Isolated blood eosinophils are routinely used to study eosinophil activation mechanisms. However, as revealed by ultrastructural analysis, different isolation protocols may yield purified eosinophils with marked variability in granule electron density. In this study, using eosinophil peroxidase (EPO) histochemistry and transmission electron microscopy (TEM), we have compared the morphology of eosinophils in immediately fixed whole blood (to represent a morphological baseline) with isolated eosinophils purified by a number of protocols. Eosinophils in whole blood contained intact specific secondary granules of which a few exhibited marginal coarsening of matrix electron density (4% (95% CI: 2 to 7) altered granules per eosinophil). By contrast, eosinophils purified according to standard protocols, which included erythrocyte lysis with either ammonium chloride or distilled water, showed moderate to extensive loss in density of secondary granule core and/or matrix (NH4Cl: 62% (95% CI: 58 to 66); dH2O: 37% (95% CI: 30 to 44) altered granules). Stepwise analysis of eosinophils during the cell separation processes indicated that the granule abnormalities seen following erythrocyte lysis were further increased following immunomagnetic separation. However, when erythrocyte lysis was omitted, by use of a two-layered Percoll gradient (1.076 g/ml/1.088 g/ml) to which diluted whole blood was applied directly, eosinophils with minimal granule abnormalities (11% (95%CI: 9 to 13) altered granules) could be obtained after immunomagnetic separation. In conclusion, to obtain eosinophils with granule morphology more closely resembling the whole blood baseline phenotype, erythrocyte lysis should be avoided when separating eosinophils from human blood. Thus it will be possible to study in vitro the early transformation of resting eosinophils into the degranulating phenotype found in diseased tissues.
Copyright 2004 Elsevier B.V.