Structure and Function of Pedal Neurons Controlling Muscle Contractions in Tritonia Diomedea

Page 1 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 Structure and function of pedal neurons controlling muscle contractions in Tritonia diomedea

Reakweeda Zazay1,2,3, Josh B. Morrison2,4, Roger Redondo2,4, James Alan Murray1,2,4 1California State University East Bay, Hayward CA 94542; 2Friday Harbor Laboratories, Friday Harbor WA 98250; 3George Washington University, Washington DC 20052; 4University of Cen- tral Arkansas, Conway AR 72035

There are 16 pairs of “identified neurons” in the pedal ganglion of any sea slug of the species Tritonia diomedea that have published behavioral functions. Many of the pedal neurons cause flexion of the ipsilateral body wall and foot when activated, but they are not thought to innervate muscle directly. The goal of this study was to examine the motor functions of brain neurons with no previously-identified functions. We described the activity of two such cells and their motor effects, and further characterized the motor effects of a previously-identified neuron (Pedal 3). We stimulated the Pedal 3 flexion neuron and characterized where and how it contracted the foot. For each neuron, we described the latency to contraction, the time to relaxation, and the distance and speed of movement. These neurons may be involved in turning during crawling, and these results will help us understand how the activity of specific neurons is translated into behavior (neuromechanics), and determine how fast the animal can respond to sensory feedback during locomotion. The relative simplicity of this brain allows us to understand how behavior is generated on a cellular basis, and to generate neural network and neuromechanical models of navigation that can be applied to robotics.

Abbreviations: SFA – spike frequency adaptation; Pd – Pedal neuron (identified neuron); PdN – pedal nerve (identified nerve) Keywords: motor control; turning; identified neuron; neuroethology

Introduction tion of muscles that may change the amount The marine gastropod Tritonia dio- of cilia contacting substrate on a given side medea (Figure 1A) is a model organism for of the animal (Murray et al., 2006). The neu- studying the neural control of behavior, as it rons that elicit movement are referred to as has giant, re-identifiable brain cells (neu- “flexion neurons” because they do not con- rons) that have been extensively studied for tact muscle directly, as would a true motor more than forty years (Willows, 1967). The neuron (Hoyle and Willows, 1973). Many of slug has a relatively simple and limited the larger neurons have been identified visu- amount of behavioral responses to environ- ally, physiologically, functionally, and are mental stimuli, such as predator, prey and morphologically mapped in the brain (Wil- conspecific odors, water flow and the geo- lows et al., 1973a; Hume et al., 1982; Mur- magnetic field (Willows, 1999, Wyeth et al., ray et al., 1992; Cain et al., 2006; Neuron- 2006; Murray et al. 2011). T. diomedea have Bank). Willows et al. (1973) published a a cilia-covered foot, and crawl by beating four-paper series describing structure and these cilia against the substrate; it is sus- function of the brain, nerves, and neurons of pected that no muscles are involved in this T. diomedea, and categorizing several “iden- process (Audesirk, 1978). However, we hy- tified neurons.” They identified 11 neurons pothesize that turning involves the contrac- in the pedal ganglion, and Page 2 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 1: Photographs and diagrams of the sea slug Tritonia diomedea, and its brain. A. T. diomedea is a nudibranch gastropod that crawls using ciliary beating and uses asymmetrical contractions of the muscular foot to turn during crawling. B. Experimental setup (Drawing from Willows AOD et al., The neuronal basis of behavior in Tritonia I. functional organization of the central nervous system. Journal of Neurobiology 4: 204-237, 1973. Used with permission, License Number 2707211025474). C. Dorsal view of the brain of Tritonia showing fused cerebral, pleural, and pedal ganglia, comprising ~7000 neurons. Each of the orange circles is the soma of a single brain cell, and the white nerves radiate out to all part of the body. D. The brain has six ganglia, and the left and right pedal ganglia contain ~1000 neurons each and most of them send axons out of Pedal Nerves (PdN 1-4) to relay motor commands to the foot muscle, cilia, and other targets.

determined the motor functions of three (Pe- (Pd21 by Audesirk 1978; Pd5 by Popescu dal neurons 1, 3, and 8; a.k.a. Pd1, Pd3, and Willows, 1999), and the remaining two Pd8). Hoyle and Willows (1973) noted that (Pd6, Pd7) are presumed to be ciliary motor these pedal neurons likely do not directly neurons (Wang et al., 2003, 2004; Cain et innervate muscle fibers, but instead work al., 2006), but definitive evidence is not yet through a peripheral nerve net that includes published. local motor neurons, and this results in de- In some of these experiments we lays in contraction. were looking to identify un-mapped neurons Six of the others (e.g. Pd5, Pd6, Pd7, that were involved in movement of the slug. Pd20, Pd21, Pd22) elicited no obvious motor Identified neurons in the pedal ganglia con- effect, though two of those neurons were tract the ipsilateral side of the foot and body later determined to be ciliary motor neurons wall when active (Willows et al., 1973a), Page 3 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 contributing to turning (Redondo and Mur- Brain cell recording ray, 2005). Pedal neuron 3 (Pd3; Figure 2A) Figure 1C shows a picture of the is a previously-identified unipolar neuron brain underneath a microscope with the out- that sends an axon into Pedal Nerve 3 er sheath removed from all ganglia and the (PdN3), which projects to the foot and lat- epineurium intact. The orange circles are eral body wall (Figure 1). This neuron has single brain cells, some of which can be several neurites that extend into the neuropil identified visually in any individual (by col- of the ganglion, and are presumed to be or, size, and position in the ganglion). We dendritic inputs since no neurons have been identified the Pd3 neuron in five slugs by shown to receive synaptic input from this position in the ganglion, size, color, sponta- neuron in the isolated brain. In this study, neous activity and motor effect (Murray et we characterized the location, extent, and al., 1992), and we characterized two new speed of contractions caused by activation of cells in one slug each when the cells did not an identified neuron (Pd3), and two newly- match published criteria. We visually- identified neurons. guided sharp glass microelectrodes to pene- trate the epineurium above a haphazardly- Material and Methods selected neuron. We characterized only one neuron per pedal ganglion. We made elec- Animals & surgery trodes with a Sutter P-97 puller and thin- We collected T. diomedea via SCU- walled glass (A-M Systems, 1.0 OD, 0.75 BA near Dash Point, Federal Way, Wash- ID) to have a resistance of 5-20 MΩ when ington. We kept the slugs at the Friday Har- filled with potassium acetate. We used elec- bor Laboratories in sea tables that had flow- trodes with the tip backfilled with the nega- ing sea water. We used seven slugs in this tively-charged fluorescent tracer AlexaFluor study. We exposed the brain for recording hydrazide 488 or 568 nm (Invitrogen, Grand while it was intact and connected to the Island, NY), or Lucifer Yellow (Vector body (a “semi-intact” preparation, see Fig- Labs, Burlingame, CA), to impale a cell, and ure 1B) by its many nerves (Figure 1C, D). the resistance was then between 20-50 MΩ. We made a one-inch incision on the dorsal We first inhibited the cells from firing by body wall above the brain. We placed the injecting them with negative DC current. animal in an acrylic tank and kept open the Once the cell stopped firing, we injected it incision with non-magnetic hooks. We made with positive current via a bridge amplifier an incision in the next layer of connective (A-M Instruments, Sequim, WA, Neuro- tissue posterior to the brain, and inserted a probe model 1600) to induce and record its non-magnetic wax-covered platform ventral firing activity. The number of trials varied to the brain. We pinned the surrounding from 10-50 for each neuron. connective tissue onto the wax to immobi- lize the brain. We cut the connective tissue Behavioral recording & analysis above the pedal ganglia (“outer sheath”) to We simultaneously recorded video of expose the tight-fitting epineurium (“inner the resulting movements (spike data were sheath”), which is the last layer of connec- recorded with a PowerLab 26T, Colorado tive tissue above the nerve cells in the pedal Springs, CO), and video was recorded with a ganglia of the brain (Figure 1D). Panasonic firewire camcorder and synchro- nized using LabChart v.6 (ADInstruments, Colorado Springs, CO), displayed on a com- puter monitor and recorded to a hard drive. Page 4 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 Movement of the body was monitored in 0.1M PBS twice for 30 min intervals, then real-time with a photovoltaic sensor (MTI dehydrated with 50%, 70%, 90%, 95%, Fotonic, Albany, NY, model 1000) placed 1- 100%, 100%, and 100% EtOH for 30 min 3 cm from the skin. This device measured each. The brain was then put in methyl sa- movement by detecting decreases in light licylate to clear the tissue for microscopy, reflection as the slug moved away from the and mounted between two cover slips using probe and put out a voltage proportional to paper clips as spacers for the ~1 mm thick reflected brightness. The instrument was not brain. calibrated for distance or speed, but provid- ed a relative index of speed of movement. Results To determine the timing of the initial onset of movement post-stimulation, we calculat- A newly-identified neuron that contracts the ed the first derivative (slope) of this voltage antero-lateral foot muscles and noted when the slope first exceeded A neuron (Figure 2A, green) was background levels, and when the slope was identified that was not similar to any of the maximal during a movement. We also quan- previously-identified neurons (numbered tified the extent and speed of the movement circles). One axon extended out of the gan- by video frame analysis (30 frames per se- glion into left Pedal Nerve 4 (which inner- cond) using ImageJ (NIH, Bethesda, MD), vates the antero-lateral body wall, Willows and plotted it with the firing rate of the pedal et al., 1973). This neuron exhibited the typi- neuron. cal morphology of T. diomedea pedal neurons with a soma just under the epineurium, Brain cell labeling & imaging a single primary neurite, one axon extending We iontophoresed AlexaFluor or Lu- out to the pedal nerves, and several pre- cifer Yellow tracer with DC current or puls- sumed dendrites extending into the central es (1-2 nA at 50-90% duty cycle for 10-60 neuropil region of the ganglion. minutes) controlled by a Getting DS1 digital Hyperpolarization of this neuron revealed pulse generator. The tracers labeled the cell spontaneous depolarizing postsynaptic po- body, axon and dendrites, so we could view tentials (Figure 3). We induced the genera- it under a (vintage) confocal microscope tion of action potentials, and some action (Bio-Rad MRC 600 with YHS and BHS fil- potentials persisted after stimulation. Firing ter blocks, 10X and 20X objectives on a Ni- rates of pedal neurons rarely exceeded 20 kon Optiphot-2 epifluorescence inverted mi- Hz using this method. This neuron caused croscope, Hercules, CA). The positions of movement with a delay from the beginning the soma and the distribution of the major of induced spikes until the onset of move- dendrites visible through the confocal are ment (Figure 4). Over eight trials, spike characteristic of specific cells. This mor- rates in the first 0.5 s, in response to 4 nA of phology, along with physiological data from current, ranged from 12-16 Hz (which was a specific cell, will allow it to be identified near maximal observed firing rates). We again in another animal of the same species. measured the latency from the peak of the After the intracellular recordings are done, first spike to earliest evidence of movement we cut out the brain, cleaned off excess ex- as 0.84± 0.31 s (mean± standard deviation), ternal sheath tissue and prepared the brain range 0.53-1.43 s, and the maximal rate of for fluorescent microscopy by fixing in 4% movement was achieved at 1.93± 0.91 s paraformaldehyde in phosphate-buffered (range 1.1-3.8 s). The spike rate and the rate saline (PBS) overnight. We rinsed it with of movement showed high correlation. Page 5 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 3: Intracellular recording from a pedal ganglion neuron. This recording of intracellular voltage ver- sus time shows spontaneous 5-10 mV postsynaptic potentials while a Tritonia neuron was being hyperpolarized with 1-5 nA of negative current to prevent spontaneous action potentials. The ~100mV action potentials were induced by the injection of 2 nA of positive current (black bar) via a bridge circuit through the recording electrode, and some action potentials persist after stimulation. Slug ID 7-20-09, trial 7.

Figure 2: A newly described flexion neuron that controls the foot. A. This neuron (green) was not similar to any of the previously identified neurons (number circles). It caused contractions of the foot, and thus was considered a flexion neuron. It was injected with AlexaFluor hydrazide to show the structure of its dendritic field and the nerve through which it projects its axon. Note that the neuron is shown on a map of the right pedal ganglion for comparison, but it was actually found in the mirror-image left ganglion. B. Ventral view of the neuron that was iontophoretically injected with AlexaFluor hydrazide. The faint outline of the pedal ganglion is seen with the very bright soma, the unipolar neurite with many branches in the central Figure 4: Stimulated spikes cause movement with an neuropil of the ganglion, and the main axon extending ~1 s delay from the beginning of spikes until the onset out of the ganglion into pedal nerve 4. The image was of movement. The red line shows the spike rate in 1 created by flattening a stack of 65 confocal images second bins when stimulated with 4nA (first spike peak: 10.96 s, last spike peak: 14.28 s). The rate of totaling 325 microns of thickness. Slug ID 7-28-09. movement (blue) was measured with a light sensor that shows an uncalibrated time course of movement (points in the curve are separated by 0.25 s). The inset shows a closer view of time 8-20 s (after beginning of this trial) that shows the ~1 s delay between the first spike (dashed line) and the onset of foot movement, and the ~2 s delay before it reaches maximum rate of movement. Slug ID 7-28-09, trial 24 (movie 22).

Page 6 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 5: Path of foot movement during activation of a novel flexion neuron from Figure 2. A. A frame from

the video of a left side view of the slug showing the Figure 6: Three frames from video of an induced foot induced foot movement taken at the maximum dis- movement shown in Figs. 4 and 5 showing (A) the placement of the foot. The point indicated by the circle posture of the slug before stimulation, (B) the maxi- (○) shows the place on the white foot margin that we mum extent of the movement, and (C) the return of the

tracked in video analysis, and the blue line shows the foot after relaxation. See supplemental video 6S (slug- direction and amount of movement of the foot with an movie- 7-28-09-trial24; arrow each second. Note that the foot lifted dorsally http://www.youtube.com/watch?v=C -4v1DFr6BE). and posteriorly, then moved back ventrally and slightly Slug ID 7-28-09, trial 24 (movie 22).

anteriorly as it relaxed. B. The tracked point on the anterior foot margin moved from ventral to dorsal, and anterior to posterior, but did not return to the origin immediately (same data show in blue in Figure original position. The foot did not fully relax for over 1 minute after the stimulation, and 5A). The starting position is 0, 0. The position of the foot margin is plotted at 1 second intervals, for 82 se- did not return to its original position. Figure conds. Slug ID 7-28-09, trial 24 (movie 22). 7B illustrates the rate of stimulated foot

movement. The foot contracted rapidly, and Figure 5 shows the movement that relaxed ~2 times more slowly. resulted from the stimulation shown in Fig- ure 4. The foot lifted dorsally and posterior- A newly-identified neuron that contracts the ly, then moved back ventrally and slightly dorso-lateral muscles anteriorly as it relaxed (Figure 5A, B, Figure We identified a second neuron (or- 6). A supplemental time-lapse video 6S ange in Figure 8A) that was not similar to (slug-movie-7-28-09-trial24) of this move- any of the previously identified neurons. ment is available online. This flexion neuron in the left pedal gangli- Figure 7A shows the total distance of on caused contraction of the dorso-lateral the circled point in Figure 5A from its body wall. Page 7 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 7: Extent and speed of movement induced by Figure 8: A. This neuron (orange) was not similar to activation of novel neuron. A. Shows the total dis- any of the previously identified neurons (number cir- tance of the circled point in Figure 5A from its origi- cles). This flexion neuron was recorded and caused nal position. Note that the foot did not fully relax for contractions of the dorso-lateral body wall. It was over 1 minute after the stimulation, and did not return injected with AlexaFluor hydrazide to show the struc- to its original position. The black bar indicates the ture of its dendritic field and the nerve through which timing of the spikes during 4 nA stimulation. Spikes it projects its axon. Note that the neuron is shown on a started at 10.96 s and ended at 14.28 s. B. Rate of map of the right pedal ganglion, but it was actually foot movement measured from video. The blue points found in the mirror-image left ganglion. B. Ventral show the rate of movement of the foot over 1 second view of neuron that was iontophoretically injected intervals and the black line shows the 3-second mov- with AlexaFluor hydrazide (oriented to match Figure ing average (to smooth apparent motion). The black 8A). The faint outline of the pedal ganglion (arrows) bar indicates when 4 nA was injected into the neuron. of unfilled neurons is seen with the very bright soma The first peak represents the high rate of movement of the filled neuron, the unipolar neurite with many during the contraction of the foot muscles, and the branches in the central neuropil of the ganglion, and second peak represents the slower rate of movement the main axon extending out of the ganglion into Pedal during the return of the foot position during relaxa- Nerve 1. The main image was created by flattening a tion. Slug ID 7-28-09, trial 24 (movie 22). stack of 248 confocal images totaling 422 microns of thickness. The inset image was made by flattening a stack of 309 images totaling 422 microns of thickness. Slug ID 7-20-09. Figure 8B shows a ventral view of a neuron that was fluorescently-labeled (ori- extended out of the ganglion into pedal ented to match Figure 8A). The neuron was nerve 1 (which innervates the antero-lateral unipolar with many branches in the central body wall and the foot margin, Willows et neuropil of the ganglion, and the main axon al., 1973).

Page 8 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 9: The spike rate in 2 second bins when stimulated with 4nA (first spike peak: 17.96 s, last spike peak: 37.86 s). The recording of the spikes (voltage vs. time) is shown in the background. Slug ID 7-20-09, trial 23.

Figure 11: Three frames from video of an induced movement shown in Figure 10. See black symbol (○) tracking location on gill (A) Posture of the slug before stimulation, (B) the maximum extent of the movement, and (C) the return of the foot after relaxation. See supplemental video 11S (slug-movie-7-20-09-trial23; http://www.youtube.com/watch?v=TYJRbKIISok). Slug ID 7-20-09, trial 23.

This neuron fired action potentials at a much slower rate (maximum 3 Hz with 4nA stimulation; Figure 9) than the previous neuron (Figure 4). Figure 10 shows a video frame taken from anterior of the slug, showing the dorsal and left sides of the slug, and indicates the Figure 10: A. A frame from the video taken from movement of the gill margin of the body anterior of the slug, showing the dorsal and left sides during stimulation. The side of the body of the slug. The point indicated by the circle (○) shows the place on the body wall that we tracked in shifted medially, then moved back laterally video analysis, and the blue line shows direction and as it relaxed (Figure 10A, B). The tracked amount of movement with an arrow for each second. point on the dorsa-lateral body wall moved Note that the side of the body was shifted medially, then moved back laterally as it relaxed. B. The from lateral to medial, and ventral to dorsal, tracked point on the dorsa-lateral body wall moved but did not return to the origin (Figure 10B, from lateral to medial, and ventral to dorsal, but did 11, same data are in blue in Figure 10A). not return to the origin within 80 s. (same data show in blue in Figure 10A). The starting position is 0,0. The position of the body wall is plotted at 1 second intervals, for 84 seconds. Slug ID 7-20-09, trial 23.

Page 9 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 13: A fluorescent image of a Pedal 3 (Pd3) neuron injected with Lucifer Yellow. This unipolar neuron sent an axon into Pedal Nerve 3 (PdN3), which projected to the foot and lateral body wall. The neuron had several neurites that extended into the neuropil of the ganglion.

Figure 12: A. This shows the total distance of the circled point in Figure 5A from its original position. Note that the foot did not fully relax for over 1 minute after the stimulation, and did not return to its original position. The black bar indicates the timing of the spikes during 4 nA stimulation. Spikes started at 10.96 s and ended at 14.28 s. B. Rate of stimulated foot movement measured from video. The blue points show the rate of movement of the foot over 1 second intervals and the black line shows the 3-second moving average (to smooth the apparent motion). The black bar indicates when 4 nA was injected into the neuron. Spikes started at 10.96 s and ended at 14.28 s. The first peak represents the high rate of movement during the contraction of the foot muscles, and the second peak represents the slower rate of movement during the return of the foot position during relaxation. Slug ID 7-20-09, trial 23 (movie 22). Figure 14: Spiking in Pd3 induced by current injection. The upper trace (red) shows spikes after they have been AC-coupled to remove the DC artifact due to an Figure 12A shows the total distance imbalanced bridge circuit. The inset shows the size and of the circled point in Figure 10 and 11 from shape of the persistent excitatory post-synaptic poten- its original position. Note that the movement tials. The lower trace (green) shows the SFA. Spike frequency dropped from a peak of 12 Hz to 8.5 Hz over begins after a latency of 1-2 s. Figure 12B the first 20 seconds, and the neuron began to burst after shows the rate of movement of the body. that. (This 5.8 nA stimulus is also the purple trace in The peak at 5 s was due to spontaneous ac- Figure 16D). tivity in a neuron we were not recording from, thus other neurons also innervated this region of the body. Page 10 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 Detailed characterization of the relationship between activity in Pedal Neuron 3 and foot contraction Pd3 elicited muscle contractions with the same 1-3 s latencies (see Redondo and Murray, 2005) as seen in the previous neurons described here. We characterized the firing rates of Pd3, its spike frequency adaptation (SFA) (a.k.a. accommodation), and the location, extent, and speed of body wall contractions while stimulating this flexion neuron with injected current. Figure 14 shows the stimulated spike Figure 15: Activation of Pd3 caused the mid-posterior rate. The inset shows 2-4 mV high, 200 ms foot margin to lift dorsally, separating foot from floor. long persistent excitatory post-synaptic po- These four video frames (A,B,C,D) show peak muscle tentials that occur ~2-4 per second. This flexion during intracellular current injections of 1, 2 , 3, and 4 nA, respectively, into the right pedal neuron Pd3. neuron exhibited SFA, which is a reduction Anterior was to the right and dorsal was upwards. in spike frequency despite constant depolar- ization. Spike frequency decreased 30% The speed of the foot lift increased when over 20 seconds, and the neuron began to spike rate in Pd3 increased, but reached a burst after (i.e., firing at an elevated rate in maximum around 4 mm/s at 5.8 nA or above between silent phases). Interburst interval (spike rates 10-12 Hz or above) about3 s af- started at 1.5 s and increased to 4.5 s over 10 ter the initial spikes were induced (Figure bursts. This bursting pattern was not evident 16C). We also stimulated Pd3 for 60 s to at lower firing rates induced by lower levels approximate the duration of activation need- of injected current (4.85 nA and lower), but ed to maintain foot lift for the time neces- the SFA was evident at the lowest current sary for a turn during crawling. Spike rates shown to elicit spikes (0.97 nA). in Pd3 during various 60 s long current in- Activation of Pd3 caused the mid- jections were higher for higher current, posterior foot margin to lift dorsally (Figure reaching up to 12 Hz, but all stimuli resulted 15), separating the foot from the floor. Larg- in SFA decreasing to a spike rate of 1-2 Hz er current injections into Pd3 resulted in after 60 s (Figure 16D). larger and faster motion of the foot (Figure The foot margin lifted vertically up 16). Spike rates were higher in Pd3 during to 33 mm when Pd3 was activated, despite 10 s long current injections as current in- large amounts of SFA (Figure 16E). After creased from ~1 to ~10 nA (Figure 16A). the lift of the foot, the foot margin slowly SFA varied from 20-30% over the 10 s in- lowered and reached its original height 50- terval. We stimulated for 10 s to approxi- 70 s after the end of stimulation. Higher cur- mate the duration of bursts observed during rent injections resulted in more lift, but the escape swimming motor pattern. reached a maximum lift with 5.8 nA of cur- The foot margin lifted vertically up rent ~30 s after the beginning of stimulation to 17 mm when Pd3 was activated, starting (Figure 16E). The speed of the foot lift in- ~1 s after the first spikes. Higher current increased when spike rate in Pd3 increased jections resulted in more lift, but reached a (Figure 16F), but reached a maximum speed maximum with 5.8 nA of current ~10 s after around 3 mm/s at 4.8 nA or above (initial the beginning of stimulation (Figure 16B). spike rates 8 Hz or above; Figure 16D). Page 11 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012

Figure 16: Larger current injections into Pd3 resulted in larger and faster motion of the foot. A. Spike rates in Pd3 during various 10 s long current injections. B. Vertical lift of the foot margin increased when spike rate in Pd3 increased, but reached a maximum. C. The speed of the foot lift increased when spike rate in Pd3 increased, but reached a maximum. D. Spike rates in Pd3 during various 60 s long current injections. Note spike-frequency adaptation (the purple trace is the same as the green trace in Figure. 14). D, E, and F had identical stimuli to A, B, and C except stimuli lasted 60 s instead of 10 s.

Discussion

The two neurons we characterized rate. Despite the differences in firing rates, for the first time here show significant func- they each produced similar amounts of body tional differences despite similar morpholo- wall contraction—presumably because of gy. Both neurons are similar in soma diam- compensatory differences in the peripheral eter, both have one axon and several den- motor neurons that these “flexion neurons” drites, but their axons extend out through synapse upon (Hoyle and Willows, 1973). different nerves. The neuron “7-20-09” con- Neuron “7-20-09” lifted the foot off trols the foot and has a higher maximal fir- the floor in a manner resembling that ob- ing rate, while neuron “7-28-09” controls served during turning while crawling. This the dorsum and has a lower maximum firing neuron may contribute to postural changes Page 12 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 made during turning (Murray et al., 2006), trolled by multiple central neurons. Thus, and we will test its role in turning by selec- the role of single neurons may be difficult or tively inactivating it in a crawling slug in impossible to determine, except in the cases future. The dorso-lateral contraction caused of neurons with extremely large motor by Neuron “7-28-09” may contribute to fields. It will be useful to monitor the activi- body curvature during turning (Redondo and ty of multiple flexion neurons during spon- Murray, 2005) and we plan to test if it is ac- taneous movements to determine how many tive during turning by using fine-wire re- are active during normal movements. This cordings of the neuron during crawling. can be done using fine wires glued over the The stimulation of Pd3 illustrated a pedal ganglion to pick up multiple field po- significant phase delay in body wall contrac- tentials (Redondo and Murray, 2005). tion following spiking. We stimulated for Willows and colleagues (1973a,b) 10 s to simulate the activity of Pd3 during its also noted that many of these pedal neurons bursting activity during the escape swim- fired bursts in phase with a network oscilla- ming motor program (Willows et al., tor central pattern generator that causes es- 1973b), and for 60 s to simulate extended cape swimming, and that Pd1 and Pd3 fire increased activity observed during turning bursts during the dorsal contraction phase of (Redondo and Murray, 2005). Moderate the swim. They noted that these neurons are SFA was observed over 10 s, but adaptation likely used in other behaviors such as crawl- was more pronounced over 60 s, especially ing and turning, which was confirmed by for larger current injections. The muscles Popescu and Frost, 2002. They showed that remained contracted long after spike rates when one Pd3 was induced to fire it pro- have fallen, thus, the muscles or peripheral duced a lateral flexion of the body that re- nervous system dynamics must be accounted sembled turning seen during crawling, but for when modeling how neurons affect the since the animal was restrained, they could mechanics of locomotion. The data collect- not prove that it contributed to turning. ed here will help inform a neuromechanical They determined the motor effects of several model of locomotion to help understand how neurons whose identity are not consistently neural activity translates into adaptive loco- determined across individuals, and described motion (Nishikawa et al., 2007) some limited somatotopic mapping of re- Hoyle and Willows (1973) noted that gions of the pedal ganglion to regions of the shortest latencies to contraction (includ- motor effects in the body. Willows et al. ing the delay caused by the propagation of (1973a) noted the inherent difficulty in de- the action potential to the periphery) were scribing the location and extent of move- ~500 ms when a single cell was stimulated ment caused by each identified pedal neu- at ~30 Hz. This was consistent with the ron. Our method of analyzing video frames shortest delay we observed being 530 ms. In correlated with spike pattern, and with a many cases, the short bursts of action poten- photosensor showing time course, is one tials that are generated naturally in these way to communicate and quantify the slugs are finished before the muscle begins movements caused by neurons in soft- to contract, so sensory feedback is not pos- bodied organisms with hydro-skeletons. sible except for subsequent motor com- It is not yet clear if the pedal flexion mands. neurons are each used in crawling, swim- The motor fields of these flexion ming, or both behaviors. Hume and col- neurons overlap considerably (Murray et al., leagues (1982) described five newly- 1992), so any part of the body can be con- identified neurons (Pd9, Pd10, Pd11, Pd12, Page 13 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 Pd13) in the pedal ganglia and determined in sal flexion phase of the swim motor pattern which phase (dorsal or ventral) of the escape (designated a DFN-A—Dorsal Flexion Neu- swim eight identified pedal neurons were ron Type A, Hume et al., 1982), and Pd13 is active (Pd1, Pd3, Pd8, Pd9, Pd10, Pd11, active during the ventral flexion phase of the Pd12, Pd13), but only one neuron (Pd8) was swim motor pattern (designated a VFN— shown to contribute significantly to the Ventral Flexion Neuron, Hume et al., 1982). swim flexions. Pd8 was shown to reduce The putative ciliary motor neurons ventral flexion amplitude and speed of the Pd5, Pd6, and Pd7 all contain a group of body by 6 and 14% respectively when it was peptides that have been shown to excite cilia hyperpolarized to prevent spiking. Hume et in the foot and gut of T. diomedea (Willows al. (1982) concluded that the whole body et al., 1997; Gaston, 1998; Wang et al., flexions were the combined result of many 2003, 2004; Cain et al., 2006; Baltzley and neurons acting in concert. They did not in- Lohmann, 2008). Popescu and Willows vestigate the functions of neurons outside of (1999) showed that activity in Pd5 causes the escape swim motor pattern, but it is pos- increased rates of ciliary transport on the sible that some of those pedal neurons affect foot. crawling. Redondo and Murray (2005) focused Our laboratory previously inves- on the function of the pair of flexion neurons tigated some of the interactions between the designated Pd3. This bilateral pair of neu- identified pedal neurons and how they re- rons produced a dorsal lift of one-third to sponded to sensory input from attractive wa- one-half of the ipsilateral foot margin that ter flow stimuli (Murray et al.,1992). They resembles those often seen during crawling showed that most of the identified neurons while turning (Murray et al., 2006; Figure respond to flow stimuli to the oral veil, and 15). They showed that increased activity of that stimulating some of the smaller identi- Pd3 on one side was correlated with turning fied neurons produced synaptic responses in towards the ipsilateral side. They also Pd5, Pd6, Pd7 that were likely to be based showed that both Pd3 neurons fired sponta- on peripheral feedback loops. They meas- neously during crawling in synchronous ured the motor effects of each of the so- bursts of 10-20 s, with a period of ~40 s, and called “flexion neurons”, each of which pro- firing rates alternating between 0 Hz and a duced movement or deformation of the body peak of 1-6 Hz. These spontaneous rates of wall or gills (Pd1, Pd2, Pd3, Pd4, Pd8, Pd9, firing are comparable to some of those we Pd10, Pd12, Pd14). Pedal 2 was shown to induced by current injection in the present withdraw gill tufts. Pedal 14 was newly study. The quiet periods in-between bursts identified, and produced a curl of the ipsilat- may allow the neuron to avoid SFA ob- eral foot margin. These motor effects were served in some pedal neurons (Murray et al., documented with a single drawing showing 1992; Katz and Frost, 1997), and/or to avoid the location of the white foot margin before synaptic depression between the pedal flex- and at the peak of movement. ion neurons and the peripheral motor neu- Pd5, Pd6, Pd7, Pd11, Pd13, Pd20, rons or between the motor neurons and the Pd21, Pd22 produced no obvious motor ef- muscle cells (Byrne, 1982). Redondo and fect, and this was consistent with the known Murray (2005) also cross-correlated the role of Pd5, Pd6, Pd7, and Pd21 as ciliary bursts of activity in Pd3 (avg. rates 1-2 Hz) motor neurons. The motor functions of with a measure of the degree of foot lift and Pd11, Pd13, Pd20, and Pd22 are unknown at found that they were significantly correlated this time, but Pd11 is active during the dor- with a 5-15 s delay. These delays are con- Page 14 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 siderably longer than those we recorded more than 60 s to return to normal posture when inducing initial firing rates of 10-15 after stimulation of spikes at rates more than Hz, which is expected based on the lower 1 Hz. The motor neurons of the peripheral firing rates observed during natural crawl- nerve net (the presence of which was hy- ing. Stimulation of the Pd3 neuron on one pothesized by Hoyle and Willows, 1973) side of the brain was sufficient to force the may maintain this body wall activation, or animal to turn towards that side, so this neu- the muscles themselves may be slow to ron may play an important role in effecting change. Despite SFA of Pd3 by ~50% in the turning. When the Pd3 neuron is inactivated first 30 s of stimulation, the foot margin on both sides, the animal crawls straight and would continue to lift higher until 30-60 s does not turn as it does before Pd3 was inac- after the beginning of stimulation. Thus tivated (Murray et al., 2006). muscle contraction is greatly affected by the Peak burst firing rates during crawl- recent history of pedal neuron activation, ing are much lower than those observed dur- even when those pedal neurons are currently ing escape swimming, which usually range inactive. from 10-20 Hz, but can approach 30 Hz in Hume et al. (1982) estimated that the some neurons. Burst periods during the pedal ganglion contained a maximum of 50 swim range from 6-9 s, more than ten times dorsal and 13 ventral flexion neurons (DFNs faster than the burst period in Pd3 during and VFNs) that burst during the escape crawling. Thus, injections of current above swim motor pattern. Brown et al. (2008) 2 nA induce firing rates above 4 Hz that are observed up to 62 DFNs and 22 VFNs using faster than those observed in Pd3 during optical recording techniques. But the pedal turning while crawling (Redondo and Mur- ganglion contains 200-500 neurons, approx- ray, 2005). At currents above 4 nA, we ob- imately half of which do not burst during the served bursting in Pd3 in the absence of pe- escape swimming motor pattern (Brown et riodic synaptic drive (as when in escape al., 2008). We plan to use fine-wire field swimming motor pattern). Thus, we con- potential recordings of the pedal ganglion to clude that this bursting is due to intrinsic determine how many of these neurons may properties of Pd3 and not due to synaptic change their firing rates during turning, and input. As a flexion neuron, Pd3 sends its to continue to identify pedal neurons that only axon out the nerve and has no known contract the foot and may contribute to turn- synaptic targets other than the body wall ing. It is possible that some of these neurons (peripheral nerve net; Hoyle and Willows, may function during more than one behavior 1973). One potential mechanism for gener- (e.g., crawling and swimming; Popescu and ating bursting in T. diomedea neurons was Frost 2002). described by Thompson et al., 1986. They The Tritonia model system is partic- described a potassium conductance that is ularly useful because there are relatively responsible for SFA, as well as hyperpolar- fewer neurons than in vertebrates, making it izing the cell during intervals between easier to determine the roles of individual bursts. We can test for the role of potassium neurons in behavior. Our long-term goal is by using voltage clamp and varying ionic to discover how neural networks control composition near the cell using microperfu- guided behavior such as navigation, and to sion (Thompson et al., 1986). generate general principles of information We noted that the body wall of T. di- processing that could be used to program omedea is both slow to respond to activation autonomous robots. of pedal neurons, and slow to relax, taking Page 15 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 Acknowledgements withdrawal reflex in Aplysia californica, J Neurophys 48:431-438. We wish to thank the California State University Cain SD, Wang JH, Lohmann KJ (2006) Immu- East Bay for research equipment and support. nochemical and electrophysio-logical The Friday Harbor Laboratories and their donors analyses of magnetically responsive neu- also supported student research over several rons in the mollusc Tritonia diomedea, J summers. This research was supported by a Comp Physiol A Neuroethology Sens CSUEB Faculty Support Grant (to JAM), an Neur Behav Physiol 192:235-245. Arkansas State Undergraduate Research Fellow- Gaston MR (1998) Neuropeptide TPep action on ship (to JBM), a Society for Integrative and salivary duct ciliary beating rate in the Comparative Biology Libbie Hyman Summer nudibranch mollusc Tritonia diomedea, Research Fellowship (to RR), and a 2009 Friday Invert Neurosci 3:327-333. Harbor Labs Blinks Fellowship (to RZ). We Hoyle G, Willows AOD (1973) Neuronal basis thank Scott Schwinge and Dr. Sophie George for of behavior in Tritonia. II. Relationship of their efforts in the Blinks Program, and the muscular contraction to nerve impulse anonymous reviewers for multiple suggestions pattern, J Neurobiol 4:239-254 that improved the manuscript. Hume RI, Getting PA, Del Beccaro MA (1982) Motor organization of Tritonia diomedea swimming I. Quantitative analysis of

swim behavior and flexion neuron firing Corresponding Author patterns, J Neurophysiol47:60-74. Katz PS, Frost WN (1997) Removal of spike James A. Murray, Ph.D. frequency adaptation via neuromodulation South Science 303C, intrinsic to the Tritonia escape swim cen- Department of Biological Sciences tral pattern generator, J Neurosci 17:7703- California State University East Bay 7713. 25800 Carlos Bee Blvd Murray JA, Jones AP, Links AC, Willows AOD Hayward CA 94542 (2011) Daily tracking of the locomotion [email protected] of the nudibranch Tritonia tetraquetra (Pallas 1788=Tritonia diomedea Bergh References 1894) in nature and the influence of water flow on orientation, crawling, and drag, Audesirk G (1978) Properties of central motor Marine Freshwater Behav Physiol 44:265- neurons exciting locomotory cilia in Tri- 288. tonia diomedea, J Comp Physiol A. Murray JA, Hewes RS, Willows AOD (1992) 128(3):259-268. Water-flow sensitive pedal neurons in Tri- Baltzley MJ, Lohmann KJ (2008) Comparative tonia: role in rheotaxis, J Comp Physiol A study of TPep-like immunoreactive neu- 171:373-385. rons in the central nervous system of nu- Murray, JA, Willows, AOD (1996) Function of dibranch molluscs, Brain, Behav Evol identified nerves in orientation to water 72:192-206. flow in Tritonia diomedea, J Comp Phys- Brown GD, Yamada S, Nakashima M, Moore- iol A 178:201-209. Kochlacs C, Sejnowski TJ, Shiono S Murray JA, Cain SD, Estepp JD (2006) Advanc- (2008) Independent Component Analysis es in the neural bases of orientation and of Optical Recordings from Tritonia navigation, Integ Comp Biol 46(6):871- Swimming Neurons. Institute for Neural 879. Computation Technical Report INC-08- Nishikawa K, Biewener AA, Aerts P, Ahn AN, 001, Salk Institute, La Jolla, CA, 20 pp. Chiel HJ, Daley MA, Daniel TL, Full RJ, Byrne JH (1982) Analysis of synaptic depres- Hale ME, Hedrick TL, Lappin AK, Nich- sion contributing to habituation of gill- ols TR, Quinn RD, Satterlie RA, Szymik B (2007) Neuromechanics: an integrative Page 16 of 16 Impulse: The Premier Undergraduate Neuroscience Journal 2012 approach for understanding motor control. Willows AO, Pavlova GA, Phillips NE (1997) Integ Comp Biol 47:16-54. Modulation of ciliary beat frequency by Popescu IR, Willows AO (1999) Sources of neuropeptides from identified molluscan magnetic sensory input to identified neu- neurons, J Exp Biol 200:1433-1439. rons active during crawling in the marine Wyeth RC, Willows AOD (2006a) Field behav- mollusc Tritonia diomedea, J Exp Biol ior of the nudibranch mollusc Tritonia di- 202:3029-3036. omedea, Biol Bull 210:81-96. Popescu IR, Frost WN (2002) Highly dissimilar Wyeth RC, Willows AOD (2006b) Odours de- behaviors mediated by a multifunctional tected by rhinophores mediate orientation network in the marine mollusk Tritonia to flow in the nudibranch mollusc, Trito- diomedea. J Neurosci 22:1985-1993. nia diomedea, J Exp Biol 209:1441-1453. NeuronBank, http://neuronbank.org/ Wyeth RC, Woodward OM, Willows AOD Redondo RL, Murray JA (2005) A single neuron (2006) Orientation and navigation relative serves a significant role in effecting turn- to water flow, prey, conspecifics, and ing while crawling in the marine slug Tri- predators by the nudibranch mollusc Tri- tonia diomedea (Bergh), J Comp Physiol tonia diomedea, Biol Bull 210:97-108. A 191:435-444. Thompson S, Smith SJ, Johnson JW (1986) Slow outward tail currents in molluscan bursting pacemaker neurons: two compo- nents differing in temperature sensitivity, J Neurosci 6:3169-3176. Wang JH, Cain SD, Lohmann KJ (2003) Identi- fication of magnetically responsive neurons in the marine mollusc Tritonia diomedea, J Exp Biol 206:381-388. Wang JH, Cain SD, Lohmann KJ (2004) Identi- fiable neurons inhibited by Earth-strength magnetic stimuli in the mollusc Tritonia diomedea, J Exp Biol 207:1043-1049. Willows AOD (1967) Behavioral acts elicited by stimulation of single, identifiable brain cells, Science 157:570-574. Willows, AOD (1978) Physiology of feeding in Tritonia diomedea. I. Behavior and mechanics, Mar Behav Physiol 5:115-135. Willows AOD, Dorsett DA, and Hoyle G. (1973a) The neuronal basis of behavior in Tritonia I. functional organization of the central nervous system, J Neurobiol 4: 204-237. Willows AOD, Dorsett DA, Hoyle G (1973b) The neuronal basis of behavior in Trito- nia. III. Neuronal mechanism of a fixed action pattern, J Neurobiol 4:255-285. Willows, A.O.D. (1999) Shoreward orientation involving geomagnetic cues in the nudibranch mollusc Tritonia diomedea, Ma- rine Freshwater Behav. Physiol. 32:181- 192.