Fusion between different cohorts of endosomes, or between endosomes and lysosomes, can be assayed in vitro. Donor endosomes are isolated from cells exposed to b-HRP, whereas recipient vesicles, endosomes or lysosomes, are isolated from cells exposed to avidin (Ward et al., 1997 ). Fusion is identified through the formation of an avidin–b-HRP complex. Isolation of donor endosomes at specific times after release of the hypo-K⁺ block permits analysis of the fusion capabilities of differently aged endosome populations. All endosome populations have a similar density on Percoll gradients, which is quite different from that of lysosomes (Ward et al., 1990 ). Endosome populations obtained from Percoll gradients were tested for fusion in vitro. Early endosomes, formed within 4 min of internalization (E4), were capable of homotypic fusion but could not fuse with lysosomes (Figure ​(Figure1A).1A). Endosomes isolated 8 min after release of the hypo-K⁺ block (E8) were also unable to fuse with lysosomes but were capable of homotypic fusion (Figure ​(Figure1B).1B). There was a gradient in heterotypic fusion capacity, in that E4 had a greater ability to fuse with E8 than with 12-min endosomes (E12). Endosomes internalized for less than 12 min were unable to fuse with lysosomes. Later endosomes (E12) gained the ability to fuse with lysosomes (Figure ​(Figure1C).1C). These endosomes fused poorly with early endosomes (E4). Lysosomes were also capable of homotypic fusion (Figure ​(Figure1D),1D), as demonstrated previously (Ward et al., 1997 ). These results indicate that each endosome population is capable of homotypic fusion and that endosomes show a change in the selectivity of heterotypic vesicle fusion after 8 min of maturation.

The genetic information encoding proteins required for many cellular processes is conserved in yeast and human, leading us to search for mammalian homologues of yeast proteins required for fusion events. Vam3p was initially identified as a vacuolar SNARE through genetic studies that demonstrated its role in vacuolar protein sorting (Darsow et al., 1997 ; Gotte and Gallwitz, 1997 ; Wada et al., 1997 ). Vam3p was also found to be required for vacuolar inheritance, which results from homotypic fusion of newly formed vacuolar vesicles present in budding cells (Ungermann et al., 1999 ). Pep12p was identified as being required for transport to the prevacuole (Becherer et al., 1996 ; Burd et al., 1997 ). Genetic and biochemical studies have demonstrated that in addition to Vam3p, Vam7p and the vacuolar SNAREs Nyv1p, Vti1p, and Ykt6p are implicated in homotypic vacuole fusion, perhaps as a fusion complex (Ungermann et al., 1999 ). Database searches identified Syntaxin 7 as a homologue of Vam3p and Pep12p (Wang et al., 1997 ). These studies demonstrate that Syntaxin 7 is a SNARE involved in late vesicle fusion in the endocytic pathway of alveolar macrophages.

Our studies demonstrating that Syntaxin 7 is a lysosomal SNARE are based on the ability of recombinant Syntaxin 7 to inhibit homotypic lysosome fusion in a dose-dependent manner. Syntaxin 7 was one of only two SNAREs or VPS proteins we tested that showed inhibitory activity in the late endocytic pathway. An affinity-purified antibody directed against recombinant Syntaxin 7 also inhibited homotypic fusion, and this inhibition could be partially prevented by previous addition of the Syntaxin 7 protein. We suspect that the recombinant protein only partially prevented inhibition of fusion because the protein, albeit at higher levels, could itself prevent fusion. Evidence for specificity is demonstrated by the facts that nonimmune or preimmune serum did not inhibit fusion and affinity-purified antibody affected the fusion of lysosomes but not early endosomes. Although the antibody inhibited lysosome fusion, it was unable to detect a rabbit protein by either Western analysis or immunoprecipitation. However, the antibody could detect an appropriately sized antigen in human cells, and this detection was ablated by the addition of recombinant GST-Syntaxin 7. A physical interaction between Syntaxin 7 and cognate molecules was shown by the binding of recombinant GST-Syntaxin 7 to lysosomes but not early endosomes.

Further evidence that our assay faithfully identifies vesicle-unique SNAREs is demonstrated by the specificity of Syntaxin 7. Syntaxin 7 inhibits lysosomal/late endosomal fusion events but not early endosome fusion events. Advani et al. (1999) demonstrated that h-VAMP-7 was required for late endosome–lysosome fusion. We have extended those studies to show that VAMP-7 is also involved in homotypic lysosome fusion but not early endosome fusion. Thus, two vesicle-specific SNAREs were identified by our assay. Are VAMP-7 and Syntaxin 7 cognate SNAREs? Studies are currently under way to address this question.

There is a discrepancy among the published studies on the location of Syntaxin 7. Two groups suggested that Syntaxin 7 was associated with early endosomes (Wong et al., 1998 ; Prekeris et al., 1999 ), whereas a third group suggested that Syntaxin 7 was localized to late endocytic compartments (Nakamura et al., 2000 ). Each study used different cell types and different methods for organelle identification. In the study by Wong et al. (1998) , early endosomes in A431 cells were defined by the addition of a mAb to surface transferrin receptors, which were then internalized and subsequently localized by indirect immunofluorescence. The secondary antibody was directed against the anti-transferrin receptor antibody. Because the antibody is multivalent, the intracellular distribution of transferrin receptors may not represent the native distribution and may not reflect early endosomes. Furthermore, the fluorescence observed reflects the localization of the antibody, not necessarily that of the receptor. The antibody may be localized to late endosomal compartments where it may be degraded, as suggested by older studies (Lesley et al., 1989 ). Prekeris et al. (1999) colocalized Syntaxin 7 and anti-transferrin receptor antibodies by both fluorescence and electron microscopy with the use of several different cell types. More recently, Nakamura et al. (2000) , again with the use of fluorescence and electron microscopy, localized Syntaxin 7 in NIH 3T3 and NRK cells to Lamp 2–positive compartments, suggesting a late endosomal/lysosomal location. Our studies used freshly obtained alveolar macrophages, whereas the other studies used cultured cell types. Alveolar macrophages are highly endocytic, and the localization of Syntaxin 7 on lysosomes may represent some adaptation for high-efficiency endocytosis. It is possible that our results, which rely on a functional assay for SNARE identification, reflect SNARE promiscuity. Recent studies postulate that SNAREs can be promiscuous and bind to a spectrum of cognates (Fasshauer et al., 1999 ). Thus, our finding of inhibition of lysosome fusion by Syntaxin 7 may reflect an inherent lack of specificity in cognate receptors. We do not favor this explanation because the inhibitory effect of Syntaxin 7 is specific for a restricted population of endocytic vesicles. Finally, we think that the identification of Syntaxin 7 as a lysosomal SNARE is consistent with the strong biochemical and genetic data demonstrating that the yeast homologue Vam3p is a vacuolar SNARE involved in both vacuolar traffic and homotypic fusion (Becherer et al., 1996 ; Darsow et al., 1997 ; Peterson et al., 1999 ; Ungermann et al., 1999 ). Nakamura et al. (2000) demonstrated that expression of Syntaxin 7 in yeast complemented vam3 and pep12 mutants. This observation reinforces the idea that Syntaxin 7 functions in late endocytic compartments.