A robust dynamic 3D liver spheroid platform for (sub)chronic toxicity assessment: Metabolic characterization of HepG2...

In vitro assessment of (sub)chronic chemical exposure requires liver models that maintain stable functionality during long-term cultivation and whose baseline behavior is well characterized. HepG2 cells remain widely used in toxicology due to their robustness and reproducibility; however, their hepatoblastoma origin confers persistent tumor‑associated traits. Although three‑dimensional (3D) HepG2 spheroids provide greater physiological relevance than two‑dimensional (2D) cultures, their baseline behavior during prolonged dynamic cultivation remains insufficiently characterized, limiting confident interpretation of repeated-dose studies. Here, we establish a baseline characterization framework for long‑term dynamic 3D liver spheroid cultures and apply it to HepG2 cells as a proof-of-concept model in a clinostat‑based bioreactor system. Spheroids self‑assembled from single‑cell suspensions and were maintained for 28 days. Longitudinal characterization combined ¹H NMR-based exometabolomic profiling with gene expression and functional analyses to assess metabolic activity and hepatic performance over time. Despite different growth trajectories, dynamic and static spheroids reached comparable final diameters and exhibited stabilization of albumin secretion following an initial maturation phase. Dynamic HepG2 spheroids exhibited sustained energy metabolism and maintained hepatic functionality, including albumin and urea production, with a consistent functional window observed between Days 14 and 28. Persistent fetal‑ and cancer‑associated features were evident, including elevated alpha‑fetoprotein expression and increased one‑carbon metabolism reflected by progressive formate release. By establishing baseline metabolic and functional trajectories, this study provides a foundation for the design and interpretation of repeated-dose exposure studies. The framework is intended to be transferable to other liver models, enabling systematic cross-model benchmarking and supporting more robust application of in vitro systems in toxicology.