Kiebler et al [1] was the first group to describe the use of a discontinuous iodixanol gradient in the isolation of dendritic spines from mouse hippocampus. The gradient comprised four layers of 9%, 12.5%, 15% and 25% (w/v) iodixanol (equivalent to densities of 1.076, 1.095, 1.105 and 1.152 g/ml); the densest solution contained the material pelleted from the hippocampal homogenate at 900 g for 10 min. The gradient was centrifuged at 18,000 g for 20 min and it was described as a velocity flotation separation. 

The primary aim of the gradient was to isolate dendritic spines and the material that banded at the 9%/12.5% interface was further fractionated in a secondary Percoll-sucrose gradient. The basic methodology has subsequently been adapted to the isolation of synaptosomes. Bagni et al [2] were the first group to report this approach and they reversed the order of the Percoll and the iodixanol flotation gradients. 

Material at the 15-23% Percoll interface has generally been used for subsequent purification in the iodixanol gradient, which is generally modified to include a denser layer of 35% (w/v) iodixanol and the centrifugation conditions changed to 10,000 g for 20 min. Synaptosomes band at the 15%/25% iodixanol interface. Essentially the same methodology has been used by other workers [3-10].