ENTH and ANTH domain proteins are involved in budding of clathrin-coated vesicles. SNAREs are fusogenic proteins that function in the targeting and fusion of transport vesicles. In mammalian and yeast cells, ENTH domain proteins (epsinR and Ent3p) interact with SNAREs of the vti1 family (Vti1b or Vti1p). This interaction indicates that ENTH proteins could function in cargo sorting, which prompted us to search for additional SNAREs as potential cargo for Ent3p and epsinR. We carried out specific yeast two-hybrid assays, which identified interactions between epsinR and the mammalian late endosomal SNAREs syntaxin 7 and syntaxin 8 as well as between Ent3p and the endosomal SNAREs Pep12p and Syn8p from yeast. Lack of Ent3p affected the trafficking of Pep12p. Ent3p binding to Pep12p required the FSD late endosomal sorting signal in Pep12p. Inactivation of the sorting signal had a similar effect to removal of Ent3p on Pep12p stability indicating that Ent3p acts as a cargo adaptor for Pep12p by binding to the sorting signal. As Vti1p, Pep12p and Syn8p participate in a SNARE complex whereas Vti1b, syntaxin 7 and syntaxin 8 are mammalian SNARE partners, we propose that ENTH domain proteins at the TGN-endosome are cargo adaptors for these endosomal SNAREs.
The vesicle-mediated membrane transport is a multi-step process, consisting of vesicle formation (budding), targeting, tethering and membrane fusion (Bonifacino and Glick, 2004; Jahn and Scheller, 2006). Cargo proteins are concentrated at a specialized region on the donor membrane and packed into a nascent vesicle generated by the assembly of coat proteins such as clathrin into a cage-like lattice around the budding vesicle. Adaptor protein complexes (AP) are required to recruit cargo into coated vesicles thus playing an essential role in cargo selectivity of the transport vesicle in traffic between the trans-Golgi network (TGN) and endosomes (Owen et al., 2004). Gga proteins are monomeric clathrin adaptor proteins mediating TGN to the endosome transport (Nakayama and Wakatsuki, 2003). Apart from the AP complexes and the monomeric GGA adaptors, the list is expanding to new sets of adaptors, which are specific to only a particular type of cargo or to one family of cargo (Bonifacino and Rojas, 2006).
Recently, the epsin family proteins came into view as cargo-specific adaptors. The ENTH (epsin N-terminal homology) domains are phosphotidylinositol binding modules present in both mammalian epsins and in their yeast homologues Ent1p to Ent4p (Duncan and Payne, 2003; Legendre-Guillemin et al., 2004). ANTH (AP180 N-terminal homology) domains are highly related to ENTH domains (Ford et al., 2001) and present in mammalian AP180 (also known as SNAP91), CALM, HIP1, Hip1R and yeast AP180, Sla2p and Ent5p. These domains bind to different phosphoinositides. In addition to ANTH or ENTH domains, these proteins also contain binding motifs for clathrin, AP or GGA allowing them to participate in clathrin-mediated budding at the TGN, endosome or at the plasma membrane. Phylogenetic analysis of ENTH domains suggested two ENTH domain branches, mammalian epsins 1-3 and yeast Ent1p and Ent2p, which are involved in endocytosis at the plasma membrane, and enthoprotin (also known as Clint and epsinR) and yeast Ent3p functioning in transport between the TGN and endosomes (Legendre-Guillemin et al., 2004). EpsinR is localized to the TGN and in endosomal membranes (Kalthoff et al., 2002; Wasiak et al., 2002) and binds to PtdIns4P (Hirst et al., 2003; Mills et al., 2003). It also binds to clathrin, AP1 and GGA2 through its C-terminal domain (Duncan et al., 2003; Mills et al., 2003; Wasiak et al., 2002). Ent3p and Ent5p are partially redundant, bind Gga proteins and AP1 and promote formation of clathrin coats at the TGN-endosome (Costaguta et al., 2006; Duncan et al., 2003). ENTH domains of Ent3p and Ent5p bind PtdIns(3,5)P2 (Eugster et al., 2004; Friant et al., 2003) and PtdIns(4,5)P2 (Narayan and Lemmon, 2006). Ent5p associates with Vps27p and together with Ent3p is required for ubiquitin-dependent protein sorting into the interior of multivesicular bodies (MVB) (Eugster et al., 2004; Friant et al., 2003). This indicates that Ent3p and Ent5p have two different functions at the TGN-endosome and in MVB.
Previously, we reported that the ENTH domains of Ent3p and epsinR specifically interact with the N terminus of the SNARE proteins Vti1p or Vti1b, respectively (Chidambaram et al., 2004). Recently, similar interactions have been described for homologous proteins EPSIN1 with AtVTI11 and EpsinR2 with AtVTI12 in the plant Arabidopsis thaliana (Lee et al., 2007; Song et al., 2006). It offered a hint that ENTH domains could function as cargo adaptors. SNARE proteins are a family of membrane proteins that play an essential role in the membrane fusion machinery. SNAREs anchored on the transport vesicles and target membranes form a complex via their conserved SNARE motifs, which brings the two membranes in close proximity, followed by fusion of the lipid bilayers. There are four different subfamilies of SNARE proteins, R-, Qa-, Qb- and Qc-SNAREs, classified on the basis of on similarities in the amino acid sequences of their SNARE motifs. All well characterized SNARE complexes are composed of four different SNARE motifs one from each subfamily to form the four-helix bundles (Jahn and Scheller, 2006). So far, 24 SNARE proteins have been identified in yeast and more than 40 in mammalian cells. Each organelle contains of a defined set of SNAREs to coordinate transport among the different compartments. Some SNARE proteins can pair with more than one set of partners and thus participate in the formation of several different SNARE complexes. Vti1p is a yeast Qb-SNARE, which is involved in a number of transport steps as part of four different SNARE complexes (Brickner et al., 2001; Fischer von Mollard et al., 1997; Fischer von Mollard and Stevens, 1999). The SNARE complex formed by Vti1p with Pep12p (Qa), Syn8p (Qc) and Ykt6p (R) is involved in the TGN to late endosome transport (Dilcher et al., 2001; Kweon et al., 2003; Lewis and Pelham, 2002). The endosomal Qa-SNARE Pep12p controls all known membrane fusion events at the late endosomes (Blanchette et al., 2004; Gerrard et al., 2000b). Vti1p with Vam3p (Qa), Vam7p (Qc) and Nyv1p (R) or Ykt6p (R) mediates fusion with the vacuole (Ungermann et al., 1999) and Vti1p with Tlg2p (Qa), Tlg1p (Qc) and Snc1/2p (R) functions in homotypic TGN fusion (Brickner et al., 2001). Vti1p has two mammalian homologues, Vti1a and Vti1b which are related but specialized to different transport steps and also localized in different subcellular compartments (Antonin et al., 2000a; Kreykenbohm et al., 2002; Mallard et al., 2002). Both proteins have different SNARE partners. Vti1b forms complexes with syntaxin 7 (Qa), syntaxin 8 (Qc) and VAMP7 (R) or VAMP8 (R) (Pryor et al., 2004). Vti1a is found in complexes with syntaxin 13 (Qa) or syntaxin 16 (Qa), syntaxin 6 (Qc) and VAMP4 (R) (Brandhorst et al., 2006).
Proper sorting and transport of SNARE proteins are necessary to maintain intracellular traffic. Sorting can be controlled by the transmembrane domain or by sorting signals either in the SNARE motif or the N-terminal domain. Some SNARE proteins are known to utilize adaptor proteins for sorting at the TGN. Mammalian VAMP4 has a di-leucine motif which interacts with AP1 at the TGN (Peden et al., 2001) and yeast Vam3p binds AP-3 in the transport from the TGN to the vacuole (Darsow et al., 1998). Gga proteins are required for the sorting of the pre-vacuolar syntaxin Pep12p (Black and Pelham, 2000). EpsinR has been shown to be a cargo adaptor for Vti1b because depletion of epsinR reduces the incorporation of Vti1b into clathrin coated vesicles (Hirst et al., 2004).
We have investigated whether Ent3p or epsinR are able to bind and to serve as cargo adaptors for additional SNARE proteins.
The ENTH domains of epsinR and Ent3p interact with additional endosomal SNAREs
Previously, we reported the binding of the N-terminal domain of mammalian Vti1b to the ENTH domain of epsinR and of yeast Vti1p to Ent3p (Chidambaram et al., 2004). We set out to determine whether additional TGN, endosomal and lysosomal SNAREs are able to interact with epsinR or Ent3p, respectively. Two-hybrid assays were done with the N-terminal domains of SNAREs fused to the DNA binding domain of LexA and ENTH domains of epsinR, Ent3p and Ent5p fused to the VP16 activation domain. Among the mammalian proteins, a weak interaction of epsinR was observed with the Qa-SNARE syntaxin 7 and the Qc-SNARE syntaxin 8 (Fig. 1A) where Vti1b was used as a positive control. Syntaxin 7 and syntaxin 8 are complex partners of Vti1b. By contrast, syntaxin 6 (Qc), syntaxin 13 (Qa) and syntaxin 16 (Qa) did not interact with epsinR (data not shown). These SNAREs form complexes with Vti1a, which failed to interact with epsinR (Chidambaram et al., 2004). Yeast Ent3p showed a strong interaction with the N-terminal domains of Pep12p (Qa) and Syn8p (Qc). The SNAREs Ykt6p (R), Tlg1p (Qc), Vam7p (Qc) and Vam3p (Qa) failed to interact (Fig. 1B and data not shown). Pep12p and Syn8p are part of the same SNARE complex and function together with Vti1p and Ykt6p in transport to the late endosome. Vam3p, Vam7p and Vti1p are the Q-SNAREs involved in transport to the vacuole whereas Tlg1p, Tlg2p and Vti1p are required for fusion with the TGN. The interaction of Ent3p with Pep12p and Syn8p was very specific because neither the mammalian homologue epsinR nor the ANTH domain protein Ent5p interacted.
These interactions were investigated by in vitro pull down assays. Bacterially expressed and purified ENTH domain of Ent3p or epsinR with a Strep-tag were incubated with His6-tagged cytosolic domains of SNARE proteins. Ent3p was bound to Pep12p very much like Vti1p that was used as positive control (Fig. 1C). However, no in vitro binding could be observed with His6-Syn8p, which may be due to low affinity of the interaction or to folding problems. As expected, Tlg1p did not pull down Ent3p. Small amounts of epsinR were bound to syntaxin 7 and syntaxin 8 (Fig. 1D).
We tested for genetic interactions with SYN8 to investigate the functional significance of the two-hybrid interaction identified. A synthetic growth defect was observed at 37°C, when both ENT3 and ENT5 were deleted in syn8Δ mutant backgrounds (Fig. 2A). Deletion of only ENT3 or ENT5 did not show any effect. The transport of carboxypeptidase Y (CPY) from the ER to the vacuole was checked in mutant cells by pulse-chase immunoprecipitation to analyze trafficking. CPY exists in three forms during its transport. p1CPY is present in the ER (67 kDa), which is modified to p2CPY in the Golgi (69 kDa). p2CPY is transported from the TGN to the late endosome by the CPY receptor Vps10p and then to the vacuole where it is matured to mCPY (61 kDa). If there is a block in the transport from the TGN to the vacuole, less mCPY will be found in the cell (fraction I) and p2CPY will be secreted by the cells (fraction E). syn8Δ ent3Δ ent5Δ cells secreted more p2CPY compared to syn8Δ as well as ent3Δ ent5Δ cells (Fig. 2B, fraction E). The absence of Syn8p increased secretion of p2CPY in ent3Δ ent5Δ cells from 17.7% (±2.7%, n=2) to 54.5% (±4.1%) of total CPY (Fig. 2C). In syn8Δ cells, 39.0% (±2.2%) of the total CPY was secreted. Deletion of Ent3p or Ent5p alone in syn8Δ cells did not have an effect on CPY transport demonstrating a genetic interaction between ENT3 and ENT5 with SYN8 in this transport pathway.
The absence of Ent proteins affected trafficking of SNAREs
Next we wanted to test whether Ent3p or Ent5p serve as cargo adaptors for Pep12p. Pep12p is found in late endosomes. This localization is maintained by cycling between Golgi and late endosomes (Black and Pelham, 2000; Hettema et al., 2003). Pep12p can be mislocalized to early endosomes or degraded in the vacuole. Altered trafficking of Pep12p may therefore change the half-life of this protein. We examined the stability of Pep12p in ent3Δ, ent5Δ and ent3Δ ent5Δ cells by pulse-chase labeling followed by immunoprecipitation of samples after different chase periods. Interestingly, Pep12p was more stable in ent3Δ cells than in wild-type cells (Fig. 3A). After 3 hours of chase, 30% (s.d. ±5.3, n=3) of Pep12p synthesized during the pulse period and a 10 minutes chase was still present in the wild-type cells, whereas 52% (±10.2, n=3) was found in ent3Δ cells and 41% (±10.6) of the labeled Pep12p was still present in ent3Δ ent5Δ cells. The same pattern of Pep12p stability was witnessed after a 5-hour chase period (Pep12p in wild-type 14% ±0.8%, n=2, ent3Δ 31% ±2.8%, ent3Δ ent5Δ 22% ±3.8%). The absence of Ent5p did not lead to stabilization of Pep12p. Pep12p was more stable in ent3Δ, indicating that trafficking of Pep12p was affected in these cells. The stability of Pep12p was intermediate in ent3Δ ent5Δ cells. This prompted us to investigate the subcellular distribution of Pep12p by immunofluorescence microscopy and subcellular fractionation (see below).
ENTH proteins have been implicated in retrograde transport from the endosome to the TGN. EpsinR is involved in the retrograde trafficking of internalized shiga toxin to the TGN (Saint-Pol et al., 2004). ent3Δ ent5Δ cells secrete α-factor precursor instead of mature α-factor (Duncan et al., 2003), which indicates that the processing protease Kex2p is not present in the late-Golgi. Since Kex2p cycles between the TGN and the endosomes, retrograde traffic of Kex2p must be blocked in ent3Δ ent5Δ cells. The role of Ent3p and Ent5p in the retrograde transport of other cargo proteins was studied by analyzing the stability of A-ALP [the cytosolic domain of dipeptidyl aminopeptidase A (DPAP A), which localizes to the TGN, fused to the transmembrane and luminal domain of alkaline phosphatase (ALP)] and Vps10p, which recycle between the TGN and the endosomes because of their retrieval signals (Cereghino et al., 1995; Nothwehr et al., 1993). Vps10p is a receptor that sorts several different vacuolar proteins by cycling between a late Golgi compartment and the late endosome. If retrograde transport of these proteins from the late endosome to the TGN is blocked, transport continues to the vacuole, resulting in enhanced proteolytic degradation. Pulse-chase immunoprecipitations were done after different chase periods using wild-type, ent3Δ, ent5Δ and ent3Δ ent5Δ cells (Fig. 4A,B). The stability of both A-ALP and Vps10p was not strongly reduced in mutant cells. This indicated that neither Ent3p nor Ent5p are involved in retrograde transport of A-ALP and Vps10p to the TGN.
Colocalization of two proteins indicates that a permanent interaction between two proteins is possible. Endogenous Ent3p and Pep12p were detected in wild-type cells by double immunofluorescence microscopy (Fig. 5A-C). Ent3p antiserum revealed punctate structures as well as a haze probably due to cytoplasmic Ent3p. Pep12p was also found in several bright dots. However, hardly any colocalization was observed between Pep12p and Ent3p. This is consistent with a transient interaction of Ent3p and Pep12p during cargo sorting. Since Pep12p was stabilized in the absence of Ent3p, we studied the consequence of a lack of Ent3p on Pep12p localization. Pep12p was detected by immunofluorescence as scattered bright punctate structures in wild-type cells (Fig. 5D) as well as in ent3Δ cells (data not shown). By contrast, in ent3Δ ent5Δ cells, the number and also the brightness of the punctate structures were reduced and staining appeared more hazy (Fig. 5E). We investigated the effects of SYN8 deletion on the distribution of Pep12p. The localization of Pep12p in syn8Δ and syn8Δ ent3Δ cells was not affected (data not shown). However, the punctate staining of Pep12p was almost lost and a diffuse hazy pattern was observed in syn8Δ ent3Δ ent5Δ cells (Fig. 5F). The total amount of Pep12p was increased in syn8Δ ent3Δ ent5Δ and ent3Δ ent5Δ cells compared to wild-type cells as observed by western blotting (data not shown). To determine whether endosomal structure was affected in syn8Δ ent3Δ ent5Δ and ent3Δ ent5Δ cells, these cells were allowed to endocytose the membrane marker FM4-64 for 5 minutes. FM4-64 reached endosomes after this short incubation period (Fig. 5G-I). Subsequently, FM4-64 was transported to vacuoles. FM4-64 staining did not differ between wild-type and mutant cells demonstrating that endosomes were intact whereas Pep12p was redistributed.
To analyze subcellular localization with an independent and more quantitative method, we performed subcellular fractionation by sucrose density gradient centrifugation (Fig. 6). As described before (Becherer et al., 1996; Black and Pelham, 2000), most Pep12p was detected in fractions 4-8 in wild-type cells, which is typical for a late endosomal localization (Fig. 6A). In syn8Δ ent3Δ ent5Δ cells Pep12p migrated into the denser fractions, 9-11, which have been reported to contain early endosomes (Black and Pelham, 2000; Holthuis et al., 1998). In addition, Pep12p was absent from the lighter fractions, 4-8. An increase of the amount of Pep12p in fractions 9-11 was also observed in ent3Δ ent5Δ cells. Thus, a redistribution of Pep12p was observed in syn8Δ ent3Δ ent5Δ and ent3Δ ent5Δ cells confirming the results obtained by immunofluorescence microscopy. Ent3p together with Ent5p are involved in Pep12p sorting. Syn8p contributes to localization of Pep12p in the absence of Ent3p and Ent5p. Vti1p had a broad distribution on sucrose density gradient in wild-type cells (Fig. 6B) because of its presence and function in multiple organelles. In ent3Δ ent5Δ as well as in syn8Δ ent3Δ ent5Δ cells, Vti1p shifted from lighter to denser fractions. This suggests that Ent3p and Ent5p are required for correct subcellular localization of Vti1p. Absence of Syn8p did not have an additive effect on Vti1p localization in ent3Δ ent5Δ cells in contrast to Pep12p localization. This indicates that trafficking of Vti1p is independent of Syn8p. Vacuoles were found in fractions 3-5 and ER in fractions 8-11 as indicated by the marker proteins: the 60 kDa subunit of the vacuolar ATPase (Fig. 6C) and Use1p (Fig. 6D) (Dilcher et al., 2003; Jackson and Stevens, 1997).
FSD sorting signal in Pep12p was required for binding of Ent3p
A similar redistribution of Pep12p to early endosomes was observed in gga1Δ gga2Δ cells by sucrose density gradient centrifugation (Black and Pelham, 2000). Mutations in the FSD sorting signal including amino acid residues 20-26 at the N terminus of Pep12p also resulted in mislocalization of mutant Pep12p to early endosomal fractions. This demonstrates that correct sorting of Pep12p to late endosomes requires a FSD sorting signal in addition to Gga1 or Gga2p. However, no direct binding could be observed between Gga1/2p and Pep12p (Black and Pelham, 2000). As Ent3p binds Gga1/Gga2p we investigated whether Ent3p acts as a link between Gga1/Gga2p and Pep12p. In this case, Ent3p interaction with Pep12p should be depended on an intact FSD sorting signal. A deletion of amino acid residues 19-26 (ΔFSD) or a substitution of phenylalanine residue 20 with leucine (F20L) was introduced into the two-hybrid construct with the N-terminal domain of Pep12p. The two-hybrid interaction with Ent3p was almost abolished upon deletion of the FSD motif as well as by the point mutation F20L as indicated by growth on selective plates (Fig. 7A) and quantification of β-galactosidase from liquid culture (data not shown). As an independent assay, His6-Pep12 with the point mutation F20L or a deletion of the FSD motif were expressed and purified. Pep12 with mutations in the FSD motif bound less Ent3-Strep than wild-type Pep12 in an in vitro binding assay (Fig. 7B). These data indicate that Ent3p requires the FSD sorting signal for binding to Pep12p. Full-length Pep12p with the point mutation F20L was expressed under its own promoter in yeast in the absence of wild-type Pep12p to analyze the consequences of the sorting motif mutation in vivo. The stability of Pep12p F20L in wild-type cells was compared to that of wild-type Pep12p in wild-type cells and in ent3Δ cells by pulse-chase immunoprecipitation (Fig. 8). After 3 and 5 hours of chase, 16% (s.d. ±1.9, n=3) and 9% (s.d. ±2.9), respectively, of Pep12p remained. Pep12p F20L was more stable, with 28% (s.d. ±2.7) remaining after 3 hours and 26% (s.d. ±3.1) after 5 hours. Pep12p F20L stability was comparable to that of wild-type Pep12p in ent3Δ cells [37% (s.d. ±12.5) after 3 hours and 27% (s.d. ±6.2) after 5 hours]. These data indicate that inactivation of the FSD sorting signal had the same effect on Pep12p trafficking as removal of Ent3p, providing in vivo evidence for interaction of Pep12p via the FSD sorting signal with Ent3p.
Despite the fact that significant progress has been made in studying the roles of SNARE proteins, less is know about how cells localize and sort SNAREs to establish proper steady state distribution of each SNARE in its correct membrane compartment. Here we describe the physical and functional interaction between endosomal SNAREs and ENTH domain proteins and show that Ent3p acts as a cargo adaptor for Pep12p by binding to the late endosomal sorting signal of Pep12p.
Interactions of Ent3p and epsinR with SNAREs
We have demonstrated by two-hybrid analysis that Ent3p interacts with the N-terminal domains of Pep12p and Syn8p in addition to an earlier study identifying the interaction of Ent3p with Vti1p (Chidambaram et al., 2004). These three SNAREs belong to different subfamilies of SNAREs, Qa, Qc and Qb, respectively. In vitro pull-down assays confirmed binding between Ent3p with Pep12p. Recently, this interaction was also seen in vivo as Pep12p was found in a complex with Ent3p-TAP (Copic et al., 2007). We were not able to show binding between Syn8p and Ent3p by in vitro pull down. This may be due to the character of the interaction because the binding between adaptor protein and cargo is very dynamic and is of low affinity to allow for dissociation after sorting. The functional significance of the two-hybrid interactions detected between Ent3p and Syn8p was supported by synthetic defects in CPY transport and synthetic growth defects in syn8Δ ent3Δ ent5Δ cells. Since Ent3p and Ent5p are partially redundant proteins (Duncan et al., 2003) and Syn8p can be replaced by Tlg1p (Lewis and Pelham, 2002), ent3Δ ent5Δ double deletion was required to observe a genetic interaction with syn8Δ, though only Ent3p interacted with Syn8p in the two-hybrid assay.
Similar interactions between endosomal ENTH domain proteins and SNAREs were detected in the mammalian system: syntaxin 7 (Qa) and syntaxin 8 (Qc) showed weak interactions with epsinR, which also binds Vti1b (Qb) (Chidambaram et al., 2004). The N-terminal domains of these SNAREs do not have homologous amino acid sequences or share common sequence motifs but their structures are most likely similar. The N-terminal domains of syntaxin 7 and Vti1b form a bundle of three α-helices (Antonin et al., 2002). The N-terminal domains of syntaxin 8, Pep12p, Syn8p and Vti1p are assumed to adopt comparable three helix bundles as indicated by secondary structure prediction programs and a high α-helical content of Pep12p and Vti1p (Tishgarten et al., 1999). Therefore, different sequence motifs or the three dimensional structure of the folded N-terminal domains may be important for recognition of different SNAREs by Ent3p or epsinR, respectively. However, these interactions are specific, because the Qa SNARE Vam3p does not interact with Ent3p even though it forms a N-terminal three helix bundle (Dulubova et al., 2001). In addition, the N-terminal domains of Tlg1p, syntaxin 6, syntaxin 13 and syntaxin16 are predicted to form these structures but do not bind Ent3p or epsinR, respectively. In the two-hybrid assay, Ent3p did not interact with Ykt6p (R-SNARE), which is present in the same endosomal SNARE complex as Syn8p, Pep12p and Vti1p. The structure of the N-terminal domain of Ykt6p is very distinct from that of other partners in the endosomal SNARE complex. Ykt6p has a mixed α-helical-β-sheet profilin-like fold (Tochio et al., 2001).
Interestingly, the Ent3p interaction partners Pep12p, Syn8p and Vti1p are part of the same SNARE complex (Lewis and Pelham, 2002). The absence of interaction between Tlg1p and Ent3p suggests that Ent3p is not involved in sorting of SNAREs required for retrograde transport to the TGN. In addition, Ent3p did not interact with Vam3p and Vam7p, which are involved in several fusion steps with the vacuole. The epsinR interaction partners Vti1b, syntaxin 7 and syntaxin 8 also participate in the same SNARE complexes (Antonin et al., 2000a). This means that these ENTH domain proteins interact with all Q-SNAREs of a single SNARE complex but not with SNAREs of other complexes (Fig. 9).
It is assumed that the N-terminal domains of SNAREs do not interact with each other in SNARE complexes suggesting that all ENTH domain binding sites of these SNAREs are accessible in SNARE complexes. Ent3p and epsinR assemble into large coat structures because they have several binding sites for Gga proteins, AP1 and clathrin. Therefore, we speculate that Ent3p and epsinR coat structures may have a higher avidity for these endosomal SNARE complexes or for possible subcomplexes containing two or three Q-SNAREs than for single SNAREs. Ent3p and epsinR may contribute to trafficking of SNARE complexes before their dissociation by NSF and α-SNAP. It is not known whether endosomal Q-SNAREs form acceptor complexes which function as binding sites for R-SNAREs as observed in synaptic vesicle fusion (Jahn and Scheller, 2006). It is also unknown which of the SNAREs function on transport vesicles and which on the late endosomes. The R-SNARE Ykt6p on its own is probably unable to cause fusion because Ykt6p does not have a transmembrane domain, and in vitro data indicate that fusion requires transmembrane SNAREs on both membranes. Therefore two or three endosomal SNAREs may form fusion-competent subcomplexes, which may have to be sorted into transport vesicles. Our two-hybrid and in vitro pull-down data indicate that epsinR and Ent3p are able to bind free SNAREs. Owing to low affinities it would be very difficult to determine whether ENTH domains bind to SNARE complexes and the exact binding sites.
Ent3p function as cargo adaptor for Pep12p requires the FSD sorting signal
In mammalian cells epsinR acts as a cargo adaptor for Vti1b (Hirst et al., 2004). The depletion of epsinR caused a redistribution of Vti1b from the Golgi region to the cell periphery. Less Vti1b was recruited into the clathrin-coated vesicles isolated from epsinR-depleted cells.
Therefore, we used different experimental approaches to determine whether Ent3p functions as a cargo adaptor for Pep12p and Vti1p. We showed that Pep12p was stabilized in ent3Δ cells, indicating that trafficking of Pep12p is affected. This resembles protein stabilization in cells defective in vesicular fusion with endosomes. For example, a mutant CPY receptor Vps10p, lacking the TGN retrieval signal is stabilized in pep12Δ, vps45 and vps21 mutant cells because vacuolar transport is prevented by a block in the fusion of TGN-derived vesicles with late endosomes (Bryant and Stevens, 1998; Deloche et al., 2001; Gerrard et al., 2000a). Therefore, the increased stability of Pep12p in ent3Δ cells may indicate a block in the forward transport of Pep12p to late endosomes.
The redistribution of Pep12p observed by immunofluorescence and sucrose density gradient fractionation in ent3Δ ent5Δ and syn8Δ ent3Δ ent5Δ cells supports the hypothesis that Ent3p is a cargo adaptor for Pep12p. Our data indicate that Ent5p and Syn8p also contribute to Pep12p localization. The more dramatic redistribution of Pep12p in ent3Δ ent5Δ cells is probably the reason for lower stability of Pep12p in ent3Δ ent5Δ cells compared to ent3Δ cells. Similar effects on Pep12p localization have been reported recently. Distribution of Pep12p on a sucrose density gradient was slightly affected in ent3Δ cells and more severely affected in ent3Δ ent5Δ cells (Copic et al., 2007). Pep12p was shifted to the position of early endosomes in sucrose density gradients. A similar redistribution of Pep12p to early endosomes in sucrose density gradients was observed in gga1Δ gga2Δ cells as well as upon mutation of the sorting signal FSDSPEF (FSD; amino acid residues 20-26) in the N terminus of Pep12p (Black and Pelham, 2000). However, no direct binding could be observed between Gga1/Gga2p and Pep12p. Absence of the FSD motif in Pep12p, absence of Gga1p/Gga2p as well as absence of Ent3p and Ent5p all resulted in similar mislocalization of Pep12p. Therefore, we tested whether the FSD motif is involved in binding of Pep12p to Ent3p. Deletion of the FSD motif, as well as exchanging the critical phenylalanine for a leucine residue (F20L), almost abolished the two-hybrid interaction between Pep12p and Ent3p and reduced binding in the in vitro pull-down assay, indicating that the FSD motif is required for binding of Pep12p to Ent3p. Pep12 F20L was stabilized to a similar degree as Pep12p in the absence of Ent3p indicating that both manipulations had the same consequences on trafficking. Therefore, binding, as well as functional data, indicate that Ent3p provides a link between Gga1/2p and Pep12p. The same conclusion was reached by an independent approach: Gga2p-TAP bound less Pep12p in ent3Δ cells than in wild-type cells (Copic et al., 2007). We showed that Vti1p was also redistributed in ent3Δ ent5Δ cells suggesting that Ent3p together with Ent5p functions in sorting of Vti1p as well. The N-terminal domains of Vti1p and Syn8p do not contain similar FSD motifs. Therefore it is likely that Ent3p recognizes different sequences or a structural motif in these SNAREs.
Ent3p is a selective cargo adaptor
Almost all CPY reached the vacuole in ent3Δ cells and only a tiny amount of p2CPY was secreted (Chidambaram et al., 2004). Only 20% of total CPY was secreted in ent3Δ ent5Δ cells suggesting that the CPY receptor Vps10p must have been sorted correctly into vesicles budding from the TGN to allow for efficient CPY transport to the vacuole. Therefore, TGN to endosome transport is intact in the absence of Ent3p. Cargo selectivity was also observed for retrograde traffic from endosomes to the TGN. Retrograde traffic of the processing protease Kex2p from endosomes to the TGN must be blocked in ent3Δ ent5Δ cells since they secrete α-factor precursor (Duncan et al., 2003). However, we observed that Vps10p and A-ALP, which recycle between TGN and endosomes, were not destabilized in ent3Δ and ent3Δ ent5Δ cells, indicating that they were retrieved correctly to the TGN and not mislocalized to the vacuole. Ent5p also functions as a selective cargo adaptor. The chitin synthase Chs3p, which travels between TGN, plasma membrane and early endosomes, binds to Ent5p-TAP, but not Ent3p-TAP (Copic et al., 2007). Chs3p localization was affected in ent3Δ ent5Δ cells, but not in the single mutant cells indicating that Ent3p and Ent5p have redundant functions in Chs3p trafficking (Copic et al., 2007). Additional adaptors with restricted cargo selectivity have been identified for these traffic steps in yeast. One example is the sorting nexin Grd19p (also known as Snx3p) required for retrieval of Kex2p and A-ALP, but not Vps10p from endosomes to the TGN (Voos and Stevens, 1998). An ANTH domain protein may act as a cargo receptor for a SNARE in endocytosis. Genetic disruption of the ANTH domain protein UNC-11, the C. elegans AP180 orthologue, leads to reduced endocytosis of synaptobrevin from the presynaptic plasma membrane (Dittman and Kaplan, 2006; Nonet et al., 1999). However, it is unclear, whether there is direct binding of UNC-11 to synaptobrevin.
Ent3p functions with Gga proteins in TGN to endosome transport (Costaguta et al., 2006; Duncan et al., 2003). The interaction partners of Ent3p, Pep12p, Syn8p and Vti1p are part of a SNARE complex required for TGN to late endosome transport, indicating that Ent3p functions as a cargo adaptor for SNAREs active in this transport step (Fig. 9). In mammalian cells, epsinR is required for retrograde transport from the early endosome to the TGN (Saint-Pol et al., 2004). EpsinR does not bind to SNAREs involved in this transport step, Vti1a, syntaxin 6 and syntaxin 16 (Mallard et al., 2002). By contrast, the epsinR binding partners Vti1b, syntaxin 7 and syntaxin 8 are involved in late endosome fusion and transport from late endosomes to lysosomes (Antonin et al., 2000a; Pryor et al., 2004). As epsinR functions as a cargo adaptor for Vti1b (Hirst et al., 2004) epsinR may be involved in transporting these SNAREs in an inactive form to the correct compartment if they are missorted. Alternatively, epsinR may function in anterograde transport from the TGN to late endosomes as well as in retrograde transport. In this case epsinR would act as a cargo adaptor for active SNAREs paralleling the role of Ent3p in yeast.
The physical and genetic analyses of ENTH domain proteins and SNARE interactions lead us to propose that Ent3p and epsinR are cargo adaptors for specific endosomal SNAREs at the TGN and endosome.
Thus, several lines of evidences indicate that incorporation of SNARE proteins into the nascent vesicle is regulated by components of the vesicle budding machinery. This could be used as a quality control mechanism to ensure that the vesicle is competent for fusion with the target organelle.
Materials and Methods
Reagents were obtained from the following sources: enzymes for DNA manipulation from New England Biolabs (Beverly, MA), [35S]methionine from Amersham Pharmacia (Braunschweig, Germany), fixed Staphylacoccus aureus cells (Pansorbin) from Calbiochem (San Diego, CA), zymolyase from Seikagaku (Tokyo, Japan). The Strep-tag system was obtained from IBA (Göttingen, Germany). All other reagents were purchased from Sigma (St Louis, MO).
Plasmid manipulations were performed in the E. coli strains XL1Blue or BL21(DE3) using standard media.
Yeast strains (Table 1) were grown in rich medium (1% yeast extract, 1% peptone, 2% dextrose, YEPD) or standard minimal medium (SD) with appropriate supplements.
Strains and plasmids
Yeast strains were constructed using standard genetic techniques. Single deletion strains in BY4742 background were obtained from the Euroscarf strain collection (Winzeler et al., 1999). Precise deletions were generated by PCR amplification of a marker gene using oligonucleotides also homologous to flanking regions of the target gene and genomic integration. Alternatively, a mutant locus was PCR amplified from genomic DNA with about 200 bp of flanking regions and integrated. SSY1, SSY2 and SSY3 strains were generated by PCR amplification and integration of pep12Δ::URA3 from RPY38 (Robert C Piper, University of Iowa, USA) into SCY13, SCY25 and SCY26. Plasmids used in this study are described in Table 2.
Yeast two-hybrid assays
The yeast two-hybrid screen was performed as described previously (Vojtek and Hollenberg, 1995) in L40 cells using pLexN with the N-terminal domain of SNAREs such as yeast Vti1p (Chidambaram et al., 2004), Syn8p (AA 1-169), Pep12p (AA 1-200) and Tlg1p (AA 1-137) as bait vectors, and the ENTH domain of Ent3p or EpsinR (AA 1-162) in pVP16-3 (Chidambaram et al., 2004) as prey vector. Yeast cells containing pairs of LexA DNA-binding domains and VP16 activation domain fusions were streaked out on plates with minimal medium lacking uracil, lysine, tryptophan, leucine and histidine (mammalian SNAREs), which was complemented with 2.5 mM 3-aminotriazole (yeast SNAREs) to study specific two-hybrid interactions. pMB285 and pMB286 (Black and Pelham, 2000) were used as templates for PCR amplification to generate pLexN-Pep12p (AA 1-200) with F20L point mutation or a deletion of amino acid residues 19-26. Liquid cultures were assayed for β-galactosidase using ONPG (ortho-nitrophenyl-β-D-galactopyranoside) as substrate, as described previously (Schneider et al., 1996).
In vitro binding assays
ENTH domain of Ent3p and epsinR with a C-terminal Strep-tag was expressed in E. coli using the plasmid pASK-IBA3 and purified using Strep-Tactin Sepharose (IBA, Göttingen, Germany). The cytosolic domains of SNAREs were expressed with a N-terminal His6 tag in E. coli and purified using Ni-NTA agarose (Qiagen, Hilden, Germany). The purified recombinant proteins were analyzed by SDS-PAGE and Coomassie Blue staining for their purity. The in vitro binding assay was done as described previously (Chidambaram et al., 2004). Briefly, purified fusion proteins consisting of His6-tagged SNAREs (1.5 μM) and Strep-tagged ENTH domain of Ent3p (1 μM) were incubated in PBS 1% Triton X-100, 10 mM imidazole for 30 minutes at 4°C, Ni-NTA beads (preincubated with 1% bovine serum albumin in PBS) added and the incubation continued for another 30 minutes. The following conditions were used for syntaxin 7 and syntaxin 8 pull downs: 2 μM SNAREs, 1 μM epsinR in 20 mM NaPO4 pH 7.4, 400 mM NaCl, 1% Triton X-100, 20 mM imidazole. Washed pellets were separated by SDS-PAGE, transferred to nitrocellulose and Ent3p-Strep-tag or epsinR-Strep detected using an antiserum directed against Ent3p or Strep-Tactin HRP and ECL.
Generation of antisera against Pep12p and Ent3p
Anti-Pep12p rabbit serum was raised against GST-Pep12p (cytosolic domain, residues 1-263). The fusion protein was expressed in E. coli and purified using glutathione-agarose. Anti-Pep12p serum recognized a protein with the expected molecular mass for Pep12p in a wild-type but not in pep12Δ cell extracts. A non-specific higher molecular mass band was also recognized along with Pep12p. Anti-Ent3p rabbit serum was raised against Ent3p-Strep-tag (ENTH domain, residues 1-172). The fusion protein was expressed in E. coli and purified using Strep-Tactin Sepharose (IBA, Göttingen, Germany). Anti-Ent3p serum recognized a protein with the molecular mass of 50 kDa in wild-type but not in ent3Δ cell extracts.
Yeast cells were grown in log-phase at 24°C and 0.5 OD preincubated at the indicated temperature for 15 minutes. Cells were labeled with [35S]methionine (100 μCi/0.5 OD) for 10 minutes, chased for 30 minutes and CPY was immunoprecipitated from cellular extracts (I) and medium (E). For Pep12p, Vps10p or A-ALP stability assays, the cells were labeled for 25 minutes, chased for the indicated times and immunoprecipitated from cellular extracts (I) as described previously (Fischer von Mollard and Stevens, 1999; Nothwehr et al., 1993; Vater et al., 1992). CPY, ALP, Vps10p and Vti1p antisera were kindly provided by T. H. Stevens (University of Oregon, USA). Immunoprecipitates were analyzed by SDS-PAGE and autoradiography. A BAS-1800 II (Fuji) was used for quantification.
Indirect immunofluorescence was performed with wild-type syn8Δ, ent3Δ, syn8Δ ent3Δ, ent3Δ ent5Δ and syn8Δ ent3Δ ent5Δ cells as described previously (Raymond et al., 1992) using a monoclonal antibody against Pep12p (Molecular probes). Primary antibodies were preabsorbed with yeast cells lacking Pep12p or Ent3p, respectively. Cells were viewed with a Leica DM5000 B fluorescence microscope equipped with a Leica DFC350 FX CCD camera.
Subcellular fractionation was performed by sucrose density gradient centrifugation. Wild-type, syn8Δ, ent3Δ ent5Δ and syn8Δ ent3Δ ent5Δ cells were spheroplasted, osmotically lysed and centrifuged at 500 g to remove debris (homogenate H). The homogenates were separated by sucrose density gradient centrifugation (Becherer et al., 1996). The gradient consisted of the following steps: 0.5 ml 50%, 1 ml 42%, 1 ml 37%, 1.5 ml 34%, 2 ml 32%, 1.5 ml 29%, 1 ml 27%, 1 ml 22% (w/w) sucrose in 10 mM Hepes-NaOH pH 7.6, with centrifugation for 16 hours at 135,000 gmax (27,000 r.p.m. in a AH627 rotor). Fractions were separated by SDS-PAGE and immunoblotted using HRP-conjugated secondary antibodies and ECL.
Endosomal staining by FM4-64
Cells were grown at 30°C and according to published procedures (Vida and Emr, 1995), 65 mM FM4-64 (Molecular Probes, Eugene, OR) was added and incubated for 5 minutes. Cells were washed once and viewed immediately under a fluorescence microscope.
We thank Beate Veith, Claudia Prange and Christiane Wiegand for excellent technical assistance. Sabine Kossmann and Torsten Wundenberg are acknowledged for the construction of plasmids and purification of recombinant proteins. We thank T. H. Stevens (Eugene, OR, USA) H. R. B. Pelham (MRC Cambridge, UK) and S. F. Nothwehr (Columbia, MO, USA) for the gift of plasmids. This work was supported by grant GRK 521 from the DFG to G.F.v.M.
↵* Present address: Nephrology Division, Department of Medicine, Harvard Medical School and Massachusetts General Hospital, Boston, MA 02129, USA
- Accepted November 12, 2007.
- © The Company of Biologists Limited 2008