Conditional cdc1(Ts) mutants of S. cerevisiae arrest with a phenotype similar to that exhibited by Mn2+-depleted cells. Sequence similarity between Cdc1p and a class of Mn2+-dependent phosphoesterases, as well as the observation that conditional cdc1(Ts) growth can be ameliorated by Mn2+ supplement, suggests that Cdc1p activity is sensitive to intracellular Mn2+ levels. This article identifies several previously uncharacterized cdc1(Ts) suppressors as class E vps (vacuolar protein sorting) mutants and shows that these, as well as other vps mutants, accumulate high levels of intracellular Mn2+. Yeast VPS genes play a role in delivery of membrane transporters to the vacuole for degradation, and we show that the vps mutants accumulate elevated levels of the high-affinity Mn2+ transporter Smf1p. cdc1(Ts) conditional growth is also alleviated by mutations, including doa4 and ubc4, that compromise protein ubiquitination, and these ubiquitination defects are associated with Smf1p accumulation. Epistasis studies show that these suppressors require functional Smf1p to alleviate the cdc1(Ts) growth defect, whereas Smf1p is dispensable for cdc1(Ts) suppression by a mutation (cos16/per1) that does not influence intracellular Mn2+ levels. Because Smf1p is ubiquitinated in vivo, we propose that Smf1p is targeted to the vacuole for degradation by ubiquitination-dependent protein sorting.
THE trace element Mn2+ is essential for growth of the yeast Saccharomyces cerevisiae (Loukin and Kung 1995). Although the essential Mn2+-dependent process is not known, Mn2+-requiring enzymes include a Golgiresident glycosyltransferase that is required for bud growth at extreme temperatures (Nakayamaet al. 1992) and an essential metalloprotease that is involved in mitochondrial protein import (Supeket al. 1996). Because Mn2+ is also toxic at high concentrations, microbes have complex mechanisms to regulate intracellular Mn2+ levels in the face of wide fluctuations in environmental Mn2+.
Yeast high-affinity Mn2+ uptake is mediated by Smf1p, a protein that bears structural homology with the Nramp (natural resistance associated macrophage protein) family of mammalian metal ion transporters (Supeket al. 1996). Strains containing SMF1 on a high-copy plasmid exhibit elevated levels of Mn2+, whereas deletion of SMF1 lowers intracellular Mn2+ concentration and imparts sensitivity to the chelator EGTA (Supeket al. 1996). Functional Smf1p is a plasma membrane protein that is tightly regulated by external Mn2+ (Westet al. 1992; Supeket al. 1996; Liu and Culotta 1999a,b). In Mn2+-limiting medium, Smf1p accumulates on the plasma membrane, whereas in Mn2+-rich medium Smf1p undergoes rapid proteolysis in the vacuole (Liu and Culotta 1999b).
The mechanism by which Mn2+ regulates Smf1p localization is not fully understood, although recent studies have implicated the endoplasmic reticulum (ER) resident protein, Bsd2p (Liu and Culotta 1994; Liuet al. 1997). bsd2 mutants contain elevated Mn2+ levels and accumulate Smf1p, even when extracellular Mn2+ levels are high (Liuet al. 1997; Liu and Culotta 1999b). These results support a model in which Bsd2p is required for proper sorting of Smf1p between the plasma membrane and the vacuole (Liu and Culotta 1999b). Interestingly, a mutant form of the plasma membrane protein Pma1p, which is sent to the vacuole along the biosynthetic route of the vacuolar protein sorting (VPS) pathway, is rerouted back to the plasma membrane in a bsd2 mutant (Luo and Chang 1997). This last observation suggests that Bsd2p plays a general role in membrane protein targeting, controlling the trafficking of Pma1p and Smf1p between the plasma membrane and the vacuole. Although it is also consistent with the notion that Smf1p is delivered to the vacuole by the same targeting pathway as aberrant Pma1p, current evidence suggests only that endocytosis does not contribute significantly to Smf1p localization or turnover (Liu and Culotta 1999b).
Mn2+ homeostasis is also regulated by intracellular storage. For example, the Golgi protein Pmr1p, a Ca2+-ATPase homolog, regulates cytosolic Mn2+ levels by pumping Mn2+ from the cytosol into the Golgi (Lapinskaset al. 1995). This has the dual property of reducing cytosolic Mn2+ levels and increasing Mn2+ levels in the Golgi, where it is required and can be exported out of the cell (Lapinskaset al. 1995; Durret al. 1998; Weiet al. 1999). A second Golgi protein, Atx2p, plays a positive role in Mn2+ homeostasis (Lin and Culotta 1996). ATX2 was first identified in a high-copy screen for genes whose overexpression increased intracellular Mn2+ (Lin and Culotta 1996) and was subsequently shown to be essential for the maintenance of normal intracellular Mn2+ levels (Lin and Culotta 1996). The mechanism by which Atx2p regulates intracellular Mn2+ levels is not known. Finally, Mn2+ is transported from the cytoplasm to the vacuole by the vacuolar membrane protein Ccc1p (Lapinskaset al. 1996; Liet al. 2001).
Another yeast protein implicated in Mn2+ metabolism/function is Cdc1p. That attribution was based on the observation that cdc1(Ts) mutants exhibit a smallbud growth arrest that is identical to the phenotype displayed by Mn2+-depleted cells (Loukin and Kung 1995; Paidhungat and Garrett 1998a). Consistent with that hypothesis, cdc1(Ts) growth is exacerbated or alleviated by manipulations that respectively deplete or supplement intracellular Mn2+ (Loukin and Kung 1995; Supeket al. 1996; Paidhungat and Garrett 1998a). Although we initially proposed that Cdc1p played a central role in Mn2+ homeostasis (Paidhungat and Garrett 1998a), these observations can also be accommodated by a model in which Cdc1p is a Mn2+-dependent protein (Supeket al. 1996). The latter model is supported by the weak, but significant, sequence similarity between Cdc1p and Mn2+-dependent phosphoesterases (Rusnak 2000), as well as by the observation that Cdc1p activity does not correlate with cellular Mn2+ levels (Paidhungat and Garrett 1998a).
Several suppressors of the cdc1(Ts) growth defect identified genes implicated in vacuole protein sorting (Paidhungat and Garrett 1998b). Strains lacking VPS gene function exhibit a myriad of defects in vacuolar protein trafficking and function (Bryant and Stevens 1998). In this article, we show that genes corresponding to three previously uncharacterized cdc1(Ts) suppressors (Paidhungat and Garrett 1998b) are identical to members of a group of VPS genes (the so-called class E VPS genes) that are required for proper processing of membrane proteins through the late endosome, or prevacuolar, compartment (Bantaet al. 1988; Raymondet al. 1992; Babstet al. 1998) into the vacuole. These and other vps mutants accumulate high levels of intracellular Mn2+, as well as elevated amounts of the high-affinity Mn2+ transporter, Smf1p. Because cdc1(Ts) suppression is dependent upon Smf1p, these results suggest that the vps mutations suppress the cdc1(Ts) growth defect by increasing Smf1p-dependent Mn2+ levels. A second group of cdc1(Ts) suppressors defines a class of proteins that function in ubiquitination of proteins targeted for vacuolar turnover. These ubiquitin processing mutants also accumulate Smf1p, which is consistent with the observation that Smf1p is ubiquitinated in vivo.
MATERIALS AND METHODS
Media and standard genetic techniques: Standard yeast media were prepared as described (Kaiseret al. 1994). LiCl was added to rich yeast medium (YEPD) after autoclaving. Medium containing CaCl2 or MnCl2 was made by autoclaving YEPD, buffering to pH 5.5 with 50 mm succinate/NaOH, and adding CaCl2 or MnCl2 to the indicated concentration. Genetic and molecular manipulations of yeast strains were carried out by standard techniques (Kaiseret al. 1994).
Yeast strains: Yeast strains are listed in Table 1. Disruption derivatives of FY11, FY70, FY388, and YR98 were made by transformation with digested plasmid DNA or gel-purified PCR products from the Yeast Deletion Consortium KanMx disruption strains using the recommended A and D primers (Open Biosystems). Strain 27038a (generously provided by C. Michaels) contains the rsp5/npi1 mutation and was crossed with FY11 to determine if the cdc1-1(Ts) growth defect could be suppressed by the rps5 mutation. Strain Y152 was isolated from a cross between FY11 and 12992 (BY4742 end3Δ::KanMx) from Open Biosystems.
Plasmids: The high-copy (YEp13-CDC1) and low-copy (YCp50-CDC1) CDC1 plasmids have been described (Paidhungat and Garrett 1998b). The high-copy SMF1 (YEp24-SMF1) and smf1 disruption (smf1Δ::URA3) plasmids (Supeket al. 1996), as well as the low-copy SMF1-HA2 (pSF10) plasmid (Liu and Culotta 1999a), were generous gifts of V. Culotta (The Johns Hopkins University). Plasmid YEp105 contains a myctagged allele of ubiquitin gene UBI4 expressed from the copper-inducible CUP1 promoter (Medintzet al. 1998). The highcopy SMF1-HA plasmid pLE59 was constructed by cloning the SacI-XhoI fragment containing SMF1-HA from pSF10 into the same sites of the high-copy URA3 vector pRS202. A high-copy SMF1-GFP plasmid was constructed in several steps. SMF1 was amplified by PCR using primers 5′-CTCGAGCGAAATCTCTC CAAAGGGTG-3′ and 5′-GGATTCCACTGATATCACCATGA GACATGCC-3′, which left XhoI and BamHI sites 700 bp upstream of the promoter and at the last codon, respectively. The PCR fragment was digested with XhoI and BamHI and cloned into the corresponding sites of pRS306 to generate pYZ18. A SMF1-GFP fusion was created by inserting an inframe 800-bp BglII-NotI fragment containing green fluorescent protein (GFP) into the BamHI and NotI sites to generate pYZ20. Finally, the 2.8-kb XhoI-NotI fragment containing SMF1-GFP was removed from pYZ20 and inserted into the same sites of pRS202 to generate pLE61. The vps18Δ::HIS3, vps4Δ::RA3, and cos16Δ::LEU2 disruption plasmids (pFB107, pFB275, and pFB371) have been described (Paidhungat and Garrett 1998b).
Cloning VPS36 (COS3), VPS32/SNF7 (COS14), and VPS20 (COS1): The wild-type VPS36 (COS3), VPS32 (COS14), and VPS20 (COS1) genes were isolated from a low-copy, yeast genomic library (Roseet al. 1987) by their ability to confer Ca2+ and Li+ tolerance to cdc1-1(Ts) cos3-3 [YEp13-CDC1], cdc1-1(Ts) cos14-3 [YEp13-CDC1], and cdc1-1(Ts) cos1-1 [YEp13-CDC1] strains, respectively. One VPS36 plasmid (pFB171) was recovered from 6000 Ura+ transformants. The vps36Δ::LEU2 plasmid, pFB281, was generated by cloning the 2.9-kb SalI fragment from plasmid pFB171 into pUC8 and replacing the 1.3-kb EcoRV fragment of VPS36 with the 2.2-kb HpaI LEU2 fragment. Plasmid pFB281 was digested with HindIII prior to transformation.
The VPS32/COS14 gene was identified from four complementing plasmids with overlapping genomic inserts (pFB175-pFB178) that were recovered from a total of 4000 cdc1-1(Ts) cos14-103 [YEp13-CDC1] Ura+ transformants. The snf7Δ::URA3 plasmid pUU4 (Tuet al. 1993) was obtained from M. Carlson, Columbia University. A vps32/snf7Δ::LEU2 plasmid was constructed by inserting the 2.2-kb HpaI LEU2 fragment at the unique StuI site in the URA3 region of pUU4. The resulting plasmid, pFB316, was digested with PvuII and SphI prior to transformation.
The VPS20/COS1-containing plasmid (pFB173) was recovered from 5000 Ura+ transformants of the cdc1-1(Ts) cos1-1 [YEp13-CDC1] strain. Plasmid pFB209 was constructed by inserting a 2.3-kb XhoI-BamHI fragment containing VPS20 into the lowcopy URA3 vector pRS316 (Sikorski and Hieter 1989). The vps20Δ::LEU2 plasmid was constructed in two steps. The 2.3-kb XhoI-BamHI VPS20 fragment from pFB209 was subcloned into pBSKII+ (Stratagene, La Jolla, CA) to obtain plasmid pFB244. The 0.3-kb EcoRI fragment in pFB244 was replaced with a 1.2-kb EcoRI URA3 fragment to produce plasmid pFB245. Finally, a 2.2-kb HpaI LEU2 fragment was inserted into the StuI site within the URA3 marker in plasmid pFB245 to generate the vps20Δ::LEU2 plasmid pFB337. pFB337 was digested with XbaI prior to transformation.
Cellular Ca2+ and Mn2+ content: Cellular Ca2+ content was measured using radionuclide 45Ca2+ tracer as described (Paidhungat and Garrett 1997). Cellular Mn2+ content was measured by atomic absorption spectroscopy as described (Paidhungat and Garrett 1998b).
Western blot analysis and immunoprecipitation: The hemagglutinin (HA)-tagged Smf1 protein was detected by Western blot analysis according to published protocols (Lui and Culotta 1999b). Strains containing the low-copy SMF1-HA2 plasmid, pSF10, or the high-copy SMF1-HA2 plasmid, pLE59, were grown in selective medium until midlog phase. Extracts were prepared by alkalai lysis and then separated by SDS-polyacrylamide electrophoresis. The Smf1p-HA2 protein was visualized using 1/1000 dilution of ascites fluid anti-HA monoclonal antibody 12CA5 (Covance) diluted 1/1000 and 1/10,000 secondary HRP-congugated goat-anti-mouse antibody (Covance).
The Smf1-HA2 protein was immunoprecipitated from extracts using ascites fluid of the anti-HA monoclonal antibody 12CA5 diluted 1/200. Extracts were prepared by bead beating and centrifugation (Garrettet al. 1991) and then incubated with anti-HA antibody at 1/200 dilution for 1 hr at 4°. Antigen-antibody complexes were isolated by adding protein G Sepharose (Invitrogen, San Diego) to the extract for 1 hr at 4° and then washing six times in immunoprecipitation buffer (Garrettet al. 1991) before separating proteins by SDS-polyacrylamide electrophoresis and then transferring to nitrocellulose by Western blot. The nitrocellulose filters were incubated with 1/1000 dilution of the monoclonal anti-Myc antibody 9E-10 to visualize Myc-tagged ubiquitin, stripped, and then incubated with 1/1000 dilution of the monoclonal anti-HA antibody 12CA5 to visualize HA-tagged Smf1p.
Immunofluorescence: Cells containing the Smf1-GFP protein were grown to midlog phase (OD600 = 0.5), incubated with Hoechst 33258 (for DNA) or FM4-64 (for vacuole membrane or class E compartment) for 10 min, washed, and then incubated for 30 min before visualizing by epifluorescence or light microscopy.
cdc1(Ts) suppression by class E vps mutants: Genetic analysis showed that the cdc1-1(Ts) growth defect can be alleviated by mutations in class C and class D VPS genes (Paidhungat and Garrett 1998b). Three unidentified cdc1 suppressors (cos1, cos3, and cos14) conferred sensitivity to ions such as Ca2+ and Li+ (data not shown) and missorted the vacuolar protein CPY (Paidhungat and Garrett 1998b). Wild-type genes for these suppressors were cloned by isolating plasmids that could restore Li+ and Ca2+ resistance to the appropriate CDC1 cos strain, as well as temperature-sensitive growth to the corresponding cdc1-1(Ts) cos mutant. Physical and genetic analysis (materials and methods) revealed that all three contain mutations in class E VPS genes: COS3 was allelic with VPS36, COS14 was identical to VPS32, and COS1 was allelic with VPS20 (data not shown). Disruption of each gene recapitulated cdc1-1(Ts) suppression of the original recessive suppressors (Figure 1A). To determine if cdc1(Ts) suppression and ion sensitivity are general characteristics of class E vps mutations, we tested several phenotypes of the class E mutants listed in Table 2. Interestingly, although cdc1(Ts) suppression and temperature-sensitive growth are general attributes of class E vps mutations (Table 2), a different pattern emerged when class E mutants were characterized according to ion sensitivity. Mutants lacking individual components of ESCRT-I and ESCRT-II were extremely sensitive to Ca2+, Li+, and Mn2+, as were strains lacking ESCRT-III components Vps20p and Vps32p (see Figure 1B and Table 2). By contrast, mutants lacking ESCRT-III components Vps24p and Vps2p, components of the Vps27p/Hsc1p complex, and the AAA-ATPase Vps4p exhibited wild-type resistance to Ca2+ and Li+ and only moderate sensitivity to Mn2+ (Figure 1B; Table 2).
vps mutations and ion homeostasis: Conditional cdc1(Ts) growth can be partially alleviated by mutations that restrict Ca2+ accumulation (Paidhungat and Garrett 1997); however, Ca2+ levels of vps36Δ::LEU2, vps32Δ::URA3, and vps20Δ::URA3 mutants were similar to the Ca2+ level of the wild-type parent (Table 3), indicating that vps mutations do not alleviate the cdc1-1(Ts) growth defect by altering intracellular Ca2+ levels. Elimination of the vacuolar proton gradient by deletion of genes VPH6 or VMA2 (Yamashiroet al. 1990; Hemenwayet al. 1995; Bryant and Stevens 1998) was synthetically lethal with the cdc1-1(Ts) conditional mutation (data not shown), indicating that vps mutations do not alleviate the cdc1-1(Ts) growth defect by inactivating V-ATPase function.
We previously proposed that class C and D vps mutants suppressed the cdc1(Ts) growth defect by mislocalizing a putative vacuolar Mn2+ transporter, Cos16p, thereby elevating cytoplasmic Mn2+ levels (Paidhungat and Garrett 1998b). However, cos16 deletion strains exhibit wild-type Mn2+ levels (Paidhungat and Garrett 1998b) and disruption of a known vacuolar Mn2+ transporter, CCC1 (Liet al. 2001), failed to suppress conditional growth of either the cdc1-1(Ts) or the cdc1-2(Ts) mutant (Figure 3B; Table 4). Thus, vacuole Mn2+ sequestration does not appear to play an important role in cdc1(Ts) growth.
To determine if vps mutations alter intracellular Mn2+ levels, we examined Mn2+ contents of several vps mutants by atomic absorption spectroscopy. The vps32Δ and vps20Δ mutants exhibited threefold higher intracellular Mn2+ levels than the isogenic wild-type strain, and accumulation was independent of the CDC1 allele (Table 3). At least two other cdc1-1(Ts) suppressors that affect vacuole protein sorting, vps18Δ and vps4Δ, also resulted in high levels of intracellular Mn2+ (Table 3), suggesting that Mn2+ accumulation is not restricted to class E vps mutants.
vps mutants accumulate the high-affinity Mn2+ transporter, Smf1p: Except under Mn2+-limiting conditions, the high-affinity Mn2+ transporter, Smf1p, is directed to the vacuole where it is rapidly degraded (Liu and Culotta 1999a,b). The route and mechanism of Smf1p vacuolar delivery is unknown; however, endocytosis does not play a significant role (Liu and Culotta 1999a). To determine if Smf1p accumulation could account for elevated Mn2+ levels and cdc1(Ts) suppression of the vps mutants, we measured levels of a functional HA-tagged Smf1p protein (Liu and Culotta 1999a). Western analysis showed that the vps18Δ mutant accumulated Smf1p to levels similar to those found in a pep4Δ control, whereas Smf1p was undetectable in the wildtype strain (Figure 3). Even class E vps mutants vps20Δ, vps32Δ, and vps24Δ, which accumulate less Mn2+ (Table 3 and data not shown) and display a less dramatic trafficking defect than class C vps mutants (Bryant and Stevens 1998), contained higher levels of the HA-tagged protein than the wild-type strain (Figure 2). Thus, the increase in intracellular Mn2+ is correlated with an increase in Smf1p.
Smf1p accumulation is necessary and sufficient for cdc1-1(Ts) suppression: If vps mutations alleviate the cdc1-1(Ts) growth defect by increasing Smf1p levels, genetic alterations that result in Smf1p accumulation in the plasma membrane should restore growth. A highcopy SMF1 plasmid is capable of elevating intracellular Mn2+ levels and alleviating the EGTA sensitivity of several cdc1 mutants (Supeket al. 1996; Paidhungat and Garrett 1998a). Temperature-sensitive growth defects of cdc1-1(Ts) (Figure 3A; Table 4) and cdc1-2(Ts) (Table 4) strains were alleviated by a high-copy YEp24-SMF1 plasmid, but were unaffected by the high-copy vector YEp24 and the low-copy SMF1 plasmid pSF10. Moreover, a bsd2Δ mutant exhibited Smf1p-dependent Mn2+ accumulation (Liuet al. 1997), and both cdc1-1(Ts) bsd2Δ (Table 4) and cdc1-2(Ts) bsd2Δ (Figure 3B; Table 4) strains grew at the nonpermissive temperature.
To determine if Smf1p is necessary for suppression, we asked if the cdc1(Ts) growth defect could be alleviated in strains lacking Smf1p. A cdc1-1(Ts) smf1Δ double mutant is inviable at all temperatures (Paidhungat and Garrett 1998a), and this synthetic lethality was insensitive to each of three vps mutations [none of >40 predicted triple-mutant spores were viable in crosses between a MATa smf1Δ strain and MATα cdc1-1(Ts) strains containing vps32Δ, vps20Δ, or vps18Δ]. By contrast, the cos16Δ mutation, which alleviates the cdc1-1(Ts) growth defect without affecting intracellular Mn2+ levels (Paidhungat and Garrett 1998b), suppressed synthetic lethality and the temperature-sensitive growth of the cdc1-1(Ts) smf1Δ double mutant [six of six predicted cdc1-1(Ts) smf1Δ cos16Δ triple mutants grew at both 23° and 30°]. Tests of epistasis were also carried out in the cdc1-2(Ts) mutant, which exhibits wild-type growth at 23° and is inviable at 34° (Paidhungat and Garrett 1998a). Introduction of each of several vps mutations (vps20Δ and vps32Δ), as well as the bsd2Δ mutation, into the cdc1-2(Ts) strain alleviated the 34° growth defect (Figure 3B). By contrast, the same suppressors were unable to alleviate the 34° growth defect of the cdc1-2(Ts) smf1Δ double mutant (Figure 3B; Table 4).
cdc1-1(Ts) suppression and ubiquitin-dependent protein turnover: To identify new cdc1-1(Ts) suppressors, we transformed a high-copy yeast genomic library into the cdc1-1(Ts) strain and isolated six plasmids that conferred temperature-resistant growth. Four plasmids identified full-length VPS genes: VPS4, a class E gene (Babstet al. 1997), was isolated three times, and VPS13, a class A gene (Bryant and Stevens 1998), was isolated once (Table 4). The isolation of VPS4 as a high-copy suppressor is consistent with our previous observation that VPS4 overexpression results in a vps-like defect (Paidhungat and Garrett 1998b). A fifth plasmid contained ATX2, which was previously identified in a high-copy screen for genes that result in Mn2+ accumulation (Lin and Culotta 1996).
The sixth high-copy suppressor contained the ubiquitin hydrolase gene, UBP1 (Table 4). The Ubp1p hydrolase removes ubiquitin from protein substrates (Tobias and Varshavsky 1991), and a UBP1 homolog from Kluyveromyces lactis was recently isolated as a high-copy suppressor of a temperature-sensitive centromere-binding protein mutant of S. cerevisiae (Winkleret al. 2000). Identification of UBP1 as a high-copy suppressor suggested a role for ubiquitination in Smf1p regulation. Although proteosomal degradation does not contribute to Smf1p turnover (Liu and Culotta 1999b), ubiquitination has recently been implicated in other cellular functions, including ubiquitin-mediated endocytosis (Hicke and Riezman 1996; Hicke 1999) and protein delivery from the Golgi to the vacuole (Ameriket al. 2000; Helliwellet al. 2001; Katzmannet al. 2001). To determine if other ubiquitin-related functions affected Smf1p accumulation, we asked if the cdc1(Ts) growth defect could be alleviated by mutations in genes previously implicated in ubiquitination of vacuole-targeted proteins [DOA4, ubiquitin recycling (Ameriket al. 2000); RSP5, E3 ubiquitin ligase (Galanet al. 1996); UBC4, E2 conjugating enzyme (Hochstrasser 1996); and BRO1, unknown activity (Springaelet al. 2002)]. All four mutations alleviated the cdc1-1(Ts) and/or cdc1-2(Ts) growth defects, and suppression was dependent on the presence of a functional SMF1 gene (Figure 3; Table 4). Ubiquitination does not appear to influence cdc1(Ts) growth by affecting endocytosis because the cdc1-2(Ts) growth defect was not alleviated by an end3Δ mutation (Table 4).
Smf1p is targeted to the vacuole by ubiquitination: To determine the role of ubiquitination in cdc1(Ts) suppression, we examined Smf1p levels in ubiquitinprocessing mutant ubc4Δ and a doa4Δ pep4Δ double mutant. The DOA4 gene encodes a ubiquitin recycling protease, loss of which results in diminished ubiquitin pools. Because Doa4p rapidly removes ubiquitin from modified proteins, its loss paradoxically stabilizes the ubiquitin-modified form of a protein if the protein itself is not subject to turnover (Swaminathanet al. 1999). Because Smf1p is degraded in the vacuole, the pep4Δ mutation was used to stabilize all forms of Smf1p. As predicted, both ubc4Δ and the doa4Δ pep4Δ doublemutant strains accumulated more Smf1p protein than did the wild-type parent (Figure 4A). Moreover, a significant proportion of the cross-reacting activity from the doa4Δ pep4Δ double mutant migrated as a broad, high-molecular-weight smear (Figure 4A). Smf1p also migrated as a higher-molecular-weight species in the isogenic bsd2Δ mutant. These species are apparent in several figures of two earlier publications and were thought to be due to unknown protein-processing events (Liu and Culotta 1999a,b).
To determine the nature of the broad Smf1p-specific smear, we examined the ubiquitination status of Smf1p protein immunoprecipitated from the doa4Δ pep4Δ double mutant. Because the doa4Δ mutation limits available ubiquitin, the strain contained a plasmid (YEp105) that encodes a Cu2+-inducible, myc-tagged ubiquitin moiety (Medintzet al. 1998). As can be seen in Figure 4B, the Smf1-HA tagged protein immunoprecipitated from the doa4Δ pep4Δ double mutant migrated as a broad, highmolecular-weight smear that cross-reacted with the mycspecific antibody. Thus, Smf1p appears to be ubiquitinated.
To determine the disposition of the Smf1p that accumulates in the ubc4Δ mutant, we localized a functional Smf1-GFP fusion protein in wild-type and ubc4Δ strains. Whereas wild-type cells displayed little to no fluorescence, the ubc4Δ mutant exhibited significant staining of a contiguous internal structure that surrounded the nucleus (as delineated by Hoechst staining), as well as patches at or just beneath the cell surface (Figure 5A). This localization is consistent with the majority of Smf1p accumulating in the ER. Modest perimeter staining in the ubc4Δ mutant (Figure 5A) colocalized with the lipophilic dye FM4-64 after a brief pulse (data not shown), suggesting that some Smf1p was also accumulating at the plasma membrane; however, intense staining of ER structures at the perimeter made it impossible to state definitively that Smf1 protein was accumulating on the plasma membrane. Previous studies have shown that bsd2 mutants accumulate the bulk of Smf1p within the Golgi (Liu and Culotta 1999b). Because bsd2 mutants divert misfolded Pma1p from the vacuole to the plasma membrane (Luo and Chang 1997), accumulate high levels of Smf1p-dependent intracellular Mn2+ (Liuet al. 1997; Liu and Culotta 1999b), and suppress the cdc1(Ts) growth defect (Figure 3B), it seems likely that Smf1p also accumulates on the plasma membrane of bsd2 cells. Thus, at least some Smf1p may accumulate on the plasma membrane of ubc4 cells.
To examine the effect of a different class of cdc1(Ts) suppressors on Smf1p accumulation, we localized Smf1p in several vps mutants. In the class E mutants vps32 (Figure 5B) and vps20 (data not shown), Smf1-GFP fluorescence appeared as one or two spots that colocalized with FM4-64 staining of the class E compartment (Bryant and Stevens 1998). In the wild-type strain (Figure 5B), by contrast, Smf1-GFP was not apparent and FM4-64 was found in the limiting membrane of the vacuole. Several other membrane proteins (Ste2p and Ste6p) accumulated in both the class E compartment and the plasma membrane of class E vps mutants (Liet al. 1999; Kranzet al. 2001), suggesting that at least some Smf1p is diverted to the plasma membrane of the vps mutants.
Cdc1p and Mn2+ homeostasis: The yeast Cdc1p protein is required for growth and has been implicated in such processes as recombination, actin polarization, bud growth, shmoo formation, and spindle pole body duplication (Halbrook and Hoekstra 1994; Paidhungat and Garrett 1998a; Rossaneseet al. 2001). Although the biochemical activity of Cdc1p is not known, the essential function involves the divalent cation Mn2+, as judged by the observations that Mn2+ supplement alleviates the cdc1(Ts) temperature-sensitive growth defect and that the chelator sensitivities of cdc1 mutants are ameliorated by overproduction of the high-affinity Mn2+ transporter Smf1p (Loukin and Kung 1995; Supeket al. 1996; Paidhungat and Garrett 1998b). Consistent with this conclusion, we show here that the temperature-sensitive cdc1(Ts) growth defect can be alleviated by genetic manipulations that increase intracellular Mn2+, including additional copies of SMF1 (Figure 3A), as well as mutations that stabilize Smf1p in a functionally active form (Figures 2A and 3; Tables 3 and 4). The genetic relationship between Cdc1p-dependent growth and intracellular Mn2+ levels is significant because previously characterized suppressors alleviated the cdc1(Ts) growth defect by mechanisms that are not obviously related to intracellular Mn2+ (Paidhungat and Garrett 1997, 1998a). The relationship between Mn2+ and Cdc1p function is also intriguing in light of the weak, but significant, sequence similarity between Cdc1p and a family of Mn2+-dependent phosphotransfersase proteins (Rusnak 2000).
Although the biochemical activity and biological function of Cdc1p have yet to be defined, the observation that a cos16/per1 deletion can alleviate the cdc1Δ growth defect without altering Mn2+ levels and do so independently of Smf1p suggests that Cos16p/Per1p function might lie downstream of Cdc1p. Previous fractionation studies localized a Cos16-βGal fusion to the vacuole (Paidhungat and Garrett 1998b), whereas a largescale localization study identified the ER as the resident site of GFP-tagged Cos16p/Per1p (Huhet al. 2003). Further work will be needed to reconcile the differences between these two studies; however, residence in the ER would be consistent with the observation that a per1 (cos16) disruption induces the unfolded protein response pathway of the ER (Nget al. 2000). Precisely how this would alleviate the physiological defects of a cdc1 mutant is not known. Nevertheless, it is intriguing to note that organisms such as Arabidopsis thaliana, Schizosaccharomyces pombe, and Neurospora crassa contain proteins that bear limited, but significant, sequence identity with Cos16p/Per1p and that Mn2+ has been implicated in the function of a Cdc1p homolog (frost) from N. crassa (Sone and Griffiths 1999).
Ubiquitination and Smf1p targeting: In yeast, Mn2+ homeostasis is maintained through regulated targeting of the high-affinity Mn2+ transporter Smf1p to the vacuole (Liu and Culotta 1999a,b). Strains lacking the major vacuolar protease Pep4p accumulate high levels of Smf1p in the vacuole in an inactive and Mn2+-insensitive form (Paidhungat and Garrett 1998b; Liu and Culotta 1999b). Thus, localization, rather than abundance per se, is critical to Smf1p function. Because vacuole delivery did not require a functional endocytic pathway, Liu and Culotta (1999b) proposed a model in which Smf1p was either secreted to the plasma membrane or, under Mn2+ excess, diverted to the vacuole for degradation. In support of this model, we show here that mutations that disrupt the biosynthetic route of the vacuolar protein sorting pathway stabilize Smf1p in a functionally active form, as judged by an increase in intracellular Mn2+ (Table 3), suppression of the cdc1(Ts) growth defect (Figures 1A and 3; Tables 2 and 4), and Smf1p accumulation (Figures 2 and 5). Because defects in vacuolar protein sorting also affect endocytosis, we cannot rule out the possibility that a defect in endocytosis contributes to Smf1p accumulation in the vps strains; however, our work (Table 4) and the work of others (Liu and Culotta 1999b) suggests that endocytosis does not play a major role in Smf1p localization. Of course, these results do not address the possibility that endocytosis plays a role in rapid elimination of Smf1p from the plasma membrane during transition from Mn2+-limiting to Mn2+-rich conditions.
Many proteins that are directed to the vacuole for degradation are ubiquitinated. Proteins such as the pheromone receptor Ste2p are ubiquitinated on the plasma membrane as a signal for entry into the early stage of the endocytic pathway (Hicke and Riezman 1996; Hicke 1999). Later in the endocytic pathway, ubiquitination serves as a molecular pass for recruitment into multivesicular bodies and subsequent deposition into the vacuole lumen (Katzmannet al. 2001). Ubiquitination also plays an essential role in the vacuolar delivery of the general amino acid permease Gap1p (Helliwellet al. 2001). When nitrogen is limiting, Gap1p is targeted to the plasma membrane where it serves as a nonspecific amino acid permease. During growth on nitrogen-rich medium, Gap1p is diverted to the vacuole along the vacuolar protein sorting pathway (Helliwellet al. 2001). As it does in endocytosis, ubiquitination directs biosynthetic membrane proteins such as Gap1p into multivesicular bodies and the vacuole lumen (Katzmannet al. 2001); however, recent observations are also consistent with ubiquitination serving as a molecular marker in the Golgi complex to divert Gap1p from the late secretory pathway to the biosynthetic route of the vacuolar protein sorting pathway (Helliwellet al. 2001). Although this latter role is not as well understood as the analogous role in endocytosis, it is clear that ubiquitination plays an important function in membrane protein localization and degradation. Thus, we are intrigued by the observation (Figure 3B; Table 4) that the cdc1-1(Ts) growth defect can be alleviated in a SMF1-dependent manner by alterations (doa4, ubc4, bro1, rsp5, and high-copy UBP1) that result in reduced ubiquitination of membrane proteins (Tobias and Varshavsky 1991; Galanet al. 1996; Ameriket al. 2000; Winkleret al. 2000). Suppression by these mutations does not reflect an indirect effect on Cdc1p function because Smf1p is ubiquitinated in vivo (Figure 4) and a ubiquitination defect results in accumulation of Smf1p in the ER and the cell surface (Figures 4 and 5).
Although ubiquitination is commonly described as a signal for degradation by the proteosome, previous results suggest that proteosomal degradation does not contribute significantly to Smf1p turnover or function (Liu and Culotta 1999b). Moreover, Smf1p trafficking and function are unaffected by several endocytosis defects (Liu and Culotta 1999a; Table 4). Thus, we propose that ubiquitination serves as an early determinant in Smf1p targeting, directing it to the VPS pathway rather than to the plasma membrane, and/or as a necessary signal for delivery to multivesicular bodies and the vacuole lumen.
If ubiquitination serves as an early determinant in Smf1p trafficking, factors involved in this Mn2+-regulated step should reside in or before the Golgi complex. Although the ER-resident protein Bsd2p seems an unlikely candidate for this role, it might mark Smf1p for vacuolar targeting in the Golgi. At odds with this notion, bsd2 mutations affect the vacuolar targeting of proteins that are unresponsive to Mn2+ (Luo and Chang 1997), suggesting that Bsd2p plays a more general role in protein trafficking. By contrast, the Atx2p protein resides in the Golgi apparatus and has been associated only with Mn2+ homeostasis (Lin and Culotta 1996; Table 4). Because ATX2 effects require Smf1p (Lin and Culotta 1996), Atx2p may play a direct role in Mn2+-regulated Smf1p trafficking and degradation. In this scenario, Atx2p would direct Smf1p to the plasma membrane in medium depleted of Mn2+, either by sequestering Smf1p before it was ubiquitinated or by overriding a ubiquitin signal that was present under all conditions.
Class E vps mutants define two phenotypic classes: Individual class E mutants display qualitative differences in processing of carboxypeptidase S (CPS), a selective cargo of the MVB (Babstet al. 2002). Mutants such as vps4Δ, vps2Δ, and vps24Δ, which accumulate ESCRT-III subcomplex Vps20p-Vps32p on the class E compartment membrane, exhibit a dramatic defect in CPS processing, whereas strains lacking individual components of the Vps20p-Vps32p subcomplex mature CPS with wild-type kinetics (Babstet al. 2002). Interestingly, class E mutants vps4Δ, vps2Δ, and vps24Δ, along with strains lacking components of the Vps27p/Hse1p complex, exhibit wild-type (Ca2+ or Li+) or moderate (Mn2+) resistance to monovalent and divalent cations, whereas mutants lacking components of ESCRT-I, ESCRT-II, and ESCRT-III subcomplex Vps20p-Vps32p are extremely sensitive to those treatments (Figure 1B; Table 2). We do not fully understand the molecular natures of these ion sensitivities, although they may represent documented defects in removing high-affinity ion transporters from the plasma membrane as well as defects in ion sequestration. Nevertheless, we are intrigued by the correlation between ion sensitivity and CPS processing.
We thank C. Michaels, M. Carlson, V. Culotta, and M. Rose for yeast strains and plasmids. M.P. thanks T. Graf and J. Johnson for help and instruction with atomic absorption spectroscopy, and Y.-S. Chung thanks V. Culotta for suggestions concerning Western analysis with the HA-tagged Smf1p protein.
Communicating editor: B. J. Andrews
- Received November 10, 2003.
- Accepted January 30, 2004.
- Copyright © 2004 by the Genetics Society of America