Measurement of the density of human platelets and its relationship to volume.

New methods are described for platelet isolation and buoyant density determination using low-speed centrifugation in continuous density gradients of Percoll. The conditions used do not induce loss of granule or cytoplasmic markers and enable reproducible platelet frequency distributions to be obtained in linear density gradients. Such frequency distributions are normal with a mode of 1.0645 +/- 0.0015 g cm-3 (mean +/- SD, n=20). Platelets fixed in 0.1% glutaraldehyde show a modal density of 1.0712 +/- 0.0005 g cm-3. Content of protein, lactate dehydrogenase, beta-thromboglobulin and 3H-serotonin correlate closely with platelet numbers throughout the density distribution. The frequency distribution of platelet volume between 2.2 and 21 fl fits a log normal model and cell volume in density subfractions from the most dense to the least dense also approximate log normality. There is a positive correlation between mean platelet volume and buoyant density with a small increment between the least and the most dense extremes. Platelet subfractions separated by volume using a FACS II cell sorter differ substantially from each other in cell volume but the difference in mean density of four different volume fractions in negligible. In discontinuous density gradients of Stractan factors other than platelet density must influence the separation of platelets, as rebanding of platelets from interfaces shows a wide variation in buoyant density when analysed in continuous gradients. It is concluded that analysis of platelet buoyant density in continuous Percoll gradients supports the view that platelet density, like platelet volume, is determined primarily during thrombocytopoiesis and that volume and density are largely independent elements of platelet heterogeneity.