View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Current Biology 21, 1036–1043, June 21, 2011 ª2011 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2011.04.040 Report Different Roles for Homologous Interneurons in Exhibiting Similar Rhythmic Behaviors

Akira Sakurai,1 James M. Newcomb,1,2 Joshua L. Lillvis,1 occur spontaneously or in response to electrical stimulation and Paul S. Katz1,* of a body wall nerve [2, 4, 5]. It had an average burst period of 1Neuroscience Institute, Georgia State University, 4.1 6 0.23 s (n = 25; see Figure 2A). In the isolated PO Box 5030, Atlanta, GA 30302, USA nervous system, analogous swim-like bursting activity, con- sisting of general alternation between the bursts in the left and right pedal ganglia, occurred spontaneously or in response Summary to body wall nerve stimulation. The average burst period in Dendronotus was 4.5 6 0.30 s (n = 13; see Figure 2D), which It is often assumed that similar behaviors in related species was not significantly different from that of the Melibe motor are produced by similar neural mechanisms. To test this, we pattern (p = 0.30, Student’s unpaired t test). Based on these examined the neuronal basis of a simple swimming behavior characteristics, we conclude that the bursting activity repre- in two nudibranchs (, Opisthobranchia), Melibe sents the motor pattern underlying the Dendronotus swimming leonina and Dendronotus iris. The side-to-side swimming behavior. movements of Dendronotus [1] strongly resemble those of Melibe [2, 3]. In Melibe, it was previously shown that the Neuroanatomical Identification of Swim Interneuron 1 central pattern generator (CPG) for swimming is composed In Melibe, swim interneuron 1 (Si1Mel; http://neuronbank.org/ of two bilaterally symmetric pairs of identified interneurons, wiki/index.php/Si1) has particular anatomical characteristics swim interneuron 1 (Si1) and swim interneuron 2 (Si2), which that allow it to be unambiguously identified from to are electrically coupled ipsilaterally and mutually inhibit animal [2, 4]. There is a single Si1Mel soma on the dorsal both contralateral counterparts [2, 4]. We identified homo- surface of each of the paired cerebral ganglia (Figure 1A). logs of Si1 and Si2 in Dendronotus. (Henceforth, homolo- The Si1Mel soma is one of the largest in this region of the cere- gous neurons in each species will be distinguished by the bral ganglion and is clear of pigment. Intracellular fills with subscripts Den and Mel.) We found that Si2Den and Si2Mel Neurobiotin or biocytin showed that the axon makes a charac- play similar roles in generating the swim motor pattern. teristic posterior bend before projecting to the ipsilateral pedal However, unlike Si1Mel, Si1Den was not part of the swim ganglion (Figure 1B, arrowhead). We found that there were fine CPG, was not strongly coupled to the ipsilateral Si2Den, branches in the cerebral ganglion and longer, thicker branches and did not inhibit the contralateral neurons. Thus, species in the pedal ganglion. None of the branches were observed to differences exist in the neuronal organization of the swim exit body wall nerves, but in 8 of 16 Si1Mel neurons that were CPGs despite the similarity of the behaviors. Therefore, simi- examined, thin processes were seen in the thicker of the two larity in species-typical behavior is not necessarily predic- pedal commissures, which encircle the esophagus (PP2; tive of common neural mechanisms, even for homologous nomenclature according to [7]). neurons in closely related species. Combining intracellular Neurobiotin fills of Si1Mel with sero- tonin immunohistochemistry, we determined that the Si1Mel Results soma was always located near a set of previously identified serotonergic neurons, the CeSP neurons [7, 8] (ten Si1Mel Swimming Behaviors neurons in seven preparations) (Figure 1C, 5-HT). In addition Melibe leonina swims by flattening its body in the sagittal plane to their serotonin immunoreactivity, the CeSP neurons can and flexing from side to side [2–6] (see Figure S1A available be identified based on their electrophysiological properties online), repeatedly bending at the midpoint with a periodicity [8, 9], facilitating the identification of Si1Mel in living prepara- of 2–5 s (mean = 3.0 6 0.18 s, n = 18). This behavior can be tions by providing a landmark for locating the neuron. triggered when the foot is dislodged from the substrate or Another unique characteristic of Si1Mel is that it displayed when the body wall is contacted by a noxious stimulus, such FMRFamide-like immunoreactivity (Figure 1C, FMRFamide). as high-molarity salt solution [2, 3]. We observed that the Although this antiserum may be relatively nonspecific in that swimming behavior of Dendronotus iris occurred under the it might recognize more than the peptide FMRFamide, the same circumstances and resembled the Melibe swim (see staining pattern was very reproducible, allowing it to be used Figure S1B) with a periodicity of body flexions that ranged as a marker of cell types. In five preparations, all six Si1Mel from 2.0 to 4.4 s (mean = 2.9 6 0.14 s, n = 22). The two behav- neurons injected with either Neurobiotin or biocytin were iors were not statistically different (p = 0.53, Student’s found to double label with the antiserum against FMRFamide. unpaired t test). This suite of neuroanatomical characteristics uniquely defined Si1Mel, differentiating it from all other neurons in Melibe. In Dendronotus (Figure 1E), we found just one neuron in Fictive Swim Motor Patterns each half of the brain that shared all of the neuroanatomical In Melibe, the fictive motor pattern produced by the isolated characteristics of Si1 : a colorless, relatively large soma in nervous system, which underlies the swimming behavior, can Mel the cerebral ganglion, an ipsilaterally projecting axon with a characteristic bend near the soma (Figure 1F, arrowhead), 2Present address: Department of Biology, New England College, 98 Bridge and branching in the pedal ganglion that spread into PP2 Street, Henniker, NH 03242, USA (n = 17) (Figure 1F, arrows). To determine whether there were *Correspondence: [email protected] other neurons with this morphology, in 19 preparations, we Homologous Neurons and Similar Behaviors 1037

Figure 1. Comparison of the Neuroanatomy and Immunoreactivity of Swim Interneurons 1 and 2 in Melibe and Dendronotus (A and E) Schematic drawings of the Melibe brain (A) and the Dendronotus brain (E) showing cerebral, pleural, and pedal ganglia; the location of swim inter- neuron 1 (Si1, pink) and 2 (Si2, blue) somata and their neurites; and the locations of the serotonergic CeSP somata (green). The gross anatomy of the pedal ganglia in Dendronotus differs from that of Melibe in that the pedal ganglia are bilobed and the pedal commissures are much shorter than in Melibe.To encircle the esophagus, the distal lobes curl around the esophagus. (B and F) Neurobiotin fills of Si1 show the location of the soma in the cerebral ganglion of Melibe (B) and Dendronotus (F). The axon in both species has a characteristic posterior bend (arrowhead) before projecting to the ipsilateral pedal ganglion. (C and G) Left: Si1, labeled by intracellular injection of Neurobiotin (pink), is surrounded by the CeSP neurons (green), which are immunoreactive to serotonin (5-HT). Right: Si1, labeled by intracellular injection of Neurobiotin (pink), is doubled labeled (white) by antisera against the neuropeptide FMRFamide (green).

(D and H) Both Si2Mel (D) and Si2Den (H) have a soma in the pedal ganglion and send a thick axon through the pedal commissure (PP2) to the opposite pedal ganglion, where they have a distinctive linearly shaped terminal arbor (arrows). injected a total of 43 neurons in this region with either Neuro- of this neuron was located near the serotonergic CeSP biotin or biocytin, and we did not see more than one neuron neurons (eleven neurons from eight preparations) (Figure 1G, on each side with this characteristic morphology. The soma 5-HT). Furthermore, in all six neurons in four preparations Current Biology Vol 21 No 12 1038

where it was examined, this neuron displayed FMRFamide-like A Melibe D Dendronotus immunoreactivity (Figure 1G, FMRFamide). On the basis of these properties, which uniquely define this neuron, we call it Si1Mel Si1Den 50 50 mV Si1Den and consider it a putative homolog of Si1Mel. (L) mV (L)

Si1Den Identification of Swim Interneuron 2 Si1Mel 50 50 (R) mV In Melibe, there is a single swim interneuron 2 (Si2Mel; http:// (R) mV neuronbank.org/wiki/index.php/Si2) soma on the dorsal Si2 Si2 surface of each pedal ganglion [4]. Its axon projects to the Mel 50 Den 50 contralateral pedal ganglion through PP2 [4](Figures 1A and (R) mV (R) mV 1D). We found that Si2 had a characteristic linear, dense Mel PdGN arborization in the pedal ganglion (Figure 1D, arrows) (n = 4). 50 PdGN 50 mV We identified one neuron in Dendronotus in each side of the (L) (L) mV brain that shared similar anatomical characteristics to Si2Mel PdN2 PdN2 (n = 5) (Figures 1E and 1H). There was one soma on the dorsal (L) (R) surface of the proximal lobe of each pedal ganglion, which had 5 s 5 s an axon that projected through PP2 and terminated in the proximal lobe of the contralateral pedal ganglion. There, it BE had an axonal arborization similar to that of Si2Mel (Figure 1H, arrows). Based on these characteristics and the electrophysi- Si2Mel 50 Si2Den 50 mV mV ological characteristics (see below), we named this neuron (R) (R) 10 10 Si2Den and consider it a putative homolog of Si2Mel. nA nA Si2Mel Si2Den (L) 50 (L) 50 Si2Den but Not Si1Den Is a Member of the Swim Central mV mV Pattern Generator 5 s 5 s In Melibe, both Si1Mel and Si2Mel are core members of the swim central pattern generator (CPG), and both display bursting CF activity in phase with the swimming movements [2, 4]. Both neurons fire in phase with ipsilateral pedal motor neurons Si2Mel 50 Si2Den 50 mV mV and in antiphase with their contralateral counterparts (Fig- (R) (R) 10 10 ure 2A). Brief depolarization or hyperpolarization of either nA nA neuron resets the swim motor pattern (Figures 2B and 2C) [4]. Si2 Si2 In Dendronotus, we found that Si2 exhibited properties Mel 50 Den 50 Den (L) (L) consistent with it being a member of the swim CPG. It was mV mV rhythmically active at a constant phase relation to pedal 5 s 5 s efferent neurons and nerve activity (Figure 2D), and the two contralateral Si2Den neurons fired bursts in strict alternation Figure 2. Comparison of Si1 and Si2 Activity in Melibe and Dendronotus (Figures 2E and 2F). Brief depolarization (Figure 2E) or hyper- (A) In Melibe, simultaneous intracellular microelectrode recordings from the polarization (Figure 2F) of either Si2Den reset the motor pattern. left (L) and right (R) Si1Mel, the right Si2Mel, and a left efferent pedal ganglion In contrast, Si1Den did not fire rhythmic bursts of action poten- neuron (PdGN) and an extracellular recording from a left body wall nerve tials during the swim motor pattern but instead fired irregularly (PdN2) show that the left and right Si1Mel burst in alternation with each other. at 2–15 Hz (Figure 2D; n = 32). Brief depolarization or hyperpo- The right Si1Mel fired bursts synchronously with the right Si2Den and in anti- phase with a left PdGN and extracellularly recorded bursts in the left PdN2. larization of Si1Den did not reset the motor pattern. Together, (B and C) Brief depolarizing (B) or hyperpolarizing (C) current pulses into the these observations led us to conclude that although Si2 is Den right Si2Mel reset the motor pattern. The red dots indicate the expected a member of the swim CPG, Si1Den is not. times of left Si2Mel bursts. The phase relationship between the contralateral Si2Mel neurons was not disrupted; they fired in antiphase before and after Si1Den Modifies the Swim Motor Pattern the reset (gray vertical bars). The effect of Si1 activity on an ongoing swim motor pattern (D) In Dendronotus, simultaneous intracellular microelectrode recordings from the left and right Si1 , the right Si2 , and a left PdGN and an extra- differed in the two species. In Melibe, injection of depolarizing Den Den cellular recording from the right PdN2 show that both Si1Mel neurons fired current into Si1 slowed down the swim motor pattern, Mel tonically while the right Si2Den burst in alternation with the left PdGN and halting it if enough current was injected (Figure 3A). Injection synchronously with bursts recorded on the right PdN2. of hyperpolarizing current had a similar effect (Figure 3B). (E and F) Injection of brief depolarizing (E) or hyperpolarizing (F) current The result was an inverted U-shaped relation of change in pulses into the right Si2Den reset the motor pattern as indicated by the red Si2 cycle frequency to the amount of current injected into dots. The contralateral Si2Den burst in antiphase before and after the reset Mel (gray vertical bars). Si1Mel (Figure 3C). In contrast, in Dendronotus, injection of depolarizing current into Si1Den did not stop the bursting of Si2Den but rather increased the burst frequency and the intra- burst spike frequency (Figure 3D). Hyperpolarization of Si1Den Species Differences in Synaptic Connectivity decreased burst frequency and intraburst spike frequency of We compared the synaptic connectivity of Si1 and Si2 in Si2Den but did not halt ongoing regular bursting (Figure 3E). Dendronotus to that in Melibe to determine whether network The result was a monotonic change in Si2Den burst frequency differences could account for the disparity in the actions of as a function of current injected into Si1Den (Figure 3F). Thus, the neurons. In Melibe, it was previously reported that Si1Mel the effects of these homologous neurons on similar swim and Si2Mel each form inhibitory synaptic connections with motor patterns differed in these two species. both contralateral Si1Mel and Si2Mel counterparts and are Homologous Neurons and Similar Behaviors 1039

Melibe Dendronotus Figure 3. Si1 Has Different Effects on the Swim Motor A D Pattern in Melibe and Dendronotus

3 nA 4 nA (A and B) Intracellular recordings of Si1Mel and Si2Mel, with corresponding plots of the instantaneous spike frequency Si1Mel Si1Den (black) and burst frequency (pink) of Si2Mel. Depolarization (L) 50 (L) 50 of Si1Mel (3 nA) arrested the swim motor pattern (A). The mV mV swim also halted when Si1Mel was hyperpolarized by a nega- tive current injection (22 nA, B). Si2Mel Si2Den (C) Relationship between the percent change in burst 50 50 (R) mV (R) mV frequency recorded in Si2Mel or in a pedal ganglion neuron in response to the steady current injection into Si1Mel. Both 15 0.3 15 0.3 depolarizing and hyperpolarizing current injection into Si1Mel 10 0.2 decreased the cycle frequency of the swim motor pattern, 10 0.2 resulting in an inverted U-shaped relationship. Hz Hz 5 0.1 5 0.1 (D and E) Intracellular recordings of Si1Den and Si2Den, with 0 0.0 0 0.0 corresponding plots of the instantaneous spike frequency 0 102030405060 0 10203040506070 (black) and burst frequency (pink) of Si2Den. Depolarization s s of Si1Den (4 nA) accelerated the swim motor pattern (D), whereas the swim slowed down when Si1Den was hyperpo- larized by a negative current injection (24 nA, E). B E (F) There was a monotonic increase in the percent change in -2 nA -4 nA the Si2Den burst frequency in response to increasing current injection into Si1Den. In (C) and (F), each point represents the Si1Mel Si1Den mean 6 standard deviation (SD) of data obtained from 3–14 50 20 (L) mV (L) mV preparations.

Si2Mel Si2Den (R) 50 (R) 50 mV mV electrical connections among all of the swim 15 0.3 20 0.3 CPG neurons as well (Figure 4G), although the 10 0.2 15 0.2 relative strengths of coupling differed from those 10 Hz 5 0.1 5 0.1 in Melibe (Figure 4H). In particular, the coupling 0 0.0 0 0.0 between the ipsilateral Si1Den and Si2Den (Fig- 0 102030405060 0102030405060 ure 4G3) was much weaker than in Melibe (Fig- s s ure 4C3), and the coupling between the contralat- eral Si1Den (Figure 4G1) was stronger than in C F Melibe (Figure 4C1). Thus, there are substantial differences in the 60 60 extent of electrical coupling and the presence of 40 40 inhibitory synapses between Si1 and Si2 in these 20 20 two species: the connections in Melibe are domi- 0 0 nated by strong ipsilateral electrical coupling and -20 -20 contralateral inhibition (Figure 4A), whereas in -40 -40 -60 -60 Dendronotus the ipsilateral electrical coupling is Swim cycle (%) (%) cycle Swim (%) cycle Swim weaker and only Si2Den exhibits contralateral inhi- Δ -80 Δ -80 -100 -100 bition (Figure 4E). -3 -2 -1 0 1 2 3 4 -4 -3 -2 -1 0 1 2 3 4 Current injected Current injected Discussion into Si1Mel (nA) into Si1Den (nA) Similar Behaviors, Different Neural Circuitry Although the swimming behaviors of Melibe and Dendronotus are similar, there are important electrically coupled ipsilaterally [4](Figure 4A). In this study, distinctions in the functions of homologous interneurons we confirmed the previous results by showing that spiking in caused by dissimilarities in neuronal connectivity. In Melibe, each neuron evoked synaptic inhibition contralaterally Si1Mel and Si2Mel both participate as members of the CPG (Figures 4B1 and 4B2) and depolarization ipsilaterally (Fig- through contralateral inhibition and strong ipsilateral electrical ure 4B3). In contrast, in Dendronotus, Si1Den made functionally coupling [2, 4]. Homologs of these neurons were identified in excitatory connections both contralaterally and ipsilaterally Dendronotus based on unique anatomical and neurochemical (Figures 4F1 and 4F3); only the Si2Den neurons exhibited features. However, based on their activity patterns and contralateral inhibition (Figure 4F2). ability to reset the motor pattern, only Si2Den was determined In Melibe, it was previously reported that the ipsilateral to be a member of the CPG. Si1Den did not fire rhythmic bursts Si1Mel and Si2Mel are electrically coupled [4]. Here we found in phase with the swim motor pattern and influenced the that in addition to the ipsilateral Si1Mel–Si2Mel coupling (Fig- bursting of the CPG differently than did Si1Mel. Si1Den lacked ure 4C3), there were electrical connections among all of the contralateral inhibition and strong coupling to the ipsilateral swim CPG neurons in Melibe (Figures 4C1–4C3). However, Si2Den seen among Melibe homologs. Despite the possibility the ipsilateral connection between Si1Mel and Si2Mel was by that there may be additional neurons that participate in the far the strongest (Figure 4D). In Dendronotus, there were CPG in one or both species, the present results demonstrate Current Biology Vol 21 No 12 1040

A Melibe E Dendronotus

Si1 Si1 Si1 Si1

LR LR

Si2 Si2 Si2 Si2

B1 cont. B2 cont. B3 ipsi. F1 cont. F2 cont. F3 ipsi.

Si1Mel Si2Mel Si1Mel Si1Den Si2Den Si1Den (L) (L) (L) (L) (L) (L) 40mV Si1 Si2 Si2 Si1Den Si2Den Si2 Mel Mel Mel 5mV Den (R) (R) (R) (R) (R) (L) 4s 40mV 5mV C1 cont. C2 cont. C3 ipsi. G1 cont. G2 cont. G3 ipsi. 4s -4 nA -4 nA -4 nA -3 nA -4 nA -4 nA

Si1Mel Si2Mel Si1Mel Si1Den Si2Den Si1Den (L) (L) (L) (L) (L) (L)

Si1Mel Si2Mel Si2Mel Si1Den Si2Den Si2Den (R) (R) (L) 40mV (R) (R) (L) 5mV 40mV 5mV 2s 4s DH

Si1↔Si1 (cont.) Si1↔Si1 (cont.) Si2↔Si2 (cont.) Si2↔Si2 (cont.) Si1→Si2 (cont.) Si1→Si2 (cont.) Si1←Si2 (cont.) Si1←Si2 (cont.) Si1→Si2 (ipsi.) Si1→Si2 (ipsi.) Si1←Si2 (ipsi.) Si1←Si2 (ipsi.)

0.0 0.1 0.2 0.0 0.1 0.2 Coupling coeffecient Coupling coeffecient

Figure 4. Melibe and Dendronotus Differ in Synaptic Connectivity (A) A schematic diagram of the Melibe swim circuit (gray area). Circles represent inhibitory connections; resistor symbols represent electrical coupling. The line thickness for the resistors reflects the strength of electrical coupling.

(B) Spikes in Si1Mel (B1) and Si2Mel (B2) evoke one-for-one inhibitory postsynaptic potentials (IPSPs) in their contralateral counterparts. Spikes in Si1Mel evoke a depolarization in the ipsilateral Si2Mel (B3). (C) Hyperpolarization reveals weak electrical coupling contralaterally (C1 and C2) and strong coupling ipsilaterally (C3). (D) Bar graph showing the coupling coefficients for each cell pair. Each bar represents the mean 6 SD of data obtained from 3–6 preparations. The ipsilateral couplings between Si1Mel and Si2Mel were by far the strongest. (E) Circuit diagram summarizing the connectivity in Dendronotus. The gray area indicates the neurons that are components of the swim central pattern generator.

(F) Spikes in Si1Den depolarize the contralateral Si1Den (F1) and Si2Den (F3), whereas spikes in Si2Den evoke one-for-one IPSPs in its contralateral counterpart (F2). (G) Hyperpolarization of each neuron reveals electrical coupling both contralaterally and ipsilaterally. (H) Coupling coefficients between the swim interneurons in Dendronotus. Each bar represents the mean 6 SD of data obtained from 4–7 preparations. A two-way analysis of variance showed a significant difference between the two species (F = 55.4, p < 0.001) and the tested neural connections (F = 52.8, p < 0.001). There was a statistically significant interaction between the species and the magnitude of electrical connections (F = 58.6, p < 0.001). Post hoc analyses (Fisher’s least significant difference) showed significant differences within and between the two species (p < 0.001). Within

Melibe, the coupling coefficients between Si1Mel and Si2Mel were significantly greater than for all other pairs (p < 0.001). Within Dendronotus, the coupling coefficient of Si1Den4Si1Den (0.060 6 0.018, n = 4) was significantly greater than for all other pairs (p < 0.01). Across species, the coupling coefficients between Si1Mel and Si2Mel were significantly greater than corresponding connections between Si1Den and Si2Den (p < 0.005), whereas the coupling coefficient of Si1Den4Si1Den was significantly greater than that of Si1Mel4Si1Mel (p < 0.001). All recordings in this figure and measurements for the graph were made in high-divalent-cation saline. that the neural circuitries differ, likely causing homologous Species Differences in Other Neural Circuits neurons to function differently in the production of similar In invertebrates, divergent behaviors correlate with differ- behaviors. ences in the connectivity or activity of identified neurons [10]. Homologous Neurons and Similar Behaviors 1041

For example, synaptic connections from mechanoreceptors be due to independent evolution [36–38]. Independent evolu- differ in leech species that respond differently to mechanical tion could suggest that the underlying neural mechanisms touch [11]. There are differences in the activity of homolo- are different. However, even if two behaviors are homologous, gous neurons in Melibe compared to another , the underlying neural mechanisms could have diverged. Here, Tritonia diomedea, which swims with dorsal-ventral body we found that the functions of homologous neurons in the flexions instead of side-to-side body flexions [9]. In the production of similar behaviors differ in two closely related stomatogastric nervous system of crustaceans, different species. Thus, the presence of similar behaviors in two related motor patterns are produced by homologous neurons through species does not guarantee that the underlying neural mecha- differences in neurotransmitter content [12–14] and small nisms have been conserved. differences in synaptic connectivity, particularly electrical coupling [15, 16]. Experimental Procedures Species differences in neural circuits that underlie similar behavior generally have arisen through independent evolu- Animal Collection, Maintenance, and Dissection tion. For example, a number of species within the animal Dendronotus iris (60–200 mm in body length) and Melibe leonina (30– kingdom show undulatory locomotion (e.g., nematodes, anne- 100 mm) were obtained as adults from Living Elements Ltd. or Monterey Abalone Company or were collected near Friday Harbor Laboratory, San lids, mollusks, and vertebrates). Convergent evolution of this Juan, WA. were kept in artificial seawater tanks at 10C–12C and behavior across phyla indicates that many of the same a 12 hr/12 hr light/dark cycle. mechanisms are employed but that the neural structures Animals were anesthetized by injection of 0.33 M magnesium chloride into that produce them are not homologous [17]. As another the body cavity. A cut was made in the body wall near the esophagus. The example, electrosensing evolved independently in African brain, consisting of the cerebral, pleural, and pedal ganglia, was removed by mormyriforme and South American gymnotiforme fish [18, cutting all nerve roots. The brain was transferred to a Sylgard-lined dish, where it was superfused, at a rate of 0.5 ml/min, with normal saline (in 19]. Species in both clades produce wave-like electric mM: 420 NaCl, 10 KCl, 10 CaCl2, 50 MgCl2, 11 D-glucose, 10 HEPES discharges and exhibit jamming avoidance responses, which [pH 7.6]) or with artificial sea water (Instant Ocean). also evolved independently [20]. Some parts of the circuits Connective tissue surrounding the brain was manually removed with that control this sensorimotor response differ significantly in forceps and fine scissors while keeping the brain at w4C to reduce  their neural composition, but others involve homologous brain neuronal activity. The temperature was raised to 10 C for electrophysiolog- areas [21, 22]. ical experiments.

Evolution of Side-to-Side Swimming Electrophysiology Intracellular recordings were obtained using 15–60 MU glass microelec- The precise phylogenetic relationship between Melibe and trodes filled with 3 M potassium chloride and connected to an Axoclamp Dendronotus has not been adequately resolved (Figure S2). 2B amplifier (Axon Instruments). Extracellular suction electrode recordings There is general agreement that within Nudibranchia, both were obtained by drawing individual nerves into polyethylene tubing filled species are within the monophyletic clade with normal saline or artificial seawater and connected to an A-M Systems and even within the subclade Dendronotoidea [23–25]. The Differential AC Amplifier (model 1700, A-M Systems, Inc.). Both intra- and side-to-side swimming behavior of Melibe and Dendronotus extracellular recordings were digitized (>2 kHz) with a 1401Plus or Micro1401 A/D converter (Cambridge Electronic Design). In some experi- has been observed in several other nudibranch species within ments, a biotinylated compound solution (see below) was injected into Cladobranchia, including Scyllaea, Bornella [26], Lomanotus a cell via iontophoresis for 30 min (1–10 nA, 1 Hz, 50% duty cycle).

[27], and Flabellina [28]. Even some Plocamopherus species, The effect on burst period of current injection into Si1Mel and Si1Den was which are not in Cladobranchia, swim with side-to-side or examined by injecting positive or negative current (24 nA to 6 nA) through lateral body flexions [29]. But lateral flexion is far from a bridge-balanced microelectrode for more than 10 s until the burst universal within Cladobranchia; in fact, for most species, there frequencies of the swim CPG neurons settled at a steady frequency. Synaptic connectivity and electrical coupling between the swim interneu- are no reports of swimming. Furthermore, there is the well- rons were tested in the presence of high-divalent-cation saline, which studied example of Tritonia, which swims with dorsal-ventral raises the threshold for spiking and reduces spontaneous neural firing. body flexions [30]. The composition of the high-divalent-cation saline was (in mM) 285 Based on the distribution of swimming behavior in the NaCl, 10 KCl, 25 CaCl2, 125 MgCl2, 11 D-glucose, 10 HEPES (pH 7.6). To clade, there are three possible evolutionary scenarios for the measure electrical coupling, we applied brief hyperpolarizing current steps differences in swim circuit organization between Melibe and (2–4 s, 2– 10 nA) to the presynaptic neuron through an additional micro- electrode placed in the same neuron while monitoring the membrane Dendronotus: (1) the swim CPGs in these two species inde- potential of both pre- and postsynaptic neurons. Coupling coefficients pendently evolved to include Si2; (2) the Dendronotus condi- were calculated as the change in membrane potential of the postsynaptic tion represents the ancestral state and incorporation of Si1Mel neuron divided by the change in membrane potential in the presynaptic into the CPG is a derived feature in the Melibe lineage; or (3) neuron. the Melibe condition represents the ancestral state and the Data acquisition and analysis were performed with Spike2 software removal of Si1 from the CPG is a derived feature in the (Cambridge Electronic Design) and SigmaPlot (Jandel Scientific). Statistical Den comparisons were made using Student’s t test, paired t test, or two-way Dendronotus lineage. Resolution of the phylogeny and tests analysis of variance with post hoc pairwise multiple comparisons by on outgroups need to be performed before a polarity to the Fisher’s least significant difference method. In all cases, p < 0.05 was change can be inferred. considered significant. Results are expressed as mean 6 standard deviation. Implications for Evolution of Behavior Although at times controversial [31], it is widely accepted that Tracer Injections and Immunohistochemistry behaviors, like anatomical structures, can be homologous, After intracellular recording, neurons were filled with Neurobiotin tracer meaning that they are derived from a behavior exhibited by a [N-(2-amino-ethyl) biotinamide hydrochloride; Vector Labs] or biocytin (Sigma). A microelectrode filled with 2% solution of either Neurobiotin common ancestor [32–34]. This concept can be traced back tracer or biocytin (in 0.75 M KCl) was inserted into the cell body, and bipolar to Darwin, who sought to compare emotions in humans and current pulses (from 25 nA to 5 nA at 50% duty cycle) were applied at 1 Hz other animals [35]. As with any characteristic, similarities might for 0.5–3 hr. The preparation was then incubated in running physiological Current Biology Vol 21 No 12 1042

saline for 6 hr at 10C. After incubation, the brain was fixed in 4% parafor- 9. Newcomb, J.M., and Katz, P.S. (2009). Different functions for homolo- maldehyde in 0.1 M phosphate-buffered saline (PBS, pH 7.4) for 5 hr at 4C. gous serotonergic interneurons and serotonin in species-specific After rinsing with PBS several times, the brain was treated with 4.0% Triton rhythmic behaviours. Proc. Biol. Sci. 276, 99–108. X-100 in PBS overnight and then incubated with streptavidin Alexa Fluor 10. Katz, P.S. (2007). Evolution and development of neural circuits in inver- 594 conjugate (Invitrogen) diluted to 1:200 in PBS containing 0.5% Triton tebrates. Curr. Opin. Neurobiol. 17, 59–64.  X-100 for 3–5 days at 4 C. The brain was washed six times with PBS 11. Baltzley, M.J., Gaudry, Q., and Kristan, W.B., Jr. (2010). Species-specific over 6 hr, dehydrated in a graded ethanol series, cleared by methyl salicy- behavioral patterns correlate with differences in synaptic connections late, and mounted on a slide glass with Cytoseal 60 (Electron Microscopy between homologous mechanosensory neurons. J. Comp. Physiol. A Sciences). Neuroethol. Sens. Neural Behav. Physiol. 196, 181–197. If the preparation was also used for immunohistochemistry, then after 12. Meyrand, P., Faumont, S., Simmers, J., Christie, A.E., and Nusbaum, fixation and 4.0% Triton-X treatment as described above, the ganglia M.P. (2000). Species-specific modulation of pattern-generating circuits. were incubated for 1 hr in antiserum diluent (ASD) consisting of 0.5% Triton Eur. J. Neurosci. 12, 2585–2596. X-100, 1% normal goat serum, and 1% bovine serum albumin in PBS. This 13. Fe´ nelon, V.S., Le Feuvre, Y., and Meyrand, P. (2004). Phylogenetic, was followed by 48–96 hr in primary rabbit polyclonal anti-serotonin (lot ontogenetic and adult adaptive plasticity of rhythmic neural networks: A #924005, catalog #20080, ImmunoStar) or anti-FMRFamide (lot #831001, common neuromodulatory mechanism? J. Comp. Physiol. A Neuroethol. catalog #20091, ImmunoStar) antiserum diluted 1:1000 in ASD. After several Sens. Neural Behav. Physiol. 190, 691–705. washes with 0.5% Triton X-100 in PBS, ganglia were incubated overnight in 14. Katz, P.S., and Tazaki, K. (1992). Comparative and evolutionary aspects goat anti-rabbit antiserum conjugated to Alexa 488 (Molecular Probes) of the crustacean stomatogastric system. In Dynamic Biological diluted 1:100 in ASD. Ganglia were then dehydrated and mounted on slides Networks: The Stomatogastric Nervous System, R.M. Harris-Warrick, as described above. E. Marder, A.I. Selverston, and M. Moulins, eds. (Cambridge, MA: MIT Fluorescence images were visualized by confocal microscopy (LSM 510 Press), pp. 221–261. mounted on an Axiovert 100M microscope or LSM 700 on an AxioExaminer D1 microscope, Carl Zeiss, Inc.) with a 103 or 203 objective. Fluorophores 15. Weimann, J.M., Meyrand, P., and Marder, E. (1991). Neurons that were excited with one of two lasers (488 and 543 nm), and fluorescent emis- form multiple pattern generators: identification and multiple activity sions were passed through a band-pass filter (505–550 nm) for visualization patterns of gastric/pyloric neurons in the crab stomatogastric system. of Alexa 488 and a 560 nm long-pass filter for visualization of Alexa 594. J. Neurophysiol. 65, 111–122. LSM 510 software was used to acquire images. Maximal projections of 16. Weimann, J.M., and Marder, E. (1994). Switching neurons are integral confocal stacks were exported as TIFF files and imported into Adobe members of multiple oscillatory networks. Curr. Biol. 4, 896–902. Photoshop CS. In Photoshop, projections were assembled into a montage 17. Mullins, O.J., Hackett, J.T., Buchanan, J.T., and Friesen, W.O. (2011). of the entire central nervous system, and overall brightness and contrast Neuronal control of swimming behavior: Comparison of vertebrate were adjusted. and invertebrate model systems. Prog. Neurobiol. 93, 244–269. 18. Bullock, T.H., Northcutt, R.G., and Bodznick, D.A. (1982). Evolution of electroreception. Trends Neurosci. 5, 50–53. Supplemental Information 19. New, J.G. (1997). The evolution of vertebrate electrosensory systems. Brain Behav. Evol. 50, 244–252. Supplemental Information includes two figures and can be found with this 20. Kawasaki, M. (1993). Independently evolved jamming avoidance article online at doi:10.1016/j.cub.2011.04.040. responses employ identical computational algorithms: A behavioral study of the African electric fish, Gymnarchus niloticus. J. Comp. Acknowledgments Physiol. A Neuroethol. Sens. Neural Behav. Physiol. 173, 9–22. 21. Kawasaki, M. (2009). Evolution of time-coding systems in weakly elec- The work was supported by National Science Foundation grants IOS- tric fishes. Zoolog. Sci. 26, 587–599. 0814411 and IIS-0827418. We thank Charuni Gunaratne for assistance 22. Rose, G.J. (2004). Insights into neural mechanisms and evolution of with this project. behaviour from electric fish. Nat. Rev. Neurosci. 5, 943–951. 23. Waegele, H., and Willan, R.C. (2000). Phylogeny of the nudibranchia. Received: February 7, 2011 Zool. J. Linn. Soc. 130, 83–181. Revised: April 1, 2011 24. Wollscheid-Lengeling, E., Boore, J., Brown, W., and Waegele, H. Accepted: April 26, 2011 (2001). The phylogeny of Nudibranchia (Opisthobranchia, , Published online: May 26, 2011 Mollusca) reconstructed by three molecular markers. Org. Divers. Evol. 1, 241–256. 25. Vonnemann, V., Schrodl, M., Klussmann-Kolb, A., and Wagele, H. References (2005). Reconstruction of the phylogeny of the Opisthobranchia 1. Kjerschow-Agersborg, H.P. (1922). Notes on the locomotion of the (Mollusca: Gastropoda) by means of 18S and 28S rRNA gene nudibranchiate mollusk Dendronotus giganteous O’Donoghue. Biol. sequences. J. Molluscan Stud. 71, 113–125. Bull. 42, 257–266. 26. Collingwood, C. (1878). On some new species of Nudibranchiate 2. Watson, W.H., Lawrence, K.A., and Newcomb, J.M. (2001). Molluscs from the eastern seas. Trans. Linn. Soc. Lond. 2, 123–140. Neuroethology of Melibe leonina swimming behavior. Am. Zool. 41, 27. Garstang, W. (1890). A complete list of the opisthobranchiate mollusca 1026–1035. found at Plymouth; with further observations on their morphology, 3. Lawrence, K.A., and Watson, W.H., 3rd. (2002). Swimming behavior of colours, and natural history. J. Mar. Biol. Assoc. U. K. 1, 399–457. the nudibranch Melibe leonina. Biol. Bull. 203, 144–151. 28. Farmer, W.M. (1970). Swimming gastropods (Opisthobranchia and 4. Thompson, S., and Watson, W.H., 3rd. (2005). Central pattern generator Prosobranchia). Veliger 13, 73–89. for swimming in Melibe. J. Exp. Biol. 208, 1347–1361. 29. Pease, W.H. (1860). Descriptions of a new species of mollusca from the 5. Watson, W.H., 3rd, Newcomb, J.M., and Thompson, S. (2002). Neural Sandwich Islands. Part 1. Proc. Zool. Soc. Lond. 28, 18–36. correlates of swimming behavior in Melibe leonina. Biol. Bull. 203, 30. Willows, A.O. (1967). Behavioral acts elicited by stimulation of single, 152–160. identifiable brain cells. Science 157, 570–574. 6. Hurst, A. (1968). The feeding mechanism and behavior of the opistho- 31. Atz, J.W. (1970). The application of the idea of homology to behavior. branch Melibe leonina. Symp. Zool. Soc. Lond. 22, 155–166. In Development and Evolution of Behavior: Essays in Memory of 7. Newcomb, J.M., Fickbohm, D.J., and Katz, P.S. (2006). Comparative T.C. Schneirla, L.R. Aronson, E. Tobach, D.S. Lehrman, and J.S. mapping of serotonin-immunoreactive neurons in the central nervous Rosenblatt, eds. (San Francisco: W.H. Freeman and Co.), pp. 53–74. systems of nudibranch molluscs. J. Comp. Neurol. 499, 485–505. 32. Lauder, G.V. (1986). Homology, analogy, and the evolution of behavior. 8. Newcomb, J.M., and Katz, P.S. (2007). Homologues of serotonergic In Evolution of Animal Behavior, M.H. Nitecki and J.A. Kitchell, eds. central pattern generator neurons in related nudibranch molluscs with (New York: Oxford University Press), pp. 9–40. divergent behaviors. J. Comp. Physiol. A Neuroethol. Sens. Neural 33. Wenzel, J.W. (1992). Behavioral homology and phylogeny. Annu. Rev. Behav. Physiol. 193, 425–443. Ecol. Syst. 23, 361–381. Homologous Neurons and Similar Behaviors 1043

34. Rendall, D., and Di Fiore, A. (2007). Homoplasy, homology, and the perceived special status of behavior in evolution. J. Hum. Evol. 52, 504–521. 35. Darwin, C. (1872). The Expression of the Emotions in Man and Animals (London: John Murray). 36. Sanderson, M.J., and Hufford, L. (1996). Homoplasy: The Recurrence of Similarity in Evolution (San Diego, CA: Academic Press). 37. Morris, S.C. (2006). Evolutionary convergence. Curr. Biol. 16, R826– R827. 38. Nishikawa, K.C. (2002). Evolutionary convergence in nervous systems: Insights from comparative phylogenetic studies. Brain Behav. Evol. 59, 240–249.