Objective: The impact of mutations on human development are rarely investigated due to the ethical rules linked to the use of embryos. Therefore, early disease manifestations often remain unexplored, which delays diagnosis and treatment initiation. In this context, induced pluripotent stem cells (iPSCs) derived from patients offer an attractive solution, as they can be used to recapitulate human development in a dish. Using single-cell transcriptomics, we showed that iPSCs with mutations in the DMD gene causing Duchenne muscular dystrophy (DMD) adopt a different gene expression profile at the somite stage. As a result, the single-cell trajectory splits into two branches that give rise to different cell fates. In this study, we aim to infer the gene regulatory network (GRN) explaining this profile and to validate it experimentally using the iPSC model.

Methods: Downstream of the bifurcation, 10% of the DMD cells are still located on the Healthy branch, and conversely. Thus, the two final cell states exist independently of the mutation, suggesting that cell fate choice is partially stochastic and that the DMD mutation modifies the probability to converge into a given state. We used an updated version of the WASABI framework as a causal, quantitative and probabilistic method of GRN inference. We generated resolutive single-cell RNA-Seq data spanning the bifurcation window using CITE-Seq. In addition, we measured RNA half-lives and performed proteomic experiments in order to model promoter and protein wave times.

Results: WASABI inferred a GRN composed of six core genes in a toggle-switch topology. In Healthy cells reaching the branching point, the expression of ENO1, MIF and PRTG is strongly down-regulated while the expression of ID1, PDGFRA and PTN is upregulated. In DMD cells, this gene expression switch is OFF with a higher probability. We next transfected siRNA against ENO1, MIF or PRTG in DMD cells in an attempt to force the activation of the switch. Interestingly, we observed that DMD cells transfected with ENO1 or PRTG siRNA cluster together with Healthy controls after the bifurcation.

Conclusion: Our results strengthen previous observations that DMD initiation happens during muscle development in the embryo. ENO1 and MIF are respectively involved in glycolysis and in the activation of the AMPK and PI3K/Akt signalling pathways. Thus, the GRN inferred by WASABI and unbalanced in DMD cells may lead to an impaired metabolic shift during myogenic development.