A Novel Ras Inhibitor, Eri1, Engages Yeast Ras at the Endoplasmic Reticulum

Abstract

Ras oncoproteins are monomeric GTPases that link signals from the cell surface to pathways that regulate cell proliferation and differentiation. Constitutively active mutant forms of Ras are found in ca. 30% of human tumors. Here we report the isolation of a novel gene from Saccharomyces cerevisiae, designated ERI1 (for endoplasmic reticulum-associated Ras inhibitor 1), which behaves genetically as an inhibitor of Ras signaling. ERI1 encodes a 68-amino-acid protein that associates in vivo with GTP-bound Ras in a manner that requires an intact Ras-effector loop, suggesting that Eri1 competes for the same binding site as Ras target proteins. We show that Eri1 localizes primarily to the membrane of the endoplasmic reticulum (ER), where it engages Ras. The recent demonstration that signaling from mammalian Ras is not restricted to the cell surface but can also proceed from the cytoplasmic face of the ER suggests a regulatory function for Eri1 at that membrane.

The Ras family of small GTPases comprises a group of molecular switches that, in the GTP-bound active state, transmit signals by interaction with effector proteins (44). Ras proteins play an important role in the transduction of external signals that regulate cell proliferation, differentiation, and metabolism (reviewed in reference 33). Constitutively GTP-bound mutants of Ras are found in ca. 30% of human tumors (25). In cases of colorectal or pancreatic cancers, this incidence is as high as 50 or 90%, respectively (11). The three forms of mammalian Ras (H-Ras, K-Ras, and N-Ras) are identical through their N-terminal 85 residues, which possess the structural features necessary for guanine nucleotide exchange, GTPase activity, and effector function (11). Moreover, they all interact with the same set of effectors. The best studied of these are the Raf protein kinases, whose stimulation by Ras activates the ERK mitogen-activated protein (MAP) kinase cascade, the core of the major proliferative pathway in metazoan cells (45). Other Ras effectors include phosphoinositide 3-kinase (PI-3K [44]), which prevents apoptosis through the Akt protein kinase, Ral-GEF, the guanine nucleotide exchange factor for Ral-GTPases (10a), and the 14-3-3 binding protein RIN1 (42). In all cases, effector association is dependent not only on the nucleotide-bound state of Ras but also on an intact effector loop, a highly conserved region that is exposed for effector interaction through a conformational change induced by GTP binding (45).

The budding yeast Saccharomyces cerevisiae possesses two functionally overlapping RAS genes, which share 84% sequence identity with their mammalian counterparts though their N-terminal domains (17). Although the essential effector of yeast Ras is adenylyl cyclase (39), an enzyme not regulated by Ras in animal cells, loss of yeast Ras function can be complemented by expression of mammalian Ras genes (9, 16). Conversely, a mutationally activated allele of yeast RAS1 is capable of causing malignant transformation of mouse fibroblasts (9). The conservation of biological properties among these members of the Ras gene family is reflected in the observation that the effector loops of yeast Ras are identical to those of mammalian Ras.

Newly synthesized Ras undergoes a series of evolutionarily conserved posttranslational modifications at its C-terminal CAAX motif that render it more hydrophobic (3, 6, 8, 29, 32). The first modification, prenylation of the CAAX cysteine, targets Ras to endomembranes (6). The next two steps, proteolytic removal of the AAX residues and carboxymethylation of the prenylated cysteine, occur at the endoplasmic reticulum (ER) prior to transit of the mature form to the plasma membrane (PM [27]). In contrast to conventional cargo carried in vesicle lumens, Ras must be transported on the cytoplasmic surface of vesicles. This leaves open the possibility that Ras can undergo nucleotide exchange and interact with its effectors while associated with endomembranes. Indeed, Chiu et al. (5) demonstrated that signaling from mammalian H-Ras and N-Ras is not restricted to the PM as previously thought but can proceed from the ER and Golgi compartments, resulting in differential activation of its various signaling pathways. However, nothing is known about Ras regulation at endomembranes. Here, we describe a novel yeast gene encoding a Ras inhibitor that engages Ras at the ER.

MATERIALS AND METHODS

Strains and growth conditions.

The S. cerevisiae strains used in the present study are listed in Table 1. Yeast cultures were grown in YEPD (1% Bacto yeast extract, 2% Bacto Peptone, 2% glucose) with or without 10% sorbitol. Synthetic minimal (SD) medium (31a) supplemented with the appropriate nutrients was used to select for plasmid maintenance and gene replacement. Escherichia coli DH5α was used to propagate all plasmids. E. coli cells were cultured in Luria broth medium (1% Bacto Tryptone, 0.5% Bacto yeast extract, 1% NaCl) and transformed to carbenicillin resistance by standard methods.

TABLE 1.

S. cerevisiae strains used in this study

Straina	Relevant genotype	Source or reference
1783	MATaleu2-3,112 trp1-1 ura3-52 his4 can1r	I. Herskowitz
1784	MATα	I. Herskowitz
1788	MATa/α	I. Herskowitz
DL838	MATaras2::LEU2	This study
DL2297	MATaira1Δ::LEU2	36
DL2522	MATaeri1Δ::TRP1	This study
DL2524	MATa/α eri1Δ::TRP1/eri1Δ::TRP1	This study
DL2566	MATaeri1Δ::TRP1 ras2::LEU2	This study
DL2570	MATa/α eri1Δ::TRP1 ras2::LEU2/eri1Δ::TRP1 ras2::LEU2	This study
DL2698	MATα stt1-1 (pkc1ts) eri1-1 pRS315[PKC1 ADE3]	This study
DL2709	MATaira1Δ::LEU2 ira2Δ::HIS4	This study
DL2725	MATa/α ira1Δ::LEU2 ira2Δ::HIS4/ira1Δ::LEU2 ira2Δ::HIS4	This study
YHUM120	MATα ras2::ura3::HIS3 FRE(Ty1)::lacZ::LEU2 ura3-52 leu2::hisG his3::hisG trp1::hisG	22

^aAll strains are isogenic to EG123 (35) except for DL2698, which carries ade2, ade3, leu2, ura3, trp1, and his3 mutations, and YHUM120, which is derived from Σ1278b (22).

Construction of plasmids and genomic deletions.

The ERI1 gene was isolated from a centromeric yeast library (30) by complementation of the temperature-sensitive growth defect associated with an eri1-1 double mutant (DL2698). The 498-bp intergenic region between YPL096w and YPL097w (MSY1), which carries the ERI1 gene, was amplified by PCR from genomic DNA from strain 1783 and cloned into the EcoRI site of centromeric (pRS316) and 2μ (pRS426) plasmids to yield pRS316[ERI1] (p1380) and pRS426[ERI1] (p1381), respectively.

Construction of a fully functional hemagglutinin (HA)-tagged form of Eri1 (HA-ERI1) under the transcriptional control of the GAL1 promoter, the MET25 promoter, or its own promoter, is described below. For construction of GAL-HA-ERI1, the ERI1 open reading frame (ORF) was amplified with 285 bp of 3′ sequence and subcloned into pYeF1 (URA3 [7]) by using NotI and EcoRI. This construction (pYeF1[ERI1]; p1461) fused ERI1 in frame at its N terminus with a single copy of the HA epitope and places it under the inducible control of the GAL1 promoter. To switch markers, the URA3 gene of pYeF1[ERI1] was disrupted with TRP1 by subcloning a SmaI fragment bearing TRP1 (from pUC18[TRP1]) into the EcoRV site in URA3, resulting in pYeF1::TRP1[ERI1] (p1482). ^HAEri1 was capable of complementing the eri1Δ growth defect even when repressed with 2% glucose (not shown). To create HA-ERI1 under the control of the ERI1 promoter, a BamHI-EcoRI fragment from pYeF1[ERI1] that includes HA-ERI1, but not the promoter, was first inserted into pRS316 and pRS426. Then, 680 bp of sequence 5′ to the ERI1 start codon was amplified by PCR and inserted into the BamHI site of the resulting plasmids, yielding pRS316[HA-ERI1] (p1740) and pRS426[HA-ERI1] (p1742). To create ERI1 and HA-ERI1 under the constitutive control of the MET25 promoter, the ERI1 (or HA-ERI1) coding sequence was amplified and inserted into pRS426-MET25 (2μ) or pRS416-MET25 (cen) between the MET25 promoter and the CYC1 terminator (23). The pEGKG[RAS2] (21) and pEGKG[RAS2^V19] plasmids, which express GST-Ras2 under the control of GAL1, were provided by Bob Deschenes. Effector site mutants in RAS2^V19 were constructed by the PCR overlap extension method (13) and cloned into pEGKG.

Deletion of the genomic copies of ERI1, RAS2, IRA1, and IRA2 in the strain 1783 background is described below. To delete the genomic copy of ERI1, 1,508 bp of sequence 5′ to the ERI1 start codon and 677 bp of sequence 3′ of the ERI1 stop codon were amplified in separate PCRs from genomic DNA from strain 1783. The 5′ fragment was amplified with primers that placed an EcoRI site at the end adjacent to the ERI1 coding sequence and a BamHI site at the opposite end.

The 3′ fragment was amplified with primers that placed a NotI site adjacent to the ERI1 coding sequence and a BamHI site at the opposite end. These fragments were ligated in a three-molecule reaction to the EcoRI and NotI sites of the integrative plasmid pRS304 (34) to create a unique BamHI site between the fragments. The resulting plasmid, pRS304[eri1Δ::TRP1] (p1356), was linearized with BamHI and used to transform yeast strains to tryptophan prototrophy. For disruption of RAS2, the pras2::LEU2 plasmid of Kataoka et al. (17) was used as described previously (gift of S. Powers). Deletion of IRA1 was described previously (36). For deletion of IRA2, the LEU2 gene of pRS304 was first replaced with the HIS4 gene (SmaI to EcoRV) from pBluescript2[HIS4] (provided by Susan Michaelis) by insertion of the HIS4 fragment between the HpaI and AatII sites of pRS304 to create YIpHIS4 (p1187). Next, 1.1 kb of sequence 5′ to the IRA2 start codon and 1.8 kb of sequence 3′ of the IRA2 stop codon were amplified in separate PCRs from genomic DNA. The 5′ fragment was amplified with primers that placed a NotI site at the end adjacent to the IRA2 coding sequence and an SpeI site at the opposite end. The 3′ fragment was amplified with primers that placed a BamHI site adjacent to the IRA2 coding sequence and an SpeI site at the opposite end. These fragments were ligated in a three-molecule reaction to the BamHI and NotI sites of YIpHIS4 to create a unique SpeI site between the fragments. The resulting plasmid, YIp[ira2Δ::HIS4] (p1584), was linearized with SpeI and used to transform yeast strains to histidine prototrophy. All genomic deletions were confirmed by PCR.

Immunodetection of ^HAEri1 and association of ^HAEri1 with GST-Ras2.

Indirect immunofluorescence microscopy to detect ^HAEri1 was performed as described previously (15) by using a Zeiss Axioskop filled with a fluorescein isothiocyanate filter. For subcellular fractionation experiments, transformants of yeast strain 1783 bearing pRS416[MET25-HA-ERI1] were grown to mid-log phase in YEPD, and lysates were prepared as described previously (15). Lysates were centrifuged at 100,000 × g in an SW50.1 rotor (Beckman) for 1 h. Supernatant and pellet fractions (resuspended in lysis buffer to equivalent volume as supernatant fractions) were separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) for immunoblot detection of ^HAEri1. For in vivo association experiments, double transformants bearing pYeF1::TRP1[ERI1] and pEGKG[RAS2] (or a mutant form of RAS2) were precultured in YEP with 2% raffinose, followed by induction with 4% galactose for 5 h. Lysate preparation and immunoblotting were done as described previously, except for the addition 0.5% NP-40 to the lysis buffer (15). Glutathione S-transferase (GST)-Ras2 was precipitated from lysates with glutathione-Sepharose 4B beads (Amersham Pharmacia) equilibrated in Tris-buffered saline (TBS; 50 mM Tris-HCl [pH 7.6], 8% NaCl), with 0.1% NP-40 (TBSN). Briefly, 4 ml of TBSN and 50-μl beads were added to 0.5 ml of extract (10 to 15 mg of protein/ml) and tumbled for 1 h at 4°C. Beads were washed three times with 10 ml of TBSN and eluted with SDS-PAGE loading dye (130-μl final volume). ^HAEri1 was detected (from 10 to 20 μl of sample) with mouse monoclonal antibody 12CA5 (BabCo), and GST-Ras2 was detected (from 2 to 4 μl of sample) with anti-GST (Amersham Pharmacia).

For separation of ER from the PM, ^HAEri1 and GST-Ras2^V19 were induced as described above prior to cell lysis by agitation with glass beads. Membranes were fractionated by sedimentation on a step sucrose-EDTA density gradient by a method modified as follows from that of Valdivia et al. (40). Unlysed cells were removed by centrifugation (500 × g for 5 min). Total cell lysates (0.2 ml) were overlaid on a step sucrose-EDTA gradient (0.2 ml of 55% sucrose, 0.5 ml of 45% sucrose, and 0.4 ml of 30% sucrose [wt/wt] in 20 mM triethanolamine [pH 7.2]-5 mM EDTA) and centrifuged at 46,400 rpm in an SW50.1 rotor for 5 h. Fractions (0.2 ml) were collected manually from the top and separated by SDS-PAGE for immunoblot analysis for GST-Ras2^V19, ^HAEri1, Gas1 (with rabbit anti-Gas1 serum; a gift of Laura Popolo), Dpm1 (with mouse monoclonal anti-Dpm1, 5C5; Molecular Probes), and CPY (with mouse monoclonal anti-CPY, 10A5; Molecular Probes). After the addition of nonionic detergent (0.5% NP-40) to the fractions, GST-Ras2^V19 was then precipitated from fractions and subjected to immunoblot analysis for ^HAEri1. Films from immunoblots were scanned into Adobe Photoshop (v5.0) for densitometric analysis by using Image Gauge (v3.3) software.

Heat shock sensitivity and FRE::lacZ assays.

The heat shock sensitivity of eri1Δ was assayed after 24 h growth on a YEPD plate. The ira1Δ strain was tested for heat shock sensitivity after 2 days on an SD plate. Cells were collected from plates and resuspended in YEPD at an A₆₀₀ of 1.0, and serial 10-fold dilutions were made in YEPD. Then, 1 μl of each dilution was spotted onto a YEPD plate, which was incubated in a 50°C water bath for either 30 min (ira1Δ) or 50 min (eri1Δ) by submerging the plate in a sealed bag. Colonies were allowed to grow for 36 h at room temperature before they were counted. The percentage of survivors was calculated as the CFU relative to non-heat-shocked controls. Each value represents the mean and standard deviation of at least three experiments. The ability of Eri1 overproduction to downregulate Ras2^V19-driven FRE::lacZ expression was tested as described by Mosch et al. (22). Yeast strain YHUM120 was transformed with centromeric plasmids pRS316[RAS2] or pRS316[RAS2-V19] and either multicopy plasmid pRS424[ERI1] or pRS424. Transformants were patched onto SD plates and allowed to grow at 30°C for 24 h. Lysates were made from cells collected from plates and assayed for β-galactosidase activity. Values represent the mean and standard deviation from at least three independent transformants.

RESULTS AND DISCUSSION

ERI1 is a novel yeast gene.

We isolated ERI1 through a genetic screen for mutants with growth defects that were additive with that of a conditional mutant in protein kinase C (pkc1 [30]), a regulator of cell wall biogenesis. The ERI1 gene (YPL096c-A) includes a previously nonannotated ORF (nORF) encoding only 68 amino acids (Fig. 1A). Several lines of evidence support the conclusion that ERI1 is a bona fide gene. First, searches of genome databases have revealed several putative ERI1 homologs from a variety of fungi, including Candida albicans, Schizosaccharomyces pombe, Aspergillus fumigatus, and Neurospora crassa. The closest of these to S. cerevisiae Eri1 are shown in Fig. 1B. Database searches have not revealed any metazoan Eri1 homologs. Second, a single orphan SAGE-tag (for serial analysis of gene expression) corresponding to the ERI1 locus was identified previously (41), suggesting that a polyadenylated mRNA is expressed at a low level from this gene. Indeed, we found that an ERI1 probe hybridizes to an mRNA of ca. 300 nucleotides from both log-phase and stationary-phase cells (Fig. 1C). Third, the original eri1 mutant was recessive, resulting from a frameshift mutation after the eighth codon. Fourth, a precise deletion of ERI1 results in a growth defect at elevated temperature (37°C) that is complemented by a centromeric plasmid bearing 498 bp of sequence corresponding to ERI1 and its regulatory elements only (Fig. 1D). Fifth, we can detect an epitope-tagged form of the Eri1 protein expressed from its own promoter (see Fig. 5B).

FIG. 1.

ERI1 is a bona fide gene. (A) Predicted amino acid sequence of Eri1 (YPL096c-A). Two potential transmembrane domains are underlined. (B) Alignment of S. cerevisiae Eri1 with homologs from C. tropicalis and C. albicans. Identical residues are boxed. (C) Detection of ERI1 RNA. Total RNA was prepared from wild-type strain 1788 and eri1Δ strain DL2524 cells in log-phase growth or from saturated cultures (2 days of growth). RNA (10 μg) was probed for ERI1 and ACT1 (encoding actin) as a loading control. (D) Yeast strain DL2524 was transformed with centromeric plasmid pRS316[ERI1] (pCEN[ERI1]), which contains the 498-bp intergenic region between YPL096w and YPL097w, or with pRS316 (vector). Transformants were streaked onto YEPD plates and incubated at the indicated temperatures for 3 days.

FIG. 5.

Eri1 is an ER membrane protein. (A) ^HAEri1 fractionates as an integral membrane protein. The indicated agent was added to lysates of wild-type haploid (strain 1783) yeast cells expressing ^HAEri1 from pRS426-MET25[HA-ERI1] prior to centrifugation at 100,000 × g for 1 h. ^HAEri1 from equivalent cell fractions of total lysate (T), supernatant (S), or pellet (P) was detected by immunoblotting. (B) Immunofluorescence micrographs of wild-type diploid strain 1788 yeast cells expressing ^HAEri1 from 2μm plasmid pRS426[HA-ERI1].

The eri1Δ phenotypes can be divided into two categories that reflect two distinct functions of this small protein. One set of phenotypes, described here, is associated with hyperactive Ras pathway signaling. The other phenotypes, including the growth defect at an elevated temperature, are associated with a deficiency in anchoring of glycosylphosphatidylinositol proteins and their subsequent secretion to the cell surface. The latter phenotypes are responsible for the additive growth defect observed with pkc1 mutants and will be described elsewhere.

ERI1 behaves genetically as a Ras inhibitor.

Some strains of S. cerevisiae can undergo a dimorphic shift on solid medium in response to nutrient limitation from ovoid, budding cells to a multicellular form consisting of filaments of elongated cells (12, 28). This shift is enhanced by hyperactivation of Ras pathway signaling (12, 19, 43) and specifically requires signaling from Ras2 (22). Microscopic examination of diploid eri1Δ cells grown on solid rich medium (Fig. 2A) revealed that they display the characteristic features of filamentous growth when cultivated at 34°C. This was unexpected, both because nutrients were not limiting in this setting and because the strain background used in the present study (EG123 [35]) is not known to be capable of filamentous growth. Therefore, to determine whether the observed behavior of eri1Δ cells reflects bona fide filamentous growth, we deleted RAS2 in an eri1Δ mutant. The eri1Δ ras2Δ mutant displayed a normal budding morphology, indicating that Ras2 is required for filamentation of eri1Δ cells. This result also suggested that Ras pathway signaling is hyperactivated in the eri1Δ mutant. Another behavior associated with filamentous growth is the ability to invade the agar, which can be detected after nonadeherent cells are washed from the plate (28). An eri1Δ mutant displayed agar invasion (Fig. 2B) similar to that observed for mutants in the same strain background with hyperactive Ras resulting from loss of Ras-GTPase-activating protein (Ras-GAP) function (Fig. 2C). Moreover, an eri1Δ ras2Δ mutant was suppressed for agar invasion, supporting the notion that Ras pathway signaling is hyperactive in eri1Δ mutants. In contrast, deletion of RAS2 failed to suppress the growth defect of eri1Δ cells at 37°C (not shown), suggesting that this phenotype is not the result of hyperactive Ras signaling. Additionally, hyperactive Ras pathway mutants are not temperature sensitive for growth, supporting the conclusion that this phenotype is independent of Ras function.

FIG. 2.

Hyperactivation of Ras pathway signaling in an eri1Δ mutant. (A) An eri1Δ mutant displays filamentous growth. Single colonies of diploid yeast strains were photographed after 24 h on YEPD plates grown at the indicated temperatures. The strains include wild-type strain 1788, eri1Δ strain DL2524, and eri1Δ ras2Δ strain DL2570. (B) An eri1Δ mutant displays invasive growth. Haploid yeast strains were streaked onto YEPD plates and allowed to grow at 34°C for 3 days (total growth). Nonadherent cells were washed from the plates with distilled water to reveal invasive growth. The strains include wild-type strain 1783, eri1Δ strain DL2522), ras2Δ (DL838), and eri1Δ ras2Δ (DL2566). (C) Hyperactivation of Ras signaling in the EG123 background results in invasive growth. Haploid yeast strains were treated as in panel B, except that growth was at 30°C. Mutants in the redundant Ras-GAPs encoded by IRA1 and IRA2 were used to activate Ras signaling. The strains include wild-type strain 1783, ira1Δ strain DL2297, and ira1Δ ira2Δ strain DL2709.

Hyperactive Ras signaling in yeast also results in the failure to acquire thermotolerance in preparation for stationary phase (18, 24). We reasoned that if the invasive growth phenotype of eri1Δ cells was a consequence of hyperactive Ras pathway signaling, this mutant should also display a defect in the acquisition of thermotolerance. A saturated eri1Δ culture lost 2 logs of viability relative to wild type in response to a brief heat shock at 50°C (Fig. 3A). This defect was suppressed by downregulation of the Ras pathway through expression of a dominant-negative form of RAS2 (14) or through overexpression of the Ras-GAP encoded by IRA2 (38). Conversely, the heat shock sensitivity of a mutant defective in one of two redundant Ras-GAPs (ira1Δ) was partially suppressed by overexpression of ERI1 under the control of the MET25 promoter (Fig. 3B), further supporting the conclusion that Eri1 can inhibit Ras pathway signaling. However, we were not able to demonstrate with this assay that the more severe heat shock sensitivity resulting from either a complete loss of Ras-GAP activity (in an ira1Δ ira2Δ mutant) or a constitutively active allele of RAS2 (RAS2-V19) was suppressed by ERI1 overexpression (data not shown), suggesting that the ability of Eri1 to inhibit Ras pathway signaling is limited. We detected no increase in severity of the heat shock sensitivity of an ira1Δ ira2Δ mutant in the presence of an eri1Δ mutation (not shown).

FIG. 3.

Suppression of heat shock sensitivities of eri1Δ and ira1Δ yeast strains. (A) Strain DL2524 (eri1Δ) was transformed with centromeric vector pRS316, episomal plasmid YEp24[RAS2-Y64] (D.N. RAS2), YEp24[IRA2], or pRS316[ERI1]. Transformants were subjected to a 50-min heat shock at 50°C and allowed to recover at room temperature for 36 h prior to scoring CFU. (B) Strain DL2297 (ira1Δ) was transformed with YEp24 (vector), pRS416-MET25-ERI1, or YEp24[IRA1]. Transformants were subjected to a 30-min heat shock at 50°C, allowed to recover for 36 h, and CFU were scored.

As a final test of the ability of Eri1 to regulate Ras pathway activity, we examined the expression of a Ras2-regulated reporter gene. Hyperactivation of Ras pathway signaling induces expression of a reporter under the control of the filamentous response element (FRE::lacZ [22]). This reporter has the advantage of providing a more sensitive output for Ras pathway signaling than the heat shock sensitivity assay. Expression of constitutively active RAS2-V19 from a centromeric plasmid induced FRE::lacZ expression 1.7-fold compared to the expression of wild-type RAS2 (201 ± 9 U versus 121 × 10 U). Overexpression of Eri1 from a multicopy plasmid reproducibly diminished FRE::lacZ expression in the presence of RAS2-V19 (163 ± 18 U). Further overexpression of Eri1 under the control of the MET25 promoter (yielding ∼10-fold more Eri1 than the multicopy plasmid; data not shown) did not diminish reporter activity further (161 ± 19 U), underscoring the limitation of overexpressed Eri1 to inhibit Ras pathway signaling. These results, taken in the aggregate, indicate that Eri1 is an inhibitor of Ras pathway activity that is capable of partially downregulating the output of an oncogenic form of Ras.

Eri1 associates with Ras2 in a GTP- and effector loop-dependent manner.

We were interested to determine the point at which Eri1 acts to inhibit Ras pathway signaling. An additional heat shock experiment suggested that Eri1 acts at the level of Ras. Specifically, overexpression of ERI1 failed to suppress the modest heat shock sensitivity of a constitutive mutant in adenylyl cyclase (SRA4-6 [4; data not shown]), the direct effector of Ras in S. cerevisiae. Therefore, we tested for association of Eri1 with Ras2 in vivo. GST-tagged Ras2 or a constitutively active mutant form (GST-Ras2^V19) was coexpressed in yeast cells with HA-tagged Eri1 (^HAEri1). GST-Ras2 was affinity purified from cell extracts made in the presence of nonionic detergent (0.5% NP-40) to disrupt membranes and tested for association with ^HAEri1 by immunoblot analysis. Although only a weak ^HAEri1 signal was detected in association with wild-type Ras2, we reproducibly detected a strong signal associated with the activated form of Ras2 (Fig. 4A). Because wild-type yeast Ras2 exists almost entirely in the GDP-bound state in vivo (11a), whereas the constitutively active Ras2^V19 is “locked” in the GTP-bound state (11), the observed difference in association between the two forms could reflect the difference in bound nucleotides. To determine whether Eri1 association with Ras2 is dependent on its nucleotide-bound state, we isolated GST-Ras2 from cells devoid of Ras-GAPs (ira1Δ ira2Δ). In this setting, Ras is predominantly GTP bound. In the absence of the Ras-GAPs, ^HAEri1 associated with wild-type GST-Ras2, as well as with the activated form (Fig. 4A), indicating that Eri1 associates specifically with GTP-bound Ras2. Another indication of the specificity of the interaction is that we were not able to detect association of Eri1 with a constitutive form of Rho1 (GST-Rho1^68H), the yeast GTPase most closely related to Ras (data not shown).

FIG. 4.

Eri1 associates with Ras2 in a GTP- and effector loop-dependent manner. (A) Extracts were prepared from diploid yeast strains 1788 (wild type) and DL2725 (ira1Δ ira2Δ) coexpressing ^HAEri1 (from pYeF1::TRP1[ERI1]) with GST (from pEGKG), GST-Ras2 (WT), or GST-Ras2^V19 (V19). GST and GST-Ras2 fusions were precipitated with glutathione-Sepharose beads and detected by immunoblot (upper panel). Associated ^HAEri1 was also detected by immunoblot (lower panel). (B) Extracts were prepared from yeast strain 1788 coexpressing ^HAEri1 and the indicated forms of GST-Ras2, and treated as in panel A. The Ras2 forms were GST-Ras2 (Wild type), GST-Ras2^V19 (V19), GST-Ras2^{V19 A42} (V19 A42), and GST-Ras2^{V19 N45} (V19 N45). The lower GST-Ras2 band does not reflect differential Ras modification (M. J. Romeo, unpublished results) and is therefore presumed to be a proteolysis product.

GTP-dependent association of a protein with Ras is one of two criteria used in the identification of novel Ras effectors (44). The other criterion is the requirement of an intact Ras-effector loop for interaction. Mutations in the Ras effector loop, comprised of residues 39 to 47 in yeast Ras (corresponding to residues 32 to 40 in mammalian Ras), disrupt interaction with effector proteins (1, 10, 20, 37). To determine whether Eri1 associates with GTP-Ras2 through its effector loop, we introduced either of two mutations known to disrupt Ras interaction with adenylyl cyclase (A42 or N45 [1, 37]) into the constitutive RAS2 mutant (creating RAS2-V19,A42 and RAS2-V19,N45). Figure 4B shows that both effector loop mutations blocked the GTP-Ras interaction with Eri1, indicating that Eri1 associates with Ras in an effector loop-dependent manner. To summarize, Eri1 fits the operational definition of a Ras effector but behaves genetically as a Ras inhibitor. Therefore, Eri1 may act as a competitive inhibitor of Ras signaling by shielding the effector loop of GTP-Ras from interaction with its effector.

Eri1 engages Ras at the ER membrane.

The Eri1 protein possesses two highly hydrophobic regions (residues 7 to 28 and residues 34 to 56; Fig. 1A), either of which is long enough to constitute a transmembrane domain. To determine whether Eri1 is associated with a membrane, we examined the fractionation pattern of ^HAEri1 in yeast cell extracts. Figure 5A shows that ^HAEri1 sedimented with the pellet from a 100,000 × g centrifugation. ^HAEri1 was liberated by addition of nonionic detergent (Triton X-100) to disrupt membranes but not by other treatments that would liberate peripherally associated membrane proteins (i.e., urea or high pH). Therefore, we conclude that Eri1 is an integral membrane protein. To determine with which intracellular membrane Eri1 associates, we conducted indirect immunofluorescence microscopy on cells expressing ^HAEri1 from its own promoter on a multicopy plasmid. ^HAEri1 displayed a pattern of perinuclear fluorescence with swirls out to the cell periphery (Fig. 5B), which is characteristic of proteins that reside in the ER (2, 29, 31, 32). This pattern was observed consistently in cells across the entire cell cycle.

Although the majority of Ras resides on the PM (2), recent studies indicate that it also associates with the ER, at least transiently, in both mammals and yeast (5, 6, 8, 27, 29, 32). Therefore, we tested the possibility that Eri1 engages Ras at the ER. Total membranes from cells coexpressing GST-Ras2^V19 and ^HAEri1 were sedimented on a step sucrose-EDTA density gradient. As anticipated from the immunofluorescence localization of ^HAEri1, the majority of the ^HAEri1 (56%) cosedimented with the ER marker Dpm1 in fraction 1 (Fig. 6A), with the remainder diminishing through fractions 2 to 6. The PM marker Gas1 sedimented in fractions 3 to 6, with the majority (66%) sedimenting in fractions 4 and 5. However, the immature, ER-modified form of Gas1 (26) sedimented in fraction 1. In contrast, GST-Ras2^V19 was evenly distributed across the entire gradient, suggesting its localization at endomembranes, as well as at the PM. After disruption of the membranes with detergent, GST-Ras2^V19 was affinity purified from the gradient fractions and tested for association with ^HAEri1. ^HAEri1 was found in association with GST-Ras2^V19, mainly from fraction 1 (Fig. 6B), indicating that Eri1 engages GTP-bound Ras at the ER. Some ^HAEri1 was detected with GST-Ras2^V19 isolated from fractions 3 and 4, suggesting that a pool of Eri1 may also engage Ras at a heavier membrane. The failure of ^HAEri1 to associate with GST-Ras2^V19 from fraction 2 may be explained by the predominance of the vacuolar marker CPY in that fraction. Because both proteins were overexpressed, some of each may be targeted for destruction in the vacuole, where they are not likely to interact.

FIG. 6.

Eri1 engages Ras at the ER. (A) Subcellular fractionation of ^HAEri1 and GST-Ras2^V19-containing membranes. Total membranes from yeast strain 1788 coexpressing ^HAEri1 with GST-Ras2^V19 (as in Fig. 4) were sedimented on a step sucrose-EDTA density gradient. Fractions were subjected to immunoblot analysis to detect GST-Ras2^V19, ^HAEri1, Gas1 (PM marker), Dpm1 (ER marker), and CPY (vacuolar marker). The mature (m) and immature (i) forms of Gas1 are labeled. (B) ^HAEri1 associates with Ras2^V19 at the ER. GST-Ras2^V19 was precipitated with glutathione Sepharose beads from the fractions in panel A. Precipitated GST-Ras2^V19 and associated ^HAEri1 were detected by immunoblotting.

Eri1 and Ras reside largely on separate membranes, with the majority of Eri1 in the ER but apparently no more than 20% of the total Ras at that organelle (see Fig. 6A). Therefore, to rule out the possibility that ^HAEri1 and GST-Ras2^V19 associate fortuitously in vitro only after liberation from their respective membranes, we tested for their association after expression in separate cell populations. Cultures of cells expressing GST-Ras2^V19 or ^HAEri1 were combined, and GST-Ras2^V19 was affinity purified from extracts made in the presence of detergent. In contrast to the association detected when the proteins were coexpressed in the same cell population (see Fig. 4A), we were not able to detect their association in vitro when the proteins were initially in separate membranes (data not shown). This indicates that the observed interaction occurs in vivo between the pools of Eri1 and Ras2 that reside on the same membrane. Because the GTP-Ras2/Eri1 association was detected in vivo, we do not know if this interaction is direct or bridged by another protein. Eri1 could be part of an ER protein complex that regulates Ras signaling.

If yeast Ras normally signals from endomembranes, as has been shown for mammalian Ras (5), Eri1 (or an Eri1-containing complex) may function to regulate such signaling at the ER. Alternatively, Eri1 may function as a molecular chaperone to maintain Ras in an inactive state while the GTPase is being processed at the ER. In either case, the segregation of the majority of these proteins on separate membranes is likely to explain the limited ability of Eri1 to inhibit GTP-bound Ras. It may be possible to enhance the efficacy of Ras inhibition by targeting Eri1 to the PM.