(595b) Uncontrolled Macropinocytosis Leading to Glioblastoma Cell Death Involves Disruptions to the Cytoskeletal Network.

Glioblastoma is the deadliest and most common form of adult brain cancer, with a median survival of just 15 months even with standard-of-care surgical resection, radiation, and chemotherapy. Treatment options have remained unchanged in the last twenty years, motivating a need for therapeutic strategies with novel modes of action. Our recent efforts focus on promoting methuosis, an understudied, non-apoptotic cell death mechanism characterized by uncontrolled macropinocytosis leading to catastrophic vacuolization and cell rupture. Macropinocytosis is an adaptation to metabolic stress in rapidly proliferating cancer cells that enables the non-specific uptake of essential nutrients from extracellular fluid. Macropinosome formation and intracellular trafficking rely heavily on cytoskeletal rearrangements and movement along cytoskeletal fibers, but the role of the cytoskeleton has not been studied in methuosis. In this study, we compared two methods for inducing methuosis – the small molecule MOMIPP, an inhibitor of inositide kinase PIKfyve, and doxycline-inducible expression of a constitutively active mutant form of the oncogenic GTPase HRAS. In U251 human glioblastoma cells, MOMIPP induces methuosis within hours, but inducible G12V HRAS expression requires days to achieve the same level of hyper-vacuolization, as determined by phase-contrast imaging and fluorescent imaging of fluid-phase trackers such as lucifer yellow. The difference in methuosis induction time scales between MOMIPP and G12V HRAS leads to questions about the degree to which the underlying mechanisms are common. Early indications that cytoskeletal disruptions are involved in methuosis are provided by our findings that paclitaxel (microtubule stabilizing small molecule) and latrunculin (an actin depolymerizer) forestall vacuole accumulation in MOMIPP-treated U251 cells. For both modes of methuosis induction, multi-channel fluorescence imaging reveals major disruptions to the cytoskeletal network in U251 cells due to the exclusion of substantial intracellular volume to the normal cytoskeletal network. We hypothesize the different time scales of chemically or genetically driven methuosis will result in unique rearrangements of vimentin, tubulin, and actin. To quantify these differences systematically with temporal resolution, we are developing a novel image analysis pipeline that combines phase and fluorescence high-content imaging with machine learning-based image segmentation in Ilastik and fluorescence segmentation in ImageJ and CellProfiler. The pipeline will be used to quantify and correlate vacuole accumulation and cytoskeletal protein rearrangements to determine if cytoskeletal disruptions occur prior to methuosis or result from methuosis.Moving forward, the imaging pipeline will be extended to a multi-round iterative immunofluorescence platform that will provide a systems-level approach to measure protein phosphorylation states. The expanded imaging platform will aid in the nomination of specific drug targets whose antagonism will promote catastrophic vacuolization in glioblastoma cells for therapeutic benefit.