Oxygen diffusion limitations in pancreatic islet culture and immunoisolation

Abstract

Data for oxygen consumption by rat islets of Langerhans in a batch microreactor were fitted using a numerical solution of the transient oxygen diffusion-reaction equation. Average best-fit values were 3.1 +/- 0.7 x 10-8 mol/cm3-s for the maximum oxygen consumption rate Vmax in aminoacid-free media and 1.2 +/- 0.4 x 10-14 mol/cm-mm Hg-s for the oxygen permeability in islet tissue. These parameter values, along with a 30-60% positive correction for the presence of aminoacids in Vmax, were used to predict oxygen profiles inside and around islets under perifusion, culture, and immunoisolation conditions. The difference between Michaelis-Menten and zero-order kinetics and the role of the necrosis process in modeling of oxygen profiles were found to increase with the severity of hypoxia. Internal and external diffusion limitations were characterized for homogeneously dispersed tissue. Oxygen profiles were determined with finite differences in perifused rat islets for which second-phase insulin secretion data were available. Data were fitted with ad hoc kinetic models describing the effect of local pO2 on insulin secretion. A two-step, one-parameter model that assumed that local insulin secretion rate as a function of local pO2 is first-order for pO2 < P and zero-order for PO2 > P* resulted in the best data fit for P* values between 2 and 10 mm Hg, depending on the value of Vmax used. Oxygen profiles were estimated with finite elements for the axi-symmetric problem of single islet culture and the model did an excellent job in predicting loss of viability data, obtained using Trypan blue staining,

(cont.) as a function of islet diameter both under normoxic (ambient pO2 = 142 mm Hg) and hypoxic (40 mm Hg) conditions. The model was extended to massive islet culture and the effects of islet surface density and medium depth on viability were characterized and suggestions were made for the improvement of porcine islet culture conditions. In bioartificial pancreas devices we found that there is an optimal islet surface density (NS)opt for which insulin secretion rate is maximized, while secretory efficiency decreases monotonically with tissue density above a critical value. The design tissue density must be chosen in the range between this critical value and (Ns)opt' and its value depends on whether minimization of the device size or the amount of loaded tissue is more important.