Aggregation kinetics of well and poorly differentiated human prostate cancer cells.

Aggregation of attachment-dependent animal cells represents a series of motility, collision, and adhesion events applicable to such diverse fields as tissue engineering, bioseparations, and drug testing. Aggregation of human prostate cancer cells in liquid-overlay culture was modeled using Smoluchowski's collision theory. Using well (LNCaP) and poorly differentiated (DU 145 and PC 3) cell lines, the biological relevance of the model was assessed by comparing aggregation rates with diffusive and adhesive properties. Diffusion coefficients ranged from 5 to 90 microm(2)/min for single LNCaP and PC 3 cells, respectively. Similar diffusivities were predicted by the persistent random walk model and Einstein relation, indicating random motion. LNCaP cells were the most adhesive in our study with reduced cell shedding, 100% adhesion probability, and enhanced expression of E-cadherin. There was an increase in DU 145 cells staining positive for E-cadherin from nearly 20% of single cells to uniform staining across the surface of all aggregates; under 30% of PC 3 aggregates stained positive. Aggregation rates were more consistent with adhesive properties than with motilities, suggesting that aggregation in our study was reaction-controlled. Relative to other assays employed here, aggregation rates were more sensitive to phenotypic differences in cell lines and described size-dependent changes in aggregation at a finer resolution. In particular, model results suggest similar aggregation rates for two-dimensional DU 145 and PC 3 aggregates and upwards of 4-fold higher rates for larger three-dimensional DU 145 spheroids, consistent with expression of E-cadherin. The kinetic model has application to spheroid production, to cell flocculation and as an adhesion assay. Copyright 2002 Wiley Periodicals, Inc.