Sec1p protein family members play an essential but poorly understood role in regulation of membrane fusion in eukaryotic cells (Gallwitz and Jahn, 2003
; Toonen and Verhage, 2003
; Kauppi et al., 2004
). Numerous studies in different model systems have resulted in partly contradictory conclusions on their function. Here, we studied the Sec1p binding protein Mso1p, to understand its functional link to Sec1p and regulation of membrane fusion. We previously noticed that no sporulation occurred in diploid yeast cells deleted for MSO1
(Jantti et al., 2002
). This prompted us to investigate in more detail the role of Mso1p in this process. In meiosis, formation of the spores occurs inside the mother cell. Thereby, the PSM, which are the precursors of the spore walls and the plasma membranes of the spores, become assembled de novo, discontinuously from the plasma membrane of the mother cell. The meiotic spindle pole bodies provide on their cytoplasmic side a specific scaffold, called the meiotic plaque, on top of which the earliest structures of the prospore membranes can be found (Moreno-Borchart and Knop, 2003
). Up to now, it had been enigmatic how the initial steps of PSM assembly occur. Our observation of a class of small and homogenous vesicles in the Δmso1
mutant that seem to have docked to the SPB, but not fused, suggest that PSM formation is mediated by homotypic vesicle fusion that most likely involves secretory vesicles. This is in agreement with the previous finding that sec1, sec4
, or sec8
mutants also block prospore membrane formation, although in these cases no detailed analysis of the phenotype with regard to the processes at the SPB has been carried out (Neiman, 1998
). This suggests that core exocytic machinery is involved in all steps that govern the PSM assembly pathway, including the very early assembly processes. However, during PSM initiation, the membrane fusion machinery seems to be more stringently regulated compared with constitutive exocytosis in vegetatively grown cells. This is exemplified by the fact that PSM formation depends on the presence of Mso1p, whereas in Mso1p-deleted vegetatively grown cells only a moderate accumulation of vesicles at the sites of membrane growth is visible (Aalto et al., 1997
). This essential function of Mso1p in PSM formation seems to be mediated through its interaction with Sec1p as mutant cells expressing a Sec1p binding defective form of Mso1p had a very similar phenotype compared with cells where the whole Mso1p was deleted.
The findings that Mso1p is essential for membrane fusion, most likely in association with Sec1p, prompted us to study in more detail the molecular interactions of Mso1p with Sec1p. Mapping the Sec1p interaction domain of Mso1p by the yeast two-hybrid system revealed an amino-terminal peptide (aa 37–59) to be necessary and sufficient for this protein–protein interaction. This interaction is direct because it could be reproduced in vitro with bacterially expressed components. Mso1p seems to form a rather stabile complex with Sec1p as Sec1p copurified in stoichiometric amounts with ProtA-Mso1p from yeast lysates under stringent conditions and in all experiments performed, an Mso1p–Sec1p complex was always observed. Within the Sec1p interacting peptide, T47 turned out to be a critical amino acid for Mso1p in vivo function and Sec1p–Mso1p interaction. The in vivo effect of the T47A mutation is presently unclear. Based on the in vitro binding results with bacterially expressed proteins it is evident that phosphorylation of T47 is not an absolute requirement for Sec1p binding. However, we cannot presently exclude the possibility that, phosphorylation of this residue may still serve as a means to regulate the Mso1p–Sec1p interaction in vivo. By using the mso1T47A mutant, we could show that cells expressing mso1T47A in combination with sec2 or sec4 mutations were viable, although this mutation formed a synthetic lethal combination with sec1 mutants. This is in contrast to the previous findings where complete loss of Mso1p formed a lethal combination also with sec2 and sec4 mutations. This underlines the importance of the Sec1p binding domain for Mso1p in vivo function and suggests that the carboxy-terminal part of Mso1p functionally interacts with the GTPase Sec4p and its exchange factor Sec2p.
Sec4p has been shown to interact with Sec15p, a subunit of the exocyst tethering complex (Guo et al., 1999
). Furthermore, in sec2
mutants no assembly of plasma membrane Sso1/2–Sec9–Scn1/2 SNARE complexes takes place and localization of Sec1p to the bud site is lost (Carr et al., 1999
). These results functionally link Sec4p with Sec1p and SNARE complex function. However, no physical interactions between Sec4p and the SNARE complex or the SNARE complex interacting Sec1p has been reported so far. Our immunoprecipitation experiments carried out both with Mso1p and Sec1p identified coprecipitation of Mso1p, Sec1p, the Ssop–Sec9p–Sncp SNARE complex subunits, and the exocyst subunit Sec15p. Under the conditions used, we did not detect Sec8p in the Mso1p–Sec1p-containing complexes. This suggests that only Sec15p and possibly some other subunits of the exocyst complex associate directly or indirectly with Mso1p. The coimmunoprecipitation of Mso1p and Sec1p with Sec15p is supported by the recent findings showing that Sec1p coimmunoprecipitates with the exocyst complex or even its subcomplexes lacking some of the subunits (Wiederkehr et al., 2004
). Previously, we showed that MSO1
deletion in combination with conditional mutations in exocyst subunits (sec3-2, sec5-24, sec6-4, sec8-9, sec10-2
, and sec15-1
) and the t-SNARE component sec9-4
formed a harmful combination for the cells (Aalto et al., 1997
). At the same time, no synthetic interaction was detected when combining MSO1
deletion and sec18-1
, a mutation affecting the disassembly of the SNARE complexes. These results support the close functional association of Mso1p with vesicle tethering and SNARE complex assembly. It has previously been proposed that exocyst complex assembly takes place during vesicle docking at the plasma membrane (Wiederkehr et al., 2004
). Although Sec8p was not detected in our immunoprecipitations, it is possible that Mso1p interacts with the vesicle associated exocyst complex at the early stages of vesicle tethering. Exocyst is a large protein complex (Hsu et al., 1998
), and it is likely that the complex has to be at least partially disassembled or displaced before SNARE assembly. Therefore, Mso1p also could facilitate formation of the t-SNARE complexes in a timely manner with the exocyst complex disassembly. The timing of these associations could be regulated through the association of Mso1p with Sec4p.
A functional association of Mso1p with Sec4p is supported by the findings that the combined effect of MSO1
deletion and sec4-8
mutation is lethal even at the permissive temperature. This lethality is not observed when only the Sec1p binding domain of Mso1p is mutated indicating that the Sec4p interaction is undisturbed by this mutation. We observed coimmunoprecipitation of Sec4I133 protein, a dominant negative mutant of Sec4p, with Mso1p, whereas under the conditions used, we could not observe coimmunoprecipitation with wt Sec4p, sec4-8p, or Sec4N34p. In sec4-8, SEC4N34
mutant cells association of Mso1p with Sec1p persisted, whereas at the same time in sec4-8
cells, interaction with the SNARE complex subcomponents was impaired. Our results thus suggest that the association of Mso1p with Sec1p is not regulated by the GTPase activity of Sec4p. Sec4-8p binds poorly GTP, Sec4N34 binds preferentially GDP, whereas Sec4I133p is presumably defective both in GDP and GTP binding (Walworth et al., 1989
). Although, Sec4N34p and I133 mutants bind similarly the Sec4p function regulating proteins Sec2p and Dss1p (Collins et al., 1997
; Walch-Solimena et al., 1997
), these mutants are likely to posses slightly different conformations due to their different nucleotide binding properties. This could affect their binding to Mso1p or to possible, presently unknown proteins that may mediate this interaction. We speculate that GTPase cycle defective Sec4I133 protein is locked into a Sec1p and Mso1p containing complex and that this interaction impairs Mso1p–Sec1p association with the SNARE machinery and thus assembly of functional SNARE complexes. On GTP hydrolysis, Sec4p may normally be rapidly released from Sec1p and Mso1p and thus allow assembly of the t-SNARE complexes. A role for Mso1p as a protein facilitating Sec1p and Sec4p interaction with the SNARE machinery could explain the apparently positive role for Mso1p in membrane fusion as deletion of it or inhibition of its interaction with Sec1p impairs membrane fusion. Localization of Mso1-GFP to the bud tip and the site of PSM generation suggest that Mso1p–Sec1p–Sso1/2p interactions take place at sites of active membrane fusion. Because of the nature of immunoprecipitation experiments, we do not know whether all proteins coimmunoprecipitated here are part of a large protein complex or whether several smaller Mso1p containing complexes with different compositions have been retrieved. Clearly, to fully understand Mso1p function and the detailed composition of the protein complexes it associates with, further biochemical and genetic studies are needed. These results nevertheless for the first time position Mso1p to the molecular interface of the putative vesicle tethering machinery and the SNARE machinery.
In analogy to the exocyst, the HOPS protein complex is required for vesicle tethering during homotypic vacuole fusion. Within this complex, both the SM protein Vps33p and the Rab-family small GTPase Ypt7p were found in association with the SNARE machinery (Wurmser et al., 2000
; Price et al., 2000
). Our results together with the previous results with the HOPS complex imply that in different membrane fusion events, there exists a molecular machinery that links the vesicle tethering machinery with the SNARE machinery, possibly through a small GTP binding protein. In exocytosis, Mso1p and Sec1p seem to possess molecular interactions to fulfill such a task.
Database searches identified Mso1p as a homologue of the PTB domains of mammalian Sec1p binding Mint proteins. In mammalian cells, three Mint proteins exist. Mint1 and Mint2 proteins bind to Munc18 (Okamoto and Sudhof, 1997
), the neuronal Sec1p homologue, whereas the ubiquitously expressed Mint3 has not been shown to interact with Munc18 (Okamoto and Sudhof, 1998
). Mint proteins are multidomain proteins, which contain, in addition to the PTB domain, two PDZ domains and a separate Munc18 interaction domain (Okamoto and Sudhof, 1997
). The PTB domain of Mint proteins has been reported to bind to the YENPTY peptide present in Alzheimer's disease-linked APP and to regulate cellular APP levels (Biederer and Sudhof, 2000
; King and Turner, 2004
). In addition to its association with Munc18, Mint1 protein has been shown to be part of a tripartite protein complex containing Mint1 (LIN-10), CASK (LIN-2), and Veli (LIN-7) and also to bind presenilins that are components of the APP processing enzyme γ-secretase (Borg et al., 1998
; Butz et al., 1998
; Lau et al., 2000
). Based on a mouse knockout study, Mint1 and 2 may display partially redundant neuronal functions (Ho et al., 2003
). Importantly, this study showed a clear role for Mint1 in neurotransmitter release. Although the detailed cellular function of Mint proteins remains obscure, the existing data support their role as adapter proteins that may form protein scaffolds with diverse downstream effectors in neuronal exocytosis. Although the expression of the Mint PTB domains could not rescue the loss of Mso1p in yeast cells (our unpublished data), the analogy of Sec1p binding and the sequence homology support the idea that Mso1p and Mint proteins are functional homologues. The interaction of Mso1p with Sec1p is confined to a short amino-terminal region of Mso1p, which makes it likely that the remaining part of the Mso1p has additional interaction partners. The Mint PTB domain also has been shown to bind phosphoinositol-4,5-bisphosphate and other phosphoinositol phosphates that are required for exocytosis (Okamoto and Sudhof, 1997
; Odorizzi et al., 2000
; Cockcroft and De Matteis, 2001
). Our attempts to detect a similar interaction of Mso1p with PIPs were not successful. This leaves the identification of the ligand for the putative PTB-pocket of Mso1p to future work.
Together, our results demonstrate that vesicle fusion at the SPB in meiosis is a highly regulated membrane fusion process that bears evolutionarily conserved features with regulated vesicle fusion processes in higher eukaryotes. The functional importance of Mso1p for membrane fusion in yeast meiosis indicates that tight control of Sec1p via Mso1p has evolved as a distinct way how to mediate timed vesicle fusion. This unexpected novel layer of functional conservation in exocytosis regulation in yeast indicates an important additional role for the PTB-domain like protein Mso1p in conjunction with Sec1p, the exocyst subunit Sec15p, and the small GTPase Sec4p.