|Home | About | Journals | Submit | Contact Us | Français|
Methionine residues and iron-sulphur (FeS) clusters are primary targets of reactive oxygen species in the proteins of microorganisms. Here we show that methionine redox-modifications help to preserve essential FeS cluster activities in yeast. Mutants defective for the highly conserved methionine sulphoxide reductases (MSRs; which re-reduce oxidized methionines) are sensitive to many pro-oxidants, but here exhibited an unexpected copper resistance. This phenotype was mimicked by methionine sulphoxide supplementation. Microarray analyses highlighted several Cu and Fe homeostasis genes that were upregulated in the mxrΔ double mutant, which lacks both of the yeast MSRs. Of the upregulated genes, the Cu-binding Fe-transporter Fet3p proved to be required for the Cu-resistance phenotype. FET3 is known to be regulated by the Aft1 transcription factor, which responds to low mitochondrial FeS-cluster status. Here, constitutive Aft1p expression in the wild type reproduced the Cu-resistance phenotype, and FeS cluster functions were found to be defective in the mxrΔ mutant. Genetic perturbation of FeS activity also mimicked FET3-dependent Cu resistance. 55Fe-labeling studies showed that FeS clusters are turned over more rapidly in the mxrΔ mutant than the wild type, consistent with elevated oxidative targeting of the clusters in MSR-deficient cells. The potential underlying molecular mechanisms of this targeting are discussed. Moreover, the results indicate an important new role for cellular MSR enzymes, in helping to protect the essential function of FeS clusters in aerobic settings.
Reactive oxygen species (ROS) are generated continuously through the process of respiration in all aerobic organisms, and may cause the oxidative deterioration of lipid, DNA and protein function in cells (Temple et al., 2005). ROS damage is often exacerbated during stress. To counter this, cells have evolved a range of enzymatic and non-enzymatic antioxidant defence mechanisms. The majority of these serve a preventative role in scavenging ROS before they exert damage. Others repair incurred oxidative damage, including phospholipid hydroperoxide glutathione peroxidases (lipid peroxidation) (Avery et al., 2004), and 8-oxoG DNA glycosylases (DNA oxidation) (Boiteux & Radicella, 1999). The principal enzymatic mechanism for reversing protein oxidation acts on the oxidation product of just one amino acid residue, methionine. This specificity for Met reflects the fact that Met is particularly susceptible to oxidation, compared with other amino acids. Furthermore, Met oxidation has been linked to Alzheimer’s (Schoneich, 2005) and Parkinson’s (Wassef et al., 2007) diseases. Methionine sulphoxide reductases (MSRs) are conserved across nearly all organisms from bacteria to humans (Kryukov et al., 2002), and have been the focus of considerable attention in recent years. Two MSR activities have been characterized in the yeast Saccharomyces cerevisiae: MsrA (encoded by MXR1) reduces the S stereoisomer of methionine sulphoxide (MetO), while MsrB (encoded by the YCL033c open reading frame, which we term here MXR2) reduces the R stereoisomer of MetO (Koc et al., 2004; Kryukov et al., 2002). Consistent with defence against oxidative damage, mutants deficient for MSR activity are hyper-sensitive to pro-oxidants such as H2O2, paraquat and Cr, while MSR overexpression enhances resistance (Kryukov et al., 2002; Moskovitz et al., 1997; Moskovitz et al., 1998; Sumner et al., 2005). In addition to countering protein damage caused by MetO, it has been proposed that MSR activity could act to support a ROS-scavenging function of reduced and surface-exposed Met residues (Levine et al., 1996; Melkani et al., 2006). Furthermore, reversible oxidation of Met residues may serve as a mechanism for regulating protein activity and in cellular signalling processes [for reviews see (Bigelow & Squier, 2005; Moskovitz, 2005)].
Besides methionine residues, iron-sulphur (FeS) clusters are exquisitely ROS-sensitive components of many (>120) cellular proteins. FeS clusters are highly abundant and diversely employed enzymatic cofactors that have ancient origins (Imlay, 2006). FeS cluster biogenesis is the only known process for which mitochondria are essential to yeast cells (Lill & Kispal, 2000). FeS clusters participate in electron transfer (featuring commonly in redox enzymes), substrate binding and activation, iron/sulphur storage and regulation of gene expression (Johnson et al., 2005). Haemoglobin formation in differentiating red blood cells is regulated through FeS cluster assembly (Wingert et al., 2005) and decreased activity of FeS enzymes is a characteristic feature of the neurodegenerative disease Friedreich’s ataxia (Shan et al., 2007). The vulnerability of FeS clusters to oxygen and oxidative stress greatly complicates their roles in biology. Oxidation of FeS causes loss of protein function, and at the same time releases free Fe which may participate in Fenton catalysis to exacerbate oxidative stress (Keyer & Imlay, 1996; Srinivasan et al., 2000). Denaturation of FeS clusters is not the only way in which these co-factors may influence cellular Fe availability. A signal which represses the Fe regulon (and so the uptake of Fe) at high cellular Fe levels in yeast is transmitted through a product of the mitochondrial FeS biogenesis machinery (Chen et al., 2004; Kumanovics et al., 2008; Rutherford et al., 2005). Thus, Fe levels influence FeS biogenesis which in turn regulates Fe homeostasis. Iron homeostasis in yeast also interplays closely with homeostasis of another essential but potentially-toxic Fenton catalyst, copper (Askwith & Kaplan, 1998). The high affinity Fe transporter Fet3p requires bound copper for activity and this Cu-binding contributes to cellular Cu resistance (Stoj et al., 2007). However, direct links between cellular FeS cluster status and Cu transport or toxicity have not been established to date. The same is true of FeS clusters and Met residues, despite their shared notoriety as highly oxidation-sensitive components of cellular proteins.
In this report, an unexpected Cu resistance phenotype of MSR-deficient S. cerevisiae cells was traced to activation of the Fe regulon, specifically FET3. Moreover, it was established that MSR activity helps to preserve the function of cellular FeS clusters. The finding suggests that the ubiquity of MSR enzymes could be an adaptation that enables the operation of FeS clusters in aerobic settings.
Saccharomyces cerevisiae BY4741 (MATa; his3Δ1; leu2Δ0; met15Δ0; ura3Δ0) and the isogenic deletion strains aft1Δ, aft2Δ, ctr1Δ, cup5Δ, fet3Δ, fet4Δ, mxr1Δ, mxr2Δ (YCL033cΔ), sod1Δ, and sod2Δ, in which the specified open reading frames (ORFs) are replaced with the KanMX4 marker, were obtained from Euroscarf (Frankfurt, Germany). Short flanking homology (SFH-) PCR (Longtine et al., 1998) was used to disrupt additional genes (see below) in this study; all primer sequences are available on request. SFH-PCR products were transformed into S. cerevisiae with the lithium acetate method (Gietz & Woods, 2002) and transformants selected on yeast nitrogen base (YNB) without amino acids (Formedium), supplemented as required to give appropriate selection (Ausubel et al., 2007). To select for the hphNT1 marker (see below), transformants were plated on YEPD medium (see below) supplemented with 150 μg ml-1 hygromycin (Invitrogen, UK). Diagnostic PCR (Longtine et al., 1998) was used to confirm appropriate gene disruption in transformants. Where used, colony PCR for diagnosis of gene disruption was according to (Amberg et al., 2005). An mxrΔ double mutant (mxr1::KAN, mxr2::hphNT1) isogenic with BY4741 was constructed using pFA6a-hphNT1 (Janke et al., 2004) as the template for SFH-PCR based disruption of MXR2 in the S. cerevisiae mxr1Δ background. A previously-constructed mxr double mutant strain (mxr1::URA3, mxr2::KanMX4, isogenic with BY4741) (Kryukov et al., 2002) was used for further gene deletions in the mxrΔ background. SFH-PCR was used to disrupt the following genes in the mxrΔ background and, where specified in the Results, also in the BY4741 wild type strain: CTR1, FET3, FET4, SOD1, CUP5, CUP1-1, AFT2 (all with the His3MX6 marker), AFT1 (with the hphNT1 marker). It was confirmed that the markers themselves did not influence Cu resistance. The LEU2 marker, amplified from pRS315 as template, was used for disruption of CUP1-2 in the cup1-1Δ and mxrΔ/cup1-1Δ backgrounds. FET3 was disrupted in the cup1Δ, mxrΔ/cup1Δ and sod2Δ strains with hphNT1. A disruption construct described elsewhere (Portnoy et al., 2001) was used to delete CTR2, following linearization of plasmid pJS411 with BamHI and transformation into S. cerevisiae BY4741 or mxrΔ cells.
Standard cloning procedures (Ausubel et al., 2007), reagents from New England Biolabs and the cloning host E. coli DH5α were used throughout. For complementation, a fragment comprising the MXR1 ORF and native promoter was cloned between the NotI and BamHI sites of the single copy vector pRS315-hphNT1. pRS315-hphNT1 was previously constructed by ligating the hphNT1 marker excised from pFA6a-hphNT1 between the BbsI and NarI sites of pRS315. For overexpression of SOD1 or FET3, fragments encompassing these ORFs plus their native promoters were amplified from yeast genomic DNA and ligated between the KpnI-SalI (SOD1) or EcoRI-XbaI (FET3) sites of the multicopy vector YEp352. Yeast transformants were selected on YNB-minus-uracil medium. An AFT1-1up fragment was released from pRS313-AFT1-1up (Chen & Kaplan, 2000) by digestion with SacI and SpeI. This 3.6 kb fragment was ligated into pRS315-hphNT1. Yeast transformants were selected in YEPD agar supplemented with 150 μg ml-1 hygromycin. Plasmid-transformed yeast cells were grown throughout, for maintenance or for experimental purposes, in the relevant medium selective for the plasmid.
Organisms were maintained either in liquid YEPD medium or, where specified, in YNB medium supplemented with the appropriate amino acids or nucleic acid bases (Ausubel et al., 2007). The same media were used for preparation of experimental cultures, by subculture from stationary phase culture and growth to mid-/late-exponential phase (A600 ~2.0) at 30°C, 120 rev min-1 (Bishop et al., 2007). Growth in the broth media after addition of Cu(NO3)2 or paraquat (from filter-sterilized stock solutions) was followed at A600 in 300 μl volumes in 48 well plates (Greiner Bio-One), incubated with shaking in a BioTek Powerwave microplate spectrophotometer (Smith et al., 2007). For qualitative viability assays on solid medium, dilution series from experimental cultures adjusted to A600 ~2.0, 0.2, 0.02, 0.002, 0.0002 were spotted (8 μl) on to medium solidified with agar [1.6% (w/v)] and supplemented as specified. An L-methionine sulphoxide supplement containing a mixture of R and S MetO isomers was obtained from Sigma-Aldrich. Quantitative viability tests on solid medium were based on colony forming ability, as described previously (Smith et al., 2007).
Cu(NO3)2 was added to experimental cultures (A600 ~2.0) to a final concentration of 3 mM and, after 20 min incubation, cells (25 ml samples) were harvested by centrifugation (740 g, 5 min). Control incubations at 4°C were set up in parallel to enable discrimination of Cu that was passively bound to cells (Avery et al., 1996; Lin & Kosman, 1990). For analysis by atomic absorption spectrophotometry (AAS) cell digests were prepared with HNO3 as described previously (Avery et al., 1996). Samples were filtered, diluted with 3 N HNO3, and analyzed for Cu content using a SpectrAA 220FS atomic absorption spectrophotometer (Varian) that was calibrated with 0-0.5 mg ml-1 Cu(NO3)2 standard solutions. To support AAS measurements of Cu uptake, a colorimetric assay employing bathocuproine disulphonic acid (BCS) (Ramirez et al., 2005) was used to determine Cu concentrations in supernatants after removal of cells by centrifugation (above). Cellular Cu uptake was determined by difference versus Cu determinations in parallel control flasks that lacked cells.
RNA was extracted from cells of experimental cultures (A600 ~1.0) using the RNeasy Mini kit (Qiagen), after cell breakage with a mini-Beadbeater (Biospec Products) for 3 × 30 s interspersed with incubations on ice. Residual DNA was removed by treatment with RNase free DNase (Promega; 0.05 U μg-1). For each yeast strain, RNA extracts from each of six replicate cultures were pooled into two samples (each comprising RNA from three replicate cultures), snap-frozen in liquid nitrogen and stored at -20°C until use. The integrity of RNA and cRNA was confirmed in all samples before analysis using an Agilent Bioanalyser. These analyses, RNA preparation, hybridizations to GeneChip® Yeast Genome S98 Arrays (Affymetrix) and analysis with an Affymetrix GeneChip Reader were performed as a service by the NASC Affymetrix Service, University of Nottingham. Raw signal intensity data from hybridizations were normalized after removal of the highest and lowest 2% of signal values, and gene signals scoring ‘marginal’ or ‘absent’ were removed from the datasets (Payne et al., 2008). Genes showing ≥25% difference in signal intensities between replicate samples were also removed from the analysis. Fold up/down-regulation values for individual transcripts were subsequently determined for mean signal values derived from the test (RNA from mutant) samples with reference to the corresponding control (wild type) analyses.
Cellular RNA was prepared as described above. The absence of protein or DNA contamination in RNA samples was confirmed according to A260/A280 ratios and standard PCR tests, respectively. For reverse transcription reactions, 1 μl of dNTP mix (10 mM each of dATP, dCTP, dTTP, dGTP), 1 μl 2.5 μg μl-1 oligo (dT)20 primer (Invitrogen), 1 μg RNA template were combined and made up to 13 μl with nuclease free water. The mixture was heated to 65°C for 5 min, followed by incubation on ice for ≥1 min. The contents of the tube were collected by brief centrifugation and 1 μl of 0.1-M dithiothreitol, 4 μl of 5x first strand buffer and 1 μl of superscript III reverse transcriptase (200 U μl-1; Invitrogen) were added. After incubation at 50°C for 60 min, the reaction was inactivated by heating to 70°C for 15 min. Residual RNA was removed by addition of 1 μl (2 U) E. coli RNase H (Invitrogen) and incubation at 37°C for 20 min.
The relative abundance method was used to determine resultant cDNA levels by quantitative real time (qRT)-PCR. For reactions, 0.4 μM each of gene-specific primers (HPLC purification scale, Sigma-Genosys), 2 μl cDNA template from first-strand cDNA synthesis (10-1 dilution), and 2x QuantiTect SYBR Green PCR Master Mix were combined and made up to 25 μl with RNase-free water in 0.2 ml tubes (Stratagene). PCR reactions [95°C for 10 min followed by (95°C for 30 s, 55°C for 1 min, 72°C for 30 s) for 40 cycles, followed by 95°C for 1 min, and 55°C for 20.5 min] were monitored using a MX4000 real time PCR thermocycler (Stratagene). The results were normalized and analyzed with the MX4000 software.
For semi-quantitative RT-PCR, 10-1, 10-2 and 10-3 dilutions of the cDNA were used as the template in PCR reactions; 95°C for 1 min followed by (95°C for 30 s, 55°C for 1 min, 72°C for 1 min), 30 cycles; followed by 72°C for 10 min and then 4°C. The PCR products were examined with agarose gel electrophoresis and the intensities of the bands were compared between wild type and mutant strains as specified, using template dilutions at which the reaction had not progressed to saturation. ACT1 was used as the reference mRNA.
Proteins extracts were prepared as described elsewhere (Cashikar et al., 2005). Incubations were under nitrogen to protect FeS clusters. Protein extracts in supernatants were transferred to clean tubes and protein concentrations determined (Bradford, 1976). The assay of aconitase activity in protein extracts (from 2 × 108 cells) under nitrogen was as described elsewhere (Wallace et al., 2005). The change in absorbance was recorded at 1 min intervals for 8 min in UV cuvettes and the specific activity (per mg protein) calculated from the linear portion of the resultant plot. Isopropylmalate dehydratase (Leu1) activity was assayed as described (Kohlhaw, 1988). The change in absorbance was recorded over 2 min following substrate addition, and the specific activity (per mg protein) calculated from the linear portion of the resultant plot.
Cultures (25 ml) were grown overnight in YEPD to mid-exponential phase (A600 ~0.5) in 125 ml flasks, and 12.8 μCi 55FeCl3 (Perkin-Elmer) added. For measurement of FeS cluster biosynthesis, cells were harvested after incubation for 1 or 2 h with 55Fe. For measurement of FeS cluster turnover, the 55Fe pre-loaded cells were washed and resuspended in YEPD medium lacking 55Fe, and incubated for a further 1 or 2 h before harvesting. Protein extracts were prepared and immunoprecipitations with antibodies specific for Bio2p or Leu1p (gifts from Roland Lill, University of Marburg) were performed as described (Molik et al., 2007). Incorporated isotope was quantified in 1 ml scintillation fluid (Emulsifier Safe, Perkin Elmer) using a Packard Tri-Carb 2100TR liquid scintillation analyzer.
During a study on the contribution of oxidative damage repair proteins to metal resistance in S. cerevisiae, we observed an unexpected Cu resistance phenotype in a mxr1Δ/mxr2Δ double mutant (hereafter referred to as mxrΔ), which lacks methionine sulphoxide (MetO) reducing activity. Growth of this mxrΔ mutant (Kryukov et al., 2002) was only mildly affected by Cu(NO3)2 at a concentration which strongly inhibited growth of the wild type (Fig. 1A). [Note that high (mM) Cu concentrations are needed to observe growth inhibitory effects in rich medium (YEPD) such as that used here, as much of the added Cu is biologically-unavailable in complexes with medium components (Avery et al., 1996; Hughes & Poole, 1991)]. The mxrΔ double mutant exhibited stronger Cu resistance than the mxr1Δ and mxr2Δ single mutants, which are defective for reduction of different MetO stereoisomers. The Cu resistance of the mxrΔ mutant was evident also from growth on Cu-supplemented agar (Fig. 1B), was suppressed by complementation with the MXR1 gene (Fig. 1B) and was confirmed in an independently-constructed mxrΔ double mutant from our laboratory. The Cu-resistance phenotype could be mimicked by growing wild type cells in MetO-supplemented medium, whereas Met supplementation had no effect. MetO also abolished the relative resistance of the mxrΔ mutant (Fig. 1C). These data suggested that it is specifically the accumulation of methionine sulphoxide (rather than Met depletion or similar) in mxrΔ cells that gives rise to Cu resistance. Yeast mutants lacking both MSRs have been reported previously to be sensitive to other pro-oxidants (Kryukov et al., 2002; Sumner et al., 2005). Those sensitivity phenotypes were confirmed during this study, indicating that the resistance phenotype was specific to Cu.
To explain the Cu resistance of mxrΔ cells it was first hypothesized that, in the absence of MSR activity, oxidation of Met residues in Cu uptake proteins may cause inactivation and a decrease in Cu accumulation. The high affinity Cu transporter Ctr1p is enriched in Met residues [proteins that interact with MSR are usually Met-rich (Alamuri & Maier, 2006; Le et al., 2008)], some of which are required for Cu uptake (Puig et al., 2002). However, epistasis tests showed that MXR gene deletion conferred Cu resistance also in a ctr1Δ background (data not shown), indicating that loss of Ctr1p activity is not the cause of Cu resistance in mxrΔ cells. Loss of activity of Ctr2p (the putative low-affinity vacuolar Cu transporter) was also ruled out as, unlike MXR gene deletion, CTR2 deletion did not improve growth on Cu-supplemented agar. We also demonstrated with PCR that CTR3, which encodes a high affinity Cu transporter, was disrupted by a Ty2 transposon insertion in the strains used here, as in most S. cerevisiae laboratory strains (Knight et al., 1996). Therefore, Ctr3p activity was not involved here. Epistasis experiments with mutants deficient in Fet4p, a low affinity copper transporter, suggested that Fet4p inactivation might partly contribute to Cu resistance in the mxrΔ mutant (Supplementary Fig. S1). Moreover, Cu accumulation tests indicated that an effect on Cu uptake was not the cause of Cu resistance in mxrΔ cells. According to two independent methods, calculated to exclude (Fig. 2A) or include (Fig. 2B) surface associated Cu, Cu accumulation was up to 2-fold higher in mxrΔ than in wild type cells. It was inferred that, for both Cu accumulation and Cu resistance to be elevated in mxrΔ cells, intracellular Cu must be less available to exert toxicity.
In light of the above results, it was hypothesized that gene products which modulate the availability of free Cu in the cell might be upregulated in mxrΔ versus wild type cells. To test this, the transcriptomes of the two strains were compared. A preliminary microarray experiment in our laboratory indicated that ~30 genes were upregulated ≥2-fold in mxrΔ cells versus the wild type (Supplementary Table S1). These included CCC2 which encodes a Cu transporter and, most strikingly, a set of genes of the iron regulon (Puig et al., 2005; Rutherford et al., 2003): FIT2, FIT3, ARN2, ARN3(SIT1), ARN4(ENB1), TIS11. To explore this further, we analyzed the more comprehensive microarray datasets produced elsewhere (Koc et al., 2004) for the same strain comparison. Those analyses distinguished 540 genes that were ≥2-fold upregulated in the mxrΔ mutant versus wild type. Again, Fe-regulon genes were among the most strongly upregulated, including FET3 (~3.8-fold), FIT2 (3.0), and ARN4 (2.7) (Fig. 3A). Over half of the ten most highly upregulated Fe regulon genes (Puig et al., 2005; Rutherford et al., 2003) were induced ≥1.5-fold in the mxrΔ mutant. In addition, several gene functions that specifically modulate intracellular Cu availability were upregulated in mxrΔ cells, including CUP1 (~2.0-fold), SOD1 (1.5), FET3 (3.8) and CUP5 (2.6) (Fig. 3B,C). Key transcriptomics data were validated either with qPCR or semi-quantitative RT-PCR. There were some quantitative differences in upregulation between the PCR-based and microarray assays. Nonetheless, according to both, each of the four genes tested (CUP1, CUP5, FET3 and YHB1) was markedly upregulated in mxrΔ versus wild type cells (Fig. 3B,C).
It was reasoned that the upregulation of one or more of the identified Cu resistance determinants in the mxrΔ mutant could explain these cells’ elevated Cu resistance. Consistent with this, multicopy expression of SOD1 and FET3 was confirmed here to elevate the Cu resistance of wild type S. cerevisiae (Supplementary Fig. S2). Conversely, CUP5 did not influence Cu resistance in our hands. SOD1, CUP1 and FET3 were selected initially for further study.
To test whether SOD1, CUP1 or FET3 were required for Cu resistance in the mxrΔ double mutant, in which they were each upregulated, these genes were deleted in the mxrΔ and wild type backgrounds. Deletion of SOD1 did not alter the relative Cu resistances of the mxrΔ and wild type strains (data not shown), indicating that Sod1p is not required for the mxrΔ phenotype. Although the MSR-deficient cells retained a relative Cu resistance following deletion of CUP1, the loss of resistance resulting from Cup1p deficiency was more marked for mxrΔ cells than for the wild type. Thus, no colony formation by the wild type was detectable at a Cu concentration yielding ~40% viability of mxrΔ cells (Fig. 4A), whereas cup1Δ cells retained >15% viability at a Cu concentration that yielded <40% viability in mxrΔ/cup1Δ cells (Fig. 4B). Therefore, CUP1 upregulation in mxrΔ cells (above) may partially contribute to their Cu resistance. In contrast, FET3 was required for the phenotype. As expected (Stoj et al., 2007), FET3 deletion resulted in a sensitization to Cu in both the wild type and mxrΔ backgrounds (note the Cu concentrations in Fig. 4C versus 4A). However, this sensitization was much greater in the mxrΔ mutant than in the wild type. Put another way, an absence of MSR activity had the opposite effect on Cu resistance in the fet3Δ background than in the wild type (Fig. 4A,C). The Cu sensitivity of mxrΔ/fet3Δ cells relative to fet3Δ cells was similar to that seen for other pro-oxidants in comparisons of mxr mutant and wild type cells (Kryukov et al., 2002; Moskovitz et al., 1997; Sumner et al., 2005). The results indicated that the unexpected Cu resistance of mxrΔ cells could be accounted for by the FET3 upregulation observed in this mutant. A partial role for CUP1 in the phenotype (above) was further supported by observations that removal of MSR activity in a cup1Δ/fet3Δ background caused a greater relative sensitization to Cu than in a fet3Δ background (Supplementary Fig. S3), but the more important role evident for FET3 provided the focus for subsequent experiments.
FET3 transcription is known to be under the regulation of the Aft2 and, in particular, the Aft1 transcription factors. These have overlapping roles in the regulation of iron utilization and homeostasis (Rutherford et al., 2003; Rutherford et al., 2005). The possible activation of Aft1 and/or Aft2 in mxrΔ cells was suggested not only by the upregulation of FET3 (and associated Cu resistance), but also of other Aft-dependent genes (Rutherford et al., 2003) such as ARN1, ARN4, FIT2 and FRE6 (Fig. 3A). We expressed an AFT1-1up allele, which produces constitutively active Aft1p (Yamaguchi-Iwai et al., 1995). AFT1-1up expression in wild type cells incubated in YEPD mimicked the phenotype of Cu resistance seen in mxrΔ cells (Fig. 5). In contrast, AFT1-1up expression in the mxrΔ mutant had no further effect on Cu resistance. The results were consistent with Aft activation determining Cu resistance in mxrΔ cells (see Fig. 6 for scheme).
Aft1p (and consequently FET3) is usually activated under Fe limited conditions. This activation has been traced largely to defective FeS cluster biogenesis in mitochondria (Chen et al., 2004; Rutherford et al., 2005). Accordingly, the transcriptomic responses to FeS biogenesis defects and Aft1 activation are strikingly similar (Hausmann et al., 2008). To test for FeS cluster defects in mxrΔ cells, the activities of the [4Fe-4S]-containing enzymes aconitase (Aco1p) and isopropylmalate isomerase (Leu1p) were assayed. Aconitase specific-activity was reproducibly ~30% lower in protein extracts from mxrΔ cells than in extracts from wild type cells (Fig. 7A). This activity defect was despite a ~6-fold higher level of ACO1 mRNA in mxrΔ cells (Koc et al., 2004). In conjunction with a >50% decrease in Leu1p activity in the mxrΔ mutant (Fig. 7A), these data supported the hypothesis that mxrΔ cells have a FeS cluster defect.
If, as inferred, FeS cluster defects in mxrΔ cells are the cause of FET3 upregulation and Cu resistance (Fig. 6), then it should be possible to mimic these phenotypes by manipulating the integrity of FeS clusters. This was tested by creating conditions conducive to FeS cluster degradation - achieved using a sod2Δ mutant defective for Mn-superoxide dismutase activity; like the process of FeS biosynthesis, Sod2p localizes to mitochondria where it scavenges superoxide radicals which are major antagonists of FeS cluster activity. Thus, Sod2p activity protects the integrity of mitochondrial FeS clusters (Irazusta et al., 2006; Strain et al., 1998). Consistent with this, the FET3 gene appeared to be upregulated in sod2Δ cells relative to the wild type (Fig. 7B). Furthermore, the sod2Δ mutant reproducibly showed a slightly higher copper resistance than the wild type in YEPD medium (Fig. 7C). This phenotype was similar, albeit less marked, to the phenotype of mxrΔ cells described above. To test whether FET3 was required for Cu resistance of the sod2Δ mutant, as in the mxrΔ mutant (Fig. 4A,C), the effect of SOD2 deletion was examined also in a fet3Δ background. Fet3p deficiency abolished the slight Cu resistance otherwise associated with SOD2 deletion (Fig. 7C,D). Thus, the FET3-dependent Cu resistance seen in mxrΔ cells could be reproduced, albeit less strikingly, in a mutant known to be limited for FeS cluster integrity.
It was speculated that the FeS cluster defect in mxrΔ cells could be due either to an increased rate of FeS cluster turnover or to decreased FeS cluster biosynthesis. To resolve these possibilities, the association of radiolabeled 55Fe with FeS proteins was measured. This was accomplished by immunoprecipitation of FeS proteins of interest from cells after incubation with 55FeCl3, and quantification of isotope in the immunoprecipitates. To distinguish FeS biosynthesis from FeS turnover, FeS proteins were immunoprecipitated from cells both during 55Fe incorporation (biosynthesis) and during a period after transferring 55Fe-loaded cells to 55Fe-free medium (turnover). First, we performed immunoprecipitations with the mitochondrial FeS protein biotin synthase (Bio2p). Analysis of immunoprecipitated protein with SDS-PAGE confirmed that the Bio2 protein levels in mxrΔ and wild type strains were similar (Supplementary Fig. S4). The incorporation of 55Fe into Bio2p after 1 h was decreased by about a third in the mxrΔ mutant (Fig. 8A), possibly suggesting a defect in FeS biosynthesis. A comparable difference was also observed after 2 h of 55Fe incorporation (Supplementary Fig. S5). However, analyses of 55FeS turnover indicated that a similar proportion (~30%) of the Bio2p-associated 55Fe of wild type cells was lost within 1 h in 55Fe-free medium, whereas the labelled Bio2p was stable in mxrΔ cells (Fig. 8A). These data may be rationalized by the fact that biotin synthases contain a [2Fe-2S] cluster in addition to the oxygen-labile [4Fe-4S] cluster (Jarrett, 2005). Previous data indicate that the additional ~30% of 55Fe incorporated to (and lost from) Bio2p in wild type cells, observed here, could correspond to the labile [4Fe-4S] cluster (Mühlenhoff et al., 2007). If so, then our data suggest that there is no detectable incorporation of 55Fe to the [4Fe-4S] cluster of Bio2p in mxrΔ cells, consistent with a [4Fe-4S]-specific biosynthesis defect in the mxrΔ mutant. However, considering the rapid turnover of 55Fe in Bio2p measured in the wild type, our data also leave open the possibility that the inability to detect 55Fe incorporation to Bio2p of mxrΔ cells could result from a significantly elevated [4Fe-4S] turnover rate (i.e., clusters are turned over at least as fast as they are synthesized).
To help resolve these possibilities, we also performed 55Fe-labeling and immunoprecipitation with Leu1p. Unlike Bio2p, Leu1p contains only a [4Fe-4S] cluster. Moreover, we considered that the cytosolic localization of Leu1p should help to give slower FeS turnover rates than suggested above with mitochondrial Bio2p, mitochondria being the major source of ROS in respiring cells. This was borne out by the data as, unlike with Bio2p (Fig. 8A), no turnover of 55Fe-labeled Leu1p was detected within 1 h of transferring labelled wild type cells to 55Fe-free medium (Fig. 8B). In contrast, mxrΔ cells treated in the same way exhibited a >50% decrease in 55Fe-labeled Leu1p. There was no significant difference in the incorporation of label to Leu1p of mxrΔ and wild type cells. The data collectively indicate a [4Fe-4S] cluster defect in mxrΔ cells which, at least in the case of Leu1p, can be attributed to a higher rate of [4Fe-4S] turnover than in wild type cells.
This study reveals a novel role for the MSR proteins of S. cerevisiae, in helping to preserve the function of cellular FeS clusters. Previous work has established an antioxidant function for the MSR enzymes, methionines being especially susceptible to oxidation compared with other amino acids. Such MSR activity has been proposed to protect against the damaging effects of MetO on protein function, and to support ROS scavenging via reduced Met, and as a mechanism for regulating protein activity (Moskovitz, 2005; Oien & Moskovitz, 2008). However, a link to FeS protein function has not previously been reported. This is an important discovery as the requirement to maintain the integrity of essential FeS proteins is a key challenge to organisms with an aerobic lifestyle (Imlay, 2006). Therefore, the ubiquitous MSR proteins could be an important key facilitating aerobicity.
The finding that deletion of MXR1 and, to a lesser extent, MXR2 both contributed to the Cu resistance phenotype of the mxrΔ double mutant was consistent with the relative contributions of these proteins to other phenotypes in yeast (Koc et al., 2004; Kryukov et al., 2002; Sumner et al., 2005). This indicates that both the S and R stereoisomers of MetO are a factor in the FeS cluster defects that led to Cu resistance. The Cu resistance phenotype of mxrΔ cells was despite an increased level of Cu accumulation. Therefore, Cu must be less available to exert toxicity in the mxrΔ mutant, as borne out by the upregulation of a number of Cu-binding proteins in this strain, in particular Fet3p. FET3 transcription is principally under Aft1p control (Courel et al., 2005; Rutherford et al., 2003; Yamaguchi-Iwai et al., 1995), and the microarray and AFT1-1up expression evidence was consistent with Aft1p dependency here (Fig. 6). The low affinity Cu transporter Fet4p has also been reported to be under Aft1p control (Waters & Eide, 2002) and is upregulated ~2-fold in the mxrΔ mutant (Koc et al., 2004). Therefore, Fet4p could account for this strain’s elevated Cu uptake.
The protection of both mitochondrial (Aco1p, Bio2p) and cytosolic (Leu1p) FeS proteins by MSR activity indicated here suggests that the MSRs localize to both of these subcellular compartments. Mammalian MsrA and MsrB proteins have been localized variously to mitochondria, the cytosol and nucleus (Kim & Gladyshev, 2004). To our knowledge, MSR localization has not been examined directly in yeast. Global protein localization studies (Huh et al., 2003; Kumar et al., 2002) indicate that MsrA (encoded by MXR1) occurs in the cytosol and nucleus (there were no data for yeast MsrB). Moreover, the Predotar tool for identification of N-terminal targeting sequences (Small et al., 2004), gives an 81% probability that the yeast MsrB (encoded by MXR2) localizes to mitochondria (no localization sequence was predicted for MsrA). A mammalian mitochondrion-targeted MsrB protein was sufficient to rescue a yeast mxrΔ double mutant, indicating the importance of mitochondrial MetO reduction (Kim & Gladyshev, 2004). These predictions imply that MsrA may protect FeS cluster integrity primarily in cytosolic and nuclear proteins, and MsrB in mitochondrial FeS proteins [including the FeS-related signal that is exported to Aft1p (Rutherford et al., 2005)]. It was previously questioned (Moskovitz, 2005) what is the dominant role of mitochondrial MSR activity? The present study reveals that preservation of FeS cluster function, at the site of FeS biosynthesis, could be the answer.
One question arising from our study is how does MSR activity protect FeS cluster function? FeS cluster deficiencies can result from depletion of mitochondrial Fe (Li & Kaplan, 2004). However, the possibility that a mxrΔ defect in mitochondrial Fe accumulation is responsible is unlikely as Fe supplementation in the medium up to 3 mM does not alter the relative Cu sensitivities of mxrΔ and wild type strains (unpublished data, T. Sideri and S.V. Avery). Furthermore, our 55Fe labeling studies suggested that the FeS defect in mxrΔ cells is primarily attributable to an enhanced FeS turnover rate rather than decreased FeS biosynthesis, substantiating that Fe availability is not the primary cause.
It is unlikely that the FeS defect is due to some generic antioxidant activity of MSRs, as other antioxidant genes in yeast do not yield a Cu resistance phenotype upon deletion (Avery, 2001; Jo et al., 2008). One exception is SOD2, as corroborated here. Sod2p has a specific role in preserving mitochondrial FeS cluster integrity, by scavenging mitochondrial superoxide radicals that are primary antagonists of FeS function (Irazusta et al., 2006). It has been proposed previously that MSRs may indirectly enhance scavenging of free radicals such as superoxide, by regenerating reduced Met residues that act as antioxidants (Levine et al., 1996; Melkani et al., 2006). However, Met depletion was not responsible for the mxrΔ phenotype as Met supplementation to the medium did not suppress Cu resistance. In contrast, the addition of MetO to wild type cultures mimicked the phenotype, indicating that Met oxidation (rather than depletion) was causative. A number of proteins are reported to be inactivated by Met oxidation (Alamuri & Maier, 2006; Ezraty et al., 2004; Sun et al., 1999). However, FeS proteins are not among those characterized. Furthermore, none of the MSR-interacting proteins that have been identified in bacteria (Alamuri & Maier, 2006) or in protein interaction studies with yeast (http://www.yeastgenome.org/) are FeS proteins. Bacterial MSR is reported to bind proteins with Met contents ~2-fold higher than average (Alamuri & Maier, 2006), whereas the average Met content of yeast FeS proteins is only slightly higher (at 2.6%) than that of all yeast proteins (2.3%).
If the yeast MSRs do not interact directly with FeS proteins, they may modulate the activities of proteins which themselves influence FeS function. Met oxidation can cause conformational changes in proteins, as MetO is more hydrophilic than Met (Moskovitz, 2005). The helix-breaking character of MetO is thought to be exploited in regulation and signalling (Bigelow & Squier, 2005; Ciorba et al., 1997). It should also be noted that MetO may initiate other oxidative reactions that themselves exert damage. For example, oxidation of the Met-35 residue of β-amyloid peptides is thought to be linked to catalysis of free radical production (Pogocki, 2003). Yeast lacking MsrA exhibit increased levels of protein carbonylation (Oien & Moskovitz, 2007), a marker of broader oxidative protein damage than solely Met oxidation (Sumner et al., 2005). Furthermore, a general decrease in mitochondrial ROS production and oxidative damage in calorie-restricted rats has been linked specifically to methionine (and therefore presumably MetO) restriction (Sanz et al., 2006). Although these studies did not examine FeS cluster function, they have highlighted how Met oxidation may trigger a cascade of oxidative events leading to phenotype. Considering that FeS clusters are major targets of superoxide action (Imlay, 2006), as underscored here with Sod2p manipulations, any catalysis of (mitochondrial) superoxide generation that is associated with MetO formation (Pogocki, 2003) could explain the FeS defects in mxrΔ cells.
A relationship between cellular MSR activity and FeS cluster function, revealed here, could be widely conserved. Both of these types of function are ubiquitous among almost all organisms and both have ancient origins (Delaye et al., 2007; Imlay, 2006). Furthermore, defects in both are associated with degeneration and disease. For example, aberrant Fe homeostasis (e.g., due to FeS cluster defects) contributes to the aging process (Atamna et al., 2002), while a role for MSR activity in extending lifespan is conserved across different organisms (Koc et al., 2004). In addition, a major hallmark of Friedreich’s ataxia is decreased activity of FeS enzymes (Shan et al., 2007), and ataxia symptoms have been reported in MsrA-deficient mice (Moskovitz et al., 2001). While such disease states were not a focus of this study, our results have revealed a relationship that contributes to our understanding of ROS-related degeneration and the sustenance of FeS cluster activity in aerobic settings.
This work was supported by a grant from the National Institutes of Health (R01 GM57945). We thank Roland Lill (University of Marburg) for his kind gifts of anti-Bio2p and anti-Leu1p antibodies. We also thank Victor Gladyshev (University of Nebraska) for sharing his mxrΔ double mutant, Jerry Kaplan (University of Utah) for plasmid pRS313-AFT1-1up, and Valeria Culotta (Johns Hopkins University) for plasmid pJS411. We are grateful for expert technical assistance from Lee Shunburne.
This is an author manuscript that has been accepted for publication in Microbiology, copyright Society for General Microbiology, but has not been copy-edited, formatted or proofed. Cite this article as appearing in Microbiology. This version of the manuscript may not be duplicated or reproduced, other than for personal use or within the rule of ‘Fair Use of Copyrighted Materials’ (section 17, Title 17, US Code), without permission from the copyright owner, Society for General Microbiology. The Society for General Microbiology disclaims any responsibility or liability for errors or omissions in this version of the manuscript or in any version derived from it by any other parties. The final copy-edited, published article, which is the version of record, can be found at http://mic.sgmjournals.org, and is freely available without a subscription.