90% de la mortalité par cancer provient de tumeurs disséminées, ou métastases. Ces métastases se forment à partir de cellules tumorales qui s'échappent d'une tumeur primaire, circulent dans le sang, puis quittent les vaisseaux sanguins pour enfin aller nicher dans des organes distants et former des tumeurs secondaires. Les processus par lesquels ces cellules circulantes envahissent les organes distants, remodèlent leur environnement pour créer une «niche micrométastatique», prolifèrent pour produire des métastases macroscopiques, sont mal connus, principalement en raison d'un manque de modèles expérimentaux. En effet ces événements sont rares, se produisent à une échelle microscopique et à des localisations à priori inconnues. La perte d'adhérence cellulaire des cellules tumorales se détachant des tissus tumoraux primaires est associée à un phénomène de transformation connu sous le nom de transition épithéliale-mésenchymateuse (EMT) conduisant à la perte des caractéristiques épithéliales. Dans ce travail, nous avons souhaité aborder la question du processus métastatiques par l'étude de l'influence de l'étape de circulation dans le flux sanguin sur différentes caractéristiques de cellules tumorales. Pour cela, des modèles microfluidiques contenant des constrictions mécaniques afin d'imiter la microcirculation sanguine ont été conçus et fabriqués. Nous avons soumis des cellules provenant de tumeurs primaires du sein dans des situations de confinement périodiques à l'intérieur de ces canaux microfluidiques en utilisant un système de contrôle de flux. Nous avons étudiés l'impact des déformations induites par les constrictions des canaux microfluidiques sur l'expression génétique des marqueurs EMT, la morphologie ainsi que la dynamique des changements morphologiques. Nous montrons que ces paramètres cellulaires sont touchés par la déformation mécanique imposée sous flux, suggérant que l'étape de circulation des cellules tumorales dans le sang a un rôle important dans la capacité de celles-ci à produire des métastases. / 90% of cancer mortality arises from metastases, due to cells that escape from a primary tumor, circulate in the blood as circulating tumor cells (CTCs), leave blood vessels and nest in distant organs. The processes by which CTCs invade distant organs, remodel their environment to create a “micrometastatic niche”, the eventual triggering of a proliferation leading to a macroscopic metastases, are poorly known, mostly because of a lack of experimental models. These events are rare; occur in the body at unknown places and on a microscopic scale. The loss of cell adhesion of tumor cells detaching from the primary tumor tissues will undergo a transformation phenomenon known as epithelial-to mesenchymal transition (EMT) leading to the loss of epithelial characteristics with different expression patterns of EMT markers (E-cadherin, N-cadherin, Vimentin, Snail1/2, Twist1/2, ZEB1/2). The changes in mechanical and physical properties of interacting cells during morphological and malignant transformation are investigated and their quantifications measured. Here, microfluidic models containing mechanical constrictions in order to mimic the blood microcirculation have been designed and fabricated. Metastatic breast cancer cells are subjected and confined to the microfluidic channels using a flow control system. These cells are circulated under optimal culture conditions, and monitored in the channels for the observance of biophysical occurrences from continuous mechanical cellular deformations. The biophysical effects of circulation and confinement on tumor cell morphogenesis will be investigated.

Mechanical properties of cells and their environment have an undeniable impact on physiological and pathological processes such as tissue development or cancer metastasis. Hence, there is a pressing need for establishing and validating methodologies for measuring the mechanical properties of cells, as well as for deciphering the molecular underpinnings that govern the mechanical phenotype. During my doctoral research, I addressed these needs by pushing the boundaries of the field of single-cell mechanics in four projects, two of which were method-oriented and two explored important biological questions. First, I consolidated real-time deformability cytometry as a method for high-throughput single-cell mechanical phenotyping and contributed to its transformation into a versatile image-based cell characterization and sorting platform. Importantly, this platform can be used not only to sort cells based on image-derived parameters, but also to train neural networks to recognize and sort cells of interest based on raw images. Second, I performed a cross-laboratory study comparing three microfluidics-based deformability cytometry approaches operating at different timescales in two standardized assays of osmotic shock and actin disassembly. This study revealed that while all three methods are sensitive to osmotic shock-induced changes in cell deformability, the method operating at the shortest timescale is not suited for detection of actin cytoskeleton changes. Third, I demonstrated changes in cell mechanical phenotype associated with cell fate specification on the example of differentiation and de-differentiation along the neural lineage. In the process of reprogramming to pluripotency, neural precursor cells acquired progressively stiffer phenotype, that was reversed in the process of neural differentiation. The stiff phenotype of induced pluripotent stem cells was equivalent to that of embryonic stem cells, suggesting that mechanical properties of cells are inherent to their developmental stage. Finally, I identified and validated novel target genes involved in the regulation of mechanical properties of cells. The targets were identified using machine learning-based network analysis of transcriptomic profiles associated with mechanical phenotype change, and validated computationally as well as in genetic perturbation experiments. In particular, I showed that the gene with the best in silico performance, CAV1, changes the mechanical properties of cells when silenced or overexpressed. Identification of novel targets for mechanical phenotype modification is crucial for future explorations of physiological and pathological roles of cell mechanics. Together, this thesis encompasses a collection of contributions at the frontier of single-cell mechanical characterization across timescales and cell state transitions, and lays ground for turning cell mechanics from a correlative phenomenological parameter to a controllable property.:Abstract Kurzfassung List of Publications Contents Introduction Chapter 1 — Background 1.1. Mechanical properties as a marker of cell state in health and disease 1.2. Functional relevance of single-cell mechanical properties 1.3. Internal structures determining mechanical properties of cells 1.4. Cell as a viscoelastic material 1.5. Methods to measure single-cell mechanical properties Aims and scope of this thesis Chapter 2 — RT-DC as a versatile method for image-based cell characterization and sorting 2.1. RT-DC for mechanical characterization of cells 2.1.1. Operation of the RT-DC setup 2.1.2. Extracting Young’s modulus from RT-DC data 2.2. Additional functionalities implemented to the RT-DC setup 2.2.1. 1D fluorescence readout in three spectral channels 2.2.2. SSAW-based active cell sorting 2.3. Beyond assessment of cell mechanics — emerging applications 2.3.1. Deformation-assisted population separation and sorting 2.3.2. Brightness-based identification and sorting of blood cells 2.3.3. Transferring molecular specificity into label-free cell sorting 2.4. Discussion 2.5. Key conclusions 2.6. Materials and experimental procedures 2.7. Data analysis Chapter 3 — A comparison of three deformability cytometry classes operating at different timescales 3.1. Results 3.1.1. Representatives of the three deformability cytometry classes 3.1.2. Osmotic shock-induced deformability changes are detectable in all three methods 3.1.3. Ability to detect actin disassembly is method-dependent 3.1.4. Strain rate increase decreases the range of deformability response to actin disassembly in sDC 3.2. Discussion 3.3. Key conclusions 3.4. Materials and methods Chapter 4 — Mechanical journey of neural progenitor cells to pluripotency and back 4.1. Results 4.1.1. fNPCs become progressively stiffer during reprogramming to pluripotency 4.1.2. Transgene-dependent F-class cells are more compliant than ESC-like iPSCs 4.1.3. Surface markers unravel mechanical subpopulations at intermediate reprogramming stages 4.1.4. Neural differentiation of iPSCs mechanically mirrors reprogramming of fNPCs 4.1.5. The closer to the pluripotency, the higher the cell stiffness 4.2. Discussion 4.3. Key conclusions 4.4. Materials and methods Chapter 5 — Data-driven approach for de novo identification of cell mechanics regulators 5.1. Results 5.1.1. An overview of the mechanomics approach 5.1.2. Model systems characterized by mechanical phenotype changes 5.1.3. Discriminative network analysis on discovery datasets 5.1.4. Conserved functional network module comprises five genes 5.1.5. CAV1 performs best at classifying soft and stiff cell states in validation datasets 5.1.6. Perturbing expression levels of CAV1 changes cells stiffness 5.2. Discussion 5.3. Key conclusions 5.4. Materials and methods Conclusions and Outlook Appendix A Appendix B Supplementary Tables B.1 – B.2 Supplementary Figures B.1 – B.9 Appendix C Supplementary Tables C.1 – C.2. Supplementary Figures C.1 – C.5 Appendix D Supplementary Tables D.1 – D.6 Supplementary Figures D.1 – D.7 List of Figures List of Tables List of Abbreviations. List of Symbols References Acknowledgements