Abstract
One of the hallmarks of cancer cells is their exceptional ability to migrate within the extracellular matrix (ECM) for gaining access to the circulatory system, a critical step of cancer metastasis. RhoA, a small GTPase, is known to be a key molecular switch that toggles between actomyosin contractility and lamellipodial protrusion during cell migration. Current understanding of RhoA activity in cell migration has been largely derived from studies of cells plated on a two-dimensional (2D) substrate using a FRET biosensor. There has been increasing evidence that cells behave differently in a more physiologically relevant three-dimensional (3D) environment. However, studies of RhoA activities in 3D have been hindered by low signal-to-noise ratio in fluorescence imaging. In this paper, we present a a machine learning-assisted FRET technique to follow the spatiotemporal dynamics of RhoA activities of single breast tumor cells (MDA-MB-231) migrating in a 3D as well as a 2D environment. We found that RhoA activity is more polarized along the long axis of the cell for single cells migrating on 2D fibronectin-coated glass versus those embedded in 3D collagen matrices. In particular, RhoA activities of cells in 2D exhibit a distinct front-to-back and back-to-front movement during migration in contrast to those in 3D. Finally, regardless of dimensionality, RhoA polarization is found to be moderately correlated with cell shape.