Concerted Auto-regulation in Yeast Endosomal t-SNAREs*

In yeast, the assembly of the target (t)-SNAREs [Tlg2p/Tlg1p,Vti1p] and [Pep12p/Tlg1p,Vti1p] with the vesicular (v)-SNARE Snc2p promotes endocytic fusion. Here, selected mutations and truncations of SNARE proteins were tested in an in vitro fusion assay to identify potential regulatory regions in these proteins, and two distinct regions were found. The first is represented by the combined effect of the three t-SNARE N-terminal regions and the second is located within the Tlg1p SNARE motif. These internal controls provide a potential mechanism to enable SNARE-dependent fusion to be regulated.

terminal domain, structured in a three-helix bundle, is able to interact with the SNARE motif, generating the so-called closed conformation. This interaction blocks the binding of the light chains and inhibits the t-SNARE complex formation (11)(12)(13)(14)(15). The removal of the Syntaxin1 N-terminal domain results in an increase of fusion (16).
The Syntaxin Vam3p on the other hand, uses a different mechanism. In the yeast vacuole, the three-helix bundle of Vam3p-N-terminal domain does not interact intramolecularly with the SNARE motif (17). Nonetheless, the removal of this N-terminal domain influences the SNARE complex assembly (18). Despite the fact that the mechanisms differ (closed or opened Syntaxins), all heavy chain N-terminal domains tested act by reducing t-SNARE assembly (11)(12)(13)(14)(15)18).
The light chains also have N-terminal domains, but their role is less clear. Neither appears to adopt a closed intra-molecular conformation nor does their N-terminal domain seem to influence the rate of the t-SNARE complex formation (15,19,20).
Previously, we characterized two yeast endocytic complexes, the early endosome/trans-Golgi network complex [Tlg2p/Tlg1p,Vti1p] and the late endocytic complex [Pep12p/ Tlg1p,Vti1p], both fusing with Snc2p v-SNARE although very slowly in vitro (4,5). In the present study, using these SNARE complexes, we determined the influence of their Nterminal domains on fusion. Furthermore, we investigated whether the SNARE motif itself influences the kinetics of liposome fusion.
SNARE Reconstitution into Liposomes-All t-SNAREs were reconstituted from individual proteins as described (9). The typical lipid recovery efficiency in the recovered Nycodenz fraction was about 50% for acceptor liposomes and about 30% for donor liposomes.
Fusion Assay-The lipid mixing assay was conducted as described (9,16). For some experiments, 3.0 nmol of Snc2-C-pept (4) or VAMP8-Cpept, or 6 nmol of cytosolic domain of Snc2p or cyt-VAMP8, were added as indicated in figure legends. The data were converted to rounds of fusion as described (16). Note that the small decrease observed during the first 10 min of each fusion reaction is due to the temperature equilibration of the fluorophore.

RESULTS
The N-terminal Domains of the t-SNARE Constitute a Potential Regulatory Switch-Both yeast endocytic t-SNAREs, [Tlg2p/Tlg1p,Vti1p] and [Pep12p/Tlg1p,Vti1p], must be activated to promote fusion of liposomes (4,5). This distinguishes them from all other fusogenic SNAREs tested. In vitro, this activation is provided artificially by a peptide (Snc2-C-pept) corresponding to the C-terminal part of the v-SNARE helical motif (4). It has been demonstrated that such peptides are able to bind and restructure t-SNARE complexes (23). The heavy chain, as well as both light chains of each endocytic t-SNARE, contributes a large N-terminal domain. These Nterminal extensions are neither homologous to each other nor to other SNAREs.
We observed that the presence of the N-terminal domain of both Syntaxins Tlg2p-(36 -191) and Pep12p-(1-188) reduced the fusion rate (Fig. 1B). The results observed for both endocytic complexes are consistent with those previously obtained for Syntaxin 1 (16) and extend the potential regulatory role of the heavy chain N-terminal domain. Interestingly, the removal of the N-terminal domain of each light chain, Tlg1p-(1-117) and Vti1p-(1-118), increases the fusion kinetic as well (Fig.  1C). However, it is still necessary to activate the t-SNARE complexes with Snc2-C-pept to promote fusion.
In Yeast, the Light Chain SNARE Motif Contributes to a Second Potential Regulatory Switch-Even when the inhibition created by the combined effect of the yeast t-SNARE N-terminal domains is removed, the rate of fusion in the absence of peptide activation is still very low (Fig. 2B). Thus an additional inhibitory element may be involved. To determine whether this second element is specific for yeast or is instead a general feature of endocytic SNAREs, we investigated the fusion capacity of a corresponding mammalian late endocytic complex (8,24). We reconstituted the t-SNARE [Syntaxin7/ Syntaxin8,Vti1b] into acceptor liposomes and the v-SNARE VAMP8 into donor liposomes and tested the fusion efficiency of this complex with or without VAMP8-C-pept. As shown in Fig.  3, these liposomes now fuse efficiently without any added activator, although the peptide still increases the fusion rate. Thus mammalian and yeast complexes have different inherent fusogenicities. Combined with the results presented above, we assumed that the additional inhibition of the yeast complexes arises from an element located in the SNARE motif. Mixing experiments were conducted to locate this switch. were reconstituted into acceptor liposomes and Snc2p into donor liposomes. Acceptor liposomes were preincubated for 2 h at room temperature with 0.05 unit of thrombin/l of liposomes (red curves). Control samples were incubated for 2 h at room temperature in absence of thrombin (black curves). 100 mM AEBSF was added to all samples to stop the thrombin reaction. Donor and acceptor liposomes were mixed (5:45 l), and N-(7-nitro-2,1,3-benzoxadiazol-4-yl)-1,2-dipaltmitoyl phosphatidylethanolamine fluorescence was monitored in the presence or absence of Snc2-C-pept (3 nmol) during 2 h at 37°C. The results were converted to rounds of fusion as described (16). B, the cleavable Syntaxins (indicated by an asterisk) were included in the following t-SNARE complex [Tlg2*/Tlg1,Vti1] and [Pep12*/Tlg1,Vti1], replacing the wild-type proteins. Both acceptor populations were treated with (red curves) and without (black curves) thrombin, and fusion was monitored as described for A. C, cleavable light chains were inserted in the following t-SNARE complexes [Tlg2/Tlg1*,Vti1] or [Tlg2/Tlg1,Vti1*] and [Pep12/Tlg1*,Vti1] or [Pep12/Tlg1,Vti1*], replacing the wild-type proteins. All acceptor populations were treated with (red curves) and without (black curves) thrombin, and fusion was monitored as described for A. In A-C, all constructs are described in the schematics, with the thrombin site labeled by the red arrow. Wild-type proteins are insensitive to thrombin cleavage. Both Syntaxin N-terminal domains as well as each light chain N-terminal domain exert a negative effect on fusion.
were fusogenic, the hypothesis being that if the yeast proteins carry a regulatory element in their sequence, replacing it with the mammalian homolog will release the negative regulation. We reconstituted the different t-SNARE yeast/mammal combi-nations into acceptor liposomes (Fig. 4) and tested their fusion capacity in the presence or absence of Snc2-C-pept.
Exchanging Tlg1p with its mammalian homolog Syntaxin8 released the blockage. In this situation, the complex [Pep12p/  Fig 1A. B, the t-SNARE proteins were replaced with the three cleavable counter-parts to generate [Tlg2*/Tlg1*,Vti1*] and [Pep12*/Tlg1*,Vti1*] complexes. All donor populations were treated with or without thrombin as indicated on the figure, and fusion was monitored as described in the legend to Fig. 1A. C; the percentage of increase is plotted, averaging at least three experiments. For each individual experiment, all results were normalized based on the fusion rate obtained after 2 h with the full-length proteins. The removal of two N-terminal domains has a stronger effect than the removal of a single one. The removal of all t-SNARE N-terminal domains dramatically enhances the fusion rate, establishing that multiple regulatory domains can act simultaneously. Syn8,Vti1p] becomes fusogenic without activation with Snc2-C-pept. When replacing both yeast light chains Tlg1p and Vti1p with their mammalian counterparts Syn8 and Vti1b, the resulting t-SNARE [Pep12p/Syn8,Vti1b] is even more fusogenic, suggesting a synergistic effect of both light chains for fusion. We also observe that Snc2p (Fig. 4A) and VAMP8 (Fig.  4B) can functionally replace each other. Altogether, these results suggest that within a given cellular pathway, variations in SNARE autoregulation may be specifically encoded within the light chains of the t-SNARE.
The Autoregulation of Tlg1p Is Encoded in the N-terminal Half of Its SNARE Motif-The SNARE motif is in average 60 amino acids long and is organized in ϳ16 layers based on its helical structure with the zero layer determining the center of the SNARE motif (7,8). To identify the region responsible for the slow kinetics in yeast, we created chimeric constructs between the Syntaxin8 and the Tlg1p SNARE motif. In these constructs, the Tlg1p N-terminal part of the SNARE motif (from layers Ϫ9 to 0) was joined to the Syntaxin8 C-terminal part of the SNARE motif (from layers 0 to ϩ9) resulting in T1-S8 protein and vice versa (resulting in S8-T1 protein). Tlg1p (T1-T1) and Syntaxin8 controls (S8-S8) were constructed using the same cloning strategy. Each of these constructs is shown in Fig. 5A. These chimeras were expressed as GST-recombinant proteins and reconstituted into donor liposomes together with the truncated forms of Pep12p and Vti1p. While both controls (T1-T1 and S8-S8) retain their fusogenic properties (Fig. 5, B and C, respectively), the chimera T1-S8 requires activation by a peptide (VAMP8-C-pept) for fusion (Fig. 5D). The chimera S8-T1, on the other hand, does not require peptide activation for fusion (Fig. 5E). We note that when C-Syn8 replaces C-Tlg1p, Snc2-C-pept does not activate the complex anymore (Fig. 5D). This effect may have different causes: either Snc2-C-pept is simply unable to bind the C-terminal part of Syn8, or Snc2-C-pept can still bind but is unable to activate the complex, suggesting in this case that an adequate cognate structure on C terminus is necessary for releasing the N-terminal switch. At this point of the study, we are unable to discriminate between both possibilities.
Altogether, this demonstrates that a second switch region is buried in the Tlg1p SNARE motif and is restricted to its Nterminal portion (between the layers Ϫ9 and 0). DISCUSSION This study suggests that SNAREs possess multiple distinct autoregulatory switches situated in different regions of the t-SNARE complex. These elements can be combined in particular t-SNAREs (yeast versus mammalian endocytic complexes), providing the SNARE system with functional flexibility to answer to the specific needs of different transports pathways in different cells.
The first switch is provided by the combination of the three t-SNARE N-terminal domains. While an autoregulatory role for syntaxin N-terminal domain is well known (16) and now includes Tlg2p and Pep12p, the role of the light chain Nterminal domains has been unclear. The light chain N-terminal domains do not interact intramolecularly with the SNARE motif and have no influence on the t-SNARE complex formation in solution (15,19). In this study, we found that the N-terminal domains of the yeast endocytic light chains can additively control the kinetics of fusion.
A second potential switch is buried in the SNARE motif of the yeast light chain Tlg1p. This regulation seems to be absent from a mammalian homolog. Possibly, the tighter autoregulation in yeast relates to the multiple roles for the endocytic v-SNARE, Snc2p. Not only do both endocytic complexes share this v-SNARE with each other, they also share it with the exocytic complex [Sso1p/Sec9p] (29). By contrast, in mammals there are several distinct VAMPs which are further functionally specialized (30).
It is interesting to note that although the C-peptide binds the C-terminal part of the SNARE motif, it releases a regulatory element located within the N-terminal half. t-SNARE complexes are partially assembled across the most N-terminal portions of the SNARE domain (23,31). By using a peptide homologous to snc2-C-pept, our group showed previously that VAMP2-C-pept binds and structures the C-terminal part of the t-SNARE complex (called the tc-fusion switch), consequently increasing the fusion rate (23). In particular, Melia et al. (23) FIG. 3. The mammalian late endocytic t-SNARE complex [Syn-taxin7/Syntaxin8,Vti1b] is fully fusogenic with the v-SNARE VAMP8. Syntaxin7, Syntaxin8, and Vti1b were reconstituted into acceptor liposomes and VAMP8 into donor liposomes. Both liposome populations were mixed and fusion was monitored in presence or absence of VAMP8-C-pept (3 nmol) as described. As a control, the fusion reaction was inhibited with an excess (6 nmol) of the soluble VAMP8 (VAMP8⌬TMD). The C-peptide increases the fusion rate but is not a requirement. observed that when the v-SNARE C-peptide is bound, t-SNARE becomes as resistant to proteolysis as when the entire soluble VAMP2 is bound, suggesting that structuring the C-terminal half is sufficient to induce complete coiled-coil formation across the SNARE bundle. Although we cannot pinpoint the site of regulation for the second potential switch in the endocytic complexes, it seems likely that the N-terminal region is refolded in response to Snc2-C-pept binding. When N-Syn8 replaces N-Tlg1p and therefore compensates the Nterminal switch, peptides still have an activation effect on fusion (Fig. 5E). In this case, the peptides are still acting on the tc-fusion switch.
In conclusion there are multiple autoregulatory switches engulfed into t-SNAREs. These switches are presumably governed by additional cellular regulatory mechanisms, which could be protein or lipids. One candidate regulator for the yeast endocytic complex is the GARP (Golgi-associated retrograde proteins) multisubunit tethering complex. This complex, together with Vps52p and Vps53p, is required for retrograde transport to the late Golgi. It has been shown recently that two members of this complex, Vps51p and Vps54p, interact directly with the N-terminal domain of Tlg1p (32,33). An attractive model is that the binding of Vps51p/Vps54p to Tlg1p stabilizes the N-terminal region of the SNARE motif and releases the SNARE complex to promote fusion. FIG. 5. The second switch is buried in the N-terminal region of Tlg1p SNARE motif. A, chimeric SNARE motifs were created between the mammalian Syntaxin8 (yellow) and the yeast Tlg1p (gray). Both SNARE motif sequences are shown. In this schematic, each layer is represented by the gray box (the red Q corresponds to the zero layer). The identity between both sequences is depicted in blue. B-E, after reconstitution of each chimera together with ⌬Pep12p and ⌬Vti1p, fusion with Snc2p liposomes was monitored in the presence or absence of Snc2-C-pept (⌬Pep12/T1-T1/⌬Vti1 (B), ⌬Pep12/S8-S8/⌬Vti1 (C), ⌬Pep12/T1-S8/⌬Vti1 (D), and ⌬Pep12/S8-T1/⌬Vti1 (E)). As control, the fusion was inhibited with an excess (6 nmol) of the soluble Snc2p (Snc2cyt). The fusion was recorded during 2 h at 37°C. We observed that the regulatory element is buried in the N-terminal region of Tlg1p SNARE motif.