Firing patterns of V1 neurons in larval zebrafish. a A side view of the compound transgenic fish of Tg[en1b:Gal4] and Tg[UAS:Kaede]. The dashed lines indicate boundaries of the spinal cord. Scale bar, 20 μm. b A schematic illustration of the simultaneous recordings of a V1 neuron (loose-patch) and ventral root (VR). Fictive swimming was elicited by applying brief electrical stimulation near the tail. c An example of the recordings from the fast-type V1 neurons. The blue arrowhead shows the time point of electrical stimulation. In the right panel, the region shadowed in blue in the left panel is enlarged. d An example of the recordings from the slow-type V1 neurons. e Histogram of spike timings of fast-type V1 neurons during fast (50–65 Hz) swim (1387 swimming cycles from 29 cells). f Histogram of spike timings of slow-type V1 neurons during slow (25–35 Hz) swim (5782 swimming cycles from 10 cells). g Schematic diagram of the Kaede photo-conversion experiment (top). The bottom two panels show the recordings from early-born V1 neurons (red Kaede [shown in magenta], left) and late-born V1 neurons (green Kaede, right). Scale bar, 10 μm. h Firing probability of early-born V1 neurons (left, n = 25) and late-born V1 neurons (right, n = 22) in each cycle during fast (>50 Hz) and slow (<40 Hz) swim. Each colored circle represents each recorded cell. i Classification of the recorded V1 neurons. Out of the 25 early-born V1 neurons, n = 23 for fast-type and n = 2 for hybrid-type. Out of the 22 late-born V1 neurons, n = 9 for fast-type, n = 10 for slow-type, and n = 3 for hybrid-type. If the value of the firing probability during fast swim was more than double that during slow swim, the cell was considered a fast-type (and vice versa). If the difference was within the doubled value, the cell was considered a hybrid-type

V1 ablation reduced cycle frequency in swimming. a Fluorescent images (green channel) of Tg[en1b:GFP] and Tg[en1b:loxP-RFP-loxP-DTA] fish with (bottom panel) or without (top panel) Tg[hoxa4a/9a:Cre]. Green fluorescent protein (GFP) expression in the spinal V1 neurons (arrow) is absent in the presence of Cre (En1-DTA), with GFP expression in the brain (triangle) and slow muscle cells in the middle region of the body (arrowhead) being intact. Scale bar, 250 μm. b, c Ventral root recordings of fictive swimming elicited by electrical stimulations (ESs) in a control (b) and an En1-DTA fish (c). d Swimming frequency of control and En1-DTA fish during the initial phase of ES swim (swimming elicited by electrical stimulation). Control: 54.8 ± 5.7 Hz, number of fish = 73. En1-DTA: 28.0 ± 2.9 Hz, n = 57. **P < 0.01 (two-tailed t test, P = 6.4 × 10⁻⁶²). e Swimming frequency of control and En1-DTA fish during Non-ES swim. Swim cycles with frequencies within 20–40 Hz were picked up and averaged. Control: 30.5 ± 2.3 Hz, n = 73. En1-DTA: 27.1 ± 2.2 Hz, n = 57. **P < 0.01 (two-tailed t test, P = 2.6 × 10⁻¹⁴). Data are mean ± s.d.

Activity of ventrally located V2a neurons in control and En1-DTA fish. a Lateral view of the spinal cord of Tg[chx10:GFP]. The horizontal dashed lines indicate the dorsal and ventral boundaries of the spinal cord. Ventrally located V2a neurons (the position from 0 to 0.5) are the subjects of the recordings (top panel). The bottom two panels show the images of a loose-patch recording. Scale bar, 20 μm. b An example of simultaneous recordings between ventrally located V2a neurons (loose-patch) and ventral root (VR) in control fish. c An example of simultaneous recordings between ventrally located V2a neurons (loose-patch) and VR in En1-DTA fish. d, e Numbers of spikes (d) and firing probability in each cycle (e) during fast/strong swim. Each circle represents each cell (n = 84 for ventrally located V2a neurons in control; n = 60 for ventrally located V2a neurons in En1-DTA). **P < 0.01 (Mann–Whitney U test, P = 1.3 × 10⁻¹⁰ [d], P = 3.7 × 10⁻¹⁴ [e]). Boxes represent the interquartile range (IQR) between first and third quartiles and the line inside represents the median. Whiskers denote the lowest and highest values within 1.5 × IQR from the first and third quartiles, respectively

Voltage-clamp recordings of fast and slow muscles in control and En1-DTA fish. a A schematic illustration of the simultaneous voltage-clamp recordings of two muscle cells (one for a fast muscle, the other for a slow muscle). Neuro-muscular transmission was partially blocked in these experiments, and portions of synaptic currents remained in the recording conditions. b An image after a recording. Slow and fast muscles were morphologically distinct and could be easily distinguished. Scale bar, 20 μm. c Example of simultaneous voltage-clamp recordings in control fish. d Example of simultaneous voltage-clamp recordings in En1-DTA fish. e Normalized currents in the slow muscle cells in the cycle when the fast muscle received strong inputs (>50% of the maximum). Five paired recordings were performed. Control: 0.44 ± 0.08, En1-DTA: 0.73 ± 0.04. Data are mean ± s.d. **P < 0.01 (two-tailed t test, P = 0.00091)

Phenotype of the compound transgenic fish of Tg [en1b:loxP-RFP-loxP-DTA] and Tg[hoxa9a-3'enhancer:Cre] (related to Figures 2 and 3) a, Fluorescent images of Tg[en1b:loxP-RFP-loxP-DTA] fish without a Cre driver (top panel), with Tg [hoxa9a-3'enhancer:Cre] (middle panel), and Tg[hoxa4a/9a:Cre] (bottom panel). Arrowheads indicate the boundary between the hindbrain and the spinal cord. Scale bar, 250 μm. b, VR and slow-type MN recordings during fictive swimming elicited by electrical stimulation in a control (top panel) and a compound transgenic fish of Tg[en1b:loxP-RFP-loxP-DTA] and Tg[hoxa9a-3'enhancer:Cre] (bottom panel).