Multimodal mechano-microscopy reveals mechanical phenotypes of breast cancer spheroids in three dimensions

Abstract

Cancer cell invasion relies on an equilibrium between cell deformability and the biophysical constraints imposed by the extracellular matrix (ECM). However, there is little consensus on the nature of the local biomechanical alterations in cancer cell dissemination in the context of three-dimensional (3D) tumor microenvironments (TME). While the shortcomings of two-dimensional (2D) models in replicating in situ cell behavior are well known, 3D TME models remain underutilized because contemporary mechanical quantification tools are limited to surface measurements. Here, we overcome this major challenge by quantifying local mechanics of cancer cell spheroids in 3D TMEs. We achieve this using multimodal mechano microscopy, integrating optical coherence microscopy-based elasticity imaging with confocal fluorescence microscopy. We observe that non-metastatic cancer spheroids show no invasion while showing increased peripheral cell elasticity in both stiff and soft environments. Metastatic cancer spheroids, however, show ECM-mediated softening in a stiff microenvironment and, in a soft environment, initiate cell invasion with peripheral softening associated with early metastatic dissemination. This exemplar of live-cell 3D mechanotyping supports that invasion increases cell deformability in a 3D context, illustrating the power of multimodal mechano-microscopy for quantitative mechanobiology in situ.

Tumor spheroid elasticity estimation using mechano-microscopy combined with a conditional generative adversarial network

“…This method presents several advantages compared to the algebraic method [29,32,33] and iterative methods [20,40], as outlined below.…”

“…Here, we provide a brief overview of the algebraic method [29,32,33], followed by a description of the methods and potential benefits associated with using cGANs.…”

“…As an alternative to the algebraic method, we propose to train a cGAN to learn the correspondence between OCM phase difference images and elasticity images; that is, given the input of a phase difference image created using mechano-microscopy and a latent vector, the cGAN should output an elasticity image instance from the learned distribution conditioned upon the phase difference. This method presents several advantages compared to the algebraic method [29,32,33] and iterative methods [20,40], as outlined below.…”

“…To validate the outputs from the cGAN, samples were imaged using a mechano-microscopy system described previously [33] and summarized briefly here. The system is based on OCM, which is a high-resolution variant of OCT that uses high numerical aperture objectives to achieve high lateral resolution [32].…”

“…al. developed a variant, mechano-microscopy, that uses optical coherence microscopy (OCM) to obtain sub-cellular elasticity resolution (5 μm) [11,32,33]. However, this technique assumes a frictionless mechanical model where the stress is uniaxial (i.e., acting only along the depth axis) and axially uniform (i.e., the stress at some location within the sample equals the stress at the layer-sample interface directly above or below that location).…”