UCLA

The measurement of cellular mechanical properties can be impactful in many areas of biosciences due to the interdependencies of mechanics and cell state or function. Mechanical cues communicated via mechanosensing and mechanotransduction can regulate cellular behavior leading to biochemical and structural modifications; conversely, native cellular processes such as differentiation, activation, and malignant transformation triggered by biochemical cues have also shown to elicit significant changes to the cellular architecture. Measurement of these mechanical biophysical changes can therefore be used to infer cell state or function. To date, there are numerous approaches used to measure the mechanical properties of cells for biophysics research, yet these are limited in their translational ability as general research or clinical tools. The biggest challenges facing successful translation of these mechanical phenotyping tools are due to the technological complexity, manual sample handling and preparation requirements, and limited sample throughput. This dissertation focuses on developing a high-throughput, label-free alternative for mechano phenotyping termed deformability cytometry. For the first time in the biophysics field, the deformability cytometry platform achieves throughputs of more than 1,000 cells/second; a rate that is three orders of magnitude greater than previous techniques. Additionally, initial work involving platform validation and proof-of-concept applications for stem cell differentiation, leukocyte activation, and cancer diagnostics are explored. The dissertation also undertakes a major engineering challenge of this system - large dataset image processing - which has led to the development of process efficient image analysis algorithms to increase the utility and robustness of this mechano phenotyping technique. Lastly, a clinical proof-of-concept study consisting of 119 patient pleural effusion samples were collected and assayed using the deformability cytometry platform to determine the diagnostic performance of mechanical biomarkers for malignant effusions. Briefly, the result of this clinical study achieved an area-under-the-curve of 0.91 with sensitivity and specificity of 100% and 69%, respectively. The high diagnostic accuracy combined with the ease-of-use, minimal sample preparations needs, and large sample sizes, satisfies many translational hurdles for a label-free mechano phenotyping platform for applications in the biosciences community.