Lipid-dependent regulation of exocytosis in S. cerevisiae by OSBP homolog (Osh) 4

Polarized exocytosis is a key cellular process among eukaryotes by which membrane and proteins are delivered to a defined point on the plasma membrane (PM). This process can facilitate the elaboration of specialized structures at the cell surface, such as the apical-basolateral domains of intestinal epithelial cells or the formation of a bud in Saccharomyces cerevisiae (He and Guo, 2009; McCaffrey and Macara, 2011; Bi and Park, 2012). If this process becomes deregulated, polarized cell growth fails, leading to disruption of processes supported by these specialized structures, such as nutrient absorption in the intestine or daughter cell growth in S. cerevisiae.

Although minor differences in the mechanics of polarized exocytosis exist among eukaryotes, key events, divided between the early exocytic pathway, i.e. endoplasmic reticulum (ER) to the trans-Golgi network (TGN) and the late exocytic pathway (TGN to PM) are conserved (Keller and Simons, 1997; Barlowe and Miller, 2013). Conservation of mechanism and organization allows for the study of exocytosis in less complex organisms, such as S. cerevisiae. In the late exocytic pathway of S. cerevisiae, Golgi-derived vesicles are transported from the mother cell to sites of polarized growth by myosin motors (Govindan et al., 1995). These vesicles are then tethered to the PM by the exocyst (Guo et al., 1999), followed by the assembly of vesicle-associated (v)-SNAREs (Snc1p or Snc2p) and PM-associated target (t)-SNAREs (Sso1p or Sso2p and Sec9p) into trans-SNARE complexes (Protopopov et al., 1993; Rossi et al., 1997). These trans-SNARE complexes subsequently scaffold fusion triggers (Wickner and Rizo, 2017) that promote vesicle fusion with the PM allowing for the release of vesicle cargo into the extracellular environment.

In addition to proteins, lipids contribute to exocytosis as well. Phosphatidylinositol phosphates (PIPs) have emerged as particularly important for successful exocytosis. Sec3p and Exo70p, two PM-associated exocyst subunits, localize to specific regions of the PM based on their interaction with PM phosphatidylinositol-4,5-bisphosphate (PI4,5P) (He et al., 2007; Zhang et al., 2008). This interaction is key because assembly of the vesicle-associated exocyst subunits (Sec5p, Sec6p, Sec8p, Sec10p, Sec15p and Exo84p) with a PM-associated subunit is required, in one model, for vesicle docking at the PM (Boyd et al., 2004). An additional role for PIPs in exocytosis centers on phosphatidylinositol-4-phosphate (PI4P). When post-Golgi vesicles form, they are enriched with PI4P and marked by the Rab protein Ypt32p (Ortiz et al., 2002; Strahl and Thorner, 2007). However, successful exocytosis requires replacement of Ypt32p with the Rab protein Sec4p (Mizuno-Yamasaki et al., 2010), for which PI4P must be removed from the vesicle membrane (Mizuno-Yamasaki et al., 2010; Rossi and Brennwald, 2011). A subsequent model and study suggested that, in S. cerevisiae, the oxysterol-binding protein homolog 4 (Osh4p, officially known as Kes1p) facilitates the removal of vesicular PI4P to promote Sec4p loading onto vesicles (Beh et al., 2012; Ling et al., 2014). This observation provided an important example of an OSBP homolog binding a lipid ligand to facilitate an important cellular event, the maturation of exocytic vesicles from a docking-incompetent to a docking-competent form.

Oxysterol-binding proteins (OSBPs) and OSBP-related proteins (ORPs) belong to a large protein family conserved from yeast to humans, comprising twelve members in humans and seven in budding yeast. Members of this protein family bind one or more lipids, including PI4P (Im et al., 2005; de Saint Jean et al., 2011; Maeda et al., 2013; Mesmin et al., 2013; Tong et al., 2013; Chung et al., 2015). While OSBP and many ORPs bind two different lipid species, the binding pocket can only accommodate one lipid at a time because the binding sites for each lipid species within the pocket are mutually exclusive (de Saint Jean et al., 2011). In S. cerevisiae the seven OSBP homologs comprise the Osh protein family, absence of which results in the loss of proper cell polarization (Kozminski et al., 2006) and growth (Beh et al., 2001). Any one family member is sufficient to support cell viability (Beh et al., 2001) and Osh4p alone is sufficient to maintain proper cell polarization (Kozminski et al., 2006). As with the mammalian OSBP family, there is a variety of lipid-binding activities within the Osh protein family. All seven Osh family proteins bind or are predicted to bind PI4P. Osh6p and Osh7p also bind phosphatidylserine (PS), whereas Osh1p, Osh2p, Osh4p and Osh5p bind or are predicted to bind sterol (Im et al., 2005; Maeda et al., 2013; Tong et al., 2013; Manik et al., 2017). However, it has not been shown what essential role lipid binding by the Osh protein family fulfills in yeast (Beh et al., 2001). It has been proposed that sterol transfer is the essential role; however, considering that a subset of the Osh family proteins does not bind sterol this is unlikely (Im et al., 2005; Maeda et al., 2013; Tong et al., 2013).

Previous studies with S. cerevisiae have shown that Osh proteins are required for efficient polarized exocytosis, a key cellular process, but important questions remained (Kozminski et al., 2006; Alfaro et al., 2011). These studies did not determine which specific step(s) in the late exocytic pathway require Osh proteins or whether lipid binding by Osh proteins is a requirement for Osh protein function in exocytosis. Here, we show that a specific pathway of polarized exocytosis requires Osh protein activity in a lipid-dependent manner and that solely Osh4p can supply this activity. Further, we show that Osh protein activity is required for efficient vesicle docking at the PM and that lipid binding by Osh proteins is a requirement for this function. We describe specific in vivo roles for lipid binding by an OSBP homolog and propose a novel two-step mechanism for Osh-dependent regulation of polarized exocytosis.

At the beginning of this study, we examined two defined exocytic pathways in S. cerevisiae – the Bgl2p-marked pathway that supports polarized cell growth at the bud tip and bud neck (Harsay and Bretscher, 1995; Adamo et al., 2001), and the invertase (Suc2p)-marked pathway that supports non-polarized cell growth (Harsay and Bretscher, 1995; Adamo et al., 2001). We asked, by using quantitative assays, whether Osh4p functions in one or both of these pathways, and whether its activity requires the binding of a specific lipid. We found that Osh4p is required in one exocytic pathway, the Bgl2p-marked pathway, and that its activity in this pathway is lipid dependent.

Consistent with what had been suggested in the literature (Beh and Rine, 2004), we found no role for Osh4p in the invertase-marked exocytic pathway (Fig. 1A,B). When the exocytosis of invertase was assayed, neither cells (oshΔ background) containing wild-type OSH4 nor a temperature-sensitive allele (osh4-1^ts) displayed a difference in invertase exocytosis when shifted from 25°C to 37°C for 4 h (Fig. 1A), the time period normally used to examine osh4-1^ts at non-permissive temperatures (Beh and Rine, 2004; Kozminski et al., 2006; Alfaro et al., 2011). Functional redundancy, which exists among members of the Osh protein family (Beh et al., 2001), did not mask an exocytosis defect in the mutant strain because the cells that lack chromosomal copies of all seven OSH family genes (oshΔ) are entirely dependent upon plasmid-borne OSH4 or osh4-1^ts for Osh protein activity. An exocytosis defect was also not masked by changes in cell physiology or invertase expression over time. At earlier time points, within 90 min post temperature shift [t=60 min after invertase de-repression (glucose washout)], both strains exhibited nearly identical levels of invertase exocytosis (Fig. 1B). In contrast to these two strains, the temperature-sensitive mutant sec6-4^ts, which encodes a defective exocyst subunit and is known to be defective in invertase exocytosis at 37°C (Novick et al., 1980; TerBush and Novick, 1995), exhibited markedly different kinetics under the same conditions (Fig. 1A,B). Therefore, the exocytic pathway marked by invertase does not appear to utilize Osh4p, though our results do not preclude a requirement for other Osh protein family members in this pathway.

In contrast to the exocytosis of invertase, we found, consistent with a previous report (Kozminski et al., 2006), that the members of the Osh protein family are necessary for Bgl2p-marked exocytosis. We show for the first time that an oshΔ [osh4-1^ts] strain accumulates Bgl2p internally 45−60 min after a shift from 25°C to 37°C (Fig. 1C). In contrast, the same strain containing a wild-type copy of OSH4 on a plasmid, rather than osh4-1^ts, exhibited no appreciable accumulation of Bgl2p post temperature shift. Deletion of OSH4 alone (all other OSH family genes present) produced a negligible (<5%) block in Bgl2p-marked exocytosis.

Observing that Bgl2p-marked exocytosis requires Osh protein activity led us to ask whether lipid binding by an Osh protein is also required. To answer this question, we took advantage of the observation that Osh4p is sufficient to support Bgl2p-marked exocytosis. We introduced into the oshΔ [osh4-1^ts] strain a second plasmid that contains wild-type OSH4 or one of several osh4 alleles that express at wild-type levels but confer specific defects in lipid binding, thereby generating a panel of strains dependent upon the activity of Osh4p expressed from this second plasmid when grown at 37°C (Fig. 1D; Table S1). Within this panel of isogenic strains, we found that strains with osh4 alleles that confer a defect either in PI4P binding (H143A/H144A) (de Saint Jean et al., 2011) or, by inference, a defect in both sterol and PI4P binding (Y97F+H143A/H144A) (Im et al., 2005; de Saint Jean et al., 2011) accumulated Bgl2p internally upon temperature shift from 25°C to 37°C (Fig. 1E; Fig. S1A). In contrast, a strain containing wild-type OSH4 did not accumulate Bgl2p at either temperature. A similar result was obtained when we assayed a strain carrying osh4^Y97F (Fig. 1F; Fig. S1B), which encodes an amino acid substitution that blocks sterol binding (Im et al., 2005). Because the osh4^Y97F allele is dominant lethal (in the absence of the H143A/H144A mutation), its expression was regulated with a methionine-repressible promoter, requiring it to be assayed independently of the other strains. Taken together, these results indicate that PI4P and sterol must bind Osh4p for it to function in the Bgl2-marked exocytic pathway.

Earlier studies indicated that Osh4p has a role in the late exocytic pathway (Kozminski et al., 2006; Alfaro et al., 2011). To exclude the possibility that lipid binding by Osh4p is required in the early exocytic (i.e. pre-Golgi) pathway, we examined electron microscopy (EM) thin sections of the strains described above. We looked to see whether these cells, which show defects in Bgl2p-marked exocytosis, accumulated vesicles of ∼50 nm diameter, the signature size of pre-Golgi transport vesicles (Kaiser and Schekman, 1990), under conditions that made these strains dependent upon a mutant Osh4p with a specific lipid-binding defect. We found that cells dependent upon PI4P-binding defective Osh4p (H143A/H144A) or an Osh4p predicted defective in binding both PI4P and sterol (Y97F+H143A/H144A) accumulated vesicles (Fig. 2A) similar to a strain in which no functional Osh proteins are available (vector alone, Fig. 2A) (Alfaro et al., 2011; Ling et al., 2014). The vesicles were ∼90 nm in diameter, the signature diameter of post-Golgi exocytic vesicles, rather than ∼50 nm, the signature diameter of Golgi and pre-Golgi transport vesicles (Fig. S2A; Kaiser and Schekman, 1990). In contrast, vesicle accumulation was observed in only one-third of the cells of the strain that carries a wild-type copy of OSH4. The only cells with an Osh4p-lipid-binding defect that did not accumulate vesicles were of strains expressing dominant lethal osh4^Y97F, which confers a sterol-binding defect (Fig. 2B). This was an unexpected because these cells, under the same experimental conditions, displayed a defect in Bgl2p-marked exocytosis (Fig. 1F). This may be explained by the observed accumulation of other membranous structures in these cells, with Bgl2p perhaps accumulating in those structures.

These data strongly suggest that neither PI4P nor sterol binding by Osh4p is required in a pre-Golgi step of exocytosis. Rather, binding of PI4P and Osh4p appears essential for one or more events in the Bgl2p-marked exocytic post-Golgi pathway.

After establishing that Osh4p-dependent, Bgl2p-marked exocytosis requires lipid binding to Osh4p, we asked where this regulation occurs. A previous study showed that exocytic vesicles transit from the mother cell into the bud, in the absence of Osh4p function (Alfaro et al., 2011), making it unlikely that lipid regulation of Osh4p affects the motility or direction of vesicle trafficking significantly. In addition, our data show a post-Golgi defect of vesicle accumulation when binding between PI4P and Osh4p is impaired (Fig. 2A). Therefore, a more likely point of regulation exists when exocytic vesicles dock with the PM, at the end of their journey from mother cell to bud. As reported earlier (Alfaro et al., 2011), the time an exocytic vesicle dwells subjacent to the PM at the bud tip is greater in the absence of functional Osh proteins, suggesting that an Osh protein is necessary for the efficient docking of exocytic vesicles with the PM.

To determine whether Osh proteins are necessary for the docking of exocytic vesicles at the PM, we assayed for assembled SNARE complexes, in the presence or absence of OSH4 (oshΔ background). When an exocytic vesicle docks at the PM, v-SNAREs tightly bind PM-associated t-SNAREs to form an assembled trans-SNARE complex (Grote, Carr and Novick, 2000). Because specific SNAREs are associated with specific membranes (Hong, 2005), we used the formation of complexes between hemagglutinin (HA)-tagged Snc2p and Sso1p or Sso2p (HA-Snc2p−Sso1/2p) as a read-out for exocytic vesicle docking with the PM. Snc2p is an exocytic v-SNARE that binds directly to either PM-associated t-SNARE Sso1p or Sso2p (Rossi et al., 1997). From whole-cell lysates of S. cerevisiae expressing 6×HA epitope-tagged Snc2p (HA-Snc2p), under the control of the endogenous SNC2 promoter, we pulled-down HA-Snc2p and then assayed the ratio of Sso1/2p to HA-Snc2p by immunoblotting. In the absence of the HA-tag or primary antibody, little Sso1/2p was detected on immunoblots, indicating that the pull-down of Sso1/2p depends upon HA-Snc2p (not shown). When we shifted cells that lack functional Osh proteins at 37°C (oshΔ [osh4-1^ts][vector]), from 25°C to 37°C for 75 min, we noted a trend toward lesser SNARE complex formation (HA-Snc2p−Sso1/2p; Fig. 3A; Figs S3, S4), with an on average ∼35% decrease (P=0.081; n=3) relative to control cells that contained a wild-type copy of OSH4 (oshΔ [osh4-1^ts][OSH4]). Although this trend suggested that Osh proteins are necessary for efficient vesicle−PM docking, we anticipated a more-significant effect considering the robust exocytosis defect observed when Osh activity is absent (Figs 1E and 2A). Therefore, we considered the possibility that a defect in endocytosis alters the amount of t-SNAREs and other proteins required for docking on the PM of mutant cells relative to control, diminishing the relative difference in the amount of assembled SNARE complexes detected between strains.

To determine whether an endocytosis defect exists in any of our osh mutants, we assayed microscopically for the vacuolar accumulation of Lucifer Yellow (LY), a fluorescent marker for fluid-phase endocytosis (Dulic et al., 1991). With cells that are wild-type for endocytosis, LY accumulates in one or more large, round, vacuolar compartments shortly after the addition of LY to the culture medium (Dulic et al., 1991). We saw the same with cells expressing a wild-type copy of OSH4 in the absence of any other wild-type OSH family genes (oshΔ [osh4-1^ts][OSH4]) or when, at 25°C, the second plasmid in the oshΔ [osh4-1^ts] strain background was an empty vector (Fig. 3B), consistent with previous report (Beh and Rine, 2004). This accumulation pattern was rarely observed when cells containing the empty vector were assayed after shifting to 37°C (Fig. 3B). When fluorescence accumulation was observed in these cells post temperature shift, compartments containing LY had irregular shapes and sizes. Similar results were observed for the strain dependent upon a PI4P-binding defective Osh4p (H143A/H144A) or an Osh4p defective in binding both PI4P and sterol (Y97F+H143A/H144A), though in the latter case defects in compartment morphology were noted even prior to temperature shift. These results suggest that the inability of Osh4p to bind a specific lipid, such as PI4P, in the absence of all other Osh proteins, is sufficient to inhibit endocytosis.

To account for the potential impact of endocytosis on the availability of docking sites on the PM, we assayed docking by a different method (Fig. 4A), for which the read-out for docking is the association of an exocytic vesicle-associated protein (e.g. Sec4p, HA-Snc2p) with PM. We isolated a PM fraction and determined the ratio of the amount of a vesicle-associated protein to the total amount of PM, calculated as the ratio of Sso1 and Sso2 proteins to total membrane, which corrects for changes in t-SNARE density on the PM due to endocytosis defects in osh mutants (Fig. 4A; Fig. S4). The PM fraction showed a >30-fold enrichment of the PM marker Pma1p relative to whole-cell extract and contained barely detectable cross-contamination by other membrane compartments, as determined by quantitative immunoblots for a variety of known membrane markers (Fig. 4B). In a test of this method (Fig. 4C), we assayed a known vesicle−PM docking mutant (sec6-4^ts) (Grote et al., 2000) and a known vesicle−PM fusion mutant (sec1-1^ts) (Grote et al., 2000), along with an isogenic control strain. We found that the vesicle−PM fusion mutant showed a significant association of Sec4p, a vesicle marker, with the PM. In contrast neither the vesicle-docking mutant nor the isogenic control strain exhibited this effect, validating that this method can differentiate defects in vesicle docking with the PM from defects in vesicle−PM fusion.

When we applied the ‘PM isolation’ method to our panel of osh mutant strains, we found that the ratio of t-SNAREs (Sso1/2p) to total PM varied among strains (Fig. S4A,B) and, in some cases, in the same strain when cultured at different temperatures, possibly due to allelic variations in the rate of endocytosis (Fig. 3B). This result therefore necessitated the normalizing correction for t-SNARE levels among strains, as described above, when calculating the ratio of v-SNARE to PM. With the PM isolation method, we found that cells lacking functional Osh proteins were unable to support vesicle association with the PM to the same extent as cells with a functional Osh family protein (Fig. 4D). Using the same assay, we found that mutants encoding PI4P-binding-deficient Osh4p did not rescue this defect and had less vesicle marker present on the PM relative to wild-type cells (Fig. 4D). These results indicate that Osh protein activity and PI4P binding by Osh4p in particular is required for vesicle docking at the PM.

We also found a third line of evidence supporting a role for Osh proteins in vesicle docking – vesicle cluster formation (Fig. 5). It is known that Sec4p-positive vesicles accumulate and cluster when there is increased Sec4p activity (Salminen and Novick, 1989; Rossi and Brennwald, 2011). We observed this clustering phenotype, by thin-section electron microscopy (TEM), within populations of cells that lack functional Osh proteins (oshΔ [osh4-1^ts] [vector] at 37°C; ∼23%, n=40; Fig. 5). The clustered vesicles had a mean diameter approximately that of exocytic vesicles, though less than that of unclustered vesicles (74 nm, n=60, vs 88 nm, n=121; Fig. S2B). The significance of this difference is unknown, though it may represent greater difficulty in ascertaining the boundaries of clustered structures. Indirect immunofluorescence microscopy, which has been used previously to detect vesicle clusters (Rossi and Brennwald, 2011), confirmed the presence of Sec4p-positive vesicle clusters in cells lacking Osh protein function (Fig. S5). Clustering was partially ameliorated by the presence of a plasmid containing wild-type OSH4 or PI4P-binding defective osh4 (osh4^H143A/H144A). These observations are consistent with the idea that a deficiency in PI4P-binding by Osh4p relieves cluster formation by partially blocking the loading of Sec4p onto exocytic vesicles, leading instead to the accumulation of nonclustered vesicles (Beh et al., 2012; Ling et al., 2014).

To determine how lipid binding to Osh4p regulates polarized exocytosis and, in particular, docking of exocytic vesicles to the PM, we investigated whether lipid binding by Osh4p regulates Osh4p association with late exocytic pathway organelles. It has been reported that PI4P binding by Osh4p is required to maintain Osh4p at sites of polarized exocytosis (Ling et al., 2014). While it is known that Osh4p localizes to the PM (Alfaro et al., 2011; Ling et al., 2014) and exocytic vesicles (Alfaro et al., 2011; Ling et al., 2014), it is not known whether specific lipid ligands that bind Osh4p are required for Osh4p localization to these membranes. To address this question, we examined by fluorescence microscopy and cell fractionation osh4Δ cells that contain a wild-type or mutant allele of OSH4 on a plasmid.

We found that the ability of Osh4p to bind specific lipids affected its ability to associate with exocytic vesicles in vivo (Fig. 6). As shown by fluorescence microscopy (Fig. 6A,B) and as expected from earlier observations (Alfaro et al., 2011; Ling et al., 2014), wild-type Osh4p fused to red fluorescent protein (Osh4p-RFP) colocalized with puncta of GFP-Sec4p, a marker of exocytic vesicles. This localization pattern was also observed with Osh4p^Y97F-RFP, indicating that the association of Osh4p with exocytic vesicles does not require sterol binding. In contrast, PI4P-binding-deficient Osh4p^H143A/H144A-RFP colocalized less frequently with GFP-Sec4p puncta than wild-type Osh4p, adopting instead a diffuse distribution in the cytoplasm. These data suggested that binding of PI4P by Osh4p promotes association of Osh4p with exocytic vesicles. Intriguingly, osh4p^{Y97F+H143A/H144A}-RFP, which is modeled to be lipid free, associated with exocytic vesicles as frequently as wild-type OSH4. These results indicate that, although lipid binding is not strictly necessary for the association of Osh4p with exocytic vesicles, the competitive binding of PI4P and sterol is likely to regulate the association of Osh4p with exocytic vesicles.

To independently validate the live cell imaging data, we analyzed Osh4p-exocytic vesicle association by cell fractionation (Fig. 6C-F). We found that Osh4p co-fractionated with the exocytic vesicle markers Sec4p and Bgl2p on buoyant density gradients of membrane preparations made from osh4Δ cells containing either plasmid-borne OSH4 or an allele of OSH4 that confers a defect in lipid binding. What varied among these gradients, in the fractions containing the peak amount of Sec4p and Bgl2p, was the ratio of Osh4p to Sec4p. The average ratio of Osh4p^H143A/H144A, which is defective in PI4P binding, to Sec4p was ∼50% less than the average ratio of wild-type Osh4p to Sec4p (Fig. 6C,D). In contrast, the ratio of Osh4p^{Y97F+H143A/H144A} to Sec4p, which binds neither PI4P nor sterol, or of Osh4p^Y97F to Sec4p, which is defective in sterol binding, approximated that of wild-type Osh4p (Fig. 6C-F). These data are consistent with our in vivo observations and suggest that the association of Osh4p with Sec4p-marked exocytic vesicles is an intrinsic property of Osh4p that is only regulated by lipid binding rather than dependent on it.

We also analyzed Osh4p association with the PM by gradient fractionation (Fig. 7A,B; Fig. S6). We found that a part of a wild-type Osh4p pool co-fractionated with PM (∼1% of total Osh4p). In addition we found that PI4P-binding-deficient Osh4p^H143A/H144A accumulated significantly on the PM (∼6.5-fold relative to wild-type) (Fig. 7A; Fig. S6A), consistent with the fact that this Osh4p can still bind sterol, a lipid enriched in the PM (van Meer et al., 2008). In contrast, sterol-binding-deficient Osh4p^Y97F co-fractionated with PM at approximately wild-type levels (Fig. 7B; Fig. S6B). These observations are inconsistent with the idea that Osh proteins associate with membranes solely on the basis of their lipid-binding capacity.

We also asked whether lipid binding by Osh4p is required for its localization at sites of polarized exocytosis, in contrast to colocalization with vesicles traversing a cell − as shown in Fig. 6A. To this end, we looked for localization of wild-type or lipid-binding defective Osh4p-YFP to sites of polarized exocytosis marked with RFP-tagged Sec4p (RFP-Sec4p), i.e. small buds (Fig. 7C) and mother-bud necks (Fig. 7D) (Sheu et al., 2000; Roumanie et al., 2005; Zajac et al., 2005). As expected, we found wild-type Osh4p present at these sites, colocalizing with RFP-Sec4p (Fig. 7C-E; Alfaro et al., 2011). We also expected lipid-binding defective Osh proteins to localize to these sites with the same allele-specific pattern observed in Fig. 6A. This was the case for Osh4p^{Y97F+H143A/H144A}. Contrary to prediction, we found a different allele-specific pattern for two mutants. We found PI4P-binding-deficient Osh4p^H143A/H144A present at sites of polarized exocytosis as often as wild-type Osh4p (Fig. 7E), indicating that PI4P binding does not promote Osh4p localization at sites of polarized exocytosis (Fig. 7E). As for the other mutant, we found that sterol-binding-deficient Osh4p^Y97F did not localize to these sites as often as wild-type Osh4p-YFP (Fig. 7E), consistent with a role for sterol binding by Osh4p at the PM just prior to vesicle docking.

Although our results suggested a key role for sterol in Osh-mediated polarized exocytosis, the finding that some Osh proteins bind PS rather than sterol (Maeda et al., 2013) challenged a role for sterol or suggested that not every Osh family protein can substitute for Osh4p to support polarized exocytosis. The latter appeared less likely because polarized exocytosis is required for growth and any Osh family member can support growth in the absence of the other family member (Beh et al., 2001). We tested both possibilities by comparing the ability of two different Osh family members to support polarized exocytosis; i.e. Osh4p, which binds PI4P and sterol (Im et al., 2005; de Saint Jean et al., 2011), and Osh6p, which binds PI4P and PS (Maeda et al, 2013; von Filseck et al., 2015).

First, we tested whether Osh6p can support Bgl2p-marked exocytosis as well as Osh4p in the absence of all other Osh family members. We found similar internal levels of Bgl2p among oshΔ OSH4, oshΔ OSH6 and wild-type strains (Fig. 8A), indicating Osh4p and Osh6p can individually support polarized exocytosis in the absence of other Osh family members. This result indicates that sterol binding to an Osh protein is not required for Osh-mediated polarized exocytosis.

Second, we tested whether the ability of Osh6p to support polarized exocytosis depends upon its ability to bind lipids, by using alleles of OSH6 that confer a defect in either PS binding (osh6^L69D; Maeda et al., 2013) or PI4P binding (osh6^H157A/H158A; von Filseck et al., 2015). When introduced into an oshΔ [osh4-1^ts] strain on a second plasmid neither allele supported cell growth at 37°C, the temperature at which the cells are dependent upon the second plasmid for growth, although wild-type OSH6 did support growth (Fig. 8B). Within this panel of isogenic strains, we found that strains with osh6 alleles that confer a lipid-binding defect accumulated Bgl2p internally upon temperature shift from 25°C to 37°C similar to a strain in which the second plasmid is an empty vector (Fig. 8C). In contrast, a strain containing wild-type OSH6 did not accumulate Bgl2p at either temperature. This result indicates that Osh6p must bind PI4P and PS in order to support polarized exocytosis.

In this study we found that polarized exocytosis depends upon binding of specific lipid ligands to an Osh protein. We also determined that Osh proteins function in the process of exocytic vesicle docking at the PM, making this study the first to demonstrate an in vivo regulatory role for lipid binding by an ORP in an essential cellular process and to establish a role for an ORP in vesicle docking at a target membrane. On the basis of our data, we propose below a two-step model for how an Osh protein mediates the maturation and subsequent docking of an exocytic vesicle at the PM.

Osh4p served as our model for Osh/ORP function in polarized exocytosis. Although significant evidence points to Osh4p having a role in this cellular process (Kozminski et al., 2006; Alfaro et al., 2011), functional redundancy within the Osh protein family (Beh et al., 2001; Ling et al., 2014; this study) necessitated the use of an oshΔ background for some assays. Thus, although we can conclude from these assays that an Osh protein has a role in a given cellular process, in a lipid dependent-manner, we can only state in the context of these assays that the presence of Osh4p is sufficient rather than necessary. We also found in our in vivo model that Osh6p, which binds PI4P and PS (Maeda et al., 2013) and can exchange them between membranes (von Filseck et al., 2015), can substitute for Osh4p, which binds PI4P and sterol (Im et al., 2005; de Saint Jean et al., 2011). This result suggests that Osh-dependent regulation of polarized exocytosis does not have a strict sterol requirement. In other words, it is possible that lipid binding by an Osh protein or the Osh-mediated exchange of lipids between membranes is more important than lipid identity per se, provided the Osh protein and its cognate lipids are present together.

Another caveat we recognize in our study is that the osh4-1^ts allele may only be defective in a subset of functions. That is, osh4-1^ts may not be truly null at non-permissive temperatures. Thus, we can only state that a role for Osh4p in a given cellular process was not found, rather than excluding a role for Osh4p in the process. Even with this caveat, osh4-1^ts has proven in this study and others to be a valuable tool for providing insight into Osh protein function (Beh and Rine, 2004; Im et al., 2005; Kozminski et al., 2006; Alfaro et al., 2011; Georgiev et al., 2011; Stefan et al., 2011).

Based on our results and data in the literature we propose a two-step model of Osh protein activity in polarized exocytosis (Fig. 9). When vesicles bud from the TGN they are enriched with PI4P and are marked by the Rab Ypt32p (Ortiz et al., 2007; Strahl and Thorner, 2007). Before these vesicles can dock with the PM, it appears that they undergo a maturation process that involves a change of molecular identity.

In the first step of the model (Fig. 9), an Osh protein, in this case Osh4p, is required for PI4P removal from the vesicle membrane to facilitate the loading of Sec4p onto the vesicle to produce docking-competent vesicles, as proposed by Beh and colleagues (Beh et al., 2012) and tested in part by others (Ling et al., 2014). Our data provide support for and fulfill predictions of this model. First, a lipid-free Osh4p associates with exocytic vesicles (Fig. 6), as one would predict for an Osh protein primed to extract lipids from a membrane. This is also consistent with the requirement of PI4P for association of Osh4p with endomembrane (Mousley et al., 2012). Second, in accord with our prediction, PI4P-binding-deficient Osh4p is not sufficient to support polarized exocytosis (Figs 1E and 2A). Third, expression of a PI4P-binding-deficient Osh4p protein ameliorates the clustering of docking-competent (Sec4p-positive) vesicles (Fig. 5) − which is expected if mutant Osh4p does not remodel the lipid composition of vesicles and the number of docking-competent vesicles decreases. These observations are consistent with the first step of our model, in which Osh4p promotes vesicle maturation. It is important to note that vesicle maturation, including the loading of Sec4p onto vesicles, occurs to some degree in the absence of Osh protein activity (Alfaro et al., 2011). This observation suggests a later, second, Osh-dependent step in polarized exocytosis.

The existence of a second step is inferred from a series of observations. First, Alfaro et al. (2011) showed that in the absence of Osh protein family activity, Sec4p-positive puncta accumulate in the cell, indicating that GFP-Sec4p loaded onto vesicles. The same study also showed that, in the absence of Osh protein activity, vesicles transited into the bud, indicating that the myosin Myo2p was properly attached to the vesicle, and that Sec5p-GFP puncta arrived at the PM, indicating that the exocyst is assembled on the vesicle. Neither event would have occurred without Sec4p being bound to the vesicle (Guo et al., 1999; Santiago-Tirado et al., 2011). Furthermore, we show in this study that osh4-1^ts at restrictive temperature promotes the formation of vesicle clusters (Fig. 5), an event that is dependent on Sec4p loading onto a vesicle (Salminen and Novick, 1989; Rossi and Brennwald, 2011).

For the second step of the model, we propose that Osh4p-bound PI4P must be exchanged for PM sterol, so that vesicles can efficiently transition from a state of exocyst-mediated tethering to a state of SNARE-mediated docking for vesicle−PM fusion to proceed. Existence of these two states was proposed by Merz and colleagues (Lo et al., 2011). At the start of the second step in the model, a docking-competent Sec4p-positive vesicle already exists, having transited to the site of polarized exocytosis as it matured. When the vesicle with PI4P-bound Osh4p associates with the PM, in essence forming a transient membrane contact site, Osh4p exchanges its bound PI4P for PM sterol (Fig. 9). Examples of OSBP/ORP-mediated PI4P-sterol exchange at membrane contact sites have been found in mammalian cells (Mesmin et al., 2013; Chung et al., 2015). We posit that lipid exchange occurs after the exocyst tethers a vesicle to the PM, though further study is needed to validate this idea. This terminal lipid exchange event serves as a spatial signal to indicate that the vesicle is at a proper distance to dock at the target membrane, one highly enriched in sterol. Thus, an Osh protein in this capacity would serve as a spatial regulator in the formation of trans-SNARE complexes.

A less-direct interpretation of available data is that an Osh protein is regulating cis-SNARE complex disassembly. This is a less likely possibility because an Osh deficiency increases the dwell time of exocytic vesicles subjacent to the PM (Alfaro et al., 2011). If SNARE complex disassembly was Osh-dependent, we would predict wild-type and osh4-1^ts cells to have equivalent vesicle dwell times, until the available pool unassembled SNAREs is exhausted. Moreover, our data do not point to a membrane fusion defect. Such a defect would be expected if SNAREs became limiting.

The existence of a terminal lipid exchange event is supported by a number of observations from this and previous studies. First, Drin and colleagues showed that Osh4p can exchange PI4P and sterol between membranes (de Saint Jean et al., 2011). Second, vesicle dwell time at the PM increases in the absence of Osh family activity, suggesting Osh proteins need to function at the PM for efficient exocytosis (Alfaro et al., 2011). Third, our study showed that the absence of Osh protein activity leads to a defect in vesicle docking, which places the observed exocytosis defect at the PM in proximity to where the lipid-exchange event would occur but, with regard to time, after Sec4p loads onto vesicles (Fig. 4). Fourth, we observed that sterol-binding-deficient osh4p^Y97F does not accumulate at sites of exocytosis to the same extent as sterol-binding competent Osh4p (Fig. 7C-E). This observation suggests that sterol binding by Osh4p is required for the interaction of Osh4p with the PM at sites of exocytosis, consistent with a model in which lipid exchange by Osh4p at the PM is needed for successful vesicle docking.

Other lipids enriched at the PM could facilitate vesicle docking at sites of polarized exocytosis as well. PS-PI4P exchange by Osh6p or Osh7p could serve the same role as sterol-PI4P exchange by Osh4p because PS is enriched at sites of polarized exocytosis in S. cerevisiae, just as sterol is enriched at sites of polarized exocytosis in S. pombe (Tiedje et al., 2007; Fairn et al., 2011; Makushok et al., 2016). Consistent with this idea is our observation that Osh6p can substitute for Osh4p to support polarized exocytosis (Fig. 8A) in a PI4P- and PS-dependent manner (Fig. 8C). These observations imply a significant role for lipids as mediators of polarized exocytosis.

If vesicle-associated Osh4p must exchange its bound PI4P for PM sterol, why does the sterol-binding-deficient Osh4p^Y97F not lead to vesicle accumulation? This apparent inconsistency can be explained by the nature of the osh4^Y97F mutation. Because osh4^Y97F is a dominant allele that is hyperactive, it might interfere with an unknown upstream process that leads to a block in vesicle formation (Alfaro et al., 2011). This idea is consistent with the absence of vesicle accumulation in cells expressing osh4^Y97F (Fig. 2B).

One essential function the Osh proteins share is support of polarized exocytosis, a function that is only lost upon removal of all functional Osh proteins (Kozminski et al., 2006). Therefore, it can be inferred that all seven Osh proteins must contribute to polarized exocytosis to some extent and that, even if Osh4p is not the primary exocytic Osh protein, Osh4p is clearly sufficient to fill the role and serve as a model for Osh protein function in polarized exocytosis. Because polarized exocytosis is a conserved essential function, and OSBPs and ORPs comprise a conserved protein family, we anticipate that other studies will identify a role for OSBPs and ORPs in exocytosis in other cell types.

S. cerevisiae strains were grown in synthetic media at 25°C unless otherwise stated (Sherman et al., 1986). When temperature-sensitive alleles were used, 25°C was permissive and 37°C restrictive. When using strains with genes under the control of the pMET25 promoter, methionine at 100 mg/l was added to repress transcription (Mumberg et al., 1994; Alfaro et al., 2011).

All S. cerevisiae strains used in this study are described in Table S2. To generate strains with N-terminal 6xHA-tagged SNC2, we used the protocol of Gauss et al. (2005), using primers oKK355 and 356 and pOM12 as a template. PCR with primers oKK357 and 358 confirmed the integration. Successful recombination was confirmed by PCR with primers oKK357 and 358 and by immunoblotting, using an anti-HA antibody (catalog no. 901501, Biolegend, San Diego, CA). For all experiments, a minimum of two independent clones or transformants were analyzed.

All plasmids and oligonucleotides used in this study are described in Tables S3 and S4, respectively.

Plasmids were made as follows. pRS316-osh4^Y97F (pKK1990) was made by subcloning the SacI/KpnI fragment from pCB662 (Im et al., 2005) into pRS316.

pRS316-osh4^{Y97F+H143A/H144A} (pKK1988) was made by PCR-mediated site-directed mutagenesis (catalog no. 210518-5, Agilent, Santa Clara, CA) using primers oKK295 and 296, to incorporate the Y97F mutation into pRS316-osh4^H143A/H144A, which was validated by DNA sequencing using primer oKK193.

All pRS316-OSH4-YFP (pKK 2089, pKK2092, pKK2093, pKK2094) plasmids were made as follows. Using primers oKK367 and oKK369, OSH4-YFP was amplified by PCR from plasmid pCB866 (pKK1965). The PCR product was cloned into the SpeI and XhoI sites of pRS316-pMET25 (pKK2005), forming pMET25-osh4^Y97F-YFP (pKK2089). The pMET25 promoter was then removed by SacI/EcoRI digest and replaced with a SacI/EcoRI fragment from plasmids containing an OSH4 promoter and an osh4 allele of interest (pKK1921, pKK1950 and pKK1988).

All pRS414-OSH4-RFP (pKK2109, pKK2110, pKK2111, pKK2112, pKK2113) plasmids were made as follows. First, using primers oKK367 and oKK322, OSH4-RFP was amplified by PCR from genomic DNA isolated from KKY1240. The PCR product was cloned into the SpeI and XhoI sites of pRS316-pMET25 (pKK2005), forming pMET25-OSH4-RFP (pKK2107). To construct pRS316-pMET25 (pKK2005), pKK1990 was digested with BamHI/XhoI to remove osh4^Y97F followed by blunt-end ligation after treatment with Klenow. The pMET25-OSH4-RFP SacI/KpnI fragment was the subcloned into pRS414 forming pMET25-OSH4-RFP (pKK2108). Following this, the pMET25 promoter was removed by SacI/EcoRI digest, and replaced with a SacI/EcoRI fragment from plasmids containing an OSH4 promoter and an osh4 allele of interest (pKK1921, pKK1950 and pKK1988).

To construct pRS316-SUC2 (pKK2012), a SUC2 containing fragment extending 125 bp upstream and 447 bp downstream of the start of the coding sequence was amplified by PCR, using S. cerevisiae genomic DNA as a template and primers oKK319 and oKK320. This fragment was cloned into the SacI/XhoI sites of pRS316.

To construct pRS316-OSH6 (pKK2122), a SacI-AclI genomic fragment containing OSH6 (S. cerevisiae chromosome XI: 445047-447320) was cloned from pCB237 into SacI-ClaI digested pRS316. osh6 alleles were synthesized as gBlocks (Integrated DNA Technologies Inc., Coralville, IA). An AgeI-SnaBI osh6^L69D fragment was cloned into the AgeI and SnaBI sites of pKK2122, forming pKK2123. A SnaBI-XbaI osh6^H157A/H158A fragment was cloned into the SnaBI and XbaI sites of pKK2122, forming pKK2124. Each cloned gBlock was validated by sequencing.

The assay for invertase was performed as per Dighe and Kozminski (2008) with the following modification. After washout of synthetic medium containing 5% glucose, cells were grown in synthetic medium containing 0.1% glucose for 4 h at 25°C or 37°C prior to analysis. For the purpose of analyzing the kinetics of invertase exocytosis at early time points after the de-repression of invertase, cells were grown in minimal medium containing 0.1% glucose, beginning 30 min after shifting them from 25°C or 37°C. Accumulation of Bgl2p was assayed as described in Kozminski et al. (2006), following the method of Harsay and Schekman (2007).

The Lucifer Yellow (LY) assay was performed as per Beh and Rine (2004).

To assay the formation of assembled SNARE complexes we used the protocol by Grote et al. (2000). 30 OD_600nm units of cells were harvested and mixed with ice cold 10× TAF (Tris-azide-flouride; 100 mM Tris-Cl pH 7.5, 100 mM NaN₃, 100 mM NaF) and left on ice for 10 min. Cells were then harvested by centrifugation for 5 min at 1750 g at 4°C and then washed with TAF and centrifuged as before. These cells were resuspended in 1 ml IP buffer [50 mM HEPES pH 7.4, 150 mM KCl, 1 mM EDTA, 1 mM DTT and 0.5% (v/v) NP-40] and vortexed at 4°C in the presence of 425−600 μm diameter glass beads (G9268, Sigma, St. Louis, MO) thrice for 4 min each with a 1 min pause on ice between beatings. Lysates were clarified in a microfuge for 20 min at 18,800 g at 4°C. The protein concentration was measured at 280 nm. Samples were then normalized by dilution to a total protein concentration of 4 mg/ml with IP buffer. Lysates were then pre-cleared with unbound protein A-conjugated agarose (Pierce 20333, ThermoFisher Scientific, Waltham MA) for 1 h 4°C with gentle rocking. Following this, protein A-conjugated agarose blocked with 1 mg/ml BSA (Fraction V, A3059, Sigma, St. Louis, MO) and pre-bound to anti-HA monoclonal antibody (MMS-101p, Convance, Princeton, NJ) was added to the lysates and rocked gently overnight at 4°C. Anti-HA-bound protein A beads were collected by centrifugation (1000 g for 30 s at 4°C) and washed 5× with ice-cold IP buffer without dithiothreitol (DTT), then resuspended in SNARE pull-down sample buffer [60 mM Tris-Cl pH 6.8, 100 mM DTT, 2% (w/v) SDS, 100 mg/ml sucrose, 0.05% (w/v) Bromophenol Blue; Carr et al., 1999]. Samples were then boiled for 15 min and clarified for 1 min at 18,000g prior to SDS-PAGE.

The vesicle isolation assay was performed as per Alfaro et al. (2011).

To isolate PMs, the protocol of Alfaro et al. (2011) for vesicle isolation was used with the following modifications. The sample was loaded on top of a discontinuous sucrose gradient (1.10 M, 1.65 M and 2.25 M sucrose, with 0.8 M D-sorbitol) and centrifuged in a Beckman SW-41 rotor at 80,000 g for 14 h at 4°C. Successive layers of 3.5 ml of 2.25 M, 1.65 M and 1.1 M sucrose in Tris-EDTA (pH 7.5) buffer containing 0.8 M D-sorbitol formed the gradient (Panaretou and Piper, 2006). Fractions were taken from the 2.25 M /1.65 M interface and the 1.65 M/1.1 M interface, diluted 1:4 with Tris-EDTA pH 7.5 and pelleted for 40 min at 30,000 g in a Beckman TLA100.2 rotor prior to resuspension in 4×SDS-PAGE loading buffer [125 mM Tris-Cl pH 6.8, 1.43 M β-mercaptoethanol, 4% (w/v) SDS, 20% (v/v) glycerol, 0.005% (w/v) Bromophenol Blue]. Samples were then heated at 65°C for 10 min and clarified by centrifugation for 1 min at 18,000 g prior to SDS-PAGE. To determine the amount of total membrane, sucrose fractions were diluted ×4 with 25 mM Tris-Cl pH 7.5, after which 200 μg/ml FM4-64 (catalog no. T13320, Molecular Probes, Eugene, OR) diluted in DMSO was added to 18 μg/ml. After 1 h of gentle shaking at room temperature, fluorescence was measured using a PTI 814 photomultiplier detection system (515 nm/650 nm).

Samples were prepared, imaged and processed for TEM as per Dighe and Kozminski (2008). Images were processed and vesicle diameters measured using using Image J (NIH) to uniformly adjust brightness and contrast.

Sec4p was detected as per Orlando et al. (2011) with the following modification. After the initial fixation in 4.4% (v/v) formaldehyde (F79-500, ThermoFisher, Waltham, MA), the cells were fixed for an additional hour in 5 ml of buffered fixative [PBS, 2% glucose, 20 mM EGTA, 3.7% (v/v) formaldehyde (F79, ThermoFisher, Waltham, MA)]. Next, the cells were probed with anti-Sec4p monoclonal antibody [kind gift from P. Brennwald (University of North Carolina); Rossi and Brennwald, 2011] diluted 1:100 in PBS with 1 mg/ml BSA and then with FITC-conjugated anti-mouse IgG (catalog no. 715-095-150, Jackson ImmunoLabs, West Grove, PA) diluted 1:100 in PBS with 1 mg/ml BSA. Mounting medium consisted of PBS, 90% (v/v) glycerol and 1 mg/ml ρ-phenylenediamine (P-6001, Sigma, St. Louis, MO). Images were processed using Image J (NIH); brightness and contrast adjustments were uniformly applied.

Single-channel microscopy was performed on a Zeiss Axioplan 2 microscope with a 100× (Plan-Fluor, N.A. 1.45) objective with a Zeiss AxioCam MRM camera (Zeiss, Oberkochen, Germany). Dual-channel microscopy for colocalization was performed on a Zeiss Axiovert S.100 microscope with a 100× (Plan-Apochromat, N.A. 1.4) objective with a Hamamatsu EM-CCD (C9100 13) ImageM camera (Hamamatsu Photonics, Japan). Bleed through between channels was not detected. Images were processed using Image J (NIH); brightness and contrast adjustments were uniformly applied.

Polypeptides were separated using standard SDS-PAGE on 15% gels. Nitrocellulose membranes were probed with primary antibody and then probed with near-infrared fluorescent anti-rabbit or anti mouse secondary antibody (catalog no. 92632211 and catalog no. 926-32210, Licor, Lincoln, NE). The probed membranes were scanned on an Odyssey^® Infrared Imaging System (Licor, Lincoln, NE) and band intensity was measured using Image Studio™ (catalog no. 9202-500, Licor, Lincoln, NE). Primary antibodies used were anti-Pma1p (sc-57978, Santa Cruz Biotech, Dallas, TX), anti-Kex2p (ab34772, Abcam, Cambridge, UK), anti-Dpm1p (A-6429, ThermoFisher Scientific, Waltham, MA), anti-Pho8p (ab113688, Abcam, Cambridge, UK), anti-Sec4p and anti-Sso1/2p antibodies (kind gifts of P. Brennwald, University of North Carolina), anti-β-tubulin antibody (kind gift of A. Frankfurter, University of Virginia), anti-Bgl2p (Alfaro et al., 2011), anti-HA (901502, BioLegend, San Diego, CA) and anti-Osh4p (kind gift of C. Beh, Simon Fraser University).

Thanks to C. Beh (Simon Fraser University), P. Brennwald (University of North Carolina), R. Collins (Cornell University), Anthony Frankfurter (University of Virginia), J. Gerst (Weizmann Institute), and R. Schekman (University of California, Berkeley) for reagents; D. Schafer (University of Virginia) for use of her fluorimeter; I. Provencio and G. Bloom (University of Virginia) for use of their microscopes; and R. Deutscher, S. Dighe, J. McDaniels, and A. Norambuena (all University of Virginia) for technical assistance. Parts of this work were completed by R.J.S. in partial fulfilment of the requirements for the degree Doctor of Philosophy (University of Virginia).