Shared versus specialized glycinergic spinal interneurons in axial motor circuits of larval zebrafish

Abstract

The neuronal networks in spinal cord can produce a diverse array of motor behaviors. In aquatic vertebrates such as fishes and tadpoles, these include escape behaviors, swimming across a range of speeds, and struggling. We addressed the question of whether these behaviors are accomplished by a shared set of spinal interneurons activated in different patterns or, instead, involve specialized spinal interneurons that may shape the motor output to produce particular behaviors. We used larval zebrafish because they are capable of several distinct axial motor behaviors using a common periphery and a relatively small set of spinal neurons, easing the task of exploring the extent to which cell types are specialized for particular motor patterns. We performed targeted in vivo whole-cell patch recordings in 3 d post fertilization larvae to reveal the activity pattern of four commissural glycinergic interneuron types during escape, swimming and struggling behaviors. While some neuronal classes were shared among different motor patterns, we found others that were active only during a single one. These specialized neurons had morphological and functional properties consistent with a role in shaping key features of the motor behavior in which they were active. Our results, in combination with other evidence from excitatory interneurons, support the idea that patterns of activity in a core network of shared spinal neurons may be shaped by more specialized interneurons to produce an assortment of motor behaviors.

Figure 1.

A, Experimental setup showing placement of electrical stimulus, patch clamp and ventral motor root electrodes, and tungsten holding pins (black circles) on a 3 dpf zebrafish larva. B, Paired ventral motor root recordings (segments 9 and 17) of the escape response (arrow) show near simultaneous left side motor activity immediately following a right side electrical stimulus (*stimulus artifact). C, During swimming, rhythmic bursts of motor activity propagate from head to tail in response to an electrical stimulus 57 ms before the first motor burst (off the page to the left of trace, data not shown). D, Struggling involves a caudorostral propagation of motor activity with substantially longer interburst intervals (brackets) and burst durations than swimming. Stimulus occurred 153 ms before the first motor burst (data not shown). Panels B-D are from the same animal. Kinematics of swimming (E) and struggling (F) illustrating the difference in body amplitude and wavelength between the two behaviors. Digital images were captured with a high-speed video camera (250 frames/s) and outlines were generated using a custom written Matlab program. G, Scatter plot showing that frequency and duration of swimming (gray circles, n = 4847 trials) and struggling (gray squares, n = 393 trials) bursts from a single ventral motor root electrode are similar to paired ventral root recording of swimming (open circles) and struggling (black squares). Inset graphs show that average swimming burst duration was significantly shorter (p < 0.001) than struggling burst duration (5.6 ± 3.6 ms vs 26.2 ± 7.2 ms), and that average swimming frequency was significantly higher (p < 0.001) than struggling frequency (42.7 Hz ± 12.2 vs 18.5 Hz ± 3.5).

Figure 2.

CoLAs are active only during struggling. Ai, Depolarizing current injections showing tonic spiking with spike frequency adaption. Aii, Tracing in a lateral view of spinal cord (rostral to the left) illustrates the large, tear-drop shaped soma from which a prominent caudal dendrite extends. The ascending axon sends dorsal processes (white triangles) proximal to the soma and ventral processes (gray triangles) distal to the soma, presumably to contact postsynaptic targets in the mid-lateral region of the cord (black indicates contralateral portions of the axon). Bi, The average dorsoventral position of the somata in the cord is 71%, where 100% represents the dorsal edge of the spinal cord. Bii, Pie chart illustrating the percentage of cells active during a particular motor pattern or repertoire of motor patterns. CoLAs are active only during struggling (st) and not during swimming and escapes. Biii, Graph showing that CoLAs fire during all struggling trials elicited. C, Simultaneous whole-cell (bottom, segment 12) and ventral motor root (top, segment 15) recordings show that CoLAs depolarize but do not fire during an escape behavior (arrow) in response to an electrical pulse (*stimulus artifact). Note that an episode of swimming follows the escape response. D, During swimming CoLAs show a depolarized membrane potential but do not fire action potentials. E, CoLAs fire multiple spikes on top of depolarized plateau potentials during struggling, but stop firing when the motor pattern transitions to swimming. Stimulus occurs 87 ms before the first motor burst (to left of trace, data not shown). F, Plot of motor root duration and frequency along with average values ± SDs (inset graphs) for swimming (circles; tested duration = 6.4 ± 3.5 ms, tested frequency = 34.7 Hz ± 8.9) and struggling (squares; duration = 25.1 ± 6.7 ms, frequency = 18.3) for all cells for all trials. Black fill indicates that the cell fired an action potential during the behavior while a gray outline indicates that the cell did not fire. CoLAs fire during struggling over its burst range (filled squares) but not during swimming (open circles). All electrophysiology traces are from the same cell.

Figure 3.

Tracing of a CoLA (A) and CoBL (B) interneuron along with their potential postsynaptic motoneuron, generated from a confocal z-stack image of a wild-type embryo coinjected with the Hb9 GFP and Glyt2 DsRed construct. Close-up images (arrows) from single 0.85 micron slices show boutons of the ascending glycinergic axons of each neuron (red) surrounding the motoneuron soma (green).

Figure 4.

CoLos are active only during escapes. Ai, Depolarizing current injections showing a single, small amplitude spike (arrow) once threshold is reached, probably small because the spike-initiation site is located relatively far away from the soma (see Results, Commissural local interneurons). Aii, Tracing of a CoLo in a lateral view of spinal cord (rostral to the left) showing a short descending axon with multiple processes, which runs parallel to the Mauthner axon (black portions are contralateral). Bi, The average dorsoventral position of CoLos in the spinal cord is 71%, where 100% represents the dorsal edge of cord. (ii) Pie chart illustrating that CoLos fire only during escapes (e) and not during swimming (sw) or struggling (st). Note that a fraction of CoLos showed no activity during any motor pattern in the trials examined. (iii) Graph showing that CoLos fired 55% of the time that an escape behavior was elicited (averaged over the 30 cells that showed activity). C, Simultaneous motor root and whole-cell recordings of a CoLo firing small-amplitude action potentials during an escape behavior (arrow) in response to an electrical pulse (*stimulus artifact). CoLos do not fire during swimming (D; stimulus occurs 59 ms before first motor burst) or struggling (E; stimulus occurs 28 ms before first motor burst)). F, Plot of tested motor root burst duration and frequency along with average values ± SDs (inset graphs) for swimming (open circles; duration = 6.8 ± 3.0 ms, frequency = 38.6 Hz ± 11.7) and struggling (open squares; duration = 27.7 ± 6.8 ms, frequency = 17.7 Hz ± 3.6) for all cells in all trials to show the range tested, over which no activity was found. All electrophysiology traces are from the same cell.

Figure 5.

CoSAs are multifunctional. Ai, Depolarizing current injections show tonic spiking near threshold and above. Aii, Tracing in a lateral view of spinal cord (rostral to the left) showing ipsilateral dendrites distal to the soma. The main axon hooks caudally before crossing to ascend (black portions are contralateral). Bi, Dorsoventral position of the somata in the cord, where 100% represents the dorsal edge of the spinal cord. CoSAs have the most dorsal positions in the cord (80%) of all cells in this study. Bii, As a group, CoSAs are active during all three motor patterns tested, with the majority of cells firing to both swimming (sw) and struggling (st) and a smaller percentage during escapes (e). Biii, Graph showing the percentage of times that CoSAs fired when escape (19%), swimming (52%) and struggling (62%) motor patterns were elicited. C, Simultaneous motor root and whole-cell recordings of a CoSA firing during an escape behavior (arrow) elicited by an electrical pulse (*stimulus artifact), followed by swimming. D, CoSAs usually fire a single spike at the beginning of slower swimming frequencies (stimulus occurs to left of trace, 63 ms before first motor burst, data not shown), in contrast to multiple spikes fired during struggling (E). F, Plot of motor root duration versus frequency for swimming (circles) and struggling (squares) for all cells for all trials. Black filled symbols indicate that the cell fired, while gray outlined symbols indicate that the cell did not fire. Due to the heterogeneity of CoSA functionality, these values overlap frequently. Average duration and frequency values ± SDs (inset graphs) are shown for the trials in which the cell fired an action potential during swimming (duration = 7.6 ± 4.4 ms, frequency = 43.1 Hz ± 13.4) and struggling (duration = 28.4 ± 6.4 ms, frequency = 18.8 Hz ± 3.6). All electrophysiology traces are from the same cell.

Figure 6.

CoBLs are largely shared between swimming and struggling. Ai, Depolarizing current injections showing tonic spiking once threshold is reached. Aii, Tracing of a CoBL in a lateral view of spinal cord (rostral to the left) showing a characteristic contralateral bifurcation of the main axon into an ascending and descending projection (black portions are contralateral). Bi, CoBL somata are distributed widely across the dorsal half of the spinal cord with an average position of 73%, where 100% represents the dorsal edge of the spinal cord. Bii, Pie chart showing that the majority of CoBLs (88%) are active during both swimming and struggling. Biii, Graph showing the percentage of times a cell fires during escapes (e; 3%) swimming (sw; 49%) and struggling (st; 81%). C, Simultaneous motor root and whole-cell recordings during an escape behavior (arrow) initiated immediately after an electrical pulse (stimulus artifact). The CoBL does not fire in the initial escape burst or during the subsequent low frequency swimming episode. D, The CoBL fires action potentials that ride on top of a depolarized potential during the start of a swimming bout (relatively high frequency) but stops firing at lower swimming frequencies at the end of the bout (see Results, Activity in commissural bifurcating interneurons, for explanation). Stimulus is located to the left of trace (data not shown), 75 ms before first motor burst. E, The CoBL fires a long series of spikes during struggling bursts. Stimulus is located to the left of trace (data not shown), 53 ms before the first struggling motor burst. Duration of activity and spike number decreases as the motor pattern transitions from struggling (first four bursts) to swimming, and spiking stops completely during lower frequency swimming. F, Plot of ventral motor root burst duration versus frequency for swimming (circles) and struggling (squares) for all cells for all trials. Black filled symbols indicate that the cell fired, while gray outlined symbols indicate that the cell did not fire. CoBLs were active over the range of burst characteristics for both swimming (duration range = 1–24 ms and frequency range = 15–75 Hz) and struggling (duration range = 13–46 ms and frequency range = 12–26 Hz), and were most reliably activated at higher swimming frequencies. Inset graphs show that the average value ± SD for swimming duration is shorter than for struggling (6.3 ± 3.7 ms vs 26.0 ± 7.9 ms). Average swimming frequency is higher than struggling (41.5 Hz ± 13.4 vs 18.8 Hz ± 3.4). All electrophysiology traces are from the same cell.

Figure 7.

Current summary of inhibitory (white filled boxes) and excitatory (gray filled boxes) spinal interneuron activity during different axial motor patterns in larval zebrafish. The top five rows (bold box) illustrate commissural cell types where black fill represents the contralateral portions of the axon. The bottom two rows are ipsilateral interneurons. When known, transmitter phenotype and special attributes are indicated. Two interneuronal classes, both inhibitory, are each active during only one of three motor patterns tested in this study (CoLAs for struggling and CoLos for escapes). Some inhibitory (CoSAs, CoBLs) and excitatory (CiDs) interneurons are multifunctional and shared between all motor patterns, although their contribution can vary with frequency within a motor pattern. MCoDs are active in slow, but not fast swimming (McLean et al., 2007); their activity during escape and struggling are currently unknown. Circumferential ascending interneurons (CiAs) are inhibitory and make monosynaptic connections with sensory pathway neurons (Higashijima et al., 2004c) and motoneurons, potentially playing a role in sensory gating and burst termination. CiAs are active during swimming, but their contribution to struggling and escape behaviors remains unknown.