Comparison of dilute and nondilute osmotic equilibrium models for erythrocytes.

Successful cryopreservation requires the addition of cryoprotective agents (CPAs). The addition of permeating CPAs, such as glycerol, is associated with some risk to the cells and tissues. These risks are both related to the CPA themselves (CPA toxicity) and to the volume response of the cell (osmotic damage). To minimize the potential for damage during cryopreservation, mathematical models are often employed to understand the interactions between protocols and cell volume responses. In the literature, this volume response is usually captured using ideal and dilute approximations of chemical potential and osmolality, an approach that has been called into question for cells in high concentrations of CPAs. To address this, the relevance of non-ideal and non-dilute models has been explored in a number of cell types in the presence of permeating CPAs. However, it has not been explored in erythrocytes, which have a cytosolic hemoglobin content of more than 20% by volume and are cryopreserved in 40% glycerol. Because hemoglobin has been suggested to be a highly non-ideal solute, if the non-ideal and non-dilute transport model is relevant to any cells, it should be relevant to erythrocytes. Here we investigate the use, and accuracy, of both the dilute and non-dilute models in predicting cell volume changes during CPA equilibration in erythrocytes, and demonstrate that using published values for the non-ideal and non-dilute model, applied to erythrocytes, leads to model predictions inconsistent with experimental data, whereas dilute approximations align well with experimental data.

Curve fitting approach for measurement of cellular osmotic properties by the electrical sensing zone method. I. Osmotically inactive volume.

We have investigated the confounding effects of dynamic range limitations on measurement of the osmotically inactive volume using electrical sensing zone instruments (e.g., Coulter counters), and propose an improved approach to parameter estimation. The conventional approach for analysis of cell size distributions measured by such particle sizing instruments requires data truncation: the mean cell volume is computed after exclusion of data below a specified lower bound (typically chosen to remove artifacts due to small-volume noise) and above an upper bound (typically governed by instrument limitations). The osmotically inactive volume is then estimated from a Boyle-van't Hoff plot of the averaged volume data obtained after exposure to various solution osmolalities. We demonstrate that systematic exclusion of data in the conventional approach introduces bias that results in erroneously high estimates of the osmotically inactive volume fraction. To minimize this source of error, we have devised a new algorithm based on fitting a bimodal distribution model to the non-truncated volume data. In experiments with mouse insulinoma (MIN6) cells, the osmotically inactive volume fraction was estimated to be 0.15+/-0.01 using the new method, which was significantly smaller than the estimate of 0.37+/-0.02 obtained using the conventional method (p<0.05). In silico experiments indicated that the parameter estimate obtained by the new method was accurate within 5%, whereas the error associated with the conventional approach was approximately 150%. Parametric analysis was used to elucidate the sensitivity of errors to variations in instrument dynamic range and cell volume distribution width.