sus1 and sus2 are mutations that change one amino acid in HSCA2 and ISU1

To better understand the molecular mechanism of SSR1 in affecting root development, the loss-of-function mutant ssr1-2 (FLAG_356A08) was mutagenized by ethyl methanesulfonate and the M2 progenies were used for screening suppressors of the short-root phenotype of ssr1-2. From over a dozen suppressors, two lines, designated as ssr1-2 sus1 and ssr1-2 sus2 (sus: suppressor of ssr1-2), have the best recovered root length (about 85% and 100%, respectively) at seedling stage as well as a complete rescue of the dwarf phenotype at the flowering stage (Fig 1A and 1D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Identification of ssr1-2 suppressor mutant genes sus1 (HSCA2^G87D) and sus2 (ISU1^T55M).

Two suppressors (sus1 and sus2) were identified to rescue the growth defect of ssr1-2 and the whole genome resequencing identified sus1 as HSCA2^G87D while sus2 as ISU1^T55M. (A) sus1 and sus2 rescued both the root and shoot growth defects of ssr1-2 to the levels comparable to wild type (WS: Wassilewskija). Top, 10-day-old seedlings; middle, 20-day-old seedlings; bottom, 40-day-old plants. Scale bar = 1 cm. (B) sus1 and sus2 rescued the expression defects in ssr1-2 of auxin-related marker genes PIN1::PIN1GFP, PIN2::PIN2GFP, WOX5::WOX5GFP, and DR5::GFP. These genes were labelled as PIN1, PIN2, WOX5, and DR5, respectively, and the levels of expression were found to be comparable to those observed in WS. (C) sus1 and sus2 mutant genes were cloned, designated as HSCA2^G87D and ISU1^T55M, respectively, and re-transformed back to ssr1-2 plants to rescue the short root phenotype. Top, images of meristem zone of different lines at 10-day-old. Middle, elongation zone length, bottom, meristem zone cell number. In transgenic lines, the transgenes are associated with a hygromycin resistant gene and therefore designated as Hyg-HSCA2^G87D or Hyg-ISU1^T55M. It should be noted that ssr1-2 Hyg-HSCA2^G87D still contains the wild type HSCA2 allele and ssr1-2 Hyg-ISU1^T55M contains wild type ISU1 allele. (D) Primary root length of different lines grown for 10 days. Lines analyzed are the same as in 1C. Scale bars in 1B and 1C are 50 μm. Yellow arrows point to the end of meristem zone in top images of 1C. In 1C and 1D, 15 and 100 roots of each were used for statistical analysis by ANOVA. Statistical value is shown as box & whiskers with 5–95 percentile. Asterisk “*”, “**” and “***” represent statistics analysis with p values < 0.05, 0.01, 0.001 respectively. “ns” represents statistically not significant.

https://doi.org/10.1371/journal.pgen.1011597.g001

It has been previously reported that several auxin-related markers were abnormally expressed in ssr1-2 seedlings [29]. To examine whether sus1 and sus2 also rescued the abnormal expression of some auxin-related genes, PIN1-GFP, PIN2-GFP, Wox5-GFP and DR5-GFP marker lines in ssr1-2 background were crossed to double mutants ssr1-2 sus1 and ssr1-2 sus2. The resulting F3 progenies which are homozygous for both sus and marker genes were identified. Confocal microscopic analyses showed that the expression patterns of all the four auxin-related genes in double mutants were restored and resembled those observed in the wild-type background (Fig 1B). Correspondingly, the defect of ssr1-2 in meristem cell number and elongation zone observed in the ssr1-2 mutant was also restored by sus1 and sus2 (Fig 1C). These data indicate that sus1 and sus2 are two suppressor mutants that could indeed rescue the defects of ssr1-2 and therefore were chosen for further analysis.

To investigate the genetic nature of sus1 and sus2, we conducted a cross-breeding in which ssr1-2 sus1 and ssr1-2 sus2 were crossed with ssr1-2, respectively. All ssr1-2 sus1/SUS1 F1 seedlings exhibited similar root length to ssr1-2 sus1 suggesting that sus1 may be a dominant mutation (S1A Fig). In contrast, the primary roots of ssr1-2 sus2/SUS2 F1 seedlings were shorter than that of ssr1-2 sus2, indicating that sus2 may be a semi-dominant mutation (S1A Fig). Further analysis of self-pollinated F2 progenies of ssr1-2 sus1/SUS1 and ssr1-2 sus2/SUS2 confirmed our speculations, with clear 3:1 and 1:2:1 segregation, respectively, in terms of the root lengths (S1B and S1C Fig and S1 Table).

A total of five candidate genes were selected from each suppressor line based on super bulked segregant analysis (S2 and S3 Tables). These genes were then verified by genetic complementation. Since sus1 and sus2 are dominant and semi-dominant mutations, respectively, the complementation analysis was performed by re-introducing the full-length genomic fragments of candidate genes from the suppressor lines into ssr1-2. It turned out that the reintroduction of At5g09590 from ssr1-2 sus1 and At4g22220 from ssr1-2 sus2 was the only means of rescuing the short root phenotype and the defects of apical meristem cell number and elongation zone observed in ssr1-2 (Fig 1C and 1D), thus confirming that At5g09590 and At4g22220 are SUS1 and SUS2, respectively. At5g09590 and At4g22220 encode a mitochondrial heat shock cognate 70 (mtHSC70-2 or HSCA2) and an ISU1 protein, respectively. Both proteins are core components of the ISC machinery involved in the assembly of Fe-S clusters in mitochondria [2,22]. sus1 and sus2 each bears a point substitution mutation in the coding region of the corresponding HSCA2 and ISU1 genes. sus1 carries a G to A transition at position 260, resulting in an HSCA2^G87D mutant protein. sus2 carries a C to T transition at position 164, resulting in an ISU1^T55M mutant protein.

With the identification of the two suppressor genes which cause amino acid substitution in HSCA2 (HSCA2^G87D) and ISU1 (ISU1^T55M), we took a different approach to analyze the other candidate suppressors obtained from our initial screen. HSCA2 and ISU1 genes from other suppressor lines were all cloned and re-sequenced. Interestingly, only sus3 does not carry mutations in HSCA2 or ISU1. The additional suppressors carry single point mutations either in HSCA2 or ISU1, resulting in ISU1^A143V, ISU1^G106D, ISU1^A143T, HSCA2^R394C and ISU1^A140V, respectively (S4 Table). Similar to sus1 and sus2, the suppressor function of ISU1^G106D and ISU1^A143T were investigated and verified by genetic complementation (S2A Fig).

HSCA1 is a homolog of HSCA2, and also located in mitochondria [2]. The hsca1 mutant was recently reported to have a deficiency in embryo development and root growth as well [31,32]. To test whether HCSA1 performs a similar function to HSCA2 in root growth, a mutant form of HSCA1, designated HSCA1^G82D, was created and introduced into ssr1-2 mutants under the control of its native promoter. This mutant corresponds to HSCA2^G87D.The short-root phenotype of ssr1-2 seedlings was substantially suppressed by the expression of HSCA1^G82D (S2B Fig), suggesting that HSCA1 may have overlapping function with that of HSCA2 in vivo. Furthermore, we obtained a knock-out mutant of HSCA2, hsca2 (CS479451 from ABRC), and observed that hsca2 seedlings did not exhibit any significant phenotype regarding primary root length (S2C Fig). All these data indicate HSCA1 and HSCA2 are functionally redundant in root development, and HSCA1 likely plays a dominant role.

We would like to understand whether sus1 or sus2 can promote root growth in the wild-type background. To achieve this, we isolated single mutants of sus1 and sus2 after crossing ssr1-2 sus1 and ssr1-2 sus2 with the wild-type Wassilewskija (WS). Then the root length of these mutants was investigated. Under the optimal growth conditions, the root length of the sus1 and sus2 single mutants was comparable to that of the wild type (S2D Fig), implying that the identified suppressors specifically rescued the defect caused by the SSR1 mutation rather than promoting root growth in a parallel pathway.

SSR1 interacts with HSCA2 and ISU1 and promotes HSCA2-ISU1 association

Since both suppressor genes, HSCA2 and ISU1, encode components in ISC machinery, it is possible that SSR1 may be also involved in Fe-S cluster assembly in mitochondria. To understand the function of SSR1 in the ISC machinery, we analyzed possible protein-protein interactions between SSR1, HSCA2 and ISU1 by bimolecular fluorescence complementation (BiFC) assay using Arabidopsis mesophyll protoplasts. Fluorescence signals were clearly visible in protoplasts as discrete bright spots that overlapped with mitochondria when cCFP-tagged SSR1 was co-expressed with nVenus-tagged HSCA2 or ISU1 (Fig 2A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. SSR1 interacts with HSCA2 and ISU1 and promotes HSCA2-ISU1 association.

(A, B) in vivo interactions between SSR1 and HSCA2 or ISU1 were analyzed in Arabidopsis protoplasts by BiFC and co-immunoprecipitation assays. In (A), full length of SSR1, HSCA2 or ISU1 was used for fusion with reporter genes. The mitochondria are labeled by MitoTracker and emit red light. FP represents reconstitute fluorescent proteins. In (B), stable transgenic lines co-expressing Flag-tagged full-length SSR1 or Flag alone under CaMV 35S promoter and Myc-tagged full-length HSCA2 under its native promoter were immunoprecipitated with anti-Flag antibody, and the co-purified proteins were immunodetected with anti-Myc and anti-ISU antibodies. (C) In vitro pull-down assays were conducted using Ni-NTA Sepharose with His-tagged proteins expressed and purified from E. coli as baits. The prey proteins were also expressed and purified from E. coli, after which the His-tag was excised. Co-purified proteins were detected with anti-Myc antibody. The immunoblotting signals of ISU1-Myc or ISU1^T55M-Myc were quantified using the ImageJ software, and the relative intensities are presented under the respective lanes with three replicates. The mitochondrial targeting peptide coding sequences of HSCA2, SSR1 and ISU1 were removed for protein expression in E. coli. The statistical significance of the results was determined through the use of a student t-test. Variations were considered significant if P <0.05(*), 0.01(**) or 0.001(***).

https://doi.org/10.1371/journal.pgen.1011597.g002

Co-IP assays were then performed using transgenic plants to verify above interaction in vivo. Given the relatively weak activity of the native SSR1 promoter, the SSR1-Flag construct driven by the 35S promoter was employed and co-integrated into ssr1-2 mutants with the HSCA2-Myc construct driven by the native promoter via transformation and subsequent crossing. Protein extracts from transgenic plants were incubated with anti-Flag antibody resin to coprecipitate SSR1-Flag associated proteins. Subsequent immunoblotting analysis with anti-Myc or anti-ISU1 antibodies demonstrated that both HSCA2-Myc and ISU1 were coprecipitated with SSR1-Flag, though with much less ISU1 being co-purified compared with HSCA2 (Fig 2B).

Additionally, protein-protein interaction was further investigated by in vitro pull-down assays using purified proteins. A variety of combinations of the proteins His-HSCA2-Myc, His-HSCA2^G87D-Myc, His-SSR-Myc, SSR-Myc, ISU1-Myc, and ISU1^T55M-Myc were incubated, followed by pull-down with Ni-Sepharose and analyzed by Western blot with anti-Myc antibody. The results further demonstrated that HSCA2, SSR1, and ISU1 can interact directly with one another (Fig 2C). It is noteworthy that ISU1 exhibited a more robust interaction with HSCA2^G87D than with HSCA2, while HSCA2 demonstrated a stronger interaction with ISU1^T55M than with ISU1 (Fig 2C). Additionally, the presence of SSR1 further promoted the interaction between ISU/ISU1^T55M and HSCA2/HSCA2^G87D (Fig 2C).

The ISC machinery is a dynamic and complex system composed of multiple components. We postulate that the effect of SSR1 on the affinities of protein interaction between ISU/ISU1^T55M and HSCA2/HSCA2^G87D may be more pronounced in the presence of other cellular cofactors. Therefore, pull-down assay was also tried with purified His-HSCA2-Myc/His-HSCA2^G87D-Myc proteins and E. coli lysates containing SSR-Myc, ISU1-Myc, and ISU1^T55M-Myc, respectively. As expected, ISU1 exhibited a more robust interaction with HSCA2^G87D than with HSCA2, while HSCA2 demonstrated a stronger interaction with ISU1^T55M than with ISU1, and the presence of SSR1 significantly promoted the interaction between ISU/ISU1^T55M and HSCA2/HSCA2^G87D (S3 Fig).

We tried to model the interaction between HSCA2 and ISU1 using AlphaFold3 to understand the effect of mutations on the interactions. Notably, previous studies have identified the LPPVK sequence in ISU1 as a critical site for HSCA2 binding [16,17,19]; however, this LPPVK region does not resided at the interaction interface with HSCA2 in the predicted interaction model, underscoring the misalignment in the predicted relative orientations of the two proteins (S1 Dataset). Therefore, it is not feasible to compare the effects of amino acid variations on the interaction between ISU1 and HSCA2 using this method.

Next, the combination of blue native polyacrylamide gel electrophoresis (BN-PAGE) and SDS-PAGE was employed to investigate the potential for His-SSR1-Myc, His-HSCA2-Myc, and ISU1-Myc proteins to form complexes. Although the immunoblotting signal is weak, the signal of three proteins is clearly observed at approximately 400 kDa simultaneously (S4A Fig). However, the co-existence of His-SSR1-Myc, His-HSCA2-Myc, and ISU1-Myc at approximately 400 kDa was not observed in any other combinations (S4B–S4G Fig). These results indicate that the HSCA2-SSR1-ISU1 complex does exist, but it may be not stable.

The SSR1 function is dependent on the interaction between HSCA and ISU1

Based on the protein-protein interaction, we speculated that the enhanced affinity between ISU1 and HSCA2, when a point mutation in either ISU1 (ISU1^T55M) or HSCA2 (HSCA2^G87D) is present, may be responsible for the suppression of the root growth defect in ssr1-2. To verify this hypothesis, amino acid substitutions were introduced into ISU1^T55M at the LPPVK motif, which was previously reported to mediate the interaction with HSCA [16,17,19]. In vitro pull-down assays showed that L126A, P127A, or P128S mutation slightly impaired the interaction between ISU1^T55M and HSCA2, while V129E or K130A mutation dramatically reduced the affinity between ISU1^T55M and HSCA2 (Fig 3A). In addition, simultaneous substitution of PVK to AAA almost completely abolished the interaction between ISU1^T55M and HSCA2 (Fig 3A). Subsequently, the ISU1 mutant constructs were introduced into ssr1-2 plant to test their ability to suppress the short-root phenotype of ssr1-2. It is interesting to note that ISU1^T55ML126A, ISU1^T55MP127A, and ISU1^T55MP128S, demonstrate partial rescue of ssr1-2, whereas ISU1^T55MV129E, ISU1^T55MK130A, and ISU1^T55MAAA are unable to do so (Fig 3B). This aligns with their capacity to interact with HSCA2. Taken all these together, we have demonstrated that enhanced interaction between ISU1 and HSCA2 is likely the cause of the rescue of ssr1-2 phenotype.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Interaction between ISU1^T55M and HSCA2 is essential for the rescue of ssr1-2 growth defect by ISU1^T55M.

(A) Ni-NTA in vitro pull-down assays with His-tagged HSCA2 expressing in E. coli as the bait. Different Myc-tagged ISU1 mutation variants within the LPPVK motif were expressed in E. coli and differentially co-purified with HSCA2. The mitochondrial targeting peptide coding sequences of HSCA2 and ISU1 were removed for protein expression in E. coli. (B) Root length of the wild type (WS) and ssr1-2 mutants carrying sus2 allele or transformed with an additional ISU1^T55M gene with differentially mutated LPPVK motif. Here, the full length of ISU1 with different mutations was employed for transformation. Top, representative seedlings at 8-days-old. Bottom, root lengths shown as bar graphs with error bars representing standard deviations from about 30 seedlings. Statistical significance was determined through student t-test. The asterisk “***” represents a statistically significant result with p-values < 0.001. The symbol “ns” represents a statistical insignificant result with p-values < 0.05.

https://doi.org/10.1371/journal.pgen.1011597.g003

SSR1 displays a chaperone-like function

The involvement of HSCA and HSCB chaperone system in the biosynthesis of Fe-S clusters has been well documented in bacteria, yeast and plant [2,7,14,22]. The release of mature Fe-S cluster from ISU1 requires the binding of chaperone HSCA, and this process can be stimulated by cochaperone HSCB [14]. Since we have shown that SSR1 interacts with HSCA2 and ISU1, and promotes the interaction between ISU1 and HSCA2, we speculate that SSR1 may display a chaperone-like activity. To explore this possibility, we used purified SSR1, ISU1, HSCA2 and HSCA2^G87D from E. coli to test their general chaperone activity in preventing heat-induced substrate protein from aggregation.

By using citrate synthase (CS) as a model substrate, it was shown that both HSCA2 and HSCA2^G87D have a strong general chaperone activity, and the difference between the wild-type and the mutant form is subtle (Figs 4A and S5). ISU1, however, unexpectedly inhibited heat-induced aggregation of CS, although not as efficiently as HSCA2 (Fig 4B). When both ISU1 and HSCA2/HSCA2^G87D were added to CS, much more aggregates accumulated especially at the later time (Fig 4B and 4C). This implied that certain physical interactions occurred between HSCA2 and ISU1. Interestingly, whenever SSR1 is present, no aggregate formed, indicating SSR1 has a powerful chaperone-like activity (Fig 4B and 4C). Further titration analysis demonstrated that SSR1 itself displayed a very strong chaperone activity in preventing heat-induced citrate synthase (CS) from aggregating even at a very low concentration (Fig 4D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. SSR1 appears strong general molecular chaperone activity.

Heat-induced aggregation of citrate synthase (CS) was performed at 45°C for 90 min and monitored by increased light scattering at 340 nm. The molecular ratios of CS to tested proteins are indicated within brackets following each protein sample. Control samples without adding CS or those tested protein mixtures that did not show significant protein aggregation are showing curves crowded along the basal line, and therefore shown together without referring individual curve. Only one representative absorbance curve is shown for each tested sample mixture. (A) Heat-induced CS aggregation with HSCA2 and HSCA2^G87D. (B) Heat-induced CS aggregation with HSCA2 in combination with ISU1 and/or SSR1. (C) Heat-induced CS aggregation with HSCA2^G87D in combination with ISU1 and/or SSR1. (D) Heat-induced CS aggregation with different amounts of SSR1. TAH1 is a TPR-containing yeast protein and used as a negative control that does not inhibit heat-induced CS aggregation. These tests have been repeated 3 times.

https://doi.org/10.1371/journal.pgen.1011597.g004

SSR1 affects iron homeostasis

The ISC machinery plays a key role in maintaining cellular iron homeostasis by regulating the uptake, storage, and distribution of iron [33,34]. This helps to prevent iron overload or deficiency, which can lead to oxidative stress and damage to cellular structures. Overexpression of AtHSCB led to higher iron accumulation in roots and lower iron content in shoots, whereas hscb knockdown mutants exhibited reduced iron uptake in roots but increased iron levels in shoots [35]. Perls’ prussian blue staining showed that ssr1-2 has an unusually high iron content in the root tips (Fig 5A). Iron content in root tips and leaves was further measured using inductively coupled plasma-mass spectrometry (ICP-MS). ssr1-2 had higher iron content in the tips and leaves than the wild type (Fig 5B and S5 Table). Furthermore, the accumulation of iron around the quiescent center and the stem cell niche of the root was obviously higher in ssr1-2 than in the wild-type (Fig 5C). As expected, SSR1, sus1, and sus2 well restored the iron accumulation phenotype in the root tips and leaves of ssr1-2 mutants (Fig 5A–5C and S5 Table).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. SSR1 affects iron homeostasis.

(A) Perls’ prussian blue staining of root tips. (B) Iron content in root tips (about 0.5 cm). (C) Semi-thin (1-mm) longitudinal sections of Perls/DAB-stained root tips. This method allows for the sensitive detection of trace amounts of iron in tissues. The seedlings utilized in Fig 5 were 10-days old and were cultivated on MS medium. Scale bars in 5A and 5C are 50 μm. In Fig 5B, statistical significance was determined through student t-test. The asterisk “***” represents a statistically significant result with p-values < 0.001. The symbol “ns” represents a statistical insignificant result with p-values < 0.05.

https://doi.org/10.1371/journal.pgen.1011597.g005

Loss-of-function of SSR1 causes mitochondrial dysfunction

The interaction of HSCA and ISU are important for the release of Fe-S clusters from ISU to apo-proteins [2,20]. Additionally, our recent study on another weak allele of SSR1 indicated that SSR1 is important in regulating the function of mitochondrial electron-transport chain complexes, a few of which are well-known Fe-S proteins [30]. We therefore hypothesized that the mitochondrial Fe-S cluster biosynthesis or the biosynthesis of Fe-S proteins are impaired in the ssr1-2 mutant.

We then investigated the activities and/or protein levels of some Fe-S proteins, namely mitochondrial complex I (CI), aconitase (ACO) and succinate dehydrogenase 2 (SDH2), in ssr1-2 seedlings. Given that aconitase is located in mitochondria and cytosol, the aconitase activity from both compartments was measured. The results showed that the enzymatic activities of CI, aconitase, and the protein level of SDH2 in ssr1-2 were dramatically low compared to that in the wild type, ssr1-2 sus1, and ssr1-2 sus2 (Fig 6A–6C). However, the enzymatic activity of malate dehydrogenase, which is not an iron-sulfur protein, displayed no difference between the wild type and ssr1-2 (Fig 6D). Additionally, the protein level of ATP5A, a Fe-S cluster-independent subunit of complex V, was higher in ssr1-2, while ISU1 exhibited comparable levels between the samples (Fig 6D). Mitochondrial dysfunction often affects mitochondrial membrane potential (MMP). To minimize the interference from chloroplasts, seedlings grown on MS medium in the dark were used for crude mitochondria preparation. A flow cytometry analysis revealed that the MMP was significantly diminished in ssr1-2 compared to the wild type, reaching a mere 30% of the wild type level. However, in ssr1-2 sus1 and ssr1-2 sus2 double mutants, the MMP exhibited a partial restoration, approaching 55% of the wild type level (Fig 6E). Taken all data together, it is concluded that SSR1 is required for Fe-S proteins activity, thus further confirming the role of SSR1 in the maintenance of mitochondrial electron-transport chain as revealed by analyzing a phenotypically weak mutant ssr1-1 under the proline treatment [30].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The activity of representative Fe-S containing enzymes and mitochondrial membrane potential decreased in ssr1-2 mutant, and were restored by sus1 or sus2.

Mitochondrial complex I (A) and aconitase (bottom panel of 6C) enzyme activities. Cytosolic aconitase (top panel of 6C) and malic dehydrogenase (D) enzyme activities. (B) Protein expression levels of SDH2, ISU, and ATP5A as detected by corresponding antibodies and the total protein shown by Coomassie brilliant blue (CBB) staining. (E) Mitochondrial membrane potential (MMP). CCCP, a mitochondrial oxidative phosphorylation uncoupler, was used as positive control for mitochondria membrane depolarization. Flow cytometry was used for MMP detection, with over ten thousand mitochondria from each sample measured in every replicate. Error bars represent standard deviation (n = 3). Statistical significance was determined through student t-test. Asterisks indicate a significant difference to the WS or between two designated groups. **P < 0.01; *P < 0.05. “ns” represents statistically insignificant.

https://doi.org/10.1371/journal.pgen.1011597.g006

SSR1 only exists in plant lineages and may be important for environmental adaptability

It has been shown that ssr1 mutants are hypersensitive to osmotic stress and proline treatment [29,30], which were believed to cause electron overflow in mitochondrial electron transport chain (mETC) due to decreased mETC activity or elevated electron supply [36,37], and consequently resulted in the burst of reactive oxygen species (ROS) [38]. Actually, ISC machinery is prone to inhibition by ROS under various environmental stresses [39]. Our observations indicated that the sus1 and sus2 were unable to effectively rescue the hypersensitive phenotype of the ssr1-2 mutants to proline treatment, in contrast to their performance under no-stress conditions (Figs 7A, 7B, 1A and 1D). Moreover, SSR1 was induced by proline, dehydration, mannitol, and NaCl treatments (Fig 7C–7F), indicating that SSR1 may be important for environmental adaptability.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. SSR1 co-evolves with the ISC machinery in plant lineages and may be important for environmental adaptability.

(A, B) Phenotype of 10-day-old seedlings under proline (20 mM) treatment. (C-F) The relative transcript level of AtSSR1 under different stresses. 10-day-old seedlings were used for treatment with proline (40 mM), mannitol (150 mM), and NaCl (100 mM). For dehydration treatment, detached leaves from 30-day-old plants were placed in a greenhouse for natural dehydration, and the control group was floated in water. The control group was used to normalize the data of the experimental group at each time point. Actin7 and eIF4a were used as the internal reference genes. (G) Phylogenetic tree was constructed using SSR1 protein sequences from representative species at different evolutionary status. (H-M) Pairwise genetic distances (GD) among designated proteins in representative species (S6 Table) are indicated by green circles. R, Pearson correlation coefficient; P, Significant differences between the Maximum Likelihood trees (one-sided t-test).

https://doi.org/10.1371/journal.pgen.1011597.g007

The ISC machinery is evolutionarily conserved from bacteria, fungi to higher plants and mammals. This prompted us to investigate the evolutionary origin of SSR1 and the evolutionary relationship between SSR1 and the ISC system. It is notable that SSR1 is not inherited from microorganisms. Rather, it emerged as a novel gene present only in plant lineages with a distant homolog appearing already in green algae (Fig 7G). Comparison of the evolutionary distances between the sequences of the associated protein families is an efficient way to examine possible protein–protein functional interactions [28]. After alignments of the predicted orthologs and constructions of the phylogenetic trees of NFS1, FH, HSCA1, HSCA2, and ISU1, we found that the phylogenetic trees of SSR1 and proteins of the ISC machinery share similar topologies (S6 Fig and S6 Table), implying that their evolution is highly coordinated. These tree similarities were further quantified by the Tol-Mirror-Tree analysis. Notably, strong correlations (R > 0.600) were observed between the SSR1-NFS1, SSR1-FH, SSR1-HSCA2, and SSR1-ISU1 trees compared to the SSR1-HSCA1 tree (Fig 7H–7M), indicating that SSR1 co-evolves with the ISC machinery and that HSCA2 rather than HSCA1 may be a major component of the ISC machinery.

In order to test whether SSR1 could play any adaptive roles under stress conditions, SSR1 was introduced into yeast cells whose ISC machinery doesn’t possess SSR orthologs. SSR1-GFP could be detected in the mitochondria of yeast cells (S7A and S7B Fig). However, its expression had an adverse effect on the normal growth of yeast and was unable to enhance the tolerance of yeast to various stress (S7C Fig).

SSR1 plays a role in ISC machinery by facilitating the interaction between HSCA and ISU

SSR1 is a mitochondrial TPR domain-containing protein and was previously reported to be required for the function of mitochondria electron transport chain [30] and primary root elongation [29]. However, the molecular mechanism underlying the function of SSR1 remained elusive. In this report, we have pinpointed, by suppressor characterization, that the function of SSR1 in mitochondria as a crucial chaperone-like component to promote the association of HSCA2 with ISU1, with the possibility to facilitate the mitochondrial Fe-S clusters assembly.

It has been well documented that the binding of HSCA to ISU is required for the release of Fe-S clusters from ISU to apo-proteins [16,17,19,20,22]. Our results demonstrated that the optimal binding of HSCA2 to ISU1 is largely SSR1-dependent, as evidenced by the significantly weaker binding observed in the absence of SSR1. However, HSCA2^G87D and ISU1^T55M could interact well with ISU1 and HSCA2, respectively, thus by-passing the need of SSR1. It appeared that the enhanced affinity between HSCA2 and ISU1 in two suppressor mutants was responsible for suppressing the defect of ssr1-2. This was demonstrated by artificially reducing the affinity between HSCA2 and ISU1^T55M by certain amino acid substitution in the LPPVK motif of ISU1^T55M, which render ISU1^T55M incapable of suppressing the ssr1-2 phenotype.

Fe-S clusters are ubiquitous and essential protein cofactors for many proteins engaged in a multitude of processes associated with bioenergetics, metabolism, DNA synthesis and phytohormone biosynthesis [9,33]. Therefore, it was not surprising that ssr1-2 mutant displayed pleiotropic growth defect, such as short roots, brush-like roots, bushy shoots and growth retardation [29]. Various degrees of growth retardation have been observed in other mutants deficient in the key components of Fe-S cluster assembly system, such as AtNFS1 and AtISU1. The loss of function of AtNFS1 is lethal [25]. The knockdown of AtNFS1 expression resulted in a number of striking visible phenotypes, including dwarf stature, chlorotic spots on the leaves, scalloped edges on new leaves, disorganized inflorescences, and a noticeable increase in axillary shoot development at the axils of both rosette and cauline leaves, resulting in a bushy shoot [25]. The knockdown of AtISU1 expression resulted in similar phenotypes as that of AtNFS1 [25]. The dysfunction of the Fe-S clusters biosynthesis machinery also disrupts iron and sulfur homeostasis [33,34]. Genes involved in Fe and S uptakes, assimilation, and regulation were up-regulated in AtNFS1 overexpressing plants and down-regulated in the knockdown plants [40]. Specifically, the overexpression of AtNFS1 resulted in elevated Fe and S accumulation, while the knockdown of AtNFS1 led to reduced Fe and S accumulation [40]. ATM3, an ATP-binding cassette transporter in mitochondria, was proposed to transport some unknown intermediates for Fe-S cluster assembly in cytosol. ATM3 T-DNA insertional mutant, atm3-1, showed dramatically reduced content of abscisic acid and the enzymatic activities of some cytosolic Fe-S proteins, resulting in a pleiotropic growth defect, such as small stature, lower chloroplast numbers and male sterility [41,42]. Strikingly, in contrast to mutants in the yeast and mammalian orthologs, Arabidopsis atm3 mutants did not display a dramatic iron homeostasis defect and did not accumulate iron in mitochondria [41,42]. Two DnaJ chaperones, DJA6 and DJA5, have recently been identified as essential for chloroplast Fe-S cluster biogenesis. DJA6 and DJA5 proteins are capable of binding iron through their conserved cysteine residues and facilitate iron incorporation into Fe-S clusters via interactions with the SUF apparatus through their J domain [28]. The absence of these two proteins results in severe defects in the accumulation of chloroplast Fe-S proteins, a dysfunction of photosynthesis, a significant intracellular iron accumulation, and growth inhibition [28]. Notably, ssr1-2 mutants accumulated higher levels of iron in their root tips. The increased iron deposition may be detrimental to the maintenance of the quiescent center and stem cell niche, as the excess ferrous iron may trigger the Fenton reaction, leading to oxidative stress and damage to cellular structures.

It was well-established in microorganisms that HSCB/Jac1 could promote the interaction between HSCA and ISU, and stimulate the release of Fe-S clusters from ISU to apo-proteins [43,44]. The knockout mutants of AtHSCB exhibit significantly reduced activities of the Fe-S cluster containing enzymes ACO and SDH [26,35]. However, the activities of the Fe-S cluster-containing enzymes, such as ACO and CI, are not impacted by the dysfunction of GRX15, which is proposed to be the primary carrier of [2Fe-2S] cluster during its release from the ISU scaffold [45]. Conversely, grx15 primarily influences the activity of lipoyl-dependent proteins [45,46]. Overexpression of AtHSCB resulted in the activation of the iron uptake system and iron accumulation in roots, without concomitant transport to shoots. This ultimately led to a reduction in iron content in the aerial parts of plants. By contrast, hscb knockdown mutants exhibited the opposite phenotype, with iron accumulation in shoots despite the reduced levels of iron uptake in roots [35]. The SSR1 knockout mutants have been observed to suppress the expression of iron deficiency response genes and to increase iron accumulation in root tips. The jac1 yeast mutant is lethal, indicating that Jac1 is an essential gene in yeast [15]. However, the defective phenotype of athscb mutant is relatively mild [26,35]; this suggests the possible existence of functional redundancy genes. Could SSR1 serve as a functionally redundant cochaperone of HSCB in Arabidopsis. HSCB does not exhibit intrinsic chaperone activity and only function as a cochaperone for HSCA [47]. The robust chaperone function of SSR1 and the different role of SSR1 and HSCB in iron homeostasis indicated that SSR1 did not function the same as HSCB. However, in the process of SSR1-assisted binding of HSCA2 to ISU1, it seems more appropriate to call SSR1 as a cochaperone. This aligns with previous reports that the TPR domain binds to the C-terminus of HSP70 [48]. Since SSR1 can bind to and protect citrate synthase as a chaperone, it may also bind to certain other proteins and perform various biological functions.

SSR1 may be a plant specific component in ISC machinery for stress adaptation

Unlike most of the other components in ISC system, SSR1 is not inherited from microorganism. Rather, it emerged as a new gene present only in plant kingdom with a distant homolog appearing already in green algae. Currently, we are not clear about the evolutionary advantage for plant to possess SSR1. One clue is that ssr1 is hypersensitive to osmotic stress [29] and proline treatment [30]. Both of these factors were previously believed to cause electron overflow in the mitochondrial electron transport chain (mETC) due to inhibited mETC activity or elevated electron supply [36,37]. Actually, ISC machinery is prone to inhibition by ROS under various environmental stresses during their sessile life cycle, to which plants must respond and adapt [39]. It has been recently reported that hypoxia can rescue FH/frataxin loss-of-function caused dysfunction of ISC machinery [49]. Although HSCA2^G87D or ISU1^T55M can effectively rescue the defect of ssr1-2, they did not rescue the sensitive phenotype of ssr1-2 to proline treatment. We propose that SSR1 may be required for protecting ISC machinery from stresses, which cannot be completely replaced by HSCA2^G87D or ISU1^T55M. Components of Fe-S cluster assembly machineries were recently reported to be robust phylogenetic markers [50]. The complexification of the ISC machinery and expansion of its components are indicative of the capacity of organisms to adapt to the environment [9]. Therefore, the emerging of SSR1 in plant lineages may confer adaptive advantages on ISC machinery in response to environmental stress. In line with this, SSR1 was shown to be transcriptionally up-regulated by several abiotic stresses. However, our observations did not indicate that the expression of AtSSR1 in yeast cells could enhance the stress adaptation of yeast. Conversely, the expression of AtSSR1 inhibited yeast growth. We proposed that this may be due to the lack of co-evolution of other components (such as HSCA and ISU) of the ISC machinery in yeast that are compatible with AtSSR1.

In conclusion, we have presented evidence that SSR1 is a new and vital cochaperone-like component for mitochondrial Fe-S clusters biosynthesis in plants.

Plant materials and growth conditions

For seedling growth experiments, surface-sterilized seeds of Arabidopsis wild type (WS), transgenic plants and mutants were sown on solid Murashige and Skoog (MS) medium, containing 1% sucrose and 0.25% phytogel, and stratified at 4°C for 2 days in the dark before being transferred to an incubator for germination at 22–24°C, 16-h light/ 8-h dark photoperiod with a light intensity of 110 μmol.m^-2.sec^-1. To grow to maturity, 4-day-old seedlings were planted in soil and grown under similar condition as in the incubator. All transgenic plants used in this study were listed in S7 Table.

Cloning of SUS1 and SUS2

ssr1-2 sus1 and ssr1-2 sus2 were backcrossed twice with the ssr1-2 parent to obtain F1 plants. The F1 plant was self-pollinated to obtain F2 progeny, which were further planted and harvested individually. The root length of the F3 population was investigated. Populations exhibiting no apparent segregation of root lengths were deemed homozygous for the SUS1 and SUS2 loci. The 30 F3 populations with the longest root length were selected, and an equal number of plants from each population was mixed to extract DNA to construct a long-root pool. A total of 30 F3 populations exhibiting the same root length as ssr1-2 were selected, and an equal number of plants from each population was mixed to extract DNA to construct a short-root pool. DNAs isolated from ssr1-2 and two pools were subjected to genome re-sequencing. Super Bulked Segregant Analysis [51] was used to screen for candidate mutation sites that are correlated with the root length phenotype (S2 and S3 Tables). Trait-related candidate regions were selected based on the differences in genotype frequency in mixed pools, and were calculated using the Euler distance (ED) correlation algorithm. The quantile-based method was used to select the threshold value. All fitted values were sorted from the smallest to the largest, and the SNP labeled ED value greater than 99% or 99.9% was selected as the threshold value during screening. Subsequently, the genes with SNPs that were most correlated with root length were subjected to further analysis based on their functional annotation and prediction of protein subcellular localization. The five most likely candidate genes were cloned from each suppressor mutant and introduced into ssr1-2 for genetic complementation (S2 and S3 Tables).

Plasmids construction and plant transformation

All constructs and primers used in this study were listed in S8 Table. Briefly, for genetic complementation, the genomic sequences for sus1, sus2, sus5 and sus6, which encode HSCA2^G87D, ISU1^T55M, ISU1^G106D and ISU1^A143T respectively, were amplified and cloned from corresponding suppressor mutants. To generate HSCA1^G82D from wild type HSCA1, the HSCA1 coding sequence (CDS) was amplified in two halves separately, and then fused together by overlap extension PCR with the simultaneous introduction of the mutation. All complementation constructs were based on the binary vector pCAMBIA1300 and pCAMBIA1300-super-Flag. For protein expression in E. coli, the coding sequences of HSCA2, SSR1 and ISU1, with or without mitochondrial targeting peptide coding sequences, were cloned into vector pRSET and fused with the Myc-tag. Amino acid substitution mutants around the LPPVK motif of ISU1 were also generated with overlap extension PCR. To generate constructs, transfected and transgenic plants for co-IP analysis, the promoter and CDS of HSCA2 was amplified and first inserted into pBI121-cMyc. Then the Myc-tagged expression cassette was amplified and inserted into pCAMBIA1300. The pSSR1-Flag was described previously [29]. For BiFC assays, the CDS of SSR1, HSCA2 and ISU1 from wild type were inserted into pE3242 or pE3228 to generate either nVenus-tagged or cCFP-tagged constructs, respectively. See Lee’s work for more details about vectors pE3228 and pE3242 [52].

Whenever needed, Arabidopsis plants were transformed with Agrobacteria-mediated floral dipping method [53] or the protoplasts were transfected with expression constructs as described [54]. The trangenic plants were screened on kanamycin- or hygromycin B-containing MS medium depending on the vector used. The integration of the transgene was confirmed by PCR.

Protein expression and purification from E. coli

His-tagged protein expression and purification from E. coil was carried out as described previously [22]. Briefly, BL21(DE3) bacterial strains with respective constructs were cultured in LB liquid medium at 37°C to OD_600nm≈0.5, and then induced with IPTG at a final concentration of 1 mM for 6 h at 28°C. Cells were harvested, re-suspended in buffer A [20 mM Tris–HCl, 200 mM NaCl, 30 mM imidazole and 1 mM phenylmethylsulfonyl fluoride (PMSF), pH 7.4] and then disrupted by sonication. The suspensions were centrifuged at 10,000 ×g for 15 min at 4°C. The supernatants of the His-tagged proteins were purified using His-tag protein purification columns (Cytiva: HisTrap HP). The recombinant proteins were eluted with buffer B (500 mM imidazole in buffer A). In order for proteins to be utilized in in vitro chaperone activity assays, the eluants were subjected to further application of size exclusion chromatography with the Superdex 75 or Superdex 200 column, utilizing the ÄKTA Purifier 10 FPLC system (GE Healthcare). The E. coli lysates containing the recombinant proteins without His-tag were directly used for the pull-down assays.

Protein-protein interaction assays

Pull-down assay was carried out as described previously [22]. For purified proteins, His-tagged proteins and tag-excised proteins were incubated together equally (about 10 μg of each protein in a total volume of 200 μL). The incubation buffer is formulated as: 50 mM tris-HCl (PH7.4), 50 mM NaCl, 10 mM KCl, 0.1% Tween-20, and cocktail protease inhibitors. For the recombinant proteins without His-tag, about 10 μg purified His-tagged HSCA2, HSCA2^G87D or SSR1 was incubated with BL21(DE3) cell lysates expressing SSR1, ISU1 or ISU1^T55M without His-tag for 1 hour at 4°C. Next, 20 μL Ni Sepharose pre-equilibrated with buffer A [20 mM Tris–HCl, 200 mM NaCl, 30 mM imidazole and 1 mM phenylmethylsulfonyl fluoride (PMSF), pH 7.4] were added and incubated for another 2 hours at 4°C. After washing twice with buffer A, the proteins binding to Ni-NTA Sepharose (GE Healthcare: Ni Sepharose 6 Fast Flow) were eluted by appropriate volume of buffer B (500 mM imidazole in buffer A), and analyzed by Western blot.

The first dimension of BN-PAGE and second dimension of SDS-PAGE were performed as described by Wittig et al. [55]. Briefly, Different designated combinations of proteins were mixed (about 10 μg of each protein in a total volume of 200 μL) and incubated at 4°C for one hour, after which they were subjected to a BN-PAGE assay. The incubation buffer is formulated as: 50 mM tris-HCl (PH7.4), 50 mM NaCl, 10 mM KCl, 0.1% Tween-20, 100 μM ATP, and cocktail protease inhibitors. Cathode buffer and anode buffer were the same as described by Wittig et al. [55]. A gradient separation gel (5%-13% acrylamide) was employed for the first dimension of BN-PAGE, while a 10% acrylamide separation gel with β-mercaptoethanol was utilized for the second dimension of SDS-PAGE.

BiFC assays were performed via transient expression in Arabidopsis mesophyll protoplasts as previously described [54]. Co-IP assays from transgenic plants were performed with C-Myc Isolation Kit (130-091-123) and DYKDDDDK (Flag) Isolation Kit (130-101-591) from Miltenyi Biotec. Briefly, 12-day-old transgenic seedlings containing HSCA2_pro:HSCA2-Myc and 35S_pro:SSR1-Flag were ground to fine powder in liquid nitrogen. About 500 mg powder was transferred to 2 mL Eppendorf tube, before 1.5 mL pre-cooled lysis buffer (50 mM Tris-HCl, pH7.9, 120 mM NaCl, 10% glycerol, 10 mM DTT, 5 mM EDTA, 1% PVP, 1% NP-40, 1 mM PMSF, protease inhibitor cocktail) was added and mixed well with the powder. The homogenates were incubated for 30 minutes on ice with occasional mixing and then centrifuged for 10 minutes at 10,000 ×g at 4°C to get rid of the debris. Total input samples were taken from the supernatant, and the remaining supernatant was transferred to a fresh 1.5 mL tube with 50 μL antibody and was incubated on ice for 40 minutes. Place μ column (130-042-701) in the magnetic field of the μMACS Separator and prepare the μ column by applying 200 μL lysis buffer on the column. Transfer the supernatant into the μ column and let the lysate run through. Rinse the column with 4×200 μL of lysis buffer and 1×100 μL wash buffer (20 mM Tris-HCl, pH 7.5). Pre-heated elution buffer (from Kit) was subsequently added to elute the immunoprecipitated, which was used for Western blot.

The antibodies used for Western blot were purchased from Abcam (anti-ISU: ab154060; anti-SDH2: ab154974; anti-ATP5A: ab14748) and Sigma-Aldrich (anti-Myc: M4439; anti-Flag: F3165).

In vitro chaperone activity assay

All tested proteins and citrate synthase (CS, Sigma, C3260) was dialyzed in 20 mM HEPES-KOH, pH 7.5, 150 mM KCl, 10 mM MgCl₂ before being used for the heat-induced aggregation assay. CS (500 nM) was prepared in a final volume of 150 μL 20 mM HEPES-KOH (pH 7.5), 2.8 mM β-mercaptoethanol with different amounts of tested proteins. The mixtures were added to a 96-well microplate and heated at 45°C. Light scattering at 340 nm was monitored at 45°C in a Synergy 4 spectrophotometer (BioTek) for 90min. Control measurements were performed with purified test proteins alone in the absence of CS and a His₆-tagged yeast TPR-containing Tah1 protein expressed and purified from E. coli was also used as a negative control [56].

Isolation of mitochondria

Mitochondria was isolated as previously described [57] with minor modification using 10-day-old seedlings grown on solid MS medium. Briefly, plant tissue was homogenized in Mito Homogenization buffer (0.35 M mannitol; 30 mM MOPS (pH7.3); 1 mM EDTA (pH8.0); adding 0.2% (w/v) BSA, 0.6% (w/v) polyvinylpyrrolidone-40 and 0.3% (w/v) ascorbic acid in cooled autoclaved buffer), filtrated through one layer of Miracloth and two layers of muslin to remove starch, cell debris, and unbroken cells, and centrifuged at 5000 x g for 5 min. The supernatant was gently transferred into fresh centrifuge tubes and centrifuged at 25000x g for 15 min. The pellet was gently resuspended in 2 mL Mito Homogenization buffer, loaded on a cooled Percoll gradient in 10 mL Oak ridge tubes (bottom to top, 2 mL of 45% Percoll, 3 mL of 27% Percoll, 2 mL of 21% Percoll in 0.25 M sucrose, 10 mM MOPS (pH7.3), 0.2% (w/v) BSA), and centrifuged at 25000 x g for 30 min. Crude mitochondria will form a white/pale band between 45% and 27% Percoll or in 45% Percoll toward the bottom.

Enzyme activity and MMP analysis

Crude mitochondria from seedlings grown in light/dark photo-period were used for analysis of enzymatic activity for complex I (Catalog number: FHTA-2-Y), aconitase (Catalog number: ACO-2-Z), and malate dehydrogenase (Catalog number: NMDH-2-Y) with kits from Suzhou Comin Biotechnology Co. Ltd according to the instruction manuals.

Crude mitochondria from dark-grown seedlings were used for mitochondrial membrane potential (MMP) measurement using kit from Beyotime (Catalog number: C2006) with JC-1, a fluorescent indicator. Flow cytometry (Beckman: Moflo XDP) was used to detect the MMP. The ratio of red fluorescence to green fluorescence can be utilized to assess the potential of the mitochondrial membrane. A smaller ratio is indicative of a mitochondrial membrane that is in a depolarized state.

Histochemical Fe staining

The Fe-specific Perls staining was adapted from Muller et al [58]. The seedlings were washed twice with distilled water (ddH₂O) and incubated in a 10% neutral formaldehyde solution for two hours. For the intensification of the 3,3’-diaminobenzidine (DAB) reaction and the preparation of semi-thin (1 mm) sections, the roots were fixed with a 4% glutaraldehyde solution for four hours. Subsequently, the seedlings were washed with ddH₂O twice and transferred to a staining buffer [4%(v/v) HCl, 4% (w/v) K-ferrocyanide] for a 30 min incubation at room temperature. The seedlings were then washed (dH₂O) and incubated (1 hour) in methanol containing 10 mM sodium azide and 0.3% (v/v) H₂O₂. Following a wash with 100 mM Na-phosphate buffer (pH 7.4), the plants were incubated for up to 30 min in the same buffer containing 0.025% (w/v) DAB and 0.005% (v/v) H₂O₂. The reaction was stopped by washing with water and incubation with chloral hydrate (1g/mL in 15% glycerol). Perls/DAB-stained roots and sections were analyzed on a Zeiss Axio Imager M2.

Measurement of Fe content

The Fe content was quantified via ICP-MS (PerkinElmer: NexION 350). The root tips (0.5 cm) of 10-day-old seedlings were harvested and washed twice in ddH₂O. The root tips were finely ground in a 1.5 mL tube. Then, 1 mL of 2% nitric acid was added to it and incubated for one day. Subsequently, the samples were subjected to centrifugation at 12,000 g, after which the supernatant was collected for subsequent analysis. Each sample comprised 100 root tips, and three replicates were prepared.

Gene expression analysis

Primers used for real-time qPCR and T-DNA insertion mutant detection were listed in S9 Table.

Evolutionary analysis

The amino acid sequence of Arabidopsis SSR1 was used to search for orthologous proteins from microorganism to higher organisms using UniPro BLAST and NCBI BLASTP. The similarity of ISC components between E. coli and Arabidopsis (ISU1/2/3, HSCA1/2, HSCB, ADX1/2 and FH) was used as a reference for threshold value (E-value less than 1*10⁻⁴; identity larger than 20%) to filter out the returned proteins displaying low similarity to SSR1. Then, the candidates from different representative species were used as query to search in Arabidopsis database with the same threshold value. If Arabidopsis SSR1 was returned as subject, the candidate would be selected as orthologs.

Maximum Likelihood phylogenetic trees of SSR1 orthologs, NFS1 orthologs, FH orthologs, HSCA2 orthologs, ISU1 orthologs and HSCA1 orthologs were constructed using MEGA 7.0.26 [59] with parameters of the Jones-Taylor-Thornton (JTT) model, complete deletion and 1,000 replicates of bootstrap.

Accession numbers

Sequence data for genes and proteins presented in this article can be found in the Arabidopsis Genome Initiative of GenBank/EMBL database under the following accession numbers: SSR1 (AT5G02130), HSCA2 or mtHSC70-2 (At5g09590), ISU1 (At4g22220), HSCA1 or mtHSC70-1 (AT4G37910), HSP70b (AT1G16030), cpHSC70-2 (AT5G49910), HSP70-1 (AT5G02500), HSP70T-1 (AT1G56410), cpHSC70-1 (AT4G24280), HSP70T-2 (AT2G32120), NFS1 (AT5G65720), FH/frataxin (AT4G03240), Actin 7 (AT5G09810), eIF4a (AT3G19760).