Dual Localized AtHscB Involved in Iron Sulfur Protein Biogenesis in Arabidopsis

Background Iron-sulfur clusters are ubiquitous structures which act as prosthetic groups for numerous proteins involved in several fundamental biological processes including respiration and photosynthesis. Although simple in structure both the assembly and insertion of clusters into apoproteins requires complex biochemical pathways involving a diverse set of proteins. In yeast, the J-type chaperone Jac1 plays a key role in the biogenesis of iron sulfur clusters in mitochondria. Methodology/Principal Findings In this study we demonstrate that AtHscB from Arabidopsis can rescue the Jac1 yeast knockout mutant suggesting a role for AtHscB in iron sulfur protein biogenesis in plants. In contrast to mitochondrial Jac1, AtHscB localizes to both mitochondria and the cytosol. AtHscB interacts with AtIscU1, an Isu-like scaffold protein involved in iron-sulfur cluster biogenesis, and through this interaction AtIscU1 is most probably retained in the cytosol. The chaperone AtHscA can functionally complement the yeast Ssq1knockout mutant and its ATPase activity is enhanced by AtHscB and AtIscU1. Interestingly, AtHscA is also localized in both mitochondria and the cytosol. Furthermore, AtHscB is highly expressed in anthers and trichomes and an AtHscB T-DNA insertion mutant shows reduced seed set, a waxless phenotype and inappropriate trichome development as well as dramatically reduced activities of the iron-sulfur enzymes aconitase and succinate dehydrogenase. Conclusions Our data suggest that AtHscB together with AtHscA and AtIscU1 plays an important role in the biogenesis of iron-sulfur proteins in both mitochondria and the cytosol.

Compared to chloroplasts, [Fe-S] biogenesis in plant mitochondria has attracted much less attention. Kushir et al pioneered this field by identifying the Arabidopsis Sta1 as an Atm1p-like ABC transporter of yeast supporting the maturation of [Fe-S] protein in mitochondria [39]. Further efforts have now identified several components in plant mitochondria that are evolutionarily conserved and similar to that of the yeast ISC-like (iron sulfur cluster) system [31,40,41]. In this context it is worth mentioning that the SufE-like protein AtSufE is localized to both mitochondria and chloroplasts where it activates both mitochondrial and chloroplastic cysteine desulfurase [27] indicating a possible spatial link between [Fe-S] biogenesis systems in Arabidopsis [42].
In the plant cytosol, [Fe-S] biogenesis is much less well understood. However, recent work by Balk and colleagues [43] has started to unravel cytosolic [Fe-S] biogenesis. They report that AtNBP35, similar to the NBP35 protein which is part of the cytosolic Cfd1-Nbp35 complex mediating Fe-S cluster assembly in yeast [12], has retained similar Fe-S cluster binding and transfer properties and performs an essential function [43]. However, much work is still required in order to assemble a model of [Fe-S] biogenesis in the plant cytosol.
In bacteria and yeast, the HscA/Ssq1 chaperones and the HscB/Jac1 co-chaperones are important elements of the ISC-like system. HscA/Ssq1 are ATPases, stimulated by the J-type cochaperone HscB/Jac1 and have been shown to interact with the scaffold protein IscU/Isu, which is regulated by HscB/Jac1 by binding to IscU/Isu to assist [Fe-S] delivery to the chaperone [12,44]. Yeast Jac1, Ssq1 and Isu have been confirmed to be mitochondrial proteins [12].
Here we demonstrate that Arabidopsis contains a functional AtHscA1/AtHscB/AtIscU1 protein cluster involved in [Fe-S] protein biogenesis. In contrast to yeast, the AtHscA1/AtHscB/ AtIscU1 protein cluster is localized to both mitochondria and the cytosol of Arabidopsis suggesting a dual action between these two spatially separate compartments.

AtHscB can rescue yeast Jac1knockout mutant
A full-length cDNA (759 nt) encoding the At5g06410 open reading frame was cloned and its predicted amino acid sequence compared to E. coli HscB and yeast Jac1 showing 30% and 24% identity, respectively ( Figure S1A). The At5g06410 amino acid sequence contains the HPD motif, essential for Jac1 function in yeast [45] and a predicted 59 amino acid N-terminal mitochondrial targeting peptide according to the CBS Prediction Server [46,47]. As the name Jac1 in Arabidopsis has been assigned to another protein we named At5g06410 AtHscB.
AtHscB and AtHscA1 are localized to both mitochondria and the cytosol To analyze the subcellular localization of AtHscB we fused AtHscB to the N-terminus of YFP (Yellow Fluorescence Protein) and transiently expressed this transgene in tobacco cells ( Figure 2). As expected, based on the predicted mitochondrial targeting signal ( Figure S1A), fluorescence was observed in mitochondria ( Figure 2C). To verify that the fluorescent signal observed was indeed mitochondrial we performed mitotracker experiments which showed that the red fluorescence of mitotracker colocalized with the YFP signal ( Figure 2C). However, AtHscB-derived YFP fluorescence was also observed in the cytosol ( Figure 2C). To verify that the observed AtHscB dual localization also occurs in Arabidopsis we expressed the AtHscB-YFP fusion protein in transgenic Arabidopsis plants with identical results ( Figure S1B). To further confirm the dual localization pattern of endogenous AtHscB protein, we performed immunogold labeling experiments in wild-type Arabidopsis leaves using electron microscopy and an anti-AtHscB antibody. Figure 2E shows the presence of gold particles in both mitochondria and the cytosol strengthening the observed dual localization of AtHscB.
To verify the specificity of the anti-AtHscB antibody we performed western blot analysis of total cell extract from E. coli expressing AtHscB and from wild-type Arabidopsis showing the presence of one protein band ( Figure S1D). In addition, we performed immunogold labeling experiments using the preimmune serum showing no signal ( Figure S1E).
In Arabidopsis there are two HscA/Ssq1-like genes (At4g37910 and At5g09590) whose products are predicted to be mitochondrial  Table S1). doi:10.1371/journal.pone.0007662.g001 ( Figure S2) and we named them AtHscA1 and AtHscA2, respectively. As for AtHscB, transient expression analysis of an AtHscA1-YFP fusion protein in tobacco cells showed fluorescence in both mitochondria and cytosol ( Figure 2D) as did stable expression of the same fusion protein in transgenic Arabidopsis plants ( Figure S1C). Combined this suggest that AtHscB and AtHscA1 may play important roles in both mitochondria and the cytosol in contrast to observations made in yeast where both proteins are exclusively mitochondrial.

AtHscB can interact with AtIscU1 in mitochondria and in the cytosol
The scaffold protein IscU/Isu has been shown to play a key role in [Fe-S] biogenesis [48,49]. To analyse whether AtHscB could interact with AtIscU1, AtIscU2, and AtIscU3 we performed Yeast Two-Hybrid (YTH) and Bimolecular Fluorescence Complementation (BiFC) assays. YTH demonstrated that AtHscB did not interact with AtIscU2 or AtIscU3 but could interact with AtIscU1 and that AtIscU1 could also interact with itself ( Figure 3A). Further, transient BiFC experiments in tobacco cells ( Figure 3C) and stable BiFC experiments in transgenic Arabidopsis plants ( Figure 3D, Figure S1F) demonstrated clear interaction between AtHscB and AtIscU1 not only in mitochondria but also in the cytosol.
These results were surprising as AtIscU1-GFP fusion experiments have previously shown that AtIscU1 is exclusively localized to mitochondria [40,41,50]. Indeed we also show here that an AtIscU1-YFP fusion protein appears exclusively targeted to mitochondria ( Figure 2B). However, the dual localization of AtIscU1, when in combination with overexpressed AtHscB in BiFC assays, suggest that AtIscU1 is retained by the high levels of AtHscB in the cytosol through direct protein-protein interactions ( Figure 3C, D). In wild-type Arabidopsis, AtHscB is only expressed at low levels which would most probably not allow for sufficient retention of an AtIscU1-YFP fusion protein and hence no detection of AtIscU1-YFP-mediated fluorescence in the cytosol. Combined our data demonstrate that AtHscB can interact with AtIscU1 in both mitochondria and in the cytosol.

AtHscA1 can complement yeast Ssq1
AtHscA1 has high similarity to HscA/Ssq1 of bacteria/yeast ( Figure S2). To confirm that AtHscA1 can functionally complement HscA/Ssq1, the Ssq1 knockout mutant Dssq1 [48] was transformed with pGADT7-AtHscA1 and both strains placed on YPD medium and on YPD containing 4 mM H 2 O 2 and incubated at 34uC for 4 days ( Figure 4A). On YPD medium both Dssq1 and Dssq1/AtHscA1 grew well as expected. By contrast, ssq1 failed to grow on YPD containing H 2 O 2 whilst Dssq1/AtHscA1 showed

AtHscA1 is an ATPase stimulated by AtHscB and AtIscU1
HscA in bacteria and Ssq1 in yeast are both ATPases [49,51] however, in plants the enzyme activity of AtHscA1 is unknown. We heterologously expressed AtHscA1 in E. coli Rosette(DE3)pLysS by auto-induction and affinity purified the protein to .95% purity ( Figure 4B). Using c-P 32 labeled ATP as a substrate the purified AtHscA1 can clearly hydrolyze ATP ( Figure 4D) giving a Km of 49.5 mM and a Vmax of 1.08 mM/min ( Figure 4C).
AtHscB and AtIscU1 were also heterologously expressed and purified as for AtHscA1 ( Figure 4B). To test the effect of AtHscB or AtIscU1 on AtHscA1 enzyme activity we added purified AtHscB or purified AtIscU1 to AtHscA1 in a 1:1 stochiometric ratio and performed ATPase assays. From these experiments it was shown that individually AtHscB and AtIscU1 can promote AtHscA1 ATPase activity approximately two-fold. In contrast, when AtHscB, AtIscU1 and AtHscA1 were combined in equal stochiometric concentrations, the ATPase activity of AtHscA1 increased 22-fold ( Figure 4D). These experiments clearly demonstrate that AtHscA1 is an ATPase and that in combination AtHscB and AtIscU1 can significantly stimulate AtHscA1-mediated ATP hydrolysis.

AtHscB expression patterns and AtHscB T-DNA mutant phenotypes
A 1041 nucleotide DNA promoter fragment (1961160-1962200 nt) directly upstream of the AtHscB start codon was PCR-cloned into the b-glucoronidase (GUS) binary vector pBADG and transformed into wild type Arabidopsis. GUS staining of T2 lines showed that AtHscB is universally expressed at low levels but with relatively high levels of expression in anthers and trichomes (Fig. 5A).
An Arabidopsis T-DNA insertion line (SALK_085159) was identified and analyzed by PCR with the T-DNA specific primer LBb1 and the AtHscB-specific primers LP585159 and RP585159. Two PCR fragments were obtained and sequenced revealing the presence of two T-DNAs inserted 0 and 5 nt downstream of the AtHscB stop codon ( Figure 5B). Although the seed set was dramatically reduced in N585159 plants ( Figure 5I) several homozygous N585159 plants were isolated ( Figure 5B) showing severe down-regulated AtHscB transcript and undetectable AtHscB protein ( Figure 5C). Phenotypic analysis revealed that homozygous mutants had stems with a shiny bright green appearance, indicating the absence of the epicuticular wax layer ( Figure 5D), a similar phenotype to that observed in CUT1 sense suppressed plants [52]. Indeed, scanning electron microscopy revealed that the stem surface contained much fewer wax crystals than wild type plants ( Figure 5E). Homozygous mutants are also conditional sterile, in agreement with the observed reduced seed set, as siliques fail to develop under normal growth conditions whilst in a humid environment siliques develop as in wild-type ( Figure 5G, Figure  S3A). These data suggests that AtHscB-deficiency results in reduced or diminished very-long-chain fatty acids (VLCFAs) biosynthesis. To confirm that the observed mutant phenotypes were caused by AtHscB-deficiency, the homozygous mutant was transformed with wild-type AtHscB followed by phenotypic characterization. More than 95% (20 out of 21) of transformed resistant plants showed a wild type phenotype ( Figure 5D, Figure  S3B, 3C) confirming that AtHscB-deficiency is responsible for the observed mutant phenotypes. Due to the high level of AtHscB expression in trichomes ( Figure 5A), we monitored trichome development in homozygous N585159 mutant plants. In agreement with the AtHscB gene expression patterns homozygous mutants not only have fewer trichomes than wild-type ( Figure 5F) but these were also smaller in size and often distorted ( Figure 5H).
To test whether AtHscB indeed has an effect on iron sulfur proteins in Arabidopsis, we assayed both aconitase and succinate dehydrogenase (SDH) activities in wild type plants and in homozygous N585159 mutant plants. Figure 6 shows that in the homozygous mutants both aconitase and SDH enzyme activities are reduced to approximately 10% of wild-type levels. These data, combined with other finding shown in this study, support the notion that AtHscB has a role in iron sulfur protein biogenesis in Arabidopsis.

Discussion
In this study we provide data suggesting that the Jac1-homolog AtHscB is involved in iron sulfur protein biogenesis in Arabidopsis. Several points of evidence demonstrates that AtHscB is indeed a Jac1-like protein: (i) AtHscB can rescue the yeast Djac1 mutant  Figure 6). Aconitase in yeast seems to have a dual function as it has an influence on the glyoxylate shunt in the cytosol and the TCA cycle in mitochondria, corroborating our finding that aconitase activity is highly diminished in an AtHscB mutant, lacking AtHscB in both the cytosol and mitochondria.
Because of the clear relationship between AtHscB and AtHscA1 in Arabidopsis we also demonstrated, through yeast complementation experiments, that AtHscA1 is an HscA/Ssq1-like protein (Figure 4 and Figure S2). Arabidopsis contains a second HscA/ Ssq1-like protein (At5g09590), AtHscA2, and it will be interesting to examine its connection with AtHscB and its possible role in iron-sulfur protein biogenesis ( Figure S2).
Surprisingly, both AtHscB and AtHscA1 show dual localization where they are present in both mitochondria and the cytosol as revealed by YFP fusion protein localization studies and immunogold labeling experiment (Figure 2 and Sup. Figure 1B-E). The fact that immunogold experiments demonstrate that endogenous AtHscB is present in both mitochondria and the cytosol eliminates the dual localization being due to overexpression of the transgene. Many dual localized proteins have been shown to have a low MitoProtII score and the AtHscB mitochondrial targeting sequence has a low MitoProtII value (http://ihg.gsf.de/ihg/mitoprot.html). Based on our findings it is reasonable to suggest that AtHscB, AtHscA1 and AtIscU1 may function in both mitochondria, and through retention of AtIscU1 by AtHscB (Figure 3), in the cytosol. The dual localization of the AtHscB/AtHscA1/AtIscU protein cluster suggests spatial coordination of [Fe-S] biogenesis in plants between subcellular compartments which has been suggested for sulfur acquisition within the ISC system [42]. The fact that AtHscB possibly retains AtIscU in the cytosol implies that AtHscB may act as a control point for the balance of AtHscB/AtHscA1/AtIscU1mediated [Fe-S] biogenesis in Arabidopsis. Based on the data presented AtIscU1 is only detected in the cytosol when AtHscB accumulates and it will be interesting to examine cytosolic retention of AtIscU1 during plant development and during conditions that favor cytosolic [Fe-S] biogenesis. Progress has been made on the biogenesis of cytosolic [Fe-S] [43] where AtNBP35 plays a key role in this process. How AtNBP35 relates to AtHscB/AtHscA1/ AtIscU1 in the cytosol remains an exciting challenge.
Based on the evolutionary conservation of AtHscB and AtHscA1 it is reasonable to assume that the interaction of AtIscU1 with AtHscA1 is regulated by AtHscB as seen in bacteria and we have shown here that AtHscB stimulates the ATPase activity of the AtHscA1/AtIscU1 complex ( Figure 4D). However, the molecular mechanism and function of AtHscB-mediated enhancement of AtHscA1 ATPase activity is unknown but may contribute to substrate specificity.
The fact that AtHscB/AtHscA1/AtIscU1 are localized to the cytosol and mitochondria may indicate some form of crosstalk between [Fe-S] biogenesis systems in these two spatially separated subcellular compartments. Such crosstalk would be essential in light of the importance of maintaining iron and sulfur homeostasis.
Iron-sulfur proteins are involved in many fundamental and diverse biological processes dictating that the consequences of inappropriate [Fe-S] biogenesis may have dramatic and pleotrophic effects on plants. Indeed several reports have shown severe developmental defects in Arabidopsis in response to inappropriate [Fe-S] biogenesis such as embryo lethality in both AtSufE and AtSufC loss-of-function mutants [26,27]. AtHscB mutant plants have dramatically reduced seed set ( Figure 5I) and this is in good agreement with previous studies showing that reduced SDH activity leads to altered gametophytic development [53]. However and somewhat surprisingly, dramatically reduced levels of AtHscB also results in relatively specific phenotypic consequences. The molecular basis for the waxless phenotype (decreased wax crystals and conditional sterility) of the homozygous AtHscB mutant ( Figure 5E) is intriguing and links iron-sulfur protein biogenesis to VLCFA biosynthesis in Arabidopsis. The role of AtHscB in VLCFA biosynthesis is unclear however VLCFA biosynthesis most probably involves [Fe-S] enzymes that are compromised in plants lacking AtHscB. Indeed it will be interesting to investigate VLCFA biosynthetic enzymes in terms of [Fe-S] content. Similarly, the altered trichome structure in AtHscB mutant plants, accompanied by the fact that AtHscB expression occurs in these cells ( Figure 5A), suggests that [Fe-S] biogenesis is important for this specialized process.
The intriguing phenotypic effects of AtHscB deficiency on plant development and the unexpected dual localization of both AtHscB and AtHscA1, highlights the complex nature of [Fe-S] biogenesis in Arabidopsis. It further implies that although iron-sulfur protein biogenesis represents a fundamental biological process, proteins involved in [Fe-S] biogenesis are also highly specific in terms of controlling defined developmental processes with seemingly intricate regulatory circuits.

Plants and growth conditions
Wild-type Arabidopsis, transgenic Arabidopsis and the AtHscB T-DNA insertion mutant Salk_085159 (N585159) were grown at 21uC with 16 h of light (100 mmol photons m -2 s -1 ) per day and 60% humidity unless otherwise stated. The AtHscB T-DNA insertion mutant Salk_085159 (N585159) [54] was obtained from NASC (European Arabidopsis Stock Centre, Nottingham).

Subcellular localization and BiFC assays
pWEN18, pWEN-N-YFP (pWEN-NY) and pWEN-C-YFP (pWEN-CY) containing AtHscB, AtIscU1 and AtHscA1 were transiently expressed in tobacco cells. To test for AtHscB and AtIscU1 interactions pWEN-AtHscB-NY/pWEN-AtIscU1-CY or pWEN-AtHscB-CY/pWEN-AtIscU1-NY were used to bombard and transiently expressed in tobacco leaves. For stable expression analysis, AtHscB, AtHscA1 and AtIscU1 fused to YFP, NY and CY, were cloned into AscI/PacI of pBA002 and pER10 and transformed into Arabidopsis [56]. All fluorescence analysis was performed on a Nikon TE-2000U inverted microscope (Nikon, Japan) and Volocity II software (Improvision, UK). For electron microscopy/immunogold analysis standard protocols were followed using a JEOL 1220 (Electron Microscopy Laboratory, University of Leicester, UK) and 17 images analyzed. Pre-immune serum was used as a control. An AtHscB polyclonal antibody was generated following standard protocols as instructed by Harlan Laboratories.

Yeast Transformation
pGBKT7 and pGADT7 containing AtHscB and AtIscU1 and control vector controls were transformed into HF7c and tested for His auxotrophy restoration following the Matchmaker two-hybrid system III manual (Clontech).