The mitochondrial monothiol glutaredoxin S15 is essential for iron-sulfur protein maturation in Arabidopsis thaliana

Significance

Monothiol glutaredoxins have been considered as components of the iron-sulfur protein assembly machinery but lack of suitable mutants, and partial functional redundancy has hampered the functional analysis in plants. Here, we report the identification of previously unrecognized embryonic lethal Arabidopsis mutants deficient in mitochondrial glutaredoxin S15 (GRXS15). Recombinant GRXS15 coordinates an iron-sulfur cluster in the presence of reduced glutathione as a cofactor. Genetic interference with glutathione binding through targeted mutagenesis diminished the ability of the protein to complement a yeast mutant lacking the homologous mitochondrial Grx5p. Similarly, mutated GRXS15 variants were not able to fully complement the lethal Arabidopsis mutants. Partial complementation resulting in a dwarf phenotype and severely diminished aconitase activity uncovered the role of mitochondrial GRXS15 in iron-sulfur protein maturation.

Abstract

The iron-sulfur cluster (ISC) is an ancient and essential cofactor of many proteins involved in electron transfer and metabolic reactions. In Arabidopsis, three pathways exist for the maturation of iron-sulfur proteins in the cytosol, plastids, and mitochondria. We functionally characterized the role of mitochondrial glutaredoxin S15 (GRXS15) in biogenesis of ISC containing aconitase through a combination of genetic, physiological, and biochemical approaches. Two Arabidopsis T-DNA insertion mutants were identified as null mutants with early embryonic lethal phenotypes that could be rescued by GRXS15. Furthermore, we showed that recombinant GRXS15 is able to coordinate and transfer an ISC and that this coordination depends on reduced glutathione (GSH). We found the Arabidopsis GRXS15 able to complement growth defects based on disturbed ISC protein assembly of a yeast Δgrx5 mutant. Modeling of GRXS15 onto the crystal structures of related nonplant proteins highlighted amino acid residues that after mutation diminished GSH and subsequently ISC coordination, as well as the ability to rescue the yeast mutant. When used for plant complementation, one of these mutant variants, GRXS15_K83/A, led to severe developmental delay and a pronounced decrease in aconitase activity by approximately 65%. These results indicate that mitochondrial GRXS15 is an essential protein in Arabidopsis, required for full activity of iron-sulfur proteins.

Iron-sulfur cluster (ISC) containing proteins conduct essential metabolic processes in all organisms. In plants, autonomous pathways for ISC assembly are present in plastids and mitochondria, whereas ISC biosynthesis and incorporation in cytosolic and nuclear proteins relies on export of bound sulfide from mitochondria (1). Because ISCs are sensitive to superoxide and its reaction products formed by aerobic metabolism, increasing oxidation of the atmosphere led to evolution of sophisticated machineries mediating and controlling the assembly and the transfer of ISCs to acceptor proteins (2). The entire machinery consisting of more than 14 proteins in plastids and 19 proteins in mitochondria includes proteins providing sulfur and iron atoms, scaffold proteins for cluster assembly and transfer proteins that insert ISCs into recipient apoproteins (3). The fundamental role of ISC assembly for building the machineries that are at the center of maintaining life is emphasized by the fact that mutants affecting genes of the ISC assembly pathways are frequently embryo-lethal (3). Among the proteins considered as transfer proteins are representatives of the type II subset of glutaredoxins (GRX), which are characterized by their CGFS monothiol active site motif and their ability to bind glutathione-bridged ISCs (4).

Although evidence exists for the function of GRXs in the assembly of Fe-S proteins in yeast cells and vertebrates (5, 6), their significance in planta is still unclear. Null mutants for plastidic monothiol GRXS14 and GRXS16 are viable, which may be explained by partially overlapping activities (3). Nevertheless, the ability of both proteins to complement the yeast Δgrx5 mutant that lacks the mitochondrial monothiol Grx5p strongly hints to involvement in maturation of Fe-S proteins (7). Similarly, it has been shown that the cytosolic and nuclear GRXS17, both from poplar and Arabidopsis, can also complement the Δgrx5 mutant (7, 8). However, Arabidopsis grxs17 null mutants had only a minor decrease in cytosolic Fe-S enzyme activities, whereas a severe developmental phenotype is visible under elevated temperature and extended daylight. These data indicate that GRXS17 is likely not required for de novo ISC assembly in the cytosol, although it could play a role in cluster repair. Rather, it may be that GRXS17 functions as an oxidoreductase, regulating the function of its partners as BolA2, NF-YC11, or other unidentified targets (8, 9).

However, the involvement of GRXS15 in the maturation of Fe-S proteins in mitochondria remains elusive in plants because, among poplar monothiol GRXs, it is the only isoform failing to rescue most phenotypes of the yeast Δgrx5 mutant (7). Furthermore, the subcellular localization of GRXS15 is ambiguous. Independent targeting experiments have reported GFP-tagged poplar GRXS15 in mitochondria (7) and Arabidopsis GRXS15 in the plastid stroma (10) or dual-targeted plastidic mitochondrial in bifunctional fluorescence complementation (BiFC) experiments with BolA4 (9). In proteome studies, GRXS15 has been repeatedly found in the mitochondria of Arabidopsis (11) and potato (12), but also in the chloroplast proteome of maize (13). The only phenotype of grxs15-null mutants described thus far is sensitivity toward H₂O₂, which led to the suggestion that GRXS15 may be involved in the maintenance of growth and development under oxidative stress conditions (10).

Here, we provide evidence that GRXS15 is an essential component of the mitochondrial ISC machinery that for a long time has been controversial because of the lack of suitable mutants and the failure to complement yeast cells lacking the orthologous gene. The results from in vitro ISC reconstitution and transfer assays and complementation of null mutants in yeast and Arabidopsis with mutated variants of GRXS15 demonstrate that GRXS15 coordinates an ISC and is likely essential for the delivery of ISC to apoproteins in mitochondria, serving in particular for the maturation of ISC into aconitase.

Results

Glutaredoxin GRXS15 Is Localized in Mitochondria.

The Arabidopsis genome encodes 31 GRXs that cluster in three subgroups (Fig. S1). In this gene family, the monothiol GRXS15 could be the only member in mitochondria, but as explained above, its subcellular localization remains controversial. To resolve this uncertainty, the predicted 37-aa target peptide (TP_GRXS15; based on TargetP) and the full-length sequence of GRXS15 were cloned in frame with GFP under control of the 35S promoter and stably expressed in Arabidopsis. In both cases, the GFP signal was exclusively localized in mitochondria (Fig. 1A). This result was further corroborated through protein gel blot analysis in which GRXS15 was only detectable in isolated mitochondria but not in chloroplasts (Fig. 1B). In whole-leaf extracts of wild-type plants, no GRXS15 was detectable consistent with decreased relative abundance of mitochondrial proteins.

Fig. S1.

Classification and subcellular localization of glutaredoxins in Arabidopsis. An alignment of GRX full-length amino acid sequences, which were retrieved from TAIR, was generated with MUSCLE. The evolutionary history was inferred by using the maximum likelihood method based on the JTT matrix-based model. The tree with the highest log likelihood (−3940.9281) is shown. The percentage of trees in which the associated taxa clustered together is shown next to the branches. Initial tree(s) for the heuristic search were obtained automatically by applying neighbor-join and BioNJ algorithms to a matrix of pairwise distances estimated by using a JTT model, and then selecting the topology with superior log likelihood value. A discrete gamma distribution was used to model evolutionary rate differences among sites [five categories (+G, parameter = 4.3811)]. The rate variation model allowed for some sites to be evolutionarily invariable ([+I], 2.7321% sites). The tree is drawn to scale, with branch lengths measured in the number of substitutions per site. The analysis involved 32 amino acid sequences. All positions with less than 95% site coverage were eliminated. Thus, less than 5% alignment gaps, missing data, and ambiguous bases were allowed at any position. There were a total of 92 positions in the final dataset. Evolutionary analyses were conducted in MEGA6. Bootstrap value: 5,000. The localization of the respective GRXs indicated by the color code is based on predictions from SUBA3 (The SUBcellular localization database for Arabidopsis proteins). (Scale bar: 0.2 amino acid substitution per site.)

Fig. 1.

Subcellular localization of GRXS15 in Arabidopsis. (A) Expression of 35S_pro:TP_GRXS15:GFP and 35S_pro:GRXS15:GFP in leaf epidermal cells. GFP, green; MitoTracker, red; chlorophyll autofluorescence, magenta. (Scale bars: 1 µm.) (B) Protein gel blot analysis with antiserum raised against Arabidopsis GRXS15. Fifteen micrograms of protein isolated from whole leaves of a GRXS15 overexpression plant (OE) and a wild-type plant (Wt) as well as proteins of isolated wild-type chloroplasts (C) and wild-type mitochondria (M) were loaded along with 0.5 µg of recombinant protein (RP). Immunoreactivity of the large subunit of RuBisCO (RbcL) served as control for purity of the mitochondrial preparation.

Disruption of GRXS15 Causes Early Embryo Abortion.

To gain insight into the physiological role of GRXS15 in Arabidopsis, three T-DNA lines for the gene locus At3g15660 were isolated. These mutants included two lines with insertions in the 5′-UTR, denominated as grxs15-1 and grxs15-3 as well as a line with an insertion in the first intron of the coding sequence (grxs15-2; Fig. 2A). Left border flanking sequences of the T-DNA insertions in grxs15-1, grxs15-3, and grxs15-2 were sequenced, and the insertion sites were mapped to positions −141, −173, and +322 bp relative to the start codon, respectively. Homozygous grxs15-1 plants were phenotypically indistinguishable from wild type under normal growth conditions (Fig. 2C and Fig. S2). Whereas in wild-type plants the gene is transcribed into two different transcripts annotated as At3g15660.1 and At3g15660.2, grxs15-1 contains only one transcript with a truncated 5′-UTR without obvious changes in transcript abundance (Fig. S2). Because this result implies that grxs15-1 is not a null mutant, it was excluded from further analysis. In contrast, no homozygous grxs15-2 and grxs15-3 mutants were found. In both cases, selfed heterozygous plants segregated in a 1:2 (susceptible:resistant) pattern (Fig. 2B). The mutant grxs15-2 was generated with a T-DNA containing a GUS gene driven by a pollen-specific promoter in the quartet (qrt) background that prevents separation of pollen after meiosis. Here, heterozygous plants always produced pollen tetrads with two GUS-positive pollen, which strongly suggests the absence of a second unlinked T-DNA insertion (Fig. 2D). Analysis of developing seeds from selfed grxs15-2 and grxs15-3 plants revealed frequent abortion (Fig. 2C). Whereas 23% of aborted grxs15-2 seeds were still supporting the hypothesis of a 3:1 (viable:aborted) segregation, abortion in grxs15-3 was slightly less frequent with approximately 20% (Fig. 2B). Interestingly, abortion in grxs15-2 occurred at an early stage after fertilization, whereas the presumed homozygous seeds in grxs15-3 progressed significantly further in development and even initiated endosperm formation. The seeds nevertheless stayed transparent because embryo development was arrested at globular stage (Fig. 2E).

Fig. 2.

Isolation and characterization of grxs15 mutants. (A) Physical map of the GRXS15 gene according to the gene model At3g15660.2. Introns are represented by lines and exons as boxes. Both UTRs are depicted in gray. The primers used for genotyping are indicated by numbered arrows. The T-DNA insertions of the mutant lines grxs15-1, grxs15-2, and grxs15-3 are shown as inverted triangles with the identified insertion site relative to the start codon indicated. Nonnumbered arrows depict the left border (LB) primers used for genotyping and indicate the orientation of the T-DNA. (B) Segregation pattern and seed development of grxs15 T-DNA insertion mutants. χ² values for the progeny genotype are calculated on an expected ratio of 1:2 and for seed development on an expected ratio of 3:1. The degree of freedom is in both cases 2. (C) Opened siliques from wild-type, grxs15-1, grxs15-2, and grxs15-3 plants. (D) GUS labeling of qrt pollen tetrads from a segregated plant homozygous for the wild-type allele GRXS15 (Left) or from a heterozygous grxs15-2 plant (Right). (E) Differential interference contrast image of normal (Left) and transparent (Right) seed from the same silique of a grxs15-3 plant. (Scale bar: 150 µm.)

Fig. S2.

Characterization of grxs15-1. (A) Analysis of the T-DNA insertion in grxs15-1. The T-DNA is inserted 141 bp upstream of the start codon and localized in an intron of the 5′ UTR of At3g15660.2. Numbered arrows depict the primers used for RT-PCR in D. (B) Sequence alignment of the 5′ UTR of the transcripts At3g15660.1 and At3g15660.2. The start codon is indicated by a red ATG. Forward primers used for PCR are underlined. The sequence alignment was performed with MUSCLE. (C) Analysis of the 5′ UTR of both isoforms in wild type and grxs15-1. cDNA of wild type and grxs15-1 was used as a template for a PCR with the primers shown in B and an exon-exon spanning reverse primer (R2) annealing to exon two and exon three in the coding region of GRXS15 depicted in A. Using primer R3 annealing upstream of the intron leads to the amplification of two fragments (395 bp and 318 bp) in wild type (lane 1), but not in grxs15-1 (lane 4), whereas primer R4 and R5 annealing downstream of the T-DNA insertion in both cases generates a PCR fragment of 279 or 356 bp irrespective of whether the primer anneals in the second exon (lanes 2+5) or the intron (lanes 3+6) of the 5′ UTR. (D) Semiquantitative analysis of GRXS15 expression in grxs15-1. For RT-PCR, a forward primer (R1) annealing to the start of the coding region and exon-exon spanning reverse primer (R2) annealing to exon two and exon three in the coding region of GRXS15 were used. PCR was carried out over 34 cycles on cDNA of 3-wk-old wild-type and homozygous grxs15-1 plants with GRXS15 and SAND (At2g28390) specific primers. (E and F) Phenotypic comparison of 4-wk-old plants (D) and 7-wk-old plants (E) grown on soil under long day conditions. (G and H) Complementation of grxs15 by UBQ10_pro:GRXS15. (G) Homozygous grxs15 UBQ10_pro:GRXS15 T₂ plants and the respective wild type were grown for 4 wk on soil, revealing no obvious differences under these growth conditions. (H) PCR analysis of T₂ progeny from three independent homozygous grxs15-2 UBQ10_pro:GRXS15 plants. Using a forward primer binding in the UBQ10_pro and an exon-exon spanning reverse primer annealing to exon 2 and exon 3 in the coding region of GRXS15 consistently results in the anticipated PCR product of 591 bp.

To further confirm that the observed embryo arrest was caused specifically by disruption of GRXS15, both null mutants were complemented with wild-type GRXS15 driven by the UBQ10 promoter through transformation of the female gametophyte. All complemented plants were phenotypically normal, demonstrating the importance of GRXS15 for plant growth and development (Fig. S2G). The viable T₂ progeny of complemented grxs15-2 plants all contain the GRXS15 transgene (Fig. S2H), indicating that a loss of the complementation construct due to segregation causes lethality. These complementation data validate the conclusion that the embryo lethality is due to GRXS15 deficiency.

GRXS15 Partially Complements Δgrx5 Yeast Cells.

Because GRXS15 is the only confirmed GRX in Arabidopsis mitochondria and shares 33% amino acid identity with the yeast mitochondrial monothiol Grx5p (Fig. S3A), yeast Δgrx5 strains displaying distinct growth defects were exploited for functional complementation studies. The original GRXS15 target peptide of 37 aa is sufficient for targeting GFP or a GRXS15-GFP fusion to mitochondria in yeast (Fig. 3A), as in plants. Therefore, the full-length sequence of GRXS15 was used for detailed complementation studies. Multiple complementation experiments with different Δgrx5 deletion strains consistently resulted in partial rescue of Δgrx5 (Fig. 3B and Fig. S3B). Expression of GRXS15 also diminished the sensitivity of Δgrx5 to the oxidative agent diamide but not the respiratory growth defect on glycerol (Fig. 3B). These observations strongly point at a partial functional conservation of GRXS15 between yeast and Arabidopsis.

Fig. S3.

Complementation of the yeast Δgrx5 mutant by Arabidopsis GRXS15. (A) Amino acid alignment of GRXS15 and ScGrx5p (NP_015266). The sequence alignment was performed with MUSCLE. Conserved amino acids are highlighted by a black background. Predicted targeting peptides (GRXS15: TargetP (www.cbs.dtu.dk/services/TargetP/); Grx5p (5): are indicated in red. (B) Complementation of Δgrx5 mutants with different genetic backgrounds (Materials and Methods). Wild type (CML235, W303.1A and BY4742) and the respective Δgrx5 mutant (MML1500, MML100 and YPL059w), transformed with an empty vector, and the Δgrx5 mutant transformed with GRXS15 were spotted onto solid dropout medium containing glucose in fivefold serial dilutions. Colonies were visualized after incubating plates for 2 d at 30 °C.

Fig. 3.

Complementation of the yeast Δgrx5 mutant by Arabidopsis GRXS15. (A) Subcellular localization of TP_GRXS15:GFP or GRXS15:GFP in Saccharomyces cerevisiae cells. Cultures were grown in liquid drop-out medium before incubation with the mitochondrial marker MitoTracker. GFP, green; MitoTracker, red. (Scale bars: 2 µm.) (B) Yeast growth on drop-out medium. Serial fivefold dilutions of wild type (BY4742) and the respective Δgrx5 mutant (YPL059w), transformed with an empty vector, and the Δgrx5 mutant transformed with GPD_pro:GRXS15 were spotted on plates containing glucose (Left) or glucose with 1.25 mM diamide (Right) and grown at 30 °C. No growth was observed on drop-out medium containing glycerol instead of glucose to enforce respiratory growth (Center). One representative experiment from three independently performed experiments is shown.

Recombinant GRXS15 Lacks Oxidoreductase Activity but Binds an Iron-Sulfur Cluster.

GRXS15 has been found among potential thioredoxin targets of plant mitochondria by affinity chromatography (14), suggesting that GRXS15 may act as an oxidoreductase, as proposed for yeast Grx5p (15). Thus, recombinant GRXS15 lacking the first 37 aa was analyzed for potential oxidoreductase activity. In the 2-hydroxyethyl disulfide (HED) assay, reductive activity of GRXS15 was only just above background and less than 3% of the activity observed for the Arabidopsis dithiol GRXC1 used as a reference (Fig. S4A). In an alternative assay with redox-sensitive GFP2 (roGFP2; ref. 16), no catalytic activity could be detected for the reduction of oxidized roGFP2 with reduced glutathione (GSH), whereas a residual catalytic effect was observed for the oxidation of roGFP2 with glutathione disulfide (GSSG). This effect, however, was approximately 30 times lower than catalysis by GRXC1 (Fig. S4 C and D). These results indicate that GRXS15 does not efficiently catalyze the formation or reduction of GSH-mixed disulfides.

Fig. S4.

Biochemical characterization of GRXS15. Reduction of HED and roGFP2 or oxidation of roGFP2 as well as ISC coordination was assayed after purification of recombinant GRXS15 and GRXC1, respectively. (A) NADPH consumption (ΔOD min⁻¹) of GRXS15 mutants and GRXC1 using HED as substrate and measuring the NADPH oxidation by 3 μM GRX. Basal background activities without GRX were subtracted. Means ± SD (n = 4). (B) Relative absorption at 420 nm of GRXS15 mutants. Absorption was measured after in vitro reconstitution. Means ± SD (n = 3). (C and D) roGFP2-interaction assay with GRXC1 and GRXS15. Reduced roGFP2 was mixed with or without 3 µM GRXC1 and GRXS15. Four minutes after start of the measurement, freshly prepared glutathione disulfide (GSSG) solution was added to a final concentration of 40 μM (C). In contrast, 2 mM reduced glutathione (GSH) was used for reduction of the oxidized sensor (B). When working with oxidized roGFP2, a highly negative redox state of the glutathione buffer was maintained by addition of 1 U GR and 100 μM NADPH. Furthermore, H₂O₂ and DTT were used at a final concentration of 10 mM to define maximum oxidation and reduction of roGFP2. Filter-based excitation at 390 and 480 nm and detection of emitted light at 520 nm was followed over time (n = 4). Gain settings for detection of roGFP2 fluorescence after excitation with 390 and 480 nm were always adjusted before each experiment and, thus, the absolute ratio values calculated might differ between experiments.

To test the alternative hypothesis that GRXS15 is involved in supplying ISC to mitochondrial proteins, we first analyzed the capacity of a recombinant GRXS15 to bind an ISC. Escherichia coli cells expressing GRXS15 did not display the strong characteristic brownish color associated with the presence of an ISC in overexpressed proteins. However, when the recombinant protein was purified in the presence of 4 mM GSH, the UV-visible spectrum showed a pronounced shoulder at 420 nm, indicating that ISC coordination by GRXS15 occurred in E. coli. Because GRXS15 ISC incorporation was likely far from completion, we have performed in vitro reconstitution assays under anaerobic conditions by using the purified apoprotein. In this case, the visible part of the absorption spectrum of the reconstituted protein presented a prominent absorbance peak at approximately 420 nm compared with the apoprotein (Fig. 4A), which is characteristic for the presence of an [2Fe-2S]²⁺ cluster or a mixture of [2Fe-2S]²⁺ and linear [3Fe-4S]⁺ clusters (17). Next, we evaluated the capacity of GRXS15 to transfer its ISC to an acceptor protein. Among mitochondrial Fe-S proteins, the mitochondrial ferredoxin 1 (MFDX1), which can bind an [2Fe-2S]²⁺ cluster, was an obvious candidate. After purifying the recombinant protein and stripping the ISC by acidic precipitation, ISC transfer to the apoprotein was monitored over time by visualizing the newly formed holo-MFDX1 on native polyacrylamide gels (Fig. 4C). There was a clear gradual increase of holo-MFDX1. The initial transfer reaction was fast because at the 0 min time-point, which was taken after mixing, spinning, and pipetting times, an appreciable amount of holo-MFDX1 had already formed. However, no further formation of holo-MFDX1 was observed after 45 min although some apo-MFDX1 was still available. The identity of the final product as MFDX1 containing the native [2Fe-2S]²⁺ cofactor was confirmed by CD spectroscopy (Fig. 4D).

Fig. 4.

Reconstitution of ISC in GRXS15 and ISC transfer to ferredoxin. (A) Reconstitution of an ISC in GRXS15. UV-visible spectra of apo- (dashed line) and holo-GRXS15 (straight black line), the K₈₃/A (red), and the C₉₁/S substituted protein (yellow) 2 h after reconstitution or directly after purification from E. coli in the presence of 4 mM GSH (gray). The spectra were normalized to the absorbance at 278 nm. (B) Structure modeling of GRXS15. A homology model was built by using Phyre2 with human mitochondrial GLRX5 as template. Highlighted are the amino acids K₈₃ and K₁₂₀ that may form hydrogen bonds with the carboxyl group of GSH as well as C₉₁ that directly interacts with the ISC. (C) ISC transfer from holo-GRXS15 to apo-MFDX1. The transfer reaction was followed over a time-course of 90 min under anaerobic conditions by mixing holo-GRXS15 with apo-MFDX1 to a ∼1.5:1 ratio in 200-µL reactions. Aliquots of 25 µL were removed at each time-point and separated on a native polyacrylamide gel. Lanes 1–3 correspond to approximately 20 µg of holo-GRXS15 (1), holo-MFDX1 (2), and apo-MFDX1 (3). C, Upper shows an unstained gel displaying the characteristic red-brownish color of holo-MFDX1. In C, Lower, a Coomassie blue-stained gel shows the relative proportions of apo- and holo-MFDX1. (D) CD spectrum comparison. Blue, initial CD spectrum of holo-GRXS15; black, CD spectrum after 60-min incubation of holo-GRXS15 and apo-MFDX1 at a ∼1:1.5 ratio; red, reference CD spectrum for holo-MFDX1.

Diminished GSH Coordination by GRXS15 Limits the Ability To Complement Yeast Δgrx5.

Based on the ability of GRXS15 to coordinate an ISC in vitro, we built a homology model with human GLRX5 as template (Fig. S5A). Monothiol GLRX5 coordinates an ISC, is mitochondria-resident and shares 37% identity with GRXS15. We then compared GRXS15 candidate residues for noncovalent binding of GSH as a prerequisite for ISC coordination to other GRX structures (Fig. 4B). Whereas position K₈₃ is fully conserved in all analyzed GRX structures, positions K₁₂₀ and D₁₄₆ are more variable but usually retain residues that may participate in GSH binding through hydrogen bonding (Fig. S5 B and C).

Fig. S5.

Candidate side chains involved in GSH binding by GRXS15. (A) Overall structural conservation of GRXS15 and human GLRX5. Shown is the model monomer of GRXS15 (blue and green) superimposed with its template GLRX5 (gray). (B) Comparative modeling of GRXS15 based on GLRX5. GSH and ISC are copied from GLRX5 after superimposition with the GRXS15 homology model (Phyre2). Although distances between the GRXS15 model and GSH are therefore not exact, four side chains were chosen as candidates for the involvement in GSH binding (K₈₃; K₁₂₀; K₁₂₄; D₁₄₆). (C) Amino acid alignment of GRXS15 and other ISC binding GRXs core region. The sequence alignment was performed with MUSCLE. Enumeration of the amino acids is based on GRXS15.

The incorporation of an ISC led us to hypothesize that the ability of GRXS15 to complement the Δgrx5 yeast mutant is based on its capacity to coordinate an ISC rather than on its reductase properties. Hence, GRXS15 was mutated to weaken GSH binding and, thus, interfere with ISC coordination. We reasoned that if ISC coordination is a critical function of GRXS15, less-efficient complementation should be expected as a result of the mutations. Based on the modeled structure, several substitutions were carried out to manipulate the direct environment of the putative ISC binding in a targeted manner. The mutations included charge inversions to gain drastic effects and substitutions by alanine that were anticipated to be less severe (Fig. S6A). Mutants lacking the active site cysteine (C₉₁) or carrying a K₈₃/E substitution were no longer able to complement the Δgrx5 mutant, indicating the essential role of both amino acids (Fig. 5A). Substitution K₈₃/A turned out to be less severe and still allowed for residual complementation (Fig. 5A). Indeed, although the residual reducing activity was similar in the mutant variants compared with the native GRXS15, the K₈₃/A mutant was still able to coordinate an ISC, albeit less efficient than the wild type but better than the C₉₁/S mutant (Fig. S4B and Fig. 4A). Mutations of K₁₂₀ generally had only diminutive effects on the ability of GRXS15 to complement Δgrx5. Mutagenesis of K₁₂₄ and D₁₄₆ had no influence on the ability to complement Δgrx5 (Fig. S6 B and C).

Fig. S6.

Substitutions in GRXS15 and rescue of the yeast Δgrx5 mutant defects by mutated GRXS15. (A) Performed substitutions in GRXS15 to influence the ability of coordinating the ISC or to bind GSH by hydrogen bonds. (B) Growth of the yeast Δgrx5 mutant complemented with different versions of mutated GRXS15. Exponentially grown cultures were spotted onto solid drop-out medium containing glucose in fivefold serial dilutions. Colonies were visualized after incubating plates for 2 d at 30 °C. One representative experiment from three independently performed experiments is shown. (C) Growth of the yeast Δgrx5 mutant complemented with GRXS15 substituted in K₁₂₄ or D₁₄₆ in liquid drop-out medium at 28 °C. The absorbance at 600 nm was followed over time.

Fig. 5.

Rescue of the yeast Δgrx5 mutant defects by mutated GRXS15. (A) Growth of the yeast Δgrx5 mutant, complemented with different versions of GRXS15. The absorbance at 600 nm was followed over time. One representative experiment from three independently performed experiments is shown. (B) Normalized ratio of aconitase (ACO)/malate dehydrogenase (MDH) activity. Enzyme activity was measured in total yeast cell extract of complemented Δgrx5 mutant. Mean ± SEM (n = 3), Asterisks indicate statistically significant differences (Student t test: *P ≤ 0.1; **P ≤ 0.05) compared with wild type or GRXS15-complemented Δgrx5 mutant.

To explore the functional impact of GRXS15 on cellular ISC homeostasis and maintenance of ISC proteins, yeast strains expressing the mutant variants in the Δgrx5 background were analyzed for the activity of ISC-containing aconitase (ACO) as a marker. Whereas malate dehydrogenase (MDH), a mitochondrial non-ISC enzyme used as control, showed similar activity in all complemented lines and the wild type, ACO activity strongly depended on the mutations. Less severe mutations led to a minor proportional decrease in ACO/MDH activity ratio, whereas mutant variants that did not rescue Δgrx5 growth showed low ACO activity (Fig. 5B and Fig. S7). The background activity may reflect a combination of residual mitochondrial ACO and cytosolic ACO. This result is in good correlation with the degree of growth rescue of Δgrx5 expressing the different Arabidopsis GRXS15 variants.

Fig. S7.

Enzyme activity of aconitase and malate dehydrogenase. (A) Aconitase activity was measured in cell lysates of exponentially grown yeast cultures at 30 °C. Mean ± SEM (n = 3). A coupled assay was performed in which cis-aconitate was used as a substrate, and the reduction of NADP⁺ was measured. Absorbance at 340 nm was followed over time. (B) Malate dehydrogenase activity was quantified in cell lysates of exponentially grown yeast cultures at 30 °C. Mean ± SEM (n = 3). Oxidation of NADH was followed over time by monitoring the absorbance at 340 nm.

To analyze whether the diminished function of GRXS15 has also an influence on the maturation of Fe-S proteins in the plant context, we complemented the Arabidopsis grxs15-3 line with the mutated GRXS15 variants K₁₂₀/E and K₈₃/A based on the observation that both mutations lead to different degrees of partial complementation of the yeast Δgrx5 mutant. In both cases, we could obtain homozygous plants. The grxs15-3 UBQ10_pro:GRXS15 K₁₂₀/E plants showed no obvious differences in phenotype nor in ACO activity (Fig. S8 A, B, and D) despite differences in the expression level of GRXS15 (Fig. S8C). In contrast, grxs15-3 UBQ10_pro:GRXS15 K₈₃/A plants showed severely reduced growth and reduced ACO activity (Fig. 6). Despite similar amounts of ACO protein detectable in leave extracts, ACO activity was decreased to a similar degree of approximately 35% residual activity in all complemented mutant plants (Fig. 6 B and C). This result may indicate that a low level of functional ACO in mitochondria is sufficient to maintain growth under nonstress conditions. The phenotypic differences between the complementation lines may be explained by GRXS15 transferring the ISC also to other apoproteins.

Fig. S8.

Complementation of grx15-3 with different GRXS15 variants. (A) Growth of T₁ homozygous grxs15-3 UBQ10_pro:GRXS15 K₁₂₀/E plants compared with the respective overexpressing wild type. (B) Enzyme activity in the T₁ grxs15-3 UBQ10_pro:GRXS15 K₈₃/A and K₁₂₀/E plants compared with overexpressing wild type (n = 4–5 ± SEM). (C) Protein expression of transgenic GRXS15 in grxs15-3 UBQ10_pro:GRXS15 K₁₂₀/E plants compared with overexpressing wild type. In Lower, GRXS15 was detected by specific antibodies in total protein extract (15 µg protein was loaded). In Upper, amido black staining of the membrane is shown focusing on the large subunit of RuBisCO as a control for protein loading. (D) Enzyme activity of aconitase and malate dehydrogenase in the T₂ grxs15-3 UBQ10_pro:GRXS15 K₁₂₀/E plants compared with wild type (n = 3 ± SEM).

Fig. 6.

Complementation of grxs15-3 with GRXS15 K₈₃/A. (A) Growth of 7-wk-old homozygous grxs15-3 UBQ10_pro:GRXS15 K₈₃/A T₂ plants compared with the respective wild type. Shown are two plants per complemented line. (Scale bars: 2 cm.) (B) Protein expression of transgenic GRXS15 in grxs15-3 UBQ10_pro:GRXS15 K₈₃/A plants compared with a nontransgenic wild type. In Upper aconitase and in Lower GRXS15 were detected by specific antibodies in total protein extract (15 µg of protein was loaded). In Middle, amido black staining of the membrane is shown focusing on the large subunit of RuBisCO (RbcL) as a control for protein loading. (C) Enzyme activity of aconitase in the grxs15-3 UBQ10_pro:GRXS15 K₈₃/A plants compared with wild type (n = 3 ± SEM).

Discussion

In plant mitochondria, the redox balance is primarily maintained by ascorbate and glutathione acting together in the ascorbate–glutathione cycle (18). Although GSH is synthesized in plastids and in the cytosol, immunocytochemical staining suggested that mitochondria contain particularly high GSH concentrations (19). GRXs are generally considered to mediate reactions involving GSH as a cofactor for ISC coordination or for deglutathionylation/glutathionylation of target proteins (20). The yeast homolog of GRXS15, Grx5p, has been reported to act as a thiol reductase deglutathionylating substrate proteins (15). GRXS15 and the two plastidic GRXs, GRXS14 and GRXS16, are also capable of deglutathionylating the sulfurtransferase SUFE1 in vitro albeit at much lower rate than the plastidic type I GRXs, GRXC5 and GRXS12 (9). Our experiments from different GRX assays, however, suggest that GRXS15 has no reducing activity and, if at all, only a minor oxidation activity. A 20-fold lower activity observed for yeast Grx5p compared with dithiol Grx1 with GSH as electron donor has been explained by inefficient reduction of monothiol GRXs by GSH (15). This result is in line with independent observations showing low or even negligible levels of thiol-disulfide oxidoreductase activity of several monothiol GRXs (21), suggesting that this function became secondary during evolution. However, GRXS17, a cytosolic multidomain GRX, is able to reduce the glutathionylated and the homodimeric form of BolA2, suggesting that the in vivo activity of BolA proteins could be modulated in a redox-dependent manner (9). For GRXS15, an interaction with the BolA domain of SufE1 has been shown by Y2H. This result, however, remains inconclusive because SufE1-YFP was reported to be dual-targeted to plastids and mitochondria when expressed heterologously in tobacco leaves (22) but so far has never been reported in proteome analyses from Arabidopsis mitochondria. Indeed no such interaction was found in BiFC assays in Arabidopsis protoplasts (9). In contrast, interaction of GRXS15 with the dual-targeted BolA4 was found in Y2H and BiFC assays in both mitochondria and chloroplasts (9). In this case, the plastidic interaction might be due to the overexpression of GRXS15, which is routed to chloroplasts via the BolA4 interaction. Little is known about BolAs in plants, but in human cells, the mitochondrial BOLA3 has been suggested to interact with GLRX5 in the maturation of lipoate-containing 2-oxoacid dehydrogenases and assembly of respiratory chain complexes (23). Regarding 2-oxoacid dehydrogenases, lipoic acid synthase (LIAS) catalyzes the lipoylation of the octanoylated apoproteins. LIAS contains a Fe₄S₄ cluster, suggesting that a BOLA3/GLRX5 heterodimer is required for its maturation. Given that Arabidopsis contains also a mitochondrial-targeted lipoic acid synthase (At2g20860), a similar function for GRXS15 appears conceivable.

Among poplar monothiol GRXs, the two plastidic isoforms, GRXS14 and GRXS16, complement the yeast Δgrx5 mutant, whereas poplar GRXS15 was shown to be relatively inefficient and there was no indication that it could incorporate an ISC (7). Triggered by the identification of embryonic lethal Arabidopsis grxs15 null mutants and the assumption that GRXS15 is the only mitochondrial GRX, we investigated the ability of Arabidopsis GRXS15 to complement yeast Δgrx5 mutants. With this isoform, complementation of different Δgrx5 yeast strains worked, although we never achieved a 100% complementation. Interestingly, the increase of the ACO/MDH activity for GRXS15 complemented Δgrx5 (∼0.59) compared with the empty vector control (∼0.17) was higher than the normalized ratio for Δgrx5 complemented with poplar GRXS15 (∼0.26; ref. 7). The reason for only partial complementation is unknown, but one can speculate about differences between yeast and Arabidopsis proteins leading to less stable ISC coordination or partially impaired protein–protein interactions as a result of the coevolution between the interacting proteins within a given organism. For instance, compared with yeast Grx5p, plant GRXS15 isoforms have a clear N-terminal extension containing many conserved Asp residues, whereas they have a slightly shorter C-terminal part missing charged residues found in Grx5p. Considering that the C-terminal part of monothiol GRXs was shown to be responsible for protein–protein interaction (24), variations in this protein region might be crucial. Another important difference is that yeast Grx5p contains an additional cysteine residue (C₉₀ or C₁₁₇ including the targeting sequence) implicated in intramolecular disulfide exchange reactions (15). Whereas genetic analyses indicated that this cysteine does not seem essential for ISC biogenesis (25), Zhang et al. showed in mutational studies that this cysteine is required for the assembly of Fe₄S₄ cluster and that the Fe₄S₄ cluster-bound form of Grx5p is competent for restoring the activity of recombinant ACO in vitro (17). Although this cysteine is present in most algal isoforms, it is replaced by a serine in orthologs from terrestrial plants, except Selaginella moellendorffii (26). The ability to complement the Δgrx5 mutant likely implies that GRXS15 is involved in ISC biosynthesis or transfer in Arabidopsis mitochondria. Reconstitution assays confirmed the incorporation of an ISC and the resulting A₄₂₀/A₂₈₀ ratio of 0.36 was in a similar range as for other monothiol GRXs coordinating ISCs [GRXS14: 0.31 ± 0.04 (7) and GRXS17: 0.29 (8)]. The ability to coordinate ISCs and the ability to transfer the ISC to MFDX1 in vitro shown here as well as the interaction with the transfer protein ISA (27) all point to a role of GRXS15 in ISC transfer similar to yeast Grx5p. This interpretation is strongly supported by the fact that mutations in GRXS15 interfering with the binding of the GSH cofactor required for ISC coordination impair the ability to complement the Δgrx5 yeast mutant. Because the role of GRXS15 seems restricted to ISC maturation, future work needs to explain how the interaction that has been shown to occur efficiently between the glutathione pool and thiol proteins is mediated (28, 29).

In contrast to Δgrx5 yeast cells, loss of GRXS15 in Arabidopsis causes embryonic lethality. Viability of Δgrx5 is maintained by the mitochondrial dithiol Grx2p as a backup, and Δgrx2Δgrx5 mutants are synthetic lethal (30). In zebrafish, where Grx2 seems to be mainly localized in the cytosol, deficiency of Grx5 is lethal between 7 and 10 d after fertilization (6, 31). Taken together, the data presented in this paper conclusively show that GRXS15 is essential in plant mitochondria. Furthermore, the knockout of other genes that are essential in the mitochondrial ISC machinery like frataxin or IBA57 results in an embryo-lethal phenotype (32, 33). Despite the necessity of GRXS15 during the vegetative phase, the segregation patterns indicate that GRXS15 is less critical during fertilization. This phenomenon may be explained by residual ISC assembled before meiosis or by the fact that energy metabolism in pollen tubes can undergo fast rearrangements shifting from aerobic respiration to ethanol fermentation (34). Pollen tube growth has been shown to continue even under anaerobic conditions or on inhibitors of the respiratory electron transport or ATP synthase albeit at lower speed (34, 35). This effect may explain the slight deviations from Mendelian segregation observed for selfed grxs15-3 mutants.

In summary, this work expands the role of GRXs in the biogenesis of Fe-S proteins by identifying GRXS15 as an essential component of the mitochondrial ISC assembly machinery in Arabidopsis. The isolation of null mutants and their characterization as being embryonic lethal solves a long-standing mystery about the role of GRXS15.

Materials and Methods

See SI Materials and Methods and Tables S1 and S2 for biological materials, growth conditions, genetic analysis, and cloning details.

Table S1.

Oligonucleotides used in this study

Name	Oligonucleotide sequence 5′ → 3′
Genotyping
P1	CACAGAGCCTAACGCCAATAG
P2	CGGACGTATACTTTGGTGACC
P3	TGAAGCATACTTTTGGGATGG
P4	ATTCAAAACCATACGCTCACG
P5	GGAGATTCAGGGACACCTTTC
P6	ATGGTCCACTTCGTATGTTGG
LB (grxs15-1)	GACCGCTTGCTGCAACTCTCTCAGG
LB (grxs15-2)	ATTTTGCCGATTTCGGAAC
LB (grxs15-3)	CCCATTTGGACGTGAATGTAGACAC
Gateway cloning (attB sites are underlined)
GRXS151–169_forward	GGGGACAAGTTTGTACAAAAAAGCAGGCTTTATGGCGGCTTCTTTATCGAGC
GRXS15_reverse	GGGGACCACTTTGTACAAGAAAGCTGGGTCTCAATCTTGGTTTCCGGAGAC
GRXS15_reverse	GGGGACCACTTTGTACAAGAAAGCTGGGTCATCTTGGTTTCCGGAGACGTC
GFP_reverse	GGGGACCACTTTGTACAAGAAAGCTGGGTCTTACTTGTACAGCTCGTCCATG
GRXS1538_169_forward	GGGGACAAGTTTGTACAAAAAAGCAGGCTTTTCAACAGTGCCAAGTGATTCAG
Site-directed mutagenesis (forward/reverse)
GRXS15 K83/A	ATCTACATGGCTGGTGTCCCT/AGGGACACCAGCCATGTAGAT
GRXS15 K83/E	ATCTACATGGAAGGTGTCCCT/AGGGACACCTTCCATGTAGAT
GRXS15 K120/A	CAAGAGTTGGCTAACGCTGTG/CACAGCGTTAGCCAACTCTTG
GRXS15 K120/E	CAAGAGTTGGAAAACGCTGTG/CACAGCGTTTTCCAACTCTTG
GRXS15 K124/A	AACGCTGTGGCTTCCTTCAGC/GCTGAAGGAAGCCACAGCGTT
GRXS15 K124/E	AACGCTGTGGAATCCTTCAGC/GCTGAAGGATTCCACAGCGTT
GRXS15 D146/A	GGCGGCTCAGCTATCATCCTT/AAGGATGATAGCTGAGCCGCC
GRXS15 D146/R	GGCGGCTCAAGAATCATCCTT/AAGGATGATTCTTGAGCCGCC
GRXS15 C91/S	CCTGAATCTCCTCAGTCTGGGTTTAGCTCACT/AGTGAGCTAAACCCAGACTGAGGAGATTCAGG
Others
MFDX169_197 forward	CCCCCCCCCATATGTCCTCTGAGAATGGTGAT
MFDX1 reverse	CCCCGGATCCCTAATGAGGTTTTGGAACAAACCC

Table S2.

Yeast strains used in this study

Strain	Relevant genotype	Comment
CML235	MATa ura3-52 leu2Δ1 his3Δ200	Wild type (30); spore from FY1679 (S288C derivative strain)
MML1500	MATa grx5:: kanMX4	Deletion of Grx5p in CML235
W303.1A	MATa ura3-1 ade2-1 leu2-3,112 trp1-1 his3-11,15	Wild type (5), W303 derivative strain
MML100	MATa grx5:: kanMX4	Deletion of Grx5p in W303-1A (5)
BY4742	MATα ; his3Δ1; leu2Δ0; lys2Δ0; ura3Δ0	Wild type, S288C derivative strain, EUROSCARF
YPL059w	MATa; grx5:: kanMX4	Deletion of Grx5p in BY4742; EUROSCARF

Microscopy.

Subcellular localization of GRXS15 was observed after GFP tagging by confocal microscopy on a Zeiss LSM780 with instruments settings detailed in SI Materials and Methods.

Protein Purification.

Recombinant proteins were expressed and purified as described (16). Purification of GRXS15 was performed under aerobic conditions with or without 4 mM GSH. The protein content was quantified by the Bradford assay with BSA as standard. For all other proteins, details are described in SI Materials and Methods.

ISC Reconstitution and Transfer.

The ISC reconstitution assay was performed as described (36). UV-VIS spectra were recorded by using a plate reader (CLARIOstar; BMG). Transfer of ISC from GRXS15 to MFDX1 was followed by native PAGE and CD spectrometry. Full details are described in SI Materials and Methods.

Enzyme Assays.

All assays were carried out at 25 °C on a plate reader (POLARstar Omega; BMG). Interaction of roGFP2 with GRXS15 or GRXC1 from Arabidopsis was performed as described (37) by using 1 µM roGFP2 as well as 3 µM GRXS15 or GRXC1. For all other assays, details are described in SI Materials and Methods.

SI Materials and Methods

Plant Material and Growth Conditions.

All plant lines used in the study were in the A. thaliana Columbia background. Seeds of grxs15-1 (SALK_112767C) and grxs15-3 (GK-837C05) T-DNA insertion alleles were obtained from Nottingham Arabidopsis Stock Centre. Seeds of the grxs15-2 allele (SAIL_431_H03) were kindly provided by Ninghui Cheng, Baylor College of Medicine, Houston (10). Genotyping of the respective T-DNA mutants was performed by using a left border (LB) primer specific for each mutant line in combination with gene specific primers (P1-6) (grxs15-1: LB+P6 and P5+P6; grxs15-2: LB+P2 and P1+P2; grxs15-3: LB+P4 and P3+P4). The location and orientation of the primers are shown in Fig. 2A, and primer sequences are given in Table S1.

Arabidopsis plants were grown on a soil:sand:vermiculite mixture in the ratio 10:1:1 and kept in controlled growth chambers under long day conditions with a diurnal cycle of 16-h light at 22 °C and 8-h dark at 18 °C. The light intensity was 75 μE⋅m⁻²⋅s⁻¹ and 50% air humidity. For growing plants on agar plates, seeds were sterilized with 70% (vol/vol) ethanol and put on nutrient medium [5 mM KNO₃, 2.5 mM KH₂PO₄ pH 5.6, 2 mM MgSO₄, 2 mM Ca(NO₃)₂, 50 μM Fe-EDTA, 0.1% (vol/vol) micronutrient mix, 0.8% (wt/vol) Phytagel, pH 5.8 (5)] and antibiotics when required for selection.

Plant Methods.

Genomic DNA was extracted according to the method described by Edwards et al. (38).

RNA isolation was performed by using the Total RNA isolation kit (Macherey-Nagel) following the manufacturer’s protocol. The RNA was eluted in 50 μL of dH₂O and quantified with the spectrophotometer NanoDrop 2000. For reverse transcription of mRNA to cDNA, the Ambion M-MLV Reverse Transcriptase was used following the manufacturer’s protocol.

Defective seed development was noted by dissecting siliques of self-pollinated plants and counting the number of normal and aborted seeds present in each silique.

To determine the phenotypes of embryos, whole siliques were destained with Hoyer’s solution (7.5 g of gum arabic, 100 g of chloral hydrate, 5 mL of glycerol, 60 mL of water) overnight. Siliques were analyzed with a stereomicroscope (Leica M165 FC) equipped with a camera (DFC 425 C) by using the software LAS V3.8 (Leica Application Suite).

Histochemical staining of pollen was performed for the confirmation of grxs15-2 mutants by exploiting the presence of a Lat52 promoter–β-glucuronidase fusion on the T-DNA pCSA110 used for transformation. The pollen were placed in a solution containing 10 μM X-Gluc (Duchefa Biochemie), 50 mM potassium ferricyanide, and 50 mM potassium ferrocyanide in 100 mM sodium phosphate buffer pH 7.0 and incubated in the dark at 37 °C overnight. Subsequently, pollen were analyzed by bright field microscopy (Axio Observer.Z1; Carl Zeiss Microscopy) equipped with a 20× lens (LD Plan-Neofluar 20×/0.4 Korr, Carl Zeiss Microscopy) and a camera (Zeiss AxioCam MRc) using the software Palm Robo V4.5.

Plant transformation was done by floral dip according to established protocols (39) with Agrobacterium solutions containing 5% (wt/vol) sucrose and 0.02% Silwet L-77 as surfactant. For complementation of lethal mutants, heterozygous plants were transformed and allowed for subsequent self-fertilization. Seeds were harvested and putative transformants were selected by using hygromycin B as a marker. Complemented homozygous null mutants were identified by PCR genotyping.

Plasmids, Cloning, and Site-Directed Mutagenesis.

For subcellular localization of GRXS15:GFP in Arabidopsis, full-length GRXS15 was cloned in frame with GFP via Gateway Cloning (Invitrogen) by using the pSS01 vector (40). For expression of the TP_GRXS15:GFP fusion protein, the nucleotide sequence of the first 37 aa were fused to GFP by an overlapping fusion PCR and finally cloned in the pB7WG2.0 vector (41). Substitutions of specific base pairs in GRXS15 were introduced by site-directed mutagenesis. Complementary oligonucleotide pairs harboring the desired nucleotide exchanges were used (Table S1). The mutated gene variant was subsequently cloned via Gateway cloning in the pAG415GPD-ccdB vector (Susan Lindquist; Addgene plasmid 14146) for expression in yeast or in the vector pSS02 (derivative of pMDC32; ref. 42) for expression in planta, respectively. For expression of recombinant proteins, the nucleotide sequences of GRXS15 and mitochondrial ferredoxin 1 (MFDX1), lacking the target sequences of 37 and 69 aa, respectively, were cloned in the pETG-10a vector (EMBL Heidelberg) and in pET12a vector (Novagen) with primers listed in Table S1.

Subcellular Localization and Fluorescence Microscopy.

Fluorescent plants and yeast colonies were detected on a stereomicroscope (Leica M165 FC) equipped with a GFP filter and documented with an attached camera (Leica DFC425 C). For colocalization in mitochondria, samples were incubated with 0.5 μM MitoTracker Orange CM-H₂TMRos for at least 15 min. Yeast cells were transferred to a slide and immobilized with 0.1% (wt/vol) agarose. A confocal laser scanning microscope (Zeiss LSM 780, attached to an Axio Observer.Z1; Carl Zeiss Microscopy) was used for confocal imaging. Images were collected with a 40× (C-Apochromat 40×/1.2 W Korr) or a 63× lens (Plan-Apochromat 63×/1.40 Oil DIC). GFP was excited at 488 nm and MitoTracker at 543 nm. GFP fluorescence was collected with a 505–530 nm bandpass filter and MitoTracker fluorescence with a bandpass filter of 560–620 nm. Chlorophyll autofluorescence was excited with 488 nm and detected with a bandpass filter of 647–745 nm.

Purification of Mitochondria and Plastids.

Mitochondria were purified as described (43) with slight modifications. After homogenization of the seedlings, the homogenate was filtered (Miracloth; Merck Millipore) and cellular debris was pelleted by centrifugation for 5 min at 1,100 × g. The supernatant was centrifuged for 20 min at 18,000 × g, and the pellet of crude mitochondria was resuspended in wash buffer (0.3 M sucrose, 10 mM TES, pH 7.5) and centrifuged for 5 min at 1,100 × g. The supernatant was transferred into a new tube and centrifuged for 20 min at 18,000 × g. The pellet was resuspended in wash buffer and loaded directly on a 0–6% Percoll gradient and centrifuged for 40 min at 40,000 × g.

For isolation of plastids, the same extraction buffer was used. After homogenization of the seedlings, the homogenate was filtered (Miracloth; Merck Millipore) and cellular debris was pelleted by centrifugation for 5 min at 1,100 × g. The pellet was resuspended in wash buffer (0.33 M sorbitol, 20 mM tricine, 2.5 mM EDTA, 5 mM MgCl₂, pH 7.6). After centrifugation for 1 min at 2,500 × g, the plastid fraction was loaded on a step gradient of 50% and 80% Percoll and centrifuged for 10 min at 2,500 × g. Chloroplasts were transferred into a new tube and washed once with wash buffer.

Protein Methods.

Total proteins for protein blot analysis were isolated from Arabidopsis leaves as described (44). For protein blot analysis, total cell extract or purified organelles were heated for 5 min and separated on standard SDS/PAGE gels. Proteins were transferred to a membrane (BioTrace PVDF Transfer Membrane; Pall Corporation) and labeled with polyclonal antibodies raised in rabbit against the GRXS15 recombinant protein. Immunolabeling was detected by chemiluminescence by using secondary horseradish peroxidase-conjugated antibodies and Pierce ECL Western Blotting Substrate.

ISC Transfer Experiments.

Reconstitution of Arabidopsis GRXS15 was accomplished in a glove box under anaerobic conditions. One hundred micromolar apo-GRXS15 was first reduced at room temperature for 3 h in the presence of 10 mM DTT and 0.5 mM GSH. Then, reconstitution was achieved in 2 mL of 50 mM Tris⋅HCl pH 8.0, 50 mM NaCl with 5 mM DTT, 200 µM GSH, 400 µM ferrous ammonium citrate, and 400 µM lithium sulfide (Li₂S). Fe-citrate was added before Li₂S for approximately 5 min until the protein solution becomes pink/red. Then, Li₂S was added and the entire mixture was incubated for 1 h. Reagents in excess were removed from holo-GRXS15 by desalting on PD10 columns equilibrated with 50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, and 10% (vol/vol) glycerol. The protein was finally concentrated by ultrafiltration dialysis.

Apo-MFDX1 was prepared by precipitating the holoprotein (approximately 1 mg/mL) with 0.5 M HCl in the presence of 1% β-mercaptoethanol at room temperature under anaerobic conditions (45). After centrifugation, the pellet was quickly rinsed with acidified water, resuspended in a small volume of 100 mM Tris⋅HCl pH 9.0, 50 mM NaCl, and finally resolved in 100 mM Tris⋅HCl pH 8.0 and 50 mM NaCl containing 1 mM DTT. Then, the ISC transfer reaction from holo-GRXS15 to apo-MFDX1 was assessed by native PAGE and CD spectrometry. In the first assay, the transfer reaction was achieved at 30 °C under anaerobic conditions by mixing in 200 µL of 50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, 10% (vol/vol) glycerol, 1 mM DTT, 70 µM apo-MFDX1 with 100 µM holo-GRXS15, and documented on the basis of Fe-S cluster content in MFDX1. The reaction was started by adding holo-GRXS15. The time-course of ISC transfer from [2Fe-2S]-loaded GRXS15 to apo-MFDX1 was monitored by taking 25 µL of the reaction mixture at 0, 5, 15, 30, 45, 60, and 90 min before adding 1 µL of 0.2% bromophenol blue, 25 mM EDTA, and keeping the tubes on ice. The presence of reconstituted holo-MFDX1 was assessed by separating all proteins on 17.5% (wt/vol) native polyacrylamide gels (46). Alternatively, the transfer reaction was achieved at 30 °C under stirring by mixing 75 µM apo-MFDX1 with 50 µM holo-GRXS15 on the basis of Fe-S cluster content in a volume of 300 µL of 50 mM Tris⋅HCl pH 8.0, 150 mM NaCl, 5% (vol/vol) glycerol, and 1 mM DTT in 3-mm path-length quartz cuvettes. Kinetics of ISC formation on MFDX1 was followed over 60 min in a J-815 CD spectrometer (JASCO) at 437 nm where the difference of the CD spectra of holo-GRXS15 and holo-MFDX1 is maximal. After 60 min, a UV-Vis CD spectrum from 350 to 700 nm was recorded. Spectra were analyzed by GraphPad Prism.

Enzyme Assays.

Enzyme assay extracts from yeast cells growing exponentially in SD medium were prepared in 0.1 M Hepes buffer, pH 7.8, 0.1% ascorbate, 0.05% β-mercaptoethanol, 10 mM EDTA, 0.01% Triton X-100 by using glass beads to break the cells. Arabidopsis leaves were homogenized in the extraction buffer [50 mM Tris⋅HCl 8.0, 50 mM KCl, 0.2% (vol/vol) Triton X-100, 2 mM sodium citrate, 1 mM DTT] and centrifuged for 5 min at 4 °C. Aconitase activity was analyzed in a coupled assay measuring NADPH formation by monitoring the increase in absorbance at 340 nm. The reaction mixture contained 50 mM Hepes pH 7.8, 2.5 mM NADP⁺, 5 mM MnCl₂ and 0.05 U isocitrate dehydrogenase. The mixture was allowed to come to equilibrium after addition of yeast extract. The reaction was started by adding 8 mM cis-aconitic acid. For measuring the malate dehydrogenase activity, the rate of change in A₃₄₀ was monitored in the following mixture: 100 mM Hepes pH 7.8, 0.5 mM NADH, 5 mM MgCl₂, 0.65% Triton X-100. The reaction was started by the addition of 750 µM oxaloacetic acid.

Reduction of HED was measured as the change in A₃₄₀ in the following mixture: potassium phosphate buffer containing 0.5 mM HED, 0.1 U GR, 500 μM NADPH, and 3 μM GRX. GSH (in 100 mM phosphate buffer, pH 7.0) was automatically injected to a final concentration of 0.5 mM by using the built-in injector.

Modeling of GRXS15 Based on GLRX5.

A homology model of Arabidopsis GRXS15 was built by using Phyre2 and human mitochondrial monothiol GLRX5 (PDB ID code: 2WUL) as a template. The coordinates of GSH and ISC were copied into the GRXS15 model after superimposition with GLRX5. Candidate side chains stabilizing GSH within GRXS15 and their existence/biological relevance in other GRXs were compared with AtGRXS14 (3IPZ), PtGRXS14 (2LKU), HsGLRX3 (3ZYW), HsGLRX5 (2WUL), ScGrx5p (3GX8), TbGrx1 (2LTK), EcGrx4 (1YKA), AtGRXC5 (3RHB), PtGRXC1 (1Z7P), PtGRXS12 (3FZA), HsGLRX2 (2HT9), ScGrx1 (3C1R), ScGrx2 (3CTF), ScGrx6 (3L4N), and EcGrx3 (3GRX).

Functional Complementation Assays in S. cerevisiae.

YPG medium [1% yeast extract, 2% (wt/vol) peptone, 2% (wt/vol) glucose] was usually used. Synthetic SD medium contains 0.67% yeast nitrogen base, the auxotrophic requirements, and 2% (wt/vol) glucose or 2% (wt/vol) glycerol, respectively. The yeast transformation of different strains detailed in Table S2 was performed by using the lithium acetate/single-stranded carrier DNA/PEG method following the protocol from ref. 47. Sensitivity to oxidants was determined on SD plates containing 1.25 mM diamide by spotting 1:5 serial dilutions of exponential cultures and recording growth after 2 d of incubation at 30 °C. Growth rates were measured in a volume of 260 µL at 28 °C on a plate reader (POLARstar Omega) monitoring the increase in absorbance at 600 nm.

Accession Numbers.

GRXC1 (At5g63030), GRXC2 (At5g40370), GRXC3 (At1g77370), GRXC4 (At5g20500), GRXC5 (At4g28730), GRXC6 (ROXY21; At4g33040), GRXC7 (ROXY1; At3g02000), GRXC8 (ROXY2; At5g14070), GRXC9 (ROXY19; At1g28480), GRXC10 (ROXY20; At5g11930), GRXC11 (ROXY4; At3g62950), GRXC12 (ROXY5; At2g47870), GRXC13 (ROXY9; At2g47880), GRXC14 (ROXY8; At3g62960), GRXS1 (ROXY16; At1g03020), GRXS2 (ROXY10; At5g18600), GRXS3 (ROXY11; At4g15700), GRXS4 (ROXY13; At4g15680), GRXS5 (ROXY12; At4g15690), GRXS6 (ROXY17; At3g62930), GRXS7 (ROXY14; At4g15670), GRXS8 (ROXY15; At4g15660), GRXS9 (ROXY7; At2g30540), GRXS10 (ROXY3; At3g21460), GRXS11 (ROXY6; At1g06830), GRXS12 (At2g20270), GRXS13 (ROXY18; At1g03850), GRXS14 (GRXcp, CXIP1, At3g54900), GRXS15 (GRX4, At3g15660), GRXS16 (GRX2, CXIP2, At2g38270), GRXS17 (At4g04950), ScGrx5p (NP_015266), MFDX1 (At4g05450).