Cell-free systems have been successfully used to clarify some of the relationships between earlier parts of the endosomal system [ I , 21. The later stages of the process, by which endocytosed material appears in lysosomes for digestion, have received less attention. Indeed, the fundamental mechanism of this process is still in question. The maturation theory [ 31 holds that early endosomes are gradually converted, first to late endosomes, and then to lysosomes. The conversion occurs by sequential addition of the various components that are characteristic of each organelle to the compartment containing the ligand and subtraction of other components from the compartment. The alternative vesicular-transfer theory [4] proposes that early endosomes, late endosomes and lysosomes preexist as discrete compartments, through which incoming ligand is sequentially transferred by vesicular traffic between one compartment and the next. We have previously shown [5] that incubation of rat-liver postmitochondrial supernatants, containing endosomes loaded with “’I-labelled asialofetuin, at 37°C in the presence of ATP, led to the appearance of radiolabel at lysosomal densities, if, and only if, lysosomes were present. However, this was a relatively crude cell-free system and the experimental system only demonstrated interaction between endosomes and lysosomes rather than fusion. We have now demonstrated that this interaction can be reproduced using isolated endosomes and lysosomes, and that true fusion is occurring. Our initial experiments using crude postmitochondrial supernatants showed that the earliest labelled endosomes are not those that are involved in the interaction with lysosomes. Perfused livers that were given a single-pass dose of asialo-orosomucoid-horseradish peroxidase conjugate to permit the visualization of ligand-containing structures under the electron microscope, have now been compared with livers that had been similarly treated with a radiolabelled ligand. Such comparisons show that, although the compartment which interacts with lysosomes is maximally labelled after only 4 min, it lies deeper in the cell than the earliest labelled endocytic structures, which are close to the blood sinusoids. The interaction between these later endosomes and lysosomes has been reconstructed using isolated fractions [6]. Maximum interaction as measured by centrifugal analysis required the presence of cytosol, an ATP-regenerating system and GTP, in addition to endosomes and lysosomes. In the absence of lysosomes, endosomal density altered on incubation, but never in such a way as to reach lysosomal density. In this, the rat-liver system differed from a similar system derived from 3T3 cells [7]. Although cytosol was required for the maximal interaction, some interaction occurred in its absence. This enabled us to examine possible membrane fusion between the endosome and lysosome fractions by a fluorescence-dequenching method [6]. Endosomal membranes loaded to self-quenching concentrations with octadecylrhodamine showed fluorescence increases, indicating dequenching and hence fusion, when incubated at 37°C with lysosomes, but not with mitochondria or unlabelled endosomes. However, these assays could only be performed in the absence of cytosol, since cytosol contains phospholipid-exchange proteins and fatty-acid-binding proteins. In order to demonstrate fusion unequivocally and in the presence of cytosol, we have therefore developed a content-mixing assay. Endosomes loaded with an avidin-asialofetuin conjugate were incubated with lysosomes containing biotinylated, radiolabelled polymeric IgA in the presence of a large excess of unlabelled biotinylated insulin. Radiolabelled polymeric IgA-biotin-avidin complexes, which could only have formed within fused organelles, were immunoprecipitated using rabbit anti-avidin and sheep anti-rabbit IgG-coated magnetic beads. Unfortunately, both ligands suffered partial digestion during the incubation, severely limiting the incubation times and making the assay relatively insensitive and variable. Nonetheless, 26 such experiments showed that 21.2 k 5.5% fusion had occurred after 10 min incubation. We have therefore demonstrated that, in cellfree preparations from rat liver, endosomes can associate and fuse with pre-existing lysosomes under conditions which do not permit the maturation of endosomes to lysosomes. This suggests that in these cells at least, some vesicular transport between pre-existing compartments occurs. 72 I