Abstract
Prp20p and Rna1p are GDP/GTP exchanging and GTPase-activating factors of Gsp1p, respectively, and their mutations, prp20-1 and rna1-1, can both be suppressed by Saccharomyces cerevisiae gtr1-11. We found that gtr1-11 caused a single amino acid substitution in Gtr1p, forming S20L, which is a putative GDP-bound mutant protein, while Gtr1p has been reported to bind to GTP alone. Consistently, gtr1-S20N, another putative GDP-bound mutant, suppressed both prp20-1 and rna1-1. On the other hand, gtr1-Q65L, a putative GTP-bound mutant, was inhibitory to prp20-1 and rna1-1. Thus, the role that Gtr1p plays in vivo appears to depend upon the nucleotide bound to it. Our data suggested that the GTP-bound Gtr1p, but not the GDP-bound Gtr1p, interacts with itself through its C-terminal tail. S. cerevisiae possesses a novel gene, GTR2, which is homologous to GTR1. Gtr2p interacts with itself in the presence of Gtr1p. The disruption of GTR2 suppressed prp20-1 and abolished the inhibitory effect of gtr1-Q65L on prp20-1. This finding, taken together with the fact that Gtr1p-S20L is a putative, inactive GDP-bound mutant, implies that Gtr1p negatively regulates the Ran/Gsp1p GTPase cycle through Gtr2p.
SACCHAROMYCES cerevisiae Gsp1p is a homologue of mammalian Ran (Ras-like nuclear G protein, Bischoff and Ponstingl 1991; Belhumeuret al. 1993). The nucleotide exchange of Gsp1p is carried out by the S. cerevisiae RCC1 homologue Prp20p, and the intrinsic GTPase of Gsp1p is activated by Rna1p (reviewed by Sazer 1996; Sekiet al. 1996). Ran was found to be essential for nuclear protein import (Moore and Blobel 1993). Indeed, all the temperature-sensitive mutants (Ts-) of Gsp1p show a defect in nuclear localization signal (NLS)-dependent nuclear protein import (Wonget al. 1997; Okiet al. 1998). Thus, Ran plays an important role in the nucleus/cytosol exchange of macromolecules (Moore and Blobel 1994; Melchior and Gerace 1995; Avis and Clarke 1996; Görlich and Mattaj 1996; Nigg 1997; Görlich 1998).
On the basis of an analogy with the Ras family (Boguski and McCormick 1993), events downstream of Ran/Gsp1p should be carried out by proteins that bind specifically to GTP-Ran/Gsp1p. The proteins that have been so far reported to bind to GTP-Ran/Gsp1p are the RanBP1 family comprising Yrb1p, Yrb2p, RanBP1, RanBP2/NUP358, and RanBP3 (Coutavaset al. 1993; Bischoffet al. 1995; Dingwallet al. 1995; Melchioret al. 1995; Schlenstedtet al. 1995; Wuet al. 1995; Yokoyamaet al. 1995; Chiet al. 1996; Noguchiet al. 1997; Muelleret al. 1998; Tauraet al. 1998), the importin β family (Görlichet al. 1997; Ullmanet al. 1997), Dis3p (Noguchiet al. 1996; Shiomiet al. 1998), and RanBPM (Nakamuraet al. 1998). The RanBP1 and importin β families are both involved in the nucleus/cytosol exchange of macromolecules, but the others are not.
The phenotype caused by a defect in the nucleotide exchange of Ran is pleiotropic. Cultures of the tsBN2 cell line, hamster rcc1 (Uchidaet al. 1990), show G1 arrest at 39.5°, the nonpermissive temperature, if shifted before the S phase (Nishitaniet al. 1991). From the S phase onwards, however, tsBN2 cells prematurely enter mitosis, resulting in premature chromatin condensation at 39.5° (Nishitaniet al. 1991). The Schizosaccharomyces pombe rcc1 homologue, pim1, shows a defect in chromosome decondensation (Sazer and Nurse 1994). Furthermore, the S. cerevisiae RCC1 homologue, PRP20, was identified as a mutant defective in the mating process (srm1-1, Clark and Sprague 1989), mRNA splicing (prp20-1, Aebiet al. 1990), and mRNA export (mtr1-2, Kadowakiet al. 1993). It is still obscure whether these phenotypes of the rcc1/pim1/prp20 mutants were caused by a defect in the nucleus/cytosol exchange of macromolecules or in other unknown pathways. To clarify this issue, we previously isolated the cold-sensitive mutants ded1-21 (Hayashiet al. 1996) and gtr1-11 (Nakashimaet al. 1996) as suppressors of srm1-1 and mtr1-2, respectively. ded1-21 also suppresses mtr1-2 (Kadowakiet al. 1993), but not prp20-1 (Aebiet al. 1990). In contrast, gtr1-11 suppresses not only all the S. cerevisiae prp20 alleles, but also rna1-1, which encodes a Ts- form of Rna1p (Hutchisonet al. 1969; Hopperet al. 1978; Amberget al. 1992; Forresteret al. 1992).
GTR1 is not essential for survival, but its loss leads to cold-sensitive growth (Bun-yaet al. 1992). Gtr1p, a protein of 36 kD, possesses three motifs conserved in the Ras family, but has been reported to bind to GTP alone (Bun-yaet al. 1992). For this reason, Gtr1p is considered to be a putative G protein. Recently, Schurmann et al. (1995) identified two human cDNA clones, encoding proteins designated as RagA and RagB, that are homologous to Gtr1p and that rescue the cold-sensitive growth of the gtr1-11 strain (Hiroseet al. 1998). Thus, Gtr1p is conserved through evolution. RagA has been identified independently as a protein that interacts with the adenovirus 14.7-kD E3 protein (E3-14.7K, Liet al. 1997), and it may be involved in cell death.
In this article, we found that gtr1-11 possessed a single amino acid substitution, S20L, that corresponds to a GDP-bound mutant of G proteins. Consistent with this finding, both prp20-1 and rna1-1 were suppressed by another putative GDP-bound mutant of Gtr1p, gtr1-S20N, but not by the putative GTP-bound mutant gtr1-Q65L. Overexpression of Gtr1p-Q65L was rather inhibitory for colony formation of prp20-1 and rna1-1 cells. We found a novel protein homologous to Gtr1p, designated as Gtr2p, that formed complexes with Gtr1p. Interestingly, the disruption of GTR2 suppressed prp20-1 and abolished the inhibitory effect of gtr1-Q65L on prp20-1.
MATERIALS AND METHODS
Strains and media: All the S. cerevisiae strains and plasmids used in this study are described in Tables 1 and 2, respectively. They were constructed by standard genetic manipulations (Kaiseret al. 1994). Transformation of S. cerevisiae was carried out by a modified LiCl method using DMSO (Hillet al. 1991). The media used for S. cerevisiae and bacteria have been described previously (Nakashimaet al. 1996).
Site-directed mutagenesis: The 3.4-kb ClaI/BamHI fragment of pL3 (Nakashimaet al. 1996) was inserted into the ClaI/BamHI sites of the pBluescript II SK(+)(Ichihara and Kurosawa 1993). The resultant pL51 was mutagenized with the Kunkel's site-directed mutagenesis method by means of the site-directed mutagenesis system Mutan-K (Takara, Kyoto, Japan) using as a mutagenic oligonucleotide 5′ TGTGGTGGGCT CGACGTGTTTATG 3′ for Q65L mutation, 5′ ATGGGCCGGG TCGGCTCCGGTAAATCGTCA for S15V, 5′ ATGGGCCGGG GCGGCTCCGGTAAATCGTCA for S15G, 5′ GGCTCCGGTA AAAACTCAATGAGGT for S20N, or 5′ CGCCCTTGAAATT TGACTTTTCGAAAGAACAAC 3′ for deletion of the C-terminal region of Gtr1p.
Construction of GTR1 and GTR2 plasmids: The 5.2-kb XbaI/SalI fragment was cut out from pL3 (Nakashimaet al. 1996) and inserted into the XbaI/SalI sites of pUC29, resulting in pL6. The 1.5-kb GTR1 fragment carried on the resultant pL6 plasmid was amplified by PCR using 5′ CAATTTAGCCATGGC GTCAAATAATAGGAAG 3′ and M13 Reverse (Toyobo, Osaka, Japan) as the primers and KOD Polymerase (Toyobo) as a polymerase. Amplified DNA fragments were digested with NcoI and BamHI enzymes and inserted into the NcoI/BamHI sites of pUC29, resulting in pL19. The 1.5-kb fragment was cut out from pL19 with NcoI and SalI enzymes and inserted into the SalI/NcoI sites of pGEX-KG (Guan and Dixon 1991), resulting in pL20. To express GST-fused GTR1 in S. cerevisiae, 1.5 kb of GST-fused GTR1 was cut out from pL20 by SmaI and SalI enzymes and inserted into the SmaI/SalI sites of pEG-KG (Mitchellet al. 1993), resulting in pL38.
The genomic DNA of the NBW5 strain was amplified using primers A and D as shown in Figure 6A, resulting in a 2.2-kb DNA fragment that was digested with BamHI/SalI enzymes and inserted into the BamHI/SalI sites of YEplac195. The 1.7-kb fragment of GTR2 carried on the resulting plasmid pL130 was amplified by PCR using 5′ TACACCATGGGTTTAGAGGC TACAGATTCCAAGGCAATGC 3′ and M13 Reverse as the primers and KOD Polymerase as a polymerase. The amplified DNA fragments were digested with NcoI and SalI enzymes and inserted into the NcoI/SalI sites of pEG-KG (Mitchellet al. 1993), resulting in pL153.
The 1.6-kb fragment of HA-fused GTR1 was cut out from the plasmid pL46 (Nakashimaet al. 1996) by XhoI and SmaI enzymes and inserted into the XhoI/SmaI sites of the plasmid YEplac112-XA (Funakoshiet al. 1997), resulting in pL80.
The 1.7-kb GTR2 fragment carried on pL130 was amplified as described above and inserted into the NcoI/SalI sites of pUC29. From the resulting pL136, the 1.8-kb GTR2 fragment was digested with NcoI and PstI enzymes and inserted into the NcoI/PstI sites of plasmid TKS-HA1 (Nakashimaet al. 1996). From the resulting pL152, the 1.8-kb GTR2 fragment was cut out with XhoI and SmaI enzymes and inserted into the XhoI/SmaI sites of YEplac112-XA (Funakoshiet al. 1997), resulting in pL154.
The 1.5-kb GTR1 fragments were digested from pL20 and its derivatives containing mutated gtr1 by the BamHI enzyme and then inserted into the BamHI site of pET-28a (Novagen Inc.), resulting in pL115 (wild type), pL117 (Gtr1-12Δp), and pL118 (Gtr1-13p). The 1.3-kb fragment containing His6-T7 carried on pET-28a was digested by NcoI and SmaI enzymes and inserted into the NcoI/StuI sites of pUC29, resulting in pL96. The 0.1-kb His6-T7 fragment was digested from pL96 by the SacI enzyme and inserted into the SacI site of pEG-KG, resulting in pEH7-SH. The 1.5-and 1.7-kb BamHI fragments containing GTR1 and GTR2 were then inserted into the BamHI site of pEH7-SH, resulting in pL101 and pL210, respectively.
Construction of GAL4 TADand DBD-fused GTR1: For the two-hybrid assay (Fields and Song 1989), the GAL4-promoter activation domain (TAD) and the DNA-binding domain (DBD) were fused to the wild-type and mutated GTR1 clones as follows. The 1.5-kb fragments of the wild-type and mutated GTR1 were cut out from pL19 or its derivatives containing the mutated gtr1 with NcoI and BamHI enzymes and then inserted into the NcoI/BamHI sites of pACT2 (Durfeeet al. 1993).
The 1.2-kb SacI/SalI fragment of pAS1 (Durfeeet al. 1993) containing the ADH promoter, the GAL4 DNA-binding domain, and the HA epitope was inserted into the SacI/SalI sites of pRS404. Into the NcoI/BamHI sites of the resulting pAS404, 1.5-kb NcoI/BamHI fragments containing the wild-type and mutated GTR1 were then inserted.
Disruption of GTR2: The genomic DNA of NBW5 was amplified using either primers A and B or primers C and D (see Figure 6A) to obtain the fragments AB and CD, respectively. The fragment AB was digested with BamHI/XhoI enzymes and inserted into the BamHI/XhoI sites of pUC29, resulting in pL119. The fragment CD was then digested with XhoI/SalI enzymes and inserted into the XhoI site of pL119, resulting in pL126. Finally, the LEU2 gene of YEp13 (Broachet al. 1979) was cut out with XhoI/SalI enzymes and inserted into the XhoI site of pL126, resulting in pL128. The 3.4-kb fragment containing LEU2::gtr2-1Δ was cut out with BamHI/NcoI enzymes from pL128 and then introduced into the strain N43.
Yeast strains used in this study
Purification of Gtr1p: GST or His6-fused wild-type and mutated Gtr1p were expressed in Escherichia coli and purified as described by Noguchi et al. (1997) and Nuoffer et al. (1995), respectively.
Cosedimentation of HA-fused proteins with the GST-fused proteins: The plasmids carrying the GST- and HA-fused gene were cointroduced into the strain NBW5ΔGTR1. Transformant cultures were grown to OD660 = 1.0 in SR medium (0.67% yeast nitrogen base without amino acid, 2% raffinose) lacking uracil and tryptophan, and then 2% galactose was added. After incubation for 2 hr at 30°, the cells were spun down, washed once with distilled water, and then resuspended in S buffer [50 mm potassium phosphate, pH 6.5, 120 mm NaCl, 1 mm MgCl2, 0.1% Triton X-100, 10% glycerol, 1 mm 2-mercaptoethanol, and 0.2 mm ρ-amidinophenyl-methane-sulfonyl fluoride (ρ-APMSF)]. Cells were then frozen at –80° and disrupted by glass beads. After centrifugation at 83,500 × g for 30 min, the supernatant was mixed with glutathione Sepharose-4B beads that had been saturated with S buffer, and then rotated for 1 hr. The beads were spun down and washed five times with S buffer. Proteins bound to the beads were analyzed by SDS-PAGE and immunoblotting. All procedures were carried out at 4° except where otherwise indicated.
β-Galactosidase assay: Overnight cultures of transformants of Y190 (0.3 ml) expressing GAL4-TAD-fused clones and GAL4-DBD-fused clones were inoculated into 15 ml of synthetic medium containing 0.67% yeast nitrogen base without amino acid, 2% ethanol, 3% glycerol, and the appropriate amino acids. Cultures were allowed to grow to OD660 = 0.9 and then spun down, washed once with distilled water, resuspended in 200 μl lysis buffer (0.1 m Tris-HCl, pH 8.0, 20% glycerol, 1 mm DTT, and 0.2 mm ρ-APMSF), and then frozen at –80°. Frozen cells were disrupted by beating with glass beads. After centrifugation at 83,500 × g for 30 min, the supernatant was used as the crude extract.
The enzyme assay was performed as reported (Kandels-Lewis and Seraphin 1993). The yeast crude extract (100 μl) or, as a control, the lysis buffer alone was mixed with 900 μl Z buffer (100 mm sodium phosphate buffer, pH 7.0, 10 mm KCl, 1 mm MgSO4, and 50 mm 2-mercaptoethanol). After incubation for 1–2 min at 28°, 200 μl o-nitrophenyl-β-d-galacto-pyranoside (4 mg/ml) was added, and the mixture was incubated at 28°. The reaction was stopped by the addition of 4.8 ml of 1 m Na2CO3. The amount of o-nitrophenol produced was estimated by measuring A420, and the protein content was estimated by the BCA Protein Assay kit (Pierce, Rockford, IL). One unit of β-galactosidase activity corresponds to the amount of β-galactosidase required to produce 1 nmol of o-nitro-phenol in 1 min at 28°.
Immunoblotting: Proteins were loaded on 11% SDS-PAGE, transferred onto PVDF membrane filters, and then probed with the anti-HA mAb (Babco) or anti-T7 mAb (Novagen, Inc.) as described previously (Noguchiet al. 1997).
Plasmids used in this study
RESULTS
Mutation site of gtr1-11: To clarify how gtr1-11, a cold-sensitive mutation of GTR1, suppresses both prp20 and rna1, we isolated the GTR1 gene from the gtr1-11 strain and determined its nucleotide sequence. In comparison with the wild-type GTR1, gtr1-11 was found to have thymine instead of cytosine at the 20th codon of Gtr1p, and serine was thereby changed to leucine (S20L, Figure 1 insert). No other nucleotide substitution was found in the gtr1-11 sequence. By comparison with other small G proteins, the 20th amino acid of Gtr1p, serine, was suggested to correspond to threonine, the 24th and 26th amino acid of Ran and Gsp1p, respectively (Figure 1). Hence, by analogy with the RanT24N mutant (Dassoet al. 1994), the S20L substitution of Gtr1p is likely to result in a dominant negative GDP-bound form of the protein, implying that the GDP-bound Gtr1p has some biological function in cells. To investigate the function of the Gtr1p, additional mutants that should be locked in the GTP-bound state (Gtr1p-Q65L, -S15V, and -S15G) or the GDP-bound state (Gtr1p-S20N) were constructed on the basis of mutants of Ras (Boguski and McCormick 1993). These mutated gtr1 clones were inserted into either a single-copy vector, pRS314, or a multicopy vector, YEplac195 (Table 2).
—Alignment of amino acid sequences of Gtr1p, Gsp1p, Ran, and K-Ras. Asterisks and periods indicate identical and chemically conserved amino acid residues, respectively. Also shown are the mutated sites of assumed GTP (#1 and #3) and GDP (#2) bound forms of Gtr1p. Inset: the mutation site of gtr1-11. Iso-leucin of Gtr1p (#4) was changed to the stop codon to create the C-terminal deletion.
When the gtr1 mutants carried on a single-copy vector were expressed in the gtr1-1Δ strain (Figure 2A), two putative GTP-bound mutants, gtr1-S15V and gtr1-S15G, rescued the cold sensitivity of the gtr1-1Δ strain with an efficiency similar to that of wild-type GTR1, but another putative GTP-bound mutant, gtr1-Q65L, rescued it only weakly. On the other hand, both putative GDP-bound mutants, gtr1-S20N and gtr1-S20L, did not rescue the cold sensitivity of the gtr1-1Δ strain. Thus, both gtr1-S15V and gtr1-S15G behaved like wild types, whereas the other mutants did not. The mutated and wild-type GTR1 clones carried on a multicopy vector were then introduced into the haploid strains prp20/2c (prp20-1), NN19-5B (rna1-1), and, as a control, NBW5 (wild type). Ura+ transformants were selected and plated on synthetic medium lacking uracil at the three temperatures indicated (Figure 2B)–the permissive, semipermissive, and nonpermissive temperatures for each mutated strain.
The two putative GDP-bound mutants, gtr1-S20L and gtr1-S20N, both rescued the temperature sensitivity of each of the prp20-1 and rna1-1 strains, whereas a putative GTP-bound mutant, gtr1-Q65L, did not (Figure 2B). Interestingly, gtr1-Q65L inhibited the colony formation of both prp20-1 and rna1-1 cells at 30° and 28°, the semipermissive temperatures for each mutant. The other putative GTP-bound mutants, gtr1-S15V and gtr1-S15G, did not show any inhibitory effect on prp20-1 and rna1-1 cells. Thus, both gtr1-S15V and gtr1-S15G behaved like the wild types, consistent with the finding that these mutants effectively complemented gtr1-1Δ (Figure 2A).
Gtr1p binds not only to GTP, but also to GDP: E. coliproduced GST-fused Gtr1p was purified on glutathione Sepharose-4B beads. The purified GST-fused Gtr1p was mixed with either 3H-labeled GTP or GDP. As a control, the nucleotide-binding experiments were also conducted in the presence of EDTA, which is reported to release the nucleotides from Ran (Bischoffet al. 1995; Richardset al. 1995). After incubation for 30 min at 30°, the glutathione Sepharose-4B beads were spun down, and the radioactivity that was coprecipitated with the beads was quantified using a liquid scintillation counter. E. coli-produced Gtr1p bound efficiently to [3H]GTP, but it did not bind as well to [3H]GDP when compared to the value obtained in the presence of EDTA (data not shown).
—Biological effect of the putative GTP and GDP-bound forms of Gtr1p. (A) Single-copy plasmids carrying the indicated mutated and wild-type GTR1 genes and, as a control, the vector pRS314 alone were introduced into strain gtr1-1Δ (NBW5ΔGTR1). The resulting Trp+ transformants were plated on synthetic medium lacking tryptophan and were incubated at 14° or 30° as indicated. (B) YEplac195 plasmids carrying the indicated mutated and wild-type GTR1 genes and, as a control, the vector alone were introduced into the indicated strains, wild-type (NBW5), prp20-1 (prp20/2c) (Aebiet al. 1990), or rna1-1 (NN19-5B, Noguchiet al. 1997). Ura+ transformants plated on synthetic medium lacking uracil were incubated at the indicated temperature.
To further determine the nucleotide-binding ability of Gtr1p, [3H]GTP was mixed with an increasing amount of cold GTP, GDP, or ATP, and was then incubated with E. coli-produced, GST-fused Gtr1p. After incubation at 30° for 30 min, the radioactivity coprecipitated with the glutathione Sepharose-4B beads was quantified using a liquid scintillation counter. In the presence of GTP, the amount of [3H]GTP bound to Gtr1p was greatly reduced (Figure 3A). On the other hand, ATP did not prevent [3H]GTP from binding to Gtr1p, even at the higher concentrations. Compared with the effect of ATP, GDP significantly inhibited the binding of [3H]GTP to Gtr1p, indicating that GDP bound to Gtr1p in a manner that competed with GTP.
The GDP-binding ability of Gtr1p was further confirmed using the Gtr1p-Q65L that may have a defect in GTPase similar to the Q61L mutant of rasH (Deret al. 1986), which has been reported to bind to both GTP and GDP in vitro (Temeleset al. 1985; Klebeet al. 1995a). Indeed, E. coli-produced, His6-fused Gtr1p-Q65L bound to [3H]GTP and to [3H]GDP (Figure 3B). Taken together, these results lead us to conclude that Gtr1p binds to either GTP or GDP, like a normal G protein would. We could not, however, biochemically assay the GTPase activity of Gtr1p in vitro. Hence, Gtr1p was referred to as a putative G protein.
Gtr1p interacts with itself: Gtr1p has a long C-terminal tail (from 200 to 310 aa) outside the nucleotidebinding domains (Figure 1A). It contains a large number of leucine residues and is predicted to also contain a coiled-coil motif (Lupaset al. 1991; Lupas 1996), both of which are thought to be involved in protein-protein interaction. Using the two-hybrid method, we investigated whether or not Gtr1p could interact with itself through its C terminus (Fields and Song 1989). The wild-type (WT) and the C-terminal-deleted Gtr1p (Gtr1-12Δp) were fused to either the GAL4 TAD or the GAL4 DBD, and then introduced into the Y190 strain. Considerable β-galactosidase activity was observed when TAD-GTR1 was coexpressed with DBD-GTR1 (Table 3). When the wild-type GTR1 was coexpressed with the C-terminal-deleted mutant gtr1-12Δ, however, no significant β-galactosidase activity was detected (Table 3). Similar to wild-type Gtr1p, Gtr1-12Δp was produced in S. cerevisiae to a significant degree (Figure 4A, compare lanes 1 and 6). Thus, the failure to detect β-galactosidase activity when the wild-type Gtr1p and Gtr1-12Δp were coexpressed did not result from a low expression of Gtr1-12Δp, implying that the C-terminal tail of Gtr1p is required for self-interaction of Gtr1p.
To further confirm that Gtr1p forms a complex with itself, the HA and GST-GTR1 clones were cointroduced into the NBW5ΔGTR1 (gtr1-1Δ) strain. Crude extracts were prepared, mixed with glutathione Sepharose-4B beads, and the beads were then pelleted. The resulting precipitates were analyzed for the presence of HA-Gtr1p by immunoblotting using the mAb to HA. As shown in Figure 4B, lane 4, HA-Gtr1p was coprecipitated with GST-Gtr1p.
—Nucleotide-binding ability of Gtr1p. (A) E. coliproduced GST-Gtr1p (4 nmol), which were bound to the glutathione Sepharose-4B beads, were mixed in S buffer with [3H]GTP (33.8 Ci/mmol:150 pmol) in the presence of the indicated amount of cold GTP (•), GDP (○), or ATP (□). After incubation at 30° for 30 min, the beads were spun down, washed, and the radioactivity coprecipitated with the beads was quantified using a liquid scintillation counter. The vertical axis indicates the ratio (percentage) of the CPM bound to Gtr1p in the presence of the indicated amount of cold nucleotides, on the basis of CPM bound to Gtr1p without cold nucleotides. (B) A total of 0.8 nmol of the E. coli-produced His6-Gtr1p proteins [wild type (Gtr1p), Q65L (Gtr1-13p), and C-del. (Gtr1-12Δp)] that were bound to Ni-NTA-Agarose (Qiagen) and, as a control, the Ni-NTA-Agarose beads alone were mixed in S buffer with either [3H]GDP or GTP (33.8 Ci/mmol:150 pmol). After incubation at 30° for 30 min, the beads were spun down, washed, and the radioactivity coprecipitated with the beads was quantified using a liquid scintillation counter. The vertical axis indicates the ratio of the count per minute (CPM) bound to His6-Gtr1p and its derivatives on the basis of CPM bound to beads alone (None).
Self-interaction of Gtr1p depends upon the bound nucleotide state: The amount of HA-Gtr1p coprecipitated with GST-Gtr1p was reduced in the presence of an increasing amount of EDTA (Figure 4B, lanes 5 and 6). Because EDTA releases the nucleotides from Ran (Bischoffet al. 1995; Richardset al. 1995), this finding suggests that Gtr1p's interaction with itself is dependent upon the nucleotide bound. It is also possible, however, that the self-interaction of Gtr1p was inhibited by EDTA because of the instability of nucleotide-free Gtr1p, as was the case with nucleotide-free Ran (Klebeet al. 1995b).
Self interaction of Gtr1p
To determine whether self-interaction of Gtr1p is dependent upon the bound nucleotide state, putative GDP or GTP-bound mutants of Gtr1p were fused with either the GAL4 TAD or the GAL4 DBD. As shown in Table 3, when TAD-gtr1-Q65L was coexpressed with DBD-gtr1-Q65L, a strong transactivation of β-galactosidase was observed, but there was no transactivation of lacZ when TAD-gtr1-S20L was coexpressed with DBD-gtr1-S20L or when TAD-gtr1-S20N was coexpressed with DBD-gtr1-S20N. The immunoblotting analysis of yeast lysates revealed that the mutant Gtr1p proteins were present at levels similar to those of wild-type Gtr1p (Figure 4A). These findings, therefore, imply that the ability of Gtr1p to interact with itself is dependent upon the bound nucleotide state. This interpretation prompted us to ask whether the lack of interaction shown by the C-terminal deletion was caused by the loss of nucleotide-binding ability. To address this issue, His6-tagged Gtr1-12Δp, and as controls, His6-fused wild-type and Gtr1p-Q65L, were expressed in E. coli. Purified His6-fused Gtr1p was then examined for nucleotide-binding ability. Similar to the wild-type and Gtr1p-Q65L, Gtr1-12Δp bound efficiently to GTP (Figure 3B), revealing that the inability of Gtr1-12Δp to interact with itself was not caused by the loss of nucleotide-binding ability.
Gtr1p belongs to a novel family of small G proteins: RagA and RagB have been reported to be mammalian homologues of Gtr1p (Schurmannet al. 1995). By homology search, we found a novel S. cerevisiae homologue of GTR1, designated GTR2 (GenBank accession no. AB015239, Figure 5A), and a Caenorhabditis elegans gene possessing an open reading frame homologous to Gtr1p (GenBank accession no. Z49912/CET24F1-1, Figure 5A).
Gtr1p and its homologues do not have a lipid modification site that is characteristic of Ras (Bun-yaet al. 1992; Boguski and McCormick 1993). In this regard, Gtr1p is similar to Ran/Gsp1p (Drivaset al. 1990; Bischoff and Ponstingl 1991; Belhumeuret al. 1993). Phylogenetic tree analysis (Higginset al. 1992) revealed, however, that Gtr1p belonged to a family different from the Ran/Gsp1p family (Figure 5B). Indeed, all the Gtr1p homologues so far found possess histidine in the third conserved domain, the sequence of which is HKXD (Figure 5A), unlike the NKXD sequence found in the Ran/Ras family (Valenciaet al. 1991). These findings indicate that Gtr1p defines a novel family of G proteins.
GTR2 rescues gtr1-11, but not gtr1-1Δ: To investigate potential functional interactions between Gtr1p and Gtr2p, we examined the consequence of overexpressing Gtr2p. We amplified GTR2 by PCR using the primers shown in Figure 6A and inserted it into a multicopy vector, YEplac195, under its own promoter. Overproduction of Gtr2p partially rescued the cold sensitivity of the gtr1-11 strain, but not that of the gtr1-1Δ strain (Figure 6B). Thus, Gtr1p cannot be replaced by Gtr2p.
We also examined the consequences of Gtr2p loss by creating a null allele in the diploid N43 strain (Figure 6A). Cultures of the N43ΔGTR2 strain were sporulated and subjected to tetrad analysis. Most tetrads showed a ratio of viable to nonviable segregants of 4:0, demonstrating that GTR2 was not essential for survival (data not shown). The disruption of GTR2 in a haploid strain caused the yeast to become cold sensitive (Figure 6C). On the basis of this finding, we constructed a double-null disruptant (Δgtr1 Δgtr2) of GTR1 and GTR2 that grew well at 30°, the permissive temperature for each mutant, implying that Δgtr1 was not synthetically lethal with Δgtr2.
Self-interaction of Gtr2p requires Gtr1p: Because Gtr1p interacted with itself, we asked whether Gtr2p also interacts with itself. GTR2 was fused to either the GST- or HA-tag, and the constructs were introduced into the gtr2-1Δ strain. As a control, both GST and HA-GTR1 were coexpressed in the gtr1-1Δ strain. Transformants were selected in synthetic medium lacking uracil and tryptophan, crude extracts were prepared, GST-fused Gtr2p or Gtr1p was pulled down with glutathione Sepharose-4B beads, and the resultant precipitates were analyzed for the presence of HA-Gtr2p or Gtr1p by immunoblotting using the mAb to HA. As shown in Figure 7A, HA-Gtr2p was coprecipitated with GST-Gtr2p (lanes 5–8), similar to the case of Gtr1p (lanes 1–4).
—Self-interaction of Gtr1p. (A) Expression of GAL4-DBD-HA-fused, wild-type and mutated GTR1 clones in S. cerevisiae. Plasmid, pL44 (wild type, lane 1), pL208 (gtr1-15, lane 2), pL209 (gtr1-16, lane 3), pL66 (gtr1-11, lane 4), pL106 (gtr1-13, lane 5), and pL87 (gtr1-12Δ, lane 6) were introduced into the strain Y190. The Trp+ transformants selected were cultured in YPD medium at 30° until the early log phase, and then crude extracts were prepared and analyzed by immunoblotting using the mAb to HA-tag. Lane 7 was the extract from nontransfected Y190 cells. (B) Coprecipitation of GST and HA-fused Gtr1p. The two plasmids, pL38 carrying GST-GTR1 and pL80 carrying HA-GTR1 (lanes 3–6), as well as GST vector pEG-KG and pL80 as controls (lanes 1 and 2), were coexpressed in the strain gtr1-1Δ. (NBW5ΔGTR1). Prepared crude extracts were mixed with glutathione Sepharose-4B beads, and they were pulled down either in the absence (0, lanes 2 and 4) or presence of EDTA (1.0 and 10 mm, lanes 5 and 6). Proteins bound to beads were analyzed by immunoblotting with the mAb to HA-tag. Lanes 1 and 3, total crude extract.
Because GTR2 rescued the cold sensitivity of the gtr1-11 strain, we examined whether Gtr2p interacts with Gtr1p. GST-fused GTR2 or GTR1 was coexpressed with HA-fused GTR1 or GTR2 in the NBW5ΔGTR1/ΔGTR2 (gtr1-1Δ gtr2-1Δ) strain, as indicated in Figure 7B. From the crude extracts of Ura+ and Trp+ transformants, GST-fused proteins were pulled down with glutathione Sepharose-4B beads. The resultant precipitates were analyzed for the presence of HA-fused proteins by immunoblotting using the mAb to HA. As shown in Figure 7B, Gtr1p was coprecipitated with Gtr2p, indicating that Gtr2p forms complexes with Gtr1p. Interestingly, HA-Gtr2p was not coprecipitated with GST-Gtr2p in the NBW5ΔGTR1/ΔGTR2 strain (Figure 7C, lane 4). Hence, self-interaction of Gtr2p requires Gtr1p.
Disruption of GTR2 suppresses prp20: The requirement of Gtr1p for the self-interaction of Gtr2p suggested that the function of Gtr2p is dependent upon Gtr1p. This notion is consistent with the finding that the overexpression of Gtr2p rescued gtr1-11, but not gtr1-1Δ. Although we do not know whether Gtr2p interacts with GTP-Gtr1p, Gtr2p could be an effector downstream of Gtr1p. If so, prp20 might be suppressed by a defect in Gtr2p, because gtr1-11, which encodes a presumed inactive form of Gtr1p, suppresses prp20 and rna1-1. To address this issue, GTR2 was disrupted in the strain HS203 (prp20-1), as described in materials and methods. Resultant HS203ΔGTR2 (prp20-1 gtr2-1Δ) and HS203 strains were then transfected with wild-type gtr1-S20N or gtr1-Q65L carried on the multicopy vectors. Ura+ transformants were plated on synthetic medium lacking uracil and were then incubated at 26°, 30°, or 31°, which are the permissive, semipermissive, or non-permissive temperatures, respectively, for prp20-1.
As expected, loss of GTR2 suppressed prp20-1 (Figure 8A). Moreover, loss of GTR2 suppressed the inhibitory effects of gtr1-Q65L (Figure 8A, 30°). This finding is consistent with the idea that Gtr2p functions downstream of Gtr1p (see Figure 10). However, the disruption of GTR1 did not suppress prp20-1 (Figure 8B). Furthermore, the double disruption of GTR1 and GTR2 did not increase the ability to rescue the temperature sensitivity of prp20-1 (Figure 8B). Hence, it is unlikely that Gtr1p and Gtr2p have parallel functions in the Ran GTPase cycle (see Figure 10).
No effect on nucleocytoplasmic transport: The strain prp20/2c (prp20-1) has a defect in mRNA splicing and export (Forresteret al. 1992). To clarify the effect of Gtr2p on the Ran/Gsp1p GTPase cycle, we examined whether the mRNA export defect in prp20 can be rescued by overproduction of Gtr1p-S20L or disruption of GTR2. We also examined whether overproduction of Gtr1p-Q65L exacerbates the defect of mRNA export. When cultures of the prp20/2c (prp20-1) strain were incubated at 37°, the nuclear accumulation of poly(A)+ RNA was observed (data not shown). Such a defect in mRNA export was not rescued by overexpression of Gtr1p-S20L or disruption of GTR2. Furthermore, Gtr1p-Q65L did not show any effect on mRNA export. Consistent with the finding that the GTR1/GTR2 pathway has no effect on mRNA export, both Δgtr1 and Δgtr2 do not show any defect in either NLS-dependent nuclear protein import or nuclear export signal (NES)-dependent protein export, even at the nonpermissive temperature (data not shown). Hence, we presumed that Gtr2p regulates the Ran/Gsp1p GTPase cycle through unknown pathways other than the nucleus/cytosol exchange of macromolecules.
—Gtr1p constitutes a novel family of small GTPases. (A) Alignment of amino acid sequence of Gtr1p and its homologues. ORF (C. e.) is the C. elegans genome (GenBank accession number CET24F1-1). Asterisks and periods indicate amino acid residues that are identical and chemically conserved, respectively. Boxes indicate the amino acid sequences conserved among small GTPases. (B) Phylogenetic tree analysis of Gtr1p and other small GTPases. Using the whole amino acid sequence of listed GTPases, the evolutionary relationship was calculated by the CLUSTAL W multiple alignment program (Higginset al. 1992). The bar of 0.1 indicates 10% divergence figures between each pair of sequences. X, any amino acid; O, hydrophobic amino acid; J, hydrophilic amino acid.
—Function of the GTR2 gene. (A) Genomic map of the GTR2 gene. The region of chromosome VII's right arm containing the GTR2 gene is shown. The number on the map corresponds to the number of the nucleotide. For instance, the GTR2 gene begins at 3546 bp and ends at 4568 bp. Primers in which small letters indicate the nucleotides that were changed to make an indicated restriction enzyme site were used to disrupt the GTR2 gene and to clone the GTR2 gene. Restriction enzyme sites indicated on primers were created to engineer the recombinant GTR2. (B) Overexpression of Gtr2p rescues gtr111, but not Δgtr1. GTR2 carried on a multicopy YEplac195 pL130, and, as a control, GTR1 carried on the same vector, pL63, were introduced into the gtr1-11 strain (NN7-3B) and the gtr1-1Δ strain (NBW5ΔGTR1) as indicated. Ura+ transformants were plated on synthetic medium lacking uracil and incubated at 14°. (C) Δgtr2 is cold sensitive, but not synthetically lethal with Δgtr1. WT (NBW5), gtr1-1Δ (NBW5ΔGTR1), gtr2-1Δ (NBW5ΔGTR2), and gtr1-1Δ gtr2-1Δ (NBW5ΔGTR1/ΔGTR2) strains were plated on YPD medium and then incubated at 14°, 30°, or 37°, as indicated.
Localization of Gtr2p: Gtr1p has been reported to be localized within both the nucleus and the cytoplasm (Nakashimaet al. 1996). To determine the localization of Gtr2p, T7-tagged Gtr2p was expressed in the NBW5ΔGTR2 and NBW5ΔGTR1/ΔGTR2 strains. As a control, T7-tagged Gtr1p was expressed in the NBW5ΔGTR1 and NBW5ΔGTR1/ΔGTR2 strains. T7-tagged Gtr1p and Gtr2p rescued the cold sensitivity of Δgtr1 and Δgtr2 strains, respectively. Gtr1p was distributed throughout both the cytoplasm and the nucleus, as reported previously (Nakashimaet al. 1996). On the other hand, Gtr2p was concentrated in the nucleus, while some of Gtr2p was also localized in the cytoplasm (Figure 9). The staining pattern of Gtr2p was the same in Δgtr1 Δgtr2 cells (data not shown). Because Gtr1p and Gtr2p were produced in similar amounts, these results indicate that Gtr2p has a tendency to accumulate in the nucleus.
DISCUSSION
Gtr1p has been reported to bind only to GTP, so it was thought to be a putative G protein (Bun-yaet al. 1992). In this study, the fact that Gtr1p is indeed a guanine nucleotide-binding protein was shown by direct biochemical experiments. Specifically, E. coli-produced, wild-type Gtr1p binds either to GTP or GDP, although its ability to bind to GDP is low when compared with GTP. We also showed that Gtr1p-Q65L, presumed by analogy with Ras to be locked in the GTP-bound form, binds with GTP and GDP. Furthermore, the in vitro activity of Gtr1p depends upon its bound nucleotide state as follows. First, gtr1-S20N and gtr1-S20L, which are putative GDP-bound mutants of GTR1, suppress both prp20 and rna1, but gtr1-Q65L, which is a putative GTP-bound mutant, does not. Rather, overexpression of gtr1-Q65L is inhibitory for both prp20 and rna1 cells. Second, Gtr1p interacts with itself in a manner dependent upon the bound nucleotide state. Both wild-type and Gtr1p-Q65L interact with each other, but Gtr1p-S20N and Gtr1p-S20L do not. Thus, GTP-, but not GDP-bound Gtr1p, is suggested to form a complex with itself. We suspect that Gtr1p-Q65L accumulates as a GTP-bound form in cells because of its insensitivity to the GTPase activation enzyme. Taken together, these results suggest that Gtr1p is a G protein, although we could not demonstrate its GTPase activity. Phylogenetic tree analysis indicates that Gtr1p belongs to a novel G protein family that is composed of Gtr1p, Gtr2p, the mammalian homologues RagA and RagB, and an uncharacterized C. elegans open reading frame (GenBank accession no. Z49912/CET24F1-1). The protein encoded by the C. elegans genome is 46.5% homologous to Gtr1p and 65.0% homologous to human RagA, including chemically conserved amino acid residues.
—Gtr2p binds with itself in the presence of Gtr1p. (A) HA-GTR1 (pL80) and GST-GTR1 (pL38, lanes 3 and 4) and, as a control, HA-GTR1 and GST alone (pEG-KG, lanes 1 and 2) were coexpressed in the strain gtr1-1Δ (NBW5ΔGTR1). HA-GTR2 (pL154) and GST-GTR2 (pL155, lanes 7 and 8) and, as a control, HA-GTR2 and GST alone (lanes 5 and 6) were similarly coexpressed in the strain gtr2-1Δ (NBW5ΔGTR2) as indicated. Prepared crude extracts were mixed with the glutathione Sepharose-4B beads and pulled down. Proteins bound to beads (lanes 2, 4, 6, and 8) and, as a control, the total crude extracts (lanes 1, 3, 5, and 7), were analyzed by immunoblotting with the mAb to HA-tag. (B) HA-fused GTR1 (pL80, HA-1) or GTR2 (pL154, HA-2) was coexpressed with GST-fused GTR2 (pL155, GST-2) or GTR1 (pL38, GST-1), or with GST alone (pEG-KG)(–) in the strain gtr1-1Δ gtr2-1Δ (NBW5ΔGTR1/ΔGTR2), as indicated. Prepared crude extracts were mixed with the glutathione Sepharose-4B beads and pulled down. Total crude extracts (lanes 1, 3, 5, and 7) and proteins bound to beads (lanes 2, 4, 6, and 8) were analyzed by immunoblotting with the mAb to HA-tag. (C) Either HA-GTR1 (pL80) and GST-GTR1 (pL38, lanes 1 and 2), or else HA-GTR2 (pL154) and GST-GTR2 (pL155, lanes 3 and 4), were coexpressed in the strain gtr1-1Δ gtr2-1Δ (NBW5ΔGTR1/ΔGTR2). Crude extracts were mixed with the glutathione Sepharose-4B beads and pulled down. Total crude extracts (lanes 1 and 3) and proteins bound to beads (lanes 2 and 4) were analyzed by immunoblotting with the mAb to HA-tag.
Gtr2p is homologous to Gtr1p. As for Gtr1p, Gtr2p interacts with itself. However, self-interaction of Gtr2p requires Gtr1p. This finding indicates that Gtr1p forms a complex with Gtr2p. Given that Gtr1p and Gtr2p form a complex, it is noteworthy that they exhibit some of the same genetic interactions: disruptions of GTR2 and the gtr1-11 mutation both suppress prp20-1. We assume that Gtr2p is an effector downstream of Gtr1p, as shown in Figure 10. The fact that the loss of GTR2 suppresses prp20-1 is consistent with the presumption that gtr1-11 encodes a putative GDP-bound, inactive mutant of Gtr1p. The inactive G protein could not turn on the downstream cascade; this resulted in the same effect as the loss of a downstream effector. Consistent with this interpretation, Gtr1p-Q65L, a putative GTP-bound and, therefore, active form of Gtr1p, inhibits the growth of prp20-1 and rna1-1 strains. The fact that the growth inhibitory effect of Gtr1p-Q65L on prp20-1 is abolished by the disruption of GTR2 is consistent with the notion that Gtr2p is a downstream effector of Gtr1p. These results suggest that GTP-Gtr1p has a negative effect on both Prp20p and Rna1p. Taking account of the fact that Prp20p and Rna1p are the GDP/GTP-exchanging and GTPase-activating factors of Gsp1p, respectively, it would seem that Gtr1p may negatively regulate the Ran/Gsp1p cycle through Gtr2p (Figure 10).
—Effect of the gene disruptions gtr2-1Δ and gtr1-1Δ on prp20-1. (A) YEplac195 plasmids carrying the indicated mutated and wild-type GTR1 genes and, as a control, the vector alone were introduced into the strains HS203 (prp20-1) and HS203ΔGTR2(prp20-1 gtr2-1Δ), respectively. Ura+ transformants plated on synthetic medium lacking uracil were incubated at the indicated temperature. (B) Strains HS203 (prp20-1), HS203ΔGTR1 (prp20-1 gtr1-1Δ), HS203ΔGTR2 (prp20-1 gtr2-1Δ), and HS203ΔGTR1,2 (prp20-1 gtr1-1Δ gtr2-1Δ) were plated on a YPD medium plate and incubated at 31° and 26°, as indicated.
The fact that the disruption of GTR1 does not suppress prp20-1, however, suggests that an inhibitory function of Gtr2p on the Ran/Gsp1p cycle is also activated by some factor other than Gtr1p (Figure 10, X). This finding also indicates that Gtr1p-S20L dominantly abolishes the negative effect of Gtr2p on the Ran/Gsp1p cycle. In this regard, both GTP and GDP-bound Gtr1p may make a complex with Gtr2p. It has been reported recently that the Rho family members Cdc42 and Rac2 form homodimers in the GTP-bound state, and that one of the GTP-bound proteins stimulates the GTP hydrolysis of the other protein (Zhang and Zheng 1998). This way, the GDP-Cdc42/GTP-Cdc42 dimer is produced as a transient state. We presume that Gtr1p-S20L forms a complex with GTP-Gtr2p and inhibits the GTPase activity of Gtr2p, although we do not know for sure that Gtr2p is a G protein.
—Nuclear localization of Gtr2p. T7-tagged GTR1 and GTR2 were introduced into strains gtr1-1Δ and gtr2-1Δ, respectively. Ura+ transformants selected in synthetic medium lacking uracil were fixed and stained with the mAb to T7 tag as described (Nakashimaet al. 1996). DNA was stained with 4′,6-daimidino-2-phenylindole (DAPI).
The finding that the nucleus/cytosol exchange of macromolecules is not affected by Gtr1p-S20L, Gtr1p-Q65L, or Δgtr2 suggests that the unknown pathways of Ran/Gsp1p other than the nucleus/cytosol exchange of macromolecules are regulated by the Gtr1p/Gtr2p pathway. Dis3p and RanBPM have previously been reported to interact with Gsp1p and Ran. Dis3p is localized in the nucleolus and is suggested to be required for ribosomal RNA processing (Mitchellet al. 1997; Shiomiet al. 1998). On the other hand, RanBPM is localized in the centrosome, which suggests that it may be involved in microtubule aster formation (Nakamuraet al. 1998). Although Gtr1p is distributed in both the cytoplasm and the nucleus, Gtr2p seems to be accumulated in the nucleus. We previously found that the human Gtr1p homologue RagA changes its cellular localization depending upon the bound nucleotide state. An interesting question is whether the nuclear localization of Gtr2p is also dependent upon the bound nucleotide state to regulate the Ran/Gsp1p cycle.
—Interaction of the Gtr1p/Gtr2p cascade with the Gsp1p GTPase cycle.
Acknowledgments
We thank M. Funakoshi (Kyushu University) for plasmid YEplac112-XA; H. Sumimoto (Kyushu University) for helpful discussion; and C. Nowak, S. Malak, C. Villa-Braslavsky, J. Kuhlmann, and A. Wittinghofer (Max Planck Institut, Dortmund, Germany) for their help in analyzing E. coli-produced Gtr1p at the beginning of this work. This work was supported by Grants-in-Aid for Specially Promoted Research and by Human Frontier Science Program. The English used in this manuscript was revised by K. Miller (Royal English Language Centre, Fukuoka, Japan).
Footnotes
-
Communicating editor: A. G. Hinnebusch
- Received September 24, 1998.
- Accepted March 18, 1999.
- Copyright © 1999 by the Genetics Society of America
