The Suf Iron-Sulfur Cluster Synthesis Pathway Is Required for Apicoplast Maintenance in Malaria Parasites

The apicoplast organelle of the malaria parasite Plasmodium falciparum contains metabolic pathways critical for liver-stage and blood-stage development. During the blood stages, parasites lacking an apicoplast can grow in the presence of isopentenyl pyrophosphate (IPP), demonstrating that isoprenoids are the only metabolites produced in the apicoplast which are needed outside of the organelle. Two of the isoprenoid biosynthesis enzymes are predicted to rely on iron-sulfur (FeS) cluster cofactors, however, little is known about FeS cluster synthesis in the parasite or the roles that FeS cluster proteins play in parasite biology. We investigated two putative FeS cluster synthesis pathways (Isc and Suf) focusing on the initial step of sulfur acquisition. In other eukaryotes, these proteins can be located in multiple subcellular compartments, raising the possibility of cross-talk between the pathways or redundant functions. In P. falciparum, SufS and its partner SufE were found exclusively the apicoplast and SufS was shown to have cysteine desulfurase activity in a complementation assay. IscS and its effector Isd11 were solely mitochondrial, suggesting that the Isc pathway cannot contribute to apicoplast FeS cluster synthesis. The Suf pathway was disrupted with a dominant negative mutant resulting in parasites that were only viable when supplemented with IPP. These parasites lacked the apicoplast organelle and its organellar genome – a phenotype not observed when isoprenoid biosynthesis was specifically inhibited with fosmidomycin. Taken together, these results demonstrate that the Suf pathway is essential for parasite survival and has a fundamental role in maintaining the apicoplast organelle in addition to any role in isoprenoid biosynthesis.


Introduction
Iron-sulfur (FeS) clusters are ancient protein cofactors found in most organisms. These cofactors have a variety of roles including the transfer of single electrons, donation of sulfur atoms, initiation of free radical chemistry, oxygen sensing, and purely structural roles [1,2]. FeS clusters are found in a variety of forms, but the most common are cubane 4Fe-4S, cuboidal 3Fe-4S, and binuclear 2Fe-2S clusters [3]. Proteins typically bind these clusters through cysteine residues, although other amino acids have been shown to be involved in coordinating the cofactor [1].
Proteins containing FeS clusters are typically sensitive to oxygen and the clusters rapidly degrade in extracellular environments. Thus, clusters are synthesized de novo by one of three known FeS biosynthetic pathways. The Nif pathway, the first synthesis pathway described, is primarily found in nitrogen-fixing bacteria [4]. The Isc and Suf pathways are the dominant FeS cluster synthesis pathways found in eukaryotes, and are also present in bacteria and archaea [5,6]. In eukaryotes, the Isc pathway is mitochondrial [5] while the Suf pathway has thus far been found in species harboring a plastid organelle and has been localized to the chloroplast in Arabidopsis thaliana [7,8]. The protozoan parasite Blastocystis, which lacks a plastid, contains components of the Suf pathway in the cytosol [9]. While the protein components of the Isc and Suf machinery are quite different, both pathways follow the same basic steps of sulfur mobilization, cluster assembly, and cluster transfer ( Figure 1).
The Isc and Suf systems both depend on a cysteine desulfurase to mobilize sulfur from L-cysteine. The cysteine desulfurases of the eukaryotic Isc pathway (IscS) and of the Suf pathway (SufS) are only active when in complex with a partner protein ( Figure 1). Isd11, a component of eukaryotic Isc pathways, is essential for mitochondrial FeS cluster synthesis in Saccharomyces cerevisiae and Trypanosoma brucei [10,11,12] but is not present in prokaryotes [13]. In the absence of Isd11, yeast IscS is prone to aggregation [10,11]. Isd11 has a conserved LYK/R motif that is essential for its ability to activate IscS cysteine desulfurase activity [14]. SufS is activated by SufE, an accessory protein which is found in both prokaryotic and eukaryotic Suf pathways. Unlike Isd11, SufE forms a persulfide bond with the mobilized sulfur atom and acts to transfer the persulfide sulfur to the SufBCD assembly machinery [15]. Bacterial SufE has been shown to accelerate the cysteine desulfurase activity of SufS [16,17]. In the presence of SufE, the V max of Escherichia coli SufS is increased eight fold and an additional rate enhancement of 32 fold is observed when the assembly machinery (SufBCD complex) is present to accept the sulfur from SufE [16]. In E. coli, SufE does not interact with the Isc cysteine desulfurase [16] while in A. thaliana SufE has been shown to localize to both mitochondria as well as chloroplasts and serves to activate both cysteine desulfurases [18].
Malaria parasites harbor a plastid organelle called the apicoplast that is thought to have arisen from two sequential endosymbiotic events [19]. The apicoplast harbors biochemical pathways of prokaryotic origin such as type II fatty acid synthesis (FASII), lipoate synthesis, tRNA modification, and 2-C-methyl-D-erythritol 4-phosphate (MEP) isoprenoid biosynthesis [20]. Enzymes in these pathways are predicted to require FeS cluster cofactors. In prokaryotes, lipoate synthase (LipA), the tRNA modification enzyme MiaB, as well as the MEP enzymes IspG and IspH contain 4Fe-4S clusters [21,22,23,24,25]. The activity of these FeS proteins is in turn thought to be dependent on the 2Fe-2S electron transfer protein ferredoxin (Fd) [26,27]. In malaria parasites, only Fd and IspH have thus far been shown to contain FeS clusters [27].
The MEP isoprenoid biosynthesis pathway, the target of the antimalarial fosmidomycin, was recently shown to be essential for the survival of erythrocytic stage malaria parasites [28,29]. Parasites cultured in the presence of the MEP pathway product IPP (isopentenyl pyrophosphate) were no longer sensitive to fosmidomycin. Additionally, supplementation with IPP allowed parasites to survive without the apicoplast organelle, demonstrating that isoprenoids are the only metabolites produced in the apicoplast that are needed outside the organelle [28]. FeS cluster proteins are likely required for the production of essential isoprenoids. However, the synthesis of FeS clusters themselves has not been well characterized in malaria parasites. Only the P. falciparum SufC protein, part of the SufBCD assembly complex, has been studied to date, and was demonstrated to be an active ATPase localized to the apicoplast [30].

Author Summary
Iron is essential for the survival of blood stage P. falciparum and is used primarily in the synthesis of iron-sulfur (FeS) cluster cofactors. We investigated the role that (FeS) clusters play in malaria parasites. We demonstrated that the synthesis of FeS clusters is partitioned between two organelles: the Isc pathway is mitochondrial while the Suf pathway is found exclusively in the apicoplast organelle. Attempts to interfere with the Suf pathway through a dominant negative approach were only successful when parasite cultures were supplemented with an isoprenoid product. This result demonstrates that isoprenoid biosynthesis depends on a functional Suf pathway. Unexpectedly, we also observed the complete loss of the apicoplast organelle when we disrupted the Suf pathway. This phenotype does not result from inhibition of isoprenoid biosynthesis; we treated parasites with high levels of the isoprenoid inhibitor fosmidomycin without any loss of the apicoplast organelle. These results demonstrate that the Suf pathway has a fundamental role in maintaining the apicoplast organelle in addition to any role in isoprenoid biosynthesis. Inhibition of the Suf pathway, which is not found in humans, will block the growth of malaria parasites.
In this report, we investigated two putative FeS cluster synthesis pathways (Isc and Suf), focusing on the initial step of sulfur acquisition. In P. falciparum, SufS and its partner SufE were found exclusively in the apicoplast and SufS was shown to have cysteine desulfurase activity in a complementation assay. IscS and its effector Isd11 were solely mitochondrial, suggesting that the Isc pathway does not contribute to apicoplast FeS cluster synthesis. We disrupted the Suf pathway using a dominant negative mutant of SufC and showed that these parasites only survive when cultured in the presence of IPP. Furthermore, these parasites lack the apicoplast organelle and its organellar genome -a phenotype not observed when isoprenoid biosynthesis was specifically inhibited with fosmidomycin. Taken together, these results demonstrate that the Suf pathway has a fundamental role in maintaining the apicoplast organelle in addition to any role in isoprenoid biosynthesis.

Results
The Plasmodium falciparum genome encodes two distinct FeS cluster synthesis pathways Bioinformatic studies suggest that the genomes of Plasmodium spp. encode both Isc and Suf proteins, including candidate cysteine desulfurases [20,31,32,33,34]. In most eukaryotes the cysteine desulfurases of the Isc and Suf pathways act in complex with the effector proteins Isd11 and SufE, respectively. SufE is essential for Suf FeS cluster synthesis in E. coli [15,16,35], but was originally thought to be absent from malaria parasites [32,34]. More recent bioinformatic studies identified a potential sufE gene [30,36] and a candidate isd11 gene [33,37]. We used the PATS [38], PlasmoAP [39], and PlasMit [40] algorithms to predict the subcellular localization of P. falciparum Suf and Isc pathway proteins ( Table 1). Most of the Suf pathway proteins were predicted to be apicoplast localized while the Isc proteins were predicted to be mitochondrial. In other systems, however, there is precedence for dual localization and crosstalk between components of Isc and Suf pathways. In Arabidopsis, SufE is dually localized to chloroplasts and mitochondria and activates both the Isc and Suf cysteine desulfurases [18]. In E. coli, SufE serves only the Suf pathway; however, the cluster transfer proteins are interchangeable between the pathways. SufA can rescue an IscA knockout, demonstrating that SufA can interact with the rest of the Isc pathway proteins; likewise, IscA can interact with the Suf machinery [35]. In S. cerevisiae, IscS has been localized to the mitochondria as well as the nucleus where it has a poorly defined but essential role [41]. In order to understand how the P. falciparum Suf and Isc pathways are partitioned in the parasite, we localized the IscS and SufS cysteine desulfurases and their effector proteins Isd11 and SufE in blood stage parasites.

SufS and SufE are exclusively located in the apicoplast
We localized the SufS and SufE proteins in P. falciparum by expressing protein constructs fused to a C-terminal green fluorescent protein (GFP) tag. For SufS, the leader peptide (SufS lp ) consisting of the first 59 amino acids was appended to GFP, since this region was predicted by the PATS algorithm [38] to contain the organellar targeting peptide. The mycobacteriophage Bxb1 integrase method was used to generate parasite strains with a single copy of SufS lp -GFP integrated into a specific recombination site in the P. falciparum genome [42,43]. Live fluorescence microscopy demonstrated the presence of GFP fluorescence in an elongated organelle distinct from the parasite mitochondrion, which is typical of apicoplast morphology (Figure 2A). To verify localization to the apicoplast, we performed immunofluorescence analysis using antibodies against the apicoplast marker acyl carrier protein (ACP) (Figures 2B and S1). We also visualized the processing of this fusion protein upon import into the apicoplast by western blot using an antibody against GFP ( Figure 2C). There was a small amount of unprocessed SufS lp -GFP while the majority of the fusion protein ran as a smaller processed species consistent with a cleavage event that occurs upon import into the apicoplast [44]. Full length SufE (SufE fl -GFP) could not be expressed in P. falciparum when driven by the strong calmodulin (CaM) promoter. Therefore, we used the lower strength ribosomal L2 protein (RL2) promoter [45]. SufE fl -GFP parasites displayed the same ramified pattern as SufS expressing parasites by live microscopy ( Figure 3A). Detection by immunofluorescence demonstrated co-localization of SufE fl -GFP with the ACP apicoplast marker ( Figures 3B and S2). As observed for SufS, SufE fl -GFP also appears to be processed, consistent with import into the apicoplast ( Figure 3C). These results demonstrate that SufS and SufE are localized to the apicoplast of erythrocytic stage P. falciparum and that SufE does not appear to be dually localized as observed in A. thaliana [18]. P. falciparum SufS is an active cysteine desulfurase that complements the loss of E. coli SufS In E. coli, the Isc and Suf pathways are partially redundant; deletions of essential elements of either pathway result in conditional lethality while deletion of both pathways is lethal [35]. E. coli deficient in the Suf pathway are more sensitive to iron starvation and oxidative stress than wild type or Isc deficient strains [35]. We used the iron starvation phenotype to test the cysteine desulfurase activity of SufS in E. coli. DsufS E. coli transformed with the mature (processed) form of SufS (pGEXT-SufS 60 ) were able to grow in the presence of an iron chelator (2,29dipyridyl) while DsufS E. coli transformed with empty vector (pGEXT) were unable to grow ( Figure 4). Thus, SufS can complement the loss of EcSufS, demonstrating that the parasite protein has cysteine desulfurase activity. This result also demonstrates that SufS is able to participate in an active E. coli Suf complex, even though mature SufS is only 30% identical to EcSufS.

IscS and Isd11 are exclusively located in the mitochondrion
We next wanted to know whether SufS is the only cysteine desulfurase that functions in the apicoplast. We localized IscS, the only other candidate cysteine desulfurase in malaria parasites, and The parasites were stained with mitotracker to identify mitochondria and DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. GFP fluorescence localizes to an elongated organelle distinct from the mitochondrion in late ring (top panel), late trophozoite or early schizont (middle), and schizont (bottom) stage parasites. B) Co-localization of SufS lp -GFP with endogenous ACP. An antibody specific for GFP co-localizes with aACP antibodies, demonstrating apicoplast localization. C) Proteolytic processing of SufS lp -GFP. An aGFP western blot identifies the mature form (black arrow) as well as a minor population of unprocessed protein (grey arrow) prior to cleavage of the apicoplast transit peptide. doi:10.1371/journal.ppat.1003655.g002 its effector protein Isd11, using the same strategy described above for the Suf proteins. A full-length IscS construct (IscS fl ) fused to GFP co-localized with mitotracker in live fluorescence microscopy ( Figure 5A). Additionally, the 35 amino acid leader peptide of IscS (IscS lp , as predicted by PlasMit [40]) is sufficient to target GFP to the mitochondrion ( Figure S3). We used the same integration The parasites were stained with mitotracker to identify mitochondria and DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. GFP fluorescence localizes to an elongated organelle distinct from the mitochondrion in late ring (top panel), late trophozoite or early schizont (middle), and schizont (bottom) stage parasites. B) Co-localization of SufE fl -GFP with endogenous ACP. An antibody specific for GFP co-localized with aACP antibodies, demonstrating apicoplast localization. C) Proteolytic processing of SufE fl -GFP. An aGFP western blot identifies the mature form (black arrow) as well as a minor population of unprocessed protein (grey arrow) prior to cleavage of the apicoplast transit peptide. doi:10.1371/journal.ppat.1003655.g003 strategy to localize a full-length construct of Isd11 (Isd11 fl ), which in yeast is necessary to activate IscS. Live fluorescence showed complete co-localization with mitotracker indicating the exclusive presence of Isd11 in the mitochondrion ( Figure 5B). Thus, both IscS and Isd11 are mitochondrial and there is no evidence of additional nuclear localization of IscS as reported for S. cerevisiae IscS [41]. Taken together, these results suggest that SufS and SufE are solely responsible for sulfur acquisition for FeS synthesis in the apicoplast and we next attempted to determine whether this activity is essential in blood stage malaria parasites. SufE(C154S)-HA is trafficked to the apicoplast but fails to elicit a dominant negative phenotype A conserved cysteine (at residue 51) in E. coli SufE is required for rapid transfer of sulfur from SufS to the SufBCD complex ( Figure 1), and mutant SufE (C51S) binds to SufS and the The parasites were stained with mitotracker to identify mitochondria and DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. B) Epifluorescent images similar to those in (A) of parasites expressing GFP fused to full length Isd11 (Isd11 fl -GFP). IscS and Isd11 co-localize with mitotracker in all erythrocytic stages: late ring (top panels), late trophozoite or early schizont (middle panels), and schizont (lower panels) stage parasites. doi:10.1371/journal.ppat.1003655.g005 SufBCD complex in a nonproductive manner [15,16]. We attempted to interfere with iron-sulfur cluster synthesis and downstream metabolic pathways by generating an overexpression construct of P. falciparum SufE with the equivalent cysteine substituted with serine, SufE(C154S)-HA. We were able to select parasites expressing SufE(C154S)-HA ( Figure S4A) in the presence of 200 mM IPP, however this parasite line was not dependent on IPP for growth ( Figure S4B). Western blot analysis identified two protein bands for SufE(C154S)-HA, consistent with processing of the apicoplast leader peptide, and immunofluorescence showed that the protein co-localized with the apicoplast marker ACP ( Figure S4C). Thus, SufE(C154S)-HA was expressed and properly trafficked to the apicoplast organelle, but ultimately failed to interfere with apicoplast metabolism enough to make these parasites dependent on IPP supplementation. Although SufE(C154S)-HA failed to act as a dominant negative mutant, this construct helps to confirm the apicoplast localization observed with SufE-GFP in Figure 3.

SufC dominant negative parasites rely on IPP for growth
We designed another dominant negative mutant based on the recent finding that an active site lysine in E. coli SufC is required for ATPase activity and for accumulation of iron on the SufBCD assembly complex [46]. We generated a construct of P. falciparum SufC driven by the calmodulin promoter (CaM) with the active site lysine substituted with alanine, SufC(K140A)-HA CaM . We were unable to select parasites expressing the SufC(K140A)-HA CaM construct, suggesting that expression of this construct is toxic. To bypass this toxicity, we then transfected parasites with the SufC(K140A)-HA CaM construct in the presence of 200 mM IPP and were able to select transgenic parasites. Western blot analysis showed a single band consistent with the expected molecular weight of our dominant negative construct ( Figure 6A). However, unlike the apicoplast proteins shown in Figures 2C, 3C and S4A, there was no indication of apicoplast leader peptide processing with the SufC(K140A)-HA CaM construct. Presumably, this construct was initially targeted to the apicoplast, but over time its expression led to apicoplast dysfunction and a loss of apicoplast leader peptide processing. Consistent with the effects of a dominant negative likely disrupting isoprenoid biosynthesis, these parasites were only able to grow when supplied with IPP ( Figure 6B).

SufC dominant negative parasites no longer contain an apicoplast organelle
We next examined the condition of the apicoplast in dominant negative parasites. We generated a control parasite line expressing the apicoplast targeting peptide of the acyl carrier protein (ACP) fused to the red fluorescent protein mCherry. Parasites expressing ACP-mCherry (ACP-mCh) were cultured for six days with IPP in the presence or absence of 100 nM azithromycin (16 IC 50 ). Azithromycin treatment is known to result in loss of the apicoplast organelle [28,47]. Parasites treated with azithromycin (16 Az) were compared to untreated parasites and dominant negative parasites using a PCR assay. We amplified genes from the apicoplast genome (sufB), the mitochondrial genome (cox1) and the nuclear genome (sufS). All parasite lines maintained their mitochondrial and nuclear genomes. However, in contrast to the wild type parasites, the azithromycin-treated parasites and the SufC(K140A)-HA CaM dominant negative parasites no longer contained sufB, indicating that both strains had lost the apicoplast genome ( Figure 6C). The localization of apicoplast marker ACP in dominant negative parasites, and the SufC(K140A)-HA protein itself, were examined by immunofluorescence and found to be present in multiple foci spread throughout the cell rather than in a single apicoplast organelle ( Figures 6D and S5). The same phenotype was observed when the apicoplast was chemically disrupted [28], confirming that the apicoplast had been similarly disrupted in the dominant negative parasites.
The SufC(K140A)-HA phenotype is specifically caused by the K140A mutation It is possible that high level expression of SufC(K140A)-HA CaM could interfere with secretory pathway function, leading to general toxicity and loss of the apicoplast. To test this, we generated a parasite line expressing SufC(K140A)-HA from the weaker strength RL2 (ribosomal protein L2) promoter [45], SufC(K140A)-HA RL2 . In contrast to the CaM driven construct presented in Figure 6A, we were unable to detect the protein by western blot, even when the blot was loaded with ten-fold more SufC(K140A)-HA RL2 parasite material. Despite the lower expression level, this mutant line was also dependent on continuous supplementation with IPP for growth ( Figure 7A) and PCR analysis indicated that these parasites no longer contained the apicoplast sufB gene ( Figure 7B). Furthermore, immunofluorescence analysis of this line showed localization of ACP to multiple puncta spread throughout the cell similar to that seen with the strong promoter ( Figure 7C). Thus, even when expressed at a lower level, the SufC(K140A)-HA RL2 dominant negative construct still causes the loss of the apicoplast organelle.
We next generated a parasite line expressing wild type SufC-HA driven by the CaM promoter, SufC-HA CaM , to test whether overexpression of this construct would lead to loss of the apicoplast. Unlike the SufC(K140A)-HA CaM line shown in Figure 6, the SufC-HA CaM line is not dependent on IPP for growth, has not lost the sufB gene, and appears to contain a single intact apicoplast organelle (Figure 7). The SufC-HA CaM construct is expressed in this parasite line and co-localizes with ACP in the apicoplast organelle ( Figure S6). Notably, the SufC-HA CaM construct is processed in a manner consistent with apicoplast import ( Figure S6A) where as the SufC(K140A)-HA CaM construct is not ( Figure 6A). Similarly, endogenous ACP protein is processed in the SufC-HA CaM line, but not in the dominant negative line ( Figure S7). Taken together, these data demonstrate that the toxicity of the dominant negative construct is not due to the expression level or the presence of the HA tag, but rather depends solely on the K140A mutation.

Inhibition of the MEP isoprenoid biosynthesis pathway does not result in loss of the apicoplast
Isoprenoids produced by the MEP pathway are the only metabolites produced in the apicoplast that are required outside of the organelle during the erythrocytic stages [28]. It is not known, however, whether the MEP pathway is required for maintenance of the apicoplast itself. To test this, we treated parasites with either azithromycin, to target all apicoplast functions, or fosmidomycin, which specifically targets the MEP pathway. We inhibited the MEP pathway in the presence of IPP by treating ACP-mCh parasites with 50 mM fosmidomycin (1006 IC 50 ), 100 mM fosmidomycin (2006 IC 50 ), 100 nM azithromycin (16 IC 50 ), or no drug for six days. In subsequent growth experiments, only the azithromycin-treated parasites were dependent on IPP for growth ( Figure 8A). Consistent with this growth phenotype, these parasites also lacked the apicoplast gene sufB ( Figure 8B). These results indicate that treatment with azithromycin leads to loss of the apicoplast organelle while treatment with fosmidomycin does not. The ACP-mCherry produced by fosmidomycin-treated parasites and untreated control parasites is trafficked to a single branched organelle consistent with normal apicoplast morphology ( Figure 8C). By contrast, parasites treated in parallel with azithromycin contain ACP-mCherry in multiple foci throughout the cell ( Figure 8C). Thus, inhibition of the MEP pathway with fosmidomycin does not lead to loss of the apicoplast organelle. The dominant negative disruption of the Suf pathway results in similar molecular and cellular phenotypes as the general disruption of the apicoplast by azithromycin and not the specific inhibition of the MEP pathway by fosmidomycin (Figures 6 and 7). These results suggest that in addition to providing FeS clusters for isoprenoid biosynthesis enzymes, the Suf pathway also plays a role in the maintenance of the apicoplast organelle.

Discussion
In the apicoplast of Plasmodium falciparum there are several pathways that are predicted to rely on FeS cluster cofactors (Figure 9), and one of these pathways is known to be essential for erythrocytic stage growth. An early step in MEP isoprenoid synthesis is the target for the antimalarial fosmidomycin [29] which is currently being evaluated in human trials as a partner drug with piperaquine. Recently, it was shown that supplementing parasites with isopentenyl pyrophosphate (IPP, one of the two final products of the MEP pathway) rescues sensitivity to antibiotics targeting apicoplast maintenance (e.g. chloramphenicol, clindamycin, doxycycline), demonstrating that isoprenoid synthesis is essential for blood stage parasite growth [28]. Antibiotic-treated parasites no longer contain an intact apicoplast or the organellar genome, however, these abnormalities should not affect the expression of the MEP pathway proteins. All of the enzymes in the MEP pathway are nuclear encoded and should still be produced under conditions in which the apicoplast is disrupted by antibiotic treatment. This is certainly true for the nuclear encoded apicoplast protein ACP, which is still produced regardless of whether the apicoplast is disrupted (Figures 6, 7 and S7). Unlike ACP, the enzymes that catalyze the penultimate and final steps of isoprenoid synthesis (IspG and IspH, respectively) should both contain FeS clusters [22,27]. As described below, these clusters should not be available in parasites that lack an apicoplast.
SufB is one of the few non-housekeeping genes encoded in the apicoplast genome. In other systems, SufB plays an essential role in FeS cluster assembly and is the scaffold on which the clusters are built [35,46]. When apicoplast maintenance is disrupted, SufB, and thereby FeS cluster synthesis, should be lost; this would then lead to disruption of the MEP pathway. Consistent with these expectations, we found that disruption of the Suf pathway with the SufC(K140A)-HA CaM dominant negative mutant was toxic to blood stage malaria parasites. Parasites were only viable if supplemented with IPP, indicating that disruption of the Suf pathway ultimately leads to loss of the MEP isoprenoid biosynthesis pathway ( Figure 6). Thus, the Suf pathway supports the MEP pathway and is essential for the survival of blood stage malaria parasites.
In addition to the MEP pathway, the apicoplast of malaria parasites harbors a type II fatty acid synthesis (FASII) pathway which is essential for liver stage development [48,49]. The FASII pathway consumes acetyl-CoA [50] which is produced by the apicoplast-localized pyruvate dehydrogenase (PDH) enzyme complex [51]. Like the FASII pathway, a complete PDH complex (composed of four proteins) is essential during liver stage parasite development [52]. PDH is modified with the protein cofactor lipoate [53] which should be required for enzymatic activity. The synthesis of lipoate in the apicoplast is catalyzed by lipoate synthase (LipA), which we have shown contains 4Fe-4S clusters ( Figure S8). These FeS clusters not only need to be synthesized, but they probably also need to be continuously repaired. One of the FeS clusters in E. coli LipA is destroyed every time lipoate is formed, making turnover of LipA dependent on replacing this FeS cluster [54]. Thus, FeS cluster synthesis in the apicoplast should ultimately be required for lipoate synthesis, PDH activity, and the function of the FASII pathway known to be critical for liver stage development in rodent and human malaria parasite species.
In organisms expressing both an Isc and a Suf pathway, such as E. coli, the Isc pathway acts as the default FeS synthesis pathway while the Suf pathway is expressed under conditions of prolonged oxidative stress and iron starvation [35]. It has been suggested that the Suf pathway is more efficient than the Isc pathway under conditions of oxidative stress [55]; this would be an attractive  . Disruption of the isoprenoid pathway does not lead to apicoplast loss. A) IPP growth dependence of azithromycin (Az) and fosmidomycin (FOS) treated parasites. Parasites expressing apicoplast targeted mCherry protein (ACP-mCh) were treated with 1006 IC 50 fosmidomycin (green) and 2006IC 50 fosmidomycin (blue) for six days. Both lines grew in the presence of exogenous IPP (solid lines) and its absence (dashed lines), similar to untreated parasites (red). ACP-mCh parasites treated with 16 IC 50 azithromycin for six days survived when supplemented with IPP (solid black line) but failed to grow when IPP was withdrawn (dashed black line). B) Loss of the apicoplast genome. Genes from the mitochondrial genome (cox1), the apicoplast genome (sufB) and the nuclear genome (sufS) were amplified from the four ACP-mCh lines. The apicoplast gene sufB is present in parasites treated with fosmidomycin (1006FOS and 2006FOS), but is not present in azithromycin-treated parasites (16 Az). C) Loss of apicoplast organelle morphology. Epifluorescent images of live parasites grown in the presence of IPP are shown with nuclei stained with DAPI. The apicoplast as marked by mCherry fluorescence no longer appears as a branched continuous organelle in azithromycin-treated parasites while the apicoplast remains intact in untreated and fosmidomycin-treated parasites. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. doi:10.1371/journal.ppat.1003655.g008 characteristic of the FeS cluster synthesis pathway expressed in oxygen producing compartments such as the plant chloroplast or the ancestral photosynthetic apicoplast [6,56]. However, the modern apicoplast appears to maintain a reducing environment and is highly resistant to oxidative stress [57]. This protective environment may enhance the activity of enzymes sensitive to oxygen, such as LipA ( Figure S8), but it is not clear whether the parasite Suf pathway retains the tolerance to oxidative stress conditions displayed by its orthologs in plants and bacteria.
FeS clusters are synthesized by ancient, highly conserved pathways, at least one of which is found in all organisms [58]. We have confirmed the presence of the Suf pathway in the P. falciparum apicoplast (Figures 2, 3, S1 and S2) and demonstrated the activity of the cysteine desulfurase SufS, the first enzyme in the pathway (Figure 4). In 2003, another group localized IscS as a test of a transfection method [59]. They fused GFP to what was at the time predicted to be the first 135 amino acids of IscS, however, the amino-terminus of the current gene model differs from the sequence used in that study. We repeated the localization using the current gene model which aligns more closely with eukaryotic IscS sequences. P. falciparum IscS and Isd11 both localized exclusively to the mitochondrion (Figures 5 and S3). The subcellular partitioning of the Isc and Suf pathways demonstrates that they function independently of each other, and are likely both essential for erythrocytic stage parasite growth. The same general pattern of organellar partitioning of the Isc and Suf pathways is observed in the only other plastid-containing organism in which both pathway components have been localized, A. thaliana [6]. P. falciparum appears to differ from A. thaliana, however, in that we observe SufE solely in the apicoplast while one of the Arabidopsis SufE homologs Figure 9. Disruption of the Suf pathway leads to apicoplast loss. The Isc pathway is located in the mitochondrion while the Suf pathway is located in the apicoplast. The exclusive division of the two pathways between the two endosymbiont organelles suggests that the FeS-dependent proteins the apicoplast rely solely on the Suf FeS cluster synthesis pathway. When the Suf pathway was disrupted by expressing the dominant negative probe SufC(K140A) the apicoplast was no longer maintained. One or more of the FeS cluster dependent proteins present in the apicoplast could be required for maintenance of the organelle. doi:10.1371/journal.ppat.1003655.g009 appears to be dually localized between the chloroplasts and the mitochondria and has been shown to activate mitochondrial AtIscS [18]. AtIsd11 is only 18% identical to Isd11 from P. falciparum and AtIscS lacks the extended amino terminus of IscS present in P. falciparum and S. cerevisiae. There may be functional differences between these IscS homologs that affect their ability to be stimulated by effector proteins.
FeS cluster modified proteins in the P. falciparum mitochondrion are involved in redox regulation, metabolism, and participate in the electron transport chain. Complex III (cytochrome bc1) is the target of the antimalarial atovaquone, which prevents binding of reduced ubiquinone and also blocks electron transfer from the Rieske type 2Fe-2S cluster, implying that the Isc pathway is essential for blood stage parasite growth [60]. This makes the Isc pathway an attractive drug target, however it is closely related to the host Isc pathway. Closer study of the Isc pathway found in parasites may identify exploitable differences between mitochondrial FeS cluster synthesis in the parasite and in the human host.
The Suf pathway is not found in humans, and the work presented here shows that it is required for the maintenance of the apicoplast organelle. If the Suf pathway was only needed to activate certain MEP enzymes, we would expect disruption of the Suf pathway to have similar effects as inhibition of the MEP pathway. This, however, was not the case. As shown in Figure 8, inhibition of the MEP pathway by the specific inhibitor fosmidomycin does not lead to dependence on IPP for growth, loss of the apicoplast gene sufB, or observable changes in organelle morphology. In contrast to fosmidomycin treatment, disruption of the Suf pathway with the dominant negative mutant SufC(K140A)-HA CaM results in loss of the apicoplast organelle. Dominant negative parasites depend on IPP for growth, have lost the sufB gene, and no longer contain an intact apicoplast organelle (Figures 6).
One possible explanation for this broader phenotype is that high level expression of SufC(K140A)-HA CaM interfered with secretory pathway function, leading to general toxicity. This seems unlikely, however, since these parasites still traffic ACP into punctate foci in the cell (Figures 6C and S5), consistent with the membranebound secretory vesicles observed by Yeh and coworkers [28]. To address this issue, we generated two additional parasite lines. The first expressed the same dominant negative mutant driven by the lower strength RL2 promoter. This parasite line, SufC(K140A)-HA RL2 , displayed the same loss of apicoplast phenotype, demonstrating the potency of the dominant negative SufC(K140A) mutation ( Figure 7). We also generated a parasite line expressing a wild type construct of SufC driven by the strong calmodulin promoter. This SufC-HA CaM construct differs from the toxic dominant negative SufC(K140A)-HA CaM construct by a single amino acid, yet had none of the molecular and cellular phenotypes associated with loss of the apicoplast organelle (Figures 7 and  S6). Thus, the K140A point mutation is solely responsible for disrupting apicoplast metabolism leading to loss of the organelle.
How does the dominant negative mutant interfere with apicoplast metabolism? SufC is known to bind to SufB [30] and presumably forms the SufBCD iron-sulfur cluster assembly complex observed in other organisms (Figure 1). The SufC(K140A) mutant was designed to form a nonproductive complex with endogenous SufB and SufD, thereby limiting the availability of these proteins for cluster assembly. The dominant negative mutant should decrease cluster synthesis, but it could also affect iron homeostasis in the apicoplast, a phenomenon that is difficult to study since organellar iron import and storage mechanisms are not known. We attempted to interfere with the Suf pathway at an earlier step (sulfur acquisition) with the SufE(C154S) mutant, but this construct did not have a dominant negative phenotype, even when overexpressed with the strong calmodulin promoter ( Figure S4). It may be that sulfur acquisition is not the rate limiting step in the parasite Suf pathway or that the SufE mutant does not interact with other Suf proteins as observed in the E. coli system.
The most likely effect of the SufC(K140A) dominant negative mutant is inactivation of apicoplast FeS proteins. Known and predicted FeS proteins are shown in Figure 9, including four FeS enzymes (LipA, IspH, IspG and MiaB) and ferredoxin (Fd). Are any of these proteins likely to be required for apicoplast maintenance during blood stage parasite growth? As described above, LipA is responsible for lipoylating the PDH and ultimately supporting fatty acid biosynthesis in the apicoplast. Since components of the FASII pathway and subunits of the PDH complex (albeit not the lipoylated E2 subunit) have been successfully deleted in blood stage malaria parasites [48,49,52], LipA is presumed to be similarly dispensable and not required for apicoplast maintenance. Although IspH and IspG should be essential for isoprenoid biosynthesis in blood stage parasites, loss of these enzymes should have the same effect as inhibition with fosmidomycin. As shown in Figure 8, inhibition of isoprenoid biosynthesis does not result in loss of the apicoplast organelle.
This result also suggests that the final FeS enzyme, MiaB, is not required for apicoplast maintenance. MiaB presumably functions in conjunction with an upstream enzyme, MiaA, in the maturation of tRNAs. MiaA has not been studied in malaria parasites, but in most eukaryotes and bacteria this enzyme transfers isopentenyl groups to a specific adenosine base in the anticodon loop of certain tRNAs [61,62]. MiaB is a methylthiolase that further modifies the isopentenyladenosine tRNA base with a CH 3 S group [63,64]. If P. falciparum MiaB functions in an analogous way, then its activity depends on isoprenoid biosynthesis, a pathway that we have shown is not required for apicoplast maintenance. Importantly, MiaA enzymes use the MEP pathway product DMAPP (dimethylallyl pyrophosphate) as the source of isopentenyl groups and would not be able to use the IPP that we supply in our parasite culture conditions unless there is an IPP/DMAPP isomerase present. Thus, based on their predicted activities (these enzymes could have additional noncanonical activities), these four FeS enzymes do not appear to be good candidates to explain why disrupting FeS cluster synthesis leads to loss of the apicoplast organelle.
Among the predicted apicoplast FeS proteins in Figure 9, ferredoxin stands out as the most integral to apicoplast function. P. falciparum Fd contains a 2Fe-2S cluster and has been shown to act as an electron donor to IspH [27]. Other apicoplast pathways may also depend on Fd, since it is predicted to be the preferred electron transfer partner for the other apicoplast FeS enzymes (LipA, IspG and MiaB) and may be required to provide reducing equivalents during certain steps of FeS cluster biosynthesis [36,65]. Because of its role in FeS synthesis, reduced Fd metalation could have an exaggerated effect by further limiting the production of its own FeS clusters. Even if Fd is required for FeS synthesis, it still does not provide an explanation for how the apicoplast is lost since the downstream FeS enzymes do not have obvious roles in apicoplast maintenance. Loss of the organelle may instead be linked to how redox balance is maintained in the apicoplast. Fd in conjunction with its associated reductase, ferredoxin-NADP + -reductase (FNR), is the only known redox system in the apicoplast [66]. Perturbation of the Fd/FNR system could lead to increased sensitivity to oxidative stress, as observed in other systems [67]. Since the apicoplast is known to be a highly reducing environment [57], failure of this protective system could lead to oxidative damage, particularly of the organellar DNA, and subsequent loss of the organelle. Regardless of the mechanism, it is clear that Suf pathway dysfunction results in a disruption of apicoplast maintenance. Since the enzymes which comprise the Suf pathway are distinct from anything found in the human host, they are attractive targets for inhibition. The Suf pathway appears to lie at the root of apicoplast metabolic function and inhibition of the pathway should block the growth of blood stage and liver stage malaria parasites.

Generation of P. falciparum transfection constructs
The genes in this study were amplified from gDNA or cDNA prepared from blood stage P. falciparum Dd2 strain parasites and inserted into the pLN-GFP transfection plasmid described by Nkrumah and coworkers [42]. In some cases, the calmodulin (CaM) promoter of pLN-GFP was substituted with the weaker strength ribosomal L2 protein (RL2) promoter [45], and in other cases the GFP tag was removed or replaced with mCherry (mCh) or a hemagglutinin tag (HA).
The iscS gene (PF3D7_0727200) was amplified from gDNA with primers IscS.AvrII.F and IscS.fl.BsiWI.R and inserted into pRL2-GFP, generating plasmid pRL2-IscS fl -GFP (see Table S1 for primer sequences). Nucleotides encoding the 35 amino acid IscS leader peptide (IscS lp ) were amplified from pRL2-IscS fl -GFP vector using primers IscS.AvrII.F and IscS.35.BsiWI.R and ligated into pLN-GFP to generate pLN-IscS lp -GFP. The in-frame intron in isd11 (PF3D7_1311000) was confirmed by amplifying this gene from cDNA with primers Isd11.AvrII.F and Isd11.BsiWI.R and inserting into pRL2-GFP, generating plasmid pRL2-Isd11 fl -GFP. The sufS gene (PF3D7_0716600) was amplified from cDNA using the primers SufS.TOPO.F and SufS.TOPO.R and ligated into cloning vector pET100/D-TOPO (Invitrogen). Nucleotides encoding the leader peptide of SufS were amplified using the primers SufS.AvrII.F and SufS.59.BsiWI.R and ligated into pLN-GFP, generating plasmid pLN-SufS lp -GFP. Amplification of sufE (PF3D7_0206100) from cDNA confirmed the four exon gene model, but consistently resulted in a frame-shifted amplicon. Gene synthesis (GeneArt) was used to generate the sufE gene flanked by AvrII and BsiWI endonuclease sites which were used to subclone into the pRL2-GFP transfection plasmid generating pRL2-SufE fl -GFP.
A transfection vector was created to express mCherry red fluorescent protein in the apicoplast organelle. The gene encoding mCherry was amplified with primers mCh.BsiWI.F and mCh.A-flII.R and inserted into the pLN-TP-ACP-GFP vector described by Gallagher et al. [68]. The resulting transfection vector, pLN-TP-ACP-mCh, encodes mCherry instead of GFP. Constucts SufE(C154S), SufC (PF3D7_1413500) and SufC(K140A) were synthezised (GeneArt) with flanking AvrII and BsiWI sites and inserted into a pLN plasmid modified to have a carboxy-terminal single HA tag, generating pLN-SufE(C154S)-HA, pLN-SufC-HA and pLN-SufC(K140A)-HA. The SufC(K140A)-HA coding region was digested from this plasmid with AvrII and BsiWI and inserted into pRL2 to generate pRL2-SufC(K140A)-HA. P. falciparum transfection and maintenance P. falciparum transfections were performed using the Bxb1 mycobacteriophage integrase system in Dd2 strain parasites containing the attB recombination site [42] in combination with a red blood cell (RBC) preloading technique [43]. Infected red blood cells (iRBC) were first observed between 11 and 27 days after beginning selection with 2.5 mg/mL blasticidin. Insertion of the transgene at the attB site was confirmed by PCR using the primers P1, P2, P3, and P4 (Table S1) as described by Spalding et al. [43]. Genomic DNA from each integrated parasite line was purified and used to verify the transgene sequence with primers GFP.R or pLN.790.R and either RL2.F or CaM.F, as appropriate (Table S1).
Parasites were maintained in human red blood cell culture at 2% hematocrit using the general method described by Trager and Jensen [69]. Briefly, blood stage parasites were cultured in RPMI 1640 supplemented with 10% human serum, 28 mM NaCO 3 H, 25 mM HEPES, and 0.09 mM hypoxanthine. Cultures were gassed with 92% N 2 , 3% O 2 , 5% CO 2 and incubated in sealed 75 cm 2 flasks at 37uC. For the chemical bypass experiments, 0.5 ml or 1 ml parasite cultures were maintained in 24 or 48 well plates and supplemented during daily feedings with 200 mM isopentenyl pyrophosphate (Sigma).

Epifluorescent microscopy and western analysis
Parasite cultures with a parasitemia between 2% and 15% were incubated for 30 minutes at 37uC with 12.5 nM mitotracker CMX-Ros (Invitrogen) and 1 mg/mL 49, 6-diamidino-2-phenylindole (DAPI). Cells were washed three times for 5 minutes at 37uC with RPMI or PBS and then sealed on a slide for observation on a Nikon Eclipse 90i equipped with an automated z-stage. A series of images spanning 4 mm were acquired with 0.2 mm spacing and images were deconvolved with VOLOCITY software (PerkinElmer) to report a single combined z -stack image.
Parasites were fixed and permeabilized for immunofluorescence studies. Live parasites were mixed with 4% paraformaldehyde and 0.0075% glutaraldehyde in PBS and placed on poly-lysine coated glass slides for 30 minutes at room temperature. The slides were then incubated with 1% Triton X-100 in PBS for 10 minutes and then washed three times for 5 minutes with PBS. Sodium borohydride (0.1 g/L in PBS) was used to reduce any remaining unreacted aldehydes followed by three more 5 minutes washes in PBS. The slides were then blocked with 3% bovine serum albumin for an hour and then probed with the appropriate primary antibodies [1:500 rabbit or rat aACP [57], 1:50 Living Colors mouse aGFP JL-8 (Clontech), or 1:50 rat aHA mAb 3F10 (Roche)]. Slides were washed three times for 5 minutes with PBS, and then incubated with the appropriate secondary antibodies [1:3,000 goat aRabbit IgG Alexa Fluor 594, 1:1,000 goat aMouse IgG Alexa Fluor 488 (Invitrogen), or 1:1,000 goat aRat IgG Alexa Fluor 488 (Invitrogen)] for one hour at room temperature. The slides were washed three times for five minutes with PBS, then mounted with Prolong Gold antifade reagent with DAPI (Invitrogen).
Expression of SufS and SufE in transgenic parasites was verified by western blot. Host RBCs from 5 mL cultures at 5-15% parasitemia were permeabilized with 0.2% saponin in PBS for 5 min on ice and then washed repeatedly in PBS until the supernatant was clear. Purified parasites were then lysed in gel loading buffer and parasite proteins were resolved on a NuPage 4-12% Bis-Tris reducing gel (Invitrogen) and transferred onto nitrocellulose. The nitrocellulose membrane was blocked for at least one hour with 5% milk in PBS and probed overnight at 4uC with 1:5,000 Living Colors mouse aGFP JL-8 (Clonetech) in 1% milk. The membrane was washed with PBS three times and probed with 1:20,000 sheep aMouse IgG horseradish peroxidase (HRP) secondary antibody (GE Healthcare) for at least one hour at room temperature. After three additional washes, the blot was visualized with SuperSignal West Pico detection solution (Thermo Scientific) and exposed to film.

Generation of E. coli expression constructs
All constructs expressed in E. coli were cloned into the pGEXT vector which expresses the parasite proteins fused to a cleavable amino terminal glutathione-s-transferase (GST) tag [70]. Mature lipA (encoding residues 89 to 415 of PF3D7_1344600) was amplified from cDNA using Pfu polymerase and primers LipA.EcoRI.F and LipA.PstI.R (Table S1). This amplicon was digested with PstI and EcoRI and ligated into vector pMALcHT [71]. Primers LipA.BamHI.F and LipA.EcoRI.R (Table S1) were used to subclone LipA, generating plasmid pGEXT-LipA 89 . A construct of mature sufS (encoding residues 60 to 546 of PF3D7_0716600) was amplified from vector pET100/D-TOPO (described above) using primers SufS.BamHI.F and SufS.EcoRI.R (Table S1), generating plasmid pGEXT-SufS 60 .
Complementation of E. coli sufS E. coli containing a deletion of sufS (DsufS, Keio collection JW1670) were transformed with either empty vector, pGEXT, or pGEXT-SufS 60 . Each strain was grown overnight at 37uC in MinE medium as modified by Allary et al. [53]. The overnight culture was used to plate 1 mL of 1.0 OD 600 on MinE agar plates containing 100 mM 2,29-dipyridyl. The plates were incubated for 48 hrs at 30uC and inspected for bacterial growth.
Anaerobic expression and purification of LipA BL21 Star (DE3) E. coli containing the pLysE plasmid were transformed with pGEXT-LipA 89 construct produced above. In order to culture the protein in conditions of minimal oxygen, E. coli were grown in flat bottom flasks filled three quarters full with LB medium. When cells reached an OD 600 of 0.6 they were induced with 0.4 mM IPTG for 10 hours at 20uC. Cells were harvested by centrifugation, flash frozen in liquid nitrogen, and stored under the liquid layer. The cell pellet was transferred to a Bactron IV (Shell Labs) anaerobic chamber flooded with 5% hydrogen, 5% carbon dioxide, and 90% nitrogen. A palladium catalyst was used to maintain ,30 ppm oxygen. Cells were resuspended in anaerobic lysis buffer (20 mM Na/K phosphate [pH 7.5], 200 mM NaCl, 2 g/L lysozyme, and 1 mM phenylmethylsulfonyl fluoride [PMSF]) and incubated at room temperature until cell lysis was apparent. After lysis, 2.5 mg/mL DNase I was added and incubated for 30 minutes at room temperature. The lysate was transferred to an air tight container and centrifuged to separate the soluble and insoluble fractions. GST-LipA 89 was purified using a 5 mL GST-Trap HP column (GE Healthcare) connected to a peristaltic pump in the anaerobic chamber.

Characterization of dominant negative and control parasite lines
Plasmids pLN-SufE(C154S)-HA, pLN-SufC(K140A)-HA, pRL2-SufC(K140A)-HA and pLN-SufC-HA were used to generate transgenic parasite lines. As described above, these parasite lines were maintained in the presence of 200 mM IPP. Protein expression was confirmed by western blot using the methods described above with 1:1,000 rat aHA mAb 3F10 (Roche) and 1:20,000 goat aRat IgG horseradish peroxidase (HRP) secondary antibody (GE Healthcare) secondary antibody. Growth assays were conducted in triplicate using 24 well culture plates and initiated at a parasitemia of 0.5%. Over a six day period, parasitemia was assessed by flow cytometry using a FACSCalibur cell sorting machine (Becton Dickinson). Samples of 10 ml from each well were incubated with 10 ml of 5 mM dihydroethidium for 15 minutes at 37uC in the dark. Results were analyzed by FlowJo software (Tree Star Inc., Ashland, OR). Whole cell PCR was used to amplify representative genes from the nuclear (sufS), apicoplast (sufB), and mitochondrial (cox1) genomes (Table S1). Phusion High-Fidelity DNA Polymerase (New England BioLabs) was used in accordance with the manufacturer's directions in 25 mL reactions containing 1 mL of parasite culture. The processing of endogenous ACP was visualized by western blot using the 4% formaldehyde 0.1% glutaraldehyde fixation conditions previously described [57] to prevent ACP from diffusing out of the blot membrane. The blot was probed with 1:5,000 rabbit aACP [57] primary and 1:3,500 donkey aRabbit IgG horseradish peroxidase (HRP) secondary antibody (GE Healthcare).

Generation of the inhibitor-treated ACP-mCherry lines
Parasites transfected with the pLN-TA-ACP-mCherry vector were supplemented with 200 mM IPP and treated with 100 nM azithromycin, 50 mM fosmidomycin, 100 mM fosmidomycin, or no drug for 6 days. All four ACP-mCherry (ACP-mCh) lines were then tested for IPP dependence and analyzed by live epifluorescence microscopy and whole cell PCR as descibed above. Figure S1 Immunofluorescence co-localization of SufS lp -GFP and endogenous ACP. An antibody specific for GFP co-localized with aACP antibodies, demonstrating apicoplast localization in late ring (top panel), late trophozoite or early schizont (middle), and schizont (bottom) stage parasites. The parasites were stained with DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. (TIF) Figure S2 Immunofluorescence co-localization of SufE fl -GFP and endogenous ACP. An antibody specific for GFP colocalized with aACP antibodies, demonstrating apicoplast localization in late ring (top panel), late trophozoite or early schizont (middle), and schizont (bottom) stage parasites. The parasites were stained with DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. (TIF) Figure S3 Subcellular localization of the IscS leader peptide to the mitochondrion of P. falciparum. Epifluorescent images of live P. falciparum erythrocytic-stage parasites expressing GFP fused to the leader peptide of IscS (amino acids . The parasites were stained with mitotracker to identify mitochondria and DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. (TIF) Figure S4 Overexpression of SufE(C154S)-HA in the apicoplast. A) Expression of SufE(C154S)-HA. An aHA western blot confirms the expression of the SufE(C154S)-HA construct and identifies the mature form (black arrow) as well as a minor population of unprocessed protein (grey arrow) prior to cleavage of the apicoplast transit peptide. B) IPP growth dependence of SufE(C154S)-HA parasites. SufE(C154S)-HA expressing parasites survive when supplemented with IPP (solid line) and when IPP is withdrawn (dashed line). Error bars represent SEM of triplicate measurements. C) Co-localization of SufE(C154S)-HA with endogenous ACP. Antibodies specific for the apicoplast marker ACP were used to visualize the apicoplast in blood stage parasites. Co-localization with an antibody specific for the HA tag shows that SufE(C154S)-HA is located in the apicoplast in late ring (top panel), trophozoite (middle), and early schizonts (bottom). Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. (TIF) Figure S5 Subcellular localization of ACP and SufC(K140A)-HA shows disrupted apicoplast morphology. A) Localization of endogenous ACP in SufC(K140A)-HA CaM parasites. Antibodies specific for the apicoplast marker ACP were used to visualize punctate vesicles in late ring (top panel), trophozoite or early schizont (middle), and schizont (bottom) stage parasites. B) Co-localization of SufC(K140A)-HA protein and endogenous ACP. An antibody specific for the HA tag indicates that SufC(K140A)-HA is co-localized with ACP in late ring (top panel), early schizont (middle), and schizont (bottom) stage parasites. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. (TIF) Figure S6 Subcellular localization of wildtype SufC to the apicoplast of P. falciparum. A) Expression of SufC-HA CaM . An aHA western blot confirms the expression of the SufC-HA protein and identifies the mature form (black arrow) as well as a minor population of unprocessed protein (grey arrow) prior to cleavage of the apicoplast transit peptide. B) Colocalization of SufC-HA protein with endogenous ACP. An antibody specific for the HA tag co-localized with aACP antibodies, demonstrating apicoplast localization in late ring (top panel), late trophozoite or early schizont (middle), and schizont (bottom) stage parasites. The parasites were stained with DAPI to identify nuclei. Image z-stacks were deconvolved and then presented as a single combined image. Scale bar = 2 mm. (TIF) Figure S7 Processing of endogenous ACP. Antibodies specific for the Acyl Carrier Protein (ACP) were used to identify the mature form (black arrow) and the unprocessed protein (grey arrow) prior to cleavage of the apicoplast transit peptide. Only the unprocessed trafficking intermediate is present in the dominant negative SufC(K140A)-HA RL2 parasite line, demonstrating loss of this apicoplast function. (TIF) Figure S8 Purification and characterization of recombinant LipA protein. A) Lipoate synthase (LipA), was expressed as a GST fusion protein and purified from E. coli. Analysis by SDS-PAGE shows that GST-LipA 89 migrates close to its predicted molecular weight of 63 kDa. Additional protein bands in the GST-LipA 89 sample (asterisk) cross-react with antibodies specific for GST and likely result from proteolytic cleavage or incomplete translation. B) The UV-Vis absorption spectrum for anaerobically purified GST-LipA 89 displays a broad peak at 440 nm typical of 4Fe-4S proteins. Because the LipA clusters are highly sensitive to oxygen, the signature 4Fe-4S UV-VIS signal degrades over time when exposed to air. (TIF)