3D microphysiological model to evaluate the effects of atovaquone as a therapeutic agent for high-grade serous ovarian carcinoma (HGSOC).

Introduction: High-grade serous ovarian carcinoma, or HGSOC, is the gynecological malignancy with the highest mortality rate. Due to a lack of reliable tests and vague external symptoms, HGSOC is typically diagnosed at advanced stages. When detected, treatment options are limited and result in a 60% recurrence rate. Consequently, there is an urgent need for new therapeutic agents for HGSOC. Atovaquone is an FDA-approved anti-malarial drug that has been proven effective against HGSOC in 2D models. Atovaquone is an OXPHOS inhibitor that binds to complex III and induces reactive oxygen species (ROS) leading to apoptosis. However, our knowledge about this mechanism is incomplete. HGSOC models present limitations to investigate the mechanisms underlying disease progression and real-time response to therapies. For instance, 2D in-vitro models fail to include crucial components of the tumor microenvironment (TME) like blood vessels for drug delivery and oxygen/nutrient gradients, which are known to deeply impact drug response. On the other hand, in-vivo models have shortcomings for studying the pharmacodynamics of drugs. To bridge this gap, we have designed a microphysiological system (MPS) to investigate the role of the TME in HGSOC response to therapeutics. Here, we use an MPS including an HGSOC cell spheroid and an adjacent 3D tubular blood vessel to study the delivery and mechanism of action of atovaquone. Using this HGSOC MPS, we will contribute to better understand the mechanism of action of atovaquone and accelerate its translation into the clinic as a sole or combination therapy agent.

Materials and

Methods: Microfluidic device fabrication and model generation: The microfluidic device consists of 2 layers of polydimethylsiloxane (PDMS, 10:1 base to curing agent ratio) sandwiching a 300µm rod and bonded to a glass plate to define the main chamber. The device was fabricated using standard soft lithography methodology, where we poured PDMS over an SU-8 wafer with the design of our device. The 2 layers of cured PDMS were plasma bonded to the glass dish. We used three cell lines (OVCAR5, SKOV3, and Kuramochi) to generate spheroids using the hanging drop method. Briefly, cells were resuspended in relevant media and 20% methylcellulose to form 20µl droplets with 2,500 cells each and incubated for 48h. Next, we harvested and seeded the spheroids and embedded them into the hydrogel containing 4.5 mg/ml collagen, neutralized with NaOH 0.5M, and with PBS 10X added to adjust osmolarity. After hydrogel polymerization, we removed the rod to reveal a tubular structure to mimic the blood vessel. Next, we added the atovaquone treatment through this structure (30µM).
Model readouts: Cells were stained with CellTracker (CMFDA) at a dilution of 1:1000 before forming spheroids to track growth overtime. To assess viability, we stained devices with propidium iodide (PI, 2.5 ug/ml) and calcein AM (CAM, 2 ug/ml) for 30 min prior to imaging. Next, spheroids were fixed (PFA, 4%, 20 min) to perform immunofluorescence to probe for Ki67 (anti-mouse, 1:50). Z-stacks of the MPS were captured with a Leica Thunder microscope and analyzed using ImageJ to quantify area measurements.

Results, Conclusions, and Discussions: We evaluated the effects of atovaquone on HGSOC cell spheroids after perfusing the drug through a tubular structure that mimics a blood vessel. Imaging revealed that treatment with atovaquone decreased the area invaded by spheroids in SKOV3, but not in OVCAR5, which suggests that a combination therapy may be needed with atovaquone. However, atovaquone did impact spheroid size growth over time, consistently with previous data reported in mice. This difference suggests that atovaquone was more effective in non-migratory cells. Ki-67 immunostaining revealed low proliferation of the HGSOC cells in spheroids in vehicle controls. Atovaquone-treated spheroids trended towards decreased cell proliferation in Kuramochi and OVCAR5 cancer cell spheroids (n=2), which is consistent with the effects of atovaquone in vitro. Our MPS model also allowed us to evaluate the timeline needed for the drug to diffuse from the blood vessel structure and affect the spheroids, which was shown to occur within 24h of addition. Future studies will leverage fluorescent reporters to evaluate the effects of atovaquone on specific ROS production (e.g., superoxide, singlet oxygen) and subsequent increase in apoptosis.

Conclusions: Our data revealed that atovaquone decreased spheroid growth but did not decrease the migration of HGSOC cell spheroids. Our model was successful in modeling the primary ovarian tumor microenvironment and provides insight into the pharmacodynamics of atovaquone in real time.

Acknowledgements (Optional):