A mitochondrial ATM1 homolog is conserved across Apicomplexa

To confirm the identity of ATM1 homologs in P. falciparum and T. gondii, we conducted homology searches using sequences from ScATM1, human ABCB7, and AtATM3 as queries [51] in PlasmoDB and ToxoDB databases. TGGT1_269000 was identified as the putative T. gondii ATM1; two homologs, PfABCB2 (PfMDR2, PF3D7_1447900) and PfABCB6 (PfMDR6, PF3D7_1352100) were initially considered in P. falciparum. However, a mitochondrial targeting sequence (MTS) was not identifiable in PfABCB2 whereas PfABCB6 had a predicted MTS (Mitoprot II prediction scores of 0.0218 and 0.9434 for PfABCB2 and PfABCB6, respectively). Phylogenetic analysis with ATM1 and HMT (heavy metal tolerance) family proteins from Drosophila melanogaster, yeast, human and Arabidopsis thaliana also suggested closer proximity of PfABCB2 to HMTs compared to PfABCB6 (S1A Fig). PfABCB2 formed a separate clade with eukaryotic HMTs, sharing a more recent common ancestor with them. PfABCB2 has been previously reported to localize to the parasite plasma membrane/food vacuole [52, 53]. It has also been demonstrated to play a role in heavy metal resistance in the parasite by mediating active extrusion of Cd²⁺ out of the cytosol [54], and likely belongs to the closely related HMT family. Thus, PfABCB6/PfMDR6 was identified as the putative PfATM1.

Subcellular localization data obtained from isotope tagging (LOPIT) indicated mitochondrial localization for TgATM1 [38]. Additionally, whole-genome mutant screens in both parasites revealed a very low fitness score for these proteins, suggesting an important role in parasite survival [39, 55]. Homologs of these proteins were identified in other apicomplexan parasites through BLASTp searches, including the ABC transporter of Theileria orientalis, ABCB25 of Babesia bovis, and ATM1-like protein of Cryptosporidium parvum (S1 Table). Overall, this transporter is highly conserved within the phylum, even in species harboring a remnant mitochondrion (mitosome) that lacks a functional heme synthesis pathway.

In contrast to the well-characterized α-proteobacterium Novosphingobium aromaticivorans ATM1, PfATM1 and TgATM1 exhibit a substantial unconserved N-terminal extension (NTE) preceding the N-terminal domain (NTD) and transmembrane region (Figs 1A and S1B). Alignment with homologs from yeast, human, and plants revealed the preservation of crucial structural components of ATMs (S1B Fig). Conserved glutathione binding sites and residues demonstrated to be complexed with [2Fe-2S] in NaATM1 and other homologs were identified between TM4 and TM5 and downstream of TM5 [16, 56] (S1B Fig). The C-terminal ATPase domain (CTD) contains conserved Walker A, Walker B and ABC signature motifs. PfATM1 and TgATM1 carry diverged insertions between the transmembrane region and CTD in the former, and between TM1 and TM2 in the latter (Figs 1A and S1B). The evolutionary proximity of apicomplexan ATM1 homologs was inferred by constructing a maximum likelihood phylogenetic tree (Fig 1B). The apicomplexan sequences formed a separate clade branching at the root of the tree and are sister to the basal clade of all other ATM1 homologs.

Download:

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

Fig 1. Phylogeny and localization of PfATM1 and TgATM1.

(A) Domain organization and N-terminal extensions (NTE) in ATM1 homologs from apicomplexan parasites (P. falciparum, T. gondii, C. parvum), Saccharomyces cerevisiae, A. thaliana, Homo sapiens and the bacterium N. aromaticivorans. (B) Phylogeny of ATM1 homologs generated by PhyML 3.0. The tree was rooted at the EcHlyB branch. Bootstrap values of 1000 replicates are indicated at each branch. (C) Western blot of a TgATM1-Ty-U1 parasite lysate probed with anti-Ty antibody. (D) Immunofluorescence assay of parental (DiCreU1) and TgATM1-Ty-U1 parasites stained with anti-Ty antibody and a mitochondrial marker (mitochondrial heat shock protein, mHSP70). The degree of colocalization is given. Images shown in C and D are representative of three independent biological replicates. (E) Western blot of P. falciparum lysate probed with anti-PfATM1 serum recognizes a specific band at the expected size of unprocessed PfATM1 (123 kDa) and a closely migrating lower band likely representing processed protein. (F) Immunofluorescence assay for subcellular localization of PfATM1 at different intra-erythrocytic stages. Scale (0–3 μm) is shown in the merged image. DAPI is the nuclear stain, Mitotracker Red is the mitochondrial marker dye, anti-PfATM1 antibodies and anti-PfHU serum detect PfATM1 and the apicoplast marker PfHU, respectively. Colocalization rate is given for overlay of PfATM1 and Mitotracker Red signals.

https://doi.org/10.1371/journal.ppat.1012593.g001

To assess the localization and significance of TgATM1, we inserted 3-Ty tags into the native locus for localization studies, alongside loxP sites and a U1 recognition site for conditional downregulation upon rapamycin (Rapa) induction [57] (S2A Fig). The modification was performed using CRISPR/Cas9 editing [58] in a line stably expressing the dimerizable Cre recombinase (DiCreU1) and the resulting strain was labelled as TgATM1-Ty-U1 [57]. Transfected parasites were selected and cloned through limited dilution. Correct integration in the 3′ untranslated region (UTR) of TgATM1 was validated in a clonal population by genomic PCR (S2B Fig, S2 Table). TgATM1-Ty was detected by western blot in TgATM1-Ty-U1 parasites, running slightly smaller than its anticipated size of 126 kDa, a common occurrence for transmembrane proteins (Fig 1C). Immunofluorescence assays (IFAs) coupled with confocal microscopy revealed that TgATM1-Ty exhibited co-localization with the mitochondrial marker mitochondrial heat shock protein 70 (mHsp70) [59] (Fig 1D), confirming its mitochondrial localization consistent with the previously determined localization by isotope tagging (LOPIT) [38].

To ascertain the subcellular localization of PfATM1, we expressed its unconserved NTE domain (PfATM1-NTE, 33 kDa) as a 6X-His tagged protein in E. coli (S3A Fig) and utilized the purified protein to produce antibodies in rabbits. The antiserum detected a distinct band around ~123 kDa, aligning with the anticipated size of unprocessed PfATM1 (Fig 1E). Additionally, a closely migrating lower band was observed, possibly indicating a processed or degraded product. IFAs using purified PfATM1 Abs showed clear partitioning of PfATM1 signal to the parasite mitochondrion, merging with Mitotracker Red across various asexual stages (Fig 1F). No overlap was seen with the apicoplast marker PfHU [60] (Fig 1F), confirming that PfATM1, like TgATM1, localizes to the mitochondrion.

TgATM1 and PfATM1 are required for optimal parasite growth

Next, we employed conditional downregulation to probe the importance and function(s) of TgATM1. Effective depletion of TgATM1-Ty upon Rapa induction in TgATM1-Ty-U1 parasites was validated via IFA (Fig 2A), with no detectable signal observed after 48 hours of Rapa treatment. Efficient downregulation of TgATM1-Ty was further confirmed by western blot, revealing a gradual depletion and undetectable levels after 72 hours of Rapa (Fig 2B). Intriguingly, two bands were observed for TgATM1 in some but not all western blots, with the lower band at 70 kDa likely reflecting a degradation product or a processed form. Based on the kinetics of TgATM1 downregulation, most of the subsequent assays were performed after 72 h. To assess the significance of TgATM1 in the lytic cycle of T. gondii, we conducted plaque assays using human foreskin fibroblasts (HFFs) infected with the parental strain (DiCreU1) and TgATM1-Ty-U1 parasites with or without Rapa (+Rapa and −Rapa), respectively. Monolayers were fixed and stained with crystal violet 7 days post infection. Depletion of TgATM1 severely affected the formation of plaques confirming its crucial role in one or more steps of the parasite’s lytic cycle (Fig 2C). To assess if the hindered lytic cycle results from a growth deficiency, we quantified the parasite number per vacuole after 24 hours of intracellular growth and varying durations of Rapa treatment (Fig 2D). A subtle but significant growth defect was evident after just 24 hours of TgATM1 downregulation. The average number of parasites per vacuole decreased from >9 in the controls (DiCreU1 +Rapa and TgATM-Ty-U1 −Rapa) to 7.7, 4.7 and 4.1 after 24, 48 hours and 72 hours of Rapa treatment, respectively (Fig 2D). These findings validate the significance of TgATM1 in the parasite’s lytic cycle and replication.

Download:

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

Fig 2. TgATM1 and PfATM1 are essential for parasite growth.

(A) Immunofluorescence assays of TgATM1-Ty-U1 parasites stained with anti-Ty antibody and a mitochondrial marker (mitochondrial heat shock protein, mHSP70) following 0 or 48 hours of rapamycin (Rapa) treatment. (B) Western blot of parental (DiCreU1) and TgATM1-Ty-U1 parasite lysates following varying durations of Rapa treatment. The membrane was probed with anti-Ty antibody and anti-catalase antibody as loading control. Two major bands were observed for TgATM1-Ty, labelled TgATM1-Ty¹ and TgATM1-Ty². (C) Stained host cell monolayers revealing plaques formed by parental (DiCreU1) or TgATM1-Ty-U1 parasites over seven days in presence or absence of Rapa. A-C: Representative images are shown from one of 3 independent biological replicates. (D) Intracellular growth assay displaying the proportion of vacuoles containing 2, 4, 8, 16 and 32 parasites after 24 hours of intracellular growth and varying durations of Rapa treatment for parental (DiCreU1) and TgATM1-Ty-U1 parasites. The means and standard deviations from three independent biological replicates are shown. A two-sided Student’s t-test was used to compare the average number of parasites between the indicated conditions. Significant differences (p <0.05) are highlighted in bold.

https://doi.org/10.1371/journal.ppat.1012593.g002

The essentiality of PfATM1 in parasite physiology is suggested from results of random piggyBac transposon mutagenesis screen in P. falciparum [55]. There are, however, contradictory reports concerning the P. berghei ortholog (PbABCB6, PBANKA_1364800); it has been deemed essential through direct genetic knockout (PlasmoGEM, [46]), but another study described that its deletion does not affect viability of blood stage parasites [47]. We thus attempted downregulation of PfATM1 using the Vivo-morpholino approach, a method previously reported for targeted translation inhibition in P. falciparum [61]. A specific Vivo-morpholino (PfATM1-V_m) was designed (by GeneTools, USA) to inhibit the translation of PfATM1 (S3B Fig). Treatment of P. falciparum 3D7-infected cells with PfATM1-V_m resulted in lower levels of the target at 72 hours and 96 hours post treatment (S3C and S3D Fig). Quantitative analysis of the western blots indicated significant reduction (~57%) in actin-normalized PfATM1 levels in PfATM1-V_m treated cells at the 96 hours’ time point when compared with untreated cells and control V_m (S3D Fig). The control V_m is a random sequence with no significant identity to the P. falciparum genome. Comparison of erythrocytic stages at different time points post-treatment revealed a growth delay in treated cells, particularly noticeable at 96 and 108 hours (S3E and S3F Fig). At 96 h, untreated and control V_m-treated cells progressed to the early/mid-trophozoite stage, while PfATM1-V_m-treated cells remained predominantly at the ring stage. By 108 hours, control cells were mainly at the mid- to late-trophozoite stage, whereas PfATM1-V_m-treated cells were still in the early trophozoite stage (S3E and S3F Fig). Neither a significant reduction in parasitemia (S3F Fig), nor apparent alterations in mitochondrial morphology were observed in PfATM1-V_m treated cells at these time points (S3G Fig). At the later time point of 144 h, control and V_m-treated cells showed similar stages suggesting that PfATM1-V_m had lost its effect after the third cycle (S3E and S3F Fig). This was corroborated by western blot analysis, which showed that the levels of PfATM1 in PfATM1-V_m-treated cells returned to normal by 144 hours (S3C and S3D Fig). Reducing PfATM1 levels results in a slowdown of asexual parasite growth in blood stages. Direct genetic knockdown of PfATM1 (e.g., using the aptamer/TetR-DOZI system) would be required for conclusive evidence of its essentiality in the parasite.

Taken together, these findings reveal the essential role of TgATM1; the slow growth observed in PfATM1 knockdown and the reported essentiality of P. berghei ABCB6/MDR6 [46] supports their role as mitochondrial transporters required for optimal development of both parasites.

TgATM1 and PfATM1 participate in [Fe-S] proteins homeostasis but not in mitochondrial function

Given the mitochondrial localization of TgATM1, we assessed if its downregulation affects the organelle’s morphology. Previous studies have reported obvious morphological defects in mitochondria, including ‘ball-shaped’ and ‘broken’ mitochondria, upon downregulation of transmembrane proteins in the inner and outer mitochondrial membrane [62, 63]. However, downregulation of TgATM1 for 72 hours was not associated with any apparent morphological abnormalities, with parasites exhibiting the typical ‘lasso-shaped’ organelle, as observed in the controls (Fig 3A). We further scrutinized if the downregulation of TgATM1 impairs mitochondrial functions. To this end, we assessed the oxygen consumption rate (OCR) as an indicator of mitochondrial respiration using an extracellular flux analyser (Seahorse, Agilent) in both the parental DiCreU1 and TgATM1-Ty-U1 parasites under conditions of no treatment or after 72 hours of Rapa treatment (Fig 3B, S3 Table). The OCR decreased to about 70–80% upon TgATM1 downregulation compared to its controls (Fig 3B). The extracellular acidification rate (ECAR), which was simultaneously quantified, as a measure for glycolysis (Fig 3C, S3 Table), similarly decreased to about 70–80% upon downregulation of TgATM1, compared to its controls.

Download:

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

Fig 3. Downregulation of ATM1 affects iron-sulfur cluster proteins in Toxoplasma and iron levels in Plasmodium.

(A) Immunofluorescence assays of DiCreU1 (parental control) and TgATM1-Ty-U1 parasites following 0 and 72 hours of rapamycin (Rapa) treatment. Parasites were visualized using anti-actin (Act) antibodies, while mitochondria were stained using antibodies against the mitochondrial heat shock protein (mHSP70). (B) and (C) Basal oxygen consumption rate (OCR) (B) and extracellular acidification rate (ECAR) (C) following 0 or 72 hours of Rapa treatment in parental (DiCreU1) and TgATM1-Ty-U1 parasites, as determined using an extracellular flux analyser. (D) Relative abundance of the tricarboxylic acid (TCA) cycle intermediate succinate as determined by gas chromatography-mass spectrometry (GC-MS) of parental (DiCreU1) and TgATM1-Ty-U1 parasite extracts following 0 or 72 hours of Rapa treatment. B-D) The means and standard deviation of three independent biological experiments are shown. Bars represent the means of the three experiments. Conditions were compared using a One-way ANOVA followed by Tukey multiple pair wise comparisons. Significant differences (p <0.05) are highlighted in bold. (E)-(F) Volcano plot highlighting changes in protein levels of TgATM1-Ty-U1 parasites after 72 hours of rapamycin treatment compared to parental parasites (DiCreU1) treated equally with Rapa (E) or TgATM1-Ty-U1 parasites not treated with Rapa (F). Unchanged proteins are displayed in black, while significantly altered proteins are displayed in grey (< 1.5-fold change (FC), local false discovery rate, LFDR < 0.05) or in blue (>1.5-fold increased; LFDR < 0.05) or red (> 1.5-fold decreased; LFDR < 0.05). The number of significantly increased and decreased proteins is given and the NEET protein CDGSH1 is indicated. (G) Table listing proteins that are significantly up- or downregulated in both comparisons shown in E and F. [Fe-S] proteins are highlighted in red. Gene accession numbers (identifiers, IDs), the fitness score from a genome-wide CRISPR screen, the putative localization from a spatial proteomics study (localization of organelle proteins by isotope tagging, LOPIT), the number of transmembrane domains (TMDs), as well as the LFDR and FC for both comparisons shown in E and F are given.

https://doi.org/10.1371/journal.ppat.1012593.g003

To further probe mitochondrial function in parasites depleted in TgATM1, we measured the relative abundance of metabolites related to mitochondrial metabolism, including tricarboxylic acid (TCA) cycle intermediates, via gas chromatography-mass spectrometry (GC-MS). The abundance of succinate (Fig 3D), citrate, malate, γ-aminobutyric acid (GABA), aspartate, glutamate and 5-aminolevulinic acid (5-ALA)) did not change significantly (p <0.05, fold change >1.5) upon downregulation of TgATM1, when compared to both controls (DiCreU1 +Rapa and TgATM1-Ty-U1 −Rapa) (S4 Table). Taken together, these results speak against a specific mitochondrial defect in parasites depleted in TgATM1. Instead, the decrease in OCR alongside a similar decrease in ECAR indicates a general fitness impairment following TgATM1 depletion, which is consistent with the growth defect observed at this time point (Fig 2D). In contrast to the relatively subtle changes observed here, we have previously shown that a defect in heme synthesis causes a dramatic drop in mitochondrial respiration in T. gondii [64]. Thus, we conclude that TgATM1 is likely not implicated in heme synthesis, i.e., the trafficking of heme synthesis intermediates (protoporphyrinogen IX) into the mitochondrion.

In other organisms, ATM1 and its homologs are implicated in the export of sulfur-containing compounds from the mitochondrion to facilitate [Fe-S] synthesis in the cytosol [7, 65, 66]. Previous studies in T. gondii and other organisms have shown that disruption of [Fe-S] synthesis leads to destabilization and concomitant decreased abundance of [Fe-S] in the affected subcellular compartment [35, 67, 68]. Thus, to explore the function of TgATM1, we conducted quantitative proteomic analyses comparing parental DiCreU1 and TgATM1-Ty-U1 parasites under untreated conditions or after 72 hours of Rapa treatment (S5 Table). In TgATM1-Ty-U1 parasites subjected to 72 hours of Rapa treatment, 277 proteins showed significant downregulation, while 60 proteins exhibited significant upregulation compared to equally treated DiCreU1 parasites (p <0.05 and >1.5-fold change) (Fig 3E). Conversely, in TgATM1-Ty-U1 parasites, exposed to Rapa for 72 hours, 11 proteins displayed significant decrease, whereas 107 proteins showed significant increase, compared to untreated TgATM1-Ty-U1 parasites (Fig 3F). To exclude non-specific effects caused by the strain or Rapa, we only considered proteins that exhibited significant changes in both comparisons. Consequently, the list was narrowed down to 15 upregulated and five downregulated proteins (Fig 3G). Notably, TgATM1 itself was not among the significantly altered proteins as it was not detected in all replicates but exhibited a clear downregulation upon Rapa treatment (S5 Table). Assessing the subcellular localization [38] and performing GO-term enrichment analysis on the significantly altered proteins, revealed no specific impact of TgATM1 downregulation, with diverse subcellular compartments (Fig 3G, S5 Table) and biological processes affected (S5 Table). However, of the 20 significantly altered proteins, three were [Fe-S] proteins, including a NEET protein homolog, here termed CDGSH1, which has been implicated in transferring [Fe-S] to cytosolic proteins at the outer mitochondrial membrane in mammalian cells [69–73] (Fig 3E and 3F). Considering that T. gondii expresses approximately 64 [Fe-S] proteins [33], the presence of three in this shortlist of 20 proteins constitutes an overrepresentation. This led us to specifically assess the impact of TgATM1 downregulation on the abundance of [Fe-S] proteins, exploiting a previously published list [33]. In addition to the three significantly altered proteins (p <0.05 and >1.5-fold change), five others changed statistically significantly, but did not meet the 1.5-fold change threshold (S5 Table). Notably, the total of five [Fe-S] proteins downregulated in response to TgATM1 depletion were exclusively of nuclear localization based on the LOPIT dataset, and included ABCE1, the delta catalytic subunit of the DNA polymerase (POLD1) and two subunits of the RNA polymerase (POLR1), all of which were previously proposed to be downregulated in response to disruption of the CIA pathway through downregulation of TgNBP35 or TgATM1 [35, 36]. We speculate that a more striking fold-change and a global impact on cytosolic [Fe-S] was not observed here, due to the relatively short time of downregulation (72 hours). Comparable studies have assessed the abundance of cytosolic [Fe-S] proteins at later time points of depletion of the protein of interest (four days) and/ or used a faster system of downregulation (based on the western blots presented in those studies) [35, 36].

To assess the function of ATM1 in P. falciparum, we tested whether depletion of PfATM1 causes changes in mitochondrial iron and glutathione levels. Total organelle fractions from PfATM1-V_m-treated and control parasites were prepared by treatment with low concentration of digitonin. The organellar pellet contained mitochondrial, apicoplast and low levels of nuclear marker proteins, and was free of the cytosolic marker tubulin (S3H and S3I Fig). At 108 hours post-treatment, the organellar fractions from PfATM1-V_m-treated cells exhibited a notable rise in iron content compared to untreated controls (S3J Fig). While total glutathione levels showed a slight increase in treated cell fractions compared to controls, the difference was not statistically significant (S3K Fig). Accumulation of organellar iron was observed upon the knockdown of PfATM1; this accumulation could be due loss of [Fe-S] export via ATM1, although contributions from other phenomena are possible, such as increased uptake of iron into the mitochondrion stimulated by the decrease in extra-mitochondrial Fe-S proteins.

TgATM1 depletion is rescued by homologs from Plasmodium and Yeast, and TgATM1 overexpression reduces NEET protein levels

The above functional analyses indicate that TgATM1 and PfATM1 may be implicated in the export of [Fe-S] from the mitochondrion to the cytosol, as previously described for the yeast homolog ScATM1 [11]. To provide more compelling evidence that TgATM1 and PfATM1 are indeed functional homologs of the well-characterized ScATM1, we generated TgATM1-Ty-U1 complemented cell lines which constitutively express ScATM1 4-myc tagged (cScATM1-myc), in the uracil phosphoribosyltransferase (UPRT) locus [74] (S4A Fig). Functional complementation, similar to what was done with ScATM1, was also performed with PfATM1 4-myc tagged (cPfATM1-myc), along with a second copy of TgATM1 4-myc tagged (cTgATM1-myc) as a control. Transfected parasites were selected and cloned through limited dilution and correct integration was confirmed by PCR (S4A and S4B Fig). Products of the second copy ATM1 genes were efficiently detected and primarily localized to the mitochondrion of T. gondii, as confirmed by their co-localization with mHsp70 through IFA (Fig 4A). Effective downregulation of the endogenous TgATM1-Ty was verified following 72 hours of Rapa treatment in TgATM1-Ty-U1 parasites expressing the second copy ATM1s (S4C Fig). As for the IFA, western blot analysis confirmed that endogenous TgATM1 was efficiently downregulated upon addition of Rapa in the complemented strain and cTgATM1 was detected at the same size as the endogenous TgATM1-Ty. cScATM1 migrated close to 70 kDa, as reported previously (Fig 4B) [23], and also exhibited a band at approximately 30 kDa, likely corresponding to a degradation product. On the other hand, cPfATM1-myc ran above the 100 kDa marker. This is in line with the size of the major PfATM1 band detected in the P. falciparum lysate (Fig 1E). The intensity of bands of cScATM1-myc and cPfATM1-myc appeared to be weaker than cTgATM1-myc, indicating lower levels of expression (Fig 4B). To determine whether ScATM1 and PfATM1 function similarly to TgATM1 and whether their expression can rescue the severe phenotype linked to the depletion of endogenous TgATM1, we conducted plaque assays with the second copy complemented strains. All three strains formed plaques of lysis upon downregulation of endogenous TgATM1, confirming functional equivalence to ScATM1 and the genuine homology of PfATM1 to TgATM1 (Fig 4C and 4D). Notably, plaques formed by parasites complemented with the second copy of TgATM1-myc, following downregulation of endogenous TgATM1-Ty, were comparable to those formed by the parental line (Fig 4C and 4D). In contrast, parasites complemented with ScATM1-myc or PfATM1-myc formed slightly smaller plaques (Fig 4C and 4D). This difference could be due to lower expression levels (Fig 4B) or less efficient localization to the mitochondrion, or potentially a slight variation in substrate affinity between the ATMs of these species.

Download:

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

Fig 4. TgATM1-depleted Toxoplasma are functionally complemented through expression of the Plasmodium or yeast homolog.

(A) Immunofluorescence assays of TgATM1-Ty-U1 parasites expressing a second copy of myc-tagged Toxoplasma- (TgATM1-Ty-U1/cTgATM1-myc), Plasmodium- (TgATM1-Ty-U1/cPfATM1-myc) or yeast ATM1 (TgATM1-Ty-U1/cScATM1-myc). IFAs were stained with anti-myc antibody, as well as antibodies against a mitochondrial marker, mitochondrial heat shock protein (mHSP70). (B) Western blots of protein lysates from parasites as listed in (A) following 0 or 72 hours of downregulation of endogenous TgATM1 through addition of rapamycin (Rapa). Membranes were probed with anti-Ty and anti-myc antibodies as well as anti-catalase antibodies as loading control. The ATM1 proteins detected at varying molecular weights are indicated. The panel shows blots that were performed on the same membrane and exposed for identical durations. (C) Stained host cell monolayers revealing plaques formed by parental (DiCreU1), uncomplemented TgATM1-Ty-U1 parasites or TgATM1-Ty-U1 parasites expressing a second copy of TgATM1 as listed in (A) over seven days in presence or absence of Rapa. A-C: Representative images are shown from one of three independent biological replicates. (D) Plaque sizes quantified from assays as shown in (C) from three independent experiments. Sizes of individual plaques are shown relative to the mean of the parental line DiCreU1 not treated with Rapa. Means are indicated through horizontal bars. Conditions were compared using a One-way ANOVA followed by Tukey multiple pair wise comparisons. Significant differences (p <0.05) are highlighted in bold. (E and F) Basal oxygen consumption rate (OCR) (E) and basal extracellular acidification rate (ECAR) (F), as determined using an extracellular flux analyser following 0 or 72 hours of Rapa treatment of parasite lines listed in (C). (G) Relative abundance of the TCA cycle intermediate succinate, as determined by gas chromatography-mass spectrometry (GC-MS) of parental (DiCreU1) and TgATM1-Ty-U1 parasite extracts following 0 or 72 hours of Rapa treatment of parasite lines listed in (C). (E-G) The means and standard deviation of three independent biological experiments are shown. Bars represent the means of the three experiments. Conditions were compared using a One-way ANOVA followed by Tukey multiple pair wise comparisons. Significant differences (p <0.05) are highlighted in bold. Note that values for DiCreU1 and TgATM1-Ty-U1 parasites are as in Fig 3 C-H and were re-plotted. (H) Volcano plot highlighting changes in protein levels of TgATM1 overexpressing parasites (TgATM1-Ty-U1/cTgATM1-myc) compared to TgATM1-Ty-U1 parasites, both not treated with Rapa. Unchanged proteins are displayed in black, while significantly altered proteins are displayed in grey (< 1.5-fold change (FC), local false discovery rate, LFDR < 0.05) or in blue (>1.5-fold increased; LFDR < 0.05) or red (> 1.5-fold decreased; LFDR < 0.05). The number of significantly increased and decreased proteins is given and the NEET proteins CDGSH1 and CDGSH2 as well as TgATM1 are indicated.

https://doi.org/10.1371/journal.ppat.1012593.g004

To evaluate the impact of expressing second copy ATMs on mitochondrial functions, we conducted extracellular flux analyses and GC-MS analyses as described earlier. Notably, the downregulation of endogenous TgATM1 resulted in a slight decrease in mitochondrial respiration and glycolytic activity, regardless of the presence of a second copy ATM (Figs 4E, 4F, S4D and S4E, S3 Table), despite the observed rescue in fitness (Fig 4C and 4D). These differences argue against a mitochondrial defect but rather suggest a general slight decrease in fitness upon downregulation of endogenous TgATM1. Concordant with the absence of mitochondrial defect upon expression of the second copy ATM1 proteins, we observed relatively modest or no differences in the abundance of TCA cycle intermediates and other mitochondrion-related metabolites, such as succinate (Fig 4G, S4 Table) in parasites expressing a second copy ATM1 and depleted in the endogenous protein or not. In summary, PfATM1 and ScATM1 serve as functional counterparts of TgATM1, capable of rescuing its depletion. The fact that expression of the second copy itself (ScATM1) or downregulation of the endogenous gene in second copy expressing strains (PfATM1) is associated with a decrease in plaques sizes emphasizes the necessity for precise regulation of ATM1 expression or subtle disparities in the transporters’ specific roles.

To expand on the relatively limited insights obtained through proteomic analyses of parasites depleted in TgATM1, we additionally performed quantitative proteomic analyses of parasites overexpressing TgATM1 (TgATM1-U1-Ty/cTgATM1-myc). Quantifying the abundance of proteins in TgATM1-U1-Ty/cTgATM1-myc parasites compared to TgATM1-U1-Ty parasites, both not treated with Rapa, revealed 59 proteins to be significantly decreased and 40 significantly increased (p <0.05 and >1.5-fold change) in parasites overexpressing TgATM1 (Fig 4H, S6 Table). The most significantly upregulated protein was TgATM1 itself, exhibiting a 21-fold increase in its abundance. As observed during TgATM1 downregulation, TgTAM1 overexpression impacted on proteins in various subcellular compartments and of diverse functions (S6 Table). Strikingly, amongst the significantly downregulated proteins were two [Fe-S] proteins, here termed CDGSH1 and CDGSH2, which belong to the NEET family (Fig 4H, S6 Table). These exhibited a 2.4-fold and 20.2-fold decrease in abundance, respectively, upon TgATM1 overexpression. In mammalian cells, CDGSH1 and 2 have been shown to localise to the outer mitochondrial membrane and endoplasmic reticulum-mitochondria interface, respectively, where they donate their uniquely coordinated and relatively labile [Fe-S] to other proteins [69–73]. CDGSH1 was also amongst only 20 significantly altered proteins upon TgATM1 downregulation (Fig 3E-3G, S5 Table), where it was upregulated (1.85-fold, average), suggesting that its abundance is inversely correlated to that of TgATM1.

PfATM1 and TgATM1 form homodimers

It is well-established that ScATM1, ABCB7, and AtATM3 form homodimers [75–77]. To verify that TgATM1 forms homodimers, we performed a co-immunoprecipitation assay (Co-IP) with TgATM1-Ty-U1-cTgATM1-myc using an α-myc antibody (Fig 5A). The potential for cross-species heterodimer formation between TgATM1-Ty and the expressed ScATM1-myc and PfATM1-myc in TgATM1-Ty-U1 strains could have a notable impact on the functional characterization described above (Fig 4). This hypothesis was tested by performing Co-IPs with TgATM1-Ty-U1/cPfATM1-myc (Fig 5B) and TgATM1-Ty-U1/cScATM1-myc (Fig 5C). The absence of TgATM-Ty in the pull-down assays suggests that no cross-species heterodimers are formed between the different ATMs even if myc-tagged ATM1 is readily detectable in the lysate (S5A-S5C Fig).

Download:

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

Fig 5. TgATM1 an PfATM1 form homodimers but no cross-species dimers are observed.

(A-C) Western blots with material from an anti-myc co-immunoprecipitation assay (Co-IP) and the corresponding flow through (FT) from parasite lysates of TgATM1-Ty-U1 parasites expressing a second copy of myc-tagged Toxoplasma- (TgATM1-Ty-U1/cTgATM1-myc), Plasmodium- (TgATM1-Ty-U1/cPfATM1-myc) or yeast ATM1 (TgATM1-Ty-U1/cScATM1-myc). Membranes were probed with anti-Ty antibody. TgATM1-Ty is indicated. Blots shown are representative of three independent experiments. (D) Schematic representation of PfATM1 domains. Full-length PfATM1 was expressed as a fusion protein in β-pET28a(+)-β. (E) Purified PfATM1-CTD (~30 kDa) with a dimeric form (~60 kDa) seen in Coomassie-stained SDS-PAGE, which are also detected in western blot with anti-6XHis Ab. (F) Purified dimeric and monomeric forms of PfATM1 after size exclusion chromatography separated on SDS-PAGE. PfATM1 dimer, of expected size 304 kDa, migrated between 158 and 440 kDa on native PAGE. (G) Negative staining TEM detected PfATM1 dimers. Arrows indicate electron densities revealing two-fold symmetry. (H) UV-VIS spectra of PfATM1, PfATM1-CTD and PfATM1-NTE. PfATM1 purified as a reddish-brown protein (inset). (I) and (J) PfATM1 (I) and TgATM1 (J) modelled on NaATM1 (PDB:4MRS). The 128 aa insertion in between the transmembrane region and CTD of PfATM1 and the insertion between TM1 and TM2 of TgATM1 could not be modelled and are shown in red. A portion (aa 231–305) of the unique PfATM1 N-terminal extension is in green. An ATP molecule is docked in the ATPase domain (CTD) of each chain and two GSSG molecules are docked at the PfATM1 dimer interface in the transmembrane region. (K) 3-D interaction plot of two GSSG molecules positioned between the two PfATM1 chains. Arg534 and Arg647 (circled in red) in each PfATM1 chain have electrostatic interactions with the two GSSG molecules. (L) Change in intrinsic tryptophan fluorescence of PfATM1 in the presence of increasing concentrations of GSSG. a.u., fluorescence intensity in arbitrary units. In the corresponding graph, change in fluorescence intensity (ΔF/F₀) at 335 nm is plotted against GSSG concentrations with curve-fitting using GraphPad Prism 5. (M) [4Fe-4S]GS₄ manually docked into the central cavity of PfATM1 modelled on CtATM1 (PDB:7PRO).

https://doi.org/10.1371/journal.ppat.1012593.g005

Dimerization of PfATM1 was tested using recombinant protein. PfATM1-CTD (30 kDa) was expressed as a 6X-His tagged protein in E. coli and purified through affinity chromatography (Fig 5D and 5E). PfATM1-CTD formed homodimers which did not completely break in SDS-PAGE but could be broken with urea (S5D Fig). Since stable full-length PfATM1 could not be expressed as a simple 6X-His fusion in bacteria, expression of soluble protein was accomplished by using the E.coli YbeL hydrophilic fusion domain (β) and 6XHis-tags fused to the N- and C-termini [78, 79] (Fig 5D). PfATM1 monomers (152 kDa) and dimers (~300 kDa) were separated by size exclusion chromatography (SEC) (Figs 5F and S5E) and a band of ~300 kDa was detected in the dimer peak in native-PAGE (Fig 5F), indicating that PfATM1 forms dimers in solution. The PfATM1 monomer fractionated with additional degradation product of ~100 kDa (Fig 5F). CD spectra of purified dimeric PfATM1 was as expected for a protein with predominantly alpha-helical secondary structural elements (S5F Fig). Negative stain transmission electron microscopy (TEM) confirmed its homodimeric state in aqueous buffer (Fig 5G). The purified complexes were observed in different orientations in the TEM micrographs and the dimers were clearly visible from the side view projections of the complex (Fig 5G). Here, the electron densities appear as two stripes or transverse striations revealing a clear two-fold symmetry which indicates that the complex is a dimer. PfATM1 purified as a reddish-brown protein suggesting the presence of bound iron (Fig 5H, inset). UV-VIS scan showed an intense peak at 420 nm indicating the presence of bound [4Fe-4S] (Fig 5H). Additional peak at 456 nm and shoulder at 322 nm expected for [2Fe-2S] were not seen. A lower intensity [4Fe-4S] peak was also observed in PfATM1-CTD, and none in PfATM1-NTE (Fig 5H).

Structure homology models of PfATM1 (231–1049 aa) and TgATM1 (410 to 1133 aa) were generated on the crystal structure of bacterial NaATM1 (PDB id: 4MRS) [56]. Ramachandran plot analysis of the PfATM1 model indicated that 87% residues were in the favoured region, 10% residues in the additional allowed region, 1.9% residues in the generously allowed region and 1.1% residues in the disallowed region with RMSD score of 0.961 Å. The TgATM1 model had 89.7% residues in the favoured region, 7.8% residues in the additional allowed region, 1.2% residues in the generously allowed region and 1.3% residues in the disallowed region, with RMSD score of 0.755 Å. Only a part of the unique NTE of PfATM1 could be modelled and insertions in TgATM1 and PfATM1 appeared as unstructured loops. The full-length TgATM1 and PfATM1 structures available in the AlphaFold2 Protein Structure Database [80] also represent these stretches as extruded loops with some α-helices. Structural folds in all major domains were largely conserved (Figs 5I, 5J, S6A and S6B). Since PfATM1 has observed and predicted glutathione binding sites, the interaction of GSSG to the central cavity was tested in silico by simultaneously docking two molecules of GSSG in the PfATM1 model (binding affinity or free energy of binding = -20.073 kcal/mol) (Figs 5K, S6C and S6D). The predicted binding affinity of GSSG to PfATM1 compares well with re-dock of two GSSG molecules on the NaATM1 co-crystal structure (binding affinity of -17.9 kcal/mol) and simultaneous docking of two GSSG molecules on the analogous site in CtATM1 (binding affinity of -20.59 kcal/mol). Out of the conserved arginine residues for glutathione binding in bacterial, fungal and yeast homologs, two PfATM1 arginine residues (Arg534, Arg647 downstream of TM4 and TM5, respectively) in each chain have electrostatic interactions with GSSG (Figs 5K and S6D). Two ATP molecules (one in each CTD) were also simultaneously docked in the PfATM1 model with a docking score of -25.465 kcal/mol in the best docking pose (Fig 5I). Molecular dynamics simulation confirmed the stability of PfATM1 with two GSSG molecules and two ATP molecules (S6C Fig). The binding of GSSG to PfATM1 was confirmed by spectrofluorometric analysis which showed a marked GSSG concentration-dependent quenching of intrinsic tryptophan fluorescence indicative of conformational changes upon interaction (Figs 5L and S6E). NaATM1 has been shown to undergo conformational changes in transmembrane helix 6 (TM6) which modify the glutathione-binding site [56]. The altered TM6 conformation pre- and post-ATP hydrolysis is directly coupled to preclusion of glutathione binding and sets the directionality of transport. Although transmembrane domain prediction did not identify TM6 in PfATM1, the corresponding region in the molecular structure model superimposed on NaATM1 TM6 (S6A Fig) suggesting a conserved role in GSSG binding and transport.

The possibility of a negatively charged [4Fe-4S]GS₄ complex interacting with the positively charged central cavity of PfATM1 (S6F Fig) was addressed by modelling the protein on the recently available cryo-EM structure of the inward open configuration of fungal Chaetomium thermophilum ATM1 (CtATM1) (PDB: 7PRO) which binds [2Fe-2S]GS₄ [17]_. The size of the central cavity of PfATM1 was predicted to be >9000 Å, as compared to >7000 Å in CtATM1. [4Fe-4S]GS₄ was manually docked into the PfATM1 cavity (Fig 5M). The gatekeeper residues Phe396 and Ser399 in CtATM1 (Met317 and 320 in NaATM1) align with Met641 and Thr644 in PfATM1 (S1B Fig) and occupy a position above the central cavity (Fig 5M). MD simulation indicates RMS (root mean square) fluctuation of >1 Å in the putative gatekeeper residues in PfATM1 suggesting conserved function (S6G Fig). Structure modelling and docking results indicate that PfATM1 likely can accommodate GSSG or glutathione-complexed [4Fe-4S].

PfATM1 interacts with components of the mitochondrial ISC and cytosolic CIA pathway

NBP35 is a scaffold protein, important for [Fe-S] assembly in the CIA pathway [81]. In T. gondii, TgNBP35 (TGGT1_280730) was localized to the outer mitochondrial membrane [35]. Thus, we speculated that a direct interaction between TgATM1 and TgNBP35 might occur. To test this, an HA-tag was fused to the C-terminus of TgNBP35 via CRIPR Cas9 editing at the endogenous locus of TgNBP35 in TgATM1-Ty-U1 and TgATM1-Ty-U1/cTgATM1-myc parasites, thus generating TgATM1-Ty-U1/TgNBP35-HA and TgATM1-Ty-U1/cTgATM1-myc/TgNBP35-HA lines. These lines were used to establish by IFA that mitochondrial localization of NBP35-HA was not altered through downregulation of TgATM1 (Fig 6A). Also, co-IP with α-HA antibody did not pull down second copy expressed TgATM1-myc in TgATM1-Ty-U1/cTgATM1-myc/TgNBP35-HA parasites (Fig 6B). Hence, we conclude that no stable interaction occurs between TgNBP35 and TgATM1.

Download:

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

Fig 6. PfATM1 but not TgATM1 interacts directly with components of the cytosolic CIA pathway.

(A) Immunofluorescence assay of TgATM1-Ty-U1/TgNBP35-HA parasites following 0 and 72 hours of rapamycin (Rapa) treatment. Parasites were stained with anti-Ty and anti-HA antibodies as well the mitochondrial marker mitochondrial heat shock protein (mHSP70). (B) Western blots with material from an anti-HA co-immunoprecipitation assay (Co-IP) and the corresponding supernatant (SN), pellet (P) and flow through (FT) from lysates of TgATM1-Ty-U1/cTgATM1-myc/TgNBP35-HA parasites. Membranes were probed with antibodies against HA and myc. TgATM1-myc and TgNBP35-HA are indicated. Images shown in A and B are representative of three independent biological replicates, (C) Pull-down from P. falciparum lysate using PfATM1-CTD as bait. Anti-6XHis antibodies were used to detect PfATM1-CTD; anti-PfIscA2, anti-PfIscU and anti-PfHU sera detected corresponding proteins in western blots. Eluates from negative control EcEngA-tagged beads and beads alone were also loaded. (D) and (E) In vitro transfer of [4Fe-4S] to recipient apo-PfATM1 from PfIscU (D) and PfIscA2 (E) in the presence or absence of DTT.

https://doi.org/10.1371/journal.ppat.1012593.g006

Similarly, to investigate the interaction between PfATM1 and components of the cytosolic CIA machinery, pull-down assays were carried out. Since the mitochondrial matrix side of PfATM1 comprises the CTD, PfATM1-CTD bound to Ni-NTA beads was used as bait to pull down interacting proteins from the parasite lysate. Western blots showed that PfATM1-CTD could pull down both PfIscU and PfIscA2 (Fig 6C). No signals for these proteins were seen in eluates of control sets using blank beads or unrelated protein EcEngA as bait. Recombinant PfIscU and PfIscA2, expressed as GST fusion proteins and thrombin-cleaved to remove the GST tag (S7A Fig) were chemically reconstituted. Reconstituted PfIscU and PfIscA2 carrying [4Fe-4S] were then used as cluster donors for PfATM1 in transfer reactions in vitro, in the presence or absence of DTT (Fig 6D and 6E). Apo-PfATM1 could receive [4Fe-4S] from both donor proteins and was converted to the holo-form even in the absence of DTT, thus confirming its role as a recipient of [4Fe-4S] from ISC pathway cluster carriers.

To probe whether PfATM1 can interact with components of the CIA pathway to transport [4Fe-4S], P. falciparum NBP35 and CFD1 homologs were recombinantly expressed in E. coli (S7B-S7E Fig). NBP35 and CFD1 are ATPases that can exist as homodimers, heterodimers and also form a heterotetrameric [Fe-S] scaffold complex as part of the CIA machinery in yeast [9, 82]. NBP35 is annotated in PlasmoDB, and a P. falciparum CFD1 homolog is currently annotated as HCF101. Both proteins show conservation of major ATP binding and hydrolysis motifs as well as a CXXC motif each for binding [Fe-S] (S8 Fig). Antibodies generated against PfNBP35 in rabbit recognized a specific band of the expected size of ~51 kDa in the parasite lysate (S7I Fig). The interaction of full-length PfATM1 with PfNBP35 and/or PfCFD1 in vitro was checked by incubating the recombinant proteins and resolving the complexes by native PAGE. PfNBP35 and PfCFD1 mixed in equimolar ratios formed a complex migrating above the 240 kDa marker protein, close to the expected heterotetrameric complex size of ~200 kDa (S7F Fig). The PfATM1 dimer formed a complex with PfNBP35 as evident from an upward shift in migration in native PAGE, but did not interact with PfCFD1 alone (S7G Fig). When the pre-formed PfNBP35-PfCFD1 heterotetramer was incubated with PfATM1, a complex migrating between 669 and 440 kDa was obtained (S7G Fig). This likely represents a complex formed by interaction of PfATM1 dimer and heterotetrameric PfNBP35-PfCFD1 with an expected total size of ~500 kDa. To check whether PfATM1-NTE which might be exposed in the mitochondrial intermembrane space can interact with PfNBP35, in vitro complex formation was checked with purified PfATM1-NTE or -CTD (S7H Fig). An interaction was detected with PfNBP35 and PfATM1-NTE (S7H Fig, lane 4), but not with PfATM1-CTD (S7H Fig, lane 5).

PfATM1-NTE was next used as bait in pull-down experiments. Western blot confirmed that PfATM1-NTE specifically pulled down PfNBP35 from the parasite lysate (S7J-S7K Fig). PfNBP35 was not pulled down with beads alone or with PfEngA as bait. The results suggest that PfATM1 can interact with the cytosolic CIA component PfNBP35 which is part of the PfNBP35-PfCFD1 scaffold complex and that its unique NTE can mediate this interaction in vitro. The orientation of the PfATM1-NTE domain in relation to the inner mitochondrial membrane is unclear. The NTE contains a hydrophobic stretch from aa 154 to 163 which is part of a predicted α-helix (aa 135 to 163) (S1B Fig); if this region traverses the membrane, a part of the NTE could be exposed to the intermembrane space. However, in the absence of direct interaction data with parasite proteins in situ, the physiological relevance of the PfATM1-PfNBP35 interaction remains to be established.

ATP hydrolysis is stimulated by GSSG and [4Fe-4S] and enhances GSSG transport by PfATM1

The ATPase activity of PfATM1 and its modulation by different ligands was assayed. We first determined that the ATPase activity of the protein was time- and concentration-dependent (S9A and S9B Fig), followed by evaluation of the effect of oxidized and reduced glutathione (GSSG and GSH) on ATP hydrolysis. Significant enhancement of ATPase activity was observed only in the presence of GSSG (P = 0.001) (Fig 7A). Kinetics of ATP hydrolysis showed that GSSG-bound PfATM1 had lower K_M (greater affinity for ATP) and higher V_max (Fig 7B). The catalytic efficiency of PfATM1 (K_cat/K_M) increased ~11-fold in the presence of GSSG (0.52 μM^-1 min^-1 versus 0.043 μM^-1 min^-1 in the presence and absence of GSSG, respectively) indicating conformational changes transmitting to the ATPase domain upon GSSG binding to its cognate site in the transmembrane region.

Download:

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

Fig 7. ATPase activity of PfATM1 and GSSG transport by PfATM1-proteoliposomes.

(A) ATP hydrolysis by PfATM1 in the presence of GSH or GSSG as measured by Pi release. (B) Kinetics of ATP hydrolysis by PfATM1 in the presence or absence of GSSG. (C) Comparison of basal ATPase activity of the wild-type and PfATM1-E972Q mutant. ATP hydrolysis by PfATM1–R534E mutant was compared with wild-type protein in the presence of GSSG. (D) Comparison of ATPase activity of holo- and apo-PfATM1 (inset) in the presence or absence of GSSG. (E) Negative staining TEM of control liposomes and PfATM1-proteoliposomes. Incorporated protein is indicated by arrows. A section of the proteoliposome with PfATM1 dimer is enlarged. (F) Time-dependent GSSG uptake in the presence or absence of ATP by proteoliposomes carrying PfATM1-wild type or PfATM1-E972Q. LS, control liposome without protein. (G-H) Transport activity (G) and transport rate (H) were calculated. P values (t-test) from two biological replicates are indicated.

https://doi.org/10.1371/journal.ppat.1012593.g007

The specificity of ATPase activity by purified PfATM1 was confirmed by generating the Walker B ATP hydrolysis mutant E972Q (Fig 7C). Mutation of the GSSG-binding site Arg534 identified in GSSG-PfATM1 docking (Fig 5) also significantly reduced ATP hydrolysis in the presence of GSSG (Fig 7C). Since PfATM1 strongly bound [4Fe-4S], the effect of the cluster on the ATPase activity was assayed by comparing apo- and holo-forms of the protein (Fig 7D). Removal of the cluster caused a significant reduction in ATP hydrolysis which could not be restored by addition of GSSG, suggesting that enhancement of ATPase activity by GSSG is contingent upon the presence of [4Fe-4S]. This was corroborated by spectrofluorometric analysis of conformational changes upon GSSG interaction with apo-PfATM1 (S9C Fig). Unlike holo-PfATM1 (Fig 5L), there was no drop in fluorescence at low GSSG concentrations; the drop seen at 3 mM GSSG did not increase further at higher concentrations. The results suggest that GSSG-induced conformational changes in PfATM1 are more prominent in the presence of [4Fe-4S]. Prior presence of the cluster might hold PfATM1 in a conformation that allows GSSG easier access to the central cavity.

To analyse whether PfATM1 can transport GSSG as reported for bacterial, yeast and plant ATM homologs [12, 56], we incorporated PfATM1 and PfATM1-E972Q in unilamellar proteoliposomes. The average diameter of PfATM1-liposomes deduced by DLS was ~312 nm (S9D Fig). Protein incorporation was confirmed by negative stain TEM (Fig 7E) and SDS-PAGE (S9E Fig). GSSG uptake by PfATM1-proteoliposomes was clearly enhanced by ATP (Fig 7F). The effect of ATP hydrolysis on GSSG transport was confirmed by the fact that PfATM1-E972Q proteoliposomes functioned identically in the presence or absence of ATP. Low levels of background GSSG uptake seen with blank liposomes is possibly attributable to limited diffusion across the membrane (Fig 7F). Transport activity was linear with time (Fig 7G). The maximum transport rate of GSSG in the presence of ATP was 4.6 ± 0.09 nmol GSSG mg^-1 min^-1. The transport rate was halved to 1.83 ± 0.14 nmol GSSG mg^-1 min^-1 when ATP was not added to the reaction (Fig 7H). PfATM1-E972Q showed no significant change in transport rate of GSSG with or without ATP.

Ethics statement

All animal experiments were conducted in compliance with guidelines of the Committee for Control and Supervision of Experiments on Animals (CCSEA), India. Approval was obtained from the Institutional Animal Ethics Committee of the Central Drug Research Institute (CDRI) (#IAEC/2007/126/Renew-10(234/17). Approval for use of RBCs from healthy human volunteers was obtained from the CDRI Institutional Ethics Committee (Human Research) (#CDRI/IEC/2017/A4); formal verbal consent was obtained from the volunteers.

In-silico analyses of ABCB superfamily members and phylogeny

The Plasmodium and Toxoplasma databases (http://plasmodb.org; http://toxodb.org) were mined for sequences of putative members of ABCB superfamily in P. falciparum and T. gondii. Targeting predictions were made using PlasmoAP (https://plasmodb.org/plasmo/app/plasmoap; a predictor for the bipartite apicoplast-targeting sequence comprising signal and transit peptide), Signal P (http://cbs.dtu.dk/services/SignalP/), Mitoprot II (https://ihg.gsf.de/ihg/mitoprot.html) and TargetP-2.0 (http://www.cbs.dtu.dk/services/TargetP; used to predict presence of N-terminal mitochondrial transit peptide- mTP). TMHMM- 2.0 [92] was used to predict membrane spanning regions in the protein.

A maximum-likelihood phylogenetic genetic tree was constructed with 1000 bootstrap replicates using LG+R+F model by ATGC: SMS (Smart Model Selection) in PhyML 3. 0 [93]. A total of 30 protein sequences (S1 Table), including apicomplexan ATMs, were used for constructing the phylogenetic tree. Amino acid sequences were taken from NCBI database (www.ncbi.nlm.nih.gov) after BLASTp searches [94] for homologs of PfATM1 (PfABCB6) from bacteria to higher eukaryotes.

Sequence analysis, molecular structure modelling and docking

Sequence alignment of PfATM1 with its homologs was done using Clustal Omega [95]. DISOPRED 3 (http://bioinf.cs.ucl.ac.uk/psipred/) was used to analyse the overall disordered regions in PfATM1. Molecular structure homology models of PfATM1 and TgATM1 were generated using Modeller 10.1 [96] based on the crystal structures of ATM1 of the bacterium N. aromaticivorans (PDB: 4MRS) [56] or the filamentous fungus C. thermophilum (PDB: 7PRO) [17]. 50 models were generated. The final model was selected according to DOPE (discrete optimized protein energy) score and the modeler objective function molpdf. PDBsum server [97] was used for external evaluation of the selected best model. Root mean square deviation (RMSD) score was estimated by alignment with template structure on PyMOL (The PyMOL Molecular Graphics System, Version 2.2, Schrödinger, LLC, 2018). Secondary structural analysis was done by Ramachandran plot statistics of the best selected model through RAMPAGE [98]. Vina version 1.2.0 [99] was used for simultaneous docking of two GSSG ligands around the analogous space in the PfATM1 model using Autodock4 forcefield. Two ATPs were simultaneously docked within a grid containing all putative ATP binding residues on each chain using Vina version 1.2.0 and Autodock 4 forcefield [100]. The docked complex consisting of protein, two GSSG molecules and two ATP molecules were used to perform molecular dynamics simulation using GROMACS version 2020.1 [101]. The simulation box was energy minimized, neutralized and equilibrated and subjected to molecular dynamics simulation using Charmm36 forcefield.

For interaction with glutathione-coordinated [4Fe-4S], PfATM1 was modelled on CtATM1 inward-open structure (PDB: 7PRO) [17]. A total of 50 models were generated and the best model was chosen on the basis of modeller objective function (molpdf). Cavity size was measured by CastP3.0 [102] with 2.5 Å probe. [4Fe-4S] was extracted from the pdb structure of [4Fe-4S]-dependent thiouracil desulfidase (PDB: 6Z92) [103]; 3D structure of glutathione was retrieved from PubChem (CID: 124886). The structure of glutathione was energy minimized using gaff forcefield [104] in Open Babel v3.0.0 by employing steepest descent algorithm with 1500 steps. Structures of a [4Fe-4S] and four molecules of glutathione were then joined as [4-Fe-4S]GS₄. The final ligand was manually docked in the analogous pocket defined in the CtATM1 structure [17] using Chimera v1.6.

Parasite strains and culture

P. falciparum 3D7 was cultured in human erythrocytes maintained in complete RPMI-1640 medium (Sigma-Aldrich, USA) supplemented with 0.5% (w/v) Albumax II (Invitrogen, Thermo Fisher Scientific, USA) and 0.2% sodium bicarbonate (Sigma-Aldrich, USA) at 2% hematocrit. Parasite genomic DNA was isolated by phenol-chloroform extraction after 0.05% saponin lysis of infected RBCs. Parasites were synchronized with 5% sorbitol at early- to mid-ring stage.

The T. gondii DiCre strain was a generous gift from Moritz Treeck. Parasites were maintained under standard tachyzoite growth conditions through regular passage in confluent human foreskin fibroblasts (HFFs) in Dulbecco Modified Eagle Medium (DMEM, Gibco, Thermo Fisher Scientific, USA) supplemented with 5% fetal bovine serum (FBS, Gibco, Thermo Fisher Scientific), 2 mM L-glutamine (Gibco, Thermo Fisher Scientific, USA) and 25 μg/ml gentamycin (Gibco, Thermo Fisher Scientific, USA). Cultures were maintained in humidified incubators at 37°C and 5% CO₂.

Vivo-morpholino (VMO) mediated PfATM1 knockdown

A specific Vivo-morpholino (PfATM1-V_m; 5’-ACTCTCTATAATATTTCATCATGACA-3’) complementary to 5’UTR region of PfATM1 mRNA was designed and synthesized by Gene Tools (LLC, USA) and used as a translation blocking agent. 5’-AAGCCTATTGCATAGAAAACGTACG-3’ was the non-specific control morpholino. PfATM1-V_m (3 μM) was added to late ring-stage synchronized P. falciparum 3D7 culture at 10% parasitemia and incubated at 37°C for 6 hours [61]. The morpholino was removed by media replacement. Parasite asexual blood stage development was monitored till 144 hours of parasite growth. After 72, 96 and 144 hours, parasites were harvested and PfATM1 expression was analysed by western blotting using anti-PfATM1 antibody. PfATM1 and actin levels were quantified by densitometry using ImageQuant TL software (Cytiva, USA). Parasite asexual growth stages were monitored by Giemsa staining.

Organellar fractions were prepared by treating parasites with low digitonin concentration (0.05%) as described previously [105, 106]. Total organellar fractions of morpholino-treated and untreated cells (from culture dishes containing synchronized parasites with equal parasitemia) were electrophoresed on 10–14% SDS-PAGE and western blotting was carried out with anti-α tubulin Ab and anti-histone 3 Ab (Sigma-Aldrich, USA). The blot was stripped and re-probed with anti-PfHU Ab and anti-PfIscA2 [31]. Iron in organellar fractions from control and PfATM1-V_m treated parasites was quantified using ferrozine as described earlier [31, 107]. The same total amount of protein (~1mg) from both treatments was assayed for quantification of iron and glutathione. Total glutathione was assayed using the Glutathione Assay kit (Cayman Chemical Company, USA).

Generation of transgenic T. gondii

Tachyzoites were transfected by electroporation [108]. The TgATM1-Ty-U1 strain was generated by co-transfection of 20 μg of a CRISPR/Cas9 expression plasmid [109] targeting the 3′-UTR of the gene of interest (TGGT1_269000), and a PCR-generated homology repair template encoding the 3-Ty-U1 tagging construct as well as a hypoxanthine-xanthine-guanine phosphoribosyl transferase (HXGPRT) resistance cassette [58, 110]. The corresponding primers are listed in S2 Table. Transfected parasites were selected for 1 week with 25 μg/ml mycophenolic acid and 50 μg/ml xanthine for HXGPRT positive selection, and stable transfectants were cloned by serial dilution [111].

Codon remodelled copies of TgATM1-myc, ScATM1-myc, and PfATM1-myc were ordered from GenScript and inserted into a pUC57 vector. The copies were excised from the pUC57 vector using EcoRI and EcoRV sites and inserted in a linearized (EcoRI and EcoRV) pUPRT-pTUB-PEPCK-4Myc vector. Integration was checked by PCR. For transfection, 60 μg of pUPRT-pTUB-XXATM1-4myc vector were linearized by NdeI and mixed with 20 μg of gRNA targeting the UPRT locus. Transfected parasites were selected for one week with 5-fluorodeoxyuridine (FUDR) for positive selection, and stable transfectants were cloned by serial dilution.

NBP35 (TGGT1_280730) was tagged at its 3′ by end through insertion of a 3-HA tag and a dihydrofolate reductase (DHFR) resistance cassette [112] using the CRISPR/Cas9 strategy as above. Transfected parasites were selected for 1 week with 1 μM pyrimethamine, and stable transfectants were cloned by serial dilution.

Genomic DNA extraction and integration PCR

Genomic DNA was extracted from T. gondii parasites using the Wizard Genomic DNA Purification Kit (Promega, USA). Integration of all constructs was confirmed by PCR (primers in S2 Table) using GoTaq DNA Polymerase (Promega, USA) with the integration primers listed in S2 Table.

Plaque assays and intracellular growth assays

Serial dilutions of freshly egressed T. gondii culture were incubated on confluent monolayers of HFF cells and left to grow for 7 days in the absence or presence of Rapa (added to a final concentration of 50 nM). After 7 days, host cell monolayers were fixed with 4% paraformaldehyde (PFA) for 10 min, washed with phosphate-buffered saline (PBS), and plaques were revealed by staining with a crystal violet solution (12.5 g crystal violet, 125 ml ethanol mixed with 500 ml water containing 1% (w/v) ammonium oxalate). Plaque sizes were quantified using the polygonal selection icon after setting the scale and entering the distance in ImageJ.

For growth assays, confluent HFFs grown on coverslips were inoculated with 10 μl of a freshly lysing T. gondii culture. Parasites were pre-treated with Rapa and treated with Rapa during the growth assay as indicated. Twenty-four hours after inoculation, parasites were fixed with 4% PFA-glutaraldehyde for 10 min, prior to quenching with 0.1 M glycine/PBS. Infected host cells were permeabilized (0.2% Triton X-100/PBS), blocked (2% BSA/0.2% Triton X-100/PBS), and probed with a polyclonal rabbit antibody against the pellicle marker GAP45 [113] diluted in 2% BSA/0.2% Triton X-100/PBS for 1 hour (polyclonal rabbit anti-IMC (1:2000). Samples were washed (3 × 5 min, 0.2% Triton X-100/PBS) and probed with a secondary antibody prior to washing as for the primary antibody (anti-rabbit Alexa fluor 594, Invitrogen, USA). The coverslip was mounted on microscopy slides using DAPI Fluoromount-G (Invitrogen, USA). Slides were viewed on a Nikon Eclipse Ti inverted microscope (Nikon, Japan). The number of parasites was counted in 100 vacuoles per condition in triplicates and replicated in three independent biological replicates.

Recombinant protein purification

Gene segments encoding the predicted mature form of PfATM1 (Plasmo DB ID: PF3D7_1352100, amino acids 43–1049) and its domains, PfATM1-NTE (aa 43–305) and PfATM1-CTD (aa 807–1049) were PCR-amplified using primers listed in S2 Table. The sequence encoding PfNBP35 (PF3D7_0910800, aa 23–356) and PfCFD1 (PF3D7_1128500, aa 96–568) were also PCR-amplified from genomic DNA (primers in S2 Table). The resulting PCR product of PfATM1 (3018 bp) was cloned in β-pET28a(+)-β expression vector (gift from Prof. Hiroshi Omote, Okayama University, Japan) [78] to express recombinant PfATM1 with C-terminal and N-terminal 6XHis-tags (Fig 5D). The clone was confirmed by SacI and SalI and DNA sequencing. PfATM1-NTE (786 bp), PfATM1-CTD (729 bp), PfNBP35 (993 bp) and PfCFD1 (1479 bp) were cloned into pET23a expression vector, and confirmed by restriction digestion using SacI and SalI followed by DNA sequencing. DNA segments encoding PfIscU and PfIscA2 [31] were sub-cloned in pGEX-KG vector and confirmed by BamHI and SalI digestion.

For expression of 6X-His tagged β-PfATM1-β (henceforth referred to as recombinant PfATM1), E. coli C43 strain was co-transformed with pET28a-β-PfATM1-β and the RIG plasmid, and grown in LB media containing 50 μg/mL kanamycin sulfate until OD₆₀₀ reached 0.6–0.8 followed by induction with 1 mM IPTG at 20°C for 16 h. For purification, we followed the procedure of [79]. Briefly, after sonication and removal of cell debris and insoluble fraction by centrifugation at 5860 × g, the supernatant was ultracentrifuged (SW41 Ti rotor, Beckman–Coulter) at 150,000 × g for 1 h. The membrane protein containing pellet was suspended and adjusted to 2 mg/ml with buffer (20 mM Tris-Cl pH 7.5, 100 mM NaCl, 10 mM KCl, 5% glycerol), followed by addition of combination of integral membrane protein detergents DDM (0.5%) and Triton X-114 (1%) (Cayman Chemical Company, USA). Membrane proteins were solubilized by incubation on magnetic stirrer for 30 min at 4°C followed by ultracentrifugation at 150,000 × g at 4°C for 1 h. The supernatant containing extracted membrane proteins was diluted two-fold and then purified using Ni-nitrilotriacetic acid (Ni-NTA) Superflow (Qiagen, Germany) pre-equilibrated with the buffer consisting of 20 mM Tris·Cl pH 8.0, 100 mM NaCl, 10 mM KCl and 5% glycerol. After protein binding, the column was washed with buffer containing 10 mM imidazole and 1% octyl glucoside (Cayman Chemical Company, USA) and eluted with buffer containing 300 mM imidazole and 1% octyl glucoside. The purity of eluted fractions was checked on 8% SDS-PAGE. For further purification of PfATM1, size exclusion chromatography (SEC) on Superdex S-200 column (GE Healthcare/Cytiva, USA) was carried out in an AKTA purifier system (GE Healthcare/Cytiva, USA). High molecular weight protein markers (GE Healthcare, USA) were used to plot the standard curve for estimation of molecular weight of PfATM1. To determine the oligomeric status of PfATM1, fractions collected during size exclusion chromatography were electrophoresed on 6% native-PAGE.

For expression, E. coli BL21 (DE3) expression host was co-transformed with RIG plasmid and pET23a-PfATM1-CTD. PfATM1-NTE, PfNBP35 and PfCFD1 were expressed in E. coli BL21 Codon Plus. The transformed cells were grown in LB media supplemented with ampicillin (100 μg/ml) and chloramphenicol (25 μg/ml) at 37°C until OD₆₀₀ reached 0.5, followed by induction with 0.5 mM IPTG for 16 hours at 16°C. pGEX-KG-PfIscU was expressed in E. coli BL21 (DE3) and pGEX-KG-PfIscA2 was expressed in E. coli C43, co-transformed with the RIG plasmid, with induction at 20°C.

Recombinant 6XHis-tagged proteins were purified by affinity chromatography using Ni-NTA Superflow and eluted in 20 mM Tris-Cl pH 7.5, 100 mM NaCl, 10 mM KCl, 5% glycerol and 300 mM imidazole. The GST-tagged proteins (PfIscU and PfIscA2) were purified using glutathione-agarose resin (Macherey-Nagel, Germany) and eluted in 20 mM Tris-Cl pH 7.5, 100 mM NaCl, 10 mM KCl, 5% glycerol and 20 mM reduced glutathione. PfATM1-CTD, PfNBP35 and PfCFD1 were further purified by SEC through Superdex S-75 column (GE Healthcare, USA).

Purified proteins were checked on 10% SDS-PAGE and confirmed by western blotting using anti-6XHis antibody (Ab) (Santa Cruz Biotechnology, USA) for PfATM1, PfATM1-NTE, PfATM1-CTD, PfNBP35 and PfCFD1 and anti-GST antibody (Sigma-Aldrich, USA) for GST-tagged proteins PfIscU and PfIscA2. To cleave the GST-tag, proteins were treated overnight with 100 U of thrombin at 4°C on the affinity column and eluted. The concentration of all purified proteins was determined by Pierce^TM BCA protein assay kit (Thermo Fisher Scientific, USA).

PfATM1 mutants (E972Q and R534E) were generated using Q5 site-directed mutagenesis kit (New England Biolabs, USA) (primers in S2 Table) and confirmed by DNA sequencing.

CD spectroscopy

CD spectra of PfATM1 (5 μM) in 20 mM Tris-Cl pH8 and 100 mM NaCl was scanned at 25°C with a spectral scan range of 190 to 250 nm in a CD spectropolarimeter (J-1500, JASCO, Japan). Total of three accumulations were taken during the scan.

TEM

PfATM1 and reconstituted proteoliposomes were visualized using the negative stain TEM technique. Purified PfATM1 (1mg/ml protein in 20 mM Tris-Cl pH 7.5, 100 mM NaCl) or proteoliposomes (in the same buffer) were allowed to adsorb to freshly glow-discharged carbon coated 400 mesh copper grids and blotted off the edge of the grids using a Whatman filter paper. 2% (w/v) aqueous uranyl acetate was used to perform negative staining followed by air drying. The grids were analysed under a TEM (Jeol JEM1400; Jeol, Japan) at 100 kV after performing astigmatism correction and magnification calibration. Images were collected using an Orius SC200B CCD camera (Gatan, UK) and analysed using Digital Micrograph software (Gatan, UK).

Polyclonal antisera and western blotting

Purified recombinant PfATM1-NTE and PfNBP35 were used as antigens to generate antisera in rabbit (New Zealand White) as described earlier [114]. Antibodies from rabbit polyclonal immune serum were affinity purified using PfATM1-NTE and PfNBP35 proteins immobilized on nitrocellulose membrane.

For western blotting of the parasite lysate, parasites at the mid- to late-trophozoite stage were released from erythrocytes by 0.05% saponin lysis. After washing with cold PBS, the parasite pellet was resuspended in RIPA buffer (50 mM Tris-Cl, 150 mM NaCl, 1.0% (v/v) NP-40, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS at pH 7.4) with 1X protease inhibitor cocktail (Sigma-Aldrich, USA), incubated for 1 hour at 4°C on a tube rotator, sonicated for 2 min, and centrifuged at 13,800 × g for 15 min. The lysate (supernatant) was suspended in 1X Laemmli buffer and electrophoresed on 8 or 10% SDS-PAGE. For western blotting, purified anti-PfATM1 and anti-PfNBP35 Abs were used as primary Abs, and HRP-tagged goat anti-rabbit Ab (1:10,000, Santa Cruz Biotechnology, USA) as secondary Ab. Immobilon Western Chemiluminescent HRP Substrate was used to probe blots using a chemiluminescence detection system (Amersham Image Quant 800, Cytiva, USA).

For immunoblot analysis in T. gondii, pellets from extracellular tachyzoites were resuspended in Laemeli buffer under a reducing condition (1 mM of DTT) and then boiled (at 98°C) for 5 min before loading onto a gradient SDS–PAGE gel for separation. Proteins were transferred to a hybond ECL nitrocellulose membrane using a wet transfer system (Bio-Rad Laboratories, Hercules, USA). Antibodies were diluted in PBS/0.05% Tween20/ 5% skimmed milk. Primary antibodies were monoclonal hybridoma mouse α-Ty (1:10, BB2, Invitrogen, USA) and monoclonal hybridoma mouse anti-myc (1:10, 9E10, Invitrogen, USA). Polyclonal rabbit α-catalase was used as a loading control (1:5,000) [115]. Secondary antibodies (goat anti-mouse, horse radish peroxidase (HRP)-conjugated, Sigma Aldrich, USA; and goat anti-rabbit, HRP conjugated, Sigma Aldrich, USA) were visualized using the SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific, USA). Images were acquired using the Bio-Rad ChemiDoc MP Imaging System and images were processed using Bio-Rad Image Lab software.

[Fe-S] reconstitution and transfer assay

UV visible spectroscopy (spectral scan from 270–700 nm) was performed in Epoch (BioTek, USA). Conversion of holo- to apo-form of PfATM1 was carried out as described earlier [29]. Chemical Reconstitution of [4Fe-4S] on thrombin-cleaved PfIscU and PfIscA2 was carried out as previously described [29, 116]. In-vitro [Fe-S] transfer assays were performed in an anaerobic chamber (PLAS-Labs, USA). For direct transfer of [4Fe-4S] from holo-PfIscU/ holo-PfIscA2 on apo-PfATM1, apo-PfATM1 was incubated with four molar excess of holo-PfIscU or holo-PfIscA2 with or without 2 mM DTT in buffer containing 25 mM Tris–Cl pH 7.5 and 100 mM NaCl for 6 hours at 4°C [31]. The reaction mixture was loaded on Ni-NTA beads, bound PfATM1 was eluted using imidazole, and analysed by UV-VIS scan.

ATPase assay and kinetics

ATPase activity of PfATM1 and its mutants was measured using EnzCheck Phosphate Assay Kit (Invitrogen, Thermo Fisher Scientific, USA). In reactions with GSH/GSSG, 3 mM of each glutathione derivative was used with 500 nM of holo/apo PfATM1 in 200 μl reaction mixture (containing 40 μl MESG substrate, 2 μl PNP enzyme, 10 μl 20 X reaction buffer and 20 μl 10X ATPase buffer). All background control values were subtracted from the readings. Kinetics of ATP hydrolysis in the presence of GSSG was determined with increasing concentrations of ATP and 500 nM PfATM1. K_m, K_cat and catalytic efficiencies (K_cat/K_m)) were calculated using GraphPad Prism 5.

Spectrofluorometric analysis

Variation in intrinsic tryptophan fluorescence was monitored in a solution containing 10 μM PfATM1, 20 mM Tris-Cl pH 8, 100 mM NaCl and 10 mM KCl with increasing concentrations of GSSG (1–10 mM). After excitation at 280 nm, emission spectra were recorded from 300–400 nm in Cary Eclipse Fluorescence Spectrophotometer (Agilent Technologies, USA). Buffer alone with the corresponding GSSG concentration was used as control.

Immunofluorescence assays

P. falciparum-infected erythrocytes at different asexual stages were fixed in glutaraldehyde and paraformaldehyde, followed by permeabilization with 0.2% Triton X-100, and processed as described earlier [106, 117]. After blocking, cells were incubated overnight at 4°C in purified anti-PfATM1 Ab (1:100) and mouse anti PfHU (1:200) as apicoplast marker [60]. For co-labelling with mitochondrial marker dye, cells were incubated with 50 nM Mitotracker Red CMXRos (Invitrogen, Thermo Fisher scientific, USA) prior to fixation. AlexaFluor 488-tagged anti-mouse Ab (1:1000, Invitrogen) and AlexaFluor 568-tagged anti-rabbit Ab (1:1000, Invitrogen) were used to detect PfHU and PfATM1, respectively. AlexaFluor 514- tagged anti-rabbit Ab (1:1000, Invitrogen) was used as secondary antibody to detect PfATM1 in Mitotracker Red labelled cells. Nuclei were stained with DAPI (0.2μg/mL, Sigma-Aldrich). Cells were layered on poly-L-Lysine coated coverslips, washed and mounted in anti-fade mounting media (Invitrogen), and scanned on Leica SP8 confocal microscope under 63X oil immersion objective. Co-localization was quantified using LasX software.

Mitochondria and nuclei of Vivo-morpholino (PfATM1-V_m)-treated P. falciparum 3D7 blood cultured parasites were labelled with 50 nM Mitotracker Red CMXRos (Invitrogen) and DAPI (0.2 μg/ml) for 30 min at 37°C. After three washes with 1X PBS, cells were diluted and spotted on poly-L-Lysine coated coverslips. After a 2 hours incubation and two more washes with 1XPBS, the cover-slips were mounted and scanned as above.

For T. gondii, HFF monolayers grown on coverslips were inoculated with tachyzoites and grown at 37°C. Cells were subsequently fixed with 4% PFA-Glu for 10 min, then quenched with 0.1 M glycine/PBS. Cells were permeabilized (0.2% Triton X-100/PBS), blocked (2% BSA/0.2%Triton/PBS), and probed with primary antibodies (1 h) prior to washing (0.2%Triton/PBS) and incubation with secondary antibodies (Alexa488 or Alexa594 conjugated goat anti-mouse/rabbit, respectively). Coverslips were mounted in DAPI FluoromountG (Invitrogen, USA). Confocal images were generated with a LSM700 laser scanning confocal microscope (Zeiss, Germany). Image stacks were processed with ImageJ. Antibodies used in the study (used dilution): polyclonal rabbit anti-GAP45 (1:5000) [113], monoclonal hybridoma mouse α-Ty (1:10, BB2, Invitrogen, USA), monoclonal hybridoma mouse anti-myc (1:10, 9E10, Invitrogen, USA), polyclonal rabbit anti-chaperone 60 (CPN60, 1:2000) [118], polyclonal rabbit anti- mitochondrial heat shock protein 70 (mHsp70, 1:2,000) [59]. Secondary anti-mouse Alexa fluor 488 (Invitrogen, USA), anti-rabbit Alexa fluor 594 (Invitrogen, USA).

Protein pull-down assay

6XHis-tagged PfATM1-CTD or PfATM1-NTE bound to Ni-NTA beads were used as bait to pull down interacting partners from the parasite lysate as described previously [114]. After elution, proteins were separated on 14% SDS–PAGE followed by western blotting using anti-PfIscU, anti-PfIscA2 [31] and anti-6XHis antibodies in the experiment with PfATM1-CTD as bait. Anti-PfHU was used as control to confirm that PfATM1-CTD bound specifically to ISC pathway components. After pull down from parasite lysate using PfATM1-NTE as bait, samples were loaded on 10% SDS-PAGE. Western blotting was carried out using anti-PfNBP35 and anti-6XHis Abs. Beads alone (without bait protein) and beads tagged with 6XHis-EcEngA (kind gift from Dr. Balaji Prakash) served as negative control baits.

In vitro complexation assay

Purified PfATM1, PfNBP35 and PfCFD1 were used for complex formation in vitro. Equimolar ratio of PfNBP35 and PfCFD1 was incubated for 1 hour or 16 hours at 4°C to detect formation of PfNBP35-PfCFD1 complex. Further, 40 nM each of PfNBP35 with PfCFD1, PfATM1 with either PfNBP35 or PfCFD1, and PfATM1 with PfNBP35+PfCFD1 complex were incubated for 16 hours at 4°C. Native-PAGE loading dye (2.5X TBE, 50% glycerol and 0.1% bromophenol blue) was added to the samples at 1X final concentration. For investigating PfNBP35-PfCFD1 complex formation, samples were electrophoresed on 6–10% gradient native-PAGE. PfATM1 complexation with PfNBP35 and PfCFD1 was detected on 6–8% native-PAGE. Gels were stained with Coomassie Brilliant Blue R-250.

PfATM1-liposomes and transport assay

Purified PfATM1 or PfATM1-E972Q were reconstituted into proteoliposomes using the detergent-mediated method as described by [119]. Briefly, 3:1 molar ratio of cholesterol and phosphatidylethanolamine (PE) (Sigma-Aldrich) dissolved in chloroform [78] was divided into 70 μl aliquots 10 ml test tubes. Nitrogen gas was purged to remove all chloroform. The thin lipid films were placed under vacuum for 4–5 h. The lipid films were then hydrated in 350 μl transport buffer (17 mM Tris-Cl pH7.5, 85 mM NaCl) and bath-sonicated until homogeneity [120]. To prepare unilamellar liposomes, 250 μl of lipid sample was extruded in Avanti mini-extruder with 400 nm polycarbonate filter, pre-equilibrated with transport buffer. Liposomes were stored at 4°C. Purified PfATM1 and PfATM1-E972Q were dialysed in 1% n-octylglucoside (n-OG) containing transport buffer. Unilamellar liposomes were also saturated with 1% n-OG [120]. 230 μl of n-OG saturated unilamellar liposomes and 770 μl of dialysed protein were mixed together. The protein-lipid-detergent mixture was incubated at room temperature for 2 hours and then applied to an NAP5 desalting column (GE Healthcare, USA) pre-equilibrated with transport buffer. The cloudy void fraction which contains proteoliposomes was collected [121]. Detergent was further removed using 20 mg/ml slurry of polystyrene biobeads SM-2 (Bio-Rad Laboratories, USA) added to the proteoliposomes mixture and nutated overnight at 4°C. The supernatant was subjected to ultracentrifugation at 140,000 × g (SW41 Ti rotor, Beckman Coulter) for 45 min to pellet proteoliposomes. Proteoliposomes resuspended in transport buffer were analysed by dynamic light scattering (DLS) scan in Zetasizer Nano ZS (Malvern Panalytical, UK). Control liposomes lacking protein were prepared following the same procedure.

Transport activity was of proteoliposomes was assayed as described by Fan and colleagues [56]. In brief, 1 ml reaction mixture was prepared in transport buffer (85 mM NaCl, and 17 mM Tris-Cl pH 7.5) with 10 mM Mg-ATP [122], 2.5 mM GSSG and 5 mg/ml of proteoliposomes. All reactions were carried out at 37°C for a total duration of 90 min with sample taken out at 15 min intervals. 150 μl sample taken at each time point was added to 10 ml of cold transport buffer. All the samples were ultracentrifuged at 140,000 × g for 10 min at 4°C. Liposomes without protein were taken as a negative control. PfATM1 proteoliposomes without GSSG were used as background control. The pellet obtained after ultracentrifugation was rinsed five times with cold transport buffer, resuspended in 100 μl of solubilization buffer (transport buffer with 2% N-lauroyl sarcosine), and incubated for 2 hours at 4°C. Samples were centrifuged at 14,800 × g to remove bubbles. 50 μl of the supernatant of detergent-lysed liposomes was used for GSSG quantification using the Glutathione Assay Kit (Cayman Chemical Company, USA). GSSG standard curve was used to estimate transported GSSG concentration in all samples. Reactions were carried out in triplicate. The transport rates were not corrected for the orientation of the protein in reconstituted liposomes.

Statistical analysis

Graphs were generated using Origin 5.0 and GraphPad-Prism 10. P values were calculated using two tailed, unpaired t-test.

Cell harvest for mass spectrometry analyses

Following the indicated treatment, freshly egressing parasites (independent triplicates per condition) were harvested. To this end, culture medium was gently removed without disturbing the host cell monolayer, the cell layer was scraped, collected, and passed 3x through 26G needles before removing host-debris through filtration (3 μm pore size, Millipore, USA). Filters were rinsed with additional ice-cold PBS to maximize the yield of parasites. The parasite solution was centrifuged at 2,500 × g for 25 min at 4°C. The supernatant was removed, and parasite numbers adjusted to 1.5x10⁸ parasites per sample. Two additional wash steps (2,500 × g for 10 min) were carried out swiftly at 4°C with cold PBS for GC-MS analyses or with PBS containing protease inhibitor (Pierce Protease Inhibitor, EDTA free, Thermo Scientific, USA) for proteomic analyses. Parasite pellets were snap frozen and stored at −80°C until metabolite extraction.

Metabolite profiling by GC-MS

Proteomics

Extracellular flux analysis (Seahorse XFe96)

The extracellular flux analysis was carried out as previously outlined [64] and described, in brief, here. Freshly egressed parasites were harvested and washed in Seahorse XF Base Medium (Agilent Technologies, USA). The Base Medium used throughout the assay was supplemented with 10 mM glucose (Agilent Technologies, USA), 2 mM glutamine (Agilent Technologies, USA), and 1 mM pyruvate (Agilent Technologies, USA) and adjusted to pH 7.4. Parasites were counted and numbers adjusted to 3x10⁷ parasites/ml. Of this cell solution, 30 μl (9x10⁵) were seeded into wells of a XFe96 culture plate (Agilent Technologies, USA), following its coating with Cell-Tak cell adhesive (Corning, USA), according to the manufacturer’s guidelines. Additional 150 μl of XF Base Medium were added to each well and plates briefly incubated in a 37°C incubator without additional CO₂. Injection ports of XFe96 sensor cartridges were loaded with Oligomycin (port A), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP) (port B), and rotenone/antimycin A (port C) resuspended in XF Base medium to obtain final concentrations of 2.5 μM, 2 μM, and 0.5 μM, respectively. The assays were run as follows: 3 basal measurement cycles, followed by 3 measurement cycles after injection of each compound, with each cycle consisting of 3 min mixing and 3 min measurement of the oxygen consumption rate (pmol O₂/min) and media acidification rate (mpH/min). Each strain/condition was analysed in 8–14 technical replicates in 3 independent experiments. The data was analysed using Wave Software (Agilent Technologies, USA) and Excel (Microsoft, USA). The basal oxygen consumption rate and extracellular acidification rate were determined based on the three initial measurements, while the maximal OCR represents the values from the three measurements after injection of FCCP and the spare capacity is the difference between the basal and maximal respiration rate. Means and standard deviation are provided for each experiment. Values between conditions were compared by performing a One-way ANOVA followed by Tukey multiple pair wise comparisons, using Graph Pad Prism 10. p-values are given for relevant comparisons.

Co-immunoprecipitation

For co-immunoprecipitation, freshly egressed T. gondii parasites were pelleted and lysed in 950 μl of 10 mM HEPES, pH 7.4, 2 mM EDTA, 150 mM NaCl and 10% glycerol with 0.1% of Triton X-100. 50 μl of the lysate was used. After separation of the soluble and insoluble fraction, the pellet was resuspended in 300 μl of 1x of Laemmli buffer. The soluble fraction was separated into two microtubes and incubated on a shaker for 2 hours at 40°C with either anti-Ty (mouse) or anti-myc (mouse) antibodies, kindly gifted by Prof. Ermmano Candolfi, University of Strasbourg, France. Protein A Sepharose CL-4B antibody purification beads (Cytiva, USA) were added following an incubation of 2 hours at 40°C on a shaker. The solution was washed three times with the above buffer. 50 μl were sampled after the first wash for the flow-through. After the last wash, beads were resuspended in 1x Laemmli buffer and loaded into a gradient SDS-PAGE gel. Membranes were incubated with the primary antibodies anti-Ty (rabbit, Sigma-Aldrich) or anti-myc (rabbit), kindly gifted by the Prof. Chris Tonkin, WEHI, Melbourne, Australia.