A in that it increases relaxation rate, i.e., their actions converge. However, it less effectively depresses contraction amplitude. The net effect of this arrangement is that with both peptides present, it is possible to produce contractions that could not be produced by a single modulator alone. Thus, when convergence and divergence are combined, the potential for plasticity is increased.

In summary, studies of the ARC neuromuscular system played an important role in establishing that peptide cotransmitters impact the execution of normal behavior. They also illustrated how this form of modulation can greatly enhance the ability of a system to produce multiple, behaviorally appropriate outputs.

PEPTIDES IN INPUTS TO THE FEEDING CENTRAL PATTERN GENERATOR (CPG)

As described above, initial studies of peptide cotransmitters in the feeding system were conducted in neuromuscular systems, which have clear technical advantages. More recent research has focused on the CNS, in part on peptides that are present in “egestive” and “ingestive” inputs to the circuit that mediates consummatory feeding, i.e., the feeding central pattern generator (CPG) (Figure 3A).

Figure 3.

(A) Ingestive motor activity is triggered when food activates sensory neurons that excite cerebral buccal interneurons (CBIs) such as CBI-2. CBI-2 is a cholinergic neuron that contains FCAP (feeding circuit activating peptide) and CP-2 (cerebral peptide-2). CBI-2 activates the feeding central pattern generator (CPG) and with repeated stimulation, an ingestive motor program is induced. Egestive motor activity is triggered when afferents with processes in the esophageal nerve (EN) are activated, e.g., by the presence of an inedible object. EN afferents also activate the feeding CPG and contain the modulatory peptides SCP (small cardioactive peptides), apNPY (Aplysia Neuropeptide Y), FMRFamide, and the RFamide peptides. With repeated stimulation of the EN, egestive motor programs are triggered. (B) In the CNS, released peptides were collected using solid phase extraction beads, which were positioned on the buccal ganglion in the vicinity of the EN. Peptide release was induced by stimulating the EN. (C) To identify peptides transported from the cerebral to the buccal ganglion the cerebral buccal connective (CBC) was cut. Hours later, MALDI was performed on the cut ends of the CBC. After Figure 1A from ref 99.

Identification of Peptides in “Egestive” Inputs to the CPG.

The egestive response that has been commonly studied in Aplysia occurs when an inappropriate object is ingested and makes contact with the esophagus (Figure 3A).^31,35 The esophagus is innervated, at least in part, by the esophageal nerve (EN).⁸⁷ EN afferents contain some of the peptides that are present in neuromuscular systems, i.e., the SCPs,⁸⁸ FMRFamide, and the RFamide peptides (Figure 3A).⁴⁸ Additionally, they contain an Aplysia version of a peptide that alters food intake in other species, neuropeptide Y.⁸⁹

Aplysia neuropeptide Y (apNPY) was originally purified from the abdominal ganglion.⁹⁰ It was subsequently localized to EN afferents, and its CNS release in response to neural activation demonstrated.⁸⁹ CNS release experiments could not utilize the methods developed to study release in the periphery. In the periphery, the ARC muscle was separated from the recording chamber and perfused using an attached artery (Figure 1E). The perfusate formed drops that fell directly into RIA test tubes. This prevented losses that would have occurred as a result of handling and transfer. Anatomical differences make it impossible to use this approach in the CNS. Instead, solid-phase extraction beads were placed on the rostral surface of the buccal ganglion near the EN, where the neuropile is close to ganglion surface (Figure 3B).⁸⁹ Released peptides were eluted from these beads directly onto a matrix-assisted laser desorption/ionization (MALDI) target surface. MALDI time-of-flight mass spectrometry (MALDI-TOF MS) was used to identify apNPY.⁸⁹ This technique has also been used to monitor peptide release in a vertebrate model system.⁹¹ Further, in Aplysia, it has been used to monitor release of the other peptides present in EN afferents.⁸⁸

In summary, the EN, which is an egestive input to the feeding CPG, contains multiple peptides, i.e., FMRFamide, the FRFamides, the SCPs, and apNPY.

Identification of Peptides in “Ingestive” Inputs to the CPG.

Ingestive behavior is triggered when Aplysia make contact with food. This activates commandlike neurons in the cerebral ganglion such as cerebral-buccal interneuron 2 (CBI-2) (Figure 3A).^92,93 CBI-2 contains multiple peptides including cerebral peptide 2 (CP-2).^94–96 CP-2 was originally purified using an approach designed to characterize peptides synthesized in the cerebral ganglion and transported to other ganglia.⁹⁷ More specifically, the cerebral ganglion was incubated in ³⁵S methionine for sufficient time to allow for both synthesis of radiolabeled peptides and fast axonal transport to other parts of the CNS. Other ganglia were subjected to RP-HPLC, and fractions containing radioactivity were used to guide the purification of CP-2 from extracts of cerebral ganglia. Subsequently, the CP2 distribution was mapped using immunocytochemistry⁹⁴ and in situ hybridization after the precursor was cloned.⁹⁵

Other peptides localized to CBI-2 are the feeding circuit activating peptides (FCAPs)⁹⁸ (Figure 3A). The FCAPs were also identified by targeting peptides transported out of the cerebral ganglion. In this situation, however, MALDI TOF MS was used to detect peptides in the cerebral buccal connective (CBC) that had a unique molecular mass⁹⁹ (Figure 3C). Thus, the CBC was severed, and preparations were placed in organ culture to allow time for transport and accumulation of material at the CBC cut ends (Figure 3C). Since peptides that were being transported to the buccal ganglion were of particular interest, unidentified peaks on the cerebral side of the cut end were chosen for further characterization.⁹⁹ Two FCAPs were partially purified from buccal ganglia, and their amino acid sequences were used to clone a precursor, which encodes eight FCAPs.⁹⁹ Interestingly, the FCAPs can initiate feeding motor programs.⁹⁹ They are present in CBI-11,¹⁰⁰ as well as CBI-2.

In summary, one set of peptides has been localized to egestive inputs to the feeding CPG. A second set has been localized to the command-like neuron CBI-2, which is an “ingestive” input to the CPG. This suggests that a form of chemical coding is utilized in this context.¹⁰¹

Physiological Effects of “Egestive” and “Ingestive” Peptides.

“Egestive” and “ingestive” peptides play an important role in configuring activity in the feeding circuit. Interestingly, this occurs with slow dynamics. For example, when motor programs are triggered by stimulating CBI-2, the first cycle of activity that is generated is actually not ingestive. Instead, motor activity is poorly “articulated”, e.g., antagonistic motor neurons are coactive (Figure 4A).^102–105 As CBI-2 stimulation continues, however, the FCAPs and CP-2 are released⁹⁸ and exert modulatory effects.^{96,104,106,107} In general, they upregulate the excitability of neurons that tend to make programs ingestive, and decrease the excitability of interneurons that tend to make programs egestive. This occurs gradually with cycles of activity becoming progressively more and more ingestive (e.g., refs 102 and 103) (Figure 4B).

Figure 4.

(A) Cycles of motor activity were triggered by stimulating CBI-2 and classified as intermediate or ingestive by determining the firing frequency of a radula closer motor neuron (RC) during the radula protraction phase of the motor program (X-axis) vs the radula retraction phase (Y-axis) and computing the ratio (e.g., ref 123). Activity is intermediate when the radula closer motor neuron fires at a low frequency during both protraction and retraction. Activity is ingestive when there is an increase in the frequency of the radula closer during retraction. Insets are typical intracellular recordings from a radula closer motor neuron (top trace) and extracellular recordings from the I2 nerve (I2) (bottom trace). Activity in the I2 nerve marks the protraction phase of the motor program. Note that the first cycle that was triggered (cycle 1) was intermediate. With repeated input activation, however, activity became clearly ingestive. Data are replotted from ref 104. (B) Progressive changes in motor programs (bottom trace) are observed when they are triggered with a short interstimulus interval (ISI) (as is the case during the time indicated by the gray panel), e.g., activity becomes progressively more ingestive when it is triggered by CBI-2. This results in progressive improvements in performance, which can be measured in a number of ways (e.g., as progressive increases in response magnitude (top trace)). It has been suggested that progressive effects result from cumulative effects of modulators (middle panel), which summate when there is a short ISI. After Figure 5 from ref 108. (C) Representational difference analysis (RDA). Left: Procedures for RDA. Identified cells are isolated. The cell that contains the peptide of interest is the tester; the other cell is the driver. cDNA is amplified from RNA from both cells. RDA is then performed using an excess of driver cDNA. Right: Abundant RNAs that are present in tester neurons but not driver neurons (tester-driver) appear as prominent difference bands. After Figure 2 from ref 22.

When the egestive input to the CPG (the EN) is stimulated, motor programs are also configured with slow dynamics. Again, the first cycle of the motor program generated is poorly articulated, but as stimulation is maintained, program definition occurs. EN peptides up regulate the excitability of neurons that tend to make programs egestive, and down regulate the excitability of neurons that tend to make programs ingestive (e.g., refs 88 and 101). Data suggest that, with both EN and CBI-2 stimulation, slow dynamics reflect the fact that effects of peptides are second messenger mediated and persistent.¹⁰⁸ Consequently, they summate and become progressively larger as cycles of activity are repeatedly evoked (Figure 4B).

Because effects of peptides outlast periods of neural activity, repeated activation of the feeding network results in a persistent change in its “state”, i.e., in its circuit parameters. As a result, consequences of input activation depend on recent history. For example, although CBI-2 triggers poorly articulated motor activity if it is stimulated in a quiescent preparation, it triggers ingestive activity if it is stimulated in a preparation in which it has recently been active (e.g., ref 102). Thus, the pre-existing state has a “positive” effect if the same input is repeatedly activated. Interestingly, the opposite effect can be observed when there is a change in input activation, as will occur when animals task switch. For example, if the EN is repeatedly stimulated to create an egestive state, subsequent stimulation of CBI-2 produces an egestive (rather than ingestive) cycle of motor activity.¹⁰²

In summary, studies of peptide cotransmission in the CNS have demonstrated that peptides play an important role in configuring motor activity and determining the current state of the feeding network. In turn, the network state is critical for determining network output.

OTHER PEPTIDES IN THE FEEDING CIRCUIT

Other peptides have been localized to feeding circuitry but exert effects that are not as well understood at the cellular/molecular level. Some of these peptides were identified in experiments that sought to determine whether Aplysia express peptides present in other molluscs. For example, immunoreactivity to the MIPs (Mytilus inhibitory peptides) was observed in a number of neurons in the cerebral and buccal ganglia motivating the search for Aplysia homologues. Five Aplysia MIP-related peptides (AMRPs) were subsequently purified from the CNS using HPLC and an immunoassay to recognize fractions of interest.¹⁰⁹ Further, the AMRP precursor was cloned, and nine additional peptides were identified, all of which are expressed.¹⁰⁹ The AMRPs have a widespread distribution in Aplysia and are present in peripheral tissues such as the digestive tract (as well as the CNS).¹⁰⁹

In another study, MALDI-MS was used to purify a substance that data suggested might be the Aplysia insulin (AI).¹¹⁰ The sequence of this peptide was then used to clone the AI precursor protein.¹¹⁰ AI is present in neurons in the cerebral ganglion and is transported to neurohemal release sites.¹¹⁰ AI mRNA decreases with food deprivation, and injections of AI reduce hemolymph glucose levels.¹¹⁰

Other experiments characterized CNS peptides that are novel. For example, cerebral peptide-1 was purified in early experiments that sought to identify peptides transported to the buccal ganglion.¹¹¹ The CP-1 precursor was subsequently cloned, which led to the identification of another neuropeptide APGWamide located on the same precursor.¹¹² APGWamide is of particular interest, since it is present in a CBI that is electrically coupled to CBI-2, the neuron CBI-3.¹¹³ CBI-3 plays a role in motor program switching,.^113–115

Other novel peptides were discovered based on their bioactivity and presence in both the CNS and gut. Ten peptides identified in this manner (the enterins) mostly share the sequence HSFVamide at their C terminus.¹¹⁶ These peptides were confirmed as a family with the cloning of the enterin precursor, which encodes a total of 35 copies of 20 different peptides.¹¹⁶ The enterins are present in neurons in the buccal and cerebral ganglia and tend to make motor programs more ingestive.¹¹⁶ Other brain/gut peptides are related to the pentapeptide Pro-Arg-Gln-Phe-Val-amide, which is referred to as PRQFVamide.¹¹⁷ PRQFVamide is present as 33 copies on its precursor protein, along with four structurally related pentapeptides.¹¹⁷ Interestingly, PRQFVamide is present in B50, an identified neuron that both initiates and modulates feeding motor programs.¹¹⁸

Other novel peptides were characterized using enhanced representational difference analysis (RDA). This approach utilizes two types of identified neurons; neurons that are not peptidergic (referred to as driver neurons) and neurons that potentially contain novel peptides (referred to as tester neurons) (Figure 4C). The general idea is to identify differences in the mRNA sequences of the two types of cells. To accomplish this, driver sequences are subtracted from tester sequences. As a result, abundant RNAs that are present in the tester (but not the driver) appear as difference bands. These bands are then cloned. This approach was used to characterize a precursor that contains an Aplysia urotensin II (apUII), which has been localized to cerebral A-cluster cells and buccal sensory neurons.¹¹⁹ Electrophysiological data suggest that apUII plays a role in satiety and/or aversive signaling.¹¹⁹ Other peptides characterized with RDA are Aplysia allatotropin-related peptide (ATRP)²² and the Aplysia leucokinin-like peptides (ALKs).¹²⁰ ATRP and the ALKs modify parameters of feeding motor programs.^22,120 Interestingly, ATRP has been localized to a CBI, CBI-4.²²

Lastly, techniques have been developed to detect d-amino acid containing peptides (DAACPs). These peptides result from a post-translational modification of an l-amino-acid-containing counterpart and are notoriously difficult to detect in any species. To address this issue, Livnat et al.¹²¹ developed a discovery funnel that consists of three steps: the identification of peptides that are resistant to digestion by aminopeptidase M, a chiral analysis, and a confirmation step that consists of synthesizing the putative peptide and comparing its chromatographic properties to those of a standard. This method was used to identify a d-amino acid peptide bioactive in the feeding circuit, GdYFD.¹²¹ GdYFD and a peptide on the same precursor (GdFFD¹²²) are some of the first DAACPs identified with a clear function in a well-defined neural circuit.

In general, these experiments indicate that peptides do more than simply specify whether activity is ingestive or egestive. Additionally, peptides modify other parameters of feeding motor programs.

INSIGHTS DERIVED FROM THIS WORK

As a result of work in the feeding system a number of novel neuropeptides were characterized that have subsequently been localized to other species. Perhaps more importantly, however, the research described in this report has made a major contribution to the effort to define the unique physiological role of peptide cotransmission. For example, neuromuscular work clearly established that peptides are released during normal behavior and do not solely mediate responses to stress. Instead they greatly increase the potential for plasticity during normal feeding behavior. This research demonstrated how this could be accomplished via modifications of the neuromuscular transform. Further, subsequent research conducted in the CNS demonstrated that peptides act centrally as well as peripherally and modify the ingestive vs egestive state of the feeding network. State changes are relatively persistent (i.e., last for minutes), and research in the feeding system demonstrated that they can play an important (or predominant) role in determining network output.