Identification of a New Neuropeptide Precursor Reveals a Novel Source of Extrinsic Modulation in the Feeding System of Aplysia

Introduction

In addition to classical neurotransmitters, many neurons also contain and release neuropeptides (Strand, 1999). These neuropeptides modulate a variety of processes in the nervous system and in the periphery. Furthermore, multiple neuropeptides often converge to modulate the very same processes. In this respect, one of the best-studied preparations is the Aplysia feeding system, in which a number of novel neuropeptides have been identified (Cropper et al., 1987a,b, 1988, 1994; Church and Lloyd, 1991; Church et al., 1993; Brezina et al., 1995, 1996; Vilim et al., 1996, 2000; Evans et al., 1999; Hurwitz et al., 2000; Morgan et al., 2000, 2002; Jing and Weiss, 2001; Sweedler et al., 2002; Koh et al., 2003).

The experimental tractability of the Aplysia system has then made it possible to investigate how the convergence and interaction of these multiple modulators contribute to behavioral plasticity.

The myomodulins (MMs) have been particularly well studied. The MMs were originally purified from one of the feeding muscles, the accessory radula closer (ARC) muscle (Cropper et al., 1987b, 1991; Brezina et al., 1995). A gene that codes for MM peptides has been cloned (Miller et al., 1993). An antibody against the MMs stains one of the ARC motor neurons, neuron B16 (Miller et al., 1991). Immuno-electron microscopy has localized the MMs to dense-core vesicles at the B16-ARC neuromuscular junction (Vilim et al., 2000). Stimulation of B16 at physiological frequencies results in MM release (Vilim et al., 2000). Once released, the MMs modulate several ionic currents in the ARC muscle fibers (Brezina et al., 1994a,b) and alter a number of parameters of the ARC muscle contraction (Cropper et al., 1987b, 1991; Brezina et al., 1995; Orekhova et al., 2003). Computer modeling has shown how convergent actions of the MMs and other peptides that also modulate the ARC muscle contribute to behavioral plasticity (Brezina et al., 2003a,b, 2005).

However, our understanding of MM action has been incomplete. The origin of two of the MMs, MMc and MMe, has been elusive. Although purified from the ARC muscle (Brezina et al., 1995), they are not encoded on the known MM gene (Miller et al., 1993). [In previously published abstracts of this work, these two MMs were therefore referred to as “orphan myomodulins” (Friedman et al., 2003).]

In this study, we identify a gene that encodes these two missing members of the MM family. We refer to this gene as MM gene 2 (MMG2). In addition to MMc and MMe, the gene encodes four other neuropeptides, structurally unrelated to the MMs. Using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI), we demonstrate that these novel MMG2-derived peptides (MMG2-DPs), as well as MMc and MMe, are actually synthesized and cleaved from the MMG2 precursor. We then demonstrate that some of the MMG2-DPs are bioactive in the feeding central pattern generator (CPG) and on the ARC muscle. Their actions are qualitatively distinct from those of MMc and MMe, and of the other MMs.

Surprisingly, however, MMG2 is not expressed in motor neuron B16 or B15, the other motor neuron of the ARC muscle. Instead, MMG2 is expressed in a population of neurons in the pedal ganglion. Through a combination of electrophysiological and immunohistochemical techniques, we demonstrate that these neurons project to the buccal ganglion and very likely to the ARC and other feeding muscles. In answering the question of the origin of MMc and MMe, this study therefore reveals an unexpected new source of modulation that converges onto the ARC muscle and the Aplysia feeding system generally.

Materials and Methods

Animals. Aplysia californica weighing 100-350 g were obtained from the University of Miami Aplysia Research Facility (Miami, FL), Pacific Biomarine (Venice, CA), and Marinus (Long Beach, CA) and maintained at 14°C.

Cloning. Standard molecular techniques (Sambrook et al., 1989) were used, except where noted. Seminested rapid amplification of cDNA ends (RACE) was performed on a Robocycler Gradient 40 (Stratagene, La Jolla, CA) with multiple annealing temperatures, as described previously (Furukawa et al., 2001; Sweedler et al., 2002). Degenerate antisense oligos designed to amino acid sequences of MMc and MMe were used in sequential PCR with two bluescript vector primers (BS1, ACCATGATTACGCCAAG; BS2, AATTAACCCTCACTAAAG). Either oligo-dT-primed Aplysia ganglionic cDNA Uni-Zap (Stratagene) lambda phage library or random-primed Aplysia ganglionic cDNA lambda phage library was used as the target for the PCR. Oligonucleotides were obtained from Operon (Alameda, CA) and used at a final concentration of 0.5-1 μm. The PCR products were ligated into a T/A cloning vector (Invitrogen, Carlsbad, CA). Because some of the degenerate oligos were expected to amplify the known MM cDNA, we excluded the clones containing these inserts by hybridization with ³²P-labeled random-primed cDNA of the known MM gene. Nonhybridizing clones were also screened by PCR to determine their size, and selected clones were then sequenced using dye termination. Using this approach, we identified a clone that contained the correct sequence of MMc and MMe flanked by dibasic cleavage sites as well as other likely peptide sequences. The insert from this clone was primed with random hexamers, labeled with ³²P, and used to screen the Aplysia ganglionic cDNA libraries to identify clones encoding the full-length cDNA. At least three independent library clones were sequenced to obtain a full-length consensus sequence of the MMG2 precursor. The uni-Zap Aplysia ganglia cDNA libraries were a kind gift from Dr. G. Nagle (Marine Biomedical Institute, Galveston, TX), and the lamda Zap random-primed cDNA library was a kind gift from Dr. W. Sossin (McGill University, Montreal, Quebec, Canada).

Northern blot analyses. RNA was isolated using the acid-phenol method (Chomczynski and Sacchi, 1987). Northern blot analysis was performed on the RNA using formaldehyde/3-morpholinepropanesulfonic acid denaturing agarose gels (1.5%) and downward transferred with 20× saline-sodium phosphate-EDTA (SSPE) to positively charged nylon membranes (Biodyne B; Invitrogen, Rockville MD). The RNA was immobilized with UV (Stratalinker; Stratagene) and visualized by staining with 0.02% methylene blue in 0.3 m Na-acetate, pH 5.5. After destaining with 1% SDS, 1 mm EDTA, and 50 mm Na₃PO₄, pH 7.2, the blot was prehybridized for 1 h at 50°C using 50% formamide, 10% dextran sulfate, 7% SDS, 10 mm EDTA, 50 μg/ml salmon sperm DNA, and 250 mm Na₃PO₄, pH 7.2. Heat-denatured, random-primed (New England Biolabs, Beverly, MA), ³²P-dCTP-labeled cDNA probe (either MMG2 or MM) was added, and hybridization was continued overnight at 50°C. Blots were washed twice for 15 min at room temperature with 2× SSPE and 0.1% SDS, washed for 60 min at 50°C with 0.1× SSPE and 0.1% SDS, and exposed to film. Autoradiographs and methylene blue-stained blots were scanned into Photoshop (Adobe Systems, San Jose, CA) and compiled into the final figures.

In situ hybridization. In situ hybridization (ISH) was performed as described previously (Vilim et al., 2001). Aplysia ganglia were digested with 1% protease type IX (Sigma-Aldrich, St. Louis, MO) for 1 h at 30°C to loosen and facilitate the removal of connective tissue. After digestion, the ganglia were washed with artificial seawater (ASW; in mm: 460 NaCl, 10 KCl, 55 MgCl₂, 11 CaCl₂, and 10 HEPES, pH 7.6) and fixed overnight at 4°C with 4% paraformaldehyde (Electron Microscopy Sciences, Fort Washington, PA) in PBS. The ganglia were washed with several changes of PBS and 0.1% Tween 20 (PBT) and desheathed. The ganglia were dehydrated in an ascending methanol series and stored in 100% methanol overnight at -20°C. After rehydration in a descending methanol series, the ganglia were washed with PBS and 0.3% Triton and refixed for 20 min with 4% paraformaldehyde and PBS. The ganglia were then washed with PBS, and the reactive groups were neutralized by washing first with 2 mg/ml glycine in PBS, then with 0.1 m triethanolamine-HCl (TEA-HCl), pH 8.0, and finally with 1 ml of TEA-HCl containing 2.5 μl of acetic anhydride. After several washes with PBT, the ganglia were prehybridized in Hyb-buffer (50% formamide, 5 mm EDTA, 5× SSC, 1× Dernhardt's solution, 0.1% Tween 20, and 0.5 mg/ml yeast tRNA) overnight at room temperature and for 6-8 h at 50°C with fresh Hyb-buffer.

Digoxigenin-labeled cRNA was synthesized with T7 RNA polymerase using the MMG2 cDNA clone as a template and used as a probe at 2 μg/ml in Hyb-buffer. The ganglia were hybridized overnight at 50°C and washed at 50°C (30 min each in 50% formamide, 5× SSC, and 1% SDS; in 50% formamide, 2× SSC, and 1% SDS; in 50% formamide, 2× SSC, and 1% SDS; and in 0.2× SSC) to remove the unhybridized probe. The ganglia were washed three times for 10 min with PBT containing 0.2% BSA and blocked for 2 h with PBT containing 0.2% BSA and 10% normal goat serum (NGS). The ganglia were then incubated overnight at 4°C with a 1:200 dilution of alkaline phosphatase-conjugated antidigoxigenin antibody in PBT containing 0.2% BSA and 1% NGS. After washes with PBT to remove unbound antibody, the ganglia were washed with detection buffer (100 mm NaCl, 50 mm MgCl₂, 0.1% Tween 20, 1 mm levamisol, and 100 mm TrisHCl, pH 9.5) and developed with 4.5 μl of nitroblue tetrazolium and 3.5 μl of 5-bromo-4-chloro-3-indolyl phosphate in 1 ml of detection buffer (Roche Molecular Biochemicals, Indianapolis, IN). The staining reaction was monitored visually and stopped by washing with PBT when the level of staining was adequate. The stained ganglia were observed and photographed using a Nikon microscope (Morrell Instruments, Melville, NY) with epi-illumination against a white background. Photographs were taken with a Nikon CoolPix 990 digital camera, imported into Photoshop, and compiled into figures.

Antibodies. Antibodies to several of the peptides that were predicted from the MMG2 precursor were produced in rats as described previously (Fujisawa et al., 1999; Furukawa et al., 2001; Li et al., 2001; Sweedler et al., 2002). Briefly, the antigen was prepared by coupling synthetic peptide (SynPep, Dublin, CA) to BSA (catalog #A0281; Sigma-Aldrich) using 1-ethyl-3-(dimethylaminopropyl)carbodiimide (EDC; catalog #E7750; Sigma-Aldrich). The coupling was performed in 1 ml of 50 mm NaH₂PO₄, pH 7.2, containing 10 mg of BSA, 1 mg of peptide, and 25 mg of EDC. The mixture was allowed to react overnight at 4°C, and the coupled antigen was purified from the reaction using a Microcon-30 spinning at 13,800 × g for 30 min at 4°C. After washing the retentate four times with 0.4 ml of 50 mm NaH₂PO₄, pH 7.2, it was resuspended in 0.5 ml of the same buffer and transferred to a new tube.

For each antigen, two male Sprague Dawley rats (250-300 g; Taconic, Germantown, NY) were immunized by intraperitoneal injection with either 12.5 μl (∼250 μg; rat 1) or 25 μl (∼500 μg; rat 2) of antigen in an emulsion of 0.3 ml of PBS and 0.3 ml of Freund's complete adjuvant. At days 21 and 42, the rats were boosted by intraperitoneal injection with either 6.25 μl (∼125 μg; rat 1) or 12.5 μl (∼250 μg; rat 2) of antigen in an emulsion of 0.3 ml of PBS and 0.3 ml of Freund's incomplete adjuvant. The animals were killed by decapitation at 49 d, and the blood was harvested and processed for serum. Sera were aliquoted, frozen and lyophilized, or stored at 4°C with EDTA (25 mm final concentration) and thimerosal (0.1% final concentration) added as stabilizers. Specificity of immunostaining was confirmed by preincubation of the primary antibody with 10^-4 m of the corresponding synthetic peptide, which abolished the staining.

Immunocytochemistry. Immunocytochemistry was performed as described previously (Vilim et al., 1996; Fujisawa et al., 1999; Furukawa et al., 2001). Tissues were fixed in freshly prepared fixative (4% paraformaldehyde, 0.2% picric acid, 25% sucrose, and 0.1 m NaH₂PO₄, pH 7.6) for either 3 h at room temperature or overnight at 4°C. After washes with PBS to remove the fixative, ganglia from large animals were desheathed to expose the neurons. Ganglia from small animals (10-15 g) were processed without desheathing. All subsequent incubations were done at room temperature. Tissue was permeabilized and blocked by overnight incubation in blocking buffer (BB; 10% normal donkey serum, 2% Triton X-100, 1% BSA, 154 mm NaCl, 50 mm EDTA, 0.01% thimerosal, and 10 mm Na₂HPO₄, pH 7.4). The primary antibody was diluted 1:250 in BB and incubated with the tissue for 4-7 d. The tissue was then washed twice per day for 2-3 d with washing buffer (WB; 2% Triton X-100, 1% BSA, 154 mm NaCl, 50 mm EDTA, 0.01% thimerosal, and 10 mm Na₂HPO₄, pH 7.4). After the washes, the tissue was incubated with a 1:500 dilution of secondary antibody (lissamine rhodamine donkey anti-rat; Jackson ImmunoResearch, West Grove, PA) for 2-3 d. The tissue was then washed twice with WB for 1 d and four times with storage buffer (1% BSA, 154 mm NaCl, 50 mm EDTA, 0.01% thimerosal, and 10 mm Na₂HPO₄, pH 7.4) for 1 d.

Double-labeling experiments were performed as described previously (Vilim et al., 1996, 2000). In these experiments, in addition to the anti-MMG2 antibodies, we have used either anti-MMa antibodies (Miller et al., 1991; Vilim et al., 2000) or anti-buccalin antibodies (Miller et al., 1992). Both the anti-MMa and anti-buccalin antibodies were prepared in rabbits. Thus, in addition to the rhodamine donkey anti-rat antibody, we stained the tissues with fluoroscein donkey anti-rabbit secondary antibody (Jackson ImmunoResearch).

The tissues were then stored at 4°C and viewed and photographed on a Nikon microscope equipped with epifluorescence. The black and white images were collected using filters with characteristics appropriate for distinguishing between the fluorophors. The negatives of the images collected with the rhodamine and fluoroscein filters are shown in the figures to enhance the visibility of the stained neuronal processes. The same field of view is shown for the images collected with the rhodamine and fluoroscein filters.

Backfills. Backfills, as described previously (Furukawa et al., 2001; Sweedler et al., 2002), did not produce sufficient intensity of staining and had to be modified to an older version of the method. Cerebral ganglia with attached pleural and pedal ganglia from 10 g animals were pinned in Sylgard-lined dishes. The cerebro-buccal connective (CBC) was isolated from the ganglia in a silicone grease well and a subchamber constructed from a 1 cm cylinder cut from a 5 cc syringe. The subchamber was then washed with 0.5 ml of distilled water to remove salt, the backfill solution (20 μl of saturated solution of biocytin in water) was applied to the CBC, and the preparation was incubated for 24 h at 15°C. The incubation medium for the ganglion was ASW containing 1% BSA (catalog #A9647; Sigma-Aldrich), 0.1% amicase (catalog #A2427; Sigma-Aldrich), and 0.03% dextrose. The incubation medium also contained 0.1% collagenase, which loosened the sheath during the incubation. The ganglia were washed several times with ASW before fixation and processed for immunostaining, as above except that streptavidin-fluorescein was also included with the secondary antibody to detect the backfilled biocytin.

Mass spectrometry. Based on the mRNA expression results, individual neurons from pedal ganglia were isolated and prepared as described previously (Li et al., 2000). Briefly, the pedal ganglion was immersed in regular 2,S-dihydroxybenzoic acid (DHB) matrix solution (10 mg/ml deionized water). Isolated neurons were transferred onto a MALDI sample plate containing 0.5 μl of either regular DHB matrix solution or concentrated DHB (50 mg/ml) in acetone:water (4:1). The neurons on the plate were smashed before drying at ambient temperature. Samples were either frozen or analyzed immediately. MALDI mass spectra were obtained using a Voyager DE STR mass spectrometer (PE Biosystems, Framingham, MA). A pulsed nitrogen laser (337 nm) was used as the desorption/ionization source. A custom-designed pinhole was used to reduce the laser spot size to ∼25 μm diameter on the sample plate, and positive-ion mass spectra were acquired using reflectron mode. A 20 kV accelerating voltage, 71.8% grid voltage, 0.03% guide wire voltage, and a delay time of 200 ns were used. The spectra were smoothed by averaging 100-256 scans. For MALDI post-source decay (PSD) analysis, 20 kV total acceleration voltage, 75% grid voltage, 0.03% guide wire voltage, and 75 ns delay time was used. Using a timed ion selector, different precursor ions can be selected and subjected to fragmentation. Under these experimental conditions, the mass accuracy of the precursor ion was within 30 ppm, and the average error on the mass assignment of the PSD ions was better than 0.3 U. Spectra were obtained by accumulating data from 100-256 laser shots. To obtain complete PSD spectra, a series of reflectron spectral segments were acquired, each optimized to focus fragment ions within different m/z ranges (Spengler, 1997). Each segment was stitched together using the Biospectrometry Workstation software to generate a composite PSD spectrum. For MALDI in-source decay (ISD) analysis (Brown and Lennon, 1995) of liquid chromatography (LC) fractions, 110 Aplysia buccal ganglia were homogenized, extracted for peptides, and separated using microbore HPLC. Individual LC fractions were screened by MALDI; 0.5 μl of each LC fraction was deposited onto a MALDI sample plate, followed by the same volume of 2,5-dihydroxybenzoic acid matrix solution. Thus, >95% of each fraction was available for additional assays. ISD mass spectra were acquired in the linear mode, the acceleration voltage was 20 kV, and laser power was applied well above the threshold value to facilitate fragmentation. The delay time used was 300-400 ns. Mass spectra were averaged over 100 laser shots and smoothed (19 points, Savitsky-Golay).

Electrophysiology. Animals weighing 150-250 g were anesthetized by injection of isotonic MgCl₂ (337 mm). Cerebral and buccal ganglia were desheathed to expose the cells of interest. The ganglia were then pinned out in a Sylgard-lined recording dish that contained ∼1.5 ml of ASW (in mm: 460 NaCl, 10 KCl, 55 MgCl₂, 11 CaCl₂, and 10 HEPES, pH 7.6). The preparation was perfused continuously with ASW at 0.3 ml/min and cooled to 14-17°C.

Intracellular recordings were made with glass microelectrodes filled with recording solution (2 m K-acetate and 200 mm KCl) and beveled to 8-15 MΩ. In some cases, the tip of the microelectrode was filled with 3% carboxyfluoroscein. The dye was injected into the neurons by passing negative current for 15 min. These dye-injected preparations were then stained with the anti-MMG2-DP antibodies.

To monitor feeding motor programs, we used a combination of intracellular and extracellular recording. Intracellular voltage recordings were amplified with either an Axoclamp 2B amplifier (Molecular Devices, Union City, CA) or Getting 5A amplifiers. Extracellular recordings and stimulation were performed using suction electrodes manufactured from polyethylene tubing connected to a differential AC amplifier (model 1700; A-M Systems, Carlsborg, WA). All signals were acquired simultaneously at 5 kHz with a Digidata 1322A data-acquisition system (Molecular Devices). Signals were then analyzed using Clampfit 9 (Molecular Devices). Axum 5 (Mathsoft, Cambridge, MA) and SigmaPlot 8 (Systat Software, Point Richmond, CA) were used to generate figures.

Identification and characterization of feeding motor programs. In Aplysia, each cycle of the feeding motor program begins with the radula protraction phase, followed by the radula retraction phase (Hurwitz and Susswein, 1996; Hurwitz et al., 1996, 1997; Jing and Weiss, 2001, 2002; Morgan et al., 2002; Koh et al., 2003; Proekt and Weiss, 2003; Proekt et al., 2004). The radula protraction phase was defined to be coincident with the burst of electrical activity in the I2 nerve (Hurwitz and Susswein, 1996; Hurwitz et al., 1996, 1997). The radula retraction phase was defined as lasting from the end of the I2 nerve burst to the end of the buccal nerve 2 (Bn2) burst (Morton and Chiel, 1993a,b). Ingestive and egestive motor programs are distinguished by the way in which radula closure activity is coupled to the protraction-retraction sequence. In ingestive programs, radula closure occurs predominantly in the retraction phase, whereas in egestive programs, it occurs predominantly during the protraction phase (Morton and Chiel, 1993a,b; Jing and Weiss, 2001, 2002; Morgan et al., 2002; Proekt et al., 2004). We monitored radula closure by recording firing of the radula closer motor neuron B8 (Morton and Chiel, 1993a,b). We also recorded firing of neurons B4/B5.

We used two previously characterized input stimuli to elicit the motor programs: the command-like cerebro-buccal interneuron 2 (CBI-2) (Rosen et al., 1991; Morgan et al., 2000, 2002; Jing and Weiss, 2001) and the esophageal nerve (Chiel et al., 1988; Horn et al., 2004; Proekt et al., 2004). CBI-2 was stimulated to fire action potentials at 8-12 Hz for the duration of the protraction phase so as to elicit a single motor program every 90 s. The esophageal nerve was stimulated with 3 ms current pulses delivered at 6-7 Hz. A 5 s burst of these pulses was delivered every minute. The amplitude of the pulses was adjusted such that each burst elicited exactly one motor program.

ARC neuromuscular preparation. We used the standard preparation for recording motor neuron-elicited contractions of the ARC muscle (Weiss et al., 1979; Orekhova et al., 2003). Briefly, the preparation consisted of the buccal ganglia, the ARC muscles, and the connecting buccal nerve 3, through which the ARC is innervated by the motor neurons B15 and B16. One of the ARC muscles was pinned out in a separate chamber and connected to an isotonic transducer (model 60-3000; Harvard Apparatus, Holliston MA) to measure the length of the muscle contracting against a light counterbalancing load. Motor neuron B15 was stimulated to fire 1.5 or 2 s bursts of spikes at a frequency necessary to produce contractions of moderate amplitude every 20 s. Peptides were added to the muscle chamber only.

Results

Discussion

Here, we have described a new gene, MMG2, that encodes a set of neuropeptides that can modulate both the CPG and the neuromuscular system involved in the generation of feeding behaviors in Aplysia. This gene encodes MMc and MMe, two neuropeptides that were originally purified from the ARC muscle, and four neuropeptides that were unknown previously (MMG2-DPa, MMG2-DPb, MMG2-DPd, and MMG2-DPf). MALDI analysis of the neurons that express MMG2 has revealed peaks that correspond to the amidated forms of both MMc and MMe as well as MMG2-DPa-f, arguing that these peptides are actually cleaved from the precursor and undergo post-translational modification typical of bona fide neuropeptides. Cloning and sequencing of the MMG2 precursor has allowed us to match the MMG2-DPs to unassigned MALDI peaks detected previously in the analysis of the cerebro-buccal and pedal-abdominal connectives (Li et al., 1998; Sweedler et al., 2002). The presence of the MMG2-DPs in the connectives suggests that they are actively transported to different parts of the nervous system.

Indeed, using the newly generated MMG2-DP-specific antibodies, we have shown that neuronal processes containing these peptides are found in the neuropil of all of the major ganglia and in all of the buccal nerves. In contrast to the very broad distribution of the MMG2-DPs, ISH has suggested that their source is limited to a very small set of neurons in the pedal and cerebral ganglion. Because the neurons in the cerebral ganglion were not located in the areas of the ganglion that contain the neurons that project to the feeding circuitry, here we have focused on the MMG2-DP-staining neurons in the pedal ganglion. By a combination of immunohistochemical and electrophysiological techniques, we have identified some of these neurons as P-BPNs. It has been shown previously by Robie et al. (2003) that these neurons project not only to the buccal ganglion but through most of the buccal nerves out to the buccal musculature. Our data, combined with those of Robie et al. (2003), make it very likely that these neurons are the source of the MMG2-DP-, MMc-, and MMe-containing innervation both centrally and peripherally in the feeding system, including in the ARC muscle from which MMc and MMe were originally purified.

Several sources of neuromodulation are already known to converge on the ARC muscle to affect various parameters of the ARC muscle contraction. In some cases, different peptides modulate the very same parameters. For instance, both the MMs and SCPs increase the amplitude and the relaxation rate of ARC contractions (Cropper et al., 1987a,b, 1991; Brezina et al., 1996). Multiple converging modulatory inputs with overlapping effects may seem redundant. Yet such convergence exists in many other systems (Nusbaum and Beenhakker, 2002). In the ARC system, we have demonstrated previously that the redundancy is only superficial. At the quantitative level, the effects of the modulators are not identical but only partially overlapping and furthermore have different dynamics. Using a combination of experimental and theoretical approaches, we have shown that the convergence of the modulators with these properties allows the ARC muscle contraction waveform to sample regions of the parameter space that would not be accessible with any single modulator (Brezina et al., 1996, 2005). Thus the combinatorial control by convergent modulation allows for greater flexibility of response.

At first glance, the effects of the peptides derived from MMG2 on the ARC muscle may also seem redundant. MMG2-DPb induces a baseline contracture, whereas previously we found (Brezina et al., 1995) that MMc and MMe have the opposite effect: they relax the contraction baseline. In addition, however, MMc and MMe also increase the amplitude and relaxation rate of the contractions, whereas MMG2-DPb does not. Thus, the overall effects of the peptides are not redundant. Furthermore, the two effects on the baseline contraction have qualitatively different dynamics. Although the relaxation of the baseline induced by MMc and MMe does not require activity of the motor neurons or muscle, the MMG2-DPb-induced contracture does. This type of activity dependence of modulator action has not been observed previously in the ARC muscle. The activity dependence allows these peptides derived from the same gene and likely to be coreleased to act preferentially in different contexts to endow the ARC muscle contractions with greater flexibility.

The modulation of the ARC muscle by neuropeptides released from the ARC motor neurons, such as the MMs and the small cardioactive peptides (SCPs), is a classic example of what has been termed intrinsic modulation (Cropper et al., 1987a; Katz et al., 1994). By its nature, intrinsic modulation produces local changes, such as changes in the contraction of a single muscle. In contrast, extrinsic modulation, stemming from higher-level neurons that are themselves not part of the circuit they modulate, can act globally at multiple levels of organization of the nervous system.

Here, we have identified the neurons that contain the MMG2-DPs as well as MMc and MMe as P-BPNs. These neurons are not rhythmically active during feeding motor programs (Robie et al., 2003) and are thus not an integral part of the feeding circuitry. Yet the P-BPNs send projections to the feeding system, including the buccal ganglion that contains the feeding CPG and through the buccal nerves to the feeding musculature. They are thus likely to modulate the feeding system both centrally and peripherally. Indeed, in addition of the peripheral effects, here we have demonstrated that the MMG2-DPs affect a number of different parameters of the feeding motor programs. The MMG2-DPs lengthen the protraction phase and switch egestive motor programs to ingestive ones. We have identified one target of MMG2-DP action, neuron B4/5, which may contribute to the switch in the character of the motor programs. Together with the effects on the musculature, these observations suggest that the MMG2-DP-containing P-BPNs are extrinsic modulatory neurons that act at multiple levels within the feeding system.

Previous work has identified another extrinsic modulatory influence on the feeding circuitry, the serotonergic MCC neurons (Weiss et al., 1975, 1978). The modulation by serotonin and the MCCs converges with that of the peptides derived from MMG2 on several levels. Serotonin and the MCCs modulate ARC muscle contractions, increasing their amplitude and relaxation rate (Weiss et al., 1978; Brezina et al., 1996). These effects therefore resemble those of MMc and MMe, which have also been identified here as extrinsic modulators. MCC stimulation and serotonin leads to several changes in the characteristics of the feeding motor programs, including a shortening of the protraction phase (Morgan et al., 2000). In this respect, the effect of the MCCs and serotonin is opposite to that of the MMG2-DPs. Like the MMG2-DPs, the MCCs also modulate the neurons B4/5 (Weiss and Kupfermann, 1976). The feeding system may be simultaneously modulated by both extrinsic systems, because both the MCC and the P-BPNS are activated by C-PR, a neuron important for the appetitive phase of feeding behavior (Teyke et al., 1990; Hurwitz et al., 1999; Robie et al., 2003).

Thus, combined with the previous work, the discovery of the MMG2-DPs and the localization of these neuropeptides to the P-BPNs demonstrate that multiple extrinsic modulatory inputs converge both at the level of the feeding CPG and in the periphery. As with the intrinsic modulators of the ARC muscle, some of the effects are similar, but some appear superficially antagonistic. Thus, the analytical methods that we have applied in the ARC muscle system may be applicable more generally. Combined with the previous data, our results suggest that the principles of combinatorial control through convergent modulation observed in the ARC muscle system may be recapitulated at multiple levels of organization of the nervous system.

Convergence of multiple modulatory systems is a ubiquitous neuroanatomical motif observed both in invertebrate and vertebrate nervous systems. For instance, serotonergic fibers originating from the raphe nuclei and noradrenergic fibers originating from the locus ceruleus converge to modulate motor neurons in the spinal cord (Holstege and Kuypers, 1987). Thus, extending the understanding of the consequences of convergence of multiple modulatory systems from the peripheral structures into the CNS in experimentally tractable preparations is likely to provide insight into the general principles of combinatorial control of behavioral plasticity.