Differential activation of an identified motor neuron and neuromodulation provide Aplysia's retractor muscle an additional function

Abstract

To survive, animals must use the same peripheral structures to perform a variety of tasks. How does a nervous system employ one muscle to perform multiple functions? We addressed this question through work on the I3 jaw muscle of the marine mollusk Aplysia californica's feeding system. This muscle mediates retraction of Aplysia's food grasper in multiple feeding responses and is innervated by a pool of identified neurons that activate different muscle regions. One I3 motor neuron, B38, is active in the protraction phase, rather than the retraction phase, suggesting the muscle has an additional function. We used intracellular, extracellular, and muscle force recordings in several in vitro preparations as well as recordings of nerve and muscle activity from intact, behaving animals to characterize B38's activation of the muscle and its activity in different behavior types. We show that B38 specifically activates the anterior region of I3 and is specifically recruited during one behavior, swallowing. The function of this protraction-phase jaw muscle contraction is to hold food; thus the I3 muscle has an additional function beyond mediating retraction. We additionally show that B38's typical activity during in vivo swallowing is insufficient to generate force in an unmodulated muscle and that intrinsic and extrinsic modulation shift the force-frequency relationship to allow contraction. Using methods that traverse levels from individual neuron to muscle to intact animal, we show how regional muscle activation, differential motor neuron recruitment, and neuromodulation are key components in Aplysia's generation of multifunctionality.

multifunctionality, which has been defined as the ability of the same peripheral structure to produce multiple, distinct behaviors (Neustadter et al. 2007), is critical for animals to survive in ever-changing environments. How do nervous systems generate different behaviors using the same structures? This has been a topic of interest to researchers for years and has been addressed through work in many systems. It has been shown in studies of locomotion that muscles can perform a variety of functions, acting, for example, as struts, springs, brakes, and motors (Dickinson et al. 2000). In the cockroach, two leg extensor muscles that are closely related anatomically serve different functions: one acts as a motor and the other as a brake (Ahn and Full 2002).

Many studies have examined changes in the activation patterns of muscles used by animals in different behaviors. In toads, the hindlimbs are used for propulsion in both jumping and swimming. One study showed that the activation patterns of certain hindlimb muscles changed between the two behaviors, and the authors concluded that these muscles serve different functions in jumping vs. swimming (Gillis and Biewener 2000). As humans switch from forward to backward pedaling of a bicycle, the phase of activation of certain muscles changes, whereas the phase of activation does not change for several other muscles. These changes are related to whether or not the muscles have different functions in the two pedaling conditions (Ting et al. 1999). Such differences in activation occur not just at the level of the whole muscle but also at the level of muscle regions. As the helmeted guinea fowl transitions from walking to running, the proximal region of the medial gastrocnemius increases its workload, yet the distal region displays no change (Higham and Biewener 2008). In cat biceps femoris, only the anterior region is active in slow walking; as locomotion speed increases, the middle and then the posterior regions become active (Chanaud et al. 1991).

Although spatial localization of motor units has been observed (Vieira et al. 2011), it is difficult to study regional changes in muscle activation at the level of single, identified neurons in vertebrates. The marine mollusk Aplysia californica provides a tractable system for such studies. Because of its accessible nervous system containing large, identifiable neurons, Aplysia has been used as a model organism for the study of many phenomena, such as neuromodulation (Cropper et al. 1987; Rosen et al. 2000; Weiss et al. 1978), the neuromuscular transform (Brezina et al. 2000), operant conditioning (Nargeot et al. 1997, 1999a, 1999b), and the role of nitric oxide in behavior regulation (Susswein and Chiel 2012), as well as the generation of multifunctionality. Aplysia's multifunctional feeding system (buccal mass; see Fig. 1) produces several distinct behaviors by utilizing the same musculature in different ways. These behaviors include ingestion (bites and swallows) and egestion (rejections) (Kupfermann 1974). Bites are attempts to grasp food, during which the animal's food grasper (radula) strongly protracts while open and then closes before retracting. Swallows occur after food has been grasped; grasper protraction is weaker, and retraction is stronger, as the animal transports food to the esophagus (Neustadter et al. 2002, 2007). In rejections, by contrast, the grasper protracts while closed and then opens to expel an object from the buccal cavity. This shift between ingestion and egestion has been tied to a change in the activity of identified neurons: the phase of activation of the B8 motor neurons, which cause closure of the grasper, changes between ingestive and egestive programs (Morton and Chiel 1993b).

Fig. 1.

Schematic cut-through drawing of a side view of the Aplysia buccal mass with key structures labeled. Below the buccal mass, the anterior-posterior axis is indicated; note that the anterior direction is toward the jaws. The radula is the pincers-like food grasper, whose 2 halves close together to grasp food. During behaviors, the I2 muscle, a thin, sheetlike muscle that wraps around the back of the buccal mass, contracts to push the grasper toward the jaws (protraction). During protraction, the grasper moves into the lumen of I3, the ring-shaped muscle at the front of the buccal mass that is the muscle of focus in this study. During the second phase of behaviors, the I3 muscle is activated to push the grasper back toward the esophagus (retraction). [Modified from schematic drawn by Dr. Richard F. Drushel for Fig. 1A in Sutton et al. (2004) with kind permission from Springer Science and Business Media.]

Another part of the feeding system for which changes in activation may be important to the generation of multifunctionality is the I3 jaw muscle. During all feeding behaviors, the food grasper first protracts into the open I3 lumen. In the second phase of the behavior, the grasper retracts, and contraction of I3 contributes to the retraction (Morton and Chiel 1993a). Thus I3 is known as a retractor muscle. Church and Lloyd (1994) described six identified motor neurons activating I3 with different, overlapping innervation regions. Through the deployment of different motor neurons, I3 can therefore be activated either as a whole unit or as smaller subunits. Consistent with I3's known role as a retractor, the neurons activating I3 were classified as retraction-group neurons, except one: B38, a protraction-group neuron causing anterior jaw closure. Why is a neuron for a retractor muscle active in the opposite phase of motor programs, when it would presumably block grasper protraction?

B38 has been extensively studied in experiments elucidating the role of modulation in muscle activation. Fox and Lloyd (1997, 1998, 2001) showed that both intrinsic and extrinsic modulation (Katz 1995), via release of small cardioactive peptide (SCP) from B38 and serotonin from the metacerebral cell, greatly increase B38-elicited muscle contractions. However, Fox and Lloyd lacked data showing B38's in vivo activity. What is the behavioral significance of the neuromodulation?

We propose a two-part hypothesis: 1) Grasper retraction is not the I3 muscle's sole function. During protraction of swallowing patterns, I3's anterior region, via activation by B38, has an entirely different function: closing to hold food. 2) Modulation is necessary for B38's in vivo activity patterns to generate functionally effective force.

We report that B38 specifically activates I3's anterior and is specifically active during swallowing. Intense B38 bursting is associated with the jaws closing on food. Additionally, we show that B38's typical in vivo activity is insufficient to generate force without modulation. Both intrinsic and extrinsic modulation dramatically shift this force-frequency relationship. Thus regional muscle activation, differential activation of an identified motor neuron, and neuromodulation enable one muscle to produce two functions.

MATERIALS AND METHODS

Animals.

Aplysia californica (size: 200–350 g for in vitro experiments, 350–450 g for in vivo experiments) were obtained from Marinus (Garden Grove, CA) and kept in large aquarium tanks filled with circulating artificial sea water (Instant Ocean; Aquarium Systems, Mentor, OH) at ∼16°C. Animals were fed every other day with large strips of dried seaweed (laver). Before experiments, animals were presented with seaweed, and animals that displayed strong bites at 3- to 5-s intervals were selected for use.

Electrodes.

Hook electrodes consisted of two wrapped, enamel-coated 0.001-in.-diameter stainless steel wires (California Fine Wire, Grover City, CA) coated in household silicone glue (GE). Before each experiment, the insulation was removed from the ends of the wires. One wire was attached to the target nerve or muscle with the use of Quick Gel Super Glue (Henkel, Avon, OH) to insulate the wire from the saline and hold it in place; the other wire served as a reference. Signals from these and other extracellular electrodes were amplified using an AC-coupled differential amplifier (model 1700; A-M Systems, Everett, WA). For in vivo recordings, a 1,000-Hz low-pass filter was used; a 500-Hz low-pass filter was used for in vitro recordings. Both in vivo and in vitro, a 300-Hz high-pass filter was used for nerve recordings, and a 10-Hz high-pass filter was used for muscle recordings.

Suction electrodes were made from polyethylene tubing (catalog no. 427421, inner diameter 0.86 mm, outer diameter 1.27 mm; Becton Dickinson, Sparks, MD) pulled over a Bunsen burner. Electrodes were filled with Aplysia saline (460 mM NaCl, 10 mM KCl, 22 mM MgCl₂, 33 mM MgSO₄, 10 mM CaCl₂, 10 mM glucose, and 10 mM MOPS, pH 7.5).

Extracellular glass electrodes were made from single-barreled capillary glass (catalog no. 6150; A-M Systems) pulled on a Flaming-Brown micropipette puller (model P-80/PC; Sutter Instruments, Novato, CA). The electrodes' inner diameters were about 40 μm, and their resistances were about 0.1 MΩ. Electrodes were filled with Aplysia saline. A stimulus isolator (model A-360; WPI, Sarasota, FL) was used for stimulation currents.

Intracellular glass electrodes were also made from single-barreled capillary glass (catalog no. 6150; A-M Systems) pulled on the Flaming-Brown micropipette puller. The electrodes' resistances were 3–6 MΩ, and they were filled with 3 M potassium acetate. The bridge was balanced for recording and stimulation. A DC-coupled amplifier (model 1600; A-M Systems) was used to amplify intracellular signals.

Experimental preparations.

Four types of experimental preparation were used to study B38's role in activating the I3 muscle during Aplysia feeding. An in vivo multielectrode recording preparation (Cullins and Chiel 2010) was used to relate electroneurogram (ENG) and electromyogram (EMG) activity to the behavioral responses of whole animals and to quantify B38's activity during different types of feeding behavior. An in vitro semi-intact suspended buccal mass preparation (McManus et al. 2012) provided the ability to record directly from B38 during feeding-like movements. In this preparation, B38's activity could be related to the muscle movements in feeding-like responses, and the nerve activity produced by B38 could be characterized to aid analysis of recordings from intact animals. Two in vitro reduced preparations, EMG and muscle force recordings, were used to characterize B38's activation of the muscle, in terms of both the regional specificity of muscle activation and the forces elicited by specific B38 firing frequencies. These four preparations are described in the following sections.

Recording and analysis of behavior from intact animals.

Nerve and muscle recordings and corresponding video from intact, behaving animals were obtained as described by Cullins and Chiel (2010). Briefly, animals were anesthetized with an injection of 30% body wt isotonic MgCl₂, and an incision was made in the head to expose the buccal mass. Hook electrodes were implanted on the I2 muscle, radular nerve (RN), and buccal nerves 2 and 3 (BN2 and BN3). Activity in the I2 muscle indicates the protraction phase of motor patterns (Hurwitz et al. 1996). Large-unit activity on the RN indicates the timing of grasper closure and can be used to distinguish egestive from ingestive patterns (Morton and Chiel 1993a, 1993b). Activity on BN2 represents the activity of I1/I3 motor neurons (Church and Lloyd 1991, 1994; Scott et al. 1991) and can be used as an indication of the retraction phase of motor patterns (Morton and Chiel 1993a, 1993b). Activity on BN3 represents the activity of important multiaction neurons B4/B5 (Warman and Chiel 1995) and motor neurons for several muscles, including I3 (Church and Lloyd 1991, 1994), and is also useful for characterizing motor patterns. Note that in the publication by Scott et al. (1991), the RN, BN2, and BN3 were referred to as nerves 1, 5, and 4, respectively. The incision was sutured shut (Ethicon, Somerville, NJ), and the animals were allowed to recover. Biting responses were induced by presenting an animal with pieces of dried seaweed but not allowing it to ingest the seaweed. Swallowing responses were induced by allowing the animal to grasp strips of seaweed (0.25, 0.5, or 1 cm wide) during a bite. Rejection responses were induced by allowing the animal to grasp a polyethylene tube during a bite; after several swallows, the animal began to reject the tube (Morton and Chiel 1993a; Ye et al. 2006b).

Behaviors were classified as bites, normal swallows, visible radula swallows, or rejections on the basis of video recordings of the animals. Behaviors in which movement was unclear or not visible were not placed into these categories and were not used for further analysis. In bites, the radula protracted while open, closed, and then retracted, with no seaweed in the mouth. In normal swallows, seaweed in the mouth had a net inward movement, and the radula did not protrude through the jaws at its peak protraction. Visible radula swallows were similarly characterized by net inward seaweed movement but had stronger protractions so that the radula became visible. Note that Ye et al. (2006a) also described swallows with smaller or larger protractions, termed type A and type B swallows, respectively. In the study by Ye et al., type A vs. type B swallows were distinguished by measuring the rotation of a polyethylene tube being swallowed; this criterion was not applicable to our study, in which the swallows analyzed were of seaweed strips, not tubing. Last, in rejections, polyethylene tubing, or in some cases seaweed, had a net outward movement.

Neuromuscular recordings were imported into Mathematica (Wolfram, Champaign, IL) for analysis. Briefly, units on the different nerves and I2 muscle were captured by using size windows based on the heights of action potentials. Instantaneous and average firing frequencies for different units were then calculated. A total of 150 bites, 130 normal swallows, 61 visible radula swallows, and 53 rejections from n = 6 animals were analyzed. To create plots of averaged instantaneous firing frequencies, a subset of these patterns was selected to best capture the typical activity of B38 during a series of patterns. Patterns with a duration from the end of the previous pattern's retraction phase to the onset of the current pattern's protraction phase of <0.2 s were excluded (if this duration was extremely short, this could lead to anomalies in the averaging of instantaneous firing frequencies). If this duration was >3 s, patterns were also excluded (a long gap between patterns would not be indicative of the activity within a series of patterns, and we wanted to capture the typical activity within a series because B38 bursting tends to begin at the end of one pattern's retraction phase and continue into the subsequent pattern's protraction phase). Furthermore, only patterns in which B38 fired at least once during this time range were included. These criteria resulted in 31 bites, 84 normal swallows, 40 visible radula swallows, and 20 rejections being analyzed for the plots of averaged instantaneous firing frequencies.

Suspended buccal mass.

In vitro motor programs with corresponding feeding-like movements and nerve, muscle, and extracellular neuron soma recordings were obtained using a preparation described by McManus et al. (2012). Animals were anesthetized with an injection of 50% body wt isotonic MgCl₂, and the buccal mass was removed with the buccal and cerebral ganglia still attached. Hook electrodes were attached to the same nerves and muscle as in the in vivo recordings, along with branch a of buccal nerve 2 (BN2-a; Warman and Chiel 1995), which can be stimulated to elicit motor programs (Nargeot et al. 1997; in that publication, the branch is referred to as branch 3; in the suspended buccal mass preparation, we found that stimulation of BN2-a consistently generated egestive programs). In most in vitro experiments, electrodes were placed on BN2 and BN3 on both sides of the buccal mass. The buccal mass was placed in a round 100 × 50-mm Pyrex dish containing a front chamber for the buccal mass, a middle Sylgard (Dow Corning, Midland, MI) platform onto which the buccal ganglia were pinned, and an isolated back chamber in which the cerebral ganglion was also pinned to Sylgard. Each chamber was filled with Aplysia saline. A notch in the Sylgard wall connected the middle chamber and back platform, and the cerebral buccal connectives ran through this notch, which was sealed with vacuum grease (Dow Corning). The sheaths of the buccal ganglia were thinned but not removed, allowing better visualization of neuron cell bodies and access to the neurons with extracellular glass electrodes. The buccal mass was then suspended with the use of a silk suture threaded through the tissue anterodorsal to the jaws and attached to balls of modeling clay placed on the side of the dish.

To locate a neuron in the buccal ganglion for extracellular recording (Lu et al. 2008), an extracellular glass electrode was positioned above the neuron's soma and gently pressed into the sheath, a stimulating current was applied, and the electrode's channel was quickly switched to recording mode. One-for-one spikes could then be observed on the channel of the soma electrode and on any nerve to which the neuron projected. By locating a neuron and then leaving the electrode in place while motor programs were elicited, a recording of the individual neuron's activity within the motor programs could be obtained. B38 was identified by its location in the ganglion (Church and Lloyd 1991), by its production of a large unit solely in the ipsilateral BN2, and by the timing of its activity in motor programs. Extracellular soma recordings were obtained from B38 in n = 15 suspended buccal mass experiments.

Rejection-like motor programs were obtained by stimulating BN2-a with a long train of 1-ms, 2-Hz pulses. Biting-like motor programs were elicited by applying the cholinergic agonist carbachol to the cerebral ganglion (Susswein et al. 1996). Swallowing-like motor programs were obtained by placing a strip of seaweed, 0.25–0.5 cm wide, in the food grasper during a bite. The BN2 activity produced by B38 was identifiable in an additional n > 50 suspended buccal mass experiments in which B38's activity was not directly recorded from the cell body.

EMG recordings.

Animals were anesthetized with an injection of 50% body wt isotonic MgCl₂, and the buccal mass was removed with the buccal and cerebral ganglia still attached. Hook electrodes were attached to both BN2s and to BN2-a, the latter to allow nerve-stimulation induction of motor patterns. All other buccal nerves were severed at their attachment points to the buccal mass. A dorsal cut was made in the buccal mass, from the jaws through the esophagus, so the I1/I3 muscle complex could be splayed out with the inside surface facing up. Note that the I1 muscle is a thin muscle covering I3 on the outer surface of the buccal mass, and the two form a single muscle complex (I1/I3), which is innervated by BN2 [Church and Lloyd 1991; Scott et al. 1991 (in which BN2 was referred to as nerve 5)]. Throughout the results of the present article, we refer specifically to muscle I3, because B38, as well as other neurons that generate strong contractions of the I1/I3 lumen (e.g., B3, B6, and B9), innervates I3 and not I1 (Church and Lloyd 1991). The radula-odontophore was removed, and the cerebral ganglion, buccal ganglia, and I1/I3 were pinned to Sylgard. The cerebral ganglion was isolated as in the suspended buccal mass preparation. The sheaths of the buccal ganglia were thinned as in the suspended buccal mass preparation. On the inside of the buccal mass, the anterior region of I3 is covered by cartilage and the posterior by pharyngeal tissue; cartilage and pharyngeal tissue were cut and peeled away from the recording sites. Glass extracellular electrodes were used to record EMG signals from I3's anterior and posterior regions. Suction electrodes made from polyethylene tubing were used to record electrical activity from I2, RN, and BN3 (thus the same recordings were obtained as during in vivo experiments). Glass extracellular electrodes were used to stimulate and record from the somata of buccal neurons as described for the suspended buccal mass. Extracellular recording and stimulation of B38 were performed in n = 7 EMG recording experiments.

Force recordings.

In characterizing skeletal muscle contraction, it is common to measure the twitch contraction that results from a single action potential in the muscle (e.g., Gallo et al. 2004). The all-or-none nature of this twitch contraction stems from the all-or-none nature of the muscle action potential (Gelfan 1933). In contrast, molluscan muscles have different properties that must be briefly explained so that the subsequent results will be clear. Cohen et al. (1978), in work on the ARC (or I5) muscle of Aplysia, showed that action potentials in the motor neurons for the muscle generate excitatory junction potentials (EJPs) in the muscle and that depolarization of the muscle due to the summation of these EJPs results in muscle contraction. Action potentials were not observed in the muscle itself. Motor neurons for the I3 muscle, such as B38, similarly produce EJPs that summate before producing muscle contraction (Church et al. 1993; Fox and Lloyd 1997). Because contractions of the I3 muscle are not associated with muscle action potentials, in our studies it was 1) not meaningful to measure a “twitch contraction” and 2) also possible to stimulate action potentials in a motor neuron at levels that were below threshold for eliciting muscle contraction.

To begin the preparation for muscle force recordings, animals were anesthetized and the buccal mass removed as described above. Hook electrodes were attached to both BN2s, BN3s, and BN2-a. All other buccal nerves were severed at their attachment points to the buccal mass so that the only muscle innervated was I1/I3 and all forces would be caused by I1/I3 contraction. [Two I3 neurons, B6 and B9, project through both BN2 and BN3 (Church and Lloyd, 1991), but we observed that when only BN3 was left attached, forces generated in I1/I3 were minimal.] The ventral surface of the buccal mass was glued to the glass bottom of a petri dish. The I1/I3 lumen was left intact so that the muscle would be in a similar configuration to its physiological state, with a similar length to its physiological length. Note that during behaviors, I1/I3 can be stretched by the radula protracting through it, which would increase I1/I3's length. However, during swallowing, the behavior during which we hypothesize B38 acts, the radula does not pass through the anterior region of I1/I3 (i.e., the region innervated by B38) while B38 is intensely active, so when B38 is intensely active, the muscle region of focus for our studies would not be substantially stretched beyond its resting length (Neustadter et al. 2002). The buccal ganglia and cerebral ganglion were pinned to Sylgard. In most of the force recording experiments, the ganglia were isolated in a separate chamber from the buccal mass, separated by a Sylgard wall. The ganglia were bathed in high-divalent cation Aplysia saline (270 mM NaCl, 6 mM KCl, 120 mM MgCl₂, 33 mM MgSO₄, 30 mM CaCl₂, 10 mM glucose, and 10 mM MOPS, pH 7.5) to suppress spontaneous activity. The sheaths of the buccal ganglia were either thinned or removed, for access to neurons using extracellular or intracellular electrodes, respectively. In certain experiments, the cerebral ganglion was also desheathed to expose the serotonergic metacerebral cells (MCCs), which are identifiable by their size and location (Weiss and Kupfermann 1976; Weiss et al. 1978), for intracellular recording and stimulation. Force transducers (Grass Technologies, West Warwick, RI) were attached laterally to the anterior and posterior regions of I1/I3 using silk sutures. After the force transducers were attached, the muscle was stretched a minimal amount to ensure there was no slack in the sutures. In some experiments, buccal neurons were stimulated and recorded using glass extracellular electrodes. B38 was stimulated and recorded extracellularly in n = 10 muscle force recording experiments. In other experiments, buccal neurons were stimulated and recorded using glass intracellular electrodes, which provided better control over the firing frequency of stimulated neurons. B38 was stimulated and recorded intracellularly in n = 16 muscle force recording experiments.

Additionally, in some experiments (n = 3) a small polyethylene tube was attached to a servomotor (Dual-Mode Lever System model 305B; Aurora Scientific, Aurora, ON, Canada) and positioned between the jaws of the buccal mass. The servomotor was programmed, using a protocol in AxoGraph X (AxoGraph Scientific, Sydney, Australia), to move the tube outward from the jaws. The servomotor was then used to record the force on the tube as B38 was stimulated.

RESULTS

B38 specifically activates I3's anterior.

Understanding B38's function requires characterizing the nature of the muscle contractions evoked by B38. It has been previously reported that B38 innervates the anterior region of the I3 muscle (Church and Lloyd 1991; Lotshaw and Lloyd 1990). We confirmed that B38 specifically activates the anterior region, and not the posterior region, of I3 and causes strong jaw closure. When B38 was stimulated extracellularly and EMG signals were recorded from I3, one-for-one spikes appeared in B38's soma and in the ipsilateral BN2, and one-for-one EJPs appeared in I3's anterior, but not posterior, region (Fig. 2A). In both the EMG and force recording preparations, stimulation of other I3 motor neurons did elicit activity in I3's posterior, so the lack of posterior activity due to B38 stimulation was not a result of poor recordings (data not shown). When B38 was stimulated and force measurements were made from I3's anterior and posterior, an anterior-specific force was generated (Fig. 2B). When B38 was stimulated extracellularly in a suspended buccal mass preparation, the anterior jaws closed on the side ipsilateral to the stimulated B38 (Fig. 2C), but no noticeable contraction was seen in the posterior or contralateral I3.

Fig. 2.

B38 stimulation causes anterior-specific I3 muscle activation. A: the soma of B38 was stimulated extracellularly (1st trace), and the soma channel was quickly switched to recording mode. One-for-one spikes appeared on the recordings of B38's soma (1st trace) and buccal nerves 2 (BN2; 2nd trace), and one-for-one excitatory junction potentials (EJPs) appeared in I3's anterior region (I3a; 3rd trace) but not in I3's posterior region (I3p; 4th trace). The 4 values on the vertical scale bar at bottom follow the same order as, and correspond to, the 4 traces shown (a similar convention is used for the scale bars in all subsequent figures). B: B38 was stimulated intracellularly (2-s, 15-Hz pulse train) with force transducers attached to the muscle. B38 spikes appeared on the recording of the ipsilateral BN2 (1st trace), and a large force was generated in the I3 anterior, while little force was measured in the posterior (2nd trace; ant., anterior; post., posterior). Force measurements were smoothed in postprocessing to remove noise. C: the B38 in the left buccal ganglion was stimulated extracellularly in the suspended buccal mass, and the anterior jaws closed on the left side of the buccal mass (white arrows point to anterior jaw closure). Frames shown occurred at intervals of one-third of a second, and the stimulation was applied 0.25 s before the first frame, causing B38 to burst for several seconds.

B38 activity increases during in vitro swallowing, and this activity can be identified on BN2.

Knowing B38's activity in the different motor programs, and being able to identify this activity without recording directly from B38's soma, would make further data collection in vitro and especially in vivo easier. To obtain these data, in the suspended buccal mass preparation, we located B38 with an extracellular electrode and then recorded its activity during motor programs. When biting-like programs were generated via application of carbachol to the cerebral ganglion, B38 firing was very weak, occurring at the postretraction phase (Fig. 3A). When we introduced seaweed into the mouth and it was grasped by the radula, the buccal mass switched into a swallowing-like mode, and B38 firing intensified greatly. The B38 activity was not only more intense but also continued into the subsequent protraction phase (Fig. 3B), consistent with the hypothesis that B38 closes the jaws during the protraction phase of swallowing. What is more important, we also observed that when B38 is active after retraction or during protraction, it consistently produces the largest unit on BN2 during this phase of the motor program. When B38 was located and recorded with the use of an extracellular soma electrode in the suspended buccal mass preparation, the BN2 spikes produced by B38 were larger than the largest non-B38 BN2 spikes that appeared during the protraction phases or inter-pattern intervals of biting, swallowing, or rejection responses in 100% (15/15) of experiments. On average, B38's BN2 spikes were 1.40 ± 0.16 (SD) times as large as the next largest non-retraction-phase BN2 spikes; the B38 spikes were significantly larger than the largest non-B38 spikes during this time (P < 0.05, paired t-test). The characteristic large unit produced by B38 is evident on both the ipsilateral and contralateral BN2 recordings in Fig. 3B. At the onset of the B38 burst, its activity may overlap with other large BN2 units, but during the main portion of the B38 burst, no other BN2 units of a comparable size are active. As a result of this finding, B38's protraction-associated activity can be identified from a BN2 recording alone, with no corresponding soma recording, both in vitro and in vivo.

Fig. 3.

B38 activity increases during in vitro swallowing, and the activity can be identified on BN2. Recordings are shown from the I2 muscle [1st trace; mediates protraction (Hurwitz et al. 1996)], radular nerve [RN, 2nd trace; mediates grasper closure (Morton and Chiel 1993a, 1993b)], contralateral and ipsilateral buccal nerves 2 [cBN2 and iBN2, 3rd and 4th traces; innervate the I3 muscle (Morton and Chiel 1993a, 1993b; Scott et al. 1991, in which BN2 was referred to as nerve 5)], and B38's soma (5th trace). A: a recording of a carbachol-induced bite in the suspended buccal mass. Note that RN activity occurs mainly during the retraction phase, so this is an ingestive program (Morton and Chiel 1993a). B38 fired once at the end of the retraction phase and once at the end of retraction of the previous bite. These one-for-one spikes can be seen on the B38 soma and iBN2 channels (connected by gray dashed lines). Note that because the B38 soma recording was obtained using an extracellular electrode, this electrode could also record activity from adjacent neurons (the small-unit activity on the B38 soma channel). The spikes generated on this channel by the neuron over which the electrode was positioned, in this case B38, were much larger in size and clearly distinguishable from the spikes generated by adjacent neurons. B: after seaweed was introduced into the mouth, the buccal mass transitioned to swallowing. Note that RN activity again occurs mainly during the retraction phase and also that the retraction phase is longer with more intense BN2 activity. B38 activity increased dramatically and continued through the protraction phase. The spikes on the B38 soma and iBN2 channels are one-for-one (solid gray underscores; see inset). After retraction and during protraction, B38's unit is the only large unit active on BN2. This is also evident on the cBN2, where B38 activity (dashed gray underline) is highly correlated with that of the iB38 but is not one-for-one.

B38 activity during protraction of swallowing is associated with the jaws closing on food.

What is the functional role of B38-induced anterior jaw closure? In swallowing patterns in the suspended buccal mass, we observed that when B38 fired intensely, the jaws closed on the seaweed during the protraction phase (Fig. 4A). Using this in vitro preparation provided key advantages in understanding the kinematics of these B38-evoked movements during motor programs: the suspended buccal mass provides a clearer view of the jaws than that seen in most in vivo patterns. Additionally, in the suspended buccal mass, it is easier to tell when the jaws are closing due to active contractions of I3 as opposed to the mouth passively closing. We consistently observed that intense B38 activity was associated with the jaws closing on food during in vitro swallowing. An analysis of one series of swallowing patterns showed a strong correlation between the duration of the B38 burst and the amount of jaw closure during protraction (Fig. 4B). It is also important to note that during patterns with intense B38 activity, there is generally no bursting of any other BN2 motor units during the protraction phase, implying even more strongly that B38 is responsible for this jaw closure.

Fig. 4.

B38 bursting is associated with the jaws closing on food. A: a recording of a swallowing pattern in the suspended buccal mass. B38's characteristic large-unit activity is indicated in the gray box: note that the B38 burst (gray box, 3rd trace) begins before and then overlaps the I2 protraction activity (1st trace). The jaw width, normalized to the top-to-bottom length of the jaws, is shown at bottom. The anterior jaws began to close ∼1 s after the beginning of the B38 burst and fully closed on the seaweed at the end of the B38 burst, during the protraction phase. During the retraction phase, the anterior jaws exhibited relaxation (jaw width increased) as the seaweed was pulled inward. In the example shown, the jaws did not relax fully before the subsequent swallow; this varied from swallow to swallow. B: a series of 14 swallowing patterns from 1 suspended buccal mass experiment was analyzed, and the B38 burst duration and minimum jaw width during protraction were measured. There was a strong and highly significant (P < 0.001) correlation between B38 burst duration and decreasing jaw width; i.e., longer B38 bursts were associated with more closure of the anterior jaws. We consistently observed, in experiments using the suspended buccal mass preparation, that whenever very intense B38 bursting occurred during the protraction phase, it was associated with anterior jaw closure.

B38-induced muscle contraction generates a force that resists outward movement of objects in the mouth.

These data show that B38 causes the jaws to close onto food but do not show that this closure produces enough force to help keep food in the mouth. To test this, we used a servomotor to move a polyethylene tube outward from between the jaws. We extracellularly stimulated B38 during this outward movement to test whether the muscle contraction would generate force resisting the movement. Indeed, we observed an inward force on the tube corresponding to the force generated in I3's anterior (Fig. 5). The maximum inward force on the tube when B38 was stimulated was significantly greater than the maximum inward force on the tube when it was moved out of the mouth in the absence of B38 neuron stimulation (average normalized force with B38 stimulation 0.84 ± 0.13, average normalized force without B38 stimulation 0.38 ± 0.14; n = 3 experiments, each containing 3 trials with and 3 trials without stimulation; P < 0.001, paired t-test).

Fig. 5.

B38-induced I3 contraction can generate force that resists outward movement of an object in the mouth. Inset, top: schematic showing an overhead view of the setup. A polyethylene tube was attached to a servomotor, the tube was positioned between the jaws of the buccal mass, and the servomotor was programmed to move the tube out of the mouth (1st row: outward movement of the tube). A force transducer (a in schematic) was used to measure the contraction force of I3's anterior (3rd row), and the servomotor (b in schematic) measured the corresponding inward force on the tube (4th row). A: a control run in which the tube moved outward but B38 was not stimulated. The tube movement elicited small positive forces in the force recordings from I3's anterior and from the tube, due to friction between the tube and muscle. The BN2 recording (2nd row) shows the spontaneous background level of BN2 activity during the experiment, which we observed did not cause I3 to contract. (When the tube was not moved, this spontaneous background activity induced no measurable force; data not shown.) B: just after tube movement began, B38 was stimulated via an extracellular electrode positioned above its soma (2nd row: BN2 recording showing large-unit B38 activity beginning immediately after stimulation). A large force was evoked in I3's anterior by the B38 stimulation, and a corresponding force was measured on the tube, which was much larger than the force during the control run. Note that the background activity of other BN2 units was similar to that during the control run. At the magnification level shown in the main BN2 recording of B, it may appear that there is a smaller BN2 unit that was activated by the extracellular stimulation at the same time as B38. The expansion at right shows that this is not a separate smaller unit but is instead part of the B38 action potential's BN2 unit. Below the expansion, a recording obtained during the same experiment is shown, during which B38 was located using an extracellular electrode. There is a one-to-one correspondence between spikes recorded from B38's soma and from BN2 (latency from B38 soma to BN2: 7.7 ± 0.4 ms based on 23 pairs of spikes). Note that B38's BN2 action potentials have the same size (average spike height 106.2 ± 3.7 μV during the record with simultaneous B38 soma recording vs. 104.5 ± 4.4 μV without) and shape in both recordings, confirming that the extracellular stimulation activated B38 during the servomotor run.

Intense B38 bursting during protraction occurs in swallowing but not in other patterns in vivo.

Is the increase in B38 activity during protraction in swallows observable in intact animals? We obtained electrophysiological and synchronized video recordings of animals performing different feeding behaviors in vivo and classified the behaviors as bites, swallows, or rejections (Kupfermann 1974) on the basis of the movements observed. Because of B38's characteristic large-unit BN2 activity occurring after retraction and during protraction, we were able to identify B38's activity on these in vivo recordings. We saw that many swallows contained intense B38 bursting continuing through much of the protraction phase (Fig. 6B). Bites (Fig. 6A) and rejections (Fig. 6D) contained much weaker B38 activity, which usually ended before or very early in protraction. Some swallows had a stronger than usual protraction so that the radula became visible at peak protraction, unlike in normal swallows. We termed these patterns visible radula swallows. These patterns often had more B38 activity than bites or rejections but less than normal swallows, and the activity did not persist as long into the protraction phase (Fig. 6C). These observations are consistent with our hypothesis, because if B38 activity caused the jaws to be closed throughout protraction, it is unlikely the radula would become visible.

Fig. 6.

B38 is most active in normal swallowing in vivo compared with other behaviors. All recordings shown are from the same animal. Note that the timing of RN activity relative to I2 protraction activity indicates the ingestive nature of the first 3 behaviors and the egestive nature of the 4th behavior. A: a biting pattern. B38 activity (gray box) is sparse, ending early in the protraction phase (protraction phase is indicated by the large burst of I2 activity). B: a normal swallowing pattern. B38 activity is very intense, continuing through the protraction phase. Note the large overlap between B38 activity and I2 activity. C: a visible radula swallow. B38 is more active than in the bite but less than in the normal swallow. B38 activity partially overlaps the I2 activity. D: a rejection pattern. B38 is not active.

At the individual pattern level, the B38 activity was highly variable between different responses even of the same behavior type (Fig. 7A shows 20 examples each of bites and normal swallows; similar variability was also present for the other behaviors). To characterize B38's typical activity in the different patterns, we calculated the mean B38 activity from the end of the previous retraction to the end of protraction for each type of pattern (Fig. 7B). These averages clearly showed that B38 was much more active in normal swallowing than in the other patterns. Additionally, the B38 activity in swallowing tended to peak at the onset of protraction, whereas no such peak appeared in the other patterns. We also quantified the number of identifiable B38 spikes in each B38 burst (Fig. 7C) and the average frequency during each B38 burst (as opposed to the instantaneous firing frequencies averaged in Fig. 7B) for the various patterns (Fig. 7D). Both the number of B38 spikes per burst and the average frequency for normal swallowing were significantly higher than for any of the other three patterns. The number of spikes and average frequency for visible radula swallowing were also significantly higher than for bites and rejections. These results support the hypothesis that B38-induced anterior jaw closure occurs specifically in swallowing patterns.

Fig. 7.

Although there is variability between individual responses, activity of B38 in swallowing is significantly greater than in other behaviors. A total of 150 bites, 130 normal swallows, 61 visible radula swallows, and 53 rejections taken from 6 intact, behaving animals were analyzed. All pattern data are included in C and D. For B, a subset of patterns was analyzed to best capture typical activity when B38 is active during a series of patterns (see materials and methods). The criteria used resulted in 31 bites, 84 normal swallows, 40 visible radula swallows, and 20 rejections being analyzed. A: to help illustrate the variability across different responses in the data set, the B38 activity from 20 bites (left) and 20 normal swallows (right) randomly selected from the set of patterns averaged in B is plotted. In each of the 2 columns, each row represents a single response. Each black tick mark represents a single B38 spike. Responses were aligned by the beginning of the protraction phase (center thick gray vertical line), identified from the I2 muscle recording. The left thick gray line in each row represents the end of the previous pattern's retraction phase, identified using the end of the bursting of motor neuron B43 on the BN2 recording (Lu et al. 2013). The right thick gray line in each row represents the end of the protraction phase, identified from the I2 muscle recording. The plots show considerable variability in both the durations and the intensity of B38 activity from response to response, but they also show that B38 tends to be more active in swallowing than in biting, especially during the protraction phase. B: average instantaneous firing frequencies for biting, normal swallowing, visible radula swallowing, and rejection. For each type of behavior, all responses were normalized and then averaged together. First, the instantaneous firing frequency (IFF) for each B38 spike was calculated. The plots of IFFs were then normalized. The activity from the end of the previous retraction (left gray lines in A) to the onset of protraction (center lines in A) was normalized to the time range −100 to 0. The activity from the onset of protraction (center lines in A) to the end of protraction (right gray lines in A) was normalized to the time range 0 to 100. Note that although this normalization process results in the x-axis scale being unitless, the IFFs retained the same values in Hz that they had before normalization. After normalization, the IFF plots for all responses of each type were aligned at the 3 normalized time points (−100, 0, 100) and averaged. Solid and dashed lines show averages ± SE. In normal swallowing, average firing frequency increases and peaks early in protraction. In visible radula swallowing, average firing frequency is lower than in normal swallowing and declines toward zero earlier in protraction. In biting and rejection, average firing frequency is lower than in either type of swallowing and falls to zero very early in protraction. C: total numbers of B38 spikes in the identifiable B38 bursts for each response. Box and whisker plots show the median (center line), 1st and 3rd quartiles (edges of box), and minimum and maximum values. D: average frequency of B38's firing in the identifiable B38 bursts for each response. *P < 0.001, significantly less than normal swallowing (Mann-Whitney test). **P < 0.001, significantly less than both normal swallowing and visible radula swallowing (Mann-Whitney test). For the comparisons indicated, with Bonferroni correction, we consider P < 0.05/10 = 0.005 to be significant.

Repeated stimulation of B38 dramatically shifts the force-frequency relationship.

After quantifying B38's in vivo activity, it became possible to determine whether this activity would be sufficient to close the jaws. To test this, we generated a force-frequency curve by intracellularly stimulating B38 to fire with 2-s pulse trains at frequencies covering the in vivo range and by measuring the forces generated in I3's anterior (Fig. 8A). Two-second stimulations were used because the average measured burst duration for in vivo swallows was 1.84 s, and this is likely to be a slight underestimate because the B38 burst sometimes begins late in the previous retraction and the early B38 spikes cannot always be clearly distinguished from the retraction-phase units produced by other motor neurons. We observed that firing frequencies below 10 Hz typically generated no force. The threshold frequency for force generation typically fell in the 10- to 12-Hz range. We observed a roughly linear increase in force with frequency from the point where force began to be generated up to ∼20 Hz, after which the beginning of saturation was evident. Our results in Fig. 7B show that the in vivo median frequency for B38 in swallowing was 8.25 Hz. This would imply that even though B38 is specifically recruited in swallowing, the B38 activity in most in vivo swallows is still insufficient to cause muscle contraction. Is there a way to resolve this apparent discrepancy? Previous work showing that repeated stimulation of B38 leads to large force increases due to B38's release of small cardioactive peptide (Fox and Lloyd 1997, 2001; Lotshaw and Lloyd 1990) suggests that modulation may be critical.

Fig. 8.

Repeated stimulation of B38 shifts the force-frequency relationship. A: we generated an unmodulated force-frequency curve by performing single stimulations of B38 for 2 s at a range of frequencies. Results are averaged from 5 experiments (error bars indicate ±SD). Forces were normalized to the maximum force from that experiment. The median in vivo B38 swallowing frequency is indicated (arrow); note that this frequency falls essentially at the bottom of the curve. B: forces from a single 2-s, 12-Hz B38 stimulation (left), a series of 2-s, 8-Hz stimulations every 5 s (center), and a series of 2-s, 10-Hz stimulations every 5 s (right). All data are from the same experiment. Black horizontal bars indicate times of stimulation. Repeated stimulation at 8 Hz took the force from zero to nearly the size of an unmodulated 12-Hz stimulation. Repeated stimulation at 10 Hz generated even larger forces. C: averaged forces obtained using the protocols shown in B: single 2-s, 12-Hz stimulations (left, 13 experiments), a series of 2-s, 8-Hz stimulations (center, 12 experiments), and a series of 2-s, 10-Hz stimulations (right, 11 experiments). Error bars indicate ±SE. For the 8-Hz stimulation series, the average initial force was approximately zero, and the average force increased to reach the same level as the average unmodulated 12-Hz force by the 6th stimulation. For the 10-Hz stimulation series, the average initial force was barely above zero, and the average force rapidly increased, reaching a level twice that of the average unmodulated 12-Hz force.

To test the effectiveness of this modulation by B38 onto itself, which we termed “self-modulation” because the neuron itself is responsible for its own modulation [a subtype of intrinsic modulation, which has been defined (Cropper et al. 1987; Katz 1995) as modulation from within the same circuit], we stimulated B38 repeatedly, using 2-s pulse trains with 3-s gaps in between trains. The resulting one stimulation per 5-s matches the most common inter-swallow interval in vivo (Hurwitz and Susswein 1992; our observations were consistent with these results). We used frequencies of 8 and 10 Hz, near the median in vivo frequency of 8.25 Hz, so that our stimulations simulated in vivo swallowing series. In the absence of modulation, a 2-s, 8-Hz B38 stimulation elicited muscle contraction in just 6.3% (1/16) of experiments. Even a 2-s, 10-Hz stimulation elicited muscle contraction in just 18.8% (3/16) of experiments. Repeated stimulations caused a rapid increase in force, with the first train generating little to no force and subsequent trains generating increasingly large forces (Fig. 8B). The increased forces tended to level off around the fourth to fifth stimulation. When we tested these repeated stimulations, 8-Hz B38 activation generated force in 83.3% (10/12) of experiments and 10-Hz stimulation in 90.9% (10/11) of experiments. By contrast, repeatedly stimulating with 5-Hz, 2-s pulse trains did not generate any force (data not shown). Combined with our in vivo data on B38's activity (Fig. 7), these results suggest that modulation allows B38 to cause anterior jaw closure in the protraction phase of many, but not all, in vivo swallows, but not in bites or rejections.

Modulation via the metacerebral cell and via B38's biting activity also aid B38-evoked contractions in swallowing.

In a series of swallows, B38's self-modulation helps it generate force, but this does not explain how B38 would generate force in the first swallow of a series. There are two potential sources of modulation that could be deployed before any swallows occur. When an animal contacts food, it becomes aroused and attempts to grasp food through biting; once food is grasped, the transition to swallowing occurs. B38 is not entirely silent during biting; rather, it fires at levels far below the threshold to elicit muscle contraction. Figure 6A shows an example of this low-frequency B38 activity. Could the weak B38 activity during biting modulate the muscle contractions evoked by B38 during swallowing? Another source of modulation toward B38-evoked contractions is the MCC (Fox and Lloyd 1998), a serotonergic neuron that begins to burst during food arousal. The MCC's in vivo activity has been recorded in two previous studies (Kupfermann and Weiss 1982; Weiss et al. 1978). Results from these studies suggest that the MCC initially bursts at an average frequency of ∼3 Hz for about 20 s upon the animal's first contact with food, and that the average frequency then drops to ∼1 Hz before gradually declining over the course of the meal. Could this MCC activity also help prepare for B38-evoked contractions at the onset of a swallowing series?

To answer these questions, we stimulated B38 for 2 s at a frequency just above the threshold for unmodulated force generation in a given experiment, then stimulated B38 at the same frequency after applying either a weak B38 stimulation or a stimulation of the MCC, and measured the change in force. Specifically, we waited 200 s between B38 stimulations, which is the duration we observed was necessary for self-modulation's effects to fully dissipate after a single 2-s stimulation. To simulate B38 activity in a bite followed by a swallow, we stimulated B38 at 2 Hz for 2 s and then stimulated B38 at the suprathreshold frequency after a 3-s gap. We observed that this weak B38 activation caused a large and significant increase in the force generated by the subsequent B38 stimulation (Fig. 9, A and D). For the MCC, we simulated the initial burst by using a 20-s, 3-Hz stimulation to see whether the MCC's modulation would have already taken effect if the animal transitioned to swallowing very early in an ingestive sequence. This stimulation caused a significant increase in the B38-evoked force (Fig. 9, B and D). We also tested a 20-s, 3-Hz stimulation followed by 60-s, 1-Hz MCC stimulation to see whether the MCC modulation would increase as the animal spent more time feeding. This resulted in a much larger increase in the force (Fig. 9, C and D), which is not surprising in light of the finding by Weiss et al. (1978) that there is a time lag before the peak modulation from the MCC onto Aplysia muscle I5 is reached. Additionally, we observed that adding MCC and self-modulation together resulted in larger forces than applying either one individually (data not shown). These results suggest that although B38 does not have any active role in eliciting muscle contraction during biting itself, activity of both B38 and the MCC during the initial biting stage of a meal builds up modulation to help B38 evoke I3 contractions when the animal switches to swallowing.

Fig. 9.

Weak B38 activity during biting and extrinsic modulation via the metacerebral cell (MCC) provide additional modulation. A: we stimulated B38 at a frequency just above threshold for force generation every 200 s. We preceded every other stimulation with a 2-Hz stimulation, a frequency far below threshold, to simulate a bite preceding a swallow. A single pair of stimulations from a series is shown. This weak B38 stimulation caused a large increase in force. B: we stimulated B38 every 200 s at a frequency just above threshold without and with a preceding stimulation of the MCC. A 20-s, 3-Hz MCC stimulation caused an increase in force. C: a 20-s, 3-Hz MCC stimulation followed by a 60-s, 1-Hz MCC stimulation caused a larger increase in the force generated by B38 stimulation. D: average modulated forces for each of the 3 treatments shown in A–C as indicated. Forces were normalized to the average pretreatment force for the given treatment. Bar A, 12 stimulation pairs from 3 experiments; bar B, 9 stimulation pairs from 3 experiments; bar C, 9 stimulation pairs from 3 experiments. Error bars indicate ±SE. *P < 0.01, significantly different from pretreatment value (paired t-test).

DISCUSSION

Our results strongly support the conclusion that B38 causes anterior jaw closure during protraction of swallowing in Aplysia. We have shown that protraction-group neuron B38 specifically activates the I3 jaw muscle's anterior, that B38 activity dramatically increases during swallowing, and that B38's activity in swallowing but not in other patterns can generate contractions with the aid of neuromodulation. Muscle I3 is known as a retractor, contraction of which causes Aplysia's food grasper to retract (Morton and Chiel 1993a). Because B38-induced contractions occur during protraction, this strongly implies that I3 has an additional function. In a semi-intact preparation that provides a better view of jaw movements than is possible in vivo, the jaws appear to close on food when B38 bursts intensely. This suggests that Aplysia has evolved an intriguing mechanism for keeping hold of food throughout ingestion sequences (Fig. 10, A and B). During retraction, the closed radula is the part of the feeding apparatus that holds food. During protraction, the radula cannot hold food, because if it did, the food would be pushed back out. Thus the radula is open as it protracts. If nothing else were holding the food, it could potentially slip outward. Recruiting the anterior jaw muscles to hold food during protraction ensures that the animal does not lose its grip. Using a servomotor, we showed that B38 could induce a force to resist outward movement of an object in the mouth. Thus the grasper and jaws repeatedly “take turns” holding food during swallowing, akin to a person pulling a rope hand-over-hand. It is critical that the jaws close during protraction only in swallowing and not in other behaviors. Of biting, swallowing, and rejection, swallowing is the only behavior in which radula protraction typically does not reach the jaws. In biting, premature jaw closure would block the radula from reaching food. In rejection, premature jaw closure would block the inedible object from being expelled (Ye et al. 2006b). Furthermore, it is also critical that I3 contraction during protraction in swallowing is specific to I3's anterior, because the radula does protract into the posterior I3 lumen (Neustadter et al. 2002); contraction of the whole muscle as one unit would block protraction even in swallowing. Differential activation of motor neurons and activation of muscles as whole units or subunits may be important elements in generating multifunctionality.

Fig. 10.

Schematics showing the dual role of the I3 jaw muscle in swallowing patterns and the significance of B38's protraction phase activity (A) as well as how both intrinsic and extrinsic modulation act to allow effective muscle contraction when B38 fires in swallowing (B). A1: schematic cut-through drawing of the buccal mass (side view). The mouth is at right and the esophagus at left. The radula (yellow) and I3 muscle (red) are highlighted, and a plane is drawn to indicate the cross section represented schematically in the remaining sections of A, which show an overhead view. Note that in the left drawing, the I3 muscle is only shown on one side of the buccal mass, whereas in the remaining schematics, the I3 muscle is shown on both sides. In both A2 and A3, 2 patterns are shown in sequence. Positions of the radula (yellow) relative to the I3 muscle (red; schematic representation of the I3 muscle's 2 symmetrical halves) during protraction and retraction are based on MRI images of in vivo swallowing responses (Neustadter et al. 2002). A2: 2 swallowing patterns with B38 activity removed. In the protraction phase (a and c), the open radula moves forward. Black arrows indicate the direction of radula movement. Because nothing is holding the food (green), the food can slip outward. Green arrows indicate the direction of seaweed movement. The dashed line is provided as a reference for comparing seaweed position in corresponding panels of A2 and A3. In the retraction phase (b and d), the radula closes on the food and motor neurons such as B3 and B6 contract the whole I3 muscle to mediate radula retraction and inward food movement. A3: swallowing patterns including B38's protraction phase activity. In the protraction phase (a and c), the open radula moves forward and B38 causes the anterior jaws to close on the food and prevent its outward movement. Note the importance of B38 specifically activating I3's anterior: if the whole I3 contracted during protraction, the radula's forward movement would be blocked. In the retraction phase (b and d), the radula closes on the food and neurons such as B3 and B6 contract the whole I3 muscle. Note that net inward movement of the seaweed is greater in A3 than in A2 due to B38 preventing outward movement during protraction. [Modified from schematics developed by Dr. Hui Ye for Figs. 1, 3, and 8 in Ye et al. (2006a) with permission.] B: 2 B38 bursts are shown in sequence. If the effects of modulation were removed (no modulation), B38 would be ineffective at generating force (bottom, dashed line). Adding B38's intrinsic modulation from repeated swallowing bursts (swallowing self-modulation) would allow muscle contraction due to later B38 bursts, but the initial B38 burst would still be ineffective (middle, dashed line). When both extrinsic and intrinsic modulation are present and have begun to build up during the initial biting stage of a meal (MCC + biting self-modulation + swallowing self-modulation), as is the case in real behavior, B38 can cause muscle contraction starting from the initial swallowing burst (top, solid line).

Types of ingestive behavior.

The level of B38 activity in visible radula swallows has interesting implications regarding Aplysia's multifunctional pattern generator. These swallows have less B38 activity than normal swallows, but more than bites or rejections. Some number of visible radula swallows likely contain anterior jaw closure early in protraction. (If the jaws were closed throughout protraction, the radula could not become visible, but jaw closure early in protraction could still hold food; as Fig. 7A shows, B38 activity declines sooner in visible radula swallows.) Strikingly, visible radula swallows also fall between bites and normal swallows in other parameters (e.g., protraction strength). Bites and swallows may be two ends of an ingestive behavior continuum, not discrete types.

Sources of increased B38 activation.

What mechanism causes increased B38 activation? We observed large variability in B38 activity, suggesting that the anterior jaws close during protraction of many, but not all, swallows. Sensory feedback may contribute to the strength of B38 activation. In our experiments, animals fed on isolated seaweed strips. Aplysia in the wild encounter a much greater variety of stimuli. Food is usually attached to holdfasts and can be subject to strong tidal forces, so it could be pulled outward if the animal did not maintain an active grasp. Thus there may be certain situations where jaw closure during protraction would be particularly advantageous. We observed that visible radula swallows were less common with thicker seaweed strips (unpublished data); because visible radula swallows have less B38 activity, increased loading may be associated with increased B38 recruitment. Another possible role for anterior-specific jaw closure is that if the jaws held food as the radula moved toward peak retraction, this could allow the animal to cut food, a behavior observed in animals swallowing weighted seaweed strips (Hurwitz and Susswein 1992). B38 activity sometimes begins late in retraction, so this scenario is plausible.

Significance of neuromodulation.

Modulation is crucial for B38-induced jaw closure. Unmodulated B38 stimulations at in vivo swallowing frequencies are not functional. Repeated stimulations result in large force increases, consistent with the findings of Lotshaw and Lloyd (1990). Fox and Lloyd (2001) showed that these increases are largely caused by release of SCP from B38. Another source of modulation is the MCC, a serotonergic interneuron that begins firing when the animal first contacts food (Kupfermann and Weiss 1982). By simulating neurons' in vivo activity, our work provides a clear behavioral context for previous findings on neuromodulation (Fig. 10C). Intrinsic (i.e., from B38) and extrinsic (i.e., from the MCC) modulation work together to ensure muscle contraction can occur. Also of note, in motor program series, activity of a motor neuron with no functional role in one behavior (i.e., B38 during biting) can cause modulation to prepare for a subsequent behavior (i.e., swallowing) in which the neuron is more active.

Neuromodulation has been extensively studied in invertebrates (Calabrese 1989). Thus far, it appears that modulatory inputs to muscle are much rarer in vertebrate skeletal muscle, whereas modulatory inputs to motor neurons are common in both (Belanger 2005). Serotonergic modulation has been shown to increase motor neuron excitability in vertebrates (Heckman et al. 2009). Although our work focuses on modulation at the muscle, serotonin and SCP also presynaptically modulate B38 (Lotshaw and Lloyd 1990). Belanger (2005) proposed that modulation of the muscle may be used more by invertebrates because they cannot vary the number of motor units recruited to nearly the extent that vertebrates can. There may be a smaller scale application of this concept within I3's motor pool. B38 activates I3 alone during protraction, whereas during retraction, several neurons (e.g., B3, B6, and B9; Church and Lloyd 1994) can be recruited to activate the whole muscle. Correspondingly, serotonergic modulation of I3 is much more effective for B38 than for B3 (Fox and Lloyd 1997), B6, or B9 (unpublished data). Additionally, although neuromodulation of muscle contractions appears to be rare in vertebrate skeletal muscle, it is a well-described property of vertebrate smooth muscle (Burnstock and Verkhratsky 2010; Savineau and Marthan 1997), muscle that, like molluscan smooth muscle (Cohen et al. 1978), contracts as a result of summating excitatory junction potentials (Burnstock and Holman 1960, 1961). Findings on neuromodulation of muscle contractions in Aplysia may therefore be particularly relevant to vertebrate smooth muscle systems.

Other work in Aplysia has shown the complementary effects of extrinsic and intrinsic modulation on the animal's arousal state. Stimulation of the command-like cerebral-buccal interneuron 2 (CBI-2) initiates motor programs by causing activity in protraction motor neurons B61/62. Prior stimulation of the MCCs increases the size of excitatory postsynaptic potentials that CBI-2 evokes in B61/62 and decreases the latency to motor program initiation (Proekt and Weiss 2003). Repeated stimulation of CBI-2 has similar effects (Proekt and Weiss 2003), and this action occurs in part via neuropeptides intrinsic to CBI-2 (Koh and Weiss 2005). The extrinsic and intrinsic modulation of the central pattern generator by the MCCs and CBI-2 may therefore be seen as analogous to the extrinsic and intrinsic modulation of the I3 muscle by the MCCs and B38. Because the activity of the MCCs decreases over the course of a meal, extrinsic modulation may be more important to feeding responses early in a meal, and intrinsic modulation may be more important later in a meal. Thus intrinsic and extrinsic modulation are important at multiple levels of the feeding circuit; our results provide further evidence of their importance at the neuromuscular level.

Multifunctionality of the I3 muscle.

Our work is not the first to suggest I3 has multiple functions. In addition to its known role in retraction, it has been suggested that I3's posterior may assist protraction in biting. In vivo MRI showed that at the peak protraction of biting, the radula was forward of I3's posterior region, and a posterior-specific contraction at this time could potentially provide a protractive rather than a retractive force (Neustadter et al. 2007). If this hypothesis is true, the muscle would have three different functions. Furthermore, one function (mediating retraction) would involve the whole muscle acting as a unit, with many motor neurons firing together, whereas the other functions would involve smaller muscle regions acting separately, activated by smaller parts of the motor pool. Regional activation would be key to both of the muscle's additional functions: only the posterior is in a position to aid protraction during biting (Neustadter et al. 2007), whereas only the anterior is in a position to grasp food without preventing the grasper's forward movement during protraction in swallowing (Neustadter et al. 2002). By contrast, during retraction in multiple behaviors, the whole I3 muscle is in a position to provide a retractive force. Muscles that act as whole units or as multiple subunits have been observed elsewhere. The cat activates all regions of the biceps femoris synchronously in ear scratching, but in running, the posterior is active out of phase from the rest of the muscle (Chanaud et al. 1991). In Aplysia, the whole I3 is activated synchronously in retraction of multiple behaviors, and the anterior is active out of phase from the whole muscle in swallowing. When the activation phase of a muscle region, relative to the rest of the muscle, changes in different behaviors, this may indicate that the muscle has multiple functions.

In vertebrates such as the cat, changes in regional muscle activation in different behaviors have been studied but have not been related to single, identified neurons. Work in cockroaches (e.g., Watson and Ritzmann 1998) has shown changes in activity of single, identified motor neurons, but these cockroach neurons activate whole muscles. Our work ties changes in activation of a muscle region to a single, identified neuron. Moreover, our work demonstrates how this single neuron, by activating a muscle region out of phase from the muscle's normal activation, can give the muscle an additional function.

The ball-and-ring structure of Aplysia's radula-odontophore vis-à-vis I3 has led to the development of biologically inspired robotics (Mangan et al. 2005) based on the idea that I3's posterior alternatingly mediates protraction and retraction. The mechanism described presently, of the grasper and jaws serving as alternating contact points to ensure that grasp of an object is never lost, could also provide such inspiration. Shifting the timing of closure so that the object was held at both contact points during retraction could also allow cutting, as we propose may occur when Aplysia cut food (Hurwitz and Susswein 1992).

Multifunctionality, the ability of a peripheral structure to produce multiple, distinct behaviors, is a common feature of organisms that have adapted to ever-changing environments. In this article, we have described some of the features underlying the generation of multifunctionality in a system tractable to neurophysiological study. A single muscle (I3) traditionally thought to have one role was shown to have multiple functions. In the retraction phase of several different behaviors, the whole muscle contracts as one unit, through activity of multiple motor neurons, to cause food grasper retraction. In the protraction phase of one specific behavior (swallowing), a subregion (the anterior) of the muscle contracts to perform a different function: holding food. This activation occurs via a single, identified motor neuron: B38, and neuromodulation is critical to this neuron eliciting muscle contraction in vivo. Thus, through differential activation of a single neuron, a muscle assumes an entirely different function from its normal activity.

GRANTS

Research reported in this article was supported by National Institute of Neurological Disorders and Stroke Grant NS047073 and National Science Foundation Grants DMS-1010434 and IIS-1065489.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.M.M., H.L., M.J.C., and H.J.C. conception and design of research; J.M.M., H.L., and M.J.C. performed experiments; J.M.M., M.J.C., and H.J.C. analyzed data; J.M.M., H.L., M.J.C., and H.J.C. interpreted results of experiments; J.M.M. prepared figures; J.M.M. drafted manuscript; J.M.M. and H.J.C. edited and revised manuscript; J.M.M., H.L., M.J.C., and H.J.C. approved final version of manuscript.