|Home | About | Journals | Submit | Contact Us | Français|
In the yeast Saccharomyces cerevisiae, glutathione plays a major role in heavy metal detoxification and protection of cells against oxidative stress. We show that Gex1 is a new glutathione exchanger. Gex1 and its paralogue Gex2 belong to the major facilitator superfamily of transporters and display similarities to the Aft1-regulon family of siderophore transporters. Gex1 was found mostly at the vacuolar membrane and, to a lesser extent, at the plasma membrane. Gex1 expression was induced under conditions of iron depletion and was principally dependent on the iron-responsive transcription factor Aft2. However, a gex1Δ gex2Δ strain displayed no defect in known siderophore uptake. The deletion mutant accumulated intracellular glutathione, and cells overproducing Gex1 had low intracellular glutathione contents, with glutathione excreted into the extracellular medium. Furthermore, the strain overproducing Gex1 induced acidification of the cytosol, confirming the involvement of Gex1 in proton transport as a probable glutathione/proton antiporter. Finally, the imbalance of pH and glutathione homeostasis in the gex1Δ gex2Δ and Gex1-overproducing strains led to modulations of the cAMP/protein kinase A and protein kinase C1 mitogen-activated protein kinase signaling pathways.
The metabolism of oxygen in cells leads to the generation of reactive oxygen species (ROS), such as hydrogen peroxide (H2O2), the superoxide anion (O2.−), and the hydroxyl free radical (OH˙). ROS are inevitable by-products of oxygen metabolism but, at high concentration, they can cause oxidative damage to the cell, including protein oxidation, lipid peroxidation, and chromatin breaks (for reviews see Scandalios, 2002 ; Pocsi et al., 2004 ; Wysocki and Tamás, 2010 ).
Thiol-containing molecules play an essential role in protecting cells against oxidative damage. They react with ROS, resulting in the neutralization of these potentially dangerous molecules. In most living organisms, the tripeptide glutathione (γ-l-glutamyl-l-cysteinyl-glycine: GSH in its reduced form, GSSG in its oxidized form) is the most abundant low–molecular weight, thiol-containing molecule. Glutathione is a strong cellular redox buffer due to its low redox potential (E′0 = −240 mV), its high redox ratio (GSH:GSSG) of 16:1, and its high cellular concentration (1–10 mM) (Grant et al., 1996 ; Penninckx, 2002 ). Yeast cells grown in normal aerobic conditions have a high redox ratio (90% of glutathione in the reduced form), which is maintained through strict regulation of glutathione biosynthesis—mediated by glutathione synthetase (Gsh1 and Gsh2)—and of glutathione reduction, mediated by glutathione reductase (Glr1) in an NADPH-dependent reaction (Herrero et al., 2008 ). GSH is also involved in the detoxification of xenobiotics and resistance to heavy metal stress. This detoxification capacity is due to the highly nucleophilic properties of the thiol moiety of GSH. The role of glutathione in cadmium detoxification is well documented. Cadmium has been shown to react with two molecules of GSH to form a stable GS-Cd-SG conjugate (Polec-Pawlak et al., 2007 ). In Saccharomyces cerevisiae, this complex is removed from the cytosol by ATP-dependent glutathione S–conjugate export pumps (GS-X pumps), such as Ycf1, a vacuolar membrane protein that imports Cd(GS)2 into the vacuolar lumen, and Yor1, a plasma membrane protein that exports Cd(GS)2 from cells (Szczypka et al., 1994 ; Li et al., 1996 ; Ghosh et al., 1999 ; Gueldry et al., 2003 ; Nagy et al., 2006 ; Paumi et al., 2009 ).
In mammalian cells, GSH conjugates are transported by ATP-binding cassette proteins of the multidrug resistance–associated protein (MRP) family and by proteins of the organic anion–transporting polypeptide (OATP) family (Homolya et al., 2003 ; Ballatori et al., 2005 ). Six of the 12 members of the MRP family (MRP1–5 and cystic fibrosis transmembrane conductance regulator) have been shown to mediate glutathione-conjugate extrusion at the plasma membrane. OATP transporters function without ATP and are less well characterized. OATP transporters were originally characterized as uptake transporters, but some are probably involved in glutathione efflux from the cell.
Given the importance of glutathione in redox ratio maintenance and detoxification, the genes involved in the biosynthesis, metabolism, and transport of glutathione are tightly controlled. Exposure to oxidative stress or to heavy metals leads to activation of the stress response regulators Yap1 and Yap2, with Yap1 activating the expression of YCF1 and GSH1 (Wemmie et al., 1994 ; Wu and Moye-Rowley, 1994 ; Cohen et al., 2002 ; Temple et al., 2005 ).
Oxidative stress may be generated by toxic heavy metals, such as cadmium, but also by essential metals, such as iron, partly due to their ability to generate toxic hydroxyl radicals through the Fenton reaction (Halliwell and Gutteridge, 1984 ). There is growing evidence that iron deficiency may also lead to oxidative stress. In rats, iron deficiency has been shown to induce oxidative damage to the mitochondria and mitochondrial DNA (Walter et al., 2002 ). Yeast, a strain lacking the iron-responsive transcriptional activators Aft1 and Aft2, both of which are required for the expression of genes involved in iron uptake, is hypersensitive to oxidative stress, and this sensitivity is decreased by the addition of iron to the growth medium (Blaiseau et al., 2001 ). Yap1 has also been shown to up-regulate the expression of AFT2 (Salin et al., 2008 ). Finally, the redox status of the cell influences iron homeostasis: GSH depletion induces a specific defect in the maturation of cytoplasmic [Fe–S] proteins (Sipos et al., 2002 ). Thus iron metabolism and thiol-dependent redox homeostasis are linked, and defects in one of these processes leads to an imbalance in the other.
Iron homeostasis and uptake are tightly regulated to prevent iron toxicity. Under aerobic conditions, iron is present mostly in its ferric form (Fe3+) and is not available for use by the cell. S. cerevisiae uses two different systems to take up environmental iron: the reductive uptake mechanism and the siderophore transport system. The reductive system involves the reduction of ferric iron at the plasma membrane by reductases, followed by the uptake of ferrous iron by a high-affinity permease (Philpott, 2006 ). The nonreductive system involves iron uptake mediated by siderophores, small organic compounds that chelate ferric iron with high affinity. In S. cerevisiae, siderophore transport is mediated by four plasma membrane transporters of the major facilitator superfamily (MFS) of transporters (Goffeau et al., 1997 ). Arn1 and Taf1/Arn2 are members of this family involved in the transport of ferrichromes (FCH) and triacetylfusarinine (TAF), respectively; Sit1/Arn3 transports ferrioxamine B (FOB) and ferrichrome (FCH); and Enb1/Arn4 transports enterobactin (ENB) (Lesuisse et al., 1998 , 2001 ; Heymann et al., 1999 , 2000a , 2000b ; Yun et al., 2000b ). These four siderophore transporters belong to the Aft1 regulon (ARN) family and are regulated by the iron-responsive transcription factor Aft1 (Yun et al., 2000a ).
Phylogenetic analysis of the ARN family has revealed the existence of two uncharacterized homologues, YCL073c and YKR106w (Haas et al., 2008 ), which we have named GEX1 (glutathione exchanger) and GEX2, respectively. In this study, we show that Gex1 and Gex2 are mostly found at the vacuolar membrane. The production of Gex1 and Gex2 is induced under conditions of iron depletion and after H2O2 treatment, and Gex1 is regulated principally by Aft2. However, Gex1 and Gex2 are not involved in the transport of known siderophores. Instead, they are proton antiporters involved in glutathione exchange. Finally, we show that the deletion of GEX1 and GEX2 and the overexpression of GEX1 modulate different signaling pathways (protein kinase A [PKA] and protein kinase C1 mitogen-activated protein kinase [PKC1-MAPK]), confirming a clear connection between iron, redox equilibrium, and the stress response.
S. cerevisiae expresses four siderophore transporters of the ARN family differing in substrate specificity (Haas et al., 2008 ). These proteins are secondary transporters of the MFS subfamily (Philpott, 2006 ; Blaiseau et al., 2010 ). Two other homologues of these ARN family genes, YCL073c and YKR106w, which we have named GEX1 (glutathione exchanger) and GEX2, respectively, were not classified as ARN transporters because their expression was shown to be independent of Aft1 (Yun et al., 2000a ).
The amino acid sequences of Gex1 and Gex2 are 46% identical to that of their closest homologue, Arn1, and 33% identical to that of their most distant homologue, Enb1/Arn4. The amino acid sequences of Gex1 and Gex2 are 98% identical. Gex1 and Gex2 are predicted to have 12 transmembrane domains rather than the 14 generally observed in members of the ARN family of transporters. This is a common feature of MFS transporters (Pao et al., 1998 ). No known protein motif was identified in sequence analyses with various prediction programs.
GEX1 and GEX2 are present in the subtelomeric regions of chromosomes III and XI, respectively. This subtelomeric duplication is probably a recent divergence because the two fragments encode almost identical products (Gromadka et al., 1996 ). Finally, the expression of GEX1 and GEX2 is undetectable in normal growth conditions (Gromadka et al., 1996 ), and it has been suggested that this lack of expression may be due to the telomeric position of these genes, as genes close to the telomeres are generally silenced (Gottschling et al., 1990 ). Given the high level of sequence identity between the coding sequences of GEX1 and GEX2, it is impossible to distinguish the expression of one gene from that of the other. We therefore constructed strains bearing GEX1 and GEX2 tagged, at the chromosomal locus of the gene and at the carboxy terminus of the encoded protein, with different epitopes. Using the GEX1-GFP/GEX2-HA and GEX1-HA/GEX2-GFP strains grown under normal growth conditions (yeast extract peptone dextrose [YPD] or yeast nitrogen base [YNB]), we were unable to detect the expression of either of these genes, consistent with the findings of Gromadka et al. (1996 ) (Figure 1A). As GEX1 and GEX2 are homologous to members of the ARN family, we hypothesized that they might be induced under conditions of iron deficiency. We therefore cultured the same cells in the presence of bathophenanthroline bisulfonic acid (BPS), an iron chelator. We detected the expression of GEX1 and GEX2 after at least 16 h of growth in the presence of BPS (Figure 1A). The expression of GEX2, whether tagged with green fluorescent protein (GFP) or hemagglutinin (HA), was very weak, only about one-fourth that of GEX1.
It has been shown that, unlike other members of the ARN family, GEX1 and GEX2 do not require Aft1 for their expression (Yun et al., 2000a ). We analyzed the promoter regions of GEX1 and GEX2 and found no Aft1 binding motif. However, we did identify an Aft2 binding motif, PuCACCC (Courel et al., 2005 ), at −668 (ACACCC) in the promoter regions of the two genes. A second Aft2 binding motif was present only in the promoter region of GEX1, at −283 (ACACCC). For analysis of the regulation of GEX1 and GEX2 by Aft2 and Aft1, we constructed centromeric plasmids containing wild-type AFT1 or AFT2 or the constitutively active mutant alleles AFT1–1up and AFT2–1up. The four genes were placed under the control of the same promoter regulated by doxycycline. The plasmids were then introduced into the strains bearing tagged GEX1 and GEX2 genes, and protein extracts were analyzed by SDS–PAGE (Figure 1B). Under these experimental conditions, we were unable to detect GEX2 expression. GEX1, whether tagged with HA or GFP, was induced more strongly by AFT2 and AFT2–1up than by AFT1and AFT1–1up. The expression of GEX1 in the presence of wild-type AFT1 and AFT2 expression showed that GEX1 was a genuine target of these transcription factors, with AFT2 more efficient than AFT1.
We analyzed the distribution of Gex1 and Gex2 in strains with the corresponding genes tagged at the chromosomal locus (at the carboxy terminus of the encoded protein) with GFP or HA. The cells were grown to midexponential growth phase in the presence of BPS, and GFP fluorescence was analyzed by fluorescence microscopy (Figure 2, A and B). In the GEX1-GFP strain, fluorescence was detected at both the plasma and vacuolar membranes (Figure 2A). The distribution of the protein depended on the growth phase (Figure 2B). Very early in the exponential growth phase (optical density at 600 nm [OD600] = 0.1) Gex1-GFP was present mostly at the plasma membrane, whereas in late exponential growth phase (OD600 = 3) this protein was found mostly at the vacuolar membrane. As shown by SDS–PAGE (Figure 1A), GEX2-GFP expression levels were much lower than those for GEX1-GFP, and the corresponding protein was barely detected. Like Gex1-GFP, Gex2-GFP was present at both the plasma and vacuolar membranes. Given the technical difficulty of detecting Gex2 production and the 98% identity of the two genes, we hypothesized that the two proteins were probably involved in the same process and decided to concentrate our studies on Gex1, except in genetic studies, in which we used a strain with a double deletion of GEX1 and GEX2.
We tried to confirm the dual location of Gex1-HA by immunofluorescence studies, but we were unable to detect the protein by this approach (unpublished data). We then analyzed the distribution of the protein by sucrose gradient fractionation and Western immunoblotting on cells grown as in Figure 2A (Figure 2C). Gex1-GFP, which appeared to be located at the plasma and vacuolar membranes on the basis of GFP fluorescence, was present in fractions containing the plasma membrane ATPase Pma1 and internal fractions containing Vph1, the membrane subunit of the vacuolar ATPase. Some free GFP was detected in the soluble fraction (1–3), probably due to the endocytosis and degradation of Gex1-GFP in the vacuolar lumen (see later discussion and Supplemental Figure 2). For Gex1-HA, the protein was mostly found at the vacuolar membrane, as demonstrated by the cofractionation of Gex1-HA with Vph1 (Figure 2C). We hypothesized that Gex1-HA might, like Gex1-GFP, have a dual location, with most of the protein present at the vacuolar membrane and a smaller fraction present at the plasma membrane. We tested this hypothesis by tagging GEX1 with HA at the chromosomal locus in the end3Δ mutant, which displays impaired internalization during endocytosis, ensuring that the plasma membrane was enriched in Gex1-HA. Gex1-HA, expressed in the end3Δ mutant, was detected in fractions colocalizing with the vacuolar membrane Vph1 but also in fractions colocalizing with the plasma membrane marker Pma1 (Figure 2C). Thus Gex1 presents a dual localization. The HA tag is smaller and is generally thought not to induce mislocalization. We therefore concluded that Gex1 was located mostly at the vacuolar membrane, with a smaller fraction of the protein present at the plasma membrane. The distribution of Gex1 depended both on the tag used and on expression levels. When Gex1 tagged with HA or GFP was overproduced under the control of a galactose promoter from a centromeric plasmid, Gex1-HA and Gex1-GFP were present at the plasma membrane, as shown by sucrose gradient fractionation and GFP fluorescence (Figure 2D).
Membrane-bound proteins are sorted in the trans-Golgi network (TGN) and are directed toward the plasma membrane or the vacuolar membrane. There are two alternative pathways for targeting membrane-bound proteins to the vacuole: the AP3 and VPS pathways. The AP3 pathway is a direct route, with vesicles originating from the Golgi apparatus directly fusing with the vacuole. The VPS pathway is an indirect pathway in which vesicles originating from the Golgi compartment fuse with late endosomes, which then fuse with the vacuole. We analyzed the targeting of Gex1-GFP to the vacuolar membrane with the aps3Δ mutant, which is affected in the AP3 pathway, and the pep12Δ mutant, affected in the fusion of Golgi-derived vesicles with late endosomes (Supplemental Figure 1). In the aps3Δ mutant, Gex1-GFP was detected at the vacuolar membrane, whereas in the pep12Δ mutant, Gex1-GFP was no longer targeted to the vacuolar membrane and the internal fraction of Gex1-GFP was trapped in vesicles. Thus Gex1-GFP is targeted to the vacuolar membrane via the VPS pathway.
Some Gex1-GFP was also present at the plasma membrane. Plasma membrane proteins that are no longer required undergo ubiquitylation, followed by endocytosis and targeting to the vacuole for degradation (Dupré et al., 2004 ). We analyzed endocytosis of the plasma membrane–localized fraction of Gex1-GFP in the GEX1-GFP and GEX1-GFP/pep4Δ strains. In the pep4Δ strain, which lacks some vacuolar protease activities, protein degradation in the vacuole is delayed. The cells were cultured in the presence of BPS to very early in the exponential growth phase (OD600 = 0.1). In these growth conditions, Gex1-GFP was found principally at the plasma membrane (Supplemental Figure 2A). In the GEX1-GFP/pep4Δ strain, staining of the vacuolar lumen was observed at time 0, probably corresponding to the targeting of a small fraction of Gex1-GFP for degradation. Cells were washed and incubated in complete medium (YPD) without BPS for 2, 4, and 6 h. Gex1-GFP present at the plasma membrane at time 0 was progressively targeted for vacuolar degradation (after BPS chase) (Supplemental Figure 2, A and B). The vacuolar degradation of Gex1-GFP was delayed in the GEX1-GFP/pep4Δ strain, as shown by Western immunoblotting and fluorescence microscopy (accumulation of fluorescence in the lumen of the vacuole). Thus Gex1-GFP located at the plasma membrane underwent endocytosis when it was no longer required and was targeted to the vacuolar lumen for degradation.
The similarity of the sequences of GEX1 and GEX2 to those of other genes of the ARN family of siderophore transporters suggested a possible role for Gex1 and Gex2 in siderophore transport. We therefore measured the rate of iron uptake from various siderophores available in our laboratory in wild-type and gex1Δ gex2Δ strains: FOB, FCH, ENB, and TAF (Supplemental Figure 3A). No difference in the rate of uptake was observed between gex1Δ gex2Δ and wild-type strains for any of the siderophores studied, as suggested in a review by Winkelmann (2001 ). The presence of Gex1 at two different locations made this study more difficult. Thus, although no defect in iron uptake from siderophores was observed in the gex1Δ gex2Δ strain, we speculated that Gex1 might be involved in the vacuolar storage of intracellular siderophores. We then analyzed the intracellular distribution of internalized FOB in wild-type and mutant cells, using the gallium (GaIIII) analogue of the fluorescent derivative of FOB, FOB-NBD (Froissard et al., 2007 ) (Supplemental Figure 3B). Accumulation of the fluorescent probe in the vacuole was similar in mutant and wild-type cells (Supplemental Figure 3B). These results suggest that Gex1 and Gex2 are probably not involved in siderophore transport.
Gex1 and Gex2 belong to the MFS subfamily and may be involved in transporting metals other than iron. We therefore assessed the ability of the gex1Δ gex2Δ strain and of the strain overproducing Gex1 to grow on media containing high concentrations of divalent metals (Cd2+, Ni2+, Cu2+, Mn2+, Mg2+, Co2+, Zn2+) and in the presence of the iron chelator BPS (Figure 3A and unpublished data). The gex1Δ gex2Δ mutant grew poorly only in the presence of the toxic heavy metal cadmium (Cd2+). The strain overproducing Gex1-HA was more resistant to this metal than the wild-type strain transformed with an empty plasmid. We then quantified the cadmium, iron, and copper contents of cells grown in a Cd-supplemented medium (Figure 3B). No significant differences were found in the iron and copper contents of the various strains, whereas a clear difference between strains was observed for cadmium. The gex1Δ gex2Δ strain had a much higher cadmium content than wild-type cells, and the cells overproducing Gex1-HA had a lower intracellular cadmium content. Gex1-HA was targeted to the plasma membrane when overproduced from the pGEX1-HA plasmid (Figure 2D). A decrease in cadmium content may therefore be indicative of an increase in cadmium efflux at the plasma membrane mediated by Gex1-HA. These results are consistent with the growth phenotypes observed in the presence of cadmium, and they suggest that Gex1 may mediate cadmium export at the plasma membrane. The topology of membrane-bound proteins is acquired in the endoplasmic reticulum and remains constant, regardless of the final site to which the protein is targeted. Thus a protein with multiple membrane spans and amino and carboxy termini facing the cytosol will present the same topology at the plasma membrane and at the vacuolar membrane (schemes in Figure 3B). Gex1 at the vacuolar membrane may therefore mediate cadmium import into the vacuolar lumen and participate in cadmium detoxification.
In S. cerevisiae, cadmium detoxification is mediated principally by ATP-dependent glutathione S–conjugate pumps, which transport complexes of cadmium with glutathione from the cytosol to the extracellular medium or into the vacuolar lumen. We first analyzed total (reduced plus oxidized) and oxidized glutathione contents in the presence and absence of cadmium for the various strains (Figure 4).
Wild-type (WT) and gex1Δ gex2Δ strains grown in the absence of cadmium had redox ratios (GSH:GSSG) of 13:1 and 14:1, respectively, indicating that most of the intracellular free glutathione was present in a reduced form (Figure 4A). However, the gex1Δ gex2Δ strain had a higher intracellular glutathione content. After incubation of the cells with 1 μM cadmium, the total glutathione content of the gex1Δ gex2Δ strain was not significantly different from that of the wild-type strain, but the redox balance of the gex1Δ gex2Δ cells was shifted to the oxidized form of glutathione (GSSG), with a GSH:GSSG ratio of 5.5:1. In the gex1Δ gex2Δ strain, some reduced glutathione may have been associated with cadmium. This reduced glutathione accumulated in the cytosol of the cells, giving a ratio in favor of GSSG. The overproduction of Gex1-HA (which is known to be localized at the plasma membrane; see Figure 2D) resulted in a significantly lower total glutathione content than for the wild-type strain bearing an empty plasmid. This decrease was accompanied by an increase in the amount of the oxidized form of glutathione, giving a redox ratio of 3:1 (19:1 for the wild-type strain). No difference in total glutathione content was found between the strains bearing the empty plasmid or pGEX1-HA after exposure to cadmium, but the redox ratio favored oxidized glutathione when Gex1-HA was overproduced (GSH:GSSG of 3.5:1). Intracellular glutathione depletion or enrichment may be attributed to several factors (decrease/increase in biosynthesis, excretion, etc.). Gex1 is an MFS transporter that may be involved in transporting glutathione alone or in a complex with cadmium. We thus analyzed the excretion of glutathione into the medium in the presence and absence of cadmium for the various strains (Figure 4B). The gex1Δ gex2Δ cells excreted less glutathione into the medium than wild-type cells. In the presence of cadmium, there was no significant difference between the wild-type and gex1Δ gex2Δ strains. The two strains also presented similar amounts of oxidized glutathione in the medium. Cells overproducing Gex1-HA excreted more glutathione into the medium than the corresponding wild-type cells. Rates of excretion were five times higher than those of the wild-type strain in the presence of cadmium. Most of the glutathione excreted into the medium when Gex1-HA was overproduced was in the reduced form (GSH). Thus glutathione is excreted in its reduced form, probably as a complex with cadmium. GSH-Cd complexes are kinetically labile and dissociate rapidly in the extracellular medium, which has a lower cadmium concentration than the cells.
Low glutathione ratio (GSH:GSSG) may indicate that cells are subject to oxidative stress. We assessed the sensitivity of the various strains to oxidative stress generated by treatment with H2O2 (Figure 5A). We found that gex1Δ gex2Δ cells were hypersensitive to H2O2 and cells overproducing Gex1-HA were resistant to H2O2. This may indicate a role for Gex1 in protecting cells against exposure to H2O2. We analyzed Gex1-GFP levels after exposure to H2O2 and found that 40% of cells displayed faint GFP fluorescence after treatment with 0.1 mM H2O2 (Figure 5, B and C). Western immunoblotting analysis of Gex1-GFP levels after H2O2 treatment showed that H2O2 induced Gex1-GFP synthesis less efficiently than BPS treatment (levels after H2O2 treatment were only one-seventh those after BPS exposure). Finally, Gex1-GFP was not induced following overproduction of the transcription factors required for oxidative stress tolerance, Yap1 and Yap2 (Figure 5, B and C). Thus GEX1 was involved in protecting cells against the oxidative stress induced by H2O2 treatment, and GEX1 expression was not dependent on the stress-responsive factors Yap1 and Yap2 but may have been dependent on the iron-responsive transcription factor Aft2.
Gex1 and Gex2 are members of the MFS family of transporters and are therefore predicted to be proton antiporters or symporters (Goffeau et al., 1997 ). We assessed the ability of Gex1 to mediate proton exchange by analyzing the cytosolic pH of the gex1Δ gex2Δ strain and of the strain overproducing Gex1-HA, using a GFP variant, the ratiometric pHluorin, which is sensitive to pH (Miesenböck et al., 1998 ; Dechant et al., 2010 ) (Figure 6A). As control strains, we used the nha1Δ and nhx1Δ mutants, which have a higher and lower pH, respectively, than the wild-type strain (Sychrova et al., 1999 ; Brett et al., 2005 ). The ratiometric pHluorin variant was found in the cytosol and nucleus but was excluded from the vacuole. The gex1Δ gex2Δ cells displayed lower levels of cytosolic fluorescence, indicative of a cytosolic pH higher than that of the wild-type strain but lower than that of the nha1Δ mutant. The cells overproducing Gex1-HA (which was localized at the plasma membrane) displayed a higher level of cytosolic fluorescence than the wild-type strain (but lower than that of the nhx1Δ mutant), indicating acidification of the cytosol and thus the import of H+ ions into the cytosol, probably mediated by Gex1-HA. This acidification of the cytosol was responsible for the extensive vacuole fragmentation observed when Gex1-HA was overproduced (Supplemental Figure 4).
We further confirmed the results obtained by the quantitative analysis of fluorescence. The ratio of fluorescence at 508 nm in response to excitation at two different wavelengths (I410/I470) was proportional to cytosolic pH (Miesenböck et al., 1998 ; Brett et al., 2005 ). The ratios obtained confirmed our fluorescence observations.
A change in cytosolic pH might also lead to a change in the pH of the vacuole. We therefore assessed vacuolar acidity, using the intracellular pH indicator quinacrine (Baggett et al., 2003 ) (Figure 6C). No detectable difference in vacuolar quinacrine staining was observed between the wild-type and gex1Δ gex2Δ mutant cells. By contrast, the overproduction of Gex1-HA resulted in more intense quinacrine staining, indicative of vacuole acidification. This acidification of the vacuole was probably an indirect effect of the cytosol acidification mediated by the Gex1-HA at the plasma membrane, as shown for other mutants, including nhx1Δ (Brett et al., 2005 ).
Hence these experiments indicate that Gex1 is a proton antiporter that makes use of the proton motive force at the plasma and vacuolar membranes to mediate glutathione exchange (scheme in Figure 6).
Dechant et al. (2010 ) recently demonstrated that cytosolic pH regulates the PKA glucose signaling pathway and a genome-wide screen for mutants overexcreting glutathione identified mutants of the PKA pathway as having the highest levels of glutathione excretion (22–25 times higher than that of wild-type cells; Perrone et al., 2005 ). We showed that Gex1-HA overproduction induced acidification of the cytosol and a decrease in glutathione content. We therefore investigated the possible role of Gex1 and Gex2 in the PKA pathway. For this purpose, we transformed the gex1Δ gex2Δ deletion mutant and the strain overproducing Gex1-HA with a plasmid encoding the stress-responsive transcription factor Msn2 tagged with GFP (Görner et al., 1998 ). Msn2 is targeted to the nucleus in response to various stresses, including oxidative stress and acidification of the cytosol, through the action of several signaling cascades, including the PKA pathway. The disruption of GEX1 and GEX2 induced a defect in the targeting of Msn2-GFP to the nucleus, and Gex1-HA overproduction induced a relocalization of Msn2-GFP to the nucleus, consistent with the induction by Gex1-HA of cytosol acidification (Figure 7A). Msn2 and Msn4 have been shown to increase the expression of genes involved in glycogen synthesis (Smith et al., 1998 ). Thus defects in Msn2 targeting to the nucleus lead to a decrease in glycogen accumulation. We assessed glycogen accumulation in the cells, using an iodine solution (Figure 7B). The disruption of GEX1 and GEX2 resulted in lower glycogen content than for the wild-type strain, confirming a defect in the targeting of Msn2 to the nucleus. The wild-type strain transformed with an empty plasmid behaved like the wild-type strain transformed with pGEX1-HA because glycogen is hardly synthesized when cells are grown on galactose as the sole carbon source.
We assessed Msn2 activation in cells overproducing Gex1-HA by searching for targets of Msn2 potentially involved in oxidative stress. The PKC1-MAPK pathway is essential for cell survival in conditions of oxidative stress (Vilella et al., 2005 ). Activation of the PKC1-MAPK pathway leads to activation of Slt2/Mpk1, the last member of this cascade. Mpk1 is activated by phosphorylation, and strict timing of the phosphorylation/dephosphorylation of Mpk1 is essential for cell survival. Two phosphatases are involved in the dephosphorylation of Mpk1: Sdp1 and Msg5. Sdp1 is specific for Mpk1, and its expression is induced by Msn2 in conditions of oxidative stress (reviewed in Martín et al., 2005 ). We therefore analyzed the phosphorylation profile of Slt2/Mpk1 in the gex1Δ gex2Δ mutant strain and in a wild-type strain (Figure 7C). The phosphorylation of Mpk1 was not affected in the gex1Δ gex2Δ strain, consistent with the cytosolic location of Msn2. We further analyzed the phosphorylation of Mpk1 in WT, MPK1-GFP, and sdp1Δ strains with and without Gex1-HA overproduction. The overproduction of Gex1-HA had no effect on Mpk1-GFP levels but induced the dephosphorylation of Mpk1, whether untagged or tagged with GFP (Figure 7C). In the strain lacking SDP1, the dephosphorylation of Mpk1 was abolished, and Mpk1 phosphorylation levels were even higher than those in a strain bearing an empty plasmid. We also carried out a more general analysis of the MAP kinase pathway, using calcoﬂuor white (CFW), a chitin antagonist that interferes with cell wall assembly, and the cell wall–stressing 1,3-glucan–binding dye Congo red (CR) (Figure 7D). We observed no difference in the growth rates of the wild-type and gex1Δ gex2Δ mutant strains, whereas the overproduction of Gex1-HA rendered wild-type cells sensitive when cultured on plates in the presence of CFW or CR, confirming inhibition of the cell wall integrity pathway response (Chen et al., 2005 ; Lesage et al., 2005 ). Our results thus indicate that Gex1 and, probably, Gex2 affect the PKA and PKC1-MAPK signaling pathways by disturbing the pH and thiol-redox homeostasis of the cells.
GEX1 and GEX2 are paralogues, presenting 98% sequence identity and mapping to the subtelomeric regions of two different chromosomes (Gromadka et al., 1996 ). The other S. cerevisiae homologues of GEX1 and GEX2, ARN1–4, also map to subtelomeric regions but are farther from the end of the chromosome than GEX1 or GEX2. It has been suggested that the subtelomeric region of the chromosome is a reservoir of silenced genes involved in adaptive responses (Fabre et al., 2005 ). We found that GEX1 and GEX2 were expressed only under certain growth conditions in which the cells were subject to stress. However, despite the lack of detection of Gex1 or Gex2 by Western immunoblotting or fluorescence, the gex1Δ gex2Δ double deletant was sensitive to cadmium and H2O2 and was affected in terms of its cytosolic pH and glutathione homeostasis, consistent with the presence of Gex1 and/or Gex2 but in amounts too small to be detected.
GEX1 and GEX2 are induced under conditions of iron depletion, as expected for genes homologous to those of the ARN family of siderophore transporters. Unlike other members of the ARN family, GEX1 is induced principally by the iron-responsive transcriptional factor Aft2 and, to a lesser extent, by its paralogue Aft1. Gex1 has two specific Aft2 binding motifs in its promoter region, at −283 and −668. Only one Aft2 binding site, the farthest from the ATG codon, is present in the promoter region of GEX2. This may account for the lack of GEX2 expression in conditions of Aft2 and Aft2–1up production. Another element that could account for the lack of GEX2 expression in conditions of iron depletion is the location of GEX2, closer to the telomere than GEX1, potentially enhancing the silencing of GEX2. Silencing close to the telomeres is probably the principal system of regulation for GEX1 and GEX2 because the reduction of nucleosome content by the histone H4 depletion increases the expression of GEX1 and GEX2 (Wyrick et al., 1999 ).
There are very few examples of genes regulated principally by Aft2. Such genes include those encoding the divalent metal ion vacuolar transporter Smf3, the mitochondrial iron transporter Mrs4, and the mitochondrial iron–sulfur cluster assembly Isu1 (Rutherford et al., 2001 ; Courel et al., 2005 ). However, all these studies have reported a role for Aft2 in gene transcription only in the absence of AFT1 or in the AFT2–1up gain-of-function mutant (Rutherford et al., 2001 ; Courel et al., 2005 ). GEX1 is the first example of a gene clearly regulated by Aft2 in a wild-type genetic context. Our results are also consistent with the hypothesis that Aft2 is involved principally in the regulation of proteins involved in vacuolar and mitochondrial transport (Courel et al., 2005 ).
We also found that GEX1 was induced under H2O2 treatment and that this induction was independent of the redox-responsive YAP regulon (YAP1 and YAP2). Previous studies have shown that the aft1Δ aft2Δ double mutant presents a number of phenotypes related to oxidative stress, including H2O2 hypersensitivity. It has been suggested that AFT1 and AFT2 have overlapping functions in iron homeostasis and oxidative stress resistance, and AFT2 has been shown to be directly regulated by YAP1 (Blaiseau et al., 2001 ; Salin et al., 2008 ). Gex1 was induced after treatment with the iron chelator BPS, and iron depletion is known to induce oxidative stress. Thus GEX1 expression is probably due to oxidative stress and not directly linked to iron transport. Our study suggests that Aft2 regulates Gex1 induction under oxidative stress conditions in a Yap1-independent manner.
The location of Gex1 at two sites—the plasma membrane and the vacuolar membrane—is very unusual but not unique. A few other examples of yeast proteins located at several sites in the steady state are known. One such example is the chloride channel Gef1, a member of the voltage-activated Cl− channel group (Greene et al., 1993 ). Gef1 has been shown to function at the Golgi apparatus, vacuolar membrane, and plasma membrane (Gaxiola et al., 1998 ; Schwappach et al., 1998 ; Lopez-Rodriguez et al., 2007 ). The presence of Gef1 at multiple sites is important for the maintenance of normal Cl− homeostasis in the cell. In mammalian cells, the lysosomal transporter LYAAT-1/PAT1, an H+/amino acid symporter, is located at the lysosomal/vacuolar membrane in neurons and is involved in the efflux of amino acids from the lysosome (Sagne et al., 2001 ). By contrast, in gut cells, this transporter is located at the plasma membrane, where it functions in the absorption of amino acids (Boll et al., 2004 ).
The mechanisms involved in the dual targeting of Gex1 to the plasma and vacuolar membranes remain to be elucidated. Transmembrane proteins following the secretory pathway are sorted in the TGN, within which they may be targeted to the plasma membrane or to the multivesicular bodies (MVBs) and hence to the vacuole. Two main signals are involved in targeting proteins to the MVBs: peptide motifs present in the cytosolic domains of the proteins are involved in exit from the TGN, and the ubiquitylation of cytosolic lysine residues is responsible for the targeting to the MVBs of membrane proteins originating from the plasma membrane or the TGN (Bonifacino and Traub, 2003 ; Lauwers et al., 2010 ).
We found that Gex1 was down-regulated in the absence of BPS in the culture medium. Ubiquitin is the only signal known to induce the internalization of plasma membrane transporters in yeast, and Rsp5 has been shown to be the only ubiquitin ligase involved in the internalization of all plasma membrane transporters in yeast. Thus Gex1 from the plasma membrane is probably ubiquitylated in an Rsp5-dependent manner before being targeted to the MVBs and the vacuolar lumen. Gex1 targeted from the TGN to the vacuole via the VPS pathway was probably not ubiquitylated because it was localized at the membrane and was not present within the lumen of the vacuole. It remains unclear how this is achieved. Are there different signals involved in the targeting of Gex1 to the vacuole depending on whether it comes from the TGN or from the plasma membrane? Are there different ubiquitylation–deubiquitylation events mediating this regulation? Does the trafficking of Gex1 depend on the concentration of its substrates? It has been shown for the uracil permease Fur4 and for the divalent metal ion transporter Smf1 that the availability and/or concentration of the substrate may cause a change in the conformation of the protein on its way to the plasma membrane, resulting in direct targeting to the vacuole rather than to the plasma membrane (Liu and Culotta, 1999 ; Blondel et al., 2004 ). A similar hypothesis could be proposed for Gex1. Gex1 cotransports glutathione and H+. We have shown that BPS, and thus probably oxidative stress, induces the production of Gex1 and Gex2 only after 16 h of growth. Yeast is known to acidify the culture medium, and, after 16 h of growth, the two substrates of Gex1 are present: glutathione within the cells and H+ outside the cells. Early in the exponential growth phase, Gex1 is present mostly at the plasma membrane and mediates cytosol acidification and the extrusion of glutathione. The low pH of the cytosol induces acidification of the yeast vacuole, probably inducing the targeting of Gex1 from the TGN to the vacuolar membrane, where it can import glutathione and export H+ out of the vacuolar lumen. Thus the production and location of Gex1 probably depend on the presence of its two substrates. The molecular mechanisms involved in the routing of Gex1 remain to be deciphered.
The results of this study demonstrate that Gex1 is involved in the exchange of glutathione, a strong antioxidant and a powerful detoxifying tripeptide. We also show that Gex1 functions as a proton/glutathione antiporter. At the vacuolar membrane, H+ is exported into the cytosol and glutathione is imported through the vacuolar membrane. At the plasma membrane, H+ is imported into the cytosol and glutathione is exported to the extracellular medium. This export facilitates cadmium detoxification when cells are grown in the presence of this metal. However, glutathione is transported even in the absence of cadmium, indicating that the cadmium–glutathione complex is probably not the primary substrate of Gex1 and that Gex1 could probably transport glutathione coupled to another divalent cations.
Gex1 and Gex2 are homologous to the members of the siderophore transporter family ARN1–4. Surprisingly, the deletion of the GEX1 and GEX2 genes had no effect on iron uptake from different siderophores. Ferrichrome, the substrate of Arn1, is a cyclical hexapeptide (Philpott et al., 2002 ). Glutathione is a tripeptide, but two molecules of glutathione are required for the interaction with cadmium (Belcastro et al., 2009 ). The structures of glutathione and ferrichrome are similar, and transporters presenting sequence similarities could transport molecules with similar structures. Furthermore, sequence similarity does not inevitably lead to the transport of the same molecules. A good example of this is provided by the multidrug resistance and pleiotropic drug resistance families of transporters, which have similar sequences but transport a broad range of molecules of different categories (for review see Jungwirth and Kuchler, 2006 ; Panwar et al., 2008 ).
GEX1 is induced under stress conditions, such as iron depletion or H2O2 exposure. We have shown that the disruption of GEX1 and GEX2 or the overproduction or Gex1 leads to changes in the cytosolic pH and glutathione content of cells. These two phenomena are tightly regulated by the cells and involve different signaling pathways and different transcription factors. Msn2 is a stress response transcriptional activator involved in regulating almost 200 genes (involved in response to heat shock, osmotic shock, oxidative stress, low pH, glucose starvation, etc.) and is a target of the PKA and PKC1-MAPK signaling pathways. It was thus a perfect candidate for involvement in the response to pH and redox imbalance observed when Gex1 is overproduced or when GEX1 and GEX2 are disrupted. We showed that Msn2 was present mostly in the cytosol when GEX1 and GEX2 were disrupted and was targeted to the nucleus when Gex1 was overproduced. The PKA pathway was affected in the gex1Δ gex2Δ strain, and this defect was probably due to alkalinization of the cytosol. Down-regulation of the PKC1-MAPK pathway was impaired in the strain overproducing Gex1 due to the action of the Sdp1 phosphatase on Mpk1/Slt2. The down-regulation of the PKC1-MAPK pathway after Gex1-HA overproduction is probably not a direct consequence of the production of the protein itself, instead resulting from changes in the redox status of the cells (GSH:GSSG ratio of 3:1). We have shown that Gex1-HA is induced by H2O2. Thus overproducing a protein induced by oxidative stress mimics a physiological state in which the cells respond to this stress. Which receptors are involved in the sensing of this redox status? Are these sensors the same as those involved in the sensing of reactive oxygen species? These two questions need to be addressed.
In summary, we have shown that Gex1 is a proton antiporter directly participating in cell survival under oxidative stress conditions through the mediation of detoxification and/or glutathione homeostasis. We have also shown that Gex1 production is regulated by the iron-responsive factor Aft2 and that pH and thiol-redox changes induced by Gex1 modulate the PKA and PKC1-MAPK pathways. We believe that the cytosolic changes induced by Gex1 also influence its intracellular distribution, although this remains to be demonstrated definitively.
The S. cerevisiae strains used in this study are listed in Table 1. All are derivatives of BY4741/BY4742.
Deletion strains and strains carrying GFP- or 6HA-tagged genes were constructed by one-step gene replacement with PCR-generated cassettes (Longtine et al., 1998 ; Janke et al., 2004 ). The gex1Δ gex2Δ strain was obtained by the disruption of GEX1 and GEX2 in haploid strains (BY4741 and BY4742), which were then crossed and allowed to sporulate for tetrad analysis.
The pGEX1-GFP and pGEX1-HA plasmids were obtained by inserting a fragment encoding GEX1 in frame with the GFP or 6HA coding sequence into the centromeric p416-GAL vector (pØ) (Mumberg et al., 1994 ). These fragments were obtained by PCR, with genomic DNA from the chromosomally tagged strains used as the template.
The centromeric pCM188 plasmid was used to clone the wild-type AFT1, AFT2 and the mutant AFT1–1up, AFT2–1up alleles under the control of the doxycycline-regulated promoter (Garí et al., 1997 ). The pAFT1 and pAFT1–1up plasmids were constructed by inserting PCR-amplified fragments containing either the wild-type AFT1 or the Cys291Phe-mutated AFT1–1up coding sequence into pCM188. These fragments were obtained by PCR amplification of genomic DNA from the CM3260 and M2 strains (Yamaguchi-Iwai et al., 1995 ). The pEG202-AFT2 plasmid (Blaiseau et al., 2001 ) was used as a template to generate the AFT2–1up allele (Cys187Phe mutated) with the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA). Plasmids pAFT2 and pAFT2–1up were constructed by inserting the AFT2 and AFT2–1up coding sequences into pCM188. The other plasmids used in this study were YEp351-YAP1 (pYAP1), pAW18 (pYAP2) (Lesuisse and Labbe, 1995 ), pADH1-MSN2-GFP (Görner et al., 1998 ), and pADH1-pHluorin (Dechant et al., 2010 ).
Yeast cells were grown at 30°C in rich medium (YP): 1% yeast extract, 2% peptone, supplemented with 2% glucose (YPD) or 2% galactose (YPG). Cells transformed with plasmids were grown in minimal medium (YNB) containing 0.67% yeast nitrogen base without amino acids (Difco; BD, Franklin Lakes, NJ) and supplemented with the appropriate nutrients. The carbon source was 2% glucose, raffinose, or galactose. Unless stated, cells bearing pØ, pGEX1-HA, or pGEX1-GFP plasmids were grown in the presence of galactose as a carbon source to induce the galactose promoter. For BPS (Sigma-Aldrich, St. Louis, MO) treatments, cells were incubated with 200 μM BPS overnight.
Total protein extracts were prepared by the NaOH/trichloroacetic acid lysis technique (Volland et al., 1994 ). Proteins were separated by SDS–PAGE in 10% tricine gels or commercial precast gels (NuPAGE 4–12% Bis–Tris gels with morpholino propane sulfonic acid buffer; Invitrogen, Carlsbad, CA) and transferred onto nitrocellulose membranes. The antibodies used were monoclonal anti-GFP (clones 7.1 and 13.1; Roche Diagnostics, Indianapolis, IN) and anti-HA (clone F-7; Santa Cruz Biotechnology, Santa Cruz, CA), polyclonal anti-Vph1 (10D7, Invitrogen), anti-PGK (NE130/7S, Nordic Immunology, Tilburg, Netherlands), anti–phospho-p44/42 MAPK (4370; Cell Signaling Technology, Beverly, MA), and anti-Pma1 (a gift from Caroline Slayman) antibodies. Primary antibodies were detected with secondary antibodies conjugated to horseradish peroxidase (Sigma-Aldrich). Immunoblotting images were acquired with the LAS-4000 imaging system (Fujifilm, Tokyo, Japan), and band intensities were quantified with ImageJ software.
Sucrose density gradients were prepared as previously described (Erpapazoglou et al., 2008 ).
Fluorescence observations were performed with an Olympus BY61 microscope (Olympus, Tokyo, Japan) equipped with a fluorescein isothiocyanate (FITC) filter set (Filter 41001 FITC, Exciter HQ480/40x, Dichroic Q505LP, and Emitter HQ535/50 nm; Chroma Technology, Bellows Falls, VT) for GFP, NBD-Ga-DFOB, pHluorin, and quinacrine and a 4’,6-diamidino-2-phenylindole (DAPI) filter set (Filter 31013v2 DAPI, Exciter D365/40x, Dichroic 400DCLP, and Emitter D460/50 nm; Chroma Technology) for 7-amino-4-chloromethylcoumarin (CMAC). Cell shape was visualized with Nomarski optics. Images were acquired with a SPOT4.05 charge-coupled device camera. Images were processed with ImageJ, Adobe Photoshop CS3, and Adobe Illustrator CS3.
For cytosolic pH measurement, cells were transformed with pADH1-pHluorin encoding the ratiometric pHluorin (Miesenböck et al., 1998 ; Dechant et al., 2010 ). Cells were grown to midexponential growth phase in YNB, and fluorescence was measured with a Photon Technology International (Birmingham, NJ) monochromator and Felix 32 software. Fluorescence emission was read at 508 nm, and the ratio of the emission obtained at two excitation wavelengths (410 nm [I410] and 470 nm [I470]) was calculated. The same cells were observed by fluorescence microscopy with the FITC filter set.
The vacuole lumen was visualized by incubating the cells with 5 μM Cell Tracker Blue CMAC (Molecular Probes, Eugene, OR) for 15 min at 30°C and washing the cells with minimal medium. Acidification of the vacuole was assayed with the pH-sensitive fluorescent probe quinacrine [N′-(6-chloro-2-methoxy-acridin-9-yl)-N,N-diethyl-pentane-1,4-diamine] (Sigma-Aldrich), as previously described (Baggett et al., 2003 ). For iodine staining, cells were grown on plates for 3 d and then gently covered with a solution of 0.2% iodine/0.4% potassium iodide. Photographs were taken 3 min later. The presence of glycogen resulted in a darkening of the yeast cells.
Iron, copper, and cadmium contents were quantified for all strains by the inductively coupled plasma mass spectrometry method as carried out at the Service Central d'Analyse du CNRS (Solaize, France).
Glutathione levels were determined as previously described (Auchère et al., 2008 ).
Yeast strains were grown in minimum medium as previously described (Perrone et al., 2005 ), and cells were harvested at the appropriate growth phase by centrifugation. Extracellular glutathione concentration was measured directly in the resulting supernatant. For the estimation of total intracellular glutathione content, cell pellets were washed and resuspended in ice-cold 5% 5-sulfosalicylic acid in 50 mM potassium phosphate buffer, pH 7.8, and disrupted with glass beads. For the estimation of total intracellular glutathione concentration, cell pellets were washed and resuspended in 50 mM potassium phosphate buffer, pH 7.8, containing 5% ice-cold 5-sulfosalicylic acid and disrupted with glass beads. The resulting mixture was clarified by centrifugation (30 min, 5000 × g, 4°C), and the supernatant was used to determine total free glutathione (reduced and oxidized) concentration. A typical reaction mixture contained the cell extract, 20 mM DTNB, and 10 mM NADPH in 50 mM potassium phosphate buffer, pH 7.8. The reaction was started by adding glutathione reductase (1.5 U/ml), and the kinetics of DTNB conversion to TNB was followed spectrophotometrically at 412 nm. Glutathione concentrations were calculated from standard curves obtained with various GSSG concentrations based on rates of TNB formation and are expressed as nmol of glutathione/mg of protein. GSH:GSSG ratios were calculated using the equation GSH/GSSG = 2(total glutathione)/GSSG.
All data points in the figures and the values listed are means of at least six determinations, and Student's t test was used to identify significant differences.
We thank Michel Jacquet (Institut de Génétique et Microbiologie, CNRS/UPS UMR 8621, Paris, France) and Myriam Ruault (Institut Curie, Paris) for helpful discussions. We also thank R. Dechant (Institute of Biochemistry, ETH Zurich, Zurich, Switzerland) for the pADH1-pHluorin construct and M. Jacquet for pADH1-MSN2-GFP. We thank Zoi Erpapazoglou (Institut Jacques-Monod, Paris) for stimulating discussions. We thank members of the laboratory for critical reading of the manuscript and Alex Edelman & Associates for editorial assistance.
This work was supported by the Centre National de la Recherche Scientifique and the Université Paris-Diderot. Experiments performed in laboratory of R.H.-T. were supported by grant 3298 from the Association pour la Recherche sur le Cancer and by a grant from the European Network VI Framework (Role of Ubiquitin and Ubiquitin-like Modifiers in Cellular Regulation, contract no. 018683). Manel Dhaoui received a fellowship from the Tunisian Ministry of Research and the Region Ile-de-France.
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E10-11-0906) on April 13, 2011.