Multicopy suppressor analysis of thermosensitive YIP1 alleles implicates GOT1 in transport from the ER

Intracellular transport between membrane-bound compartments of the early secretory pathway is highly dynamic, yet organellar identity is strictly maintained. Newly made secretory proteins and lipids are selected for anterograde transport from the endoplasmic reticulum (ER) to the Golgi complex while compartmental residents are retained and/or very efficiently retrieved. Organization of the early secretory pathway is known to depend on coat protein complexes, membrane targeting and fusion factors and a host of additional components that catalyze bidirectional transport between the ER and Golgi (Bonifacino and Glick, 2004). However, the overall mechanisms that govern transport and maintain organelle structure and function in the early secretory pathway are poorly understood.

The current view of ER export suggests that folded secretory proteins are collected at specialized zones in the ER, referred to as ER-exit sites. The COPII-budding machinery is also concentrated at ER-exit sites where cargo proteins are linked directly or indirectly to coat subunits for their incorporation into COPII-derived transport intermediates (Lee et al., 2004). The levels of specific secretory cargo and lipid species appear to influence ER export dynamics (Forster et al., 2006; Runz et al., 2006). However, it is not clear how vesicle budding is regulated at ER-exit sites under a variety of these conditions. We have identified additional protein components on COPII transport vesicles that cycle between ER and Golgi compartments and influence ER export (Otte et al., 2001). One such protein, Yip1p, is efficiently packaged into vesicles and required for COPII budding (Heidtman et al., 2003).

Yip1p was initially identified as a Ypt/Rab-interacting protein and localizes to ER and Golgi membranes (Yang et al., 1998). Yip1p forms a heteromeric integral membrane complex with two other membrane proteins, Yif1p and Yos1p (Matern et al., 2000; Heidtman et al., 2005). All three members of the Yip1 complex are conserved in nature and both Yip1p and Yif1p share sequence identity with one another and a larger family of polytopic membrane proteins, known as the Yip1 family (Calero et al., 2002). Our studies have shown that the Yip1 complex acts in a later stage of vesicle budding, most likely after assembly of the COPII coat. We propose the Yip1 complex regulates COPII budding at the membrane scission stage (Heidtman et al., 2005). Yip1 family members might act similarly in other coat-dependent budding events at distinct intracellular compartments. To gain further insight on Yip1 complex function, we isolated genetic suppressors of the semi-functional yip1-2 mutation. Three multicopy suppressor genes were identified, all of which might in some manner be connected to membrane structure and function in the early secretory pathway. The strongest suppressor isolated encodes Got1p, a 16 kDa tetraspanning membrane protein implicated in Golgi transport. Based on further investigation in this study we propose that Got1p influences trafficking from the ER.

Recent biochemical, genetic and morphological studies have shown that the Yip1p complex has a crucial role in COPII vesicle biogenesis (Heidtman et al., 2003; Heidtman et al., 2005). To identify other gene products that function in the same pathway as this complex, we performed a multicopy suppression screen of the themosensitive yip1-2 strain (Yang et al., 1998). This strain contains a single point mutation (G175E) in the second predicted transmembrane domain of Yip1p, which results in growth defects and an accumulation of secretory proteins in the ER at restrictive temperatures (Yang et al., 1998). A high-copy genomic YEp24 library bearing the URA3 marker (Carlson and Botstein, 1982) was used to select for overexpressed genes that could rescue the yip1-2 growth defect at a restrictive temperature of 36°C. From a pool of ∼10,000 transformants, we identified 13 plasmids that suppressed the yip1-2 mutation upon retransformation. DNA sequencing identified the genomic insert present in each of the suppressing plasmids. Ten of these plasmids contained a region of DNA harboring the YIP1 open reading frame (ORF). The remaining three plasmids contained regions of DNA corresponding to chromosomes II, VII and XIII respectively.

The suppressing plasmid bearing a region of chromosome VII had only one complete ORF present: FYV8 (function required for yeast viability upon toxin exposure). FYV8 is a non-essential gene; however, strains lacking FYV8 exhibit hypersensitivity to K1 killer toxin (Page et al., 2003). This gene is predicted to encode a ∼90 kDa protein with moderate homology to predicted open reading frames in higher eukaryotes and might be involved in the response to ER stress (Chen et al., 2005). To more closely examine the level of FYV8 high-copy suppression activity, we performed dilution series experiments to monitor growth rates. As shown in Fig. 1A, FYV8 displayed mild but detectable suppression of the yip1-2 strain at 36°C. Multicopy FYV8 did not suppress the thermosensitive growth defects of other yip1 or yif1 conditional mutants at the restrictive temperature (data not shown). The suppression activity of FYV8 towards the yip1-2 strain was not investigated further.

The suppressing plasmid containing a region of chromosome II also had only a single complete ORF present: TSC3 (temperature-sensitive suppressor of Ca²⁺ sensitivity). The TSC3 gene encodes a ∼9 kDa transmembrane protein that localizes to ER membranes. Several lines of evidence indicate that Tsc3p participates in the early steps of sphingolipid synthesis at the ER (Gable et al., 2000). As shown in Fig. 1A, TSC3 moderately suppressed the growth defect of the yip1-2 strain at 36°C. Multicopy TSC3 did not suppress the thermosensitive growth defects of other yip1 or yif1 conditional mutants (data not shown). The mechanism by which multicopy TSC3 suppressed the yip1-2 mutation was not investigated.

The third suppressor plasmid isolated from this screen contained a region of DNA from chromosome XIII. Several complete ORFs were present in this region, including YMR291W, GOT1, YMR293C and JNM1. Subcloning revealed that GOT1 was solely responsible for the suppression of the thermosensitive growth defect of the yip1-2 strain (Fig. 1B). GOT1 (Golgi transport) is a non-essential gene encoding a tetra-spanning integral membrane protein of ∼16 kDa and was originally identified in a synthetic-lethal screen designed to isolate mutants that require the SFT2 gene for function (Conchon et al., 1999). Both Got1p and Sft2p are tetra-spanning membrane proteins that are conserved across species. Double-mutant strains that rely on conditional got1-2 or sft2-2 alleles for growth exhibit secretion defects in pulse-chase assays and accumulate ER membranes indicating a role for these proteins in the early secretory pathway (Conchon et al., 1999). As shown in Fig. 1B, we found that multicopy GOT1 strongly suppressed the growth defect of the yip1-2 strain at 36°C to the same extent as multicopy YIP1 or YIF1. Furthermore, multicopy GOT1 exhibited different degrees of suppression toward the yip1-1 (P114, G129E), yip1-4 (E70K), yif1-2 (W133Amber) and yif1-4 (V162A, S189P, L205A, V230A) mutants at their restrictive temperatures (Fig. 2; and data not shown). We also tested a conditional allele in YOS1, which encodes an additional subunit of the Yip1-Yif1 complex (Heidtman et al., 2005), and observed that multicopy GOT1 did not suppress the growth defect of a yos1-1 (N18A) strain (Table 1).

Got1p and Sft2p are proposed to have partially redundant roles in intracellular trafficking with Got1p acting in the early secretory pathway and Sft2p functioning at the late Golgi (Conchon et al., 1999). Although multicopy SFT2 was not isolated in our suppressor screen, we wanted to test directly whether a functional 2 μm version of SFT2 displayed suppressor activity toward the yip1 or yif1 conditional alleles. We observed that multicopy SFT2 was unable to suppress the growth defects associated with any of the yip1 or yif1 mutant strains tested (supplementary material Fig. S1). These results suggest that the strong suppressor activity of GOT1 toward these yip1 and yif1 alleles was relatively specific.

In order to determine if multicopy GOT1 is a specific suppressor of Yip1 complex mutants, or more generally could suppress other ER-Golgi trafficking mutants, we transformed several thermosensitive strains with multicopy GOT1 or an empty vector control. For each mutant analyzed, two independent transformants were grown to saturation in the appropriate selective media to maintain plasmids, then tenfold dilution series were spotted onto YPD plates and growth monitored at 30°C, 34°C and 38°C. In addition to yip1-4 and yos1-1 mutants, we tested a total of 20 mutant strains with deficiencies in the stages of ER vesicle budding, vesicle tethering, membrane fusion or retrograde transport. Interestingly, multicopy GOT1 moderately suppressed the growth defects of sec31-1 and sec23-1 genes involved in vesicle budding, and weakly suppressed the growth defects of sec24-11, uso1-1 and bet1-1 (Table 1; supplementary material Fig. S2).

Conchon and colleagues (Conchon et al., 1999) localized a C-terminally tagged 3MYC version of Got1p to cis-Golgi membranes by fluorescence microscopy and subcellular fractionation approaches. Interestingly, a more recent study localized a C-terminally GFP-tagged version of Got1p to the ER (Huh et al., 2003). These results, although conflicting, could be consistent with our genetic data, and suggested that Got1p might cycle between the ER and Golgi compartments. To investigate this hypothesis, we prepared polyclonal antibodies specific for Got1p to detect the endogenous untagged protein. For localization studies, cell lysates were prepared from wild-type and got1Δ strains to monitor the subcellular distribution of Got1p on sucrose gradients. Fractions were collected and analyzed for Got1p, the early Golgi marker Och1p, the ER marker Sec61p, and the COPII vesicle protein Yip1p. We observed that ∼66% of Got1p cosedimented with Sec61p and ∼34% with Och1p (Fig. 3A). This fractionation pattern was similar to that of Yip1p and other vesicle proteins that cycle between the ER and Golgi compartments (Cao and Barlowe, 2000; Heidtman et al., 2005; Otte et al., 2001; Powers and Barlowe, 1998; Schimmoller et al., 1995). No differences were observed when the sucrose gradient fractionation patterns for Yip1p, Sec61p and Och1p were compared in wild type and got1Δ strains (data not shown), indicating that GOT1 is not required for Yip1p localization.

To confirm dynamic cycling of Got1p in vivo, we analyzed the distribution of Got1p after a sec12-4 block. Proteins that cycle between the ER and Golgi compartments accumulate in the ER when export from this compartment is blocked by shifting a sec12-4 strain to the restrictive temperature (Schroder et al., 1995). Wild-type and sec12-4 cells in logarithmic-phase growth were shifted to 37°C for 45 minutes and membrane organelles resolved by differential centrifugation of cell lysates. The P13 fraction (enriched in ER membranes) and the P100 fraction (enriched in Golgi membranes) were assessed by immunoblot (Fig. 3B). We observed that the distribution of Och1p and Sec61p was not affected by the sec12-4 block. However, Got1p and the known COPII vesicle proteins Erv41p and Erv25p, accumulated in the ER fraction of sec12-4 cells. In addition, a got1Δ strain and isogenic wild-type were examined under the temperature shift and no alterations in protein distribution were observed. Taken together, these data indicate that Got1p cycles between the ER and Golgi compartments and that got1Δ does not influence the normal localization pattern of the set of proteins monitored.

Our localization data and genetic interaction analysis suggested that Got1p would be packaged into COPII vesicles. To test this idea, microsomes from wild-type and got1Δ strains were incubated in the presence (+) or absence (–) of COPII proteins with an energy-regeneration system to reconstitute budding. Membrane vesicles generated in each condition were isolated and analyzed by immunoblot as described (Belden and Barlowe, 1996). Relative levels of individual proteins were quantified by densitometry analysis of immunoblots using ImageJ (Rasband, 1997-2005), and packaging efficiencies were calculated. As shown in Fig. 4, the ER-resident Sec12p was not efficiently packaged (1%) into COPII vesicles, compared with efficient packaging of Erv41p (37%) and Erv25p (25%). Notably, we found that Got1p was also very efficiently packaged (62%) into COPII vesicles. This level of packaging indicates that Got1p probably contains a sorting signal recognized by the COPII coat or might associate with other factors that contain potent ER-export signals.

To determine whether the got1Δ mutation influences the sorting efficiency of other COPII vesicle proteins, the composition of vesicles produced from wild-type and got1Δ microsomes were compared by immunoblot. As seen in Fig. 4, the packaging efficiencies of Erv41p and Erv25p were not influenced by got1Δ mutation. In addition, no changes were detected in the packaging of other characterized ER vesicle proteins including Yip1p, Yif1p, Erv26p, Sed5p, Bet1p, Bos1p and Sec22p (data not shown). Moreover, ER-resident proteins such as Sec12p (Fig. 4), Sec61p and Kar2p (not shown) were not inappropriately packaged into COPII vesicles when microsomes were prepared from got1Δ cells. These data suggest that Got1p is not required for protein sorting at the ER, at least for the subset of proteins we examined.

We observed that multicopy GOT1 could not bypass a complete yip1Δ deletion (supplementary material Fig. S3), indicating that Got1p does not replace Yip1p function but somehow compensates for semi-functional yip1 and yif1 alleles. To explore this compensation mechanism, we tested whether elevated levels of Got1p could rescue the in vitro budding defect observed in membranes prepared from yip1-4 mutant cells. Semi-intact cell membranes prepared from wild-type and yip1-4 cells bearing empty vector or multicopy GOT1 were used in a cell free assay that measures COPII dependent vesicle budding of [³⁵S]glyco-pro-α-factor (gpαf). As shown in Fig. 5A, mutant yip1-4 membranes displayed a significant budding defect compared with wild-type membranes, which is consistent with our previous results (Heidtman et al., 2003). Interestingly, we observed that multicopy GOT1 partially rescued the budding defect of yip1-4 membranes, similarly to that observed for multicopy YOS1 rescue of the yip1-4 budding defect (Heidtman et al., 2005). Taken together, these results indicate that elevated levels of Got1p can partially compensate for reduced Yip1p activity in this COPII vesicle budding assay.

We next tested the influence of got1Δ deletion mutation on other stages in ER-Golgi transport. Previous studies indicated that membranes lacking Got1p budded COPII vesicles containing [³⁵S]gpαf at normal levels but were compromised for vesicle fusion with Golgi membranes (Conchon et al., 1999). However, our investigation of the got1Δ strain from the Research Genetics strain collection (Winzeler et al., 1999) revealed that membranes prepared from this strain did not display detectable defects in any specific transport stages or in overall ER-Golgi transport (Fig. 5B,C). Upon re-analysis of the original got1Δ strain (Conchon et al., 1999), we did observe phenotypic consequences; however, these defects could not be restored by reintroduction of the GOT1 gene. Closer examination of membranes isolated from this original got1Δ strain also revealed that the Rab-GTPases Ypt1p and Ypt7p were upregulated and predominantly found in the soluble fraction instead of associated with membranes; again, this defect was not rescued by transformation with multicopy GOT1 (supplementary material Fig. S4). These results suggest that the original got1Δ strain contains additional mutation(s) or a genetic background that contributes to the observed trafficking defects. Further studies will be required to determine the nature of these genetic differences.

We next took a biochemical approach to determine whether other proteins could be detected in stable association with Got1p. Our genetic analyses suggested that Got1p might physically associate with the Yip1 complex or other proteins involved in vesicle budding. To investigate this, we prepared microsomes from a strain expressing Yif1p-3HA as the sole source of cellular Yif1p (Heidtman et al., 2005). Microsomes from this strain were solubilized with 0.5% digitonin and Yif1p-3HA immunoprecipitated with monoclonal anti-HA. As seen in Fig. 6A, Yif1p-3HA was efficiently recovered, and Yip1p and Yos1p – both members of the Yip1 complex – were specifically co-immunoprecipitated (Heidtman et al., 2005; Matern et al., 2000). When the same immunoprecipitates were examined for the presence of Got1p using polyclonal anti-Got1p antibodies, no Got1p was detected. Based on this result we conclude that Got1p is not a stable member of the Yip1 complex.

In a reciprocal approach, we immunoprecipitated Got1p to identify potential binding partners. Using a CEN plasmid expressing Got1p-3MYC from the TPI promoter (Conchon et al., 1999), we transformed the Yif1p-3HA strain to express a tagged version of Got1p. Microsomes were prepared from these strains and then solubilized with 1% Triton X-100. The solubilized material was subjected to immunoprecipitation with monoclonal anti-Myc antibodies. As shown in Fig. 6B, Got1p-3MYC was efficiently recovered in this manner as determined by immunoblotting with anti-Myc and polyclonal anti-Got1p antibodies. Interestingly, we found that endogenous Got1p was efficiently co-immunoprecipitated with Got1p-3MYC, indicative of a homo-oligomeric association. Yif1p-3HA, Sec23p and Sec61p did not co-immunoprecipitate with Got1p-3MYC, confirming the specificity of the Got1p interaction. Further analysis of these immunoprecipitates failed to detect Sec12p, Sec13p, Sec24p, Yip1p, Erv41p, Uso1p, Sec22p, Bet1p, Bos1p or Sed5p in association with Got1p-3MYC (data not shown). Moreover, analysis of Got1p-3MYC immunoprecipitates on stained protein gels did not reveal other binding partners beyond endogenous Got1p. Similar results were obtained when immunoprecipitation experiments were performed in 0.5% digitonin in place of 1% Triton X-100 (data not shown).

To further characterize the nature of the Got1p oligomer, we determined its relative size by sucrose gradient sedimentation analysis. A post-nuclear fraction of homogenized cells was solubilized in 1% Triton X-100 and proteins resolved on a linear 1-12% sucrose velocity gradient containing 1% Triton X-100. Soluble globular molecular weight markers (RNAseA: 13.7 kDa; Aldolase: 158 kDa; and Catalase: 232 kDa) were run in parallel gradients. Erv25p was used as a control for the resolution of integral membrane protein complexes because Erv25p is known to form a ∼100 kDa complex with Emp24p, Erp1p and Erp2p (Marzioch et al., 1999). Immunoblot analysis of sucrose gradients (Fig. 6C) showed that Got1p peaked between fractions 2 and 3 at a size slightly greater than Sec22p (∼25 kDa). Taken together, our results suggest that Got1p most likely forms a low molecular weight homo-dimeric complex.

We have shown that ∼tenfold overexpression of GOT1 from a 2 μm plasmid (supplementary material Fig. S4) suppresses thermosensitive mutations in essential genes required for COPII vesicle budding (Table 1). To investigate dosage effects of GOT1 expression, we placed GOT1 under control of the inducible GAL1 promoter. In the case of GAL1-3HA-GOT1 expression, we observed that increases in galactose concentration in growth medium significantly reduced growth rates compared with wild-type strains. Indeed, strains containing GAL1-3HA-GOT1 did not grow on medium with 2% galactose. To explore the cause of this toxicity, we precultured cells in 1% glucose, 1% galactose and then shifted cultures to 2% galactose. After 1.5 hours, a reduction in growth rate was detected and after 5 hours, growth on galactose medium was strongly inhibited (Fig. 7A). At the 5 hour time-point, we monitored secretory protein transport and organelle morphology. As seen in Fig. 7B, the ER form of carboxypeptidase Y (p1 CPY) accumulated in the GAL1-regulated strain compared with the wild type. Precursor accumulation correlated with an approximate 40-fold increase in 3HA-Got1p expression over endogenous levels. This phenotype suggested that GAL1-3HA-GOT1 overexpression produces a block in ER-Golgi transport of secretory proteins.

To inspect organelle morphology, we monitored the ER marker Sec63-GFP (Prinz et al., 2000), and the late-Golgi marker Sec7-GFP (Rossanese et al., 2001) by microscopy after growth in 2% galactose for 5 hours (Fig. 8). ER morphology was assessed by focusing on the center of cells, where wild-type cells exhibited a typical perinuclear and peripheral ER morphology (Fig. 8A, WT) (Prinz et al., 2000). GAL1 overexpression of HA-Got1p produced an elaboration of ER membranes, which was accompanied in a few cases by gapped ER structures (Fig. 8A, GAL1-3HA-GOT1). After assessing ER structures in a population of cells under each condition, we determined that 60% of the GAL1-3HA-GOT1 cells displayed abnormal ER morphologies whereas only 20% of the control cells contained such structures (supplementary material Fig. S5A). Previous reports have shown that yeast mutants blocked in ER export, such as sec23-1 (Prinz et al., 2000) and yip1-4 (Heidtman et al., 2003), accumulate similar ER structures at restrictive temperatures.

Golgi membranes were assessed by monitoring the late-Golgi marker Sec7-GFP, which localizes to punctate structures throughout wild-type cells and displays minimal background GFP staining (Fig. 8B: WT) (Seron et al., 1998; Rossanese et al., 2001). By contrast, GAL1-3HA-GOT1 cells exhibited a diffuse Sec7-GFP localization pattern with punctate structures reduced or absent from cells (Fig. 8B: GAL1-3HA-GOT1). Quantification of these phenotypes indicated that 95% of the GAL1-3HA-GOT1 cells displayed abnormal Sec7-GFP distributions compared with 20% of cells in the wild-type culture (supplementary material Fig. S5B). Similar results were obtained from cells expressing the early-Golgi marker Sec21p-GFP (data not shown). Loss of early- and late-Golgi structures has been reported in yeast mutants under conditions that block ER export as well as in mutants that block vesicle fusion stages with Golgi membranes (Wooding and Pelham, 1998; Sato and Nakano, 2002; Kamena et al., 2008). This dispersal of Golgi membranes is thought to be caused by a reduced delivery of ER derived cargo to the Golgi. In summary, our observations that GAL1-3HA-GOT1 alters ER and Golgi morphologies, and causes an accumulation of the ER form of CPY, indicate high levels of 3HA-Got1p inhibit transport through the early secretory pathway. These results are consistent with a block at the ER-export stage. Alternatively, these high levels of HA-Got1p could influence subsequent stages in transport to the Golgi complex.

Previous studies implicated Yip1p and Yif1p in ER-Golgi trafficking (Yang et al., 1998; Matern et al., 2000) and more recent work has demonstrated a requirement for the Yip1 complex in COPII vesicle budding (Heidtman et al., 2003; Heidtman et al., 2005). However, the molecular mechanism by which the Yip1 complex contributes to this process is unclear. To gain a better understanding of the role of this complex in COPII vesicle budding, we performed a multicopy suppressor screen of the thermosensitive yip1-2 strain, which contains a single point mutation (G175E) in the second predicted transmembrane domain of Yip1p (Yang et al., 1998). Three multicopy suppressors of yip1-2 were identified (FYV8, TSC3 and GOT1) and the properties of GOT1 were further characterized.

The FYV8 multicopy plasmid displayed modest suppressor activity toward yip1-2. FYV8 was first identified in a screen for deletion mutants with altered sensitivity to K1 killer toxin and fyv8Δ cells are hypersensitive to K1 killer toxin, calcofluor white, hygromycin B, and SDS (Page et al., 2003). The cellular function of Fyv8p remains unclear, although sequence-profiling methods predict a function as a transcriptional regulator (Mi et al., 2005). Recent studies suggest a role in the ER-stress response, because fyv8Δ cells display increased sensitivity to tunicamycin and reducing agents (Chen et al., 2005). Other studies have implicated FYV8 in chromosome segregation (Baetz et al., 2004) and chromosome stability (Smith et al., 2004). Based on the collective findings, we speculate that multicopy FYV8 upregulates genes involved in managing ER stress and this might allow yip1-4 mutant cells to cope with a deficit in Yip1p activity.

TSC3 was initially identified in a temperature-sensitive suppressor screen of Ca²⁺-sensitive csg2Δ mutants (Beeler et al., 1998). Further investigation revealed that Tsc3p localizes to the ER where it associates with Lcb2p and is a subunit of the serine palmitoyltransferase (SPT), which catalyzes the first committed step in sphingolipid synthesis (Gable et al., 2000; Monaghan et al., 2002). Tsc3p interaction with SPT stimulates transferase activity several fold (Gable et al., 2000) by conferring substrate specificity to the enzyme (Cowart and Hannun, 2007). Interestingly, a recent study suggested a possible synthetic interaction between TSC3 and RUD3 (Schuldiner et al., 2005), which might indicate a role for sphingolipids in ER-Golgi trafficking. The identification of TSC3 as a dosage suppressor of the yip1-2 strain could be due to an increased concentration of sphingolipids at the ER caused by high levels of Tsc3p. We have proposed that the Yip1 complex acts in membrane dynamics through promotion of COPII vesicle scission (Heidtman et al., 2005). If increased sphingolipid levels at the ER rescues impaired activity of the yip1-2 mutant, this would suggest that steps in the COPII-budding process are influenced by both membrane lipid composition and Yip1 complex activity. One such possible role in budding might be to mediate membrane bending at the bud-neck to resolve membrane buds into vesicles.

GOT1 displayed the strongest multicopy suppressor activity toward yip1-2. GOT1 was originally identified in a synthetic lethal screen with sft2Δ, a multicopy suppressor of thermosensitive alleles of sed5 (Banfield et al., 1995; Conchon et al., 1999). Both Got1p and Sft2p are small tetra-spanning membrane proteins that are conserved across species and have been proposed to have a role in facilitating Sed5-dependent fusion events with the Golgi at early- and late-Golgi compartments, respectively (Conchon et al., 1999). Furthermore, both Got1p and Sft2p have been proposed to act downstream of trans-SNARE complex formation in promoting membrane fusion pores (Bayer et al., 2003). However, we observed that the multicopy suppressor activity displayed by GOT1 was distinct since multicopy SFT2 did not suppress temperature sensitive yip1 or yif1 mutants. These findings led us to further explore the role of Got1p in ER-to-Golgi transport and to gain a better understanding of its relationship with the Yip1 complex.

Our findings were unexpected. We found that in addition to yip1-2, multicopy GOT1 was a strong suppressor of other thermosensitive alleles of yip1 and yif1, a moderate suppressor of sec31-1 and sec23-1, and weak suppressor of sec24-11, bet1-1 and uso1-1. This was a surprising finding in light of the fact that Got1p was initially reported to act in membrane fusion (Conchon et al., 1999). A relative specificity to the COPII-budding stage is supported by the fact that thermosensitive mutations in 11 other genes involved in vesicle tethering, fusion and retrograde transport (COPI subunits) were not suppressed by multicopy GOT1. This suppression pattern suggested that Got1p physically interacts with components of the COPII budding machinery; however, under the conditions of our co-immunoprecipitation experiments we failed to detect such interactions, including any association with the Yip1 complex. We also observed that multicopy GOT1 could not bypass a complete yip1Δ deletion, indicating that Got1p activity was not redundant with Yip1p. However, multicopy GOT1 was able to partially rescue the budding defect of yip1-4 membranes suggesting that overexpression imparted membrane properties that compensate for deficiencies in Yip1 complex activity.

Consistent with previous reports on the localization of Got1p (Conchon et al., 1999; Huh et al., 2003), we found that Got1p was very efficiently packaged into COPII vesicles and cycled between the ER and Golgi compartments. In contrast to previous findings, however, we observed that got1Δ membranes prepared from the Research Genetics isolate did not display defects in cell free assays that monitored vesicle budding, tethering or fusion. Immunoblot analysis of this strain confirmed that Got1p was not expressed. Moreover, addition of anti-Got1p immunoglobulins or anti-HA antibodies to GOT1-3HA membranes did not inhibit in vitro transport even in a sft2Δ background (data not shown). These results suggested that Got1p does not function in ER-to-Golgi transport as initially proposed (Conchon et al., 1999). We found that the original got1Δ isolate appeared to contain mutation(s) that cause upregulation and mislocalization of the Rab GTPases Ypt1p and Ypt7p, which might explain the previously observed defect in fusion of COPII vesicles with Golgi membranes (Conchon et al., 1999). This defect was absent from the Research Genetics got1Δ strain. Together, these results indicate that Got1p influences transport though the early secretory pathway but under our in vitro conditions this function is not directly required for a single round of ER-to-Golgi transport.

A study in mammalian cells reported that overexpression of an N-terminal GFP-tagged allele of the mammalian Got1p homolog causes a delay in surface expression of VSV-G concomitant with defects in Golgi integrity (Starkuviene et al., 2004). In agreement with these findings, we observed that overexpression of 3HA-GOT1 from the GAL1 promoter resulted in ER-to-Golgi transport defects accompanied by growth arrest and altered ER and Golgi morphologies. These findings suggest that excessive levels of Got1p may titrate out an essential component of the ER-to-Golgi transport machinery. Although no clear COPII- or COPI-packaging signals were detected, Got1p was very efficiently packaged into COPII vesicles and it seems plausible that GAL1 overexpression of Got1p could saturate or somehow interfere with the normal cargo recognition functions of the Sec23-Sec24 complex (Miller et al., 2003).

GOT1 has also been identified as a multicopy suppressor of the thermosensitive ypk1-1/ykr2Δ strain. Ypk1p and Ykr2p are functionally redundant protein kinases proposed to function in the maintenance of cell wall integrity (Roelants et al., 2002). Interestingly, phytosphingosine (the product of the third step in sphingolipid biosynthesis) is an activator of Pkh1p and Pkh2p kinases, which in turn activate Ypk1 and Ykr2 kinases during the heat stress response (Dickson et al., 2006; Roelants et al., 2002). Although Got1p localizes primarily to the early secretory pathway, we suggest that this rescue may be caused by an alteration in delivery of plasma membrane lipids (possibly sphingolipids) involved in this signaling cascade.

These collective genetic findings seem to connect Got1p to a function in regulation of membrane composition. We also note that sequence-profiling programs indicate that Got1p contains a lipase-related domain found in other proteins and species (Mi et al., 2005). Based on these observations, we propose that Got1p influences membrane composition in a manner that can compensate for decreases in certain activities required for ER and/or Golgi transport. In the case of reduced Yip1 complex activity, elevated Got1p activity could change the ER lipid makeup in a way that facilitates COPII vesicle budding. Our proposal predicts that the concentration of specific lipid species in ER membranes will be altered depending on Got1p expression level. A preliminary analysis of phospholipid species from whole cells indicated alterations in acyl chain length and degree of saturation when got1Δ cells were compared with wild-type cells (unpublished). However, future experimentation will be necessary to fully assess lipid composition in different subcellular compartments when isolated from cells that have varying levels of Got1p.

Yeast strains used in this study are listed in supplementary material Table S1. Unless noted otherwise, cultures were grown at 25°C (for mutant stains) or 30°C (for wild-type strains) in rich medium (YPD: 1% Bacto yeast extract, 2% Bacto peptone, 2% dextrose), or in minimal medium (YMD: 0.7% USBiological yeast nitrogen base without amino acids, 2% dextrose and appropriate amino acid supplements). Standard yeast (Sherman, 1991) and bacterial (Ausubel et al., 1987) molecular biology methods were used. Yeast cells were transformed by the lithium acetate method (Elble, 1992). Plasmids pRS426-FYV8, pRS426-TSC3, pRS425-GOT1, pRS426-GOT1, pRS426-SFT2, pRS426-YIP1 and pRS426-YIF1 were generated by amplification of genomic DNA obtained from FY834 (primers available upon request). The GAL1-3HA-GOT1 strain was constructed in CBY740 as described (Longtine et al., 1998). The KAN^RMX6 marker of CBY1574 was exchanged to NAT^RMX4 (Tong et al., 2001) and then mated with CBY2087 to generate CBY1769. Wild-type (CBY740) and GAL1-3HA-GOT1 (CBY2577) strains were transformed pJK59 [CEN SEC63-GFP URA3] (Prinz et al., 2000), or a chromosomal GFP tag was introduced into the SEC7 locus as described (Rossanese et al., 2001) to generate CBY2537, CBY2542, CBY2538 and CBY2543, respectively.

Polyclonal antibodies were raised against a GST-Got1p fusion protein with the cytosolic C-terminus of Got1p (amino acid positions 109-138) expressed from plasmid pGEX-2T. The fusion protein was purified according to the manufacturer's specifications (Amersham Biosciences) and used to immunize rabbits by standard procedures (Covance). For western blots, anti-Got1p serum was diluted 1:500. Antibodies against CPY (Rothblatt et al., 1989), Sec61p (Stirling et al., 1992), Kar2p (Brodsky and Schekman, 1993), Sec12p (Powers and Barlowe, 1998), Sec23p (Hicke and Schekman, 1989), Sec24p (Hicke et al., 1992), Sec13p (Salama et al., 1993), Och1p, Erv41p (Otte et al., 2001), Yip1p (Heidtman et al., 2003), Yif1p (Matern et al., 2000), Yos1p (Heidtman et al., 2005), Erv25p (Belden and Barlowe, 1996), Erv26p (Bue et al., 2006), Ypt1p (Rexach et al., 1994), Ypt7p (Haas et al., 1995), Gdi1p (Garrett et al., 1994), Uso1p (Ballew et al., 2005), Sec22p (Liu and Barlowe, 2002), Bet1p, Bos1p (Sogaard et al., 1994), Sed5p (Cao et al., 1998) and actin (Eitzen et al., 2002) have been described. Monoclonal anti-HA (Sigma, clone HA-7) and anti-cMyc (Covance, 9E10) were used. Western blot analysis was performed with nitrocellulose membranes or PVDF membranes (for detection of Yos1p and Got1p), using the SuperSignal West Pico chemiluminiscent substrate (Pierce Chemical) and developed on film or with a UVP Bioimaging System.

The multicopy suppression screen of yip1-2 (strain YXY12α) was performed as described (Heidtman et al., 2005). Briefly, ∼10,000 transformants from a yeast genomic YEp24 library (Carlson and Botstein, 1982) were grown on YMD-Ura plates at 25°C. Colonies were then replica plated to YPD plates and incubated for 48 hours at 36°C. Eighteen temperature resistant transformants were transferred to YMD-Ura plates, plasmid DNA isolated and retransformed into yip1-2 to test for plasmid-linked suppression. Of the 18 isolated plasmid preparations, 13 conferred varying levels of suppression upon retransformation. Plasmids inserts were sequenced with the primers YEpF1 and YEpR3 (Heidtman et al., 2005).

Yeast semi-intact cell membranes from wild-type and mutant strains were prepared as described (Baker et al., 1988). Vesicle budding, tethering and fusion assays using [³⁵S]gpαf were performed as described (Barlowe, 1997; Cao et al., 1998). Data points are the average of duplicate determinations and the error bars represent the range. To measure protein packaging into COPII vesicles, microsomes were prepared (Wuestehube and Schekman, 1992), and preparative scale budding reactions performed as described (Otte et al., 2001).

ER (P13) and Golgi (P100) membrane fractions were collected by differential centrifugation of cell lysates prepared from wild-type and sec12-4 strains as described (Belden and Barlowe, 2001). Membrane pellets were resuspended with 0.05 ml of SDS-PAGE sample buffer and heated for 10 minutes at 37°C. Sucrose gradient fractionation of membrane organelles was performed as described (Powers and Barlowe, 1998). Cell extracts were loaded on top of 11-step 18-60% (wt/vol.) sucrose gradients and centrifuged at 164,000 × g (SW40 rotor, Beckman Instruments, Palo Alto, CA) for 3 hours at 4°C. Eighteen 0.65 ml fractions were collected from the top of the gradient and analyzed by immunoblot. Relative protein levels in each fraction were quantified by densitometry using ImageJ (Rasband, 1997-2005). The plotted optical density (OD) per fraction was calculated by dividing the OD of the fraction over the total sum of ODs throughout the gradient for that particular protein.

Velocity gradient sedimentation of detergent solubilized membrane proteins was performed at 4°C (Price et al., 2000) with minor modification. Briefly, spheroplasts in lysis buffer (Baker et al., 1988) were disrupted in a Dounce homogenizer, cleared (1500 × g, 5 minutes) and membranes sedimented (20,000 × g, 5 minutes). Pellets were resuspended with 1 ml of gradient buffer (GB: 25 mM HEPES pH 7.0, 100 mM KOAc) containing 1 mM DTT and membranes solubilized with addition of an equal volume of 2% Triton X-100 prepared in GB. Solubilized extracts were cleared (100,000 × g, 10 minutes) and 1.4 ml supernatant fluid was loaded on top of a 10.4 ml 1-12% continuous sucrose gradient prepared in GB with 1% Triton X-100. Molecular mass markers (beef pancreas ribonuclease A, rabbit muscle aldolase, and beef liver catalase; Amersham Biosciences) were loaded on parallel gradients, and all were centrifuged at 175,000 × g (SW41 rotor, Beckman Instruments) for 24 hours. Twelve fractions were collected from the top of the gradient and the last fraction was used to resuspend the pellet. Relative levels of proteins and sucrose concentration of individual fractions were determined as above.

For immunoprecipitation of Yif1p-3HA, 0.3 ml microsomes (∼1 mg/ml membrane protein) were solubilized in an equal volume of 0.5% digitonin in buffer 88-8 (Kuehn et al., 1998) in the presence of 10 mM PMSF and 5 mM EDTA. After centrifugation at 20,000 × g for 4 minutes at room temperature, the supernatant fluid (∼0.5 ml) was transferred to a fresh tube on ice. The solubilized material was diluted with 1.5 volumes of 0.05% digitonin/buffer 88-8 and proteins immunoprecipitated for 2 hours at 4°C by addition of 2 μg anti-HA antibody and 25 μl of 20% protein-A-Sepharose beads (Amersham Biosciences). After washing beads, bound protein was released by addition of 0.03 ml of SDS-PAGE sample buffer and heated at 75°C for 3 minutes. For immunoprecipitation of Got1p-3MYC, microsomes as above were solubilized on ice for 20 minutes with IP-buffer (150 mM NaCl, 15 mM Tris-HCl, pH 7.5, 1% Triton X-100, 2 mM NaN₃) in the presence of 2.5 mM EDTA, 5 mM PMSF, 0.5 μM Leupeptin and 2.6 μM Pepstatin A. Samples were processed as above and supernatants diluted 2.6-fold with IP buffer to a final volume of 1.2 ml. Anti-Myc antibody (4 μg) and protein-A-Sepharose beads (30 μl of 20% solution) were added and proteins immunoprecipitated for 3 hours at 4°C. After washes, bound protein was released from the beads by the addition of 25 μl of SDS-PAGE sample buffer and heated at 37°C for 10 minutes.

Fluorescent and bright-field images were acquired using a microscope (model BX51; Olympus) equipped with a 100 W mercury arc lamp, Plan Apochromat 60× objective (1.4 NA) and a Sensicam QE CCD camera (Cooke). The camera and microscope were controlled by the IP Lab system (Scanalytics). All images were processed in Openlab software (Improvision, Lexington, MA). To quantify ER morphologies, cells expressing Sec63-GFP were scored for the appearance of elaborated and discontinuous or gapped ER structures within focal planes near the center of cells. To assess Golgi morphologies, cells expressing Sec7-GFP were monitored for the appearance of diffuse fluorescence distributions and loss of punctate structures.