Parallel locomotor control strategies in mice and flies

Abstract

Our understanding of the neural basis of locomotor behavior can be informed by careful quantification of animal movement. Classical descriptions of legged locomotion have defined discrete locomotor gaits, characterized by distinct patterns of limb movement. Recent technical advances have enabled increasingly detailed characterization of limb kinematics across many species, imposing tighter constraints on neural control. Here, we highlight striking similarities between coordination patterns observed in two genetic model organisms: the laboratory mouse and Drosophila. Both species exhibit continuously-variable coordination patterns with similar low-dimensional structure, suggesting shared principles for limb coordination and descending neural control.

Figure 1

Discrete gaits can be represented by distinct support patterns that depend on the relationship between stance duration and interlimb phasing. In some cases, animals display distinct, preferred patterns of interlimb coordination that can vary depending on size, speed, or species. For example, horses famously alter their gaits at different speeds, with a characteristic gallop at higher speeds. Flies display a wave gait where, at slow speeds, they lift one limb from the ground at a time. Giraffes also have a characteristic slow walk, lifting each limb sequentially. Stick insects display a tetrapod gait, where four limbs touch the ground at each time point in a diagonal arrangement. Cockroaches can show an alternating tripod gait of the six limbs, where diagonal limbs on the ground at the same time. Mice move most of the time in a diagonal trot where one pair of diagonal limbs is in contact with the ground at a time. This figure is modeled after Figure 5 of [17], and uses swing-stance patterns from the studies by Machado et al., DeAngelis et al., and Collins et al. [5,6,10] to estimate the range of relative homolateral limb phases across walking speeds. Canonical stance (solid) and swing (shaded) phases of front-, mid-. and hindlimbs of the left (FL, ML, HL) and right (FR, MR, HR) sides of the body are illustrated for each species.

Figure 2

Measurements of fly (left column) and mouse (right column) limb kinematics reveal parallel continua of coordination patterns. (a, b) Continuous forward trajectories over time for a fly’s six limb-tips (a) and a mouse’s four limb-tips (aka “paws,” b). (c, d) Stance duration decreases steeply with forward walking speed. (e, f) Average relative frequencies of limb support patterns within a stride cycle change gradually across forward walking speed for both flies (e) and mice (f; note the expanded speed range). (g, h) Speed-conditioned probability distributions of relative homolateral limb phasing vary smoothly and monotonically with forward walking speed for both flies (g, fore-mid claws) and mice (h, front-hind paws). Fly walking data is adapted from the study by DeAngelis et al. [6]; mouse data from the study by Machado et al. [5].

Figure 3

Dimensionality reduction illustrates common low-dimensional structure in fly and mouse interlimb coordination patterns. (a, b) Each point in the UMAP embedding represents a random 200 ms segment of limb positions over time, colored by the mean frequency of forward walking for the fly (a) and the mouse (b). (c–f) UMAP embedding of limb kinematic data colored by instantaneous left mid limb phase for the fly (c). For the mouse (d), colors represent the time of the first stance of the front right paw within each segment as a proxy for limb phase to avoid errors in instantaneous phase estimation due to incomplete information about the stride cycle within individual segments. (e–f) same as (c–d) but illustrating the end-on view of the manifold space. Fly limb coordinate time series embeddings were adapted from the study by DeAngelis et al. [6]. Mouse limb positions over time were collected during tied-belt locomotion on a transparent treadmill, as in the study by Darmohray et al. [40]. As in the study by DeAngelis et al. [6], randomly-sampled segments of limb position timeseries were embedded into three dimensions using a Euclidean distance metric and default UMAP hyperparameters.

Figure 4

Parallel locomotor control strategies in flies (left) and mice (right). Descending information from the central nervous system, driven by external stimuli, internal state, and/or sensory feedback, modulates locomotor speed and/or direction by modulating CPG modules in the ventral nerve cord of the fly or spinal cord of the mouse, either directly or by modulating internal coupling between CPGs (represented by arrows).