Tissue-engineered skeletal muscle presents promising opportunities for developing high-fidelity in vitro models for investigating human muscle biology in the areas of regeneration and disease. In muscle regeneration, satellite cells (SCs) are essential for new muscle fiber formation;​ however, they lose their native quiescent state upon isolation, making in vitro studies of human SC function challenging. To optimally promote SC quiescence and enable exploration of SC dynamics in vitro, engineered muscle needs to recapitulate the native muscle microenvironment, which is comprised of muscle fibers, extracellular matrix, and other biochemical and mechanical cues. In disease modeling, mechanistic studies and therapeutic development are still extensively evaluated in animal models, which have limited translational relevance to patients. Specifically, Pompe disease is caused by a variety of mutations in the lysosomal enzyme acid alphaglucosidase (GAA) that varyingly affect residual GAA activity and cannot be captured in the current GAA^-/- mouse model. Therefore, human in vitro models are needed to enhance our mechanistic understanding of diseases and stimulate the development of effective therapies. To overcome these limitations, we set the dissertation goals to:​ 1) generate a pool of quiescent SCs and explore the mechanisms governing their formation and activation using an engineered skeletal muscle microenvironment and 2) develop a highfidelity tissue-engineered skeletal muscle model for Pompe disease to investigate pathological mechanisms and test candidate therapies.

To achieve these goals, we first compared methods for primary human myoblast expansion and found that p38 inhibition significantly increases the formation of Pax7+ cells in engineered 3D skeletal muscle tissues (“myobundles”). Gene expression analysis suggested that within the myobundle environment the Pax7+ cells adopt a quiescent phenotype (3D SCs), characterized by increased Pax7 expression, cell cycle exit, and Notch signaling activation relative to the original 2D expanded myoblasts. We then compared 3D SCs to previously described satellite-like cells that form alongside myotubes in 2D culture, termed reserve cells (RCs). Compared to RCs, 3D SCs showed an advanced quiescent phenotype characterized by a higher Pax7, Spry1, and Notch3 expression, as well as increased functional myogenesis demonstrated by formation of myobundles with higher contractile strength. To examine 3D SC activation, we tested several myobundle injury methods and identified treatment with a bee toxin, melittin, to robustly induce myofiber fragmentation, functional decline, and 3D SC proliferation.

To further investigate the transcriptional processes describing how 2D myoblasts acquire 3D SC phenotype (i.e. deactivate) and how 3D SCs respond to injury (i.e. reactivate), we applied single cell RNA-sequencing (scRNA-seq) from which we discovered the existence of two subpools of 3D SCs—“quiescent” (qSC) and “activated” (aSC). The qSC subpool possessed greater expression of quiescence genes Pax7, Spry1, and Hey1, whereas the aSC subpool exhibited increased expression of inflammatory and differentiation markers. Furthermore, we performed trajectory inference along the deactivation process from 2D myoblasts to qSCs and identified deactivation-associated genes, included downregulated genes for proliferation, cytoskeletal reorganization, and myogenic differentiation. In response to tissue injury, we observed a decrease in the proportion of qSCs and an increase in the proportion of aSCs and committed myogenic progenitor cells suggestive of myogenic differentiation. In addition, we observed transcriptional changes within the aSC population reflective of SC activation in vivo, namely increased TNF-a signaling, proliferation, and glycolytic and oxidative metabolism. These results strongly suggested that 3D SC heterogeneity and function recapitulate several aspects of native human SCs and could be applied to study human muscle regeneration and disease-associated SC dysfunction.

To evaluate the myobundle system in the context of disease modeling, we developed the first 3D tissue-engineered skeletal muscle model of infantile onset Pompe disease (IOPD), the most severe form of Pompe disease. Diseased myobundles demonstrated characteristic GAA enzyme deficiency, accumulation of the GAA target glycogen, and lysosome enlargement. Despite exhibiting these key biochemical and structural hallmarks of disease, IOPD myobundles did not show deficits in contractile force generation or autophagic buildup. We therefore identified metabolic stress conditions that acutely targeted disease-associated abnormalities in the lysosomes and glycogen metabolism, which revealed impairments in contractile function and glycogen mobilization. To further elucidate the biological mechanisms underlying the phenotype of IOPD myobundles, we applied RNA sequencing (RNA-seq) and observed enrichment for terms consistent with Pompe disease phenotype including downregulation of gene sets involved in muscle contraction, increased endoplasmic reticulum stress, and reduced utilization of specific metabolic pathways. We then compared the transcriptomic profiles of GAA-/- and wild-type mice to identify a Pompe disease signature and confirmed the presence of this signature in IOPD myobundles. Finally, treating IOPD myobundles with clinically used recombinant protein (rhGAA) therapy resulted in increased GAA activity, glycogen clearance, and a partial reversal of the disease signature, further confirming the utility of the myobundle system for studies of Pompe disease and therapy.

In summary, this dissertation describes novel strategies for the formation and characterization of quiescent human SCs using the myobundle system. We present firsttime application of scRNA-seq to engineered skeletal muscle, and uncover transcriptional descriptors of human myoblast deactivation and SC heterogeneity and activation. When utilizing human myobundles as a novel model of Pompe disease, we identified disease hallmarks and responses to therapy consistent with observations in Pompe patients. We anticipate that the findings and methods developed in this work will serve as a useful framework for the future engineering of regenerative human muscle for therapeutic and disease modeling applications.