Dynamical systems theory analyses reveal cell cycle-coupled paths of epithelial-to-mesenchymal transition.

Abstract

Epithelial to mesenchymal transition, EMT, is involved in numerous biological processes such as wound healing, tissue fibrosis, and cancer metastasis. Existing literatures have been debated on whether the transition proceeds through a single or multiple paths, and how cell cycle couples to EMT. To address the above questions, we first generated scRNA-seq dataset, where mammary epithelial MCF10A cells were treated with different doses of TGFβ, an EMT inducer. Then we analyzed the data with dynamo, a machine-learning based analytical framework we developed to reconstruct single cell dynamical equations (Qiu et al. Cell, 2022, 185: 690-711). From the obtained vector fields, we applied the transition path analyses, which are originally developed in studying chemical reactions, on simulated single cell trajectories. The analyses reveal two unique types of transition paths, corresponding to either an arrest in the G1/S or G2/M phase, when cells undergo EMT. The existence of two paths agrees with our previous live cell imaging studies (Wang et al., Sci. Adv. 2020, 6:eaba9309; eLife 2022, 11:e74866), but not pseudotime analyses reported in the literature. Our analyses also reveal a surprising backward cell cycle propagation of cells arrested in G2/M to a G1/S attractor through mitotic skipping. We obtained similar results with a number of other EMT scRNA-seq data sets, then confirmed with live cell imaging using a A549-Vim/RFP-PCNA-EGFP cell line. Our results provide mechanistic understanding of how a cell makes the decision between EMT and cell cycle progression, and demonstrate the importance of analyzing single cell data in the formalism of dynamical systems (Xing, Phys. Biol. 2022, 19: 061001).