a Inverted confocal images from whole-mount adult zebrafish spinal cord hemisegments showing cycling (BrdU⁺) cells in control animals (untrained), following 2 weeks of training and 2 weeks rest after training. b Similar distribution pattern of BrdU⁺ cells in the spinal cord comparing untrained, trained, and resting zebrafish. c Quantification of BrdU⁺ cells per hemisegment in different conditions show that the enhanced proliferation after training is reversible (P = 4.418E-10). d Expression pattern of her4.1:GFP (NSPCs; green) in close apposition of the adult zebrafish spinal cord’s central canal. e The vast majority (~97.5%) of the her4.1⁺ cells (green) express the stem cell marker Sox2 (magenta). Arrowheads indicate double-labeled cells. f Cycling her4.1⁺ radial glia cells (BrdU⁺, magenta; GFP, green). Training increased the number of BrdU⁺/her4.1⁺ cells per hemisegment. g Quantification of the average BrdU⁺ cells per spinal cord section co-expressing neuronal markers (mef-2, HuC/D, or NeuN) in untrained (control) and trained animals. h Proportions of BrdU⁺ cells expressing neuronal or glial markers are similar comparing untrained and trained animals. Quantification is based on the early neuronal marker mef-2. BrdU, 5-bromo-2ʹ-deoxyuridine; CC, central canal; GFP, green fluorescent protein; her4.1, hairy-related 4, tandem duplicate 1; HuC/D, elav3 + 4; mef-2, myocyte enhancer factor-2; NeuN, neuronal nuclei; NSPC, neural stem/progenitor cell; Sox2, sex-determining region Y-box 2. Data are presented as mean ± s.e.m. or as box plots showing the median with 25/75 percentile (box and line) and minimum–maximum (whiskers). ^**P < 0.01; ^***P < 0.001; ^****P < 0.0001; ns, not significant. For detailed statistics, see Supplementary Table 1.

a Images of a NSPC (her4.1:GFP⁺, white arrowhead) close to the spinal cord’s central canal. Current steps do not produce action potentials in NSPCs. Ex vivo setup of the brain-spinal cord preparation allows simultaneous recordings of a spinal cord NSPC and ipsilateral motor nerves (Nv). Electrical stimulation (10 pulses at 1 Hz) of the descending inputs elicits a swimming episode. In the absence of swimming, NSPCs do not respond (top traces). During a fictive locomotor episode, NSPCs periodically receive strong inputs (bottom traces). b NSPC responses during swim phase and the locomotor cycle during simultaneous ipsilateral and contralateral recordings. c Graph showing the activity of different NSPCs as a function of instantaneous swimming burst frequency. Individual data points represent instantaneous swimming frequencies of all swimming cycles where the respective NSPCs responded. d No apparent correlation between the amplitude of the periodic NSPC responses and the swimming frequency. e Correlation (R² = 0.9098) between the swim duration and the detected number of inputs to NSPCs. The dashed gray line represents the baseline. NSPC, neural stem/progenitor cell. For detailed statistics, see Supplementary Table 1.

a Bath application of ACh induced inward currents in all recorded NSPCs (20 out of 20). Sample traces of the muscarine- and nicotine-induced inward currents in recorded NSPCs. Significant reduction of the induced ACh currents in the presence of the α7 nicotinic receptor antagonist MLA (10 μM). Quantification of the frequency (Hz; P = 3.237E-5) and amplitude (pA; P < 0.0001) of the recorded cholinergic currents. b Exogenous application of GABA induced tonic activation of NSPCs (22 out of 22). GABAergic tonic responses were completely abolished in the presence of the GABA_A receptor antagonist, gabazine (10 μΜ). Exogenous application of muscimol (GABA_A receptor agonist) induced tonic activation of NSPCs. Quantification of the amplitude (pA; P = 7.904E-8) and duration (s; P = 1.141E-7) of the GABA-related responses. c Schematic protocol for NSPC recordings during local electrical stimulation. Ten pulses (20 Hz) were applied to increase the probability of presynaptic release. Superimposed representative sample trace (in red) out of >40 sweeps (in black) from NSPC responses under control conditions, following application of the polysynaptic blocker mephenesin, and application of the selective nicotinic receptor antagonist MLA suggesting synaptic cholinergic, but not GABAergic activation of the NSPCs. Quantification of the average number of detected EPSCs per sweep and the average amplitude of the responses in control and after the application of polysynaptic blocker (mephenesin). d Application of MLA during locomotion abolishes the regular and strong input to NSPCs, implying a predominant role of nicotinic receptors. ACh, acetylcholine; GABA, γ-aminobutyric acid; MLA, methyllycaconitine; NSPC, neural stem/progenitor cell; Nv, motor nerve recording. The dashed gray line represents the baseline. Data are presented as mean ± s.e.m. and as violin plots. ^***P < 0.001; ^****P < 0.0001; ns, not significant. For detailed statistics, see Supplementary Table 1.

a Large and dorsally located spinal cord V2a-INs (Chx10:GFP⁺, green) are cholinergic (ChAT⁺, magenta). Arrowheads indicate double-labeled neurons (Chx10:GFP⁺ChAT⁺). b A sample stack from the central canal area showing the presence of V2a-IN (Chx10:GFP⁺) axonal collaterals (green) close to CB⁺ NSPCs (magenta) with analysis of the proportion of the CB⁺ NSPCs that are in close proximity with the V2a-IN (GFP⁺) processes. c Quantification of the probability of the V2a-IN axonal collaterals in the central canal region (n = 15 zebrafish). d Representative whole-mount confocal image showing that all (25 out of 25; 100% from 8 zebrafish) long descending (dextran tracer, blue; >10 segments) cholinergic (ChAT⁺, red) neurons are V2a-INs (GFP⁺, green). Arrowheads indicate triple-labeled neurons. e Quantification and analysis of the number, size and location of long descending cholinergic V2a-INs in the adult zebrafish spinal hemisegments. f Sample average (~25 sweeps) traces from dual electrophysiological recordings between a premotor V2a-IN and NSPCs, located in the same segment (intra-segmental, 1) or 5–6 segments rostrally (inter-segmental, 2). Cholinergic connections were observed in the inter-segmental pairs (22%, 11 out of 50 pairs) but not in the intra-segmental pairs (0%, 0 out of 15). g Postsynaptic responses in the recorder NSPCs generated from suprathreshold (black, with action potential) and not from subthreshold (gray, without an action potential) short pulse depolarization of the V2a-IN. h Ex vivo setup of the brain-spinal cord preparation allows simultaneous recordings of a NSPC and ipsilateral descending V2a-INs during fictive locomotion. Sample trace of a connected pair that, while the V2a-IN discharges during fictive swimming, the NSPC receives occasional input. i Illustration of recordings acquired during electrical stimulations. Ten pulses (20 Hz) of a rostral spinal cord segment were applied to depolarize V2a-INs connected to NSCs. Representative sample trace (in red) of superimposed sweeps (~20, in black) from not responding and responding NSPCs. Bath application of the selective nicotinic antagonist MLA (10 μM) abolished the recorded currents suggesting that they are cholinergic. Changes in the proportion of the recorded NSPCs that respond to electrical stimulation observed after training (n: number of recorded NSPCs). The average number of detected events per stimulation sweep from both sites was significantly higher in trained animals, suggesting adaptive changes in the innervation and the cholinergic release to the NSCs (P < 0.0001). The dashed gray line represents the baseline. CB, calbindin D-28K; CC, central canal; ChAT, choline acetyltransferase; EPSC, excitatory postsynaptic current; INs, interneurons; MLA, methyllycaconitine; NSPC, neural stem/progenitor cell. Data are presented as mean ± s.e.m., as violin plots and as box plots showing the median with 25/75 percentile (box and line) and minimum–maximum (whiskers). ^****P < 0.0001. For detailed statistics, see Supplementary Table 1.

a Microphotographs and analysis show that nicotine and gabazine increased the her4.1⁺ (green)/BrdU⁺ (magenta) cells, whereas GABA reduced the number of her4.1⁺/BrdU⁺ cells in the examined spinal hemisegment (P < 0.0001). b In vivo administration of ACh, muscarine, nicotine, and gabazine increased the number of BrdU⁺ cells per hemisegment (P < 0.0001). Administration of GABA reduced the number of BrdU⁺ cells in the adult zebrafish spinal cord hemisegment (P < 0.0001). c Co-administration of nicotine and gabazine generated the same number of BrdU⁺ cells as the individual administration of nicotine or gabazine. Co-activation of the nicotinic-ACh receptors and the GABA_A receptors produced the same number of BrdU⁺ cells as in control (saline; P < 0.0001). d Application of ACh induced currents of the same frequency and amplitude in NSPCs before and after training, suggesting no changes in the cholinergic receptors following training. e Bath application of GABA before and after training revealed a significant reduction of the tonic activation amplitude without affecting its duration in the NSPCs. ACh, acetylcholine; BrdU, 5-bromo-2ʹ-deoxyuridine; GABA, γ-aminobutyric acid; GFP, green fluorescent protein; her4.1, hairy-related 4, tandem duplicate 1; The dashed gray line represents the baseline. Data are presented as mean ± s.e.m., as violin plots and as box plots showing the median with 25/75 percentile (box and line) and minimum–maximum (whiskers). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; ns, not significant. For detailed statistics, see Supplementary Table 1.

a Pulse-chase experiment to assess proliferation, neurogenesis, and restoration of motor functions after pharmacological manipulation of the nicotinic-ACh and GABA_A receptors in the adult zebrafish spinal cord. b Representative whole-mount confocal microphotographs showing BrdU⁺ cells in the zebrafish spinal cord. c Representative whole-mount confocal images for immunodetection of BrdU⁺/mef-2⁺ cells. Arrowheads indicate double-labeled cells. d Quantification of BrdU-incorporation after injury in control (saline) and pharmacologically treated animals (nicotine, gabazine). The dashed gray line represents the baseline (BrdU⁺ cells in uninjured animals). e Quantification of the BrdU⁺ cells express the neuronal marker mef-2⁺. f Nicotine- and gabazine-treated animals swim faster than the control (saline) fish during the critical speed test. The dashed gray line represents the baseline (critical speed of the uninjured animals). Speed is normalized (BL/s). BL, body length; BrdU, 5-bromo-2ʹ-deoxyuridine; mef-2, myocyte enhancer factor-2; SCI, spinal cord injury. Data are presented as box plots showing the median with 25/75 percentile (box and line) and minimum–maximum (whiskers). ^*P < 0.05; ^**P < 0.01. For detailed statistics, see Supplementary Table 1.

The findings link the locomotor CPG network to adult neurogenesis. Spinal cholinergic interneurons, including the premotor V2a-IN population, increase their cholinergic release to NSPCs during training. ACh acts directly on the NSPCs via nicotinic and muscarinic cholinergic receptors to activate them. Activation of NSPCs leads to the downregulation of GABA_A receptors.