Hydrodynamic mechanisms of cell damage in microcarrier bioreactors
Cherry, Robert Stephen
Doctor of Philosophy thesis
A stirred microcarrier bioreactor contains a variety of local fluid dynamic environments, including isotropic turbulence, boundary layers, and vortices trailing from the impeller blades, all of which are superimposed on a larger-scale rotation and figure eight circulation. Analysis of the interaction of a cell-covered microcarrier with this complex environment suggest there are three potentially important mechanisms of cell damage. These are interaction with turbulent fluid eddies of about the size of the microcarrier bead, collisions with other beads, and collisions against the impeller. These mechanisms may be characterized by the eddy/bead size ratio, turbulent collision severity, and impeller collision severity respectively. The severities are defined as the kinetic energy of that type of collision times the frequency at which it occurs. Measurement of growth and death rates of bovine embryonic kidney cells showed that both a modified eddy/bead size ratio and turbulent collision severity based on a shear collision mechanism fit the results for moderate to high levels of agitation. Both parameters must be calculated on the basis of power dissipation around the impeller, rather than throughout the entire reactor. Experiments with decreased bead size and increased viscosity of the medium gave improved growth rates in agreement with these parameters' predictions. At low levels of agitation the net growth rate of cells decreased. This was attributed to the formation of cellular bridges joining two or more beads together in a stable clump. The lack of branched clumps and the observation of artifacts of bridges indicate that these bridges are frequently destroyed. The increased death of cells may be explained by bridge breaking and by the clump's having a larger equivalent diameter which is predicted to be more damaging by both parameters. Modeling of the growth of contact-inhibited cells on a sphere predicts a continually decreasing apparent growth rate. Use of an exponential model for early growth gives approximately the correct growth rate. This model also predicts the observed beneficial effect of increasing the inoculum cell density.