|Home | About | Journals | Submit | Contact Us | Français|
The exocyst is an essential protein complex required for targeting and fusion of secretory vesicles to sites of exocytosis at the plasma membrane. To study the function of the exocyst complex, we performed a structure-based mutational analysis of the Saccharomyces cerevisiae exocyst subunit Sec6p. Two “patches” of highly conserved residues are present on the surface of Sec6p; mutation of either patch does not compromise protein stability. Nevertheless, replacement of SEC6 with the patch mutants results in severe temperature-sensitive growth and secretion defects. At nonpermissive conditions, although trafficking of secretory vesicles to the plasma membrane is unimpaired, none of the exocyst subunits are polarized. This is consistent with data from other exocyst temperature-sensitive mutants, which disrupt the integrity of the complex. Surprisingly, however, these patch mutations result in mislocalized exocyst complexes that remain intact. Our results indicate that assembly and polarization of the exocyst are functionally separable events, and that Sec6p is required to anchor exocyst complexes at sites of secretion.
In eukaryotic cells, membrane-bound vesicles are required for carrying protein and membrane cargo between functionally distinct organelles and to the plasma membrane for exocytosis. The precise spatial and temporal regulation of vesicle fusion is achieved by a number of essential proteins (Wickner and Schekman, 2008 and references therein). Specific SNARE proteins, present on both the vesicle (v-SNARE) and the target (t-SNARE) membranes, are critical for membrane fusion. Tethering proteins, which are proposed to bridge the vesicle and target membranes at a distance, are either dimeric coiled-coil proteins or components of multisubunit protein complexes (Sztul and Lupashin, 2006 ). These proteins interact with the donor and target membranes through specific interactions with small Ras superfamily GTPases and phospholipids (Wu et al., 2008 ). Many tethering proteins bind directly to SNAREs and the regulatory Sec1/Munc18 family proteins and likely play crucial roles in the temporal and spatial regulation of SNARE complex assembly before membrane fusion (Cai et al., 2007 ; Kummel and Heinemann, 2008 ).
The exocyst is a conserved multisubunit protein complex required for tethering and fusion of vesicles at specific sites of polarized secretion on the plasma membrane (TerBush et al., 1996 ; Guo et al., 1999 ; Munson and Novick, 2006 ; Wu et al., 2008 ). This complex is distantly related to tethering factors found in other trafficking pathways, including the GARP (Golgi-associated retrograde protein), COG (conserved oligomeric Golgi), and Dsl1 complexes (Whyte and Munro, 2002 ; Koumandou et al., 2007 ). The exocyst is comprised of eight subunits: Sec3p, Sec5p, Sec6p, Sec8p, Sec10p, Sec15p, Exo70p, and Exo84p. These proteins localize to regions of exocytosis and membrane growth in eukaryotic cells, where their activity is regulated by small Rab, Rho, and Ral GTPases (Wu et al., 2008 ). Temperature-sensitive (ts) yeast exocyst mutants accumulate secretory vesicles at sites of secretion (bud tips and mother-bud necks) at a stage before exocytic SNARE complex assembly. The exocyst also plays critical roles in Drosophila, mammals and plants, not just in exocytosis, but also in endocytic recycling and cytokinesis (Gromley et al., 2005 ; Sommer et al., 2005 ; Oztan et al., 2007 ; Hala et al., 2008 ).
Elucidation of the molecular function of the exocyst requires detailed knowledge of each of the subunits and their assembly into functional complexes. High-resolution structures of four exocyst subunits have been determined: the C-terminal domain of Drosophila melanogaster Sec15 (Wu et al., 2005 ), the nearly full-length yeast Exo70p (Dong et al., 2005 ; Hamburger et al., 2006 ; Moore et al., 2007 ), the C-terminal domain of yeast Exo84p (Dong et al., 2005 ), and the C-terminal domain of yeast Sec6p (Sivaram et al., 2006 ). They fold into remarkably similar elongated helical bundle structures with diverse surface features (Munson and Novick, 2006 ; Sivaram et al., 2006 ). Protein–protein interactions within the exocyst complex and with small GTPases have been identified genetically, by qualitative in vitro binding studies and by yeast two-hybrid analyses (reviewed in Munson and Novick, 2006 ). The exocyst associates with secretory vesicles through the interaction of Sec15p with the Sec4p GTPase (Guo et al., 1999 ), and most of its subunits appear to traffic with secretory vesicles to sites of secretion (Boyd et al., 2004 ). Exocyst function is regulated by interactions of Sec3p and Exo70p with membrane-anchored Rho GTPases and phosphoinositides (Adamo et al., 1999 ; Guo et al., 2001 ; Zhang et al., 2001 , 2008 ; He et al., 2007 ), as well as interactions with the yeast lethal giant larvae (Lgl) homologues Sro7p and Sro77p (Lehman et al., 1999 ; Zhang et al., 2005a ; Grosshans et al., 2006 ; Hattendorf et al., 2007 ). Most of the subunits are essential in yeast, except Sec3p, although sec3Δ cells grow quite slowly (Wiederkehr et al., 2003 ). In addition, the lethality of sec5Δ and exo70Δ can be bypassed by overexpression of either SEC1 or SEC4 (Wiederkehr et al., 2004 ). These structural, biochemical, and genetic studies indicate that each of the subunits plays diverse roles in exocyst function; however, many questions remain about the specific function(s) of the exocyst and its subunits.
Here, we focus on mutational studies of the yeast exocyst subunit Sec6p, which forms a direct link between the exocyst complex and the exocytic plasma membrane SNARE protein Sec9p and has been proposed to regulate SNARE complex assembly (Sivaram et al., 2005 ). Qualitative protein–protein interaction studies also indicated that Sec6p can dimerize in vitro (Sivaram et al., 2005 ), interact with the exocyst subunits Sec10p, Sec8p, and Exo70p (Dong et al., 2005 ; Sivaram et al., 2006 ), and bind Rtn1p, an interaction that is important for cortical ER inheritance (De Craene et al., 2006 ). We identified and mutated two patches of residues on the surface of Sec6p that are highly conserved from yeast to humans and show that these mutants have ts growth and secretory defects in yeast. These phenotypes are due to a loss of polarized localization of the exocyst subunits at nonpermissive conditions. However, unlike other exocyst ts mutants, exocyst complexes are intact at the nonpermissive conditions, indicating that the Sec6p patch residues are necessary to anchor functional, assembled exocyst complexes at sites of secretion. The conservation of these patch residues suggests an anchoring role for Sec6p in all eukaryotes.
Mutations (sec6-49: L418A, Y422A, W433A, Q470A, and Q474A; sec6-54: D607A, T632A, E635R, Y636A, and D639R) were introduced into the SEC6 gene by PCR, and cloned into the NdeI and BamHI sites of pET15b (Novagen, Madison, WI) containing an N-terminal His6 tag. For analysis in yeast, constructs were placed into the BamHI and NotI sites of pRS315 (LEU2 CEN) including the endogenous promoter of SEC6 (0.5 kb of flanking 5′ and 3′ genomic DNA). A C-terminal triple hemagglutinin (HA) tag was added to the pRS315 constructs for use in indirect immunofluorescence and immunoprecipitations. All clones were confirmed by sequencing. The sec6 LEU2 CEN plasmids were introduced as the sole copy of SEC6 in yeast by plasmid shuffling the wild-type SEC6 URA3 CEN plasmid out of MMY204 (Table 1). For green fluorescent protein (GFP) and synthetic lethal analyses, MMY204 was mated with strains containing either a single C-terminally GFP-tagged exocyst subunit (Huh et al., 2003 ) or a single ts mutation (gifts from P. Novick, Yale University; W. Guo, University of Pennsylvania; and J. Jäntti, University of Helsinki). Diploid strains were sporulated and dissected, and the mutant plasmids were introduced into the GFP+ sec haploids as described above. Media, growth conditions, and genetic methods for yeast were described previously (Munson et al., 2000 ). Yeast strains used in this study are listed in Table 1. Invertase secretion was performed as in Adamo et al. (1999) .
His6-Sec6p, His6-Sec6-49, and His6-Sec6-54 were expressed and purified as described for His6-Sec6p (Sivaram et al., 2005 ), with the following change for the His6-Sec6-49 anion-exchange buffers: 10 mM HEPES, pH 7.4, and 1 mM DTT ± 1 M NaCl. Protein concentrations were determined by ninhydrin protein assay (Rosen, 1957 ). Circular dichroism spectroscopy was performed as described (Sivaram et al., 2005 ). Maltose binding protein (MBP) pulldown binding assays were performed as described (Sivaram et al., 2006 ) using a 20 mM HEPES, pH 7, 100 mM NaCl, 1 mM EDTA, 0.1% NP-40, and 1 mM DTT binding buffer. For experiments with MBP-Exo84CT, the soluble C-terminal domain (residues 523-753; Dong et al., 2005 ) was subcloned into pMALc2X (New England Biolabs, Beverly, MA). MBP-Exo84CT was expressed at 20°C for 3 h after induction with 0.1 mM IPTG and purified by amylose resin affinity chromatography.
Strains containing wild-type or mutant Sec6p with C-terminal HA3 tags were grown to log phase in synthetic complete (SC) media at 25°C and either fixed or shifted to nonpermissive conditions (prewarmed YPD at 37°C for 1 h) and then fixed with 37% formaldehyde. Cells were spheroplasted, permeabilized in HS/SDS buffer (0.1 M HEPES, pH 7.4, 1.0 M sorbitol, and 0.5% SDS), and washed twice in HS buffer. Permeabilized cells were placed on slides (Electron Microscopy Sciences, Fort Washington, PA) prepared with 0.1% polylysine. A 1:400 dilution of α-HA-AlexaFluor 488 (Molecular Probes, Eugene, OR) was added to visualize Sec6p-HA3. For Sec4p, a 1:500 dilution of α-Sec4 antibody (P. Brennwald, University of North Carolina) was used, followed by washing and incubation with 1:1000 goat α-mouse-AF488 (Molecular Probes). Differential interference contrast (DIC) and fluorescent images were obtained at room temperature using an Axioskop2 plus epifluorescent microscope (Zeiss, Thornwood, NY) fitted with a 100× Plan-NEOFLUAR (Zeiss 1.30 NA oil immersion) objective lens. Images were collected using a Diagnostic Instruments camera (Sterling Heights, MI; model 2.1.1) and 3rd Party Interface Advanced (ver. 3.5.4 for MacOS) software. Immunofluorescent images were adjusted for total contrast in Adobe Photoshop (San Jose, CA).
Strains containing C-terminal GFP-tagged exocyst subunits and either wild-type or mutant Sec6p were grown to log phase in SC media at 25°C and shifted to nonpermissive conditions for 1 h. Samples were immediately resuspended in PBS containing 10% glycerol, fixed with 37% formaldehyde for 10 min, washed, and resuspended in PBS with 10% glycerol. Cells were viewed as described above. Localization of the GFP-tagged proteins was quantitated in Adobe Photoshop by counting cells with localized or mislocalized exocyst subunits. For each strain and condition, n > 100 cells were counted for two replicates each.
Strains were grown to log phase in SC media at 25°C or shifted to nonpermissive conditions for 1 h. One hundred OD units of cells were pelleted and washed with 10 mM Tris, pH 7.4, and 10 mM NaN3 at either 25 or 37°C. Supernatants were removed and pellets were resuspended in lysis buffer (50 mM HEPES, pH 7.4, 150 mM KCl, 0.5% NP-40, 1 mM DTT, and 1 mM PMSF). Cells were lysed by bead beating, and the lysates cleared by centrifugation and normalized for total protein concentration by Bradford assay (Bio-Rad, Hercules, CA). The supernatants were added to prewashed protein A beads (GE Healthcare Bio-Sciences, Piscataway, NJ) and mixed for 30 min at 4°C. Input samples were collected, the remaining supernatants were placed in fresh tubes, and 1:50 dilutions of HA antibody (Roche) were added. Samples were incubated at 4°C for 1 h. Lysate-antibody mixes were added to fresh, prewashed protein A beads. These mixtures were incubated at 4°C for 30 min, the supernatants were removed, and the beads were washed with lysis buffer. Input and bound fractions were analyzed by 10% SDS-PAGE gels and Western blotting. Exocyst subunits were detected using antisera raised against C-terminal constructs of Exo70p, Exo84p, and Sec10p (Pocono Rabbit Farm and Laboratory, Canadensis, PA) or antibodies against Sec3p, Sec5p, and Sec15p (generous gifts from P. Brennwald, University of North Carolina), followed by HRP-conjugated goat α-rabbit secondary antibody (Jackson ImmunoResearch Laboratories, West Grove, PA) and ECL (GE Healthcare Bio-Sciences), and visualized using a luminescent image analyzer (Fuji Film, Tokyo, Japan; LAS-3000).
We previously determined the x-ray crystal structure of the C-terminal domain of yeast Sec6p, residues 411-805 (Sivaram et al., 2006 ). This domain is sufficient for interactions with the exocyst subunits Exo70p and Sec10p, but not Sec8p or the t-SNARE Sec9p (Dong et al., 2005 ; Sivaram et al., 2006 ). Structure-based sequence alignments revealed two highly conserved surface “patches,” which we hypothesize are important sites for protein–protein interactions in all eukaryotes (Figure 1A; Sivaram et al., 2006 ). These patches are composed of the following amino acids: L418, Y422, W433, Q470, Q474, and V478 in Patch-1, and D607, T632, E635, Y636, and D639 in Patch-2. Patch-2 overlaps a conserved acidic region, as well as being in close proximity to the original sec6-4 ts mutation (L633P; Novick et al., 1980 ; Lamping et al., 2005 ; Sivaram et al., 2006 ). When these patch residues were individually mutagenized to alanine in the context of the full-length Sec6p and substituted for the wild-type in yeast, no apparent effects on growth rates were observed (data not shown). We reasoned that this lack of phenotype could be due to tight protein–protein interactions at these surfaces and that single alanine perturbations would be insufficient to disrupt them. Therefore, we combined mutations in each patch (Figure 1A). All the residues in Patch-1 were mutated to Ala (called sec6-49). For Patch-2, residues 607, 632, and 636 were mutated to Ala and residues 635 and 639 to Arg, causing a reversal of charge at these positions (called sec6-54). These mutations were anticipated to severely disrupt protein–protein interactions in these regions.
Because these are surface mutations, they should not significantly destabilize the α-helical structure of Sec6p. To test this, we cloned and purified the recombinant Sec6p mutant proteins from Escherichia coli and analyzed their secondary structure and stability using circular dichroism. As expected, the patch mutations did not decrease the stability of the proteins. Both of the mutant proteins showed the same characteristic α-helical signal as the wild-type Sec6p (~50% helicity) and also had similar thermal stabilities to wild-type at 37°C (Figure 1B). In contrast, the protein containing the ts mutant allele sec6-4 must be considerably destabilized, because we were unable to produce any soluble recombinant protein in E. coli (data not shown).
To test the function of these mutant proteins as the sole copy of Sec6p in yeast, we performed a plasmid shuffle experiment to replace the wild-type SEC6 with the mutants. The mutants were expressed under control of the endogenous promoter from a low-copy (CEN) plasmid and had equivalent expression levels to the endogeneous Sec6p at both 25 and 37°C (data not shown). Both patch mutants were able to replace the wild-type SEC6 URA3 plasmid when grown on SC plates containing 5-fluoro-orotic acid (5-FOA) at 25°C. However, on SC plates at higher temperatures, both mutants exhibited ts growth defects (Figure 2A). In addition, both sec6-49 and sec-54 showed severe growth inhibition when plated on rich YPD medium, even at temperatures that were permissive for growth on SC medium (Figure 2A). The growth rates of the mutants in SC liquid cultures were similar to wild-type at 25°C and were two- to threefold slower than wild-type at 37°C, and the mutants did not grow in liquid YPD at 37°C (data not shown). The slow growth on rich YPD media may be explained by the higher rate of exocytosis compared with synthetic media, which would exacerbate the secretory defect. Similar results have been observed for other exocyst mutants (Haarer et al., 1996 ; Wiederkehr et al., 2003 ). Consistent with their growth phenotypes, the mutant strains were defective for secretion of the marker protein invertase at the nonpermissive conditions (Figure 2B). Because the recombinant mutant proteins were stable both in vitro and in vivo, we conclude that Sec6p's function, rather than its stability, is compromised by mutation of residues in these conserved patches.
To determine the defect(s) caused by mutations of the conserved patch regions, we examined the phenotypes of the mutant yeast strains in more detail. First, we monitored the localization of the Sec6p mutants at the permissive and the restrictive conditions. The mutants could not be visualized using a GFP tag, because placing a GFP tag at either the N- or C-terminal end resulted in slow growth, even at conditions permissive for growth of the untagged mutant (data not shown). However, when we tagged the mutants at their C-termini with a triple HA epitope tag, these strains grew similarly to the untagged strains (data not shown). Immunofluorescence experiments demonstrated that the patch mutants were properly polarized to tips of the growing bud and mother-bud necks at the permissive conditions (Figure 3). After a shift to the nonpermissive conditions (YPD at 37°C for 1 h), the mutant proteins were completely mislocalized (Figure 3), suggesting that the mutations disrupt the ability of Sec6p to localize to, or remain localized at, sites of secretion. Similarly, the sec6-4 protein was not polarized at the nonpermissive temperature (Lamping et al., 2005 ).
To rule out that mislocalization of the Sec6p mutants was caused by a general disruption of the secretory pathway, we monitored Sec4p localization by immunofluorescence. Sec4p is the Rab family GTPase associated with secretory vesicles; its localization is sensitive to perturbations of the secretory pathway (Ayscough et al., 1997 ; Walch-Solimena et al., 1997 ; Roumanie et al., 2005 ). We found that Sec4p localization was unperturbed in the sec6-49 and sec6-54 mutant cells after 1 h at nonpermissive conditions (Figure 3), as was previously observed for sec6-4 and other exocyst ts mutants (Walch-Solimena et al., 1997 ; Roumanie et al., 2005 ). Therefore, although the Sec6p patch mutant proteins are mislocalized at the nonpermissive conditions, the sec6 mutant cells appear fully competent in localizing secretory vesicles to sites of polarized secretion. Moreover, consistent with the observation that Sec4p trafficking is normal, we observed no obvious morphological defects in the mutant cells (data not shown).
We next tested the localization of the other exocyst subunits at the nonpermissive conditions to investigate whether mislocalization of Sec6p mutants would also affect exocyst complex localization. On the basis of previous results with sec6-4 and other exocyst ts mutants (Finger et al., 1998 ; Roumanie et al., 2005 ; Zhang et al., 2005b ; He et al., 2007 ), we expected to find mislocalization of most of the subunits due to disruption of complexes at the nonpermissive conditions. The exceptions to this would be subunits proposed to localize and/or stabilize the complex at sites of secretion: Sec3p, Exo70p, and Exo84p. To visualize each of the exocyst subunits individually, we constructed strains in which the genomic copy of one of the exocyst subunits was C-terminally GFP tagged (Huh et al., 2003 ), and the only copy of SEC6 was either the wild-type or one of the patch mutants (Table 1). We observed significant mislocalization of all the exocyst subunits at the nonpermissive conditions (75–90%, except Sec3-GFP, which was ~50%; Figure 4, A and B). As a control, we also observed similar mislocalization of each of the exocyst subunits in sec6-4 cells (data not shown), consistent with previous results (Roumanie et al., 2005 ). This suggested two possibilities: 1) the exocyst complex was being disassembled and dispersed from sites of secretion; or 2) interactions with a factor(s) at sites of secretion were destabilized, and hence the subunits became mislocalized.
We reasoned that the conserved patches on Sec6p are important for interactions with other exocyst subunits. If true, mutation of these patches would therefore cause global destabilization and disassembly of the exocyst, resulting in dispersal of the subunits. This phenotype has been previously reported for the sec6-4 ts allele, as well as for ts alleles of several other exocyst subunits (TerBush and Novick, 1995 ). To examine whether the exocyst complex is disassembled at the nonpermissive conditions, we examined the state of the complex by coimmunoprecipitation analyses. For these experiments, we immunoprecipitated the wild-type and mutant Sec6p-HA proteins with α-HA antibodies, and assayed for the presence of the other exocyst subunits by Western blot analyses. Surprisingly, we found that, even after 1 h at nonpermissive conditions, similar amounts of the exocyst subunits were coimmunoprecipitated with wild-type and the Sec6p mutants—indicating that the complexes were still assembled (Figure 5A).
The presence of assembled exocyst complexes at nonpermissive conditions is clearly distinct from similar immunoprecipitation experiments using the sec6-4 ts strain; the complexes in sec6-4 cells were completely disassembled and/or degraded at the nonpermissive temperature (TerBush and Novick, 1995 ). To demonstrate that the exocyst is capable of disassembly under our experimental conditions, we immunoprecipitated wild-type Sec6p-HA in a strain containing the sec5-24 ts allele as the sole copy of SEC5. We used sec5-24 because degradation of the sec6-4 protein precluded the use of the Sec6-4-HA for immunoprecipitations. We found, at nonpermissive conditions, nearly complete loss of the other exocyst subunits bound to Sec6p (Figure 5B). Another possibility is that exocyst complexes are in a dynamic equilibrium and can reassemble after lysis during the course of our immunoprecipitation experiments. To exclude this possibility, we performed a lysate mixing experiment using two strains with differently tagged subunits (Figure 5C). The first strain contained SEC8-GFP and untagged SEC6 (Huh et al., 2003 ); the second had SEC8-MYC3 and SEC6-HA3. The strains were lysed, either kept separate or mixed in a 1:1 ratio, and then the Sec6p-HA was immunoprecipitated with α-HA antibodies. The Myc-tagged Sec8p coimmunoprecipitated with Sec6p-HA (from the same strain), but the GFP-tagged Sec8p did not bind Sec6p-HA (from the mixed lysates), indicating that exocyst complexes do not form during the immunoprecipitation experiments. Therefore, we conclude that exocyst complexes remain intact in the patch mutant strains, even at nonpermissive conditions where mislocalization occurs.
The finding that the exocyst remains assembled at nonpermissive conditions suggests that the conserved patches are not required for Sec6p to interact with other subunits of the exocyst complex. We have previously shown that the C-terminal domain of Sec6p is sufficient for binding both Exo70p and Sec10p (Sivaram et al., 2006 ). In contrast to the full-length Sec6p, however, this domain is not sufficient for binding Sec8p and the t-SNARE Sec9p, nor for dimerization (Sivaram et al., 2005 , 2006 ). Here, we examined the interactions between the full-length Sec6p patch mutants and Exo70p, Sec10p, and Sec8p using qualitative pulldown assays and found that the Sec6p mutants bound with affinities similar to that of wild type (Figure 5D). As a negative control, we found that neither the wild-type nor the mutant Sec6p proteins interacted above background with the C-terminal domain of Exo84p (residues 523-753; Dong et al., 2005 ). Furthermore, as expected from our previous data, gel filtration experiments demonstrated that the mutants were still dimeric and capable of binding Sec9p (data not shown). Together, these binding results corroborate the immunoprecipitation results, and indicate that the binding partner(s) for the conserved patches is not one of the exocyst subunits, nor Sec9p.
To test this possibility further, we examined genetic interactions between the mutants and the other exocyst subunits. We first tested for suppression of the growth phenotypes by overexpression of the other exocyst subunits, as well as SEC4, SEC1, and SEC9, the other plasma membrane t-SNARE, SSO1, and the secretory vesicle v-SNARE, SNC2, from 2μ plasmids. We also tested for suppression by overexpression of an “open” mutant of Sso1p (Munson and Hughson, 2002 ). None of the exocyst subunits or SEC4 or SEC1 suppressed the sec6 mutants (data not shown). These results support the idea that the Sec6p binding partner(s) is not a subunit of the exocyst complex. In contrast, overexpression of SEC9, SSO1, and SNC2 all suppressed the sec6-49 and -54 ts phenotypes (Figure 5E). The likely explanation is that loss of assembled exocyst complexes at sites of secretion leads to a decrease in SNARE complex assembly, which can be rescued by overexpression of the plasma membrane SNAREs. We also expected that the open Sso1p mutant would rescue this defect, because it readily forms SNARE complexes. However, similarly to the synthetic defects observed in combination with other exocyst ts mutants, this mutant failed to suppress the sec6-49 and -54 ts phenotypes (Figure 5E), likely because of a detrimental overaccumulation of SNARE complexes (Munson and Hughson, 2002 ). In contrast, the sec6-4 allele, in which the exocyst complex is completely disassembled (TerBush and Novick, 1995 ), has a more severe block in SNARE complex assembly (Grote et al., 2000 ) that cannot be suppressed by overexpression of SSO1 and SEC9 (Aalto et al., 1993 ). Additionally, unlike the sec6-4 allele, the sec6-49 and -54 mutants were synthetically lethal when crossed with most of the exocyst ts strains (Table 2). The differences in synthetic lethality between our sec6 mutants and sec6-4 (Finger and Novick, 2000 ) may reflect the severity of the ts defects in these cells, especially sec6-54, rather than specific functional differences. However, the combination of destabilization of the exocyst by the sec ts mutants, together with partial mislocalization of the complex by the sec6-49 and -54 mutations at 25°C, could result in more severe defects. Together, the biochemical, phenotypic, and genetic interactions observed for both of the patch mutants indicate that sec6-49 is very similar, albeit weaker, to the sec6-54 allele. In support of this idea, we found that combining the sec6-49 and -54 mutations together resulted in similar ts and YPD growth defects as the individual mutants (data not shown). The patch residues appear to comprise a single binding site for an as yet unknown factor, which is not an exocyst complex subunit or Sec9p.
We have identified a novel class of Sec6p mutants that disrupt the polarized localization of the exocyst complex at nonpermissive conditions, yet do not affect complex assembly. We show that mutation of conserved surface residues on the C-terminal helical bundle domain of Sec6p do not destabilize full-length Sec6p (Figure 1), yet result in severe temperature- and media-sensitive growth defects (Figure 2). At the nonpermissive conditions, vesicle trafficking to sites of secretion appears unimpaired, whereas the Sec6p patch mutants are completely mislocalized (Figure 3). Consequently, all of the other exocyst subunits are also mislocalized (Figure 4). Previous studies of exocyst temperature sensitive mutants showed that this mislocalization is caused by destabilization and disruption of the exocyst complex. However, we show that exocyst complexes remain intact in the sec6 patch mutant strains, even at the nonpermissive conditions (Figure 5).
Unexpectedly, the extensive exocyst mislocalization observed in our sec6-49 and -54 strains at the nonpermissive conditions included Sec3p, which was proposed to be the spatial landmark for exocyst localization (Finger et al., 1998 ). Sec3p-GFP appeared polarized in yeast cells in the absence of ongoing exocytosis and in other exocyst ts mutants (Finger et al., 1998 ; Zhang et al., 2005b ). However, localization of the endogeneous Sec3p by immunofluorescence indicated that the GFP tag may artificially stabilize Sec3p-GFP at sites of secretion (Roumanie et al., 2005 ), although this was recently disputed (Zhang et al., 2008 ). Localization of the endogenous Sec3p in sec6-4 cells (Roumanie et al., 2005 ) was similar to our results with Sec3-GFP in the sec6-4 strain. Several variations between our studies and previous results could account for these discrepancies: the use of plasmid-borne, instead of genomic GFP-tagged Sec3p; the testing of Sec3-GFP in the presence of the endogenous Sec3p; differences in strain backgrounds; and the fact that our construct contains a linker sequence (9 a.a.) between the C-terminus of Sec3p and the GFP (Huh et al., 2003 ). Our data, combined with the observations that Sec3p is not an essential gene and that the other exocyst subunits are polarized in sec3Δ cells at 25°C (Wiederkehr et al., 2003 ; Zhang et al., 2005b ), suggest that Sec3p is not the spatial landmark for the exocyst complex.
Other candidates for targeting the exocyst to sites of secretion are Exo70p and Exo84p (Boyd et al., 2004 ; He et al., 2007 ). However, we found that Exo70p-GFP was mislocalized in the sec6-49 and -54 strains at the nonpermissive conditions (Figure 4). Similarly, although the localization of Exo84-GFP was proposed to be independent of the exocyst subunits (e.g., not mislocalized in previous ts strains; Zhang et al., 2005b ), Exo84-GFP was also mislocalized in the sec6-49 and -54 strains (Figure 4). Therefore, we conclude that the conserved surface patches on Sec6p are required for localization and/or stability of the exocyst complex at sites of secretion and that Sec3p, Exo70p, and Exo84p are not sufficient to localize the complex in these sec6 mutant strains. Nevertheless, the presence of a landmark is not strictly necessary, because actin cables are polarized independently of the exocyst, bringing vesicles to the correct sites of exocytosis (Walch-Solimena et al., 1997 ). Therefore, we propose that exocyst subunits travel to sites of secretion where they assemble together, and that Sec6p is required to anchor the assembled exocyst complexes at these sites. Perhaps Sec6p acts in cooperation with Sec3p, Exo70p, and/or Exo84p for stabilization of the exocyst complexes on the plasma membrane or perhaps their interactions (e.g., with Rho proteins and phosphoinositides) are more regulatory in nature (Roumanie et al., 2005 ; Wu et al., 2008 ). It may be that either Exo70p or Sec3p are sufficient to localize the exocyst, when wild-type Sec6p is present to stabilize the complex at sites of secretion (Hutagalung et al., 2008 ).
The spatial and temporal regulation of exocyst assembly and disassembly is currently unclear. Most of the exocyst subunits appear to traffic to the plasma membrane with secretory vesicles (Boyd et al., 2004 ), although it is unknown whether they traffic as preassembled complexes or they assemble upon vesicle arrival. Assembly upon arrival at sites of secretion, which was suggested by FRAP data (Boyd et al., 2004 ), is compelling, because the interactions between subunits on the vesicle with those on the plasma membrane could tether the membranes together. If so, then the patch residues of wild-type Sec6p would be necessary to anchor the complex at these sites after vesicle arrival. In the case of the patch mutants, Sec4p would need to hydrolyze the bound GTP, perhaps upon exocyst assembly, to release Sec15p and therefore the rest of the exocyst complex. On the other hand, it is possible that the exocyst assembles at sites of vesicle formation and budding; the patch residues of Sec6p would be required for proper trafficking of the complex to sites of secretion. This could only be the case if vesicles are capable of transport in the absence of assembled exocyst complexes, because trafficking of vesicles containing Sec4p is unimpaired in the patch mutant strains (as well as in the sec6-4 and other exocytic sec mutants). This situation would require that Sec4p on the budding vesicle hydrolyze the bound GTP to release the complex. Further mutational and GFP localization studies will be necessary to distinguish between these scenarios.
Regardless of where assembly takes place, the finding that the complex remains assembled despite a block in exocytosis and in the absence of polarized accumulation suggests two possibilities. First, the complex cannot disassemble, because disassembly is triggered by a factor(s) at sites of secretion or by membrane fusion. If the complex is locked into an assembled conformation in the sec6-49 and -54 mutants, then it may be unable to be recycled for subsequent rounds of vesicle tethering and fusion, thus contributing to the ts phenotype. Alternatively, the complex is assembled in these mutants because it normally does not disassemble unless destabilized. In this case, mislocalization of the assembled exocyst complexes at nonpermissive conditions would lower the concentration of active exocyst complexes that are available to function at sites of secretion. This would ultimately result in decreased SNARE complex assembly and membrane fusion.
Conservation of these Sec6p patch residues suggests an important anchoring function for Sec6p in all eukaryotes. Our analyses indicate that the binding partner for the patches on Sec6p is not a subunit of the exocyst complex, nor is it the t-SNARE Sec9p (Figure 5). Therefore, further genetic and biochemical experiments will be necessary to identify the anchoring factor and analyze its role in polarizing Sec6p and the exocyst complex at sites of secretion.
In memory of Jen Songer (1979–2008), who died in a car accident on December 9, 2008. Her legacy is a passion for life and science, which continues to inspire everyone who knew her. We miss her very much. We are grateful to J. Towle, N. Croteau, and B. Farley for technical assistance. We thank P. Novick, P. Brennwald, W. Guo, and J. Jäntti for yeast strains, antibodies, and plasmids, C. R. Matthews and his lab for use of the CD, and N. Rhind for use of the fluorescence microscope and advice. We thank R. Gilmore, C. Carr, N. Rhind, and members of the Munson lab for critical reading of this manuscript and suggestions. This work was supported by National Institutes of Health Grant GM068803 to M.M.
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E08-09-0968) on December 10, 2008.