Two families of GTPases, Arfs and Ypt/rabs, are key regulators of vesicular transport. While Arf proteins are implicated in vesicle budding from the donor compartment, Ypt/rab proteins are involved in the targeting of vesicles to the acceptor compartment. Recently, we have shown a role for Ypt31/32p in exit from the yeast trans-Golgi, suggesting a possible function for Ypt/rab proteins in vesicle budding as well. Here we report the identification of a new member of the Sec7-domain family, SYT1, as a high-copy suppressor of a ypt31/32 mutation. Several proteins that belong to the Sec7-domain family, including the yeast Gea1p, have recently been shown to stimulate nucleotide exchange by Arf GTPases. Nucleotide exchange by Arf GTPases, the switch from the GDP- to the GTP-bound form, is thought to be crucial for their function. Sec7p itself has an important role in the yeast secretory pathway. However, its mechanism of action is not yet understood. We show that all members of the Sec7-domain family exhibit distinct genetic interactions with the YPT genes. Biochemical assays demonstrate that, although the homology between the members of the Sec7-domain family is relatively low (20-35%) and limited to a small domain, they all can act as guanine nucleotide exchange factors (GEFs) for Arf proteins, but not for Ypt GTPases. The Sec7-domain of Sec7p is sufficient for this activity. Interestingly, the Sec7 domain activity is inhibited by brefeldin A (BFA), a fungal metabolite that inhibits some of the Arf-GEFs, indicating that this domain is a target for BFA. These results demonstrate that the ability to act as Arf-GEFs is a general property of all Sec7-domain proteins in yeast. The genetic interactions observed between Arf GEFs and Ypt GTPases suggest the existence of a Ypt-Arf GTPase cascade in the secretory pathway.

THE movement of membranes and proteins through the secretory pathway involves their orderly progression through a series of intracellular compartments (Palade 1975). Transport between successive compartments is mediated by vesicles that bud from one compartment and fuse with the next, and the components of the machinery used in constitutive vesicular trafficking are conserved from yeast to mammals (Ferro-Novick and Jahn 1994; Rothman and Wieland 1996). Two families of small GTPases, Arfs and Ypt/rab proteins, have an important role in the regulation of vesicular transport and are implicated in vesicle budding and targeting, respectively (Schekman and Orci 1996; Novick and Zerial 1997). However, their specific mechanisms of action are still unknown.

Ypt/rab GTPases are crucial for secretion in both yeast and mammalian cells. Genetic, physiological, and biochemical approaches have demonstrated roles for Ypt/rab proteins in the different steps of the exocytic and endocytic pathways (Lazaret al. 1997; Novick and Zerial 1997). In yeast, several Ypt proteins have been shown to function in the exocytic pathway: Ypt1p in transport from the endoplasmic reticulum (ER) to cis-Golgi and cis- to medial-Golgi; Ypt31p and Ypt32p, which are functional homologues, in exit from the trans-Golgi; and Sec4p in the last step of transport to the plasma membrane (Goudet al. 1988; Segevet al. 1988; Jedd et al. 1995, 1997). A role for Ypt/rab proteins in targeting of secretory vesicles to their respective acceptor compartments has been clearly demonstrated by physiological and in vitro studies (Rexach and Schekman 1991; Segev 1991; Lupashinet al. 1996). In addition, our recent work showing a requirement for Ypt31/32p for exit from the trans-Golgi suggested a role for Ypt/rabs in the budding of vesicles (Jeddet al. 1997). To better understand the function of Ypt31/32p we undertook a genetic approach to identify proteins that interact with these GTPases. We report here that a high-copy suppressor of the ypt31/32 mutant belongs to a family of Arf nucleotide exchangers.

Arf proteins are another class of small GTPases that play a critical role in mediating vesicular transport of proteins through the secretory pathway of both yeast and mammalian cells. The yeast Saccharomyces cerevisiae has two ARF genes that are 96% identical and functionally homologous (Sewell and Kahn 1988; Stearnset al. 1990). A third gene, ARF3, encodes a protein that is less closely related (Leeet al. 1994). In vivo and in vitro studies have demonstrated a role for Arf proteins in regulation of multiple transport steps in yeast and mammalian cells (Stearnset al. 1990; Bomanet al. 1992; Tayloret al. 1992; Dascher and Balch 1994; Zhanget al. 1994; D’Souza-Schoreyet al. 1998; Gaynoret al. 1998). The mammalian Arf1p regulates the reversible association of COPI coat proteins with Golgi membranes, although it is not clear whether Arf1 carries out its function by directly interacting with COPI proteins (Zhaoet al. 1997), or if it acts indirectly by stimulating phospholipase D (Ktistakiset al. 1996; Caumontet al. 1998).

Arfs, like all other GTPases, cycle between GTP- and GDP-bound forms. Conversion from the GDP- to the GTP-bound form is achieved by nucleotide exchange, while the shift from the GTP- to the GDP-bound form is accomplished by GTP hydrolysis. Both nucleotide exchange and hydrolysis play a crucial part in Arf’s function, because mutations that interfere with either are profoundly detrimental to the cell (Dascher and Balch 1994; Kahnet al. 1995). Because these nucleotide exchange and hydrolysis reactions occur at a very low intrinsic rate, factors that stimulate these reactions, guanine nucleotide exchange proteins (GEFs) and GTPase activating proteins (GAPs), are also likely to have an important function in Arf-mediated vesicular transport.

Recently, several Arf-GEFs that share homology with a domain of Sec7p, including the yeast Gea1p, have been identified (Chardinet al. 1996; Peyrocheet al. 1996; Meacciet al. 1997; Morinagaet al. 1997; Franket al. 1998; Klarlundet al. 1998). The so-called Sec7 domain is ∼20 kD and is the only region of significant sequence similarity between Sec7p and the Arf-GEFs. In yeast, the three known members of the Sec7-domain family, Sec7p, Gea1p, and Gea2p, are known to have a role in secretion, because mutations in these genes result in clear secretory defects (Achstetteret al. 1988; Franzusoffet al. 1989; Peyrocheet al. 1996; Wolfet al. 1998). Whether the mammalian Sec7-domain proteins also function in the secretory pathway remains to be seen. Sec7p is known to interact genetically with Arfs (Stearnset al. 1990; Deitzet al. 1996). We wished to determine whether all four yeast Sec7-domain proteins can function as Arf-GEFs.

The fungal metabolite brefeldin A (BFA) is a potent inhibitor of intracellular transport and has been used as a powerful tool to study membrane traffic in eukaryotic cells (Orciet al. 1991; Klausneret al. 1992). It blocks nucleotide exchange by Arf GTPases, which is essential for vesicle budding (Donaldsonet al. 1992; Helms and Rothman 1992). Some, but not all, of the Arf-GEFs are inhibited by BFA (Achstetteret al. 1988; Chardinet al. 1996; Peyrocheet al. 1996; Meacciet al. 1997; Morinagaet al. 1997). We show that the Sec7-domain of Sec7p is sufficient for Arf-GEF activity, and this activity is inhibited by the fungal metabolite brefeldin A (BFA).

Here, we describe the identification of Syt1p, a Sec7-domain protein, as a high-copy suppressor of inactivating mutations in the Ypt31p and Ypt32p GTPases. To better understand the relationship between Ypt proteins and the Sec7-domain proteins we undertook genetic analyses that uncovered interactions between YPT1, YPT31, YPT32, and all four Sec7-domain genes in yeast. We also pursued biochemical approaches to ascertain that Sec7p and Syt1p act as Arf-GEFs as well as to ask whether they influence nucleotide exchange by Ypt GTPases. Our results provide strong evidence for an interaction between Ypt31/32p and Arf-GEFs and raise the possibility that these two families of GTPases may act together to regulate secretion.


Strains, plasmids, and materials: The following yeast strains were used in this study: NSY348 (MATa his4-539 lys2-801 ura3-52 ypt31::HIS3 ypt32-A141D), NSY403 (AFY89; MATα his3-11 leu2-3,112 ura3-1 sec7-4), NSY460 (DBY5886; MATa his3-11,15 ura3-1 trp1-1 sec7-1), NSY125 (DBY1034; MATa his4-539 lys2-801 ura3-52), NSY222 (MATα his4 ura3-52 ypt1-A136D) (Jeddet al. 1995), NSY416 (APY022; MATα his3200 leu2-3,112 ura3-52 lys2-801 gea1-6 gea2::HIS3) (Peyrocheet al. 1996), and NSY508 (NY405; MATa ura3-52 sec4-8) (Goudet al. 1988). Yeast transformations were performed as described (Gietzet al. 1992).

The yeast expression plasmids used in this study that result in high-copy expression of the respective genes from their own promoters are as follows: YEp352-YPT32 (pNS228), a URA3-marked, 2-μm plasmid was constructed by PCR amplification of the YPT32 gene plus its promoter. The following oligonucleotides were used to amplify the open reading frame (ORF) plus 650 bases upstream of the translational start site: upstream, 5′-GCC GGA TCC GCG GCT CTC CCA TCA AGA GAT CAA-3′; downstream, 5′-GCGTC TAGAG GTTAG TAATA AATAA CTTG-3′. BamHI and XbaI sites created by the amplification were used to subclone the fragment into the corresponding sites of plasmid YEp352 (Hillet al. 1986). Plasmid pNS229 is a YEp24-based URA3-marked, 2-μm plasmid containing a genomic insert that includes YPT31. YEp352-SEC7 (pNS230) was made by moving SEC7 to YEp352 from plasmid YEpTA67 (Achstetteret al. 1988) as an SphI fragment. Plasmids pCLJ88 and gms2, gifts from Catherine Jackson, are URA3-marked, 2-μm plasmids carrying GEA1 and GEA2, respectively (Peyrocheet al. 1996). Yeast plasmids for inducible expression of genes under the control of the GAL1/10 promoter are as follows: Plasmid pNS326, for expression of the wild-type YPT1 gene, was described previously (Joneset al. 1995). To construct plasmid pGAL-YPT31 (pNS198) a URA3 marked, CEN plasmid with YPT31 under the control of the GAL1/10 promoter, the YPT31 ORF was PCR amplified with BamHI and XbaI sites added at the 5′ and 3′ ends, respectively. This fragment was first inserted between the BamHI and XbaI of pBluescript and then excised with BamHI and NotI. This fragment was then inserted between the BamHI and NotI sites of pTS395 (a gift from Tim Stearns, Stanford University) to place YPT31 downstream of the GAL1/10 promoter and upstream of the ACT1 transcriptional terminator. Oligonucleotides for the amplification of the YPT31 ORF were as follows: upstream, 5′-GCC GGA TCC ATG AGC AGC GAG GAC TAC GGG-3′; and downstream, 5′-CGC TCT AGA GTC ATT CAC ATG CAA GTG CGC-3′. To construct plasmid pGAL-YPT32 (pNS250), the YPT32 ORF was PCR amplified with BamHI and XbaI ends. This fragment was used to replace YPT31 in pNS198. Oligonucleotides for the amplification of the YPT32 ORF were as follows: upstream, 5′-GCC GGA TCC ATG AGC AAC GAA GAT TAC GGA TAC G-3′; and downstream, 5′-GCGTC TAGAG GTTAG TAATA AATAA CTTG-3′. Templates for both PRC amplifications were plasmid-borne genes. Coding regions were sequenced to confirm that no mutations were introduced during construction.

The His6-T7-tagged Syt1p (pNS271) was constructed by ligating the PCR-amplified SYT1 ORF to the GAL10 promotor in the YGALSET985 vector (Enomotoet al. 1998). The SYT1 ORF, minus the initial ATG, was PCR amplified with XhoI and KpnI sites added at the 5′ and 3′ ends, respectively. Oligonucleotides for the amplification of the SYT1 ORF were as follows: upstream, 5′-G CGC TCG AGA AAT CAG TCT ATT TCT TCG TTG ATT AAG C-3′; and downstream, 5′-GCG GGT ACC TGA CGC ATT CTA TTG TTG GGA GTG-3′. The template was the plasmid-borne gene.

The Sec7-domain, including amino acids 833-1020, of wild-type and mutant forms of Sec7p were expressed as GST-fusion proteins in Escherichia coli. Plasmids for expression of these proteins (pNS231 and pNS232, respectively) were constructed by inserting PCR-amplified fragments with added BamHI and XbaI sites into the corresponding sites of pGEX-KG (Hakes and Dixon 1992). Oligonucleotides used for amplification of the Sec7 domain were as follows: upstream, 5′-GCC GGA TCC GAA TGT ATT GCT ATT TTC AAC AAT AAA CCC-3′; and downstream, 5′-GGC TCT AGA CAT TGC CTG ATG CTG TTC AGA AAT-3′. Templates for the PCR reactions were YepTA67 (wild-type SEC7) and AFB972 (sec7-4). Inserts were sequenced to confirm that no unexpected mutations were introduced during PCR amplification.

All chemical reagents were purchased from Sigma (St. Louis), unless otherwise noted. Protein concentrations were determined by the method of Bradford using reagents from Bio-Rad (Richmond, CA), with BSA as a standard. Pfu polymerase was from Stratagene (La Jolla, CA). Anti-T7-tag antibody was from Novagene (Madison, WI). BFA was from Epicentre Technologies (Madison, WI). BFA analogs were a gift from Dr. J. Donaldson (National Institutes of Health).

Culture conditions: Yeast strains were grown in rich medium (YEP, 1% yeast extract; 2% bactopeptone) or synthetic medium (0.67% yeast nitrogen base without amino acids) supplemented with the appropriate auxotrophic requirements (Roseet al. 1988). Unless otherwise noted, the carbon source was dextrose (2% w/v).

Identification of high-copy suppressors of ypt31Δypt32-A141D: Yeast strain NSY348 was transformed with a 2-μm genomic library (Carlson and Botstein 1982). Transformed cells were plated and shifted directly to the restrictive temperature (35°) to select for temperature-resistant colonies.

Disruption of SYT1 (NSY421) was performed using PCR for replacement of the entire coding region with the selectable kanr gene (Wachet al. 1994). Oligonucleotides contained 35 bp of target homology at their ends and were designed such that the structural gene was precisely deleted. Selection for G418 (0.2 mg/ml) resistance was performed as described (Wachet al. 1994). Deletions were confirmed by PCR on G418-resistant colonies, using the following oligos: GGT TAC TCG CAT CCA CTC CCT (which hybridizes to the region immediately 5′ of the SYT1 locus) and CAC TTG CGA TTG TGT GGC CTG (which hybridizes to the kanr insert).

Preparation of extracts for nucleotide exchange assays: Yeast strain NSY125 carrying 2-μm plasmids as indicated was grown in synthetic medium to an OD600 of ∼1. After washing with buffer 88 [20 mm HEPES, pH 6.8, 150 mm potassium acetate, 5 mm magnesium acetate, 250 mm sorbitol (Bakeret al. 1990)] cells were resuspended in twice the pellet volume of buffer 88 with 1 μm phenylmethylsulfonyl fluoride (PMSF) and PIC (protease inhibitor cocktail: 1 μg/ml each of leupeptin, chymostatin, pepstatin, antipain, and aprotinin) and lysed with glass beads by vortexing for 5 min continuously at 4°. Extracts were cleared of unbroken cells by centrifugation at 1000 × g for 5 min. Protein concentrations were generally 10-15 mg/ml. Yeast cells carrying YGALSET985 or YGALSET985-SYT1 were grown to an OD600 of 0.5 in synthetic medium lacking uracil, with raffinose (2%) as the carbon source. Galactose was added to 2% and the expression of SYT1 was induced for 6-12 hr. Lysates were prepared as above, except that the cells were spheroplasted with zymolyase before glass-bead lysis to improve breakage. His6-tagged Syt1p was batch purified by incubating 10 mg of yeast lysate with 100 μl Ni-NTA agarose (QIAGEN, Chatsworth, CA) equilibrated with 100 mm Tris, 150 mm NaCl, pH 7.4. Beads were washed 3 times each with 1 ml of B88, pH 7.4, and B88, pH 7.4, with 20 mm imidazole, prior to elution of the His6-tagged protein with 300 μl of 100 mm imidazole in B88, pH 7.4. Control samples were prepared in parallel from the strain carrying the empty YGALSET985 plasmid. GST fusion proteins were purified as described (Joneset al. 1995).

Nucleotide exchange assays: Recombinant, myristoylated yeast Arf proteins were purified as previously described (Randazzoet al. 1995). MyrArf1p and myrArf2p were loaded with [3H]GDP as described (Kahnet al. 1995). At the end of the incubation, samples were moved to ice and MgCl2 was added to 10 mm. GDP release assays were carried out on 1 μm Arf-[3H]GDP) in the presence of 1.5 mg/ml azolectin vesicles (Francoet al. 1996) and 1 mg/ml crude yeast or E. coli extract or purified Sec7 domain. The reaction buffer for GDP release assays contained 25 mm HEPES, pH 7.4, 100 mm NaCl, 10 mm MgCl2, 0.5 mm GDP, 0.5 mm GTP, 1 mm DTT, and 100 μg/ml BSA. For GTP binding, Arf2p was loaded with nonradioactive GDP and the reaction buffer contained 10 μm [α-32P]GTP (17 Ci/mmol; Amersham, Arlington Heights, IL). The final reaction volume was 30 μl. Reactions were carried out at 30°, unless otherwise noted. To monitor exchange, 5-μl samples containing 5 pmol Arf were removed and filtered through nitrocellulose (BA85, Schleicher & Schuell, Keene, NH) at timed intervals. GDP release assays were carried out on Ypt proteins as described (Joneset al. 1995) except that substrate concentrations in the reactions were 1 μm.


Sec7-domain genes exhibit unique genetic interactions with ypt mutants: The functional homologues Ypt31p and Ypt32p are required for exit of secretory proteins from the yeast trans-Golgi. Cells deleted for YPT31 and carrying a conditional inactivating ypt32 mutation (ypt31Δ ypt32-A141D) exhibit protein transport defects in the late exocytic pathway, but not in vacuolar protein sorting (Jeddet al. 1997). To identify factors that interact with Ypt31p and Ypt32p, we selected for genes that in high copy can suppress the temperature sensitivity of a ypt31Δ ypt32-A141D mutant. Mutant cells grow well at 26°, but fail to grow at temperatures over 35°. Transformation with a 2-μm-based yeast genomic library (Carlson and Botstein 1982) yielded 10 suppressors in addition to YPT31 and YPT32 from 12,000 transformants (in the three library pools, 50-80% of the plasmids contained inserts). One of the suppressors, designated SYT1 (for Suppressor of ypt3), has been characterized. SYT1 was isolated twice and can support growth of mutant cells at 36° (Figure 1a). Mutant cells transformed with an empty plasmid failed to grow at this temperature. The genomic insert was ∼8 kb and contained two ORFs (YPR095C and YPR096C) and a large fraction of a third ORF (YPR097W). Deletion analysis revealed that ORF YPR095C was responsible for suppression. Specifically, an internal deletion was constructed that removes all but the first 540 bases of the YPR095C ORF, leaving ORF YPR096C and the partial ORF YPR097W intact. This deletion rendered the plasmid (pNS235) incapable of suppressing ypt31Δ ypt32-A141D, thus demonstrating that YPR095C, and not YPR096C or YPR097W, is responsible for suppression of ypt31Δ ypt32-A141D.

Figure 1.

—Genetic interaction of YPT mutants and Sec7-domain genes. (a) Overexpression of SYT1, but not SEC7, GEA1, or GEA2, suppresses loss of Ypt31/32p function. A thermosensitive ypt31Δ ypt32-A141D strain (NSY348) was transformed with high-copy-number 2-μm plasmids to overexpress Sec7-domain genes or wild-type YPT32. Shown are 10-fold serial dilutions of transformed strains incubated at the permissive temperature (26°) and the nonpermissive temperature (35°). The apparent enhancement of the growth defect by 2 μm-SEC7 appears in ∼80% of transformants (24 transformants tested in three independent transformation experiments). This variability is probably due to variability in plasmid copy number in the cell. (b) Enhancement of the growth defect of ypt31/32 mutant cells by overexpressing Sec7p. Strain NSY348 was transformed with high-copy-number 2-μm plasmids to overexpress SEC7, SYT1, or wild-type YPT32 or with an empty plasmid control (YEp). Transformants (two independent transformants for each plasmid) were streaked on selective medium and incubated at the indicated temperatures. (c) Overexpression of SEC7 weakly suppresses the ypt1-A136D mutation. A thermosensitive ypt1-A136D strain (NS-Y222) was transformed with high-copy-number 2-μm plasmids to overexpress the Sec7-domain genes. Transformants were grown on selective medium and incubated at the indicated temperatures. (d) Enhancement of the growth defect of ypt1 mutant cells by overexpressing Gea2p.

Figure 2.

—Alignments of four yeast Sec7-domain proteins. (a) The similarity is restricted to the Sec7 domain of Sec7p, Syt1p, Gea1p, and Gea-2p (shaded region). The level of similarity in this domain as compared to the Sec7 domain of Sec7p is indicated. (b) Alignment of the Sec7 domains of the four yeast Sec7-domain proteins. The Sec7 domains comprise the following amino acid residues: Sec7p, 838-1020; Syt1p, 459-633; Gea1p, 552-758; and Gea2p, 570-755. Sec7 domains were aligned using the MACAW 2.0.5 multiple alignment program (Schuleret al. 1991), using Blosum 62 parameters and mean-score shading. Two blocks of particularly strong homology, indicated with capital letters, were identified using the sequence-pair overlap and Gibbs sampler methods. The asterisk denotes the position of the sec7-4 mutation (G883D; A. Franzusoff, unpublished results).

SYT1 encodes a protein with a predicted molecular weight of ∼135 kD that contains a Sec7-domain (Figure 2). Unlike the other three Sec7-domain genes in yeast, SYT1 is not essential, because deletion of the gene had no effect on viability of cells. There was no detectable phenotype associated with the SYT1 deletion at any temperature on rich medium. A search of the yeast genome database revealed no obvious homologues of Syt1p besides the other Sec7-domain proteins, Gea1p, Gea2p, and Sec7p. Homology between Syt1p and the other Sec7-domain proteins falls entirely within the Sec7 domain (Figure 2). The Sec7 domain of Syt1p is more closely related to that of Gea1p/Gea2p than to the Sec7 domain of Sec7p (25 and 20% similarity, respectively). To determine whether the other yeast Sec7-domain proteins show genetic interactions with Ypt31/32p, we compared the effects of overexpressing each of the four yeast Sec7-domain proteins on the growth phenotype of ypt31Δ ypt32-A141D mutant cells. Only Syt1p overexpression suppressed the temperature sensitivity of the ypt31/32 mutant (Figure 1a). Overexpression of Sec7p actually had a strong negative effect on growth of this mutant. Cells transformed with a high-copy plasmid carrying SEC7 grew more slowly at the permissive temperature (Figure 1a) and had a lower restrictive temperature (32° rather than 35°) than cells harboring the empty plasmid control (Figure 1b). Overexpression of Gea1p or Gea2p from a 2-μm plasmid had a very mild negative effect on growth of ypt31/32 cells (Figure 1a). Overexpression of SEC7, GEA1, GEA2, or SYT1 had no effect on the growth of wild-type cells. Thus, two of the four yeast Sec7-domain genes exhibit clear genetic interactions with YPT31/32. While overexpression of SYT1 suppresses the ypt31/32 mutation, overexpression of SEC7 enhances the severity of this mutation. Both types of genetic interaction, suppression and enhancement of mutant phenotypes, suggest functional interaction of the gene products in vivo.

Figure 3.

—(a and b) YPT1 and YPT31/32 exhibit allele-specific suppression of sec7 mutations and (c) have distinct effects on gea1 mutant cells. Temperature-sensitive yeast cells carrying (a) sec7-4, a mutation within the Sec7 domain (NSY403), (b) sec7-1, a mutation outside of the Sec7 domain (NSY460), or (c) gea1-6 (NSY416) were transformed with plasmids to overexpress the exocytic YPT genes. Overexpression was achieved by (a and b) high-copy 2-μm constructs or (c) use of the strong inducible GAL10 promoter. Ten-fold serial dilutions of transformants were spotted onto selective media and incubated for 4 days at the temperatures indicated.

We also tested the effect of overexpression of the Sec7-domain genes on ypt1-A136D and sec4-8 mutant cells. GEA2 overexpression had a negative effect on the growth of ypt1 mutant cells even at permissive temperature, while the closely related GEA1 did not affect growth of this strain (Figure 1, c and d). The difference between the effect of GEA1 and GEA2 on ypt1 mutant cells may reflect a difference in the level of expression rather than a functional difference between the two genes. Overexpression of SEC7 resulted in weak suppression of ypt1-A136D, while SYT1 overexpression had no effect (Figure 1c). In tests with a sec4-8 mutant strain, overexpression of SEC7 had a weak inhibitory effect at the semipermissive temperature of 30°; the other Sec7 domain genes had no effect.

YPT genes exhibit unique genetic interactions with mutants of Sec7-domain genes: We wished to determine whether the genetic interactions between Ypt proteins and the Sec7 domain proteins are reciprocal, because a true physiological interaction is frequently manifested by reciprocity of genetic interactions. We tested the effect of overexpression of YPT genes on growth of cells carrying mutations in SEC7 or GEA1/2. First, we tested whether overexpression of YPT31 or YPT32 could suppress the temperature sensitivity caused by two different sec7 mutations (A. Franzusoff, unpublished results). Overexpression of YPT31/32 suppressed sec7-4, a mutation within the Sec7 domain (Figure 2 and Figure 3a), but failed to suppress sec7-1, a mutation outside this region (Figure 3b). sec7-4 mutant cells transformed with a 2-μm plasmid carrying YPT31 or YPT32 grew at temperatures up to 35°, while those harboring an empty plasmid control failed to grow at this temperature. Cells transformed with the wild-type SEC7 gene grew well at temperatures up to 37°. To determine whether this suppression is unique to YPT31/32 or general to the other exocytic YPT genes, we examined the effects of overexpressing YPT1 and SEC4. Only overexpression of YPT31/32, but not YPT1 or SEC4, can suppress the growth phenotype of the sec7-4 mutation at 35°. In contrast, sec7-1 cells grew at 35° when transformed with plasmids overexpressing wild-type SEC7 or YPT1 (Figure 3b), but not YPT31/32 or SEC4. Therefore, overexpression of YPT31/32 results in allele-specific suppression of a Sec7-domain mutation, and the ability to suppress this mutation is unique to YPT31/32 among the exocytic YPT genes. In addition, these results show that the genetic interaction between the YPT31/32 and SEC7 genes is reciprocal, though the effects are opposite. Reciprocal genetic interaction is a strong indication of a physiological interaction of gene products in vivo. Allele specificity of suppression is a further indication of the significance of this interaction. Moreover, because we show that the sec7-4 mutation severely reduces the Arf-GEF activity of the Sec7 domain (see below), this allele-specific suppression suggests that the interaction between Ypt31/32p and Sec7p is related to the function of the Sec7-domain.

The genetic interaction observed between YPT1 and SEC7, suppression of a mutation outside the Sec7-domain of SEC7 by overexpression of YPT1 (Figure 3b), is different from the interaction that YPT31/32 exhibit with SEC7. YPT1 has a role early, while YPT31/32 are required late in the secretory pathway. The allele-specific differences in genetic interactions of SEC7 with YPT1 and YPT31/32 may point to different physiological interactions of Sec7p with these GTPases.

We also tested the ability of overexpression of YPT genes to affect the growth of a gea1-6 gea2Δ mutant strain (Peyrocheet al. 1996). Strong overexpression of YPT31 and YPT32, under the control of the inducible GAL10 promotor, exacerbated the growth defect of the gea1/2 mutant, lowering the restrictive temperature from 35° to 33° (Figure 3c). Overexpression of YPT31 and YPT32 from 2-μm plasmids (under the control of their own promoters) had a similar, but weaker, effect. Therefore, there is a reciprocal negative effect between YPT31/32 and GEA1/2. This does not seem to be the case with YPT1 and GEA1/2. Overexpression of SEC4 from a 2-μm plasmid had no effect on the growth of any of the mutants associated with Sec7 domain genes.

Sec7p and Syt1p can stimulate nucleotide exchange by Arf but not Ypt GTPases: The ability of several Sec7-domain proteins, including the yeast Gea1p, to stimulate nucleotide exchange on Arf GTPases has been demonstrated (see Introduction). If the two Sec7-domain proteins that show genetic interactions with Ypt31/32p, Sec7p and Syt1p, possess Arf-GEF activity, the results described above would suggest a connection between Ypt and Arf GTPases. Therefore, we wished to determine whether the ability to stimulate nucleotide exchange by Arfp is a general property of all Sec7-domain proteins in yeast. To assess their ability to act as Arf-GEFs, we assayed the effect of overexpressing each of the Sec7-domain proteins from a high-copy 2-μm plasmid on release of GDP from Arf proteins. Lysates of yeast cells that do not overexpress any Sec7-domain protein stimulated release of GDP from myrArf2p, and overexpression of any one of the Sec7-domain proteins stimulated this Arf-GEF activity further. Overexpressing GEA1 yielded the greatest stimulation of nucleotide exchange. The rate of [3H]GDP release from myrArf2p was ∼5.5-fold faster in the presence of extract from the GEA1 overexpresser strain than with the control (Figure 4a). Overexpression of GEA2, SEC7, or SYT1 stimulated [3H]GDP release 2- to 2.5-fold. The 2-μm plasmid that overexpresses Syt1p (pNS233) also contains one other ORF (YPR096C) and a part of a third (YPR097W). To ascertain that the effect was due to overexpression of Syt1p, an internal deletion was constructed that removes a large portion of the SYT1 ORF (see above). Extracts of cells carrying the plasmid deleted for the SYT1 ORF did not stimulate GDP release above the level of the empty plasmid control (Figure 4b), confirming that overexpression of Syt1p was necessary for the observed Arf-GEF activity. Thus, the ability to stimulate GDP release is a common feature of all four Sec7-domain proteins in yeast.

Figure 4.

—Increased stimulation of GDP release from Arf2p by overexpression of the four yeast Sec7-domain proteins. (a) Extracts from cells carrying 2-μm plasmids expressing Gea1p, Gea2p, Sec7p, or Syt1p were tested for Arf-GEF activity. MyrArf2p (1 μm), loaded with [3H]GDP, was incubated with 1 mg/ml crude lysate from wild-type yeast cells (NSY125) transformed with the various 2-μm plasmids. Release of [3H]GDP was measured by nitrocellulose filtration of 5 μl samples at the indicated times. The unstimulated (intrinsic) loss of [3H]GDP, measured in the absence of extract (in the presence of BSA), was subtracted from each point. The intrinsic rate of GDP release was 0.76%/min. Dotted line, empty plasmid control (YEp352); ▪, Gea1p (pCLJ88); □, Gea2p (gms2); •, Syt1p (pNS233, YEp24-SYT1); ○, Sec7p (pNS230). (b) Increased Arf-GEF activity in extracts from cells carrying SYT1 on the 2-μm plasmid is due to overexpression of Syt1p. Cells carrying a 2-μm-SYT1 plasmid with or without an internal deletion of the SYT1 ORF (pNS233, YEp24-SYT1, •; pNS235, YEp24-SYT1Δ, ▪) or empty plasmid control (YEp352, ○) were tested for Arf-GEF activity as in a. Data presented are the averages of duplicate measurements from two independent experiments and are representative of experiments with more than one lysate preparation. Error bars are the average deviation from the mean.

To show more directly that Syt1p is an Arf-GEF we constructed a His6-T7-tagged version of this protein. The tagged protein was expressed in yeast cells and was purified by binding to Ni-NTA agarose and elution with imidazole. Immunoblot analysis verified that the protein was expressed and eluted (Figure 5a). The eluted fraction possesses Arf-GEF activity as shown by stimulation of both GDP release by Arf2p (data not shown) and GTP binding to Arf2p by the Syt1 eluate, but not by the control eluate (Figure 5b).

To further characterize the Arf-stimulating GDP-release activity of Sec7p and show that the increase of Arf-GEF activity in the yeast lysates is a direct effect of overexpression of Sec7p, we purified the Sec7 domain of Sec7p and studied its ability to stimulate nucleotide exchange. Previous work showed that the Sec7 domain of ARNO is sufficient to stimulate nucleotide exchange on Arf proteins (Chardinet al. 1996). The Sec7 domain of Sec7p was purified as a GST fusion from E. coli and the GST moiety was removed by digestion with thrombin. We assayed the ability of the purified Sec7 domain to stimulate nucleotide exchange on myrArf2p, measuring GDP release as well as concurrent GTP binding. Rates for both reactions were similar, indicating that this domain does, in fact, act as a true nucleotide exchange factor (Figure 6, a and b). As seen in Figure 6a, ∼50% of the nucleotide-bound Arf2p exchanges the nucleotide within 30 min (only ∼10% of the total Arf2p in the reaction binds nucleotide). Therefore, the Sec7 domain of Sec7p can act as an Arf-GEF.

Figure 5.

—Syt1p is an Arf-GEF. (a) Purification of His6-T7-tagged Syt1p from yeast extracts. Extracts were prepared from yeast cells carrying control plasmid (YGALSET 985) or His6-T7-tagged Syt1p plasmid (pNS271) under the GAL10 promoter that was induced for the expression of the tagged protein. Lysates (L) were incubated with Ni-NTA agarose beads and bound proteins were eluted with imidazole (E). The presence of the His6-T7-tagged Syt1p was detected using immunoblot analysis and anti-T7 antibody. (b) His6-T7-tagged Syt1p possesses Arf-GEF activity. The eluted fractions from panel a were tested for the ability to stimulate GTP binding by Arf2p. myrArf2p (1 μm), loaded with nonradioactive GDP, was incubated with 10 μl control (○) or Syt1p eluate (□) in the presence of [32P]GTP at 30°. Binding of [32P]GTP was measured by nitrocellulose filtration of 5-μl samples (5 pmol Arf2p) at the indicated times. Unstimulated (intrinsic) nucleotide exchange, measured in the absence of Syt1p (in the presence of BSA), was subtracted from each point. GTP binding to the eluates in the absence of Arf2p was negligible and was subtracted out. GTP bound at the onset of the reaction (T = 0 min) was also subtracted from each point. The intrinsic rate of GTP binding was 7.7 fmol/min. Results are the average of two independent measurements and are representative of three experiments. Error bars represent the range divided by two.

Because SYT1 was identified as a suppressor of ypt31/32 mutations and SEC7 also interacts genetically with YPT31/32 and YPT1 mutations, it seemed possible that the Sec7-domain proteins might affect nucleotide exchange by Ypt/rab GTPases as well as Arfs. Lysates of yeast cells overexpressing Sec7p or Syt1p or of E. coli cells expressing a GST-Sec7-domain fusion protein (compared to GST alone) were tested for effects on nucleotide exchange activity on a number of recombinant Arfs and Ypt-GTPases. Stimulation of [3H]GDP release by Sec7p, Syt1p, or the Sec7 domain of Sec7p was observed only in the case of Arf-family substrates (Figure 7). Thus, Sec7p and Syt1p apparently act as nucleotide exchange factors only for the Arf family of GTPases. Moreover, suppression of the ypt31Δ ypt32-A141D mutant phenotype by overexpression of Syt1p does not seem to be due to stimulation of Ypt32p nucleotide exchange by Syt1p.

A mutation within the Sec7 domain of Sec7p severely reduces Arf-GEF activity: Because Ypt31/32p overexpression can specifically suppress the sec7-4 mutation, which was shown to reside within the Sec7 domain of Sec7p (A. Franzusoff, unpublished results), we wished to determine whether this inactivating mutation affects the ability of the Sec7 domain to act as an Arf-GEF. Lysates from the sec7-4 mutant yeast strain grown at permissive temperature or shifted to nonpermissive temperature showed no reduction in Arf-GEF activity relative to wild-type controls. However, because deletion of SYT1 also yielded no reduction in Arf-GEF activity, we reasoned that loss of one Arf-GEF may be compensated for in vitro by activity of the others present in the lysate. To directly assess the effect of a Sec7-domain mutation, sec7-4, on Arf-GEF activity, we cloned the Sec7 domain of the sec7-4 mutant and purified the mutant Sec7 domain as a GST fusion from E. coli. To protect against possible thermolability of the sec7-4 mutant protein (sec7-4 yeast mutant cells are temperature sensitive), fusion proteins were induced at 30°. The mutant fusion protein was expressed as abundantly as the wild-type GST-Sec7 domain, and the thrombin cleavage products were equally stable as assessed by SDS-PAGE and Coomassie blue staining (data not shown). GEF activity of the mutant Sec7 domain at 30° was severely compromised, promoting nucleotide exchange at only 20% of the wild-type level (Figure 8a). Lowering the assay temperature to 20° did not increase Arf-GEF activity of the mutant Sec7 domain (data not shown). These results indicate that the Arf-GEF activity of the Sec7 domain of Sec7p is severely diminished by a point mutation in this domain. This dramatic reduction in Arf-GEF activity is likely to be the cause of the temperature sensitivity in sec7-4 yeast strains. In addition, this result supports the idea that the specific suppression of the sec7-4 but not the sec7-1 mutation by overexpression of Ypt31/32p is related to the function of the Sec7 domain as an Arf-GEF.

Figure 6.

—The Sec7 domain of Sec7p is sufficient to stimulate exchange of nucleotides bound to Arf2p. (a) [3H]GDP released from Arf2p; release of [3H]GDP as fmol (left axis) and as a percentage of the Arf2p:[3H]GDP present at the beginning of the reaction (right axis). GDP release was measured as in Figure 4 in the presence of purified Sec7 domain (0.5 μm, □; and 3.5 μm, ○). (b) [32P]GTP binding to Arf2p was measured as in Figure 5b. The intrinsic rates of GDP release and GTP binding were 7.75 and 17.1 fmol/min, respectively. Results are the average of two independent measurements and are representative of four experiments with two preparations of Sec7-domain protein. Error bars represent the range divided by two.

BFA inhibits activity of the purified Sec7 domain: BFA inhibits a subset of Sec7 domain Arf-GEFs (Donaldsonet al. 1992; Helms and Rothman 1992; Chardinet al. 1996; Peyrocheet al. 1996; Meacciet al. 1997; Morinagaet al. 1997). To test whether the Sec7 domain itself is involved in BFA sensitivity, we assayed Sec7-domain stimulated nucleotide exchange in the presence of BFA. We found that the purified Sec7 domain of Sec7p was itself sensitive to BFA, and that the inhibition increased linearly to ∼35% with increasing BFA concentrations (Figure 8b). It was not feasible to increase the BFA concentration above 600 μm as the solvent becomes inhibitory at higher levels. As a control for the specificity of this effect of high concentrations of BFA, two inactive BFA analogs [B17 and B36 (Donaldsonet al. 1992)] were also tested at 600 μm. Neither analog yielded significant inhibition of Sec7-domain-stimulated nucleotide exchange. Thus, the observed BFA effect is specific. The sensitivity to BFA suggests that the Sec7 domain itself may be a target for this drug.


We have observed a variety of genetic interactions between Sec7-domain Arf-GEFs and Ypt GTPases. Three Sec7-domain proteins had previously been described in yeast: Sec7p, Gea1p, and Gea2p. We have identified Syt1p, the most distant relative in this family as a high-copy suppressor of a ypt31/32 mutation. All of these proteins are large (130-200 kD) and share homology in a 20-kD region termed the Sec7 domain. Because the whole yeast genome has been sequenced and no other proteins show homology to the Sec7 domain using BLAST searches, these four Sec7-domain proteins probably compose the whole family of these proteins in yeast. One of them, Gea1p, has been shown previously to possess Arf-GEF activity (Peyrocheet al. 1996). However, it had not been obvious that the other Sec7-domain proteins would also have this activity, because the similarity shared in the Sec7 domain ranges between 20-35%, and this domain comprises only about 10% of these proteins. In this study we demonstrate that Sec7p, Syt1p, and Gea2p also have Arf-GEF activity in yeast cell lysates. We also show that the purified Syt1p and Sec7 domain of Sec7p can stimulate nucleotide exchange, indicating that the increased Arf-GEF activity in the yeast cell lysates is a direct effect of overexpressing the Sec7-domain proteins. Therefore, all the members of the Sec7-domain family in yeast have the ability to activate Arf GTPases in vitro.

Figure 7.

—Syt1p, Sec7p, and the Sec7 domain of Sec7p stimulate release of GDP from Arfs, but not Ypt-GTPases. GDP release assays were carried out in the presence of (a) crude yeast lysate from wild-type cells (NSY125) transformed with 2-μm plasmid carrying SYT1 (pNS233; gray bars), SEC7 (pNS230; black bars), or 2 μm alone (YEp352; white bars) or (b) extracts of E. coli cells expressing either a GST-Sec7-domain fusion (gray bars) or GST alone (white bars). MyrArf1p, myrArf2p, Ypt1p, Ypt31p, or Ypt32p (1 μm), loaded with [3H]GDP, was incubated with 1 mg/ml lysate (for myrArf1p, myrArf2p, or Ypt1p) or 2 mg/ml lysate (for Ypt31/32p). Release of [3H]GDP was measured as in Figure 4. Activities from the linear range of the assay are expressed relative to activities in control extracts (empty plasmid, in the case of yeast lysates, or E. coli expressing GST alone). Results are the average of two to six independent measurements. Error bars represent the standard error of the mean. Independently prepared extracts gave equivalent results.

Figure 8.

—The Arf-GEF activity of the Sec7 domain of Sec7p is inhibited by a mutation within the domain, and by BFA. (a) A mutation within the Sec7 domain of Sec7p, sec7-4, diminishes Arf-GEF activity of the purified domain. GDP-release assays were carried out with purified wild type (▪, •) and sec7-4 mutant (□, ○) Sec7 domains, as described in the legend to Figure 4, except that 1 μg (•, ○) and 2 μg (▪, □) of purified protein were used. Results presented are the averages of three independent measurements; error bars represent the standard error of the mean. Two preparations of wild type and mutant Sec7 domain gave equivalent results. (b) Increasing concentrations of BFA, but not of inactive BFA analogs, inhibit the Arf-GEF activity of the purified domain. Stimulation of GDP release from Arf2p by 3.5 μm Sec7 domain was measured in the presence of an increasing amount of BFA dissolved in DMSO (▪). As a control, 600 μm of BFA analogs B17 (□) or B36 (○) were also tested for inhibition of nucleotide exchange. Intrinsic exchange in the absence of Sec7 domain was subtracted from each point. The presence of BFA had no effect on the intrinsic exchange.

We show here that the sec7-4 mutation severely diminishes the Arf-GEF activity of the Sec7 domain. This mutation changes a glycine (at position 883 of Sec7p), well-conserved among the Sec7-domain proteins, except those in yeast, to an aspartic acid (A. Franzusoff, unpublished results) The structures of the Sec7-domains of ARNO and cytohesin-1 have recently been solved (Betzet al. 1998; Cherfilset al. 1998; Mossessovaet al. 1998). The structures are similar and their analysis reveals a cleft formed by two well-conserved motifs that may be the Arf-binding region. This hypothesis is supported by the finding that mutations within this cleft dramatically reduced the Arf-GEF activity of the Sec7 domain. The mutation encoded by the sec7-4 allele introduces a change outside of the cleft. In contrast to the effects we observed with the sec7-4 mutant domain, a mutation in the adjacent glutamate (E117K in the ARNO Sec7 domain) had no effect on GEF activity (Cherfilset al. 1998). Assuming that the Sec7 domain of Sec7p has a similar structure, this difference is most likely due to the fact that while the ARNO mutation is relatively conservative, the sec7-4 mutation introduces a much larger side group. Because these residues are in close proximity to the putative Arf-binding cleft, it is possible that the sec7-4 mutation affects the structure of the cleft, thereby abrogating GEF activity.

We report here that the Arf-GEF activity of the Sec7 domain of Sec7p is sensitive to BFA. This result suggests that the Sec7 domain itself may be a target for this drug. This is surprising in light of the fact that not all Sec7-domain Arf-GEFs are sensitive to BFA and suggests that other regions of the proteins may play a role in protecting the Sec7 domain from BFA. Results similar to those we present were published recently (Sataet al. 1998). Because a relatively high concentration of BFA produced a maximum of 35% inhibition, and the concentration of BFA that is required to inhibit protein transport in vivo is ∼15-fold lower, it is also possible that other factors or regions of the protein act to enhance sensitivity to BFA.

Sec7p and Gea1/2p clearly have roles in the regulation of secretion in yeast (Esmonet al. 1981; Stevenset al. 1982; Franzusoff and Schekman 1989; Peyrocheet al. 1996). Whether Syt1p plays a significant role in secretion remains to be shown. Because deletion of SYT1 does not affect growth, it is evidently not an essential player. However, the ability of Syt1p both to act as an Arf-GEF and to suppress a ypt31/32 mutation suggests that Syt1p does have a role in protein transport. Is the Arf-GEF activity of Sec7-domain proteins required for their role in the regulation of protein transport? We found that the Sec7-domain mutation, sec7-4, which causes conditional growth and secretory defects (Deitzet al. 1996), dramatically reduced the level of Arf-GEF activity of the purified Sec7 domain in vitro. Assuming that the Arf-GEF activity of the sec7-4 mutant protein is also defective in vivo, this result suggests that the Arf-GEF activity of Sec7p is required for its role in protein transport.

Figure 9.

—Summary of genetic interactions described in this article. (a) Genetic interactions due to overexpression between YPT and Sec7-domain genes: Suppression is denoted by arrows, while enhancement of a phenotype is shown by a flat arrow (dashed lines denote weak genetic interaction). (b) A model for a GTPase cascade in the yeast secretory pathway. Overexpression of each gene can suppress mutations in genes immediately upstream and can enhance the phenotype of mutations in downstream genes. Only interactions thought to be related to the function of the Sec7 domain are shown. Thickness of the arrows indicates strength of interactions.

All four Sec7-domain Arf nucleotide exchangers interact genetically with the Ypt31/32 GTPases (Figure 9a). Overexpression of Ypt31/32p has opposite effects on mutations in different Sec7-domain genes, enhancing the growth defect caused by gea1/2 mutations and suppressing the growth defect caused by a sec7 mutation. Moreover, suppression of sec7 is allele specific. YPT31/32 overexpression suppresses the Sec7-domain mutation sec7-4, which severely diminishes the Arf-GEF activity of this domain, but not the sec7-1 mutation, which falls outside of this domain, suggesting that the functional link between these gene products is based on the Arf-GEF activity of the Sec7 domain. Overexpression of Sec7-domain genes also has distinct effects on the growth of ypt31/32 mutants. While overproduction of Syt1p suppresses the growth defect of a ypt31Δ ypt32-A141D strain, increased Sec7p expression exacerbates this defect. Gea1p and Gea2p overexpression have very mild negative effects. The fact that all Sec7-domain proteins are related by their biochemical activity, but exhibit distinct genetic interactions, suggests that Arf-GEF activity is not the sole determinant for genetic interaction with Ypt GTPases.

Sec7p and Gea2p also interact genetically with the Ypt1 GTPase. Here we show that overexpression of YPT1 suppresses sec7 mutations in an allele-specific manner (sec7-1 but not sec7-4). Overexpression of SEC7 weakly suppresses the ypt1-A136D mutant phenotype. Conversely, overexpression of GEA2 inhibits the growth of ypt1 mutant cells even at the permissive temperature. Interestingly, overexpression of GEA1, the functional homologue of GEA2, does not have this effect. In addition, synthetic lethality between YPT1 and both ARF1 and SEC7 have been reported previously (arf1Δ and ypt1-1, Stearnset al. 1990; sec7-1 and ypt1-1, Baconet al. 1989; sec7-4 and ypt1-3, Lupashinet al. 1996). The interaction of SEC4 with the genes encoding Sec7-domain proteins is limited to a very mild negative effect of overexpression of SEC7 in a sec4-8 mutant.

What might these genetic interactions mean? It has been shown that genetic interactions between SEC genes define two distinct sets that correspond to the groups involved in vesicle formation or vesicle targeting/fusion (Kaiser and Schekman 1990; Lupashinet al. 1996). The genetic interactions described here between Ypt31/32p and the two Sec7-domain proteins support the possibility that Ypt31/32p have a role in vesicle budding, because both Sec7p and Syt1p can act as Arf-GEFs, and Arf GTPases have a role in this process. We have shown previously that Ypt1p is required for two successive steps of the yeast secretory pathway. On the basis of this result, we have suggested that Ypt/rab proteins determine the specificity of individual compartments, rather than secretory steps; in the case of Ypt1p it would be the cis-Golgi compartment (Jeddet al. 1995). In the first step, ER to cis-Golgi, Ypt1p is necessary for targeting of vesicles to the acceptor compartment (Rexach and Schekman 1991; Segev 1991; Lupashinet al. 1996). However, the role of Ypt1p in the cis-to-medial Golgi step is not yet understood. The interaction of Ypt1p and Sec7p might point to a role for Ypt1p in vesicle budding from the cis-Golgi, or to a different role for Sec7p in vesicle fusion (Lupashinet al. 1996). The very limited interaction of Sec4 GTPase with Sec7p suggests that the role of Sec4p may be limited to vesicle fusion.

Differences in genetic interactions exhibited by the Sec7-domain genes with respect to ypt mutations, and by the YPT genes with respect to mutants of Sec7-domain genes, may indicate that these Arf-GEFs, like the Ypt proteins, normally act in different compartments. Physiological and morphological analyses of arf mutants suggest a role for Arf at multiple steps of the secretory pathway (Gaynoret al. 1998). Compartmentalization of Arf-GEFs would provide a mechanism for specifying the site of Arf recruitment and therefore of Arf action in the different secretory compartments. Interaction with specific Ypt/rab proteins that are functionally compartmentalized would impart greater precision in determination of the site of Arfp action. This might be especially important in yeast, which only has two Arf proteins.

In summary, we have observed unique genetic interactions between YPT genes and Sec7-domain genes encoding Arf-GEFs (Figure 9). In some cases the interaction is allele specific, and, while not reciprocal in the classical sense, effects are observed in both directions. Allele-specific interactions usually indicate that the proteins encoded by these genes interact, although not necessarily directly. We also show here that Sec7-domain proteins do not act as GEFs for the exocytic Ypt GTPases. The combined data suggest the possibility that these two families of GTPases interact in a regulatory cascade to direct secretion. Several examples of GTPase cascades have been established in which GTPases are linked directly, or through the action of accessory proteins such as exchange factors, to regulate different cell processes. These processes include cell morphogenesis and cell motility (Chant and Stowers 1995). In some cases, these accessory factors have different effects on the various GTPases involved. For example, two GTPases that interact to regulate bud formation in yeast, Bud1p and Cdc42p, are linked by Cdc24p, which serves as an inhibitor of GTP hydrolysis by Bud1/Rsr1p and acts as a GEF for Cdc42p (Zheng et al. 1994, 1995). The genetic interactions described here between Ypt proteins and Sec7-domain Arf-GEFs are complex, and the means by which they might interact biochemically is not immediately obvious. It is possible that Ypt proteins act upstream of Arf proteins to regulate their activity, or vice versa. It seems likely that the effect is through nucleotide cycling. In Figure 9b we present a model that can explain the different genetic interactions described in this article. In this model, Ypt GTPases and Sec7-domain Arf-GEFs function in successive steps of a cascade. The genetic interactions observed can be explained if overexpression of a protein can compensate for the defect of a mutant protein in the step immediately upstream (e.g., through shared interactions with Arf GTPases) and enhance the defect of a mutant protein in a step downstream (e.g., by sequestration of an essential factor). The challenge now is to determine the mechanism by which Ypt and Arf GTPases interact to regulate vesicular transport.


We are grateful to B. Glick, M. Hochstrasser, and R. Dubreuil for helpful discussions and critical reading of the manuscript. We thank Catherine Jackson for sending GEA1/GEA2 mutant strains and plasmids, Julie Donaldson for a generous gift of BFA analogs, Tim Stearns for sending plasmid pTS395, and Xinjun Zhu for help in sequencing the Sec7-domain constructs. G.J. was supported by The National Institutes of Health (NIH) predoctoral training grant No. 5T32 GM-07151-20. This research was supported by grant GM-45444 from NIH to N.S. and GM-55823 from the National Institute of General Medical Sciences to R.K.


  • Communicating editor: E. W. Jones

  • Received August 17, 1998.
  • Accepted April 23, 1999.


View Abstract