Optical microscopy-based mechanical phenotyping of heterogeneous cellular migration and contractility.

Abstract

All forms of matter, including biological entities, are bound to the law of physics. While mechanical forces have been implicated in the development of organisms almost a century ago, the recent ability to quantitatively measure forces has revealed that physical principles drive fundamental cellular processes. For specialized cells such as leukocytes or cardiomyocytes, the biological function is tightly linked to the cell's mechanical state, misregulation of which is implicated in numerous diseases. Optical microscopy empowered with fluorescence tagging of biomolecules enables kinetic visualization of the cytoplasm and the surrounding substrate. By combining physical modeling, statistical methods and experimental microscopy data at the cellular level, the heterogeneous mechanical behavior of single cells and their collectives can be quantified. This thesis describes methodological advancements in the field of cell migration profiling, cell-exerted traction force measurements and the estimation of cardiomyocyte material properties. First, the stochastic motility of leukocytes in three-dimensional interstitial environment is studied by measuring the matrix deformation measurements, random walk models and single-cell trajectory analysis. Neutrophils migrating through extravascular spaces must negotiate narrow matrix pores without losing directional tendencies to reach their target. Neutrophils in low collagen concentration matrices were observed to exert large deformation and followed relatively straight trajectories. However, as the collagen concentration increased, matrix deformation decreased and neutrophils turned often to circumvent rather than remodel matrix pores. Inhibiting protrusive or contractile forces shifted this transition to lower collagen concentration, implying that mechanics play a crucial role in defining migratory strategies. Leukocytes exhibit considerable cell-cell variability in their migration response, and thus their effector function. Hence, there is a need to identify leukocyte population heterogeneity to properly characterize the migratory behavior. A pipeline to autonomously identify clusters of cells exhibiting different random walk characteristics identified two naturally occurring migratory phenotypes: a high motility cluster with faster, persistent cells undergoing fewer turns and a low motility cluster with lower speed and increased incidence of large turns. This cluster proportion was modified by the process of transendothelial migration with an enrichment of the low-motility population. As transendothelial migration is a necessary step in the inflammation response, these observations provide insights into the biophysical importance of differential priming of neutrophil sub-populations to conduct sentinel functions. Such migratory heterogeneity was also demonstrated in mouse primary T-cells showing that proteins of the same Formin family (Formin-like 1 and mammalian diaphanous-related formin 1), despite modulating linear actin polymerization, differentially regulate T-cell motility based on the collagen density. Secondly, many biophysical measurements (such as leukocyte matrix deformation quantification) rely on image-based motion estimation and further computational modeling to estimate forces. Traction Force Microscopy (TFM) is a widely adapted method to measure cell-exerted forces by tracking fiduciary markers embedded in the substrate. TFM calculations are numerically ill-conditioned and smoothing/regularization is routinely employed to avoid overfitting input noise. While it is known that such image-derived deformation measurements are sensitive to local image features and the choice of the regularization parameter, the existing force reconstruction algorithms do not account for variations in image quality. This hinders objective comparison between experiments, requiring ad-hoc manual smoothing parameter selection and masking the variability of force readouts. TFM-UQ, a TFM uncertainty quantification pipeline is developed to estimate the local, image-dependent deformation uncertainty and propagate it to obtain traction stress error bars using a Bayesian framework. This approach enables spatially adaptive, automated regularization and provides traction stress error bars which was previously not possible. Force readout uncertainty allows decoupling biological heterogeneity from measurement variability, as well as providing a roadmap to enable automating analysis of large datasets by parameter-free input data-based regularization. The Bayesian modeling in TFM-UQ was extended to sub-cellular platelet force resolution using a reference-free TFM procedure. Finally, the TFM biomechanical model is augmented to estimate the effective stiffness of in vitro multi-cellular colonies such as human induced Pluripotent Stem Cells derived cardiomyocytes (hiPSC-CM) monolayers. While physiological assays with hiPSC-CMs are powerful tools for the development of new drugs, high throughput methods to quantify cardiomyocyte contractile force are not prevalent. Furthermore, there are no optical high-throughput methods to quantify cardiomyocyte stiffness in a multiplexed experimental format. To this end, a computational pipeline that can measure kinematic and dynamic contraction metrics including the stiffness of the monolayer in a non-invasive manner was developed. This pipeline is applied to spontaneously beating monolayers of hiPSC-CMs treated with different doses of benchmark compounds affecting contractility, diastolic tension, and stiffness, demonstrating mechanical phenotyping of hiPSC-CMs in a comprehensive and computational efficiency manner. The methods developed in this thesis advance image-based mechanical characterization and biophysical measurements in 3D cell migration, cardiomyocyte mechanics and more generally enable precision in mechanobiology applications. The use of statistical modeling is essential to model inherent uncertainties due to biological activity and allows quantitative testing of new hypotheses. Improvements to the precision of biophysical measurements with microscopy images have significant implications in personalized medicine and enable more rigorous comparison between experiments.