Repurposing FDA-approved drugs for the treatment of osteoarthritis using high-throughput screening in microphysiological models

Abstract: Osteoarthritis (OA) is a painful and disabling joint disorder affecting more than 30 million Americans. Current treatments for OA are primarily to relieve pain and maintain mobility, and none of them are clinically demonstrated to reverse OA progression. Developing novel disease-modifying OA drugs (DMOADs) is

challenging because there are no demonstrated molecular targets. The heterogeneous clinical presentations of OA add additional barriers to developing DMOADs. While global efforts are being channeled into understanding each OA endotype and discovering novel compounds, repurposing FDA-approved drugs represents a cost-efficient and time-saving solution to develop DMOADs, given that their safety in humans has

been previously validated. In particular, three primary OA endotypes have been identified, which include 1) cartilage-driven endotype, 2) bone-driven endotype, 3) inflammation-driven endotype. Selectively targeting these three endotypes represents a practical and generic strategy to discover DMOADs from the FDA-

approved drugs. Importantly, repurposing thousands of FDA-approved drugs needs to be performed on a platform possessing high-throughput screening capacity. The current OA animal models are not compatible with such a large-scale study. In addition, the preclinical efficacy of novel therapeutics defined in animal

models is poorly translated in clinical trials, raising concern about their predictive utility. Therefore, for drug repurposing, using human cell-derived in vitro models that can selectively stimulate OA endotype(s) represents a feasible and efficient strategy. Recently, we developed an OA cartilage model by chondroinducing in vitro-

aged human mesenchymal stromal cells (hMSCs), which simulated many key changes observed in OA chondrocytes. In addition, we have mixed hMSC-derived fibroblasts and human monocyte-derived macrophages to simulate the two key cell types in synovial membranes: fibroblast- and macrophage-like synoviocytes. To overcome the limitation of these single-tissue models in simulating tissue crosstalk in disease

progression and drug treatment, our team has also generated an in vitro microphysiological joint chip (miniJoint) that integrates the osteochondral, synovial, and adipose analogs. In this application, we propose to use the three in vitro models to repurpose FDA-approved drugs for OA treatment by defining their capacity to reverse

cartilage and/or inflammation-driven endotypes. The drug library (~2,400) will be first screened with cartilage and synovial models, and the most promising compounds and their combinations will be further examined in the miniJoint model. To overcome the limitations of using primary cells, such as finite cell number and donor-to-

donor variation, all three models will be created by human induced pluripotent stem cells (hiPSC). The successful completion of this study will generate a potential DMOAD pool that is ready for study in animal and human clinical trials. Testing different combinations of these drugs at varying doses in the human cell-derived

miniJoint model will also facilitate a personalized approach to the treatment of OA.