The GAP Domain and the SNARE, Coatomer and Cargo Interaction Region of the ArfGAP2/3 Glo3 are Sufficient for Glo3 Function

Abstract

The ArfGAP Glo3 is required for coat protein I vesicle generation in the Golgi–endoplasmic reticulum (ER) shuttle. The best-understood role of Glo3 is the stimulation of the GTPase activity of Arf1. In this study, we characterized functional domains of the ArfGAP Glo3 and identified an interaction interface for coatomer, SNAREs and cargo in the central region of Glo3 (BoCCS region). The GAP domain together with the BoCCS region is necessary and sufficient for all vital Glo3 functions. Expression of a truncated Glo3 lacking the GAP domain results in a dominant negative growth phenotype in glo3Δ cells at 37°C. This phenotype was alleviated by mutating either the BoCCS region or the Glo3 regulatory motif (GRM), or by overexpression of ER–Golgi SNAREs or the ArfGAP Gcs1. The GRM is not essential for Glo3 function; it may act as an intrinsic sensor coupling GAP activity to SNARE binding to avoid dead-end complex formation at the Golgi membrane. Our data suggest that membrane-interaction modules and cargo-sensing regions have evolved independently in ArfGAP1s versus ArfGAP2/3s.

In eukaryotic cells, proteins and membranes are transported between organelles and to the plasma membrane. This transport is carried out primarily through the production of transport vesicles at one membrane compartment and the fusion of these vesicles with another compartment. Although such transport vesicles exhibit distinct and characteristic protein coats, the underlying mechanism of vesicle generation is conserved, with the various coats and adaptor proteins functioning in similar fashion. In general, a small GTP-binding protein belonging to the Arf/Sar family of GTPases is activated by a guanine nucleotide exchange factor (GEF) that, in turn, allows interaction of the GTPase with the membrane that is associated with the GEF. The activated GTPase then associates with proteins, such as SNAREs and cargo, which are targeted for transport to the next compartment along the vesicular transport pathway. A complex is formed between the small GTPase and the cargo or SNARE, and this complex may be stabilized by another effector of the GTPase, the GTPase-activating protein (GAP), and/or other coat components. This protein complex has been referred to as primer (1). When a threshold amount of cargo/SNARE–GTPase complexes is achieved, the associated coat drives membrane deformation and vesicle formation is initiated. If this threshold amount cannot be achieved, the GTPase is inactivated through the action of the GAP protein and the primer complex dissociates. This cycle is repeated until a critical mass of cargo and priming complexes is available to stably deform the membrane and allow coat polymerization to form the mature transport vesicle. In this process of vesicle formation, the GAP protein plays a critical role by assessing the status of the priming complex.

Most transport vesicles require the Arf1 GTPase; therefore, ArfGAPs are predominant regulators of vesicle formation. Two major classes of ArfGAPs have been identified to date: the ArfGAP1 class, which has the GAP domain at the N-terminus, and the AZAP class, in which the GAP domain is sandwiched between ankyrin repeats and a PH domain (2). Moreover, in mammalian cells, distinct subclasses of the ArfGAP1 class have been identified. The founding member, ArfGAP1, acts at the Golgi and is involved in coat protein I (COPI)-coated vesicle formation (3,4), whereas ArfGAP2 and ArfGAP3 are less well characterized but also participate in COPI vesicle formation (5,6). Regardless of the specific roles for ArfGAP family members, the functions of these proteins appear to be conserved from yeast to mammals. The conserved N-terminal GAP domain in the ArfGAP1 class is essential for function. In addition, ArfGAP1 family members contain an ArfGAP1 Lipid Packing Sensor (ALPS) motif that can form an amphipathic helix and is responsive to membrane curvature (7,8). In contrast, ArfGAP2 and ArfGAP3 proteins do not exhibit an obvious ALPS motif but instead contain a C-terminal motif (termed the Glo3 motif) that seems to play a regulatory function (9).

In the budding yeast Saccharomyces cerevisiae, six proteins possess the conserved ArfGAP domain, and four of these family members have proven ArfGAP function by both in vivo and in vitro criteria (10–13). The Gcs1 and Glo3 members of the yeast ArfGAP family are the most extensively characterized (10,14). Gcs1 is the yeast orthologue of the mammalian ArfGAP1, and the Glo3 protein belongs to the ArfGAP2/3 class of proteins. Glo3 and Gcs1 ArfGAP proteins have partially overlapping functions in retrograde transport from the Golgi to the endoplasmic reticulum (ER) (10). In addition to stimulating the GTPase activity of Arf1, Glo3 and Gcs1 can also induce a conformational change in SNARE proteins in vitro(15). This ArfGAP-mediated change in conformation is a prerequisite for the subsequent binding of Arf1 to SNAREs, indicating an active role of ArfGAP proteins in primer complex formation. Interestingly, this ArfGAP-induced conformational change is independent of the GAP activity itself. Thus, Glo3 and Gcs1, with only limited sequence conservation outside of the conserved ArfGAP domain, perform similar functions for vesicle formation.

To expand our appreciation of ArfGAP involvement in vesicle formation and function, we carried out a structure/function analysis of the yeast Glo3 protein that belongs to the ArfGAP2/3 family of ArfGAPs. We show that Glo3 has a central region involved in coatomer and SNARE/cargo binding that we refer to as the binding of coatomer, cargo and SNARE (BoCCS) region. The BoCCS region and the N-terminal GAP domain comprise a minimal version of Glo3 that can fulfill all essential ArfGAP functions of Gcs1 and Glo3 in the cell. Furthermore, in agreement with an earlier report (9), the conserved C-terminal ‘Glo3’ motif is found to have a regulatory function. We show that this C-terminal Glo3 motif may act as a signal transducer, allowing communication between the GAP domain and the BoCCS region. Our investigation of Glo3 function suggests a general mechanism by which ArfGAPs mediate the interplay among proteins during transport-vesicle formation. Furthermore, our data provide evidence for an independent evolution of the membrane- and coatomer/cargo interaction modules.

Results

The GAP domain of the Glo3 ArfGAP is necessary but not sufficient to restore function in cells lacking GLO3 and GCS1

The ArfGAPs Glo3 and Gcs1 provide overlapping function for retrograde transport from the Golgi to the ER and are structurally related, with greatest similarity exhibited in their GAP domains (Figure 1A). Although deletion of either the GLO3 or GCS1 gene results in cells that are viable, such single-deletion cells are unable to grow at low growth temperatures (16°C), while deletion of both GLO3 and GCS1 genes is lethal (10).

Figure 1. The GAP domain of Glo3 is not sufficient to complement the concomitant absence of Glo3 and Gcs1. A) The ArfGAPs Glo3 and Gcs1. The identity levels between the GAP domains and over the entire proteins are shown. Crosshatch, BoCCS region. The positions of the domains within the proteins are indicated by amino acid numbers. B) The GAP domain of Glo3 does not alleviate the lethal phenotype of glo3Δgcs1Δ cells. Double-deletion glo3Δgcs1Δ cells, which were kept alive by the expression of GLO3 from the inducible GAL1 promotor, were transformed with plasmids expressing C-terminal deletions or a GAP-domain deletion of Glo3. Growth was assayed on GAL1-promotor repressing (glucose) and inducing (galactose) conditions.

To assess whether the Glo3 GAP domain itself is sufficient for essential Glo3 function, we expressed Glo3 residues 1–145 (encompassing the GAP domain) in glo3Δgcs1Δ double-deletion cells that are kept alive by a plasmid-borne GLO3 under control of an inducible GAL1 promoter (Figure 1B). Incubation of these mutant cells on glucose-containing medium prevents expression of GLO3. Under these conditions, expression of the Glo3 ArfGAP domain alone failed to support growth. Even expression of the GAP domain embedded in a larger portion of the Glo3 protein (the first 213 amino acids) did not allow wild-type levels of growth. However, good growth of these glo3Δgcs1Δ double-mutant cells was supported by expression of the first 375 amino acids of the 493-residue Glo3 protein. These findings indicate that the central region of the Glo3 protein is important for Glo3 function. The GAP domain of Glo3 is also essential: a truncated version of Glo3 lacking a functional GAP domain failed to support growth of glo3Δgcs1Δ double-deletion cells (Figure 1B).

The C-terminal 118 amino acids of Glo3, shown above to be dispensable for cell growth, contain the highly conserved ‘Glo3’ motif (9), referred to here as the Glo3 regulatory motif (GRM). Our findings support the suggestion by Yahara et al. that the GRM may have a regulatory function.

Dissection of Glo3 functions

Although we understand the function of the GAP domain of Glo3, there is only limited appreciation of the potential roles of other portions of the protein. As a first approach to address this limitation, we determined the ability of truncated versions of Glo3 to relieve growth defects seen for glo3Δgcs1Δ double-deletion cells that are kept alive by a temperature-sensitive allele of GCS1, gcs1-28(10). This strain does not grow at either 16°C or 37°C ( (10); Figure 2A,C). Into this mutant background, we introduced truncated versions of GLO3 under control of the inducible MET3 promotor and assessed cell growth at 16°C, 30°C and 37°C (Figure 2A,C). The MET3 promotor is turned on in the absence of methionine in the medium (-met). All constructs were expressed in the absence of methionine from the medium (supporting information Figure S1). The growth defect at elevated temperature (37°C) was alleviated by all Glo3 derivatives that contain the GAP domain (residues 7–124), suggesting that the temperature sensitivity (ts) of these mutant cells is due to inadequate stimulation of Arf1-dependent GTP hydrolysis. In contrast, the GAP domain was neither necessary nor sufficient to alleviate the cold sensitivity (cs) of glo3Δgcs1-28 cells. Instead, a 162-residue portion of the central region of Glo3 (residues 214–375, hereafter referred to as the BoCCS region; Figure 1A) was sufficient to support growth of this strain at 16°C. This Glo3 fragment contains neither the GAP domain nor GRM, indicating that the highly conserved GAP domain and GRM are not required for essential Glo3 functions at 16°C in this background.

Figure 2. The central part of Glo3 alleviates the cold sensitivity of glo3Δcells. A) glo3Δgcs1Δgcs1-28 cells harboring various plasmids expressing Glo3 truncations under the control of the MET3 promotor were grown to early log phase at 30°C, and subsequent growth was assessed at different temperatures in methionine-free medium (-met, MET3 promoter on) by serial-dilution assay. Growth on methionine-containing medium (+met, MET3 promoter off) controlled for equal spotting. B) Serial-dilution assay of glo3Δ cells harboring plasmids expressing Glo3 truncations under the control of the MET3 promoter. Strains were grown as in (A) and growth on medium ± methionine was assayed at 16°C, 30°C, and 37°C. C) Summary of growth phenotypes. The Glo3 GAP domain alleviated the temperature sensitivity and the BoCCS region (residues 214–375) alleviated the cold sensitivity of glo3Δ cells, whereas cells with Glo3 derivatives lacking the GAP domain but containing the GRM failed to grow at 37°C and 16°C. The triangle marks the R59K substitution in the GAP-dead mutant.

It is possible that the gcs1-28 temperature-sensitive mutation contributes to the cold-sensitive phenotype of glo3Δgcs1-28 cells. We therefore assessed whether the BoCCS region of Glo3(214–375) alleviates the cold-sensitivity of single-mutant glo3Δ cells (Figure 2B,C). The BoCCS region rescued the growth defect of glo3Δ at 16°C, indicating that this region provides essential functions at low temperatures. Interestingly, the growth of glo3Δgcs1-28 cells was rescued by the expression of Glo3(214–493) (Glo3ΔGAP), while the same positive effect was not observed in the glo3Δ strain. Because gcs1-28 is expressed from a CEN plasmid, we suggest that a slight increase in Gcs1-28 protein levels may rescue the phenotype. This explanation seems likely because expression of wild-type GCS1 from a CEN plasmid also rescues a Δglo3 Glo3ΔGAP strain at 16°C (Figure S4C and data not shown). More importantly, our results show that the ArfGAP Glo3 has separable functions and that distinct regions carry out these functions. The possible functions of Glo3 include ArfGAP activity, and specific interactions with membranes and proteins. Because it is well established that the N-terminal GAP domain provides GAP activity, the specific membrane- and protein-interacting functions of Glo3 are likely encoded by the C-terminal portion of Glo3; cold-sensitive phenotypes are often attributed to defects in protein–protein or protein–membrane interactions.

A cluster of basic residues in the Glo3 BoCCS region is important for coatomer and SNARE/cargo binding

To further analyze the function of the BoCCS region and the GRM domain, we first compared the sequences of Glo3 and the mammalian ArfGAP2 and ArfGAP3 proteins, which revealed, in addition to the highly conserved N-terminal GAP domain and the GRM, considerable sequence conservation within the BoCCS region (9) (Figure S2A,B). Therefore, the BoCCS region and the GRM may fulfill functions conserved from yeast to humans. We used HHpred (16,17) to predict the secondary structure of the BoCCS region and the C-terminus (residues 214–493; Glo3ΔGAP); exclusion of the GAP domain allowed more reliable predictions. The overall confidence level, however, was too low (Figure S2C) to build a high-confidence three-dimensional model of Glo3ΔGAP using different parameters and various algorithms, e.g. SAM, Phyre and HHpred (18–20). Nonetheless, large stretches were predicted to be neither β-strands nor α− helices, indicating that the BoCCS region may be rather flexible. Interestingly, a cluster of well-conserved basic residues is present at the beginning of the BoCCS region. Basic stretches can either promote membrane association (21) or engage in protein–protein interactions. To score the importance of the basic cluster, we made substitutions in two basic tracts (Figure 3A; bold RKK and KKK converted to glutamates, Figure S2C). The mutated BoCCS region [Glo3(214–375) DB] no longer supported the 16°C growth of glo3Δ or glo3Δgcs1-28 mutant cells (Figure 3B, and data not shown). This growth defect was not due to differences in expression levels (Figure S1C,D). In marked contrast, the K267P substitution in Glo3(214–375), predicted to disrupt an adjacent α−helix, did not compromise Glo3(214–375) function at 16°C (Figure 3B○). These findings indicate that the basic stretches are essential for Glo3 function at low temperatures in this context.

Figure 3. The BoCCS region of Glo3 binds coatomer and SNARE/cargo proteins. A) Secondary structure prediction for the C-terminal part of Glo3 with the BoCCS region and the adjacent GRM (residues 214–493, Glo3ΔGAP). The basic-stretch residues and the single lysine, which were mutated, are highlighted. Red indicates α-helix, the green arrow β-sheet and blue straight lines unstructured regions. B) The BoCCS basic-stretch mutants fail to alleviate glo3Δ cold sensitivity. A serial-dilution assay was carried out for glo3Δ cells expressing the indicated constructs. No growth at 16°C was observed when both BoCCS region basic tracts were substituted by glutamate (), whereas growth was supported by the BoCCS region with the K267P substitution (○). C) Glo3 binds directly to the cytoplasmic part of SNARE proteins. GST or SNARE–GST fusion proteins, in which the transmembrane domain was replaced with GST, were immobilized on GSH-agarose. Recombinant Glo3 was added and the binding to GST proteins determined by pulldown. Glo3 bound to SNAREs but not to GST. The arrowhead indicates full-length Glo3. D) The basic-stretch double mutant [Glo3(214–375) DB] no longer binds coatomer, as indicated by co-immunoprecipitation. The levels of wild-type and truncated versions of myc-tagged Glo3 in the lysates are shown on the left, and similar amounts of Glo3-myc were precipitated in all cases. Coatomer, Emp47 and Bos1 failed to bind when the BoCCS region was mutated. The GTPase Ypt7 was an immunoprecipitation control. The star indicates the double basic-stretch substitution. E) The BoCCS region is not membrane associated. Differential centrifugation was carried out on cell lysates expressing different Glo3 constructs; pellets were resuspended to the initial volume and equal volumes were subjected to SDS–PAGE. Proteins were visualized by immunoblots with antibodies against coatomer, the myc tag, the SNARE Bos1 and the soluble protein Pgk1.

To understand the function of the BoCCS region at the cellular level, we addressed the importance of the BoCCS region for protein–protein and/or protein–membrane interactions. First, we tested for the role of BoCCS in two well-established features of Glo3, namely the binding of coatomer and the association with SNARE proteins. We showed previously that coatomer interacts directly with Glo3 (14). In addition, the interaction between coatomer and mammalian ArfGAP3 has been recently reported (6). Furthermore, Glo3 is capable of interacting with and inducing a conformational change in a wide variety of SNARE proteins (15,22). Moreover, recombinant Glo3 bound directly to glutathione S-transferase (GST)–SNARE fusion proteins (Figure 3C). Both interactions of Glo3 with coatomer and SNAREs are considered important in regulating the efficiency of vesicle generation. Several Glo3 derivatives were N-terminally appended with a 3× myc-epitope tag. These tagged proteins behaved as the untagged versions in growth assays (data not shown). Interestingly, all Glo3 derivatives containing the BoCCS region (214–375) bound efficiently to coatomer (Figure 3D). Moreover, the coatomer–Glo3 interaction was abolished by the substitutions that eliminate basic residues from the BoCCS region. We were also able to co-immunoprecipitate the ER–Golgi SNAREs Bos1 and Bet1 and the cargo protein Emp47 using the BoCCS region of Glo3 (Figures 3D and 5C). These in vivo interactions with the BoCCS region are likely to reflect direct protein–protein interactions because we have shown that SNARE proteins and Glo3 can interact directly in vitro (Figure 3C) (15,22). Moreover, Glo3 interacts directly with the tails of the p24 family members Emp24 and Erv25 (23). The binding is likely to be stabilized by the recruitment of coatomer. In contrast, Bos1 and Emp47 failed to co-immunoprecipitate with the Glo3(214–375) DB mutant protein lacking the conserved basic residues (Figure 3D). Taken together, these results suggest that the BoCCS region comprises interaction sites for coatomer, SNAREs and cargo. This, however, neither addresses whether this interaction is direct nor excludes that other parts of Glo3 may help stabilize BoCCS interaction with SNARE, cargo and coatomer.

Figure 5. The Glo3 regulatory motif (GRM) is responsible for the negative growth effect of Glo3 derivatives lacking the GAP domain. A) Wild-type and mutant versions of the GRM used in this study. B) GRM substitutions alleviate the inhibitory effect of Glo3(214–493) in glo3Δ cells. Plasmids encoding Glo3(214–493) with GRM substitutions and expressed under control of the MET3 promoter were transformed into glo3Δ cells, which were then tested for growth at 16°C in the presence (MET3 promoter off) and absence (MET3 promoter on) of methionine in the growth medium. C) GRM substitutions do not affect coatomer binding. Glo3-myc immunoprecipitations from cells expressing myc-tagged Glo3(214–493) with and without the glo3-13 mutations were probed for coatomer, Emp47, Bos1, Bet1 and Ypt7. D) GRM substitutions do not interfere with the membrane association of myc-Glo3(214–493) (Glo3ΔGAP). Differential centrifugation and analysis of lysates from cells expressing myc-Glo3 and myc-Glo3(214–493) with and without the glo3-13 GRM substitutions were performed as described in Figure 3. ‘++’ mark full-length myc-Glo3 and ‘+’ indicates the migration position of the truncations.

Next, we tested whether the basic residue cluster in the BoCCS region promotes membrane association of Glo3 by differential centrifugation. Surprisingly, both the wild-type and the DB mutant versions of the BoCCS region were mostly soluble, and hence it is unlikely that BoCCS per se plays a major role in Glo3 membrane binding (Figure 3E). Even reducing the expression levels of BoCCS to that of Glo3 did not change the distribution, indicating that membrane-binding sites are not limiting under these conditions (Figure S3). Thus, we conclude that the basic stretch of the BoCCS region provides an interaction platform for coatomer, SNAREs and cargo, but is not itself sufficient to serve as the primary membrane-binding site.

Essential Glo3 functions are provided together by the GAP domain and BoCCS region

Glo3 contains at least three conserved parts: the GAP domain, the BoCCS region and the GRM. As shown above (Figure 1B), the GRM is dispensable for growth. Therefore, we wanted to test whether the GAP domain and the BoCCS region were sufficient to fulfill all essential Glo3 functions. We generated a fusion construct between the GAP domain and the BoCCS region and expressed this fusion in the glo3Δgcs1Δ strain, which is kept alive by a plasmid-borne wild-type GLO3 under control of the inducible GAL1 promoter. The GAP–BoCCS fusion rescued the growth phenotype of glo3Δgcs1Δ to the same extent as wild-type Glo3 or Glo3(1–375), which also comprises both the GAP domain and the BoCCS region (Figure 4A).

Figure 4. Analysis of the role of the BoCCS region of Glo3. A) A fusion of the N-terminal GAP domain to the BoCCS region was tested for the ability to sustain the growth of glo3Δgcs1ΔGAL1::GLO3 cells on glucose medium lacking methionine (for wild-type Glo3 shut-off and Glo3-derivative expression) and galactose medium (control). B) The GAP domain and the BoCCS region can provide all essential Glo3 function when expressed in trans. The GAP domain and the BoCCS region were expressed either individually from two different plasmids or as a fusion from the same plasmid, and growth was determined on glucose-containing medium in a glo3Δgcs1ΔGAL::GLO3 strain at 30°C. C) The basic-stretch substitutions do not affect glo3Δ complementation by full-length Glo3 protein. glo3Δgcs1Δgcs1-28 transformants carrying the various Glo3 constructs under control of the MET3 promoter were plated on medium lacking methionine. The gray star marks the changes of the first three lysines, the black the mutations of the second stretch of three lysines and the open star the mutation of both basic stretches. D) Full-length Glo3 with basic-stretch substitutions fails to bind coatomer, as indicated by immunoprecipitation. Levels of myc-tagged Glo3 and derivatives in the lysates are shown on the left, and similar amounts were precipitated in all cases. For Glo3 DB, the interactions with Bos1 and Emp47 were also lost. Symbols are as in panel (C).

Our data suggest that the GAP domain and the BoCCS region fulfill independent functions (one GAP activity; the other cargo, coaotmer and SNARE interaction), and that both functions are essential for proper Glo3 function. If this assumption was true, coexpressing both domains individually from two different plasmids should also rescue the lethality of glo3Δgcs1Δ. Expression of either domain did not support growth. However, expression of both regions in trans promoted growth of glo3Δgcs1Δ albeit to a lesser extent than the expression of the fusion or the full-length of Glo3 (Figure 4B). These data show that the GAP domain and the BoCCS region are necessary and sufficient, even in trans, to provide all essential Glo3 functions.

Stable coatomer and SNARE interactions are not essential for Glo3 function

Because mutations in the GAP domain render Glo3 inactive, we wanted to determine whether mutations in the BoCCS region would behave similarly. To address this, we introduced the basic-stretch mutations into the full-length Glo3 construct and assessed their influence on growth at 16°C as measure of functionality in glo3Δgcs1-28 or glo3Δ cells. Surprisingly, all basic-stretch mutations rescued the 16°C growth phenotype in the context of full-length Glo3, whereas under the same growth conditions only mutations in the first basic stretch in the context of the isolated BoCCS region rescued to a similar extent than wild type (Figure 4C and data not shown). These data indicate that either there is another coatomer, cargo, SNARE interaction site or the BoCCS region was tampered down into a low affinity-binding region that prevented protein–protein interactions to be detected. Alternatively, the mutated BoCCS region in isolation may not be able to fulfill its function as protein scaffold at all. We can exclude that there are alternative high affinity interaction sites for coatomer, cargo and SNAREs, as their interaction is lost with full-length Glo3 in which both basic stretches were mutated (Figure 4D). These data support the model that the BoCCS region provides the interface for coatomer, cargo and SNAREs. These results also indicate that other parts of Glo3, like GRM, contribute to Glo3 function.

GRM regulates BoCCS function

In contrast to the positive effects on growth of the BoCCS region itself, a Glo3 derivative comprising both the BoCCS region and GRM (residues 214–493, Glo3ΔGAP) did not alleviate the cold-sensitivity of glo3Δ mutant cells (Figures 2B and 5B). Moreover, expression of this construct inhibited growth at 37°C (Figure 2B). To test whether the GRM is responsible for this growth inhibition, we introduced several non-functional versions of the GRM [encoded by glo3-11, glo3-12, glo3-13, glo3-14; Figure 5A; (9)] into Glo3ΔGAP (residues 214–493). The non-functional GRM constructs rescued the negative growth effect of Glo3ΔGAP at 16°C and 37°C (Figures 5B and S4A). These findings indicate that the GRM counteracts the positive growth effect of BoCCS in the absence of the GAP domain.

Next, we determined the effect of an altered GRM on BoCCS function as a protein scaffold. Although altering the GRM did not affect coatomer binding, the altered GRM did lead to markedly decreased binding to the SNAREs Bos1 and Bet1 (Figure 5C). Importantly, mutations in the GRM, which renders it inactive (9), had no effect on membrane association (Figure 5D). Our data suggest that GRM facilitates BoCCS region association with SNAREs, potentially to participate in primer complex formation for vesicle generation.

The inhibition caused by the Glo3ΔGAP construct and the reversal of the inhibition by inactivating the GRM suggest a model in which the GAP domain would transmit the GTP hydrolysis of Arf1 to the GRM, which in turn would lead to the release of the protein interactions in the BoCCS region. This hypothesis makes several predictions. First, if the basic stretch is mutated in Glo3ΔGAP, the negative growth effect in glo3Δ at 37°C should no longer be observed, which is exactly what we observed (Figure S4D). Second, a point mutation that inactivates GAP activity should phenotypically mimic the growth inhibition caused when the entire GAP domain is missing. Lacking GAP activity (14) did not allow growth of glo3Δ mutant cells at 16°C and prevented growth at 37°C (Figures 2CΔ, 6AΔ and S4B Δ). Therefore, our data support a model in which GRM transmits a signal from the GAP domain to regulate the release of cargo/SNARE proteins from the BoCCS region. For GAP-dead versions of Glo3 or derivatives lacking the GAP domain, such signaling through the GRM is abolished, thereby preventing the normal regulation of binding to the Glo3 BoCCS region.

Figure 6. GRM-mediated growth inhibition seen for Glo3 derivatives lacking GAP activity is alleviated by increased expression Gcs1. A) Glo3 derivatives lacking GAP activity cause temperature sensitivity. Drop assay of glo3Δ cells expressing Glo3 derivatives under the control of the MET3 promoter were tested for 37°C growth under conditions in which the MET3 promoter is on (- met) or off (+ met). The triangle indicates the R59K substitution that abolishes GAP activity. B) Extra copies of the GCS1 gene alleviate the temperature sensitivity caused by Glo3 derivatives lacking GAP activity. Plasmids expressing Gcs1 from low-copy (CEN) and high-copy (2μ) plasmids were transformed into glo3Δ cells expressing the indicated Glo3 derivatives, and the resulting transformants were tested for 37°C growth.

Increased expression of either Gcs1 or ER–Golgi SNAREs overcomes the growth inhibition caused Glo3ΔGAP

Another prediction of the proposed model is the participation of Glo3ΔGAP in a dead-end complex. The inability of Glo3ΔGAP to dissociate from SNARE, coatomer and cargo would create stable complexes on membranes. These dead-end complexes would ultimately inhibit vesicles formation by virtue of sequestering critical components, consistent with the observed inhibition of growth by Glo3ΔGAP (Figure 2C) and the reversal of the growth inhibition at 37°C by mutating the basic stretches in Glo3ΔGAP (Figure S4D). To address this scenario, we postulated that elevating the abundance of the ArfGAP1 Gcs1 would outcompete Glo3ΔGAP for binding sites and thereby reducing the level of dead-end complexes. Increased expression of Gcs1 indeed restored growth at 16°C and 37°C to these mutant cells (Figures 6B and S4C, and data not shown).

Alternatively, increasing the levels of critical transport components such as SNARE proteins might also alleviate the negative growth effect, not by preventing dead-end complex formation but by supplying enough transport components to ensure efficient transport. In a previous multicopy suppressor screen, we isolated SNAREs as efficient suppressors of the cold-sensitive growth phenotype of glo3Δ cells (10). Overexpression of these SNAREs also rescued the growth phenotype conferred by the Glo3ΔGAP constructs at 37°C, indicating that excess of SNAREs can bypass the putative dead-end complexes and thus allow efficient vesicle formation (Figure 7). Moreover, our data suggest that the underlying defect producing the cold-sensitive phenotype in glo3Δ cells and the temperature-sensitive phenotype of Glo3ΔGAP expression in glo3Δ cells may be mechanistically related. Taken together, we have identified a novel conserved region in Glo3, termed the BoCCS domain that is required for the interaction with coatomer, cargo and SNARE. Interactions with the BoCCS domain are regulated by the GRM domain. The GRM domain in turn receives input from the GAP domain about the guanine nucleotide state of Arf1.

Figure 7. Overexpression of SNAREs suppresses the temperature-sensitive phenotype of glo3Δcells expressing Glo3(214–493). Multicopy genomic-library plasmids that allowed the growth of glo3Δ Glo3(214–493) cells at 37°C were selected; the active gene on the genomic insert is indicated. Pos, positive control for 37°C growth; vector, empty-vector control.

Discussion

The yeast ArfGAP Glo3, a member of the ArfGAP2/3 family, mediates retrograde vesicular transport from the cis-Golgi to the ER and is a GAP for the small GTPase Arf1. The findings described in this study show that the GAP activity of Glo3 is but one of the activities of this protein that mediates vesicular transport. We found that the central portion of the Glo3 protein, here termed the BoCCS region, is the only portion of Glo3 that is needed in vivo at 16°C, even though this portion of Glo3 lacks the GAP domain; GAP activity is likely provided by Gcs1. This BoCCS region associates with coatomer, cargo and SNARE proteins, thereby providing an important function during transport-vesicle formation. Although binding of Glo3 to coatomer (14,24) and to cargo (23) has been shown previously, the domain responsible for this binding was unknown. Recently, Kliouchnikov et al. reported that the basic-stretch region in ArfGAP3 is important for coatomer binding (6). In addition to the BoCCS domain, a sequence motif in Glo3 that is conserved among related ArfGAP proteins (9) referred to here as the GRM regulates protein binding to the BoCCS region. The GRM itself is regulated by the activity of the GAP domain. The Glo3 protein is therefore multifunctional, with GAP-dependent and GAP-independent activities.

The combination of the BoCCS region and the GAP domain provides all essential functions of Glo3. The C-terminal GRM, albeit dispensable for growth, appears to serve as a liason between the GAP domain and the BoCCS region. The GRM may be transmitting a signal from the GAP domain to the BoCCS region upon GTP hydrolysis, leading to destabilization of the interactions of the BoCCS region with cargo/SNAREs and coatomer. For a Glo3 mutant protein lacking GAP activity, this regulatory feedback is never provided, and the mutant protein remains stably bound to the membrane; in contrast, eliminating the GRM allows the BoCCS region to release cargo and SNAREs. The GRM domain has also an effect on coatomer because mutations in the GRM were not able to rescue a mutant in the β-COP subunit of coatomer SEC26(24), albeit the interaction between coatomer and mutated GRM was not compromised (Figure 5C). Communication between the different modules in Glo3 is probably achieved through conformational changes (Figure 8A).

Figure 8. Model of Glo3 regulation. A) Model of Glo3 intramolecular regulatory interactions identified here. The arrow between GRM and BoCCS indicates stabilization of binding to the BoCCS, whereas the blunt arrow from the GAP domain represents inhibition of binding to BoCCS leading to release of cargo. B) Phylogenetic tree of different ArfGAPs in yeast.

Based on our findings, we propose a model in which Glo3 acts analogously to the Sec23/24 protein complex that mediates formation of the COPII vesicle coat during anterograde transport from the ER to the Golgi. For COPII transport-vesicle formation, Sec23 is the GAP for the GTPase Sar1, whereas Sec24 recognizes cargo (25–28). The Sec23/24 complex then recruits the Sec13/31 complex, which forms the outer layer of the COPII vesicle coat (29,30). Like Sec23, Glo3 can stimulate GAP activity through its GAP domain, and like Sec24, it can bind other coat components, cargo and SNARE proteins. In this study, we show that these binding abilities are localized to the BoCCS region of Glo3. Coatomer binding by the Glo3 BoCCS region is independent of cargo/SNARE binding because eliminating the GRM reduced SNARE binding while coatomer binding was unaffected. In contrast, when coatomer binding was eliminated (by the basic-stretch mutations), SNARE and cargo binding was also abolished. A likely explanation is that coatomer stabilizes the SNARE–ArfGAP interaction. Glo3 thus displays the activities of both members of the Sec23/24 complex.

ArfGAP2/3 family members, such as Glo3, and ArfGAP1 family members, such as Gcs1, are involved in the generation of COPI-coated vesicles (4,5,10,14,31). Surprisingly, these two classes of ArfGAP proteins exhibit structural similarity only in their N-terminal GAP domains. The dissimilarities in more distal regions of these proteins suggest that portions of these proteins outside the GAP domain may mediate distinct functions or may have evolved different strategies to meet the common need to regulate GAP function (Figures 1A and 8B). For example, members of the ArfGAP1 family, including the yeast Gcs1, contain an ALPS domain (8) that is capable of detecting membrane curvature and acting as a sensor to regulate GAP activity within the proper membrane context. Although the ALPS domain is important for ArfGAP1 function, the ALPS domain is not found in ArfGAP2/3 family members. Unlike the ArfGAP1 proteins, the ArfGAP2/3 proteins contain the conserved GRM (9), which we suggest functions to regulate protein binding to Glo3 for transport-vesicle production.

The ArfGAPs Glo3 and Gcs1 have overlapping functions in retrograde transport from the Golgi to the ER (10), and both proteins can induce conformational changes in SNAREs (22,32) that are required for vesicle biogenesis. Nevertheless, there are functional differences between the classes of ArfGAPs, as indicated by our finding that Glo3 interacts directly with coatomer and is built into the COPI coat, whereas Gcs1 does not bind coatomer and is not incorporated into vesicle coats (14). Furthermore, Gcs1 binds to the GDP-bound form of Arf1 lacking the N-terminal 17 amino acids (ΔN17-Arf1T31N) (33), whereas under the same experimental conditions Glo3 binding cannot be detected. Moreover, Glo3 binds directly to the p24 family members Emp24 and Erv25, whereas Gcs1 does not (23). In contrast to the yeast situation, mammalian ArfGAP1 is associated with Golgi membranes in a coatomer-dependent manner when Arf1 is activated (4), indicating an active role for ArfGAP1 in COPI vesicle biogenesis. These different features of the ArfGAPs suggest that Glo3 and Gcs1 have specialized functions in vesicle biogenesis.

The absence of either the Glo3 or Gcs1 is tolerated, and vesicular transport occurs in both glo3Δ and gcs1Δ single-mutant cells. The ability of single-mutant cells to function without a particular ArfGAP protein may reflect parallel but different strategies adopted by Glo3 and Gcs1 to regulate function. Nonetheless, both proteins can sense the development of a transport vesicle to regulate GAP activity. Gcs1 directly senses membrane curvature through the ALPS domain, whereas Glo3 senses cargo and SNARE recruitment as part of the vesicle coat. The different strategies used by Glo3 and Gcs1 in vesicle biogenesis indicate that the sensing mechanisms evolved separately.

Perhaps, cells can survive without Glo3 because, during active growth, the amount of cargo associated with a donor membrane and the precise timing of the GTPase cycles of Arf1 may not be critical. Under these conditions, whether Glo3 is present to stabilize the coat and the priming complex may not be significant because there is always enough cargo to be transported. Analogously, the point when the coat of a vesicle is destabilized by GTP hydrolysis may not be important when there are enough transport vesicles, so that any kinetic delay for a single vesicle-fusion event can be tolerated by the cell.

Material and Methods

Yeast methods, strains and antibodies

Standard yeast genetic techniques and media were used (34). Yeast strains used in this study are listed in Table 1. The glo3Δ::HIS3 and gcs1Δ::URA3 glo3Δ::HIS3 deletions have been described (10). Polyclonal rabbit antibodies directed against coatomer, Bos1, Bet1, Glo3 (14), Emp47 andYpt7 (both gifts from HD Schmitt, Göttingen), mouse monoclonal anti-myc (Sigma) and anti-Pgk1p (Invitrogen) antibodies were used in this study.

Table 1.  Strains used in this study.

Strain	Genotype	Source
PPY51	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3	Poon et al. (10)
SL112	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28	Lewis et al. (14)
YAS950	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-GLO3(1–493)	This study
YAS951	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(102–493)	This study
YAS952	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–375)	This study
YAS953	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–213)	This study
YAS963	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–493)R59K	This study
YAS990	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(102–375)	This study
PPY126	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480)	Poon et al. (10)
YAS1243	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–145)	This study
YAS1244	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-GLO3(1–493)	This study
YAS1245	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-Glo3(1–375)	This study
YAS1246	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-Glo3(1–213)	This study
YAS1247	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-Glo3(214–493)	This study
YAS1248	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-GLO3(214–375)	This study
YAS1249	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-Glo3(1–145)	This study
YAS1257	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc	This study
YAS1258	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-3myc-GLO3(102–375)	This study
YAS1284	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc	This study
YAS1285	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-GLO3(1–493)	This study
YAS1286	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-Glo3(1–375)	This study
YAS1287	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-Glo3(1–213)	This study
YAS1288	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-Glo3(1–145)	This study
YAS1289	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-glo3(102–493)	This study
YAS1291	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-Glo3(214–493)	This study
YAS1292	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-Glo3(214–375)	This study
YAS1293	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1:: pPPL50-Gal10-Glo3(1–480) pLS439- MET3-3myc-Glo3(1–493)R59K	This study
YAS1511	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(214–375) K267P	This study
YAS1560	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(214–375) RKK223EEE	This study
YAS1561	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(214–375) KKK233EEE	This study
YAS1562	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(214–375) RKK223EEE, KKK233EEE	This study
YAS1601	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–493) RKK223EEE	This study
YAS1602	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–493) KKK233EEE	This study
YAS1603	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 gcs1::URA3 gcs1-28 pLS439- MET3-Glo3(1–493) RKK223EEE, KKK233EEE	This study
YAS1607	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3	This study
YAS1608	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-GLO3(1–493)	This study
YAS1609	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(1–213)	This study
YAS1610	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–375)	This study
YAS1611	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–493)	This study
YAS1614	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(102–493)	This study
YAS1615	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(102–375)	This study
YAS1616	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–375) RKK223EEE	This study
YAS1617	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–375) KKK233EEE	This study
YAS1618	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–375) RKK223EEE, KKK233EEE	This study
YAS1641	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(1–375)	This study
YAS1642	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(1–145)	This study
YAS1643	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–375) RKK223EEE	This study
YAS1644	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–375) KKK233EEE	This study
YAS1645	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pLS439- MET3-3myc-Glo3(1–493) R59K	This study
YAS1709	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc	This study
YAS1710	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc-GLO3(1–493)	This study
YAS1711	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc-Glo3(102–493)	This study
YAS1712	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc-Glo3(1–213)	This study
YAS1713	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc-Glo3(214–493)	This study
YAS1714	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc-Glo3(214–375)	This study
YAS1715	MATaleu2-3,112 ura3-1 his3-11,15 trp1-1 ade2-1 glo3::HIS3 pPP421-GCS1 pLS439- MET3-3myc-Glo3(1–493) R59K	This study
YAS1847	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal1-Glo3(11–493) pLS439- MET3-Glo3(1–145 and 214–375)	This study
YAS1864	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pLS439- MET3-3myc-Glo3(1–145 and 214–375)	This study
YAS1866	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–493)-11	This study
YAS1867	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–493)-12	This study
YAS1868	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–493)-13	This study
YAS1869	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–493)-14	This study
YAS1914	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pLS439- MET3-3myc-Glo3(214–432)	This study
PPY201-3	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pEP2-3-AKR1	This study
PPY201-45	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pEP2-45-MGA2
PPY201-48	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pEP2-49-GOT1	This study
PPY201-49	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pEP2-49-BET1	This study
PPY201-BOS1	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pEP2-BOS1	This study
PPY201-SEC22	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pEP2-SEC22	This study
PPY201-Neg	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pPPL150- MET3-Glo3(214–493) pRS315	This study
PPY201-Pos	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 pRS316 pRS315	This study
PPY209-vec/vec	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) YEp351 pPPL158-NatR	This study
PPY209-Glo3/vec	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-GLO3(1–493) pPPL158-NatR	This study
PPY209-GapBoCCS/ vec	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439-Glo3(1–145 and 214–375) pPPL158-NatR	This study
PPY209-GAP/vec	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-Glo3(1–145) pPPL158-NatR	This study
PPY209-vec/BoCCS	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) YEp351 pPPL159- MET3-GLO3(214–375)	This study
PPY209-Gap/BoCCS	MATaura3 leu2 trp1 his3 ade2 glo3::HIS3 gcs1::URA3 pPPL50-Gal10-Glo3(1–480) pLS439- MET3-Glo3(1–145) pPPL159- MET3-GLO3(214–375)	This study

Plasmids

GLO3 was amplified by polymerase chain reaction (PCR) using genomic DNA as a template and primers listed in Table 2 and cloned into pCR2.1-TOPO (Invitrogen). In addition to the coding region, 230 nucleotides of the 3′-UTR were added to all fragments of GLO3. Constructs lacking the C-terminus of Glo3 and therefore lacking the native Stop codon were generated by addition of a Stop codon as well as overhangs suitable for recombinant PCR in the primer. For these constructs, the 3′-UTR was amplified separately (PCR2). To generate the yeast expression vectors, the TOPO vectors were digested with HindIII and ApaI, and the Glo3-coding fragments were ligated into pSL439 (14). For expression, the initiation codon in the MCS of pSL439 was used. Site-directed mutagenesis and domain fusion was performed using the QuikChange II XL Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer's instructions. The fused domains were separated by a 7-residue linker. To produce N-terminally 3myc-tagged versions of Glo3, the 3myc tag of pYM4 (35) was amplified using primers CS173 and CS174. The PCR product was restricted with Eco31I and HindIII and ligated into HindIII-cleaved pSL439 to generate a new empty vector, pSL439-3myc. pSL439-3myc was constructed to encode a 5-residue linker separating the 3myc tag from the Glo3 fragments. All constructs were confirmed by sequencing.

Table 2.  Primer used in this study.

Restriction sites are underlined; inserted STOP codons are in boldface; sequence of linker in the domain fusion construct is in italics.
CS86	AAGCTTAGTAACGATGAAGGAGAAACATTTGC	GLO3 amplification
CS87	AAGCTTAAACAGCTACTAAATACAGCAAATGTTGAC	GLO3 amplification
CS88	GGGCCCGAACCAAATGCTACCTCGTCTG	GLO3 amplification
CS89	TATACGTAAGACTTAAGAAAATAAGTTTAATTTTCTCTTTAATTG	GLO3 amplification
CS91	ACTAAATAAGACTACTTAAGAAAATAAGTTTAATTTTCTCTTTAATTG	GLO3 amplification
CS93	GTAGTCTTATTTAGTGACAGTGGTCTTGGGTGT	GLO3 amplification
CS95	GTAGTCTTACGTATATCTTGGTCCCTGAGCTAT	GLO3 amplification
CS146	AAGCTTACGAGAAGTTCCATCTTAACGGCT	GLO3 amplification
CS169	GTAGTCTTACAAGACTAATTCTGATGGGTATAAGTC	GLO3 amplification
CS170	GTCTTGTAAGACTACTTAAGAAAATAATTTTCTCTTTAATTG	GLO3 amplification
CS173	CCCAAGCTTGACTCTAGATGATCCGTTCAAGTC	3myc amplification
CS174	GCGGGTCTCCAGCTCGGATCCTCTAGAGGTGAACAAAAGTT	3myc amplification
CS175	GGCAGAAGACTTGTTTGATCAATTTAAACCAGAGGCTCAACAAGAAAAGG	Site-directed mutagenesis
CS176	CCTTTTCTTGTTGAGCCTCTGGTTTAAATTGATCAAACAAGTCTTCTGCC	Site-directed mutagenesis
CS177	AAGCTTGGCAGAATCGCGGAAAGG	GLO3 amplification
CS181	CGAGAAGTTCCATCTTAACGGCTAGCGAAGAGGAACCGGTATTAAACTCGCAAG	Site-directed mutagenesis
CS182	CTTGCGAGTTTAATACCGGTTCCTCTTCGCTAGCCGTTAAGATGGAACTTCTCG	Site-directed mutagenesis
CS183	CCGGTATTAAACTCGCAAGATGAAGAGGAACATTCAATTTTGTCTTCGTCAAGAAAGCCC	Site-directed mutagenesis
CS184	GGGCTTTCTTGACGAAGACAAAATTGAATGTTCCTCTTCATCTTGCGAGTTTAATACCGG	Site-directed mutagenesis
CS213	GTCTTGGAATTCGGATCCGCTTACGTAACGAGAAGTTCCATCTTAACGGCT	Domain fusion
CS214	TCTCGTTACGTAAGCGGATCCGAATTCCAAGACTAATTCTGATGGGTATAACTC	Domain fusion
CS223	GAAGTGTAAGACTACTTAAGAAATAAGTTTAATTTTCTCTTTAATTG	GLO3 amplification
CS224	GTAGTCTTACACTTCTTTATCTTCTCCAAAGTATGAAGA	GLO3 amplification

Growth assays

For each plating assay, cells were grown in liquid selective medium and equilibrated to equal cell densities, followed by serial dilutions of 10-fold each. Portions of each dilution were applied as drops onto solid growth medium and incubated at indicated temperatures for appropriate times.

Immunoprecipitations and GST-pulldowns

Cells were grown to early- to mid-log phase in selective medium. Then, they were shifted to medium lacking methionine and incubated for 1 h to induce GLO3 or glo3 mutant-gene expression. Cells were then harvested by centrifugation at 3000 ×g for 3 min, washed twice with water, transferred to a microfuge tube and resuspended in B150 TW20 buffer [20 mm HEPES pH 6.8, 150 mm KOAc, 5 mm Mg(Ac)₂ and 1% Tween 20] and protease inhibitor. Glass beads were added and cells were lysed by vortexing at 4°C. Lysates were cleared by centrifugation (15 min at 13 000 ×g ) and the protein concentration was determined. Equal amounts of protein were incubated for 2 h at 4°C with anti-myc antibody covalently attached to beads. Beads were washed thrice with B150 TW20 buffer and once with B150 buffer [20 mm HEPES pH 6.8, 150 mm KAc and 5 mm Mg(Ac)₂]. Beads were boiled in sample buffer, and the bound proteins were analyzed by immunoblot.

SNARE–GST pulldown experiments were performed as described by Rein et al. (15), except for that 10 μg of SNARE proteins were immobilized and about 40 nm Glo3 was present in the reaction.

Subcellular fractionation

Overnight cultures were diluted to 0.1 OD₆₀₀ and grown for another 3–4 h at 30°C. Cells were harvested by centrifugation at 2000 ×g for 3 min at RT, washed twice with water and resuspended in 50 mL minimal medium lacking methionine and leucine to induce expression of Glo3 or derivatives. After 1 h, 20 OD₆₀₀ of cells were harvested by centrifugation at 2000 ×g for 3 min at room temperature. Cells were washed twice with water, resuspended in buffer A (100 mm Tris–HCl pH 9.4 and 10 mm DTT) and incubated for 5 min at RT. Cells were converted into spheroplasts by incubation for 30 min in buffer B (0.75× YP, 0.7 m sorbitol, 0.5% glucose, 50 mm Tris–HCl pH 7.5) containing zymolase. The collected spheroplasts (1000 ×g , 3 min) were resuspended in modified B88 buffer [20 mm HEPES pH 6.8, 250 mm sorbitol, 150 mm NaAc, 5 mm Mg(Ac)₂, 1 mm DTT and protease inhibitors], transferred to a microfuge tube and disrupted with a Dounce homogenizer. Unlysed spheroplasts were removed by centrifugation at 2000 ×g for 2 min at 4°C. The supernatant was transferred to a fresh microfuge tube and centrifuged for 10 min at 13 000 ×g at 4°C. The pellet (P13) was saved and the supernatant was transferred to an ultracentrifuge tube (Beckman) and centrifuged for 60 min at 100 000 ×g at 4°C. The supernatant (S100) and the pellet (P100) were saved. Pellets were solubilized in the starting volume of modified B88 buffer. Samples were analyzed by immunoblot.

Sequence analysis and secondary structure prediction

Multiple sequence alignments were performed using ClustalW (EMBnet-Ch server). For the phylogenetic tree construction, the alignments were loaded into Quicktree, which uses a neighbor-joining method. The output file was converted into a weighted cladogram using Drawgram. Quicktree and Drawgram are part of the PHYLIP software package. The ClustalW multiple sequence alignment of the ArfGAP2/3 was loaded into ESPript for visualization. The secondary structure of Glo3 was predicted using HHpred (17), PHYRE and SAM, and the prediction from HHpred subsequently submitted to POLYVIEW for graphical visualization.

Acknowledgments

We thank H.-D. Schmitt for antibodies, A. Nakano for the Glo3 motif mutation plasmids and J. Benjamin for critical reading of the manuscript. We are grateful to B. Antonny and D. Cassel for discussions. This work was supported by the Swiss National Fund (A. S.) and the Canadian Institutes of Health Research (G. C. J. and R. A. S.).