A gradient in endogenous rhythmicity and oscillatory drive matches recruitment order in an axial motor pool

Abstract

The rhythmic firing behavior of spinal motoneurons is a function of their electrical properties and synaptic inputs. However, the relative contribution of endogenous versus network-based rhythmogenic mechanisms to locomotion is unclear. To address this issue, we have recorded from identified motoneurons and compared their current-evoked firing patterns to network-driven ones in the larval zebrafish (Danio rerio). Zebrafish axial motoneurons are recruited topographically from the bottom of the spinal cord up. Here, we have explored differences in the morphology of axial motoneurons, their electrical properties, and their synaptic drive, to reveal how they match the topographic pattern of recruitment. More ventrally located "secondary" motoneurons generate bursts of action potentials in response to constant current steps, demonstrating a strong inherent rhythmogenesis. The membrane potential oscillations underlying bursting behavior occur in the normal frequency range of swimming. In contrast, more dorsal secondaries chatter in response to current, while the most dorsally distributed "primary" motoneurons all fire tonically. We find that systematic variations in excitability and endogenous rhythmicity are inversely related to the level of oscillatory synaptic drive within the entire axial motor pool. Specifically, bursting cells exhibit the least amount of drive, while tonic cells exhibit the most. Our data suggest that increases in swimming frequency are accomplished by the recruitment of axial motoneurons that progressively rely on instructive synaptic drive to shape their oscillatory activity appropriately. Thus, within the zebrafish spinal cord, there are differences in the relative contribution of endogenous versus network-based rhythms to locomotion and these vary predictably according to order of recruitment.

Figure 1.

Systematic sampling of axial motoneurons throughout the spinal segment. A, Schematic of a larval zebrafish at the developmental stage examined here (4–5 d of age). Larvae have just over 30 chevron-shaped muscle segments that begin just behind the ear (labeled “e”). The brain and spinal cord are shaded in gray. Motoneurons were targeted within the same region at midbody (segments 11–13). Here and elsewhere, rostral is to the left and dorsal is up. B, A contrast inverted confocal image of a zebrafish larva from the parg^mn2Et enhancer trap line in which GFP is expressed in motoneurons. The image on the left illustrates the anatomical reference points used to calculate somata location, namely the top and bottom edges of spinal cord (dashed lines) and the ventral root exit point (arrows). The asterisks indicate likely caudal primary motoneurons (CaPs). The caudal edge of the muscle segments is numbered. For clarity, only the left side of the larva is illustrated (confocal z-stack depth, 43 μm). On the right is an optical cross section from the same larva. Both the left and right pools of motoneurons are illustrated from the 12th segment. The dashed line indicates the outline of the spinal cord. C, Differential interference contrast image of the dissection for targeted whole-cell recordings. The inset image is at a different focal plane to illustrate access to the spinal cord for patch electrodes. The dorsal and ventral edges of spinal cord are normalized from 1 to 0, respectively, for anatomical measurements. Muscle segments are numbered. In the inset image, the Mauthner cell axon is indicated by the black arrow. D, The plot illustrates every motoneuron analyzed in this study (n = 106; marked with X's) with respect to rostrocaudal (RC) position, measured from the ventral root exit point (vertical gray line, V), and dorsoventral (DV) position, normalized to the top and bottom edges of spinal cord. Also illustrated here are cells whose identity could not be determined because their axons were severed during the dissection (n = 3; marked with O's). These cells were not included in the analysis. The dorsal edge of the Mauthner cell axon (horizontal gray line, M) is provided as an additional anatomical reference.

Figure 2.

Four primary motoneurons tile the dorsoventral extent of the muscle segment. A, Morphological reconstructions of primary motoneurons illustrate the somata, the main axon, and axon collaterals within the muscle. Shown here are representative reconstructions of a caudal primary (CaP), a middle primary (MiP), a rostral primary that innervates ventral musculature (vRoP), and a rostral primary that innervates dorsal musculature (dRoP). Collectively, the four classes cover the full dorsoventral extent of the muscle segment. The dashed gray line is the bottom edge of the spinal cord. The horizontal myoseptum is also indicated as a solid gray line in each reconstruction (labeled “HM”). Muscle segments are numbered in gray. Dorsal (D) is up and rostral (R) is to the left. B, Somata locations of all primary motoneurons (n = 31), separated by anatomical class and overlaid on the total population (gray X's). Consistent with their nomenclature, CaPs are the most caudal, with the MiPs falling between the CaPs and the RoP classes. V, Ventral root exit point; M, dorsal edge of the Mauthner axon. C, Quantification of the dorsal-most and ventral-most extent of primary motoneuron axon collaterals, normalized to the top (1) and bottom (0) edges of the innervated muscle segment, confirms the stereotyped innervation patterns of each primary class. The horizontal myoseptum (HM) is included as a dashed gray line for reference. D, Soma size distribution for each primary class illustrates a progressive decrease in cross-sectional area from the CaPs to the dRoPs. **p < 0.001 and *p < 0.01 using a post hoc Tukey–Kramer test. E, Histogram of the distribution of soma size for all 106 motoneurons, categorized anatomically as primary (n = 31) and secondary (n = 75). At midbody, primary motoneurons tend to be larger than secondary motoneurons, although there is a region of overlap.

Figure 3.

Primary and secondary motoneurons with intraspinal axon collaterals. A, B, On the left are morphological reconstructions of a ventrally projecting RoP (A) and dorsally projecting secondary motoneuron (B) with intraspinal axon collaterals (at arrows). The soma and dendrites are in gray, while the axon and intraspinal collaterals are in black. The dashed gray lines indicate the top and bottom edges of the spinal cord. The asterisks mark the continuation of the axons outside the plane of focus. In the middle are inverted epifluorescent images of the axon collaterals marked with arrows on the left. On the right, simultaneous differential interference contrast and epifluorescent images illustrate the location of the dorsal-most point of the axon collaterals, which can terminate in the spinal neuropil or soma layer. C, Somata locations of all motoneurons with intraspinal collaterals (n = 29), which occupy more rostral regions of the spinal segment. The open circles are overlaid on the total population for reference (gray X's). V, Ventral root exit point; M, dorsal edge of the Mauthner axon. D, Measurements of the dorsoventral (DV) position of motoneuron somata versus the dorsal-most point of their respective intraspinal axon collaterals indicate a clear correlation between the two (R₍₂₇₎ = 0.56; p < 0.01; n = 29). This suggests motoneurons can make contact with spinal cells at levels up to their own dorsoventral position. The solid gray line indicates unity.

Figure 4.

Arborization patterns can be used to categorize secondary motoneurons. A, Morphological reconstructions of secondary motoneurons illustrate the somata, the main axon, and axon collaterals within the muscle. Shown here are representative reconstructions of a dorsoventrally projecting secondary (dvS), a ventrally projecting secondary (vS), and a dorsally projecting secondary (dS). Collectively, these three classes cover the full dorsoventral extent of the muscle segment. The dashed gray line is the bottom edge of the spinal cord. The horizontal myoseptum (HM) is also indicated as a solid gray line. Muscle segments are numbered in gray. Dorsal (D) is up and rostral (R) is to the left. B, Morphological reconstructions of secondary motoneurons with axons that run along the intermyotomal boundaries. Shown here are three intermyotomal secondaries with collaterals (iS-c) and two intermyotomal secondaries with no collaterals (iS-nc). C, Somata locations of all secondary motoneurons (n = 75). The dvS class is the most dorsally distributed one in the spinal segment, followed by vS and the dS classes. For the iS class, motoneurons with axon collaterals are distributed more dorsally than ones without them. For reference, data points are overlaid on the total population (gray X's). V, Ventral root exit point; M, dorsal edge of the Mauthner axon. D, Quantification of the dorsal-most and ventral-most extent of secondary motoneuron axon branches, normalized to the top (1) and bottom (0) edges of the innervated muscle segment, confirm the overlapping muscle coverage provided by each class. For intermyotomal secondaries, the dorsal- and ventral-most points at which the axon reached the intermyotomal boundary were used for measurements. The HM is included as a dashed gray line for reference.

Figure 5.

Three common responses of motoneurons to current injection. A–D, Examples of a tonic response in a CaP motoneuron (A), a chattering response in a dvS motoneuron (B), and burst responses in a vS motoneuron (C) and dS motoneuron (D). Each response is shown at 1, 1.1, and 1.5× rheobase (R) from left to right. There is a progressive increase in the complexity of the firing response to current steps from tonic to burst cells. Rheobase for the CaP was 300 pA, for the dvS was 105 pA, for the vS was 65 pA, and for the dS was 35 pA. Voltage (V) and current (I) calibration bars are to the right, while time is presented below in D. The inset in gray on the left is the response of each class at rheobase on a faster timescale to more clearly illustrate the slower depolarization that characterizes the endogenous membrane potential oscillations of burst responses. Traces are all at the same calibration as indicated in D. Note that all voltage records include small artifacts aligned with the onset and offset of the current step, due to capacitance compensation and bridge balance correction. E–G, The firing frequency over the entire 500 ms current step for the tonic (E), chattering (F), and bursting (G) cells in A, B, and D is plotted, with responses labeled as a function of rheobase (e.g., 2R = 2× rheobase). Tonic motoneurons can fire at stable levels throughout suprathreshold current steps, while chattering and burst cells tend to rapidly accommodate at higher levels of current injection. For clarity, we have only presented the burst response in G at 1.5 and 2× rheobase. H–J, The instantaneous firing frequency of the first two spikes in response to increasing levels of current is plotted for tonic (H; n = 31), chattering (I; n = 53), and bursting motoneurons (J; n = 22). All firing classes showed an increase in firing frequency in response to increased current steps. Each response is normalized to their respective rheobases. The gray lines indicate the individual motoneuron responses, while the black lines with SDs are an average response of all cells binned at 0.5 rheobase intervals. The double gray asterisks indicate significant differences using a pairwise comparison of firing frequency at the lowest and highest value of current tested on a cell-by-cell basis (see Results for details). K, For burst neurons, the instantaneous frequency of the first two bursts is plotted at the lowest and highest level of current where bursting was observed (n = 22). Each response is normalized to rheobase. The double gray asterisks indicate a significant increase in burst frequency in response to increased current using a pairwise comparison (see Results for details). L, The instantaneous frequency of the every cellular burst over the entire 500 ms current step (1065 bursts from 22 cells) compared with the frequency of motor bursts recorded from peripheral motor nerves during fictive swimming (n = 2230 cycles from 7 different larvae). The cellular bursting data represent multiple current steps over the narrow range where bursting was most easily detected (from 1 to 1.5× rheobase). Data are normalized to the total number of observations in 10 Hz bins (% total). The intrinsic bursting behavior occurs at frequencies that overlap with motor bursts generated during fictive swimming.

Figure 6.

Current-evoked firing responses exhibit graded, overlapping measures of intrinsic excitability related to their morphologies. A, The first spike at rheobase for the tonic, chattering, and burst cells presented in Figure 5A–D. Note the progressive differences in spike amplitude, half-width, and AHP observed between the classes. A gray dashed line is provided for reference. B, Spike AHP versus spike amplitude for each firing response. There is a clear correlation between the two properties when you examine the whole population (R₍₁₀₄₎ = 0.91; p < 0.001; n = 106). There is also significant overlap between the grouped firing responses, which argues more for a continuum and less for discrete classes. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = 0.91; p < 0.001; n = 75). C, Spike half-width versus rheobase for each firing response. There is a clear correlation between the two properties when you examine the entire population (R₍₁₀₄₎ = −0.90; p < 0.001; n = 106). The inset is the same plot on a logarithmic scale to make this point clearer (both X- and Y-axes). The distribution of values between firing responses is consistent with a continuum. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = −0.90; p < 0.001; n = 75). D, Input resistance versus soma size for each firing response. There is a clear correlation between the two properties when you examine the entire population (R₍₁₀₄₎ = −0.88; p < 0.001; n = 106). The inset is the same plot on a logarithmic scale to make this point clearer (both X- and Y-axes). Again, a continuous gradient between the firing responses is evident. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = −0.73; p < 0.001; n = 75). E, On the left, quantification of the dorsoventral (DV) and rostrocaudal (RC) distribution of the somata of different firing responses. Tonic cells are more dorsally located, burst cells are more ventrally located, and chattering cells overlap the two classes. On the right, this relationship is more obvious when you plot firing class as a proportion of the total population at different dorsoventral locations (normalized from 1 to 0). V, Ventral root exit point; M, dorsal edge of the Mauthner axon. F, Dorsoventral distribution of different anatomical classes of motoneurons related to their firing responses. CaP (n = 6), MiP (n = 10), vRoP (n = 7), dRoP (n = 8), dvS (n = 20), vS (n = 18), dS (n = 11), iS-c (n = 15), and iS-nc (n = 11). Tonic cells are exclusively primary, while the chattering and burst cells are found among all anatomical classes of secondary motoneurons.

Figure 7.

Differences in synaptic drive and the reliability of firing during fictive swimming related to anatomical class and firing type. A–D, Examples rhythmic oscillatory drive during fictive swimming in a dRoP tonic motoneuron (A), a dS chattering motoneuron (B), an iS-c bursting motoneuron (C), and an iS-nc bursting motoneuron (D). On the left is a whole episode at a slower time base. In the middle and right are examples of lower versus higher frequency (in hertz) swimming, respectively, at a faster time base. In D, an asterisk marks a subthreshold oscillation, which is evident against a barrage of synaptic activity. The segments from which simultaneous motoneuron (top) and peripheral motor nerve (bottom) recordings were performed are labeled below the respective traces. The dashed gray lines on the left indicate resting membrane potential. Calibration for all records is presented in A. Note that the action potentials are truncated to more easily observe the oscillatory drive. E–G, Plots of firing frequency as a function of fictive swimming frequency from tonic (E), chattering (F), and burst (G) motoneurons. Data points represent means ± SDs plotted in 10 Hz bins. Tonic (n = 24), chattering (n = 45), burst (n = 20). Note that motoneurons were only included in this analysis if they fired two or more spikes over at least two swim frequency bin widths, so these represent a subset of the total population (89 of 106 cells). There is an increase in firing frequency as a function of swimming frequency for the different firing classes, consistent with their respective responses to increased current injection. The double black asterisks indicate significant differences using a pairwise comparison of firing frequency at the lowest and highest frequency of swimming tested on a cell-by-cell basis. Tonic: 421.3 ± 19.5 to 493 ± 22.3 Hz (t₍₂₃₎ = 2.87; p < 0.01; n = 24). Chattering: 294.7 ± 14.7 to 362.2 ± 17.5 Hz (t₍₄₄₎ = 4.93; p < 0.001; n = 45). Burst: 345.9 ± 19.8 to 398.0 ± 29.2 Hz (t₍₁₉₎ = 2.69; p < 0.05; n = 20). H, Percentage of cycles in which a motoneuron fired an action potential (firing reliability) at a particular swim frequency for iS-c and iS-nc motoneurons. For iS-c motoneurons, this represents 2627 total cycles from 15 larvae between 20 and 70 Hz, while for iS-nc motoneurons it is 1427 total cycles from 11 larvae over the same frequency range. The iS-c class is unique in that it is de-recruited during increases in swimming frequency. I, The same reliability analysis as in K, but organized according to firing type. Despite the iS-nc de-recruitment, there is an increase in firing reliability when considering the populations as defined by current-evoked firing pattern. Tonic cells, 4793 cycles from 31 larvae; chattering cells, 8714 cycles from 53 larvae; burst cells, 3005 cycles from 22 larvae. J–L, Plots of suprathreshold (Supra-T) (black) and subthreshold (Sub-T) (gray) oscillations for tonic (J), chattering (K), and burst (L) motoneurons. While tonic and chattering cells show a clear increase in subthreshold drive that eventually reaches suprathreshold levels at higher swim frequencies, burst neurons have subthreshold oscillations that do not increase significantly and are consistently closer to suprathreshold oscillations over the full swim frequency range. Data points represent the means ± SDs plotted in 10 Hz bins. To the right of the data are indications of the significance of differences using a pairwise comparison of suprathreshold and subthreshold oscillations at the lowest and highest frequency of swimming tested on a cell-by-cell basis. **Significant; n/s, not significant. Tonic: subthreshold from 2.9 ± 0.2 to 14.2 ± 0.9 mV (t₍₃₀₎ = 11.85; p < 0.001; n = 31), suprathreshold from 21.4 ± 0.7 to 21.7 ± 0.8 mV (t₍₃₀₎ = 0.62; p = 0.538; n = 31). Chattering: subthreshold, 5.6 ± 0.3 to 11.3 ± 0.6 mV (t₍₅₂₎ = 7.78; p < 0.001; n = 53); suprathreshold, 17.7 ± 0.5 to 18.0 ± 0.6 mV (t₍₅₂₎ = 0.74; p = 0.462; n = 53). Burst: subthreshold, 6.6 ± 0.7 to 9.1 ± 1.0 mV (t₍₂₁₎ = 1.98; p = 0.063; n = 22); suprathreshold, 13.6 ± 1.1 to 14.5 ± 1.0 mV (t₍₂₁₎ = 1.67; p = 0.111; n = 22).

Figure 8.

Oscillatory and depolarizing drive vary according to input resistance and rheobase. A, Input resistance versus suprathreshold oscillation amplitude for each firing class. Input resistance is negatively correlated with the amplitude of suprathreshold oscillations for the entire population of motoneurons (R₍₁₀₄₎ = −0.78; p < 0.001; n = 106). Thus, higher input resistance motoneurons exhibit smaller amplitude membrane potential oscillations during swimming. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = −0.73; p < 0.001; n = 75). B, Rheobase versus suprathreshold oscillation amplitude for each firing class. Rheobase is positively correlated with the amplitude of suprathreshold oscillations for the entire population of motoneurons (R₍₁₀₄₎ = 0.78; p < 0.001; n = 106). Higher rheobase motoneurons exhibit larger-amplitude membrane potential oscillations during swimming. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = 0.74; p < 0.001; n = 75). C, Input resistance versus suprathreshold on-cycle amplitude for each firing class. Input resistance is negatively correlated with on-cycle amplitude for the entire population of motoneurons (R₍₁₀₄₎ = −0.73; p < 0.001; n = 106). Higher input resistance motoneurons require less depolarizing drive to get them to fire. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = −0.63; p < 0.001; n = 75). D, Rheobase versus suprathreshold on-cycle amplitude for each firing class. Rheobase is positively correlated with on-cycle amplitude for the entire population of motoneurons (R₍₁₀₄₎ = 0.78; p < 0.001; n = 106). Higher rheobase motoneurons require more depolarizing drive to get them to fire. The correlation is maintained when only the secondary motoneurons are examined (R₍₇₃₎ = 0.74; p < 0.001; n = 75).

Figure 9.

A model for the organization of the larval zebrafish axial locomotor network. Shown is a simple schematic that summarizes the implications of our findings. Intrinsic cellular properties like input resistance, rheobase, and current-evoked rhythmic bursting behavior are matched by gradients in excitatory and inhibitory synaptic drive. More excitable rhythmogenic neurons likely require less patterned drive to shape their rhythmic activity during swimming. During increases in swimming frequency, less rhythmogenic neurons, which rely more on patterned synaptic drive, are engaged. These systematic central differences in connectivity, intrinsic properties, and recruitment are translated peripherally into different intensities of local segmental muscle contractions, as observed during varying speeds of swimming.