A single motor neuron determines the rhythm of early motor behavior in Ciona

Single motor neuron regulates rhythmic tail flick in prehatching Ciona embryo.

Abstract

Recent work in tunicate supports the similarity between the motor circuits of vertebrates and basal deuterostome lineages. To understand how the rhythmic activity in motor circuits is acquired during development of protochordate Ciona, we investigated the coordination of the motor response by identifying a single pair of oscillatory motor neurons (MN2/A10.64). The MN2 neurons had Ca²⁺ oscillation with an ~80-s interval that was cell autonomous even in a dissociated single cell. The Ca²⁺ oscillation of MN2 coincided with the early tail flick (ETF). The spikes of the membrane potential in MN2 gradually correlated with the rhythm of ipsilateral muscle contractions in ETFs. The optogenetic experiments indicated that MN2 is a necessary and sufficient component of ETFs. These results indicate that MN2 is indispensable for the early spontaneous rhythmic motor behavior of Ciona. Our findings shed light on the understanding of development and evolution of chordate rhythmical locomotion.

INTRODUCTION

Swimming behavior is a common behavior observed in a variety of aquatic organisms. Before the maturation of swimming, embryos exhibit several stereotyped motor behaviors; in Xenopus and zebrafish, the earliest behaviors are spontaneous (1, 2). Early spontaneous motor behaviors in vertebrates are driven by spontaneous network activity in the developing spinal cord (e.g., chicken, mouse, and zebrafish) (3–6).

Motor neurons are believed to play a crucial role in motor circuit development (7–9). The motor neurons are the first population of neurons to be active in the spinal cord, and they may recruit other types of neurons to finally form multicell coordinated Ca²⁺ activity in zebrafish (9).

While the motor circuit development in vertebrates has been studied intensively, studies of the closest relative of vertebrates (ascidian) are also notable especially in the field of evolution and developmental biology (10–16). While ascidian has many conserved characteristics with vertebrates, it has only 330 cells in the central nervous system, which can facilitate studies of motor circuit development at the single-cell level (17).

Hatched Ciona larvae swim via a somatic muscle on the left and right sides (18). Ciona larvae have only five pairs of motor neurons (10), and there are five pairs of cells immunostained by the cholinergic marker in the motor ganglion (19). Each motor neuron is named MN1 to MN5 via a numerical number from anterior to posterior (10). Of those, the cell lineage of MN1 and MN2 is considered to correspond to A11.118 and A10.64, respectively (20–24). The molecular profile and differentiation of motor neurons have also been well studied, and each motor neuron can be distinguished by specific transcription makers (21, 25–27).

Furthermore, ascending contralateral inhibitory neurons (ACINs) may play a pivotal role in the alternating left and right muscle contraction for swimming in Ciona larvae (19). ACINs have been suggested to be GABAergic/glycinergic neurons (19). The motor neurons and ACINs are considered essential components allowing hatched larvae to generate swimming (19).

Several studies have attempted to link connectomes and functional motor circuits to swimming behavior in Ciona. Currently, the connectome of the hatched larva has been completed, and a network of motor circuits has been proposed (10, 11). The descending decussating neurons (ddNs) project axons contralaterally to MN2 (11). Because the network of ddNs is suggested to be the homology with that of reticulospinal Mauthner cells, the ddNs may regulate bilateral tail flicks in the later larval stage (11). Moreover, Gnrh2-expressing cells (28) and anatomical asymmetry of the central nervous system (29) have been reported to be associated with ascidian swimming behavior.

While molecular work and circuit connectivity studies have highlighted similarities between the Ciona motor circuit and hindbrain motor circuits of vertebrates, it remains unclear how spontaneous motor behavior is acquired during motor circuit development in Ciona. The notable characteristic of spontaneous motor behavior is its rhythmicity (30). In vertebrates, the emergence of spontaneous network activity has been characterized by Ca²⁺ imaging as a transition from first sporadic and then cell-autonomous activity to coordinated rhythmic activity (e.g., zebrafish, and chicken) (9, 31, 32).

To understand how rhythmic activity and spontaneous motor behavior are acquired in the motor circuit development of Ciona, we visualized the calcium and/or membrane potential activity of the motor neuron before the matured larva. We then used optogenetic tools to investigate coordination with the early spontaneous motor behavior [early tail flick (ETF)], which is thought to mediate hatching from the egg envelope (33).

Here, we show that a single pair of motor neurons exhibits rhythmical bursting in the motor ganglion. We then determine the rhythm of the early spontaneous motor behavior during chordate early embryogenesis.

RESULTS

Ca²⁺ oscillation first observed in a pair of single cells

We first observed the Ca²⁺ transient in the whole embryo in tailbud stage. The GCaMP6s-encoded mRNA was injected into the unfertilized eggs, which allowed us to investigate the Ca²⁺ transient at least from gastrula (St. 10) to juvenile (St. 36) (34, 35). Oscillation of Ca²⁺ transients (Ca²⁺ oscillation) was observed symmetrically on the left and right sides of the motor ganglion (Fig. 1A). No reproducible Ca²⁺ oscillation was observed in any other cells from the early- to mid-tailbud stage (St. 20 to St. 22; Fig. 1B, white arrowhead, and movie S1).

Fig. 1. Ca²⁺ oscillation is observed in the motor ganglion at tailbud stage.

(A) Merged images of Ca²⁺ imaging at different time points. Ca²⁺ transients were observed on the left and right sides of the motor ganglion, respectively (yellow box; enlarged view of the white box, inset). L, left; R, right. (B) The representative time course image of the Ca²⁺ transient in the motor ganglion at St. 22. Ca²⁺ transient is indicated by a white arrowhead (second panel). The region of interest (ROI) for motor ganglion and other regions (for negative control) is indicated by black and red circles, respectively. The image in the motor ganglion (white rectangle) is magnified in the right bottom, respectively. Dashed white lines outline the embryo. (C) Fluorescence intensity of Ca²⁺ in the ROI. The ROI for the black graph is indicated in (B) (black circle). The ROI for the red graph is indicated in (B) (red circle). A.U., arbitrary units. (D) Transition of the interval between Ca²⁺ oscillations over time. As the count of Ca²⁺ transients increases, the interval between Ca²⁺ oscillations gradually decreases. The period when Ca²⁺ oscillation in the motor ganglion and tail muscle contraction coincide is indicated by an orange window. Approximate developmental stage is indicated.

In mid-tailbud II (St. 22), the duration and interval of the Ca²⁺ oscillation on one side were 23 ± 4 s (mean ± SD) and 80 ± 4 s (mean ± SD), respectively (Fig. 1, C and D). The interval of Ca²⁺ oscillations varied at St. 22 but gradually decreased to about 25 s toward late tailbud II (St. 24) (Fig. 1D). The duration of Ca²⁺ oscillation also gradually decreased: 12 ± 2 s (mean ± SD) at St. 23 and 10 ± 2 s (mean ± SD) at St. 24.

To investigate how many cells are involved in the Ca²⁺ oscillation, we expressed the nuclear-localized Ca²⁺ sensor H2B-GCaMP6s in the entire embryo. Time-lapse imaging of the lateral side by confocal microscopy revealed that Ca²⁺ oscillation occurred only in a single nucleus (Fig. 2A and movie S2); no other cells exhibited reproducible Ca²⁺ transients at the mid-tailbud II (St. 22), indicating that the Ca²⁺ oscillation was derived from only a single pair of cells at mid-tailbud II (St. 22).

Fig. 2. Ca²⁺ oscillation occurs in a pair of cells at St. 22 and coincides with Ca²⁺ elevation in tail muscle from St. 23.

(A) Three-dimensional (3D) reconstruction (left) and section (right) images of the Ca²⁺ oscillation by H2B-GCaMP6s. Ca²⁺ oscillation is indicated by the white arrowhead. Ca²⁺ oscillation occurs only in a single nucleus. Dashed white lines outline the embryo. N = 2 (H2B-GCaMP6s–expressing embryos). The data were collected from CLSM and FM. A, anterior; D, dorsal; P, posterior; V, ventral. (B) Bright-field image of a dissociated single cell (left) and time course fluorescence images of Ca²⁺ oscillation in the cell (right). The position of the cell is indicated by a white arrowhead. N = 6. (C) Change in the fluorescence intensity of Ca²⁺ in (B) over 2000 s. (D) Representative time course image of Ca²⁺ oscillation at St. 23. After a few minutes, the Ca²⁺ oscillation (white arrow, 8 s) coincides with the Ca²⁺ elevation in the ipsilateral muscle cells (black arrow, 202 s). (E) Fluorescence intensity over time in the motor ganglion and tail muscle at St. 23. ROI was set in (D) at 8 s (yellow dotted line for motor ganglion and blue dotted line for tail muscle cell). The time when the Ca²⁺ oscillation and Ca²⁺ elevation of ipsilateral muscle cells first coincided is indicated by a blue rectangle. The asterisks indicate Ca²⁺ oscillation without synchronization of the Ca²⁺ elevation in ipsilateral muscle cells after the first synchronization.

To determine whether a single cell can independently exhibit Ca²⁺ oscillation, we dissociated tailbud embryos expressing GCaMP6s into single cells and observed Ca²⁺ levels (Fig. 2B and movie S3). Unexpectedly, time-lapse imaging revealed a cell-autonomous Ca²⁺ oscillation at an interval of 40 to 80 s (Fig. 2C). Thus, the Ca²⁺ oscillation during the tailbud stage was caused by a single cell acting independently without any contribution from other cells.

Ca²⁺ oscillation coincided with the Ca²⁺ elevation in the ipsilateral tail muscle

Continuous whole-embryo Ca²⁺ imaging from St. 22 revealed that the Ca²⁺ oscillation began at the same time as Ca²⁺ elevation in the ipsilateral tail muscle from late tailbud I (St. 23) (Fig. 2D). Earlier Ca²⁺ oscillations in the motor ganglion sometimes did not elevate Ca²⁺ in the tail muscle (Fig. 2E, asterisks) but fully synchronized 10 min after the first synchronization (first synchronization; Fig. 2D, at 202 s; Fig. 2E, blue box). The tail muscle contraction (observed as the slight movement with Ca²⁺ elevation in the tail muscle) was synchronized with the ipsilateral Ca²⁺ oscillation in the motor ganglion (Fig. 1D, orange window; for example, at 2 hours 30 min in movie S4). The interval is relatively constant (interval, 25 to 70 s) during the period of synchronization between the Ca²⁺ oscillation in the motor ganglion and the tail muscle contraction starting from St. 23 (Fig. 1D). This implies a rhythmic contractile behavior. These results suggest that the pair of neurons exhibiting Ca²⁺ oscillation are motor neurons and that these cells regulate the contractions of tail muscle after late tailbud I (St. 23).

Cell lineage tracking revealed a pair of cells as MN2/A10.64

We investigated the cell lineage of a pair of cells exhibiting Ca²⁺ oscillation. We considered that the largest motor neurons A10.64 corresponded to this pair of cells (Fig. 3A). Simultaneous imaging of Neurog::Kaede-NLS and CiVACHT::GCaMP6s can track the cell lineage of A9.31 and A9.32 (mother cell of A10.64; Fig. 3, B and C) (36). The cell lineage tracking revealed that a pair of Kaede nucleus–localized signals of A10.64 appeared exclusively in the motor ganglion at late tailbud I (St. 23) (Fig. 3D, 137 min). Last, we confirmed that the Ca²⁺ oscillation in the cytoplasm overlapped with the Kaede nucleus–localized signal of A10.64 (Fig. 3E, yellow arrowhead, and movie S5). On the basis of these findings, we identified cells exhibiting Ca²⁺ oscillation as A10.64 and named as MN2 cholinergic motor neurons (20, 22).

Fig. 3. Ca²⁺ oscillation overlaps with the nucleus-localized signal of MN2/A10.64.

(A) 3D reconstruction of a Ciona embryo at mid-tailbud II (St. 22) modified from previous study (60). The region of the motor ganglion (yellow rectangle) is magnified in the right. Some cell lineages of cells including A10.64 are indicated in the right. (B) Schematic illustrations of the late gastrula (St. 13) (left) and the cell lineage of the neural plate cells (right). A9.32 cells are indicated in green color. The first and second rows of neural plate cells are indicated by light green color. (C) Late gastrula embryo expressing Kaede under the control of the Neurogenin promoter. Kaede was immunostained with antibodies (magenta). Nuclei are stained with DAPI (green). A9.32 and A9.31 cells are identified via the relative position of neural late cells (B). A9.32 and A9.31 cells overlapped with immunostained signals from Kaede. (D) Cell lineage tracking of A9.32 cells from late gastrula (St. 13) to late tailbud I (St. 23). The embryo was electroporated with Neurog::Kaede-NLS and pSP-CiVACHT::GCaMP6s. A10.64 migrates anteriorly and is localized to the motor ganglion at late tailbud I (St. 23) (22). Dashed white lines outline the embryo. (E) Representative time course images of Ca²⁺ oscillation at St. 23. The A10.64 is labeled with Kaede-NLS (white arrowhead). Ca²⁺ oscillation overlapped with the nuclear signal from Kaede-NLS (yellow arrowhead). Enlarged views of A10.64 are embedded in each panel (inset). Dashed white lines outline the embryo. N = 6.

Tail muscle contraction gradually coincided with the spikes of MN2/A10.64

We performed simultaneous recordings of Ca²⁺ and membrane potential of A10.64 using the red fluorescent Ca²⁺ indicator NES-jRGECO1a (37) and the voltage sensor ASAP2f (38) under the expression of Neurogenin promoter. In the Ca²⁺ transient (Fig. 4A and fig. S1, red graph), we observed burst firing of membrane potential in MN2 (Fig. 4A and fig. S1, green graph). Each burst consisted of multiple spikes (Fig. 4A and fig. S1, asterisks). The timing and interval of the bursts were consistent with those of the Ca²⁺ oscillations because every burst was accompanied by a single Ca²⁺ transient (Fig. 1D).

Fig. 4. The number of spikes and interspike interval in a burst decreased from St. 23 to St. 24.

(A) Representative fluorescence intensity of Ca²⁺ (red) and membrane potential (green) in MN2 at St. 23 (top) and St. 24 (lower). The fluorescence intensity of ASAP2f decreases upon depolarization (38). Asterisks indicate the timing of depolarization. (B) Box plot of numbers of spikes in bursts in MN2. Data are presented as means ± SD [n = 7, 12, and 11 for tubocurarine 0 (St. 23), 5 (St. 23), and 5 mM (St. 24), respectively; *P < 0.05; Wilcoxon rank-sum test; Bonferroni multiple comparison test]. (C) Box plot of interspike intervals in MN2. Data are presented as means ± SD [n = 7, 11, and 11 for tubocurarine 0 (St. 23), 5 (St. 23), and 5 mM (St. 24), respectively; ***P < 0.001; Wilcoxon rank-sum test; Bonferroni multiple comparison test]. n.s., not significant.

The average number of spikes in a burst was 6.3 ± 5.4 at late tailbud I (St. 23). This decreased significantly to 2.6 ± 1.4 at late tailbud II (St. 24) (P < 0.05) (Fig. 4B). The interspike interval in a burst was 0.5 ± 0.3 s at late tailbud I (St. 23) and decreased significantly to 0.2 ± 0.04 s at late tailbud II (St. 24) (P < 0.05) (Fig. 4C).

As we mentioned above, tail muscle contraction coincided with the Ca²⁺ elevation in the tail muscle. This suggested that Ca²⁺ oscillations in the MN2 correlate with the tail muscle contractions in St. 23 (Fig. 2, D and E). However, the relationship between the spikes of MN2 and muscle contractions is unclear.

Considering that the interval of the Ca²⁺ oscillation in MN2 and Ca²⁺ elevation in the muscle gradually decreased from 80 s (St. 23) to 25 s (St. 24) (movie S4 and Fig. 1D), the interval of muscle contractions would also decrease from 80 s (St. 23) to 25 s (St. 24). To observe muscle contraction in detail, we used a high-speed camera to record the muscle contraction both on the left and right sides at St. 23 and St. 24 (Fig. 5, A and B). We referred to each set of unilateral muscle contractions as an ETF. The average intervals between ETFs were 79 ± 14 s (mean ± SD) at St. 23 decreasing to 21 ± 6 s (mean ± SD) at St. 24. This is comparable to the change in the Ca²⁺ oscillation in MN2 between St. 23 and St. 24 (80 to 25 s; Fig. 1D).

Fig. 5. Muscle contraction couples with the neuronal activity of MN2 at St. 24.

(A) Three merged images of late tailbud I (St. 23, top) and late tailbud II (St. 24, bottom) as captured by a high-speed camera. The midlines of the embryos are indicated in white (resting state), red (the timing of right muscle contraction), and blue (the timing of left muscle contraction). (B) The time course of the curvature calculated from the midline at late tailbud I (St. 23, top) and late tailbud II (St. 24, bottom). Red and blue bars indicate representative right and left muscle contractions, respectively. Magnified time course of curvature at red and blue bars are indicated in the right graphs, respectively. Asterisks indicate the timing of tail muscle contraction. (C) Box plot of the number of muscle contractions in one ETFs at St. 23 and St. 24. Data are presented as means ± SD [n = 9 (St. 23) and 13 (St. 24) from three and two embryos respectively; **P < 0.01; Wilcoxon rank-sum test].

Tail muscle contraction occurred only once in each ETFs at St. 23; two or three consecutive muscle contractions occurred in each ETFs at St. 24 (Fig. 5B). The number and frequency of tail muscle contractions in each ETFs at St. 24 (Fig. 5, B and C) were comparable to those of spikes in the burst at St. 24 (Fig. 4, B and C). These observations indicated that muscle contractions in each ETFs gradually correlated with the individual spike in the burst of MN2 as the embryo developed toward St. 24.

MN2/A10.64 was necessary and sufficient for rhythmic tail muscle contraction

We hypothesized that the neuronal activity of the MN2 would be necessary and sufficient to regulate the ipsilateral ETFs at least until St. 24. To test this hypothesis, we performed single-cell photoablation of MN2 using an Light-Oxygen-Voltage (LOV)–based optogenetic tool miniSOG2, which produces singlet oxygen under laser irradiation (39). We targeted the mCherry-CAAX signal of A10.64 and expected that penetration of the laser would completely suppress the ETFs (Fig. 6, A and B). In contrast to the control embryo (electroporated only with Neurog::mCherry-CAAX), the embryo expressing miniSOG2 did not exhibit ETFs at St. 24 (Fig. 6, C and D). Early motor behavior was comparable to that in control embryos in the absence of 440-nm laser irradiation (fig. S2).

Fig. 6. Single-cell photoablation for MN2 at St. 22 abrogates the ETFs until St. 24.

(A) Schematic illustration of the scheme for photoablation of MN2. MN2 is illustrated in red. A 440-nm laser was applied from the lateral side. The upper and lower sides of MN2 are marked by black and blue asterisks, respectively. A, anterior; D, dorsal; P, posterior; V, ventral. (B) Fluorescence image (left) and merged image with differential interference contrast (DIC; right) at mid-tailbud II (St. 22) in an embryo electroporated with Neurog::mCherry-CAAX. The signal of mCherry-CAAX in MN2 is indicated by a white arrowhead. (C) Time course of the curvature as calculated from the midline for 120 s. Blue line indicates the curvature of embryos electroporated with Neurog::mCherry-CAAX (for control). Red line indicates the curvature of embryos electroporated with Neurog::miniSOG2-CAAX and Neurog::mCherry-CAAX. Magnified time course of curvature (black rectangles) is indicated in the right graphs. The asterisks indicate a muscle contraction. (D) The number of ETFs in 2 min at St. 24 under each condition. The upper side of MN2 [(A), black asterisk] is indicated as the “ablation side.” The lower side of MN2 [(A), blue asterisk] is indicated as the “opposite side.” Data are presented as means ± SD [N = 9 (electroporated with Neurog::miniSOG2-CAAX and Neurog::mCherry-CAAX) and 7 (electroporated with Neurog::mCherry-CAAX, for control); *P < 0.05; Wilcoxon rank-sum test; Bonferroni multiple comparison test].

These results collectively suggest that oscillatory neuronal activity of a single pair of cells, MN2, is necessary for ETFs until St. 24. The single-cell photoablation of MN2 abolished early spontaneous motor behavior until St. 24, but tail muscle contraction appeared at St. 26 (movie S6). The Ca²⁺ imaging by CiVACHT-H2B-GCaMP6s revealed that MN2 and another neuron [annotated as MN1 and MGIN2 (40)] are active in intact larva at St. 26 (fig. S3), suggesting that other motor neurons also regulate tail muscle contraction after St. 24 (10). However, the tail muscle contraction, which appeared at St. 26, did not show the rhythmic muscle contraction like swimming in our observation (St. 26; movie S6). This emphasizes the role of MN2 in the rhythm generation and swimming behavior.

We lastly investigated whether the stimulation of MN2 is sufficient to drive tail muscle contraction. We performed single-cell photostimulation of MN2 using Channelrhodopsin-2 (ChR2), which depolarizes neurons and evokes precisely timed action potentials (41). The Neurog::hChR2(E123T/T159C)-mCherry was induced into the embryos, and a 488-nm laser was applied for laser stimulation. Tail motion was observed during laser stimulation of the motor ganglion region (Fig. 7, A and B, corresponding to MN2, and movie S7). Tail motion was not frequently observed during laser stimulation of other regions (Fig. 7, C and D, and movie S7). We calculated the percentage of the tail that moved time during the laser stimulation in the motor ganglion and compared it with the other region (including hChR2 expressed area in other regions and no hChR2 expressed region). There was a significant difference in the percentage between the MN2 region and other regions (Fig. 7E). These results collectively suggest that the oscillatory activity of MN2 is necessary and sufficient to drive the tail muscle contraction/ETFs.

Fig. 7. Single-cell photostimulation for MN2 evokes tail motion.

(A) The representative image of embryo expressing Neurog::hChR2(E123T/T159C)-mCherry at stages later than St. 24. The image is merged with fluorescence image and DIC image. The region treated via laser stimulation was indicated by the white circle (corresponding to MN2). (B) The raster diagram of the tail moved over time in (A). Tail motion was observed in all trials. The tail moved time point is indicated by the blue colored point. The time window with laser stimulation applied was indicated by an orange window. (C) The representative image of embryo expressing Neurog::hChR2(E123T/T159C)-mCherry. The image is merged with a fluorescence image and a DIC image. The region treated via laser stimulation was indicated by the white circle (corresponding to other regions). (D) The raster diagram of the tail moved over time in (C). The tail-moved time point is indicated by the blue colored point. The time window with laser stimulation applied was indicated by an orange window. (E) The box plot of the percentage of tail movement time during the laser stimulation in the motor ganglion region (corresponding to MN2) and other regions (including hChR2-expressed area in other regions and no-hChR2-expressed region). Note that the tail moved spontaneously regardless of laser stimulation, and thus, the percentage may not be 0 in both regions. Data are shown as means ± SD [n = 12 (motor ganglion region) and 11 (other region) from five embryos; ***P < 0.001; Welch’s t test].

DISCUSSION

Development of Ciona early motor circuit

We describe how the first motor responses/ETFs emerge during development of motor circuit in Ciona. We found that MN2 first exhibits Ca²⁺ oscillation with an 80-s interval at St. 22 (Figs. 1, C and D, and 3E), and MN2 oscillates independently (Fig. 2, B and C). We further show the relationship between the Ca²⁺ oscillation and spikes of membrane potential in the early spontaneous motor behavior. On the basis of these results, we suggest that the early development of Ciona motor circuit proceeds as follows (Fig. 8A).

Fig. 8. The summary of the motor circuit in Ciona until St. 24 and its relation to the circuit of matured larva.

(A) The summary of the Ca²⁺ oscillation and membrane potential in MN2 and its relationship to the early spontaneous motor behavior from St. 22 to St. 24. Neural tubes (light blue), muscle (red), and MN2 (yellow) are indicated by black arrows. The asterisks and curved arrow indicate the number of muscle contractions in one Ca²⁺ oscillation or ETFs. (B) (left) Schematic picture of Ciona early motor circuit at St. 23 and St. 24. Ciona early motor circuit shows that MN2 first generates rhythm and regulates muscle contraction. L, left; Mu, muscle; R, right. (right) Schematic picture of motor circuit in matured larva modified from the literature (10, 11, 23). Independent rhythmic activity of MN2 in both the left and right sides in St. 23 and St. 24 probably becomes reciprocal by joining ACIN commissural neurons (19). Each name of the neuron is annotated. Colored lines indicate major chemical synapses. Dashed lines indicate major gap junctions (electrical synapses). Schematic illustration of Ca²⁺ oscillation is indicated by a red or blue colored line. CNS, central nervous system.

1. Coupling of Ca²⁺ oscillation with burst firing of membrane potential first occurs in MN2 from early to mid-tailbud II (St. 20 to St. 22) (Fig. 2A). The interval and the duration of the Ca²⁺ oscillation are approximately 80 and 23 s, respectively (Fig. 1D). No muscle contraction was observed at this stage.

2. In late tailbud I (St. 23), MN2 activity begins to coincide with the Ca²⁺ elevation of ipsilateral tail muscle (ETFs first observed) (Fig. 8A, asterisks). One tail muscle contraction is coupled with one burst firing and is associated with multiple spikes.

3. In late tailbud II (St. 24), both the interval and duration of each Ca²⁺ transient are shorter (approaching 25 s and less than 10 s, respectively; Fig. 1D). In the ETFs of St. 24, each tail muscle contraction coincides with each spike in a burst of membrane potential.

Here, the early spontaneous motor behavior of Ciona begins with the activation of the oscillatory motor neuron MN2. At later stages, muscle contraction of ETFs is gradually fine-tuned to each spike rather than each burst of membrane potential in MN2 (Fig. 8A).

The role of MN2 in swimming larva

MN2 is a necessary and sufficient component for mediating the ETFs at least until St. 24; thus, we propose the model of motor circuit at St. 23 and St. 24 (Fig. 8B, left). However, it is unclear how this simple motor circuit develops into a circuit of matured larva.

Prior work (10, 11, 23) indicates the motor circuit of matured Ciona larva (Fig. 8B, right). Matured larva has ACINs, and ACINs join the circuit. This is considered to be a necessary component for mediating the left and right alternative swimming behavior (19). The ddNs project axon moves contralaterally onto MN2, which may mediate the bilateral tail flick in matured larva (11). In the matured larva, another type of motor neurons like MN1 and interneurons MGINs are then initially considered to be active (fig. S3). However, it is unclear whether MN2 still has a role in this matured complex circuit. We consider that MN2 still plays an indispensable role in these complex circuits.

Oscillation in MN2 continued until larval stage (St. 26) (fig. S3), and the ablation of MN2 disrupted rhythmic normal left-right alternative swimming of larva (movie S6), which suggests that MN2 is necessary for the alternative swimming locomotion. Second, MN2 has the largest number of synapses on the muscle among five pairs of motor neurons in swimming larva (10), suggesting the main component for generating muscle contraction. Third, MN2 projects to the ACINs, which is thought to be necessary for alternative swimming (10). Together, MN2 are indispensable in mature alternative swimming behavior.

The behavior of matured larva becomes complex and exhibits several types of motor behavior (42–45). However, it has recently been reported that decelerated larvae only contain motor ganglion regions and exhibit periodic tail beating bursts with an interval of approximately 20 s (46). That interval is similar to the Ca²⁺ oscillation of MN2 at St. 24 (Fig. 1D). This result and fig. S3 suggest that the oscillatory activity of MN2 still plays a role in swimming behavior at the larval stage. In the hatched larva, it is possible that the oscillatory activity of MN2 may trigger periodic alternating tail beating unless the upstream region of CNS innervates MN2. The ddNs may be active together to generate tail beating in decerebrated larva and possibly mediate tail flick in addition to MN2 in matured larva. Overall, the oscillatory activity of MN2 continues until the larval stage, and it plays an indispensable role in the circuit of matured Ciona larva.

The comparison between Ciona and other chordata

The urochordate ascidian is the closest relative of vertebrate (13), and one of the main basal chordates in addition to the cephalochordate (amphioxus) and jawless vertebrate (lamprey). Thus, Ciona have shared enhancer machinery (47, 48) and gene expression patterns (49) with amphioxus and lamprey. Ciona also has similar characteristics as motor circuits with amphioxus and lamprey. For instance, the motor circuit that generates rhythmic bursts for swimming are located in the hindbrain in lamprey (50). Commissural neurons are implicated as GABAergic neurons in amphioxus (51). These characteristics are consistent with Ciona.

However, there are some differences in the network of motor circuits. In the lamprey, motor neurons bundle into ventral roots and are the final output innervating muscle (52). The lamprey has each body segment connected by an excitatory glutamatergic interneuron (53). In contrast, mature larvae of Ciona have MN2 motor neurons that ipsilaterally project axons to the ACINs (Fig. 8B, right), suggesting that motor neurons are not merely the final output but rather are incorporated into the locomotor circuit producing the alternate left-right swimming-like muscle contraction.

In this study, we found that the oscillatory activity of MN2 is rhythmic in early motor circuit development. A recent report in zebrafish suggests that motor neurons are the first population of neurons to be active in the spinal cord of vertebrates (9). Future work will study the neuronal activity of motor circuit development in different chordate animals to understand the evolution of the chordate motor neural circuit.

MATERIALS AND METHODS

Experimental Animals

Ciona robusta (Ciona intestinalis type A) adults were obtained from Maizuru Fisheries Research Station (Kyoto University), Onagawa Field Center (Tohoku University), and Misaki Marine Biological Station (The University of Tokyo) through the National Bio-Resource Project, Japan. Eggs and sperm were collected by dissection of oviducts and sperm ducts, respectively. After artificial insemination, fertilized eggs were incubated at 15° to 20°C until observation.

Preparation of reporter constructs and gene transduction

pGP-CMV::GCaMP6s (Douglas Kim Lab) (54), pGP-CMV::NES-jRGECO1a (Douglas Kim Lab) (37), ASAP2f (Michael Lin lab) (38), pcDNA3.1::miniSOG2-T2A-H2B-EGFP (Xiaokun Shu lab) (39), and pAAV-Ef1a-DIO hChR2(E123T/T159C)-mCherry (Deisseroth Lab) (41) were purchased from Addgene (USA). pSP-CiVACHT::Kaede was purchased from C. intestinalis Transgenic line RESources (CITRES). Neurog::Kaede-NLS was made as follows: A 3.6-kb upstream region of the Ciona neurogenin gene (Neurog) was amplified by polymerase chain reaction (PCR) using a pair of primers (forward primer, AGGGATCCGGAAGAGGTGTTAGA; reverse primer, GGGGATCCATTTTGTAGCAAGAGC) and cloned into pSP-Kaede-NLS vector (55). The pSP-Neurog::Kaede was used for immunostaining of Kaede at late gastrula and was made as follows: A 3.6-kb upstream region of the C. neurogenin gene (Neurog) was subcloned into the Bam HI restriction site of the pSP-Kaede vector.

Micro-injection

mRNA encoding GCaMP6s was synthesized from pSPE3::GCaMp6s as previously described (34). By injecting mRNA encoding GCaMP6s into unfertilized eggs, we can observe Ca²⁺ transients in the whole embryo from gastrula (St. 10) to juvenile (St. 36) (34, 35). For synthesis of mRNA encoding H2B-GCaMP6s, the open reading frame of H2B was PCR amplified from Foxb::H2B-CFP (forward primer, TCTGAATTCAGGCCTATGGTTGCATCCAAA; reverse primer, GGCGACTGGTGGATCTTTTGAGCTGGTGTA), and GCaMP6s was PCR amplified from pSPE3::GCaMp6s (forward primer, GATCCACCAGTCGCCACCATGGGTTCTCAT; reverse primer, AGGCCTGAATTCAGATCTGCCAAAGTTGAG). The cloning reaction for pSPE3::H2B-GCaMp6s was performed using linearized PCR products and the In-Fusion HD Cloning Kit (Takara Bio). H2B-GCaMP6s mRNA was produced and precipitated using the mMESSAGE mMACHINE T3 kit (Life Technologies, Carlsbad, CA, USA). GCaMP6s or H2B-GCAMP6s mRNA was injected into dechorionated eggs at 0.5 μg/μl.

Electroporation

pSP-CiVACHT::GCaMP6s, pSP-Neurog::ASAP2f, pSP-Neurog::NES-jRGECO1a, pSP-Neurog::miniSOG2-CAAX, pSP-Neurog::mCherry-CAAX, and pSP-Neurog::hChR2(E123T/T159C)-mCherry were synthesized as described for pSPE3::H2B-GCaMp6s. Forty microliters of each plasmid construct (20 μl each for electroporation of two constructs) at 1000 ng/μl was combined with 360 μl of 0.77 M mannitol in 10% seawater. At 30 min after fertilization, eggs in 400 μl of this solution were placed in a cuvette for electroporation. After electroporation, eggs were washed with seawater and incubated until observation.

Immunostaining of late gastrula embryo with Kaede antibody and DAPI

Immunofluorescence staining (Fig. 3C) was carried out as described previously (56). To visualize the localization of fluorescent reporter proteins, polyclonal rabbit anti-Kaede (PM012; Medical & Biological Laboratories, Nagoya, Japan; for Kaede) was used as the primary antibody (diluted 1:1000). The secondary antibody was Alexa Fluor 594–conjugated anti-rabbit IgG (immunoglobulin G) (A11012; Thermo Fisher Scientific). Samples were mounted under a coverslip in 50% glycerol-PBST with mounting medium containing 4′,6-diamidino-2-phenylindole (DAPI; Vector Laboratories).

Microscopy

The embryos were observed by fluorescence microscopy with a three–charge-coupled device (3CCD; C7800-20, Hamamatsu Photonics) camera or by confocal laser scanning microscopy (CLSM). A Nikon inverted microscope (Nikon eclipse, IX71) with a 20× or 40× objective lens (LUCPlanFLN) was used for fluorescence imaging with a U-MWBV2 mirror unit (Olympus). A SOLA light-emitting diode light (Lumencor) was used as the light source, and fluorescence images were acquired with a 3CCD camera and the AQUACOSMOS software (Hamamatsu Photonics). For Ca²⁺ imaging, the time interval was set to 1 to 5 s per frame. For membrane potential imaging, the time interval was set to 20 ms per frame. An Olympus fv1000 microscope was used for CLSM imaging. Excitation was performed at 488-nm to visualize GCaMP6s and Kaede and at 559-nm to visualize mCherry. An Olympus 20× or 40× oil immersion lens was used.

Fluorescence image analysis

Fluorescence intensities in image data were analyzed using the AQUACOSMOS software. Numerical data were exported into Microsoft Office Excel 2016 (Microsoft, Redmond, WA, USA) for graph plotting. Statistical analyses used R (CRAN) and Excel 2016 (Microsoft, Redmond, WA, USA).

Dissociation of embryos

The GCaMP6s-encoded mRNA was injected to the unfertilized eggs. Embryos expressing GCaMP6s at the tailbud stage were dissociated by 1% trypsin in EGTA-containing artificial seawater for 5 min. EGTA-containing artificial seawater was prepared according to the previous protocol (57). Embryos were pipetted for 5 min to complete dissociation of individual cells. The embryos were observed by fluorescence microscopy with a 3CCD (C7800-20, Hamamatsu Photonics) camera.

Cell lineage tracking

The Z-stack images of the dorsal side of late gastrula embryos, electroporated with Neurog::Kaede-NLS and pSP-CiVACHT::GCaMP6s, were acquired every 10 min until late tailbud I (St. 23) using an Olympus fv1000 microscope. The photoconversion of Kaede was performed by bleaching with a SIM scanner by manual operation. The Kaede signal of a pair of A9.32 cells at late gastrula (St. 13) was photoconverted by irradiation with a 405-nm laser for a few seconds.

Membrane potential imaging

At St. 23, we performed membrane potential imaging because the region of interest (ROI) did not move significantly. The ROI moved significantly at St. 24. Therefore, the antagonist of nicotinic acetylcholine receptor, d-tubocurarine, was used to inhibit the motility of larva (58). Embryos at St. 24 were paralyzed with 5 mM d-tubocurarine 30 min before observation. At St. 23, d-tubocurarine affected neither the number of spikes in the burst nor the interspike interval, compared with the non–tubocurarine-treated embryos (Fig. 4, B and C).

Single-cell photoablation of MN2

Single-cell photoablation of MN2 was performed using miniSOG2, which produces singlet oxygen by laser irradiation. pSP-Neurog::miniSOG2-CAAX was used to express miniSOG2 at MN2, and pSP-Neurog::mCherry-CAAX was used to label the MN2 for laser irradiation. Laser irradiation was performed by an Olympus fv1000 microscope.

In mid-tailbud II (St. 22), pSP-Neurog::mCherry-CAAX was expressed exclusively in MN2 (Fig. 6B). Embryos (St. 22) electroporated with pSP-Neurog::miniSOG2-CAAX and pSP-Neurog::mCherry-CAAX were ablated from the lateral side by irradiation with a laser (440 nm, 15 μW/cm²) for 10 min (Fig. 6A).

Control embryos were electroporated with pSP-Neurog::mCherry-CAAX and were ablated from the lateral side by irradiation with a laser (440 nm, 15 μW/cm²) for 10 min (Fig. 6A). The ROI for laser irradiation was set to the region of mCherry fluorescence. Ablation was performed once per minute to correct the position of the ROI. Laser power density was measured by the optical power meter 3664 (Hioki) and optical sensor 9742 (Hioki). After laser irradiation, the behavior of the embryos was recorded using a high-speed camera (WRAYCAM-VEX230M, WRAYMER, Osaka, Japan) attached to a stereoscopic microscope (OLYMPUS, SZX12, Tokyo, Japan) for 2 min every 30 min until larval stage. Optical images were captured at 200 to 500 fps to quantify the temporal change in curvature associated with early spontaneous motor behavior.

Single-cell photostimulation of MN2

Single-cell photostimulation of MN2 was performed using ChR2, which depolarizes neurons and evokes precisely timed action potentials (41). pSP-Neurog::hChR2(E123T/T159C)-mCherry was used to express hChR2(E123T/T159C)-mCherry at MN2. Embryos at stage later than St. 24 were stimulated by a laser (488 nm). The ROI for laser stimulation of MN2 was set to the region of mCherry fluorescence in the motor ganglion. The ROI for laser stimulation of other regions was set either to the region of mCherry fluorescence in other regions or to no mCherry fluorescence regions. The recording time varies depending on the embryo from 15 to 60 s. The time for laser stimulation varies in each trial from 1.2 to 8 s. The serial images are acquired by CLSM every 0.4 s. The time resolution of the CLSM is not sufficiently high to capture the ETFs. Therefore, we manually judged whether or not the tail moved in each image. The data are summarized in a raster diagram (Fig. 7, B and D).

Quantification of the ETFs

To quantify early spontaneous motor behavior, recording data were processed using the open-source program Fiji (National Institutes of Health, Bethesda, MD, USA) and MATLAB (MathWorks). Image binarization and skeletonization (59) were processed by Fiji, and then skeleton pruning used a custom-made MATLAB script (MathWorks). The midline of the embryos at each time point was extracted, and the curvature of the midline was calculated using the position of three points on the midline: start (A), mid (B), and end (C). The equation to calculate the curvature was as follows.

$$\text{curv}.=\frac{1}{R}=\frac{2\text{sin} C }{C}=\frac{2}{\mid \overrightarrow{\mathbf{A}-\mathbf{B}}\mid }\frac{(\overrightarrow{\mathbf{A}-\mathbf{C}}) \times (\overrightarrow{\mathbf{B}-\mathbf{C}})}{\mid \overrightarrow{\mathbf{A}-\mathbf{C}}\mid \mid \overrightarrow{\mathbf{B}-\mathbf{C}}\mid }$$

Supplementary Materials

This PDF file includes:

Figs. S1 to S3

Legends for movies S1 to S7

Other Supplementary Material for this manuscript includes the following:

Movies S1 to S7

View/request a protocol for this paper from Bio-protocol.