The transport of ferritin molecules by endothelial cell vesicles has been quantitatively investigated by electron microscopy. Single mesenteric capillaries of pithed frogs were perfused with solutions containing 6.7 g ferritin 100 ml.-1 for known periods before fixation in situ with osmium tetroxide. 2. Two series of experiments were carried out: in the first series the perfusate contained bovine serum albumin (1.0 g 100 ml.-1); in the second series the perfusate contained no protein other than the ferritin. To assess the molecular radius of ferritin in solution, the free diffusion coefficient of ferritin was measured in the presence and absence of albumin. 3. The free diffusion coefficient of ferritin in saline solution (110 m-mole 1.-1) was found to be 0.35 X 10(-6) cm2 sec-1 at 21 degrees C and was not affected by the presence of bovine serum albumin. This indicates that there is no significant binding of albumin to ferritin in solution and yields a value for the Stokes-Einstein radius of ferritin of 6.1 nm. 4. In all perfusion experiments the percentage of luminal vesicles containing ferritin exceeded the percentage of labelled cytoplasmic vesicles, which in turn exceeded the percentage of labelled abluminal vesicles. 5. Labelling of all vesicle populations was seen after perfusions lasting less than 1 sec. At this time luminal vesicles were more heavily labelled in the absence of albumin. 6. The labelling of luminal vesicles increased with lengthening perfusion times up to 30-40 sec, after which steady levels of labelling were achieved. The rate of rise in luminal labelling and the steady-state levels reached were both greater in the absence of albumin. By contrast cytoplasmic labelling increased above its initial value only after perfusions of longer than 10 sec. 7. In the steady state, labelled cytoplasmic vesicles contained, on average, fewer ferritin molecules than labelled luminal vesicles. This finding is inconsistent with translocation of labelled luminal vesicles across the cell. 8. It is suggested that the early constant labelling of cytoplasmic and abluminal vesicles is consistent with the existence of vesicular channels. Later cytoplasmic labelling may result from the transient fusion of cytoplasmic vesicles with labelled luminal vesicles for periods long enough to allow mixing of vesicular contents. Albumin may affect vesicular transport by its interaction with the endothelial glycocalyx.
