Three-dimensional platform-dependent modulation of stemness and EMT-associated gene expression in primary and metastatic uveal melanoma models.

Uveal melanoma is the most common primary intraocular malignancy in adults and is associated with high rates of metastatic progression. Although local tumour control is often achieved, effective therapies for metastatic disease remain limited. Three-dimensional culture systems more closely replicate aspects of the tumour microenvironment than conventional two-dimensional monolayers; however, differences between platforms and their impact on stemness- and epithelial-to-mesenchymal transition-associated transcription in uveal melanoma remain incompletely defined. Primary-derived and metastatic-derived uveal melanoma cell lines were cultured under four three-dimensional conditions, including AggreWell 800™, Biofloat™, Happy Cell Advanced Suspension Medium™, and hanging drop systems. Spheroid morphology and volume were quantified over ten days. Expression of cancer stem cell markers (CD133, CD24, CD44) and epithelial-to-mesenchymal transition-associated genes (E-cadherin, ZO-1, N-cadherin, Vimentin, Snail1, Snail2) was assessed relative to two-dimensional controls using quantitative polymerase chain reaction. All platforms supported spheroid formation, although growth dynamics differed between systems and site of tumour origin. Aggregation-based platforms produced more compact spheroids and greater volumetric expansion, particularly in metastatic-derived lines. Enhanced expression of CD133 and N-cadherin mRNA was observed in three-dimensional growth platforms, most prominently under structured aggregation conditions. Modulation of epithelial markers and transcription factors was variable and did not uniformly reflect complete epithelial-to-mesenchymal transition. Three-dimensional culture architecture was associated with differential modulation of stemness- and epithelial-to-mesenchymal transition-associated transcription in uveal melanoma. Platform selection and tumour origin were associated with differences in the phenotype observed in vitro and should be carefully aligned with specific biological endpoints when modelling disease progression.