Abstract
Similar design characterizes neuronal networks for goal-directed motor control across the complex, segmented vertebrates, insects, and polychaete annelids with jointed appendages. Evidence is lacking for whether this design evolved independently in those lineages, evolved in parallel with segmentation and appendages, or could have been present in a soft-bodied common ancestor. We examined coordination of locomotion in an unsegmented, ciliolocomoting gastropod, the sea slug Pleurobranchaea californica, which may better resemble the urbilaterian ancestor. Previously, bilateral A-cluster neurons in cerebral ganglion lobes were found to compose a multifunctional premotor network controlling the escape swim and feeding suppression, and mediating action selection for approach or avoidance turns. Serotonergic As interneurons of this cluster were critical elements for swimming, turning, and behavioral arousal. Here, known functions were extended to show that the As2/3 cells of the As group drove crawling locomotion via descending signals to pedal ganglia effector networks for ciliolocomotion and were inhibited during fictive feeding and withdrawal. Crawling was suppressed in aversive turns, defensive withdrawal, and active feeding, but not during stimulus–approach turns or prebite proboscis extension. Ciliary beating was not inhibited during escape swimming. These results show how locomotion is adaptively coordinated in tracking, handling, and consuming resources, and in defense. Taken with previous results, they also show that the A-cluster network acts similarly to the vertebrate reticular formation with its serotonergic raphe nuclei in facilitating locomotion, postural movements, and motor arousal. Thus, the general scheme controlling locomotion and posture might well have preceded the evolution of segmented bodies and articulated appendages.
SIGNIFICANCE STATEMENT Similar design in the neuronal networks for goal-directed motor control is seen across the complex, segmented vertebrates, insects, and polychaete annelids with jointed appendages. Whether that design evolved independently or in parallel with complexity in body and behavior has been unanswered. Here it is shown that a simple sea slug, with primitive ciliary locomotion and lacking segmentation and appendages, has similar modular design in network coordination as vertebrates for posture in directional turns and withdrawal, locomotion, and general arousal. This suggests that a general neuroanatomical framework for the control of locomotion and posture could have arisen early during the evolution of bilaterians.
Introduction
Did the general neural organization of decision and action selection precede or follow the evolution of complex body form and behavior? Vertebrates, insects, and polychaete annelids likely evolved segmentation, articulated appendages, and behaviors independently (Mullins et al., 2011; Hochner, 2013; Katz, 2016), yet they share analogous modular designs for computational functions that mediate decision, spatial mapping, and motivation (Loesel, 2011; Tosches and Arendt, 2013; Holland, 2016). Such functions may have originated in the ancestral bilaterian with simpler anatomy and behavior or could have evolved independently multiple times. The above lineages each have more complicated body forms than the expected urbilaterian (Hejnol and Martindale, 2008). For comparative analysis, it is useful to examine circuits in organisms more similar to the common ancestor.
One organism accessible to such examination is the sea slug Pleurobranchaea californica, which lacks complicating segmentation and appendages, has a much simpler nervous system, and expresses very primitive ciliolocomotion, where myriad cilia on the foot paddle through secreted mucus. Previous work described neuronal circuitry underlying cost–benefit and approach–avoidance decisions in foraging by Pleurobranchaea (Jing and Gillette, 2000; Gillette and Jing, 2001; Yafremava et al., 2007; Hirayama et al., 2012; Hirayama and Gillette, 2012), and these circuits shared functional analogies with the vertebrate striatum and hypothalamus (Gillette and Brown, 2015). Here we extend the analogies to coordination of locomotion in goal-directed behaviors.
The premotor organization of the locomotion and posture of gastropods bears significant resemblance to functions of the reticular activating system of the vertebrate brainstem (Gillette, 2006; Gillette and Brown, 2015; Lillesaar and Gaspar, 2019) and the head ganglia in the insect brain (Emanuel et al., 2020). The reticular formation of the vertebrate brainstem houses a premotor system mediating posture and locomotion, and motor arousal. It receives inputs from basal ganglia for select actions, adapts them to intrinsic premotor patterns, and transmits them to motor pattern generators in the spinal cord and cranial nuclei (Takakusaki et al., 2016; Brownstone and Chopek, 2018). Prominent serotonergic outputs from the raphe nuclei of the system provide critical neuromodulatory outputs to the entire CNS, including motor circuitry in spinal cord.
Neurons of the A-cluster in the cerebral lobes of the cerebropleural ganglion of Pleurobranchaea express a multifunctional premotor network that adopts multiple coordinated states to mediate different behaviors, driving escape swimming (Jing and Gillette, 1999, 2000), contributing to postural movements in approach and avoidance turning and withdrawal (Jing and Gillette, 2003; Brown, 2014), and enhancing the general arousal state (Jing and Gillette, 2000; Gillette and Jing, 2001). Descending outputs from A-cluster to effector networks in the pedal ganglia activate appropriate motor pattern generation. Serotonergic neurons of the cluster are necessary to each behavioral network state, as for the serotonergic raphe nuclei of the reticular formation (Gillette, 2006; Takakusaki et al., 2016). The neuronal mechanisms for crawling remain uncharacterized in Pleurobranchaea, but in both the distantly related sea slugs Tritonia exsulans (formerly Tritonia diomedea) and Aplysia californica, apparent homologs to the serotonergic cells As2/3 of the A-cluster have been shown to drive locomotion (Popescu and Frost, 2002; Jing et al., 2008). Given these homologies, we tested whether the As2/3 cells, an electrically coupled neuron pair of adjacent, apparently identical neurons, also drive the crawling of Pleurobranchaea and, further, how they coordinate in different goal-directed behaviors.
For Pleurobranchaea, incentive to crawl varies with stimulus quality and hunger state. Individuals crawl after turning either toward or away from a stimulus (video at https://publish.illinois.edu/slug-city/videos/) and following escape swims initiated by noxious stimuli (Jing and Gillette, 1995). Hungry Pleurobranchaea respond to appetitive stimuli with approach turns and crawling, and if sufficiently hungry will also attack noxious stimuli. In contrast, partially satiated individuals respond to noxious stimuli by transiently suppressing locomotion during stereotypic avoidance turns and then resume crawling. Finally, satiated animals actively avoid appetitive stimuli as if they were noxious (Gillette et al., 2000).
We found that the As2/3 neurons send crawling signals to the pedal ganglia to activate ciliary mucus transport on the foot, and that both ciliolocomotion and As2/3 activity are similarly coordinated with appetitive, consummatory, and aversive behaviors for adaptive expression. The present results augment the functional similarities of the A-cluster to the interactions observed in the vertebrate reticular formation for control of posture, locomotion, and general arousal.
Materials and Methods
Animals
Specimens of P. californica were trapped by the Monterey Abalone Company, shipped overnight, and housed individually in artificial seawater (ASW; Instant Ocean) at 12°C.
Experimental design and statistical analysis
Measuring ciliolocomotion.
Ciliolocomotor transport was quantified by tracking the anterior-to-posterior movement of silicon carbide grains (mesh no. 120) placed on the mucus of the foot, in both semi-intact preparations (described in the Electrophysiology subsection) and intact animals. Video recordings were made of the carbide grain movement, converted to AVI (Audio Video Interleave) files, and analyzed with the open-source software ImageJ using the MtrackJ plugin (Fig. 1; Meijering et al., 2012; Schneider et al., 2012). For analysis, XY coordinates of representative particles were tracked at 1 s intervals. To correct for changes in a the position of a particle caused by muscular movements of the foot, rather than by ciliary beating, XY coordinates of a nearby portion of the foot were also recorded and the instantaneous foot movement was subtracted from that of the particles.
Carbide particle tracking in ImageJ. The XY positions of particles were recorded at 1 s intervals for periods of interest (longer trace). To account for changes in position caused by movement of the foot itself, the XY position of a section of the foot was simultaneously recorded (shorter trace, star), and the movement of the foot was subtracted from the movement of the carbide particle.
Electrophysiology.
To assess neuronal correlates of crawling and its suppression, extracellular nerve and intracellular neuron recordings were made in both semi-intact and isolated CNS preparations. For isolated preparations, the entire CNS, with interganglionic connectives intact (Fig. 2), was removed under cold anesthesia (4°C) and pinned dorsal side up. For semi-intact preparations, buccal mass and viscera were removed through a dorsal mantle incision, and the head region was excised, including the oral veil, rhinophores, and mouth region. The foot was left intact with cerebropleural body wall nerves and all pedal ganglia nerves connected. The foot was then pinned ventral side up in a Sylgard dish, and the CNS was pinned dorsal side up. A constant flow of 12°C ASW was maintained via Marriott bottle. For intracellular recordings of the As2/3 neurons, ganglia were locally proteased for 6–8 min at room temperature with Sigma-Aldrich type XIV protease, 7% in ASW, to facilitate dissection, after which the outer and inner sheaths of the CNS were manually removed to expose the neuronal somata. Nerve recordings were made with suction electrodes of polyethylene tubing via a differential AC amplifier (model 1700, A-M Systems). Intracellular recordings were made with glass microelectrodes filled with 3 m KCl and with resistances of 12–20 MΩ, using an intracellular amplifier (model 1600, A-M Systems). In some experiments, the three pedal nerves (PNs) were severed to assess their effects on crawling. Recordings were digitized with a sampling rate of 20 kHz using a PowerLab 8/30 (ADInstruments), and displayed with LabChart software (version 7; ADInstruments).
The isolated Pleurobranchaea CNS showing the soma of a single As2/3 neuron originating in the right cerebropleural ganglion is illustrated, with its major axon collateral descending to the contralateral pedal ganglion (see Jing and Gillette, 1999). BG, Buccal ganglion; CePlG, cerebropleural ganglion; PdG, pedal ganglion (bilaterally paired).
Assessing effects of different behaviors on crawling-related activity.
Effects of escape swimming and appetitive behaviors on carbide movement in the foot mucus were assessed in intact individuals. For escape swimming, individuals, after deposition of carbide on their feet, were placed foot-down in an aquarium ∼0.5 m deep and were induced to swim by a shock train applied to the dorsal mantle (30 V; pulse duration, 4 ms; 15 Hz for 2.5 s). Individuals were allowed to move freely, and, as with the recordings in the semi-intact preparation, XY coordinates were recorded for both carbide particles and an adjacent portion of the foot, and the movement of the foot was subtracted from that of the particles to determine the net particle movement. Particles were tracked during the dorsal flexion phase of each swim cycle, but not during the ventral flexion phase when the foot was obscured (Jing and Gillette, 1999). Video recordings of particle movement were made by an iPhone 5 camera below the clear tank bottom. To observe effects of approach turning and active feeding on ciliary beating, subjects were placed ventral side up in a cooled, shallow ASW-filled chamber with the foot at the water interface, and carbide was deposited on their feet. Pieces of shrimp were gently applied to the center of the chemosensory oral veils of the animals to elicit appetitive responses, but animals were not allowed to consume the shrimp.
Effects of aversive turning on crawling and fictive crawling were assessed in both semi-intact preparations and the isolated CNS, while the effect of defensive withdrawal was assessed in only the semi-intact preparation. Fictive turns were elicited in the isolated CNS by stimulating one of the two bilateral sensory large oral veil nerves (LOVNs) that innervate the oral veil, and were recorded with suction electrodes from the posterior lateral body wall nerves (plBWNs) that carry turn motor output (Jing and Gillette, 2003; Hirayama and Gillette, 2012).
Defensive withdrawal not involving the aversive turn was induced by similarly stimulating one of the cerebropleural BWNs that innervate the posterior body wall to elicit a single, sustained contraction. Stimulation parameters differed between the two experiments and are reported separately. For the semi-intact preparations, the movement of carbide was recorded by a Celestron Digital Microscope Imager. To quantify the nerve firing rate for a given section of a recording, spike frequency over a given period was calculated with the LabChart Event Count function. Where appropriate stimulus artifacts were subtracted from the obtained number.
Effects of fictive feeding on As2/3 activity were assessed in the isolated CNS. Feeding motor output was monitored by recording from buccal ganglion nerves R1 and R3, and was induced by stimulating the stomatogastric nerve (3–5 V, 5–7 Hz) until robust feeding bursts began.
Statistical analysis.
Statistical tests were performed with the GraphPad Prism software (version 8.2.0). For all tests with stimulation of nerves or identified neurons, ANOVAs with post-tests for multiple comparisons were performed to assess changes in ciliolocomotor transport rate; repeated-measures (RM) ANOVAs with Tukey's post-tests were performed for experiments in which all datasets were normally distributed, and Friedman tests with Dunn's multiple-comparisons post-tests were used when at least one dataset was not normally distributed. For the experiments in which the firing rate in the pedal nerves was compared with the rate of ciliolocomotor transport, linear regression t tests were performed. Data are reported as the mean and SE.
Results
Anterior, medial, and posterior pedal nerves drive ciliary beating
In simultaneous records of the anterior PN (aPN), medial PN (mPN), and posterior PN (pPN) spiking and cilia-driven carbide movement, spontaneous firing rates were significantly correlated with the rate of ciliary beating (aPN, n = 5; pPN, n = 4; mPN, n = 1; R2 of at least 0.3192 and p < 0.05 for each experiment; Fig. 3, Extended Data Table 3-1). PN spike activity was tonic and did not show cyclic bursting while carbide was moving along the foot, suggesting that ciliolocomotion was not driven by a rhythmically active central pattern generator (CPG), but rather by tonic activity of ciliolocomotor neurons. En passant suction electrode stimulation of each nerve accelerated the rate of ciliary beating, which declined when stimulation stopped (Fig. 4, Extended Data Table 4-2). Finally, when each nerve was severed from the pedal ganglion and stimulated at the distal cut end, ciliary beating only increased in a portion of the foot (Fig. 5, Extended Data Table 5-3). The aPN only elicited activity in the anterior portion of the foot, the mPN in the medial portion of the foot, and the pPN in the medial and posterior portions of the foot. The greater innervation field of the pPN likely reflects its greater size, with multiple smaller branches diverging from it. Stimulating the two other pedal nerves that innervate the longitudinal body wall muscle and mediate withdrawal and turning, the anterior lateral BWN and plBWN, did not affect ciliary beating (Extended Data Table 4-2).
Pedal nerve activity correlated with crawling rate. A, Sample linear regression comparing firing rate to ciliolocomotor transport rate (F(13) = 18.70, p = 0.001). B, C, Sample recordings from a single experiment illustrating spontaneous activity during both relatively slow (B) and fast (C) ciliary transport (recordings obtained from the pPN). For this experiment, rates were calculated for 10 s bins. Summary data are found in Extended Data Table 3-1.
Table 3-1
Summary of linear regressions of PN activity versus ciliolocomotor transport rate. Download Table 3-1, DOCX file.
Pedal nerve stimulation promoted ciliolocomotor transport. A, Stimulating each pedal nerve individually (2.5 V; pulse duration, 4 ms; 15 Hz) enhanced ciliolocomotor transport. B, C, Ciliolocomotor rates before, during, and after stimulation. Stimulation of each nerve caused a significant increase in transport rate (Extended Data Table 4-2).
Table 4-2
Statistical summary of the effect of PN stimulation on ciliolocomotor transport rate. Download Table 4-2, DOCX file.
Stimulation of a given nerve (which had been severed from the ganglion) caused silicon carbide movement only in a portion of the foot. A, The anterior pedal nerve drove carbide movement only in the anterior portion of the foot. B, The medial portion of the foot did not immediately respond to medial PN stimulation, but carbide movement was significantly elevated in the 15 s following stimulation. C, The posterior PN induced carbide movement in both the medial and posterior portions of the foot, but is considerably larger than the other PNs, with a small branch that connects to the foot close to the connection point for the medial PN. Bar indicates period of stimulation. Statistical summary is shown in Extended Data Table 5-3.
Table 5-3
Statistical summary of the effect of PN stimulation on different parts of the foot. Download Table 5-3, DOCX file.
Avoidance turns, defensive withdrawals, and active biting behaviors suppress crawling
Induction of head withdrawal and avoidance turns by unilateral LOVN stimulation (3–5 V, 3–5 Hz, ∼10 s) significantly reduced ciliolocomotor transport in 13 of 14 trials in the hemipreparation (four subjects; F(2,39) = 11.85, p = 0.0002; one-way RM ANOVA with Tukey's post-test; Fig. 6A,B). The carbide transport rate decreased nearly threefold during stimulation (from 0.48 ± 0.076 mm/s down to 0.20 ± 0.034 mm/s; p = 0.0002), before recovering partially after stimulation ceased. An analogous similar effect was seen in the isolated CNS, in which stimulation of the LOVN caused reduction in the firing rate in each ipsilateral PN (Fig. 7, Extended Data Table 7-4). These data are consistent with the behavior of intact individuals performing stereotypical avoidance turns, where they cease forward movement, withdraw from a noxious stimulus and cease forward movement, perform the avoidance turn, and then crawl away (Jing and Gillette, 2003). These results implicate the defensive withdrawal component of the stereotypical avoidance turn in the inhibition of locomotion.
Avoidance turns inhibited crawling in the hemipreparation, while orienting turns had no effect on crawling in intact animals. A, Ciliolocomotor transport over time during an avoidance turn; bar indicates the period of LOVN stimulation. B, Stimulation of the LOVN significantly reduced the rate of transport, which then partially recovered after stimulation ceased. C, Ciliolocomotor transport during an approach turn, which was induced by application of shrimp to the oral veil in intact animals. The triangle is the time of shrimp application to the oral veil, and the bar indicates the period in which individuals were turning. Turns lasted 5–19 s (mean, 8.8 ± 1.4 s); for plotting of average rates on the line graph, the turn of each trial was compressed into a 5 s period. D, Transport rate did not change during the turn.
Induction of fictive avoidance turns suppressed PN activity. A, Representative recording from the ipsilateral pedal nerves and plBWN during LOVN stimulation (indicated by bar). A fictive turn was induced by LOVN stimulation; when stimulation ceased, the firing rates in the pedal nerves fell below the prestimulatory baseline. The lack of stimulus artifacts is the result of the LOVN stimulation site being centimeters distant from the pedal nerve recordings. B, LOVN stimulation significantly attenuated firing in all three nerves; firing returned to baseline for the aPN and mPN but remained depressed for the pPN. *p < 0.05; **p < 0.01; ***p < 0.001 (Extended Data Table 7-4).
Table 7-4
Statistical summary of the effect of LOVN stimulation on PN activity. Download Table 7-4, DOCX file.
In semi-intact preparations, inducing fictive defensive withdrawal by BWN stimulation (5 V, 15 Hz, ∼2 s) also suppressed ciliolocomotor transport in 12 of 14 trials (three preparations; F(2,36) = 17.32, p < 0.001; Friedman test with Dunn's multiple-comparisons post-test; Fig. 8). Carbide movement rate quickly decreased from 0.45 ± 0.11 mm/s down to 0.12 ± 0.056 mm/s (p = 0.005) in the first 10 s after stimulation and remained depressed for the following 10 s (0.15 ± 0.097 mm/s; p = 0.0003).
Induction of defensive withdrawal via BWN stimulation inhibited crawling. A, Rate of ciliolocomotor transport over time. The bar indicates the period of BWN stimulation. B, Average firing rate before, 0–10 s after, and 10–20 s after stimulation. Stimulation caused a marked decrease in transport rate, which had not recovered even 20 s after stimulation. *p < 0.05; **p < 0.01.
Finally, active feeding behavior suppressed ciliolocomotor transport in intact animals (F(4,20) = 3.97, p = 0.011; one-way RM ANOVA with Tukey's post-tests; Fig. 9). When presented with an appetitive stimulus of shrimp, individuals extended their probosces and began probing but did not significantly change their crawling rate (baseline, 1.24 ± 0.13 mm/s; vs with the proboscis extended, 0.76 ± 0.23 mm/s; p = 0.28). However, when active biting occurred, the crawling rate was significantly decreased relative to baseline (0.46 ± 0.11 mm/s, p = 0.019), remained decreased for the first 5 s after biting ceased (0.44 ± 0.11 mm/s, p = 0.015), and resumed shortly thereafter.
Active feeding behavior slowed ciliary beating. When presented with an appetitive shrimp stimulus (triangle), subjects (n = 6) began to probe with their proboscis. Ciliolocomotor transport did not differ significantly from baseline in this time but slowed significantly during the act of biting itself, before returning to baseline. Asterisks indicate periods in which ciliolocomotor transport was significantly reduced compared with baseline. *p < 0.05; **p < 01.
Neither approach turns nor escape swimming suppress crawling
In contrast to avoidance turns, approach turning did not reduce crawling. Approach turns were elicited in intact animals by applying shrimp to the oral veil (Fig. 6C,D) and caused no significant change in carbide movement at any point in the experiment (three subjects, 12 replicates; baseline, 1.75 ± 0.45 mm/s; during turn, 1.59 ± 0.27 mm/s; after turn, 1.40 ± 0.27 mm/s; F(2,39) = 0.14, p = 0.93; Friedman test).
Similarly, swimming had no detectable effect on ciliary beating. The ciliolocomotor transport rate before (0.84 ± 0.22 mm/s) and during (2.02 ± 0.89 mm/s) a swim was not significantly different (n = 5; t(9) = −1.44, p = 0.18; paired t test), with three of five individuals exhibiting increased transport during the swim.
Serotonergic As2/3 cells induce ciliary beating and are inhibited during feeding
The As2/3 neurons induced ciliary beating in semi-intact preparations, reproducing in Pleurobranchaea the previously reported role of the homologous neurons in Tritonia (Popescu and Frost, 2002). Driving a single As2/3 cell at 14–30 Hz caused a significant increase in ciliolocomotor transport [n = 13 (3 subjects); F(2,36) = 10.24, p = 0.0013; one-way RM ANOVA with Tukey's post-tests; Fig. 10]. The transport rate rose from 0.049 ± 0.0087 to 0.30 ± 0.067 mm/s (p = 0.0068), before returning toward baseline (0.16 ± 0.053 mm/s; p = 0.13). A similar effect was seen in isolated CNS preparations; stimulation of single As cells at 2–6 Hz caused a significant increase in activity in the contralateral aPN, contralateral mPN, and ipsilateral mPN (Fig. 11, Extended Data Table 11-5).
As2/3 stimulation enhanced ciliary beating. A, Rate of ciliolocomotor transport over time (bar indicates the period of As cell stimulation). B, The rate of transport increased significantly during the As cell stimulation. *p < 0.002.
As cells drove activity in the contralateral aPN and both mPNs. A1,2, Mean firing rate from each PN over time as a function of As cell stimulation (bar). In the aPN and ipsilateral mPN, the mean firing rate significantly increased during stimulation, then returned to baseline following stimulation. In the contralateral mPN, the mean firing rate significantly increased during stimulation, then dropped significantly below baseline afterward (Extended Data Table 11-5, statistical summary). B, Representative recording from both mPNs. Note that driving the As cell induced a negligible change in plBWN activity, indicating that the As cell activity, although sufficient to cause an increase in pedal nerve activity, was insufficient to cause a turn. A 2.5 Hz high-pass filter was applied to the As cell record to compensate for the inability of the bridge circuit to balance the driving electrode. *p < 0.05; ***p < 0.001 (Extended Data Table 11-5).
Table 11-5
Statistical summary of the effect of As cell stimulation on PN activity. Download Table 11-5, DOCX file.
Fictive feeding motor output inhibited As2/3 in the isolated CNS in the context of stimulus-induced turning (Fig. 12). In seven trials from five different preparations, fictive feeding induced by stimulation of the stomatogastric nerve and observed as rhythmic bursting in buccal root nerves correlated with significant decreases in spontaneous firing from As2/3 (before feeding, 1.21 ± 0.24 Hz; vs during feeding, 0.73 ± 0.19 Hz; t(10) = 3.10, p = 0.021; paired t test).
Fictive feeding inhibited As2/3 in the isolated nervous system. A, The onset of robust feeding bursts (observed in the buccal R1 and R3 nerves) coincided with the suppression of spontaneous activity in the As2/3 neuron. B, Average firing rates before and during fictive feeding. *p < 0.05.
Discussion
These results add to the known neuronal circuitry underlying behavioral transitions in the foraging decisions of Pleurobranchaea. Previous work showed that motivational state and learning regulated the decision to approach or avoid appetitive and noxious stimuli, acting via the excitation state of the feeding motor network (Davis and Gillette, 1978; London and Gillette, 1986), which switched turn response directions between approach and avoidance (Gillette et al., 2000; Hirayama and Gillette, 2012; Brown, 2014; Hirayama et al., 2014). Absent until now have been the neuronal mechanisms for integrating crawling with other forms of locomotion, and how locomotion is coordinated with approach–avoidance decisions in foraging. Here, we found these mechanisms and showed the following: (1) crawling motor signals are conveyed to the foot via three bilateral pedal nerves; (2) crawling can be induced by the serotonergic As2/3 cells; (3) defensive withdrawal, which precedes avoidance turning, and feeding inhibit crawling; (4) fictive feeding inhibits As cell activity; and (5) approach turning, prebite proboscis extension, and escape swimming exert no significant inhibitory effect on ciliary activity/crawling.
These data, taken with previous observations in Pleurobranchaea, inform the construction of a model of the coordination of locomotion with appetitive and aversive behavior (Fig. 13). The promotion or suppression of crawling depends on motivational state, context, and coordination in ongoing goal-directed behaviors. As in many animals, appetitive stimuli can induce locomotion in quiescent Pleurobranchaea, and hungry animals perform approach turns precisely directed at the estimated likely stimulus source direction (Yafremava et al., 2007). On encountering a stronger appetitive stimulus like potential prey, animals stop locomotion and begin active biting. In contrast, stereotypical avoidance turns are induced by noxious stimuli, such as contact with a larger cannibal conspecific, an electric shock, or acidic seawater. Such stereotypical avoidance is also induced in both satiated and food avoidance-trained animals by food stimuli (Gillette et al., 2000; Noboa and Gillette, 2013). During these turns, crawling locomotion is transiently suppressed, preventing the animal from moving closer to the aversive stimulus. Crawling and pedal ciliary beating cease during the defensive withdrawal phase of the avoidance turn, and during general defensive withdrawal, in which the animal recoils but does not turn. Ciliary beating is not suppressed during escape swimming, which may favor the animal's immediate period of locomotion that happens on swim cessation. Active feeding also suppresses crawling/ciliary beating, but approach turns and pre-bite appetitive behaviors do not.
A central position for the As neurons in coordinating the posture and locomotion elements of foraging behavior. Chemotactile stimuli target both the feeding premotor network and the multifunctional A-cluster containing the As neurons. Chemotactile stimuli can induce turns for avoidance or approach, locomotion, or escape swimming depending on the stimuli qualities, the hunger state of the animal, and learning experience, and set the excitation state of the feeding network between subthreshold and active consummatory biting. Blue indicates contributions from this study to the existing model in Pleurobranchaea.
Regulation of locomotion and posture centers on the serotonergic As cells. These key elements in the multi-functional A-cluster network lie in a central position of an extended network of serotonergic neurons that mediates general arousal of motor activity and appetite (Jing and Gillette, 2000). The As cells were first shown to provide neuromodulatory excitation to the swimming central pattern generator (Jing and Gillette, 1999), as also for the homologous B/C neurons in Tritonia (McClellan et al., 1994), and were subsequently shown to drive both avoidance and approach turns when asymmetrically active (Jing and Gillette, 2003). Jing and Gillette (2000) showed that they directly excite pedal serotonergic neurons presumed to drive ciliary beating, and here we confirmed their role in ciliolocomotion. The As2/3 cells are inhibited during active feeding, concomitant with crawling suppression. However, the As cells may not be the only important inputs to the locomotor neurons, and we speculate that locomotor suppression may involve not only the inhibition of excitatory drive in As2/3, but also active inhibition of the pedal ciliary cells of the foot by as-yet-unidentified neurons.
Crawling coordinates differently with other behaviors in other gastropods, where function differs with foraging strategy. The pond snail Planorbarius corneus reduces locomotion when biting at solid food but does not slow down when grazing on algae (Deliagina and Orlovsky, 1990), and the nudibranch Melibe leonina, a filter feeder, feeds readily while either crawling or stationary (Newcomb et al., 2014). This contrasts with Pleurobranchaea, which, when biting, suppresses crawling, thereby avoiding overrunning the prey. Thus, the actions of the As cell homologs in organizing locomotion may differ with feeding and other ecological strategies across the species.
These results fill out the role of the A-cluster neurons of the cerebral lobe of Pleurobranchaea as a multifunctional premotor neuronal network mediating postural and locomotive behaviors, and motor arousal (Jing and Gillette, 1995, 1999, 2000, 2003). The As2/3 neurons act as normal drivers of crawling locomotion in Pleurobranchaea, as shown earlier for their homologs in Tritonia (Popescu and Frost, 2002) and Aplysia (Jing et al., 2008). Notably, the Aplysia lineage characteristically locomotes with muscular pedal waves instead of ciliolocomotion and likely diverged from the lineage of Pleurobranchaea >300 million years ago. This is a salient example of how basic function can be conserved in the nervous system with marked evolutionary changes in how it is accomplished in the periphery (see Dickinson, 1979; Jing et al., 2009).
These findings strengthen analogies between the As interneurons and the serotonergic raphe nuclei of the vertebrate reticular system (Fig. 14), whose innervation of spinal motor circuitry is essential to control of posture and locomotion in trunk and appendages (Hornung, 2003; Pearlstein et al., 2011; Liang et al., 2015; Grillner and El Manira, 2020). As in the vertebrate reticular system (Dorocic et al., 2014), the A-cluster lies synaptically downstream of the areas that integrate sensation, motivation, and memory for cost–benefit decisions, notably the oral veil and feeding network (Hirayama et al., 2012), with similar relationships to upstream areas that integrate homeostatic state with memory to direct action selection. In gastropods, serotonergic elements are embedded in the various central pattern-generating circuits for different behaviors in which they provide neuromodulatory excitation to locomotor networks, increasing arousal (Kupfermann and Weiss, 1982; Katz and Frost, 1995; Panchin et al., 1995; Satterlie and Norekian, 1995; Popescu and Frost, 2002; Jing and Gillette, 2003; Marinesco et al., 2004). Additionally, they are electrically and chemically coupled in a distributed serotonergic network, not only providing neuromodulatory excitation to the specific CPGs in which they are located, but also through the network coupling increasing general excitation and thereby establishing a common arousal state across behavioral circuits (Kupfermann and Weiss, 1982; Jing and Gillette, 2003; Gillette, 2006; Jing et al., 2009; Hirayama and Gillette, 2012; Dyakonova et al., 2015). Thus, activity in one circuit can increase arousal in another, priming the latter to activate and facilitate the coordinated expressions of multiple behaviors. For example, as Pleurobranchaea crawls during foraging, As cell activity weakly excites the feeding network (Jing and Gillette, 2000), preparing it to respond to prey. Reciprocally, hunger increases serotonergic activity within the feeding network (Hatcher et al., 2008), which may in turn promote As cell activity and so prepare the animal to crawl in search of food.
Functional analogies for the modulatory and premotor functions of the vertebrate reticular system and the gastropod (Pleurobranchaea) A-cluster neurons. Both receive decisive inputs for action selection based on stimuli qualities, motivational state, and learning experience. Through specifically configured outputs, both control the coordination of downstream patterning motor systems for locomotion and dynamic posture. Both systems promote the arousal states of the downstream motor pattern generators through serotonergic neuromodulation.
Like the A-cluster, the reticular formation contains a serotonergic network (the raphe nuclei) and premotor networks that increase behavioral arousal (Sakai and Crochet, 2001) and drive the motor effector networks (Takakusaki et al., 2016) through neuromodulatory actions. In the vertebrates these downstream effector networks are the pattern-generating systems of the segmented spinal cord, and in the unsegmented gastropods these are the networks of the pedal ganglia for locomotion and turning, and probably righting responses as well (Davis and Mpitsos, 1971). In summary, these observations suggest that the general design for control of locomotion, posture, and motor arousal by the reticular formation might well have preceded the evolution of segmented bodies and articulated appendages.
Footnotes
This work was supported by Office of Naval Research Grant N00014-19-1-2373.
The authors declare no competing financial interests.
- Correspondence should be addressed to Rhanor Gillette at rhanor{at}illinois.edu