Background

Neuromuscular diseases are a diverse group of disorders with considerable unmet clinical need. For instance, many muscular dystrophies, such as Duchenne muscular dystrophy (DMD) that lead to progressive muscle wasting, currently lack effective treatments (Duan et al., 2021). Similarly, neurodegenerative disorders such as amyotrophic lateral sclerosis (ALS), caused by degeneration of motor neurons (MNs) that innervate and control movement of skeletal muscle, also lack effective treatments (Hardiman et al., 2017). Early peripheral phenotypes like axonal and neuromuscular junction (NMJ) dysfunction have been implicated in a wide range of divergent subsets of neuromuscular disorders, including both ALS and DMD (Fischer et al., 2004; van der Pijl et al., 2016). Emerging evidence suggests that therapeutically targeting neuromuscular synapse dysfunction is a viable option for extending lifespan and reducing disease severity in these disorders (Bruneteau et al., 2013; Cantor et al., 2018; Paredes-Redondo et al., 2021). However, a major challenge has been the generation of human-relevant disease models that accurately recapitulate in vivo human neuromuscular physiology and distinct neuromuscular disease phenotypes.

While mouse models have been widely used for studying neuromuscular physiology, growing evidence shows that there are substantial differences between human and mouse neuromuscular synapses, including: contrasting synaptic morphologies, divergent transcriptomes and proteomes, and altered homeostasis and maintenance (Jones et al., 2017).

With the development of differentiation protocols to derive enriched populations of motor neurons and myofibers (MFs) from human pluripotent stem cells (PSCs) (Du et al., 2015; Chal et al., 2016; Fernandopulle et al., 2018; Rao et al., 2018; Cheesbrough et al., 2022; Harley et al., 2022), a number of neuromuscular co-culture platforms have been established. A major challenge for generating human PSC-derived neuromuscular co-cultures has been stabilising contractile myofibers in order to prevent detachment upon contraction and allow long-term culture and maturation (Abd Al Samid et al., 2018). A notable example of an approach that overcame this issue was the development of a microphysiological neuromuscular disease model of ALS by Kamm and colleagues, in which bundles of myofibers were supported between micropillar cantilevers capable of deflecting upon contraction (Uzel et al., 2016; Osaki et al., 2018). Building on this work, we have developed a model of human neuromuscular physiology with a number of innovative features. These include a compartmentalised design comprising three chambers connected by microchannels, which mimics the spatial organization of tissues in vivo. This design enabled us to compare two genetically different motor neuron populations—one sensitive to optogenetic stimulation and the other insensitive—in their ability to innervate the same myofiber targets in response to optogenetic entrainment (Machado et al., 2019). Likewise, it allowed us to study phenotypes of the same motor neurons innervating an isogenic pair of DMD and CRISPR-corrected myofibers (Paredes-Redondo et al., 2021). Furthermore, incorporation of ALS-related TDP-43^G298S motor neurons and CRISPR-corrected controls enabled us to recapitulate key ALS-related neuromuscular phenotypes (Harley et al., 2022). In all these experiments, motor neurons were embedded in a scaffold of mouse embryonic stem cell (ESC)–derived astrocytes to promote maturation and provide neurotrophic support. In future studies, it will allow different reconfigurations of the device to support different co-cultures. Such combinations of tissues may include a full corticomotor tract (cortical neurons, motor neurons, muscle) or the incorporation of sensory neurons into a sensory feedback compartment. Secondly, the proximity of the individual compartments to the optical polymer allows easy and rapid high-power imaging without the need for additional tissue processing and microdissection. Combined with the microchannels for axonal outgrowth, this would easily allow for high-resolution live axonal transport imaging and live synaptic imaging assays to be developed in future studies. Thirdly, myofibers are stabilized by a thin layer of UV-cured resin applied to the surface of the optical polymer, similar to the anchor points in the original microdevices (Machado et al., 2019). Finally, we provide evidence that formation of functional in vitro human neuromuscular circuits is activity dependent. We found that optogenetic entrainment of the motor neurons over the course of the culture dramatically enhanced synapse formation and myofiber contractility. In future studies, the microdevices described here may be used to understand how different stimulation paradigms and competitive innervation of common synaptic targets facilitates the wiring of human neuromuscular circuits.