The Yeast Iron Regulon Is Induced upon Cobalt Stress and Crucial for Cobalt Tolerance*

To identify yeast genes involved in cobalt detoxification, we performed RNA expression profiling experiments and followed changes in gene activity upon cobalt stress on a genome-wide scale. We found that cobalt stress specifically results in an immediate and dramatic induction of genes involved in iron uptake. This response is dependent on the Aft1 protein, a transcriptional factor known to regulate a set of genes involved in iron uptake and homeostasis (iron regulon). Like iron starvation, cobalt stress induces accumulation of the Aft1 protein in the nucleus to activate transcription of its target genes. Cells lacking the AFT1 gene (aft1) are hypersensitive to cobalt as well as to other transition metals, whereas expression of the dominant AFT1–1 up allele, which results in up-regulation of AFT1-controlled genes, confers resistance. Cobalt resistance correlates with an increase in intracellular iron in AFT1–1 up cells, and sensitivity of aft1 cells is associated with a lack of iron accumulation. Furthermore, elevated iron levels in the growth medium suppress the cobalt sensitivity of the aft1 mutant cells, even though they increase cellular cobalt. Results presented indicate that yeast cells acquire cobalt tolerance by activating the Aft1p-dependent iron regulon and thereby increasing intracellular iron levels.

To identify yeast genes involved in cobalt detoxification, we performed RNA expression profiling experiments and followed changes in gene activity upon cobalt stress on a genome-wide scale. We found that cobalt stress specifically results in an immediate and dramatic induction of genes involved in iron uptake. This response is dependent on the Aft1 protein, a transcriptional factor known to regulate a set of genes involved in iron uptake and homeostasis (iron regulon). Like iron starvation, cobalt stress induces accumulation of the Aft1 protein in the nucleus to activate transcription of its target genes. Cells lacking the AFT1 gene (aft1) are hypersensitive to cobalt as well as to other transition metals, whereas expression of the dominant AFT1-1 up allele, which results in up-regulation of AFT1-controlled genes, confers resistance. Cobalt resistance correlates with an increase in intracellular iron in AFT1-1 up cells, and sensitivity of aft1 cells is associated with a lack of iron accumulation. Furthermore, elevated iron levels in the growth medium suppress the cobalt sensitivity of the aft1 mutant cells, even though they increase cellular cobalt. Results presented indicate that yeast cells acquire cobalt tolerance by activating the Aft1p-dependent iron regulon and thereby increasing intracellular iron levels.
Although cobalt is an essential micronutrient as a cofactor of vitamin B 12 (reviewed in Ref. 1) and various other enzymes in animals, yeasts, bacteria, Archaea, and plants (2), exposure to inorganic cobalt is associated with various human diseases such as contact dermitis (3), allergic asthma leading to subsequent interstitial fibrosis (4), and lung cancer (5). Exposure of humans to cobalt is widespread; cobalt compounds are used in many industrial processes such as refining and production of alloys, jet engines and gas turbines, electrochemical materials, and permanent magnets. Furthermore, cobalt is used in drying agents for lacquers and in varnishes, paints, inks, catalysts, ceramics, pigments, and surgical implants, and mineral supplements for pasture lands often carry cobalt salts (6 -8). Cobalt increases oxidative stress in cells by raising the concentration of reactive oxygen species (9 -11) and mimics or replaces ions such as magnesium and calcium in various essential reactions (12).
In molecular biology, cobalt, like iron chelators, is frequently used as a tool to mimic hypoxia. Cobalt stimulates, as does hypoxia, the production of erythropoietin, a glycoprotein hormone essential for the differentiation of red blood cells in response to hypoxia. There is evidence that the cellular oxygensensing mechanism, present in most if not all tissues, utilizes a heme protein where cobalt might substitute for the iron in the porphyrin ring, thereby decreasing its affinity to oxygen and mimicking a hypoxic environment (13,14). The notion that iron and cobalt compete for this heme protein is supported by the fact that cells grown in low-iron medium require less cobalt for erythropoietin stimulation than those exposed to iron-rich medium (15).
To understand the cellular response to cobalt stress and screen for genes involved in cobalt detoxification, we determined the transcriptional profile of Saccharomyces cerevisiae upon cobalt exposure. We found that the iron regulon, in particular, genes dependent on the transcriptional factor Aft1p, plays a vital role in cellular cobalt tolerance. The 77.7-kDa protein Aft1p has been shown to be transported into the nucleus upon iron starvation to mediate iron-regulated transcription by binding to the consensus motif PyPuCACCCPu in the 5Ј-upstream region of a number of genes involved in iron transport and homeostasis (16 -18). Whereas aft1 mutant cells are unable to switch on the iron regulon upon cobalt stress and appear sensitive to cobalt, permanent induction of these genes (via the AFT1-1 up allele) leads to cobalt resistance. Cellular iron levels coincide with cobalt tolerance; iron is decreased in aft1 disrupted cells and increased in AFT1-1 up cells, and in addition, the wild-type shows elevated iron concentrations upon cobalt stress. Furthermore, high external iron can restore cobalt tolerance of the aft1 disrupted cells.
Plasmid Constructions-To overexpress C-terminally HA 1 -tagged Cot1p, the COT1 gene was PCR-amplified from FY1679 chromosomal DNA using the mutagenic oligonucleotide primers COT1-PstI/XbaI (5Ј-CCCCCTGCAGTCTAGACTCAGCACTTTCTACATT-3Ј) and COT1-PstI (5Ј-CCCCCTGCAGGATGATCCTCTAAGCAATC-3Ј), starting at positions Ϫ425 and ϩ2008, respectively, relative to the COT1 coding sequence, and introducing PstI and XbaI sites (underlined). The product was digested with PstI and XbaI and then ligated to the PstI and XbaI linearized vector YEp351HA to obtain the 7.5-kb vector YEpCOT1HA. To clone pVTUCOT1-1HA, the HA-tagged COT1 was amplified from YEpCOT1HA using the primers COT1-X (5Ј-* This work was supported by Austrian Science Fund Project P13460-MOB. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ‡ To whom correspondence should be addressed. Tel.: 43-1-4277-54604; Fax: 43-1-4277-9546; E-mail: schweyen@gem.univie.ac.at. CTAGTCTAGAATAGTTCTGCATAGC-3Ј) and COT1-H (5Ј-CAGTGC-CAAGCTTTCAGC-3Ј) located at Ϫ61 and ϩ1432, respectively, relative to the COT1 open reading frame and featuring a XbaI and a HindIII site, respectively (underlined). The PCR product was digested with XbaI and HindIII and cloned into pVT102-U, placing the COT1 gene behind the constitutive p ADH1 promoter.
Immunofluorescence-Cells were fixed and stained with antibodies as described previously (24). Antibodies used were mouse anti-HA (19) and goat anti-mouse antibody conjugated to rhodamine (SC-2092; Santa Cruz Biotechnology). Nuclear and mitochondrial DNA was visualized by staining with 4Ј,6-diamidino-2-phenylindole (Molecular Probes). Fluorescence microscopy was done with a Zeiss Axioplan 2 microscope and a Visitron Systems Imaging system.
RNA Isolation for Microarray Hybridization-Cells grown in YPD medium at 28°C to A 600 ϭ 0.5-0.6 were treated with 100 M to 2 mM CoCl 2 or with 80 M BPS (Sigma) for 30 or 90 min at 28°C, and then the cells were harvested, and RNA was prepared essentially according to the hot acidic phenol method (25); the only difference was that three chloroform extractions were performed instead of one.
cDNA Probe Synthesis and Microarray Hybridization-The yeast DNA chips were obtained from the Ontario Cancer Institute Microarray Centre. Reverse transcription, probe cleanup, and microarray hybridization were performed according to the manufacturer's protocol.
Microarray Reading and Data Analysis-Microarrays were read using an Axon GenePix 4000B laser scanner (Axon Instruments) and analyzed using GenePix Pro 3.0 software. The Saccharomyces Genome Data Base, Yeast Proteome Data Base, and Munich Information Center for Protein Sequences data base were used to extract information on Co 2ϩ -responsive genes.
Northern Analysis-Cells grown in YPD medium at 28°C to A 600 ϭ 0.5-0.6 and then incubated with 2 mM CoCl 2 , 1.5 mM MnCl 2 , 5 mM NiSO 4 , or 10 mM ZnCl 2 for 60 min at 28°C were harvested, and RNA was prepared (25). Northern blotting and hybridization (25 g of RNA each) were performed (26) and analyzed using an Amersham Biosciences Typhoon 8600 phosphorimaging system. Probes were generated by PCR from chromosomal DNA (JS034-4C) using [␥-32 P]ATP and Determination of Cellular Iron and Cobalt-Cells grown in YPD medium at 28°C to A 600 ϭ 0.5-0.6 were incubated with CoCl 2 or FeCl 2 for 90 min. Cells were harvested, washed twice with high pressure liquid chromatography-grade water (atomic absorption spectrophotometry) or 50 mM Tris-HCl, pH 7.4, 10 mM EDTA (inductively coupled plasma-mass spectrophotometry), and then dried at 105°C for 120 min. Upon determination of the dry weight, the pellet was digested in 65% HNO 3 at 90°C for 30 min. Ions were quantified using a PerkinElmer Life Sciences 5100PC Atomic Absorption Spectrophotometer (Fe) or PerkinElmer Life Sciences Sciex ELAN 6100 inductively coupled plasma-mass spectrophotometry (Fe, Co).

RESULTS
Expressional Response to Co 2ϩ Stress-To follow the genomic response to Co 2ϩ , early logarithmic yeast wild-type cells (JS034-4C) grown in liquid YPD medium were either stressed with 100 M CoCl 2 for 30 min or stressed with 2 mM CoCl 2 for 30 and 90 min. Whereas 100 M Co 2ϩ represents mild, subtoxic Co 2ϩ stress that does not affect the growth of yeast wild-type cells, 2 mM Co 2ϩ causes a 50% reduction of growth rates (data not shown). Expression profiling experiments were repeated three times using independently grown cultures. Genes whose expression was significantly up-or down-regulated at least 2-fold are shown in Table I. The complete set of data is available at www.at.embnet.org/gem/mrs/cobalt/.
Subtoxic Co 2ϩ concentrations of 100 M were found to exclusively up-regulate a set of genes known to be involved in iron transport and iron homeostasis (Table I). With the exception of YLR205, all the genes of this group are known to be regulated by the transcription factor Aft1p (22,(27)(28)(29)(30), which is mediating iron-dependent transcription of the so-called "iron regulon" (17,22). ARN1, ARN2, and ARN4 are coding for siderophore transporters of ferriochromes, triacetylfusarinine C, and ferric enterobactin, respectively (31)(32)(33). FET3 and FTR1 constitute the high-affinity iron uptake system (34,35). The FIT1, FIT2, and FIT3 gene products are cell wall glycosylphosphatidylinositol-anchored glycoproteins facilitating iron transport (28), and ISU2 codes for a mitochondrial protein involved in ironsulfur cluster assembly (29). The protein sequence deduced from the YLR205 gene shows similarity to the mammalian heme oxygenases 1 and 2.
Six additional genes up-regulated by 2 mM Co 2ϩ but not by 100 M Co 2ϩ contain consensus Aft1p binding sites (16) in their 5Ј-non-translated region (highlighted in Table I). COT1 is involved in cellular Co 2ϩ accumulation and confers Co 2ϩ resistance when overexpressed (36). PCL5 is a G1 cyclin gene (37), whereas TIS11 codes for a putative transcription factor. YJL079, YDL124, and YPL250 are open reading frames of unknown function. YPL250 contains in its 5Ј-non-translated region two zinc responsive elements in addition to three AFT1 recognition motifs (38). None of the genes known to be AFT1dependent or possessing AFT1 recognition motifs in their promoter regions and found here to be transcriptionally regulated upon Co 2ϩ stress in the wild-type were up-regulated upon Co 2ϩ stress applied to JS018 aft1-null mutant cells (Table I).
High Co 2ϩ (2 mM), but not the subtoxic concentration of 100 M, was found to stimulate expression of a series of other genes ( Table I). Most of them are involved in stress response, amino acid synthesis pathways (particularly arginine and methionine), and vacuolar degradation.
Genes Induced upon Iron Limitation-To induce mild iron depletion, the cell-impermeable chelator BPS was added to wild-type cells (JS034-4C) to a final concentration of 80 M 90 min before harvest. The expressional response of yeast upon iron depletion was strikingly similar to the immediate response observed upon mild Co 2ϩ stress (30 min, 100 M CoCl 2 ). Genes significantly induced by BPS were the siderophore transporters ARN1 and ARN3 and the glycosylphosphatidylinositol-anchored glycoproteins FIT2 and FIT3 (Table I). ARN2, FET3, and FIT1 showed slight (ϮS.D. Ͼ 1.5) induction (data not shown).
Total Cellular Iron in Co 2ϩ -stressed Cells-To determine whether the cellular iron content is altered in response to Co 2ϩ stress, cells were grown in YPD medium to mid-log phase and stressed with 2 mM CoCl 2 for 90 min. Total cellular iron contents of the CoCl 2 -treated cells and untreated control cells were determined by atomic absorption spectrophotometry (Fig. 1). In wild-type cells (JS034-4C), iron was increased 2-fold upon Co 2ϩ stress, consistent with Co 2ϩ -induced up-regulation of the iron transport genes as observed in the microarray experiments. aft1 mutant cells (JS018) exhibited 2.3 times lower total cellular iron than wild-type cells and failed to accumulate iron when stressed with Co 2ϩ . This is consistent with earlier findings that the aft1-null mutant was found to exhibit a lower iron uptake rate than wild-type cells (39). Cells expressing the dominant AFT1-1 up allele (JS034-4C pT14), which promotes iron-independent permanent transcription of AFT1-regulated genes, contained 3.8 times more iron than wild-type cells. No significant further increase in cellular iron was observed when AFT1-1 up cells were stressed with Co 2ϩ . Taken together, Co 2ϩ stress apparently induces overaccumulation of iron via an Aft1p-controlled process.
The Subcellular Localization of Aft1p Is Dependent on Cobalt-Because the iron-dependent regulation of the Aft1p tran- , grown in YPD medium or exposed to 2 mM CoCl 2 for 90 min, were subjected to immunofluorescence microscopy. Correlating with the activation of Aft1p target genes upon Co 2ϩ stress (Table I), Aft1p was localized in the cytoplasm at standard conditions (Fig. 2, Ϫ Co) and directed to the nucleus by Co 2ϩ (Fig. 2, ϩ Co). Aft1p Is Critical for Metal Ion Tolerance-To investigate the role of the AFT1 gene product in metal ion tolerance, various strains were compared for their ability to grow on elevated ion levels. As shown in Fig. 3, the aft1-null mutant was hypersensitive to Co 2ϩ , whereas the dominant AFT1-1 up allele conferred moderate resistance to elevated Co 2ϩ levels. Whereas aft1 mutant cells were also sensitive to iron depletion (BPS), increased levels of Zn 2ϩ , Ni 2ϩ , Mn 2ϩ , and Ca 2ϩ , and elevated temperature (37°C), the AFT1-1 up allele conferred slight resistance to iron limitation and increased Zn 2ϩ and Ni 2ϩ concentrations in the media. Because expression of the Cot1 protein, which is known to be important for cellular Co 2ϩ tolerance (see Refs. 36 and 40), appeared to be Aft1p-dependent (the COT1 gene has one Aft1p consensus binding motif in the 5Ј upstream region), we overexpressed Cot1p in the aft1 strain by the constitutive p ADH1 promoter from the multicopy vector pVTU-COT1HA. Whereas COT1 overexpression conferred Co 2ϩ tolerance to wild-type cells (data not shown), it was not sufficient to suppress sensitivity of the aft1 disrupted cells to Co 2ϩ or any other stress condition tested (Fig. 3).
The Iron Regulon Is Induced by Co 2ϩ but not by Mn 2ϩ , Ni 2ϩ , or Zn 2ϩ -Because aft1 mutant cells appeared sensitive not only to Co 2ϩ but also to other metal ions, we examined whether Mn 2ϩ , Ni 2ϩ , or Zn 2ϩ is also capable of up-regulating the iron regulon. Northern analysis was performed to follow expression of the Aft1p target genes FET3 and FIT1 (28,34) upon exposure to these metal ions. FET3 expression was highly induced by Co 2ϩ (Fig. 4), but not by Mn 2ϩ or Ni 2ϩ , whereas Zn 2ϩ also slightly induced FET3. FIT1 appeared to be up-regulated by Co 2ϩ only and not by Mn 2ϩ , Ni 2ϩ , or Zn 2ϩ . Consequently, FIG. 1. Total cellular iron content of wild-type, aft1, and AFT1-1 up cells upon cobalt stress. JS034-4C (wild-type), JS018 (aft1::TRP1), and JS034-4C pT14 (AFT1-1 up ) cells were grown in YPD medium to A 600 ϭ 0.5, and then CoCl 2 was added to a final concentration of 2 mM where indicated. Incubation at 28°C was prolonged for 90 min, and then the cells were harvested, and total cellular iron was determined (g/g dry weight). Data (means Ϯ S.E.) are representative of at least three independent experiments.

FIG. 2. Cellular localization of the Aft1 protein upon cobalt stress.
Wild-type (JS034-4C) cells expressing HA-tagged Aft1p from the centromere vector pRS416-AFT1-HA were grown to late log phase and then stressed with 2 mM CoCl 2 for 90 min where indicated (ϩ Co). After fixation, cells were subjected to immunofluorescence microscopy. Aft1p was detected using anti-HA antibodies (left panels), and nuclear DNA was stained with 4Ј,6-diamidino-2-phenylindole (right panels). consistent with microarray studies (data not shown), Co 2ϩ , but not Mn 2ϩ , Ni 2ϩ , or Zn 2ϩ , specifically induces the iron regulon.
High External Iron Overrides aft1 Co 2ϩ Sensitivity-To find out whether Co 2ϩ hypersensitivity of the aft1 mutant cells may be a consequence of intracellular iron shortage and of their inability to increase cellular iron upon exposure to Co 2ϩ , we assayed changes in growth as well as cellular iron and cobalt concentrations of aft1 cells (JS018) when cells were stressed with 90 M CoCl 2 in the presence or absence of 2 mM FeCl 2 . At 90 M CoCl 2 , growth of aft1 cells was severely reduced (Fig. 5,  छ), and cellular cobalt increased dramatically (Table II). 2 mM FeCl 2 in the growth medium highly increased cellular iron (Table II) but did not affect growth (Fig. 5, OE). When 90 M CoCl 2 was provided in addition to 2 mM FeCl 2 , the aft1 mutant was no longer inhibited by Co 2ϩ (Fig. 5, ࡗ). Surprisingly, cellular cobalt was even 3-fold increased as compared with mere addition of 90 M CoCl 2 (Table II). This indicates (i) that high iron levels in the medium alone can compensate for the loss of induction of iron transporters (aft1) in cobalt tolerance and (ii) that high iron does not override cobalt toxicity by competition in uptake, but rather intracellularly. DISCUSSION This study commenced at the surprising observation that in the yeast S. cerevisiae, cobalt stress selectively induces a number of genes known as the iron regulon and coding mostly for iron transport proteins, a response strikingly similar to observations upon iron starvation. As a consequence of this response, cellular iron is increased. Classical stress response genes and others also appear to be up-regulated by cobalt, although only at much higher, growth-inhibiting cobalt concentrations, whereas the iron regulon is induced even by subtoxic cobalt doses. Thus, enhancing iron transporters represents the primary transcriptomal response to cobalt stress in yeast.
Aft1p, the common transcription factor of the iron regulon genes (see , plays a vital role in this cobalt response pathway. Upon cobalt stress, Aft1p shifts from the cytoplasm to the nucleus, just as in the case of iron starvation (18). The central role of Aft1p in the cellular response to cobalt has been further elucidated by studying an aft1 disrupted strain and the AFT1-1 up mutant, in which the iron regulon is constitutively up-regulated (18,22). The aft1 mutant fails to increase its iron level in response to cobalt and exhibits hypersensitivity to this ion. Vice versa, AFT1-1 up cells have cellular iron increased constitutively and display moderate resistance to cobalt. The dependence of cobalt tolerance on cellular iron has been further documented by the observation that the cobalt hypersensitivity of aft1 cells can be compensated by the addition of high iron to the growth media, thereby raising cellular iron concentrations. The effect of increased iron is not to lower intracellular cobalt concentrations. Thus, we conclude that in yeast, increased cellular iron can compensate for high intracellular cobalt concentrations, resulting in the tolerance of this ion up to certain concentrations. Conversely, a decrease in cellular iron, e.g. in aft1 cells, results in cobalt sensitivity. However, there may also be Aft1p target genes directly involved in cobalt detoxification, acting independently of cellular iron accumulation.
The mechanism by which cobalt exerts its toxic effects in eukaryotic cells is not well understood. Several observations indicate that cobalt might compete with iron for specific binding sites in certain proteins, thereby impairing their functions (13,14). An increase in cellular iron levels, as observed here in cobalt-treated cells, might then suffice to supersede cobalt from such sites. Likewise, there is evidence that in higher eukaryotes, cobalt might compete with iron at a heme protein FIG. 3. Cellular metal ion tolerance/dependence with regard to Aft1p. Serial 10-fold dilutions of JS034-4C (wild-type), JS018 (aft1::TRP1), JS018 pVTU-COT1HA, and JS034-4C pT14 (AFT1-1 up ) cells were spotted onto YPD medium supplemented with 1 or 2 mM CoCl 2 , 300 M or 1 mM BPS, 10 mM ZnCl 2 , 5 mM NiSO 4 , 1.5 mM MnCl 2 , 100 mM CaCl 2 , or 100 mM spermidine as indicated and then incubated at 28°C (YPD also at 37°C) for 2-4 days.
FIG. 4. The iron regulon is induced by cobalt but not manganese, nickel, or zinc. JS034-4C (wild-type) cells were grown in YPD medium to A 600 ϭ 0.5, and then CoCl 2 (2 mM), MnCl 2 (1.5 mM), NiSO 4 (5 mM), or ZnCl 2 (10 mM) was added. Incubation at 28°C was prolonged for 60 min, and then the cells were harvested, RNA was isolated, and 25 g of each were used for Northern analysis.

TABLE II
Total cellular cobalt and iron contents of aft1 cells upon addition of cobalt and iron JS018 (aft1::TRP1) cells were grown in YPD medium to A 600 ϭ 0.5, and then CoCl 2 and FeCl 2 (90 M and 2 mM final concentrations, respectively) were added when indicated. Incubation at 28°C was prolonged for 90 min, and then the cells were harvested, and total cellular iron and cobalt were determined (ppm). Data (mean Ϯ S.E.) are representative of at least three independent experiments. involved in cellular oxygen sensing, thereby decreasing its affinity to oxygen and mimicking a hypoxic environment (14). In a similar way, cobalt might displace iron in yeast iron-sensing factors, thereby modifying signaling and inducing expression of iron uptake genes. Considering the data presented in this study, particularly the similarity of the Aft1p-dependent transcriptional response to iron deprivation and cobalt stress, and the ability of cobalt to promote nuclear localization of Aft1p as iron shortage, we speculate that cobalt can interfere with ironsensing of Aft1p or an additional factor interfering with Aft1p activity.
Although only cobalt appears to have a significant effect on the transcription of the iron regulon genes, aft1 cells also show increased sensitivity to other metal ions (zinc, nickel, manganese, and calcium), and the AFT1-1 up mutation confers resistance to nickel and zinc. This is reminiscent of the finding that cells with defects in the high affinity iron uptake system (fet3) have reduced cellular iron and exhibit hypersensitivity to cobalt, zinc, manganese, and copper, which can be compensated for by high iron in the growth medium (40). Hence, toxicity of these transition metals is increased when the high affinity iron transporter is defective (fet3) or fails to be up-regulated (in the case that Aft1p is missing), consistent with the accumulation of transition metals by transporters of broad specificity (40). However, it remains to be determined whether the sensitivity of aft1 mutant cells to transition metals directly correlates with the intracellular iron concentration, or whether it is caused secondarily by low expression of one of the many Aft1p target genes as a consequence of the absence of Aft1p. In any case, cobalt appears to be unique among the divalent metal ions tested in that it acts through Aft1p or a factor associated with this transcription factor and results in a response similar to that to iron shortage, e.g. elicited by iron chelators.
Taken together, this study came across novel aspects of the so-called iron regulon.
(i) Yeast cells appear to counteract the presence of high levels of cobalt by up-regulating their iron transport system and increasing their intracellular iron content. This response is mediated by the iron-sensing mechanism and highly specific to cobalt treatment. Therefore, it is likely to reflect an evolutionarily selected mechanism, which might not be restricted to the yeast S. cerevisiae, to deal with potentially deleterious cobalt concentrations.
(ii) Counteracting cobalt toxicity by increasing cellular iron is particularly surprising because metal ion tolerance is mostly discussed with regard to exporting, sequestering, or chelating the toxic agent. Increased cellular iron may not be harmful, as long as sufficient iron sequestration pathways are available, but rather displaces cobalt, entering the cell through the low affinity iron uptake system (41,42) and the magnesium transporter Alr1p (43,44) from iron binding sites and thereby minimizing its toxic effects.
(iii) Interestingly, subtoxic doses of cobalt (100 M, applied for 30 min) elicit significant effects only on the iron regulon, and not on stress genes or others. This illustrates the high potential of expression profiling to detect subtle effects far below a recognizable toxic reaction of yeast cells. Furthermore, it indicates that cobalt can interfere with iron sensing in S. cerevisiae, reminiscent of interactions between iron and cobalt in hypoxia signaling pathways in higher eukaryotes (13,14).
It remains to be determined, especially with regard to diseases associated with exposure to inorganic cobalt, to what extent cobalt and iron homeostasis are linked by means of physiological and molecular levels in yeast as well as in higher eukaryotes.