AbstractIn the present work, we introduce a scheme for efficient mechanobiological research on the dedifferentiation of chondrocytes (CHs) using a deep neural network (DNN) model that does not require sacrificing the cells and thus, saving resources. Cells are a part of active biological systems subjected to physical stimuli such as mechanical loading. Such loading affects cellular processes including proliferation, differentiation, and the interplay with the surrounding environment. CHs are mechanosensitive cells that produce and maintain the cartilaginous matrix that allows cartilage to bear and distribute mechanical loads in joints; their role in cartilage regeneration and treatment of osteoarthritis (OA) has been the focus of research projects. Nevertheless, it remains an open challenge to fully understand the effect of mechanical stimuli on CHs. One of the unresolved mechanobiological behaviors of CHs is dedifferentiation. Dedifferentiation of CHs is the phenomenon where isolated CHs alter their phenotype when cultured in a 2D in vitro environment over passages. While intact, cobblestone‐shaped CHs mainly produce collagen II, dedifferentiated CHs with elongated shapes produce fibroblastic collagen I. This limits the scalability of a promising treatment for OA, called ‘autologous chondrocyte implantation (ACI)’. To overcome this limit, research is being carried out to understand the dedifferentiation mechanism in detail. This proposed scheme is composed of three parts: (i) the cell‐seeded specimens are loaded in a bioreactor system, (ii) an optical microscope is used to obtain the phase‐contrast images of living cells, and (iii) the images are analyzed using a DNN. This scheme provides an efficient tool to analyze the ratio of intact to dedifferentiated CHs while preserving cells on our way to elucidate the effects of mechanical loading on the dedifferentiation of CHs.
Read full abstract