Saccharomyces cerevisiae Glutaredoxin 5-deficient Cells Subjected to Continuous Oxidizing Conditions Are Affected in the Expression of Specific Sets of Genes*

The Saccharomyces cerevisiae GRX5 gene codes for a mitochondrial glutaredoxin involved in the synthesis of iron/sulfur clusters. Its absence prevents respiratory growth and causes the accumulation of iron inside cells and constitutive oxidation of proteins. Null Δgrx5 mutants were used as an example of continuously oxidized cells, as opposed to situations in which oxidative stress is instantaneously caused by addition of external oxidants. Whole transcriptome analysis was carried out in the mutant cells. The set of genes whose expression was affected by the absence of Grx5 does not significantly overlap with the set of genes affected in respiratory petite mutants. Many Aft1-dependent genes involved in iron utilization that are up-regulated in a frataxin mutant were also up-regulated in the absence of Grx5. BIO5 is another Aft1-dependent gene induced both upon iron deprivation and in Δgrx5 cells; this links iron and biotin metabolism. Other genes are specifically affected under the oxidative conditions generated by the grx5 mutation. One of these is MLP1, which codes for a homologue of the Slt2 kinase. Cells lacking MLP1 and GRX5 are hypersensitive to oxidative stress caused by external agents and exhibit increased protein oxidation in relation to single mutants. This in turn points to a role for Mlp1 in protection against oxidative stress. The genes of the Hap4 regulon, which are involved in respiratory metabolism, are down-regulated in Δgrx5 cells. This effect is suppressed by HAP4 overexpression. Inhibition of respiratory metabolism during continuous moderately oxidative conditions could be a protective response by the cell.

Cells growing aerobically are subjected to oxidative stress caused by reactive oxygen species (ROS) 1 produced at the respiratory chain in the presence of oxygen (1). Cells have devel-oped a number of enzymatic and non-enzymatic mechanisms to counteract the damage that ROS cause to the different cellular macromolecules. In the yeast Saccharomyces cerevisiae, addition of external oxidants such as hydrogen peroxide or menadione (a generator of superoxide anion) cause a rapid and generally transient transcriptional response (2,3). This consists of the induction of genes involved in hydrogen peroxide and superoxide detoxification, as well as those related to redox homeostasis within the cell. In the case of hydrogen peroxide, observations at the transcriptome level confirm studies carried out at the proteome level (4,5). Two transcriptional regulators that respond to oxidative stress, Yap1 and Skn7, are important in such a response to hydrogen peroxide (5,6). The Yap1 regulon involves two types of proteins as follows: those required for antioxidant scavenging, and enzymes that participate in the metabolic pathways that regenerate cell reducing power in the form of glutathione and NADPH. The first subset of proteins is also Skn7-dependent, whereas the second is not (5). Msn2 and Msn4 are two transcription factors that respond to different stresses, including oxidative stress, by regulating gene expression through binding to STRE promoter elements (7,8). The defense function of Msn2/4 partially overlaps with that of Yap1. However, Yap1 seems to be more important for the adaptive response induced by moderate concentrations of hydrogen peroxide, whereas Msn2/4 would be preferentially implicated in recovery from acute exposure to the oxidant (9). This is consistent with the specific role that Msn2/4 has in regulating expression of genes involved in protein degradation pathways (9).
Most studies into the results of oxidative stress on yeast cells have involved adding external oxidants to cell cultures. However, this does not necessarily reproduce the physiological effect caused by aerobic respiration or other types of oxidation at moderate but constant levels. Proteins of S. cerevisiae cells growing under aerobic conditions are more exposed to continuous oxidation than proteins from anaerobic cultures (10). The latter demonstrated continuous oxidative stress under aerobic conditions. However, the gene expression pattern of cells that shift from fermentative to respiratory metabolism at the diauxic transition (11,12) is very different from that of cells subjected to external aggression by oxidants such as hydrogen peroxide, menadione, or diamide (2,3). Thioredoxins and glutaredoxins are two thiol oxidoreductases that play an important role in protecting sulfhydryl groups in proteins against oxidation and therefore in maintaining the protein activity under these conditions (13). Grx5 is a monothiol glutaredoxin of S. cerevisiae (a single cysteine at the active site) whose absence leads to high sensitivity to oxidative agents and to constitutive carbonylation of proteins (a parameter that meas-ures protein oxidation (10)) even in the absence of external oxidative injury (14). Accumulation of glutathionated proteins has also been observed in the absence of Grx5 (15,16). This monothiol glutaredoxin is located at the mitochondria, where it is involved in the formation of iron-sulfur clusters (17). As occurs with other proteins of the iron-sulfur assembly complex, lack of Grx5 causes iron accumulation within the cell. This could be related to the observed constitutive protein oxidation due to the formation of high levels of ROS via the Fenton reaction (17). A ⌬grx5 mutant is therefore an appropriate model for studying the effect of continuous oxidative stress on gene expression. In this work, we carried out transcriptome analysis of the Grx5-defective mutant. The results show that under these conditions the expression program was significantly different from the transcription pattern described in response to external oxidants, the inhibition of respiratory genes being a relevant response.
Cells were generally grown in rich YPD medium (1% yeast extract, 2% peptone, 2% glucose) at 30°C. In some experiments 3% glycerol was used instead of glucose (YPGly medium). Synthetic complete medium (SC) contains 0.67% yeast nitrogen base (Difco), 2% glucose, and the amino acids and nitrogen base additions indicated in Ref. 21. In some experiments, SC medium was prepared by adding the individual components at the standard concentrations, with the exception of biotin, which was included at the indicated final concentrations.
Array Hybridizations and Data Analyses-Experiments representing a particular situation were independently carried out in triplicate. Total RNA from W303-1A wild type strain and the ⌬grx5 and ⌬pet117 mutants was used to obtain labeled cDNA. About 30 -40 g of RNA were retrotranscribed into cDNA by adding 200 units of RT polymerase SuperScript II (Invitrogen), 500 ng of oligo(dT) primer (5Ј-T 15 VN-3Ј), 1 l of RNaseOUT (Invitrogen), 6 l of 5ϫ First Strand Buffer (Invitrogen), 1.5 l of dNTP mix (16 mM dATP, dTTP, dGTP, and 100 M dCTP), and 5 l of [ 33 P]dCTP (10 mCi/ml) in a final reaction volume of 30 l. Labeling reaction was allowed for 1 h at 43°C. One microliter of EDTA 0.5 M was added to stop the reaction. The labeled sample was purified using an S300-HR MicroSpin column (Amersham Biosciences).
Macroarrays were made in nylon membranes containing PCR products representing full-length open reading frames for 6,049 genes of S. cerevisiae. They were prepared at the Servicio de Chips de DNA of the University of Valencia. After pre-hybridizing nylon membranes for 1 h in 5ϫ SSC, 5ϫ Denhardt's, 0.5% SDS, hybridizations were performed using the same solution containing labeled cDNA (3.5 ϫ 10 6 dpm/ml) for 16 -18 h. After hybridization, the filters were washed once in 2ϫ SSC, 0.1% SDS for 30 min, and twice in 0.2ϫ SSC, 0.1% SDS for 30 min. They were exposed to a imaging plate (BAS-MP, Fujifilm) for 24 h and read in a PhosphorImager (FLA-3000, Fujifilm) at 50 m resolution.
The images were quantified by using ArrayVision 7.0 software (Imaging Research, Inc.). Each mutant replicate was normalized against the corresponding wild type hybridized in the same membrane (to eliminate the membrane variability factor) following the Lowess method. Reproducibility of the replicates was tested by the ArrayStat software (Imaging Research, Inc.). The data were considered as inde-pendent, and the program was allowed to take a minimum number of two valid replicates in order to calculate mean values for every gene. Only one of the three replicates was allowed to be removed from further calculations. A Z test for independent data was applied in order to detect differences in individual gene expression between each mutant and the wild type. Then a Z score was obtained for every gene. A p value of 0.05 and the False Discovery Rate method were used to monitor the overall false positive error rate. Genes that changed at least two times in the mutant with respect to the wild type strain were considered for further analyses and discussion.
Data Base Analyses-Data on the expression of individual genes from global genome analyses were obtained from the Yeast Microarray Global Viewer (yMGV, transcriptome.ens.fr/ymgv). Analysis of genes whose expression is modulated throughout the cell cycle (22) was carried out with the Saccharomyces Genome Data base (genome-www. stanford.edu).
Northern Blot Studies-RNA electrophoresis, probe labeling with digoxigenin, hybridization, and signal detection was carried out as described previously (20). Signals were quantified using the Lumi-Imager equipment (Roche Applied Science) software. Gene probes were generated from genomic DNA by PCR, using oligonucleotides designed to amplify internal open reading frame regions.
Sensitivity Tests-Exponential cultures in YPD medium at 30°C (about 2 ϫ 10 7 cells per ml) were serially diluted (5-fold dilution factor), and 2 l drops of each dilution were spotted onto YPD plates containing inhibitory agents at various concentrations. Growth was tested after 3 days of incubation at 30°C. In parallel, growth was also recorded in YPD plates without an inhibitor after incubation in the same conditions.
Protein Carbonylation Analyses-Protein carbonyl groups in cell extracts were derivatized with 2,4-dinitrophenylhydrazine. This was followed by SDS-PAGE and the immunodetection of peptides using an anti-dinitrophenylhydrazone antibody, according to Ref. 10.

RESULTS
Wild type and isogenic ⌬grx5 cells were grown exponentially in rich YPD medium at 30°C. Transcriptome analyses were carried out in three independent experiments in these growth conditions for each of the strains. Table I lists the genes that showed expression levels in ⌬grx5 cells that were at least twice as high as in wild type cells. Table II shows the genes whose expression levels in ⌬grx5 cells were at least 50% lower than the expression levels in wild type cells. Cells lacking GRX5 exhibit a Ϫ petite phenotype, that is they do not grow under respiratory conditions (17). It has also been shown that S. cerevisiae mitochondrial DNA °mutants have an altered expression profile (23). As expression pattern for ⌬grx5 cells could have been influenced by their petite defects, we carried out a parallel transcriptome analysis of ⌬pet117 cells with the same W303-1A genetic background. The total number of genes either up or down-regulated in the ⌬grx5 mutant was considerably lower than in the ⌬pet117 cells. This is clearly seen in Fig. 1A, where points are more widely distributed along the x axis (⌬pet117 relative profile) than along the y axis (⌬grx5 profile). Both Tables I and II indicate the genes affected in ⌬grx5 cells that were up-or down-regulated, respectively, in ⌬pet117 with respect to wild type cells. The proportion of these genes was very low in both cases, and we can therefore conclude that the transcriptome profile of Grx5-deficient cells was not significantly influenced by the petite character of the mutant. As revealed by fluorescence-activated cell sorter analysis, 2 the ⌬grx5 mutant accumulated a large proportion of cells at the G 1 stage of the cell cycle. This fact could have influenced the expression profile of those genes whose expression changed throughout the cell cycle (22). However, the proportion of these cell cycle-dependent genes whose expression is up-or down-  (Tables I and II), and overall, this did not significantly affect the analyses.
Overexpressed Genes in ⌬grx5 Cells-A total of 99 genes showed at least 2-fold induction of expression in cells lacking Grx5 as opposed to wild type cells (Table I). For some of these genes, the results were confirmed by Northern analysis ( Fig.  2A). A group of genes implicated in ion homeostasis, especially in iron and copper transport, was induced in ⌬grx5 cells. Most of these genes were controlled by the transcriptional activator Aft1, which responds to iron deprivation by inducing the expression of genes coding for plasma membrane metalloreductases (FRE1-3), the multicopper ferroxidase (FET3), the iron permease (FTR1), copper ion transporters (ATX1, CCC2), components of the siderophore-iron uptake system (ARN1-4), cell wall-associated facilitators of iron uptake (FIT1-3), and heme oxygenase (HMX1) (24 -29). Most of these genes are also upregulated in a yeast ⌬yfh1 mutant, which lacks the yeast homologue of human frataxin (27). Absence of frataxin causes mitochondrial accumulation of iron in a form that is not available to the cells (30), and this probably signals the activation of Aft1 (27). We had shown previously that disruption of ironsulfur cluster assembly through Grx5 inactivation also caused iron accumulation (17), and the results presented in Table I support the idea that this iron pool is also in a form that is not available to the cell.
Two genes associated with biotin transport were up-regulated in Grx5-deficient cells, VHT1 and BIO5. VHT1 has been characterized as coding for a plasma membrane high affinity biotin transporter, and its expression is repressed in media with high biotin levels (31). It has been reported recently (29) that this gene is an Aft1 target. This suggests that its upregulation in ⌬grx5 cells could also be a consequence of iron deficiency in these cells. More intriguing is the up-regulation of the BIO5 gene (Table I and Fig. 2A). S. cerevisiae cells require an external source of biotin as they lack the first enzyme of the pathway that converts pimelic acid into biotin (32). However, they are able to grow on the biotin vitamers 7-keto 8-aminoperlargonic acid (KAPA), 7,8-diaminopelargonic acid, and dethiobiotin, which are sequential intermediates in the pathway that leads to biotin (32). The last enzyme in the pathway (biotin synthase) converts dethiobiotin into biotin and is the product of BIO2 in S. cerevisiae. This enzyme contains iron/sulfur centers (33,34), and its activity therefore would be very low in ⌬grx5 cells. We have observed that ⌬grx5 cell growth was no more dependent on biotin concentration in the SC medium than growth of wild type cells under the same conditions (data not shown). This seems to dissociate the absence of an active Bio2 protein from biotin requirement and is in accordance with the fact that in laboratory growth conditions biotin precursors are not able to substitute for biotin. The BIO5 product has been characterized as the transporter of KAPA, which is a vitamer that is probably not present in either natural or laboratory yeast cell growth media (32). We examined whether BIO5 expression was also dependent on biotin levels in the growth medium, as in the case of VHT1. However, this was not the case (Fig. 2B). On the other hand, BIO5 expression was up-regulated upon iron chelation by ferrozine, similarly to the well known Aft1-dependent gene FET3 (Fig. 2C). This up-regulation   did not occur in a null aft1 mutant (Fig. 2C), which is consistent with the recent demonstration that BIO5 is overexpressed in an aft1 up strain (35). Therefore, BIO5 could be a member of the Aft1 regulon, although the fact that the BIO5 promoter does not contain consensus Aft1-binding sites (35) could indicate that the relationship between BIO5 expression and Aft1 is indirect. Nevertheless, these results confirm the previous suggestion (29) that biotin synthesis and transport are related to iron metabolism in yeast cells. SNZ1 transcription was increased in ⌬grx5 cells (Table I and Fig. 2A). This gene is involved in vitamin B 6 (pyridoxine) biosynthesis and has orthologues in various prokaryotes and eukaryotes (36). Its up-regulation in the absence of Grx5 (from almost undetectable levels in wild type cells) could be related to the proposed effect of pyridoxine as a quencher of singlet oxygen (37). This hypothesis is supported by the fact that yeast snz mutants are hypersensitive to the singlet oxygen generator methylene blue (38).
YKL161C (MLP1) is one of the genes that showed highest relative induction fold in ⌬grx5 cells (Table I). The Mlp1 protein is homologous to the Slt2 (Mpk1) MAP kinase, although it lacks the conserved active site of MAP kinases (39). Overexpression of MLP1 suppresses the caffeine sensitivity phenotype of a bck1 mutant in the signal transduction pathway leading to Slt2, whereas a null mlp1 mutant has additive effects on caffeine sensitivity when combined with a slt2 mutation (39). However, a single null mlp1 mutant does not show apparent phenotype. These observations suggest that Mlp1 acts in a pathway parallel to the Slt2 pathway in the regulation of downstream targets involved in cell integrity such as Rlm1 (40). On the basis that the grx5 mutant overexpresses MLP1, we reasoned that the absence of the latter could enhance defects in cells lacking GRX5. The double ⌬grx5 ⌬mlp1 mutant (strain MML533) was no more sensitive to caffeine, to the cell wall inhibitor calcofluor white, or to heat shock (growth at 38°C) than either the single ⌬mlp1 mutant or the wild type cells (not shown). The latter are phenotypic defects characteristic of cells where the Slt2 MAP kinase pathway is affected. However, the double mutant was more sensitive than the single ⌬grx5 mutant to t-BOOH and menadione (Fig. 3A). The MLP1 function could therefore play a protective role against oxidative stress in ⌬grx5 cells. Most important, hypersensitivity to external oxidants was not observed in the single ⌬mlp1 mutant (Fig. 3A). We then tried to correlate sensitivity to oxidants with protein oxidation for the different genetic backgrounds. Carbonylation of protein residues (see "Experimental Procedures" for determination) was used as a measure of protein oxidation (10,14). As expected from previous studies (14), under normal growth conditions ⌬grx5 cells contained more carbonyl groups in proteins than wild type cells. In contrast, ⌬mlp1 cells exhibited even fewer carbonyl groups in total protein extracts than wild type cells (Fig. 3B). Protein carbonylation in the double ⌬grx5 ⌬mlp1 mutant was slightly higher than in the single ⌬grx5 mutant (Fig. 3B). When total carbonyl groups were quantified (10,14), the double mutant consistently gave values about 30% higher than ⌬grx5 cells. These differences were much larger after treatment with t-BOOH; under these conditions we observed an additive effect between the two mutations on protein carbonyl content (Fig. 3B). It is also interesting that protein oxidation in the treated single ⌬mlp1 FIG. 1. Comparative expression of S. cerevisiae genes in ⌬grx5 and ⌬pet117 cells. A, expression of each individual gene (marked by a square) in ⌬grx5 cells (MML100) in relation to its expression in ⌬pet117 cells. Values result from normalizing the expression level of the gene in the respective mutant by its expression in wild type cells (log 2 of the ratio). B, genes whose relative expression was significantly modified in divergent ways in ⌬grx5 and ⌬pet117 mutants. Only induction and repression ratios (mutant versus wild type) of two or more were considered significant.

FIG. 2. Representative genes up-regulated in ⌬grx5 cells relative to wild type cells.
A, Northern blot analysis of some of the genes that were induced (Table I) in the ⌬grx5 mutant (strain MML100) relative to wild type cells (W303-1A). The small nuclear U2 mRNA served as a loading control, and ACT1 is representative of genes whose expression did not vary in the mutant. Samples were taken from exponential cultures in YPD medium at 30°C. B, Northern blot analysis of BIO5 expression in exponential cultures of wild type and ⌬grx5 cells in SC medium at 30°C, with biotin at the indicated concentrations. C, Northern blot analysis of FET3 and BIO5 expression in wild type (W303-1A) and aft1-⌬5 (MML348) cells. Samples were obtained from exponential cultures in YPD medium at 30°C; ferrozine was added at 2 mM final concentration at time 0. Same amounts of RNA were loaded per lane, as determined by small nuclear U2 mRNA analysis (not shown). mutant was greater than in wild type cells but less than in ⌬grx5-treated cells. Altogether these results support the protective role of MLP1 against external oxidants, particularly when the Grx5 function is absent.
Genes of the Yap1 and Msn2/4 Regulons Were Not Constitutively Induced in the Absence of Grx5-In yeast, oxidative stress by external agents such as hydrogen peroxide, menadione, or diamide causes up-regulation of genes involved in glutathione and thioredoxin-based defense and repair systems against oxidative damage, and also of other antioxidants such as superoxide dismutases and catalases (see Ref. 6 for details). This induction is regulated by one or several of the transcriptional factors Msn2/4, Yap1, and Skn7. None of the above genes appeared to be up-regulated in the experiments shown in Table  I for ⌬grx5 cells. We confirmed these results by Northern blot analysis for a number of genes that are known to be up-regulated by external oxidants: DDR2 (41), GSH1 (42), HSP12 (43), CTT1 (41), TRX2 (44), and TRR1 (5). Basal expression in ⌬grx5 cells did not differ significantly from wild type cells in any of these cases (Fig. 4A). However, the Yap1 and Msn2/4 regulatory systems remained functional in the mutant, because GSH1 (whose induction by oxidants is Yap1-dependent (5, 42)) and DDR2 (Msn2/4-dependent (7) 3 ) were inducible by menadione in wild type and ⌬grx5 cells (Fig. 4B).
Repressed Genes in ⌬grx5 Cells-A total of 64 genes had an expression level in ⌬grx5 cells that was 50% or less with respect to wild type cells (Table II). Thirty four of these genes were also down-regulated by external oxidants, as shown in Refs. 2 and 3. This confirmed that the ⌬grx5 cells were subjected to metabolic oxidative stress. However, the remaining 30 genes in Table I were specifically down-regulated in the mutant. These included the YFH1 frataxin gene. Decreased expression of the latter was confirmed by Northern blot analysis (Fig. 5). Twenty of the down-regulated genes were nuclear genes involved in mito-chondrial functions, mostly in respiration (Table II). Among these, we confirmed the down-regulation of CYC1, COX13, and CYT1 in the ⌬grx5 cells by Northern blot analysis. Of the three genes, only CYC1 was also down-regulated in the ⌬yfh1 mutant, although to a lesser extent than in the ⌬grx5 mutant (Fig.  5). This confirms the results described in Ref. 27. It has been reported that petite cells display decreased expression of a number of genes related to mitochondria (23). However, none of the genes reported in that study were among the genes downregulated in the present work. Only three of the mitochondrial function genes in Table II (CYC1, COX5A, and COX17) displayed decreased expression in the ⌬pet117 mutant. We therefore concluded that for most of these mitochondrial function genes, the decreased expression in the null grx5 mutant is not due to the petite character of the latter. It can be deduced that this is a specific response of genes involved in respiration and other mitochondrial energy metabolism functions caused by 3 (Table I) in the ⌬grx5 mutant (strain MML100) relative to wild type cells (W303-1A). Expression in the ⌬yfh1 mutant MML298 is shown in parallel. The small nuclear U2 mRNA served as a loading control, and ACT1 is representative of genes whose expression did not vary in the mutant. Samples were taken from exponential cultures in YPD medium at 30°C. the continuous generation of ROS in Grx5-deficient cells. Hap4 is a transcriptional factor that up-regulates S. cerevisiae respiratory functions and that is important in the shift from fermentative to respiratory metabolism, which occurs at the diauxic shift (12,45). We analyzed previous reports (11,12) to determine whether there were any changes in the expression pattern of mitochondrial function genes down-regulated in ⌬grx5 cells during the diauxic shift (11) or after overexpression of Hap4 (12). In fact, the expression of 16 of these genes is up-regulated after the diauxic shift and/or by Hap4 overexpression (Table III). Two of the unknown function open reading frames (YDR316W and YER182W) showing strong down-regulation in ⌬grx5 cells (Table II) are also Hap4-dependent (12).
We studied the expression of one of these genes, CYC1, in conditions in which HAP4 was overexpressed in order to demonstrate that down-regulation of Hap4-dependent genes in cells deficient in Grx5 was not indirect. As expected from previous studies (46), overexpressing HAP4 in wild type cells led to an increased expression of CYC1 (Fig. 6A). Under the same conditions, ⌬grx5 cells expressed CYC1 at almost the same level (only 20% reduction) as the wild type cells, contrasting significantly with the situation in non-overexpression conditions (which experienced a 3-fold reduction) (Fig. 6A). This confirmed that the observed down-regulation of Hap4-dependent genes occurred as a direct consequence of the absence of Grx5.
Based on the hypothesis that down-regulation of respiratory genes is a response to the continuous oxidation of cell molecules that occurs in the absence of Grx5, it would be expected that a shift to respiratory metabolism by HAP4 overexpression would have a negative effect on the physiology of ⌬grx5 cells. Although these cells have no functional respiratory chains due to defects in iron-sulfur cluster biogenesis (17), ROS could still be produced as a consequence of incomplete electron transfer. The overexpression of HAP4 caused increased carbonylation of cell proteins even in a wild type background, although this effect was higher in ⌬grx5 cells (Fig. 6B). This confirmed that respiratory conditions are a source of ROS able to cause oxidative damage on proteins, and also that there are additive effects between the absence of Grx5 activity and the increased expression of respiratory genes. In accordance with these observations, overexpression of HAP4 in the ⌬grx5 mutant significantly lowered growth rate compared with mutant cells displaying normal levels of HAP4 expression. In contrast, this effect of HAP4 overexpression on cell growth was not observed in wild type cells (Fig. 6C). Altogether, these observations support the biological significance of the observed down-regulation of respiratory genes in the absence of Grx5. DISCUSSION In this work we used a Grx5-deficient mutant as a model for yeast cells continuously subjected to moderate oxidative stress, in order to study the genes whose expression was significantly modified under these conditions. As the ⌬grx5 mutant exhibits a petite phenotype (17), we had to first distinguish between the genes whose modified expression was specifically due to the petite character of the cells and those genes affected by other physiological effects of the grx5 mutation. Very few of the genes affected in the ⌬pet117 mutant were also affected in the ⌬grx5 mutant cells (Tables I and II). Furthermore, there was no overlap with the genes shown to be differentially expressed in °cells in a previous study (23). Therefore, most of the expression changes seen in ⌬grx5 cells do not seem to be due to the respiratory deficiencies in these cells. It is also interesting that a small set of genes displayed significantly altered expression in opposite ways when both mutants were compared (Fig. 1B). Most of the 12 genes that were up-regulated in ⌬grx5 cells and down-regulated in the petite mutant are of unknown function, although the group also included two genes (FIT3 and ARN2) related to iron uptake. Of the 16 genes up-regulated in the ⌬pet117 mutant and down-regulated in the ⌬grx5 cells, four (ILV3, LYS4, BAT1, and GAP1) are involved in amino acid metabolism and transport. This could be related to the fact that up to 32 genes involved in amino acid metabolism and transport were induced in the ⌬pet117 mutant. This is a situation that extends to genes for purine and pyrimidine biosynthesis (data not shown in detail). The up-regulation of genes in these functional categories was not observed in °cells (23).
A subset of the genes up-regulated in the absence of Grx5 is also up-regulated in the frataxin yfh1 mutant (27). These genes are regulated by the transcriptional activator Aft1. The latter responds to iron deprivation by regulating its nuclear localization and activating genes that participate in iron and copper utilization (24 -29, 47). In this study we added BIO5 to the list of genes that are regulated by Aft1 in a way that is dependent on iron concentration in the medium. This confirmed the previously reported relationship between biotin metabolism and iron assimilation (29). The up-regulation of Aft1-dependent genes in the yeast frataxin mutant was taken as evidence that the iron accumulating in the mitochondria of the mutant cells is in a non-metabolizable form, and that yfh1 cells are therefore nutritionally depleted of iron (27). Similarly, iron accumulates intracellularly at abnormally high levels in the ⌬grx5 mutant (17), probably as a consequence of the disruption of the Grx5 function in the iron-sulfur cluster assembly. The results shown here suggest that the accumulated iron in the grx5 cells is not available for cell metabolism. There is, however, an alternative explanation, which is based on the fact that the YFH1 gene was down-regulated in cells lacking Grx5 (Table I and Fig. 5). This could make the ⌬grx5 cells phenotypically similar to a ⌬yfh1 mutant. In this case, iron accumulation in the ⌬grx5 mutant would not be a direct consequence of disruption of iron-sulfur cluster assembly but secondary to depletion of Yfh1 molecules. We, however, favor the first alternative, because iron accumulation is common to many mutants in genes involved in the assembly of iron-sulfur clusters (48). At this respect, recent studies indicate that the primary function of yeast frataxin is the synthesis of iron-sulfur clusters (49,51) and heme groups (52). In order to carry out this function, Yfh1 would act as an iron chaperone and an iron store to provide Fe(II) (53) to the assembly complexes.
It has been reported that iron-mediated generation of ROS in  the yeast frataxin mutant causes nuclear DNA damage (54), although this does not lead to significant induction of DNA repair genes (27). Similarly, DNA damage could also occur in ⌬grx5 cells, where iron accumulation is not limited to the mitochondrial compartment (17). However, also in this case DNA repair genes are not up-regulated.
Two genes that could be related to protection against the oxidative stress generated in these conditions were up-regulated in the absence of Grx5. One of these is SNZ1, through its role in the biosynthesis of pyridoxine (36), a vitamin that acts as a protector against single oxygen (37). The second is MLP1, which codes for a homologue of the Slt2 MAP kinase (39). Slt2 plays a central role in the S. cerevisiae cell integrity pathway by activating at least two transcriptional factors, Rlm1 and SBF (55). These factors respond to stimuli that alter the integrity of the cell envelope by respectively activating the expression of cell wall biosynthesis genes and cell cycle genes (56,57). However, genes that protect against oxidants do not appear to be targets of Rlm1 or SBF. Genetic interactions between SLT2 and MLP1 suggest the existence of some parallel functions between the two gene products (39), but the hypothetical targets for the Mlp1 function are unknown. Our results showed additivity between the ⌬grx5 and ⌬mlp1 mutants in their sensitivity to oxidants and in their levels of both constitutive and oxidant-induced protein carbonylation. This supports a functional relationship between Grx5 and Mlp1. The single ⌬mlp1 mutant was no more sensitive to oxidative stress than wild type cells. Therefore, the Mlp1 hypothetical protective role against oxidative stress was only manifested under the continuous stress conditions occurring in the absence of the Grx5 function. Transcriptome analysis of the mlp1 mutant and of cells overexpressing MLP1 could help to reveal more details about the role of Mlp1 relative to oxidative stress.
On the other hand, Msn2/4, Yap1, and Skn7 regulon genes were not constitutively up-regulated in ⌬grx5 cells. This indicates that either the stress signal generated in the glutaredoxin mutant was not sufficient to activate the respective transcriptional activators, or that these were in fact only acting in response to instantaneously generated external signals and not to a sustained stress signal.
A significant number of genes whose expression was downregulated in the ⌬grx5 background are Hap4-regulated. Hap4 is the activator component of a complex in which the Hap2/3/5 proteins also participate. These are necessary for binding of the complex to specific promoter sites. The Hap complex is required for expression of respiratory genes (58). Overexpression of Hap4 is sufficient to cause the shift from fermentative to res-piratory metabolism, through induced expression of respiratory genes (12,46). We have shown in the present work that overexpression of HAP4 suppresses the down-regulation of the Hap4-dependent gene CYC1 in ⌬grx5 cells. This supports the idea that expression of the respiratory genes is repressed in the mutant cells through the inhibition of Hap4 activity. Heme could initially be proposed as the mediator involved in the process. In fact, it is required for the activation of Hap4-dependent genes (29,58). In ⌬grx5 cells, depletion of metabolically available iron and consequent up-regulation of the heme oxygenase gene HMX1 (29) could lead to heme depletion and eventually to Hap4 inactivation. However, we have shown previously (17) that heme levels are not different in the ⌬grx5 cells with respect to wild type cells. Therefore, this points to additional factors being responsible for regulating the Hap complex activity under conditions of moderate continuous oxidative stress. Iron accumulation in the ⌬grx5 mutant is not limited to the mitochondria, and proteins from different cellular compartments are oxidized in this mutant (14,17). Therefore, it is not necessary to hypothesize an oxidative damage signal transduction from mitochondria (the location of Grx5) to the nuclear Hap complex. Many aspects of the regulation of this complex remain to be elucidated, and other regulators besides heme seem to exist (58). Further studies are required to determine whether the oxidative signal acts directly on the Hap complex.
We have shown by macroarray and Northern analysis that a number of the respiratory genes that are down-regulated in the ⌬grx5 cells are not affected in the frataxin mutant, in accordance with previous studies (27). The yfh1 mutant shows lower carbonylation of cellular proteins than the ⌬grx5 mutant. 4 This adds support to the specificity of the Hap-related transcriptional response observed in the Grx5-deficient cells.
Based on the role of Grx5 in the biogenesis of iron-sulfur clusters, the respiratory deficiency of ⌬grx5 cells (17) could be initially explained by the absence of Rip1 activity. Rip1 is the iron-sulfur Rieske protein that is part of complex III in the yeast respiratory chain (59). S. cerevisiae cells have no respiratory complex I, which in mammalian cells also contains ironsulfur proteins. In yeast, complex I is replaced by the mitochondrial inner membrane Ndi1 NADH-dehydrogenase (60,61). In mammalian cells, both complex I and complex III are sources of ROS, whose production is enhanced by cytochrome oxidase inhibition (50,62). It can therefore be hypothesized that incomplete respiratory chains in Grx5-deficient cells are a source of 4 6. Effect of HAP4 overexpression in wild type (wt) and ⌬grx5 cells. Wild type (W303-1A) and ⌬grx5 (MML100) cells were transformed with plasmid pMT15 (that overexpresses HAP4 under the tet promoter). Non-transformed and pMT15-transformed cells were grown in liquid SC medium at 30°C. Samples were taken in exponential conditions. A, Northern blot analysis of HAP4 and CYC1 expression. Small nuclear U2 mRNA is shown as a loading control. B, carbonyl groups in total cell proteins, after Western blot analysis (20 g of total protein per lane). C, growth rates of the respective cultures (mean of three experiments). Values over bars indicate the optical density (at 600 nm) doubling times in minutes, in exponential growth conditions. ROS with substrates such as NADH or succinate, unless all chain components are depleted, for instance through Hap4 inhibition. The situation observed in cells lacking Grx5, in which expression of the different components of the respiratory chain (including the NDI1 gene) seems to be down-regulated through Hap4 activity, could be extrapolated to other situations in which ROS-mediated stress exists. This could be a general response to prevent respiratory production of ROS. Other models for metabolically generated continuous oxidative stress need to be tested to confirm this hypothesis.