Background/Objectives: Advanced 3D cell culture techniques enhance the physiological relevance of in vitro models, while supporting the 3Rs principles (Reduction, Refinement, and Replacement) of animal experimentation. In this context, 3D collagen-based systems mimic key extracellular matrix properties, enabling more accurate cellular organization and phenotype. However, changes in culture dimensionality can affect RT-qPCR reference gene stability, underscoring the need for careful validation when combining 2D and 3D systems. Methods: AML12 cells were cultured for 7 days under different 2D and collagen-based 3D conditions. The expression stability of nine candidate housekeeping genes was systematically evaluated using established algorithms (BestKeeper, NormFinder, geNorm, RefFinder, and ΔCt method), followed by inter-group statistical and correlation analyses of raw Ct values. Albumin gene expression was used as a target gene. Results: Although all candidate genes initially met acceptable variability thresholds, a stepwise, exclusion-based analysis revealed distinct performance differences. Hprt, Ppia, and Actb emerged as the most stable, showing no intra-group variability or interaction with Albumin expression. Nevertheless, Ywhaz and Rplp0, despite their high stability, were compromised by significant correlation with Albumin. Furthermore, Ywhaz showed significant downregulation under 3D culture conditions. B2M, Gapdh, 18S, and Hmbs exhibited increased variability, likely reflecting metabolic and microenvironmental heterogeneity associated with prolonged 2D cultivation of AML12 cells. Conclusions: Overall, this study highlights the importance of context-dependent, exclusion-based reference gene validation when comparing 2D and 3D models, and demonstrates a new approach for reliable gene expression normalization in complex in vitro culture systems.

The present study evaluated the stability of candidate reference genes during adipogenic differentiation of 3T3-L1 cells cultured on different extracellular matrices. The aim was to investigate the effects of matrix composition and differentiation stage on the expression of candidate housekeeping genes and to compare validation strategies in dynamic in vitro models. Eleven candidate reference genes (18S, Actb, B2m, Gapdh, Hmbs, Hprt, Nono, Ppia, Rplp0, Tbp, and Ywhaz) were analyzed by RT-qPCR in 3T3-L1 cells cultured on TC, collagen, gelatin, and Matrigel at Days 7 and 14 of differentiation. Gene stability was assessed using geNorm, NormFinder, RefFinder, comparative ΔCt, BestKeeper, generalized linear model (GLM), linear mixed model (LMM), and correlation analyses with the adipogenic markers Pparg and Fasn. The results demonstrated that the expression of most housekeeping genes was influenced by matrix composition, differentiation stage, or their interaction. Actb and 18S exhibited the strongest condition-dependent variability and pronounced matrix sensitivity. Gapdh and Hprt showed significant correlations with both Pparg and Fasn, while Hmbs correlated with Fasn, suggesting that these reference genes may not be fully independent of adipogenic status. Ppia demonstrated markedly contrasting rankings across analytical approaches, highlighting limitations of single-method stability assessment. The findings confirm that universal housekeeping genes are unlikely to exist across different matrix conditions and differentiation stages. The results highlight the need for multi-level validation strategies and experimentally validated normalization panels to minimize normalization bias and avoid misleading RT-qPCR expression profiles. Functional validation identified B2m and Rplp0 as the most suitable two-gene normalization panel for the experimental model evaluated, whereas Tbp remained a strong complementary reference gene candidate.