TgNFS2 and TgISU1 are functional homologs of components of the Fe-S cluster synthesis pathways

Fe-S cluster biosynthesis pathways in the mitochondrion and the plastid follow a similar general pattern: cysteine desulfurases (NFS1, NFS2) produce sulfur from L-cysteine, scaffold proteins (ISU1, SufB/C/D) provide a molecular platform allowing iron and sulfur to meet and form a cluster, and finally carrier proteins (like IscA or SufA) deliver the cluster to target apoproteins [23]. The cytosolic CIA pathway, which is responsible for the de novo formation of Fe-S clusters to be incorporated in cytosolic and nuclear proteins, is dependent on the ISC pathway, as its first step requires the import of a yet unknown sulfur-containing precursor that is translocated to the cytosol from the mitochondrion [24]. To get a general overview of the predicted components for the Fe-S cluster machinery in T. gondii, we conducted homology searches in the ToxoDB.org database [37], using well-characterized proteins from plants (Arabidopsis thaliana) belonging to the SUF, ISC and CIA pathways (S1 Table). Data from global mapping of protein subcellular location by hyperLOPIT spatial proteomics [38] was in general in good accordance with the expected localization of the homologs (with the noticeable exception of members of the NBP35/HCF101 ATP-binding proteins). Overall, our search revealed that T. gondii appears to have a good conservation of all the main components of the three ISC, SUF and CIA Fe-S synthesis pathways (S1 Table and Fig 1A). Additional information available on ToxoDB.org such as scores from a CRISPR/Cas9-based genome-wide screening [39], highlighted that most components of the three pathways are predicted to be important for parasite fitness. This suggests several Fe-S proteins localizing to the endosymbiotic organelles, but also the cytosol/nucleus, are essential for the optimal growth of tachyzoites. In order to verify this, our aim was to specifically interfere with the apicoplast-localized SUF pathway or the mitochondrion-localized ISC pathway in T. gondii tachyzoites. More precisely, we wanted to target the homologs of A. thaliana NFS2 and ISU1, which are both central (and presumably essential) to their respective pathways (Fig 1A): NFS2 is a cysteine desulfurase that provides the inorganic sulfur part of plastidic Fe-S clusters, while ISU1 is a scaffold protein important for cluster assembly in the mitochondrial pathway. Interfering with these key players of the upstream machinery would likely lead to a comparable disruption of the respective Fe-S cluster biogenesis pathways hosted by each organelle [40].

Download:

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

Fig 1. TgNFS2 and TgISU1 are functional homologs of components of iron-sulfur cluster synthesis pathways.

A) Putative Fe-S cluster synthesis pathways and associated molecular machinery in Toxoplasma. B) Functional complementation of bacterial mutants for IscU (top) and SufS (bottom). Growth of ‘wild-type’ (WT) E. coli K12 parental strain, bacterial mutant strains and strains complemented (‘comp’) by their respective T. gondii homologues (‘comp’), was assessed by monitoring the optical density at 600 nm in the presence or not of an iron chelator (2,2’-bipyridyl, ‘chel’). Values are mean from n = 3 independent experiments ±SEM. * denotes p ≤ 0.05, Student’s t-test, when comparing values obtained in the absence of chelator for mutant strains versus complemented ones.

https://doi.org/10.1371/journal.ppat.1010096.g001

As a first step, we sought to determine whether TgNFS2 (TGGT1_216170) and TgISU1 (TGGT1_237560) were real functional homologs by performing complementation assays of bacterial mutants. In many bacteria, the ISC machinery is the primary system for general Fe-S cluster biosynthesis, while the SUF system plays a similar general role, but is mostly operative under stress conditions (like iron limitation or oxidative stress). In Escherichia coli, both pathways are partially redundant, but their individual disruption results in slowed bacterial growth, especially when limiting iron availability with a specific chelator [41]. Expression of the predicted functional domains of TgNFS2 and TgISU1 in mutant strains for the corresponding E. coli proteins (named SufS and IscU, respectively) improved bacterial growth, in the presence of an iron chelator or not (Fig 1B). The complementation seemed partial as complemented strains remained more sensitive to the iron chelator than the wild-type strain. Yet, and although stationary phase was reached earlier than for the wild-type bacteria, contrarily to the mutants the complemented strains showed a bacterial density close to that of the WT at stationary phase. This suggests TgNFS2 and TgISU1, in addition to a good sequence homology with their bacterial homologues (S1 Fig), have a conserved function.

We next determined the sub-cellular localizations of TgNFS2 and TgISU1 by epitope tagging of the native proteins. This was achieved in the TATi ΔKu80 cell line, which favors homologous recombination and would allow transactivation of a Tet operator-modified promoter we would later use for generating a conditional mutant in this background [42–44]. A sequence coding for a C-terminal triple hemagglutinin (HA) epitope tag was inserted at the endogenous TgNFS2 or TgISU1 locus by homologous recombination (S2 Fig). Using the anti-HA antibody, by immunoblot we detected two products for each protein (Fig 2A and 2B), likely corresponding to their immature and mature forms (ie after cleavage of the transit peptide upon import into the organelle). Accordingly, the analysis of TgNFS2 and TgISU1 sequences with several subcellular localization and N-terminal sorting signals site predictors confirmed they likely contained sequences for plastidic and mitochondrial targeting [45], respectively, although no consensus position of the exact cleavage sites could be determined. Immunofluorescence assay (IFA) in T. gondii tachyzoites confirmed HA-tagged TgNFS2 and TgISU1 co-localize with markers of the apicoplast and the mitochondrion, respectively (Fig 2C and 2D).

Download:

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

Fig 2. TgNFS2 and TgISU1 localize to the apicoplast and the mitochondrion, respectively.

Detection by immunoblot of C-terminally HA-tagged TgNFS2 (A) and TgISU1 (B) in parasite extracts reveals the presence of both precusor (p) and mature (m) forms of the proteins. Anti-actin (TgACT1) antibody was used as a loading control. Immunofluorescence assay shows TgNFS2 co-localizes with apicoplast marker TgCPN60 (C) and TgISU1 co-localizes with mitochondrial marker F1 β ATPase (D). Scale bar represents 5 μm. DNA was labelled with DAPI. DIC: differential interference contrast.

https://doi.org/10.1371/journal.ppat.1010096.g002

NFS2 is a cysteine desulfurase whose activity is enhanced by an interaction with the SUFE protein [46]. Similar to plants that express several SUFE homologues [47], there are two putative SUFE-like proteins in T. gondii (S1 Table), one of which was already predicted to reside in the apicoplast by hyperLOPIT (TgSUFE1, TGGT1_239320). We generated a cell line expressing an HA-tagged version of the other, TgSUFE2 (TGGT1_277010, S3A–S3C Fig), whose localization was previously unknown. Like for TgNFS2, several programs predicted a plastidic transit peptide, which was confirmed by immunoblot analysis (detecting TgSUFE2 immature and mature forms, S3D Fig). IFA showed TgSUFE2 co-localizes with an apicoplast marker (S3E Fig). This further confirms that the initial steps of Fe-S cluster biogenesis in the apicoplast are likely functionally-conserved.

Disruption of either the plastidic or the mitochondrial Fe-S cluster pathway has a profound impact on parasite growth

In order to get insights into plastidic and mitochondrial Fe-S biogenesis, we generated conditional mutant cell lines in the TgNFS2-HA or TgISU1-HA-expressing TATi ΔKu80 background [44]. Replacement of the endogenous promoters by an inducible-Tet07SAG4 promoter, through a single homologous recombination at the loci of interest (S4 Fig), yielded TgNFS2 and TgISU1 conditional knock-down cell lines (cKD TgNFS2-HA and cKD TgISU1-HA, respectively). In these cell lines, the addition of anhydrotetracycline (ATc) can repress transcription through a Tet-Off system [48]. For each cKD cell line several transgenic clones were obtained and were found to behave similarly in the initial phenotypic assays we performed, so only one was analysed further. Transgenic parasites were grown for various periods of time in presence of ATc, and protein down-regulation was evaluated. Immunoblot and IFA analyses of cKD TgNFS2 -HA and cKD TgISU1-HA parasites showed that the addition of ATc efficiently down-regulated the expression of TgNFS2 (Fig 3A and 3C) and TgISU1 (Fig 3B and 3D), and most of the proteins were undetectable after two days of incubation.

Download:

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

Fig 3. Efficient down-regulation of TgNFS2 and TgISU1 expression with anhydrotetracyclin (ATc).

A) Immunoblot analysis with anti-HA antibody shows efficient down-regulation of TgNFS2 after 48h of incubation with ATc. Anti-SAG1 antibody was used as a loading control. B) Immunoblot analysis with anti-HA antibody shows efficient down-regulation of TgISU1 after 24h of incubation with ATc. Anti-SAG1 antibody was used as a loading control. C) and D) Immunofluorescence assays show TgNFS2 and TgISU1 are not detectable anymore after 48h of incubation with ATc. Scale bar represents 5 μm. DNA was labelled with DAPI. DIC: differential interference contrast.

https://doi.org/10.1371/journal.ppat.1010096.g003

We also generated complemented cell lines expressing constitutively an additional copy of TgNFS2 and TgISU1 from the uracil phosphoribosyltransferase (UPRT) locus from a tubulin promoter in their respective conditional mutant backgrounds (S5A and S5B Fig). We confirmed by semi-quantitative RT-PCR (S5C Fig) that the transcription of TgNFS2 and TgISU1 qenes was effectively repressed in the cKD cell lines upon addition of ATc, whereas the corresponding complemented cell lines exhibited a high transcription level regardless of ATc addition (due to the expression from the strong tubulin promoter).

We next evaluated the consequences of TgNFS2 and TgISU1 depletion on T. gondii growth in vitro. First, to assess the impact on the parasite lytic cycle, the capacity of the mutants and complemented parasites to produce lysis plaques was analyzed on a host cells monolayer in absence or continuous presence of ATc for 7 days (Fig 4A and 4B). Depletion of both proteins completely prevented plaque formation, which was restored in the complemented cell lines. To assess whether this defect in the lytic cycle is due to a replication problem, all cell lines were preincubated in absence or presence of ATc for 48 hours and released mechanically, before infecting new host cells and growing them for an additional 24 hours in ATc prior to parasite counting. We noted that incubation with ATc led to an accumulation of vacuoles with fewer parasites, yet that was not the case in the complemented cell lines (Fig 4C and 4D). Overall, these data show that either TgNFS2 or TgISU1 depletion impacts parasite growth.

Download:

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

Fig 4. Depletion of TgNFS2 or TgISU1 affects in vitro growth of the tachyzoites.

Plaque assays were carried out by infecting HFF monolayers with the TATi ΔKu80 cell line, the cKD TgNFS2-HA (A) or the cKD TgISU1-HA (B) cell lines, or parasites complemented with a wild-type version of the respective proteins. They were grown for 7 days ± ATc. Measurements of lysis plaque areas are shown on the right and highlight a significant defect in the lytic cycle when TgNFS2 (A) or TgISU1 (B) were depleted. Values are means of n = 3 experiments ± SEM. Mean value of TATi ΔKu80 control was set to 100% as a reference. **** denotes p ≤ 0.0001, Student’s t-test. Scale bars = 1mm. TgNFS2 (C) and TgISU1 (D) mutant and complemented cell lines, as well as their parental cell lines and the TATi ΔKu80 control, were grown in HFF in the presence or absence of ATc for 48 hours, and subsequently allowed to invade and grow in new HFF cells for an extra 24 hours in the presence of ATc. Parasites per vacuole were then counted. Values are means ± SEM from n = 3 independent experiments for which 200 vacuoles were counted for each condition. E) Plaque assays for the TgNFS2 and TgISU1 mutants were performed as described in A) and B), but ATc was washed out after 7 days (7 days+ 7 days-) or not (14 days+), and parasites were left to grow for an extra 7 days. Plaque area was measured. Data are means ± SEM from three independent experiments. *** p ≤ 0.001, Student’s t-test. Arrowheads show plaques forming in the TgISU1 upon ATc removal. Scale bar = 1mm.

https://doi.org/10.1371/journal.ppat.1010096.g004

Then, we sought to assess if the viability of the mutant parasites was irreversibly affected. We thus performed another series of plaque assays, but at the end of the 7-day incubation, we washed out the ATc, incubated the parasites for an extra 7 days in the absence of ATc and evaluated plaque formation (Fig 4E). In these conditions, cKD TgNFS2-HA parasites displayed very small plaques suggesting their viability was irreversibly impacted. In contrast, cKD TgISU1-HA parasites showed plaques, suggesting parasite growth had at least partly resumed after ATc washout, while host cell lysis remained limited if the drug was kept continuously during the same period of time. This suggests that although depletion of TgISU1 has a marked impact on parasite growth, it is not completely lethal.

We performed IFAs to assess possible morphological defects that may explain the impaired growths of cKD TgNFS2-HA and cKD TgISU1-HA parasites. We stained the apicoplast and mitochondrion of parasites kept in the continuous presence of ATc for several days. cKD TgNFS2-HA parasites managed to grow and egress after three days and were seeded onto new host cells, where there were kept for two more days in the presence of ATc. During this second phase of intracellular development, and in accordance with the replication assays (Fig 4C), growth was slowed down considerably. Strikingly, while the mitochondrial network seemed normal, we noticed a progressive loss of the apicoplast marker TgCPN60 (Fig 5A), which was quantified (Fig 5B). As this could reflect a specific impact on this protein marker rather than a loss of the organelle, we also stained the parasites with fluorescent streptavidin, which mainly detects the biotinylated apicoplast protein acetyl-CoA carboxylase [49], confirming a similar loss of signal (S6 Fig). This suggests there is a general impact of TgNFS2 depletion on the organelle. Moreover, the fact that TgNFS2-depleted parasites eventually managed to perform a first lytic cycle and reinvade host cells before being blocked (Fig 5A) is reminiscent to the “delayed death” effect observed in apicoplast-defective parasites [6,50,51]. On the other hand, we were able to grow cKD TgISU1-HA parasites for five days of continuous culture: they developed large vacuoles and showed little sign of egress from the host cells (Fig 5C). Both the mitochondrion and the apicoplast appeared otherwise normal morphologically. These large vacuoles could reflect a defect in the parasite egress stage of the lytic cycle [52]. We thus performed an egress assay on cKD TgISU1-HA parasites that were kept for up to five days in the presence of ATc, and they were able to egress normally upon addition of a calcium ionophore (Fig 5D). These large vacuoles are also reminiscent of cyst-like structures [53], so alternatively this may reflect spontaneous stage conversion. Cysts are intracellular structures that contain the slow-growing form of T. gondii, called the bradyzoite stage (which is responsible for the chronic phase of the disease), and they may appear even during in vitro growth in particular stress conditions [54]. Yet, this mutant cell line was generated in a type I T. gondii strain, which is associated with acute toxoplasmosis in the mouse model [55], and typically does not spontaneously form cysts. So, to be confirmed this hypothesis needed further investigations, as we will see later in the manuscript.

Download:

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

Fig 5. Impact of TgNFS2 and TgISU1 depletion on intracellular tachyzoites.

A) Depletion of TgNFS2 impacts the apicoplast. cKD TgNFS2-HA parasites were kept in the presence of ATc and the aspect of the apicoplast and mitochondrion was evaluated by microscopic observation using specific markers (CPN60 and F1β ATPase, respectively). After 72 hours, parasites egressed and were used to reinvade new host cells for subsequent timepoints. Scale bar represents 5 μm. DNA was labelled with DAPI. DIC: differential interference contrast. B) Quantification of apicoplast loss in vacuoles containing cKD TgNFS2-HA parasites after 72 to 120 hours of incubation with ATc. Data are mean values from n = 3 independent experiments ±SEM. ** p ≤ 0.01, **** p ≤ 0.0001, Student’s t-test. C) Depletion of TgISU1 does not impact mitochondrial and overall parasite morphologies, but affects parasite growth. cKD TgISU1-HA parasites were grown in the presence of ATc for up to five days and the aspect of the apicoplast and mitochondrion was evaluated by microscopic observation using specific markers described in A). Growth in the presence of ATc was continuous for up to five days. Scale bar represents 5 μm. DNA was labelled with DAPI. DIC: differential interference contrast. D) Egress is not affected by TgISU1depletion. An egress assay was performed using calcium ionophore A23187. On the left are representative images of vacuoles containing parasites that egressed normally or did not. GRA3 (parasitophorous vacuole marker) staining is shown in green and GAP45 (parasite periphery marker) in red. Scale bars = 10 μm. On the right is the quantification of egress for cKD TgISU1-HA parasites kept in the presence of ATc or not. Mean values ± SEM from n = 3 independent biological experiments are represented.

https://doi.org/10.1371/journal.ppat.1010096.g005

In any case, our data show that interfering with the plastidic and mitochondrial Fe-S protein pathways both had important consequences on parasite growth, but had a markedly different impact at a cellular level.

Use of label-free quantitative proteomics to identify pathways affected by TgNFS2 or TgISU1 depletion

There is a wide variety of eukaryotic cellular processes that are depending on Fe-S cluster proteins. To get an overview of the potential T. gondii Fe-S proteome, we used a computational tool able to predict metal-binding sites in protein sequences [56], and performed subsequent manual curation to refine the annotation. We identified 64 proteins encompassing various cellular functions or metabolic pathways that included, beyond the Fe-S synthesis machinery itself, several DNA and RNA polymerases, proteins involved in redox control and electron transfer, and radical S-adenosylmethionine (SAM) enzymes involved in methylation and methylthiolation (S2 Table). HyperLOPIT data or manual curation helped us assign a putative localization for these candidates. A considerable proportion (19%) of these were predicted to localize to the nucleus, where many eukaryotic Fe-S proteins are known to be involved in DNA replication and repair [57]. Yet, strikingly, most of the predicted Fe-S proteins likely localize to the endosymbiotic organelles. Several (19%) are predicted to be apicoplast-resident proteins, including radical SAM enzymes lipoate synthase (LipA) [58] and MiaB, a tRNA modification enzyme [59], as well as the IspG and IspH oxidoreductases of the non-mevalonate isoprenoid pathway [60]. Finally, for the most part (43%), candidate Fe-S proteins were predicted to be mitochondrial, with noticeably several important proteins of the respiratory chain (SDHB, the Fe-S subunit of the succinate dehydrogenase complex, Rieske protein and TgApiCox13) [61–63], but also enzymes involved in other metabolic pathways such as heme or molybdopterin synthesis. CRISPR/Cas9 fitness scores [39] confirmed many of these putative Fe-S proteins likely support essential functions for parasite growth.

We sought to confirm these results experimentally. Thus, in order to uncover the pathways primarily affected by the depletion of TgISU1 and TgNFS2, and to identify potential Fe-S protein targets, we conducted global label-free quantitative proteomic analyses. Like most plastidic or mitochondrial proteins, candidate Fe-S acceptors residing in these organelles are nuclear-encoded and thus need to be imported after translation and have to be unfolded to reach the stroma of the organelle. This not only implies the addition of the Fe-S cofactor should happen locally in the organelle, but also that this may have a role in proper folding of these proteins. We thus assumed that disrupting a specific pathway may have a direct effect on the stability and expression levels of local Fe-S proteins. Cellular downstream pathways or functions may also be affected, while other pathways may be upregulated in compensation. Parasites were treated for two days with ATc (cKD TgISU1-HA) or three days (cKD TgNFS2-HA, as it takes slightly longer to be depleted, Fig 3A), prior to a global proteomic analysis comparing protein expression with the ATc-treated TATi ΔKu80 control. For each mutant, we selected candidates with a log2(fold change) ≤-0.55 or ≥0.55 (corresponding to a ~1.47-fold change in decreased or increased expression) and a p-value <0.05 (ANOVA, n = 4 biological replicates) (S3 and S4 Tables and Fig 6A and 6B). To get a more exhaustive overview of proteins whose amounts varied drastically, we completed this dataset by selecting some candidates that were consistently and specifically absent from the mutant cell lines or only expressed in these (S3 and S4 Tables).

Download:

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

Fig 6. Change in protein expression induced by TgNFS2 and TgISU1 depletion.

Volcano plots showing the protein expression difference based on label-free quantitative proteomic data from TgNFS2 (A) and TgISU1 (B) mutants grown in the presence of ATc. X-axis shows log2 fold change versus the TATi ΔKu80 control grown in the same conditions, and the Y-axis shows -log10(p value) after ANOVA statistical test for n = 4 independent biological replicates. Selected variant protein categories are highlighted in color. C) Venn diagram representation of the shared and unique proteins whose expression is affected by the depletion of TgNFS2 and TgISU1. D) and E): mapping of less or more abundant candidates (circled dots) in the TgNFS2 and TgISU1 mutants, respectively, on the spatial proteome map representation of the hyperLOPIT data, highlighting clusters denoting putative subcellular localization. Full details available at: https://proteome.shinyapps.io/toxolopittzex/.

https://doi.org/10.1371/journal.ppat.1010096.g006

Overall, depletion of TgISU1 led to a higher variability in protein expression and while the pattern of expression was essentially specific for the respective mutants, a number of shared variant proteins were found (Fig 6C and S5 Table). For instance, common lower expressed candidates include a SAM synthase, possibly reflecting a general perturbation of SAM biosynthesis upon loss of function of Fe-S-containing radical SAM enzymes [64]. Using dedicated expression data [65,66] available on ToxoDB.org we realized that, strikingly, many of the common variant proteins were stage-specific proteins (S5 Table). For instance, the protein whose expression went down the most is SAG-related sequence (SRS) 19F. The SRS family contains GPI-anchored surface antigens related to SAG1, the first characterized T. gondii surface antigen, and whose expression is largely stage-specific [67]. This protein, SRS19F, may be most highly expressed in stages present in the definitive host [66,68]. Conversely, SRS44, also known as CST1 and one of the earliest marker of stage conversion to bradyzoites [69], was upregulated in both mutants. Several other bradyzoite proteins whose expression increased included Ank1, a tetratricopeptide-repeat protein highly upregulated in the cyst-stages but not necessary for stage conversion [70], aspartyl protease ASP1, an α-galactosidase, as well as several dense granule proteins (GRA). Dense granules are specialized organelles that secrete GRA proteins that are known to participate in nutrient acquisition, immune evasion, and host cell-cycle manipulation. Many GRA have been characterized in the tachyzoite stage, but several are stage-specific and expressed in bradyzoites [71]. It should be noted that bradyzoite-specific proteins were generally much strongly expressed upon TgISU1 depletion than TgNFS2 depletion. Nevertheless, altogether these results show that altering either the plastidic or the mitochondrial Fe-S cluster synthesis pathway led to an initial activation of the expression of some markers of the bradyzoite stage, whose involvement in the stress-mediated response is well documented [54].

Depletion of TgNFS2 has an impact on the apicoplast, but also beyond the organelle

We next focused on proteins that varied specifically upon depletion of TgNFS2 (S3 Table). Using the hyperLOPIT data available on ToxoDB.org, we assessed the putative localization of the candidates (Fig 6D and S7A Fig) and we also defined putative functional classes based on manual curation (Fig 6A and S7B Fig). Surprisingly, few apicoplast proteins were impacted. This could reflect a limited impact on apicoplast Fe-S apoproteins, but this is in contradiction with the late, yet pronounced, effect we see on the organelle in the absence of TgNFS2 (Fig 5A and 5B). There might also be a bias due to an overall low protein abundance: less than half of the apicoplast candidates of the predicted Fe-S proteome (S2 Table) were robustly detected even in the control for instance, including our target protein TgNFS2. Finally, of course it is possible that depletion of Fe-S clusters, while impacting the functionality of target proteins, did not have a considerable effect on their abundance. We sought to verify this for apicoplast stroma-localized LipA, a well-established and evolutionarily-conserved Fe-S cluster protein, which was found to be only marginally less expressed in our analysis (S3 Table). LipA is responsible for the lipoylation of a single apicoplast target protein, the E2 subunit of the pyruvate dehydrogenase (PDH) [33]. Using an anti-lipoic acid antibody on cKD TgNFS2-HA protein extracts, we could already see a marked decrease in lipoylated PDH-E2 after only one day of ATc incubation (Fig 7A). This was not due to a general demise of the apicoplast as it considerably earlier than the observed loss of the organelle (Fig 5A and 5B), and levels of the CPN60 apicoplast marker were clearly not as markedly impacted (Fig 7A). This is also unlikely to reflect a general decrease of TgPDH-E2 levels upon TgNFS2 knock-down, as our quantitative proteomics data, which was performed after 3 days of ATc incubation, show the same amount of unique peptides for TgPDH-E2 (32 ± 2.5 for the control and 32.25 ± 2.75 for the mutant, ~60% of sequence coverage). This finding suggests apicoplast Fe-S-dependent activities may be specifically affected in this mutant, which would happen before observing the general demise and loss of the organelle. Long term incubation of cKD TgNFS2-HA parasites with ATc and co-staining with apicoplast and inner membrane complex (IMC) markers, revealed general cell division defects, including organelle segregation problems and an abnormal membranous structures (Fig 7B). Overall, this suggests impacting the Fe-S cluster synthesis pathway in the apicoplast had important consequences beyond the organelle itself.

Download:

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

Fig 7. TgNFS2 depletion impacts apicoplast-related pathways and has deleterious effects on parasite replication.

A) A decrease in the lipoylation of the E2 subunit of pyruvate dehydrogenase (TgPDH-E2), which depends on the Fe-S-containing lipoyl synthase called LipA in the apicoplast, was observed by immunoblot using an anti-lipoic acid antibody on cell extracts from cKD TgNFS2-HA parasites kept with ATc for an increasing period of time. TgCPN60 was used as a control for apicoplast integrity. TgSAG1 was used as a loading control. Decrease of lipoylated TgPDH-E2 was quantified by band densitometry and normalized with the internal loading control. Data represented are mean ±SEM of n = 3 independent experiments. ** p ≤ 0.01, *** p ≤ 0.001 ANOVA comparison. B) cKD TgNFS2-HA parasites that were grown in the presence of ATc for 5 days were co-stained with anti-TgIMC3 (to outline parasites and internal buds) and anti-CPN60 (an apicoplast marker), which highlighted abnormal membrane structures and organelle segregation problems. Scale bar represents 5 μm. DNA was labelled with DAPI. DIC: differential interference contrast.

https://doi.org/10.1371/journal.ppat.1010096.g007

Depletion of TgISU1 impacts the mitochondrial respiratory chain

We also analyzed the proteins whose abundance changed upon TgISU1 depletion (S4 Table). Again, we used hyperLOPIT data to determine the localization of variant proteins (Fig 6E and S8A Fig) and we also inferred their potential function from GO terms or manual curation (Fig 6B and S8B Fig). Depletion of TgISU1 had a notable impact locally, as numerous mitochondrial proteins were found in lower abundance. Remarkably, most of these proteins were identified as members of the mitochondrial respiratory chain: while our proteomic study detected peptides corresponding to as much as 92% of the hyperLOPIT-predicted mitochondrial membrane and soluble proteins, 12% of these were found to be significantly less expressed upon TgISU1 depletion, 60% of which are known mitochondrial ETC components (Figs 6A and 8A and S4 Table). This suggests a specific effect of TgISU1 depletion on the mitochondrial ETC. This ETC comprises four complexes in Apicomplexa (which typically lack complex I), in which several Fe-S proteins have important function. As mentioned earlier, they include the Fe-S subunit of the succinate dehydrogenase complex (SDHB, part of complex II), the Rieske protein (part of complex III) and TgApiCox13 (part of complex IV) [61–63]. Not only these three Fe-S cluster proteins were found to be less expressed upon TgISU1 depletion, but about 60% and 80% of complexes III and IV components (including recently characterized parasite-specific subunits [62,63]), respectively, were also significantly less abundant (S4 Table and Fig 8A). The impact on components of complexes III and IV beyond their respective Fe-S-dependent subunits is not surprising: as shown by others in T. gondii depletion of selected members of a mitochondrial ETC complex can result in stalled assembly or impaired stability of the whole complex [61–63,72].

Download:

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

Fig 8. TgISU1 depletion impacts the mitochondrial respiratory chain.

A) Schematic representation of the T. gondii mitochondrial respiratory chain; listed are the subunits of the different complexes that were found less abundant upon TgISU1 depletion; Fe-S proteins are highlighted in green. SDH: succinate dehydrogenase; CoQ: coenzyme Q B) Immunofluorescence analysis of cKD TgISU1 parasites expressing HA-tagged TgSDHB (top) or TgApiCox13 (bottom) showing a decrease in expression in the HA-tagged candidates after 3 days of incubation with ATc. Images were taken with the same exposure time for similar channels. TgHSP29 was used as a mitochondrial marker. Scale bar represents 10 μm. DNA was labelled with DAPI. C) Immunoblot analysis of TgSDHB (top) and TgApiCox13 (bottom) levels upon depletion of TgISU1 after 3 days of incubation with ATc. TgACT1 was used as a loading control. Protein levels were quantified by band densitometry and normalized with the internal loading control. Data represented are mean ±SEM of n = 5 independent experiments. ** p ≤ 0.01, *** p ≤ 0.001, ANOVA comparison. D) Impact of TgISU1 depletion on the parasite mitochondrial membrane potential was measured by JC-1 labelling. cKD TgISU1-HA parasites were grown or not in the presence of ATc, mechanically released from their host cells and labelled with the JC-1 dye. This dye exhibits potential-dependent accumulation in the mitochondrion, indicated by a switch from green fluorescence for the monomeric form of the probe, to a concentration-dependent formation of red aggregates (inset, DNA is labelled with DAPI and shown in blue, scale = 1μm). Parasites were then analysed by flow cytometry. Unlabelled parasites (no JC-1) was used as a control for gating. Numbers represent the percentage of cells in each of the subpopulations (P1, P2, P3). One representative experiment out of n = 3 biological replicates is shown.

https://doi.org/10.1371/journal.ppat.1010096.g008

We sought to verify the impact of TgISU1 depletion on proteins of the mitochondrial respiratory chain by tagging two candidates, TgSDHB and TgApiCox13. In order to do this, due to the lack of efficient selectable marker in the cKD TgISU1-HA cell line, we first had to generate a new independent cKD TgISU1 untagged mutant from the TATi ΔKu80 cell line (S9A and S9B Fig). In this cell line, we verified proper regulation of TgISU1 expression by ATc (S9C Fig) and impact of TgISU1 depletion on parasite growth (S9D Fig). In this cKD TgISU1 mutant, a sequence coding for a C-terminal triple HA epitope tag was inserted at the endogenous TgSDHB or TgApiCox13 locus by homologous recombination (S10 Fig). We then incubated these parasites with ATc for three days and used an anti-HA antibody to detect and quantify the proteins of interest by IFA (Fig 8B) and immunoblot (Fig 8C). In accordance with the quantitative proteomics data, both TgSDHB and TgApiCox13 were found to be less expressed in absence of TgISU1.

This suggested the mitochondrial membrane potential and consequently the respiratory capacity of the mitochondrion were likely altered in the absence of a functional mitochondrial Fe-S cluster synthesis pathway. To verify this, we performed flow cytometry quantification using JC-1, a monomeric green fluorescent carbocyanine dye that accumulates as a red fluorescent aggregates in mitochondria depending on their membrane potential (Fig 8D). Depletion of TgISU1 led to a marked decrease of the parasite population displaying a strong red signal (Fig 8D). The effect was maximal after two days of ATc treatment and not further increased by a four-day treatment, which is consistent with the quantitative proteomics data already showing strong impact on proteins from complexes II, III and IV after only two days of ATc treatment. It should be noted that although we believe the drop in mitochondrial membrane potential is likely due to a specific alteration of the respiratory chain, it may also be due to a loss of parasite viability. Yet, reassuringly our results are in line with the recent findings obtained by Aw et al., who generated a mutant of mitochondrial Fe-S cluster synthesis (by depleting TgNFS1), and observed a sharp decrease in TgSDHB abundance and a clear drop in mitochondrial O₂ consumption rate [25].

Concomitantly to the lesser expression of mitochondrial respiratory chain subunits, the proteomics analysis revealed TgISU1 depletion induced a significant increase in cytosolic enzymes involved in glycolysis, as well as its branching off pentose phosphate pathway (Fig 8A and 8B and S4 Table). The upregulation of glycolytic enzymes potentially reflects a metabolic compensation for mitochondrial defects in energy production due to the impairment of the respiratory chain. Other proteins whose abundance was markedly decreased were predicted to be cytoplasmic or nuclear, including proteins involved in DNA repair and replication (S4 Table), which is perhaps unsurprising as the cytosolic CIA Fe-S cluster assembly pathway is supposedly dependent from the ISC pathway [24]. Finally, the changes in abundance of several RNA-binding proteins involved in mRNA half-life or transcription/translation regulation may also reflect adaptation to a stress (S4 Table).

Depletion of TgISU1 initiates conversion into bradyzoites

Another feature highlighted by the quantitative proteomics analysis of the TgISU1 mutant is the change in the expression of stage-specific proteins (S4 Table). The expression of several bradyzoite-specific proteins including GRAs and proteins of the SRS family, was strongly increased. At the same time, some tachyzoite-specific SRS and GRA proteins were found to be less expressed. This was supporting the idea that intracellularly developing parasites lacking TgISU1 may convert into bona fide cyst-contained bradyzoites, as suggested by our initial morphological observations (Fig 5C). To verify this experimentally, we used a lectin from the plant Dolichos biflorus (DBL), which recognizes the SRS44/CST1 glycoprotein that is exported to the nascent wall of differentiating cysts [69]. We could see that during continuous growth of cKD TgISU1-HA parasites in the presence of ATc, there was an increasing number of DBL-positive structures (Fig 9A). This was quantified during the first 48 hours of intracellular development (Fig 9B) and, interestingly, was shown to mimic the differentiation induced by nitric oxide, a known factor of stage conversion [73], and a potent damaging agent of Fe-S clusters [74]. We combined RNAseq expression data for tachyzoite and bradyzoite stages [66] to establish a hierarchical clustering of the SRS proteins detected in our quantitative proteomics experiments for the two mutants (Fig 9C). Overall, this revealed an increase in the expression of bradyzoite-specific SRS in the TgISU1 mutant, although not all bradyzoite-specific SRS were strongly increased, perhaps reflecting an atypical or incomplete stage conversion. As mentioned earlier, some were also upregulated in the TgNFS2 mutant but in much lesser proportions. The strongest increase in bradyzoite-specific SRS expression upon TgNFS2 depletion was for SRS44/CST1, which happens to be the protein DBL preferentially binds to [69]. However, contrarily to the TgISU1 mutant, labelling experiments did not indicate any detectable increase in DBL recruitment in the TgNFS2 mutant (Fig 9B), confirming that impairing the plastidic Fe-S center synthesis pathway does not trigger full stage conversion.

Download:

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

Fig 9. Depletion of TgISU1 triggers parasite differentiation.

A) cKD TgISU1-HA parasites were grown in the presence of ATc and labelled with anti-TgIMC3 (to outline parasites) and a lectin of Dolicos biflorus (DBL) to specifically outline nascent cyst walls. Scale bar represents 10 μm. DNA was labelled with DAPI. DIC: differential interference contrast. B) Quantification of DBL-positive vacuoles after 24 hours or 48 hours of culture of 1) the cKD TgISU1-HA and cKD TgNFS2-HA mutants in the presence or absence of ATc 2) the Tati ΔKu80 cell line, as a negative control, 3) the Tati ΔKu80 cell line in the presence of 100μM nitric oxide (NO), as a positive control. Data are from n = 3 independent experiments. Values are mean ±SEM. * p ≤ 0.05, ** p ≤ 0.01, Student’s t-test C) Clustering of bradyzoite (Bz) or tachyzoite (Tz)-specific proteins of the SRS family shows specific enrichment of bradyzoite proteins upon TgISU1 depletion. D) The cKD TgISU1-HA mutant was grown for up to 14 days in the presence of ATc and co-stained with early cyst wall marker DBL together with tachyzoite marker SAG1, or intermediate (P18/SAG4), or late (P21) bradyzoite markers. Scale bar represents 10 μm. DNA was labelled with DAPI. E) Measurement of the cyst area size after growing the cKD TgISU1-HA mutant for 7 and 14 days in the presence of ATc and labelling the cyst wall with DBL and measuring the surface of 60 cysts per condition. Mean ±SD is represented. One representative experiment out of three independent biological replicates is shown. **** denotes p ≤ 0.0001, Student’s t-test.

https://doi.org/10.1371/journal.ppat.1010096.g009

Stage conversion is a progressive process that happens over the course of several days, as it involves the expression of distinct transcriptomes and proteomes [54]. Markers for specific steps of in vitro cyst formation had been previously described [75], so we have used several of these to check the kinetics of stage conversion in the TgISU1-depleted parasites. We kept the cKD TgISU1-HA parasites for up to 14 days in the presence of ATc and tested for the presence of SAG1 (tachyzoite maker), DBL (early bradyzoite marker), P18/SAG4 (intermediate bradyzoite marker) and P21 (late bradyzoite marker) (Fig 9D and S11B Fig). After 7 days of ATc treatment, the DBL-positive cyst contained parasites that were still expressing SAG1 and not yet P18/SAG4, whereas after 14 days parasites with P18/SAG4 labelling were found, but there was still a residual SAG1 expression; expression of late marker P21 was, however, never detected. As controls, the TATi ΔKu80 parental cell line (derived from the RH type I strain) and the Prugniaud cystogenic type II strain were subjected to alkaline stress-induced stage conversion [76] for a similar duration (S11A and S11B Fig). While a majority of DBL-positive cysts of the type II strain expressed the P18/SAG4 and P21 markers after 14 days of differentiation, this was not the case for the type I TATi ΔKu80 cell line. Cysts containing TgISU1-depleted parasites thus displayed a somewhat intermediate phenotype with the expression of P18/SAG4, but not P21. This suggests stage conversion of these parasites progresses beyond the appearance of early cyst wall markers, but it seems incomplete. In fact, DBL-positive cysts showed a marked decrease in their mean size between the 7 and 14 days timepoints upon TgISU1 depletion (Fig 9D and S11C Fig). Smaller cyst size seemed to be a feature of the parental TATi ΔKu80 cell line when compared with the type II strain (S11C Fig), and noticeably these type I parasites also kept growing largely as tachyzoites and were able to reinitiate invasion cycles, leading to considerable host cell lysis during the course of the pH stress-induced conversion. Decrease in cyst size over time may thus reflect incomplete conversion, and be a consequence of later reactivation/reinvasion events; in the case of the TgISU1 mutant, the markedly smaller cyst size after 14 days of protein depletion (Fig 9E and S11C Fig) may also suggest a lack of fitness in the long term.

Lack of lethality and initiation of stage conversion are features shared by other mitochondrial mutants

As impairment of the mitochondrial ETC and stage conversion are the two main features observed for the TgISU1 mutant, it raised the possibility the former may be involved in triggering the latter. We thus sought to evaluate in more details viability and differentiation of other mitochondrial mutants. We used the ATc-regulatable cKD cell line for TgQCR11 [63], a complex III subunit found less abundant in absence of TgISU1 (Fig 8A), and which was found by others to be essential for mitochondrial respiration and parasite growth [62,63]. We also included in our study an ATc-regulatable cKD cell line for TgmS35, a mitoribosomal protein whose depletion impacts organelle morphology and function (including mitochondrial respiration), and overall parasite fitness [77].

Like for TgISU1 (Fig 4E), we assessed whether or not the fitness phenotype of these mutants was reversible by performing plaque assays in the presence of ATc for 7 days, and then monitored plaque formation for another 7 days upon drug removal (Fig 10A). The TgQCR11 mutant behaved very similarly to the TgISU1 mutant, with virtually no plaques formed when incubated with ATc, while it was able to reactivate the lytic cycle upon ATc washout. The TgmS35 mutant seemed less affected by protein depletion: not only it generated small plaques after 7 days in the presence of ATc, but whether or not the drug was washed out, prolonged incubation led to large lysis plaques in the host cells. Overall, this experiment showed that mitochondrial mutants, whether they are affected in a metabolic or more structural function, remain essentially viable. The similarity between the TgISU1 and the TgQCR11 mutant suggested the latter may be switching to a slow-growing, but still viable, bradyzoite stage. The slowed-down, but less hampered progression through the lytic cycle of the TgmS35 mutant suggested it might retain more tachyzoite-like growth kinetics.

Download:

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

Fig 10. Other mitochondrial mutants remain viable and initiate stage conversion.

A) Plaque assays for the TgQCR11 and TgmS35 mutants were performed by infecting HFF monolayers with cKD cell lines and letting them grow in absence (7 days -) or presence of ATc (7 days +), after 7 days ATc was washed out (7 days+ 7days-) or not (14 days+), and parasites were left to grow for an extra 7 days. Plaque area was measured. Data are means ± SEM from three independent experiments. * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, **** p ≤ 0.0001, Student’s t-test. Arrowheads show small plaques. Scale bar = 1mm. B) Tati ΔKu80 parasites were grown in the presence of atovaquone and cKD TgQCR11 and TgmS35 parasites were grown in the presence of ATc for up to two days. They were labelled with anti-TgIMC3 (to outline parasites) and Dolicos biflorus lectin (DBL) to specifically outline nascent cyst walls. Scale bar represents 10 μm. DNA was labelled with DAPI. DIC: differential interference contrast. C) Quantification of DBL-positive vacuoles after 24 hours or 48 hours of culture of 1) the cKD TgQCR11 and cKD TgmS35 mutants in the presence or absence of ATc 2) the Tati ΔKu80 cell line in the presence or not of 1μM atovaquone. Data are from n = 3 independent experiments. Values are mean ±SEM. ** p ≤ 0.01, *** p ≤ 0.001, Student’s t-test. D) The cKD TgQCR11 and cKD TgmS35 mutants were grown for up to 14 days in the presence of ATc and co-stained with early cyst wall marker DBL together with tachyzoite marker SAG1, or early (P18/SAG4), or late (P21) bradyzoite markers. Scale bar represents 10 μm. DNA was labelled with DAPI.

https://doi.org/10.1371/journal.ppat.1010096.g010

To investigate the ability of these mutants to form cysts, we performed IFAs with specific bradyzoite markers, like we did previously for the cKd TgISU1-HA cell line (Fig 9A and 9B, S11B Fig). We first looked for the recruitment of DBL to parasite-containing vacuoles during the first 48 hours of protein depletion, (Fig 10B and 10C). As a control, we also treated TATi ΔKu80 parasites with atovaquone, an inhibitor of the mitochondrial ETC that was previously shown to trigger stage conversion [78]. Interestingly, like the atovaquone-treated TATi ΔKu80 parasites and the TgISU1 mutant (Fig 9A and 9B), treatment of both TgQCR11 and TgmS35 mutants with ATc seemed to trigger stage conversion, as illustrated by an increase in DBL staining of vacuoles over time (Fig 10B and 10C). When then tried to evaluate conversion over a longer period by keeping the parasites for up to 14 days in the presence of ATc (Fig 10D and S11B Fig). For the TgmS35 mutant, this was achieved by using very low doses of parasite inoculum, which allowed preserving some host cells in spite of the extensive lysis caused by the parasites. Like for the TgISU1 mutant, even long term differentiation-inducing conditions did not allow complete disappearance of the SAG1 tachyzoite marker in the TgQCR11 and TgmS35 mutants, and staining of the late bradyzoites marker P21 was not observed. However, while a few DBL-labelled cysts remained in the TgmS35 culture after 14 days of ATc treatment, they very rarely showed staining with the intermediate bradyzoite marker P18/SAG4 (S11B Fig). On the contrary, substantial staining was observed with this marker upon long term depletion of TgQCR11, like we observed for the TgISU1 mutant (S11B Fig). Of note, as observed for the TgISU1 mutant and parental TATi ΔKu80 cell line, these other mutants generated in a type I background strain also displayed smaller cyst size than the type II parasites after long term induction of stage conversion (S11C Fig), suggesting reactivation may happen and incomplete differentiation is likely a general feature of RH-derived parasites.

In conclusion, our findings suggest that interfering with general function like mitochondrial translation, or targeting more specifically the mitochondrial ETC, does not irreversibly impair parasite viability and instead leads to an initiation of stage conversion into bradyzoites, although it may not necessarily be complete.

Parasites and cells culture

Tachyzoites of the type I RH TATi RHΔKu80 T. gondii cell line [44], as well as derived transgenic parasites generated in this study and the type II Prugniaud strain, were maintained by serial passage in human foreskin fibroblast (HFF, American Type Culture Collection, CRL 1634) cell monolayer grown in Dulbecco’s modified Eagle medium (DMEM, Gibco), supplemented with 5% decomplemented fetal bovine serum (FBS), 2-mM L-glutamine and a cocktail of penicillin-streptomycin at 100 μg/ml. The TgQCR11 [63] and TgmS35 [77] conditional mutants cell lines were generously provided by the Sheiner laboratory.

Bioinformatic analyses

Sequence alignments were performed using the MUltiple Sequence Comparison by Log-Expectation (MUSCLE) algorithm of the Geneious 6.1.8 software suite (http://www.geneious.com). Transit peptide and localization predictions were done using IPSORT (http://ipsort.hgc.jp/), Localizer 1.0.4 (http://localizer.csiro.au/), and Deeploc 1.0 (http://www.cbs.dtu.dk/services/DeepLoc-1.0/) algorithms.

The putative Fe-S proteome was predicted using the MetalPredator webserver (http://metalweb.cerm.unifi.it/tools/metalpredator/) [56]. The whole complement of T. gondii annotated proteins was downloaded in FASTA format from the ToxoDB database (https://toxodb.org [37], release 45) and used for analysis in the MetalPredator webserver. Additional manual curation included homology searches for known Fe-S proteins from plants (see appendix A in [114]), and search for homologues in the Uniprot database (https://www.uniprot.org) that were annotated as containing a Fe-S cofactor. For proteomics candidates, annotations were inferred from ToxoDB, KEGG (https://www.genome.jp/kegg/) and the Liverpool Library of Apicomplexan Metabolic Pathways (http://www.llamp.net/ [115]).

N-glycosylation predictions were done with the GlycoEP webserver (http://crdd.osdd.net/raghava/glycoep/index.html). GPI anchor predictions were done with the PredGPI (http://gpcr.biocomp.unibo.it/predgpi/) and GPI-SOM (http://gpi.unibe.ch/) webservers.

Candidate variant proteins identified by quantitative proteomics were mapped on the graphic representation of the high-resolution spatial proteome map of T. gondii using the hyperLOPIT dataset [38] (https://proteome.shinyapps.io/toxolopittzex/).

Heterologous expression in E. coli

Constructs for designing recombinant proteins were defined by aligning TgNFS2 and TgISU1 amino acid sequences with their E. coli counterparts. For TgNFS2, a 1,438 bp fragment corresponding to amino acids 271–699, was amplified by polymerase chain reaction (PCR) from T. gondii cDNA using primers ML4201/ML4012 (sequences of the primers used in this study are found in S6 Table). For TgISU1, a 393 bp fragment corresponding to amino acids 64–194, was amplified by PCR from T. gondii cDNA using primers ML4204/ML4205. The fragments were cloned into the pUC19 (Thermo Fisher Scientific) using the HindIII/BamHI and SphI/BamHI restriction sites, respectively. E. coli mutants from the Keio collection (obtained from the The Coli Genetic Stock Center at the University of Yale: stain numbers JW1670-1 for SufS, JW2513-1 for IscU), were transformed with plasmids for expressing recombinant TgNFS2 and TgISU1 and selected with ampicillin. For growth assays [41], overnight stationary phase cultures were adjusted to the same starting OD₆₀₀ of 0.6 in salt-supplemented M9 minimal media containing 0.4% glucose and varying amounts of the 2,2′-Bipyridyl iron chelator (Sigma-Aldrich). Parental E. coli (strain K12) were included as a control. Growth was monitored through OD₆₀₀ measurement after 7, 14 and 24 hours at 37°C in a shaking incubator.

Generation of HA-tagged TgNFS2, TgSUFE2 and TgISU1 cell lines

The ligation independent strategy [43] was used for C-terminal hemagglutinin (HA)₃-tagging of TgISU1. A fragment corresponding to the 3’ end of TgISU1 was amplified by PCR from genomic DNA, with the Q5 DNA polymerase (New England BioLabs) using primers ML4208/ML4209 and inserted in frame with the sequence coding for a triple HA tag, present in the pLIC-HA₃-chloramphenicol acetyltransferase (CAT) plasmid. The resulting vector was linearized and 40 μg of DNA was transfected into the TATi ΔKu80 cell line to allow integration by single homologous recombination, and transgenic parasites of the TgISU1-HA cell line were selected with chloramphenicol and cloned by serial limiting dilution.

For TgNFS2 and TgSUFE2, a CRISPR-based strategy was used. Using the pLIC-HA₃-CAT plasmid as a template, a PCR was performed with the KOD DNA polymerase (Novagen) to amplify the tag and the resistance gene expression cassette with primers ML3978/ML3979 (TgNFS2) and ML4023/ML4162 (TgSUFE2), that also carry 30 bp homology with the 3′ end of the corresponding genes. A specific single-guide RNA (sgRNA) was generated to introduce a double-stranded break at the 3′ of the respective loci. Primers used to generate the guides were ML3948/ML3949 (TgNFS2) and ML4160/ML4161 (TgSUFE2) and the protospacer sequences were introduced in the Cas9-expressing pU6-Universal plasmid (Addgene, ref #52694) [39]. Again, the TATi ΔKu80 cell line was transfected and transgenic parasites of the TgNFS2-HA or TgSUFE2-HA cell lines were selected with chloramphenicol and cloned by serial limiting dilution.

Generation of TgNFS2 and TgISU1 conditional knock-down and complemented cell lines

The conditional knock-down cell lines for TgNFS2 and TgISU1 were generated based on the Tet-Off system using the DHFR-TetO7Sag4 plasmid [116].

For TgISU1, a 930 bp 5’ region of the gene, starting with the initiation codon, was amplified from genomic DNA by PCR using Q5 polymerase (New England Biolabs) with primers ML4212/ML4213 and cloned into the DHFR-TetO7Sag4 plasmid, downstream of the anhydrotetracycline (ATc)-inducible TetO7Sag4 promoter, yielding the DHFR-TetO7Sag4-TgISU1 plasmid. The plasmid was then linearized and transfected into the TgISU1-HA cell line. Transfected parasites were selected with pyrimethamine and cloned by serial limiting dilution. A similar approach was used to generate an independent cKD TgISU1 cell line, but starting from the TATi ΔKu80 cell line instead of the TgISU1-HA cell line, to allow subsequent tagging of mitochondrial candidates.

For TgNFS2, a CRISPR-based strategy was used. Using the DHFR-TetO7Sag4 plasmid as a template, a PCR was performed with the KOD DNA polymerase (Novagen) to amplify the promoter and the resistance gene expression cassette with primers ML4154/ML4155 that also carry 30 bp homology with the 5′ end of the TgNFS2 gene. A specific sgRNA was generated to introduce a double-stranded break at the 5′ of the TgNFS2 locus. Primers used to generate the guide were ML4156/ML4157 and the protospacer sequences were introduced in the pU6-Universal plasmid (Addgene ref#52694) [39]. The TgNFS2-HA cell line was transfected with the donor sequence and the Cas9/guide RNA-expressing plasmid and transgenic parasites were selected with pyrimethamine and cloned by serial limiting dilution.

The cKD TgNFS2-HA and cKD TgISU1-HA cell lines were complemented by the addition of an extra copy of the respective genes put under the dependence of a tubulin promoter at the uracil phosphoribosyltransferase (UPRT) locus. TgNFS2 (2097 bp) and TgISU1 (657 bp) whole cDNA sequences were amplified by reverse transcription (RT)-PCR with primers ML4576/ML4577 and ML4455/ML4456, respectively. They were then cloned downstream of the tubulin promoter sequence of the pUPRT-TUB-Ty vector [44] to yield the pUPRT-TgNFS2 and pUPRT-TgISU1plasmids, respectively. These plasmids were then linearized prior to transfection of the respective mutant cell lines. The recombination efficiency was increased by co-transfecting with the Cas9-expressing pU6-UPRT plasmids generated by integrating UPRT-specific protospacer sequences (with primers ML2087/ML2088 for the 3’, and primers ML3445/ML3446 for the 5’) which were designed to allow a double-strand break at the UPRT locus. Transgenic parasites were selected using 5-fluorodeoxyuridine and cloned by serial limiting dilution to yield the cKD TgNFS2-HA comp cKD TgISU1-HA comp cell lines, respectively.

Generation of HA-tagged TgSDHB and TgApiCox13 cell lines

A CRISPR-based strategy was used. Using the pLIC-HA₃-CAT plasmid as a template, a PCR was performed with the KOD DNA polymerase (Novagen) to amplify the tag and the resistance gene expression cassette with primers ML5116/ML5117 (TgSDHB) and ML5114/ML5115 (TgApiCox13), that also carry 30 bp homology with the 3′ end of the corresponding genes. A specific sgRNA was generated to introduce a double-stranded break at the 3′ of the respective loci. Primers used to generate the guides were ML4986/ML4987 (TgSDHB) and ML4984/ML4985 (TgApiCox13) and the protospacer sequences were introduced in the Cas9-expressing pU6-Universal plasmid (Addgene, ref #52694) [39]. The cKD TgISU1 cell line was transfected and parasites were selected with chloramphenicol and cloned by serial limiting dilution.

Immunoblot analysis

Protein extracts from 10⁷ freshly egressed tachyzoites were prepared in Laemmli sample buffer, separated by SDS-PAGE and transferred onto nitrocellulose membrane using the BioRad Mini-Transblot system according to the manufacturer’s instructions. Rat monoclonal antibody (clone 3F10, Roche) was used to detect HA-tagged proteins. Other primary antibodies used were rabbit anti-lipoic acid antibody (ab58724, Abcam), mouse anti-SAG1 [117], rabbit anti-CPN60 [118] and mouse anti-actin [119]. Protein quantification was performed by band densitometry using FIJI (https://imagej.net/software/fiji/).

Immunofluorescence microscopy

For immunofluorescence assays (IFA), intracellular tachyzoites grown on coverslips containing HFF monolayers, were either fixed for 20 min with 4% (w/v) paraformaldehyde (PFA) in PBS and permeabilized for 10 min with 0.3% Triton X-100 in PBS or fixed for 5 min in cold methanol (for the use of cyst-specific antibodies). Slides/coverslips were subsequently blocked with 0.1% (w/v) BSA in PBS. Primary antibodies used (at 1/1,000, unless specified) to detect subcellular structures were rabbit anti-CPN60 [118], mouse monoclonal anti-F1-ATPase beta subunit (gift of P. Bradley), mouse monoclonal anti-GRA3 [120], rabbit anti-TgHSP28 [121], rabbit anti-GAP45 [122], mouse monoclonal anti-SAG1 [117], anti P18/SAG4 (diluted 1/200, T8 3B1) and anti P21 (diluted 1/200, T8 4G10) [123]. Rat monoclonal anti-HA antibody (clone 3F10, Roche) was used to detect epitope-tagged proteins. Staining of DNA was performed on fixed cells by incubating them for 5 min in a 1 μg/ml 4,6-diamidino-2-phenylindole (DAPI) solution. All images were acquired at the Montpellier RIO imaging facility from a Zeiss AXIO Imager Z1 epifluorescence microscope driven by the ZEN software v2.3 (Zeiss). Z-stack acquisition and maximal intensity projection was performed to visualize larger structures such as in vitro cysts. Adjustments for brightness and contrast were applied uniformly on the entire image.

Plaque assay

Confluent monolayers of HFFs were infected with freshly egressed parasites, which were left to grow for 7 days in the absence or presence of ATc (added to a final concentration of 1 μg/ml). They were then fixed with 4% v/v PFA and plaques were revealed by staining with a 0.1% crystal violet solution (V5265, Sigma-Aldrich). For the washout experiments, after 7 days the culture medium was removed from the wells and one gentle wash was performed with Hanks’ Balanced Salt Solution (HBSS), taking care not to disturb the cells; then new DMEM medium containing or not ATc was added for another 7-day incubation, before termination of the experiment by PFA fixation and crystal violet staining.

Egress assay

T. gondii tachyzoites were grown for 40 (without ATc) or 120 (with ATc) hours on HFF cells with coverslips in 24-well plates. The infected host cells were incubated for 7 min at 37°C with DMEM containing 3 μM of calcium ionophore A23187 (C7522, Sigma-Aldrich) prior to fixation with 4% PFA. Immunofluorescence assays were performed as previously described [124]: the parasites and the parasitophorous vacuole membrane were labelled with anti-GAP45 and anti-GRA3, respectively. The proportion of egressed and non-egressed vacuoles was calculated by counting 250 vacuoles in three independent experiments. Data are presented as mean values ± SEM.

Alkaline stress-induced stage conversion

Tachyzoites were seeded onto HFF monolayers grown on coverslips in 24-well plates. The inoculum was adapted to the cell line used to avoid early lysis of the host cells, with as much as 20,000 parasites per well for the type II Prugniaud strain, and as little as 300 for the type I RH TATi ΔKu80 cell line. 24 hours after infection, in order to induce to induce alkaline stress as previously described [76], the regular culture medium was changed to a pH 8.2 differentiation medium: Minimum Essential Medium (MEM) without NaHCO₃ (Gibco), 3% FBS, 50 mM HEPES, pH 8.2. Cultures were maintained at 37°C in air (to avoid pH variation due to CO₂), and the medium was changed every 48 hours for the whole duration of the experiment.

Semi-quantitative RT-PCR

Total mRNAs of freshly egressed extracellular parasites from the cKD TgNFS2-HA, cKD TgISU1-HA and their respective complemented cell lines, as well as cKD TgISU1 (incubated with or without ATc at 1.5 μg/mL for 3 days) were extracted using Nucleospin RNA II Kit (Macherey-Nagel). The cDNAs were synthesized with 450 ng of total RNA per RT-PCR reaction using High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). Specific primers for TgNFS2 (ML4686/ML4687), TgISU1 (ML4684/ML4685) and, as a control, Tubulin β (ML841/ML842) or actin (ML843/ML844) were used to amplify specific transcripts with the GoTaq DNA polymerase (Promega). PCR was performed with 21 cycles of denaturation (30 s, 95°C), annealing (20 s, 55°C), and elongation (30 s, 72°C).

Mitochondrial membrane potential measurement

Parasites grown for the indicated time with or without ATc were mechanically released from their host cells, purified on a glass wool fiber column, washed and adjusted to 10⁷ parasites/ml in phenol red-free medium, and incubated in with 1.5 μM of the JC-1 dye (5,5’,6,6’-tetrachloro-1,1’,3,3’-tetraethylbenzimidazolylcarbocyanine Iodide, T3168, Invitrogen) for 30 min at 37°C, washed phenol red-free medium and analyzed by flow cytometry or microscopy. Flow cytometry analysis was performed on a FACSAria III flow cytometer (Becton Dickinson). An unstained control was used to define gates for analysis. 50,000 events per condition were collected and data were analysed using the FlowJo Software.

Quantitative label-free mass spectrometry

Parasites of the TATi ΔKu80 and cKD TgISU1-HA cell lines were grown for two days in the presence of ATc; parasites of the cKD TgNFS2-HA were grown for three days in the presence of ATc. Then they were mechanically released from their host cells, purified on a glass wool fiber column, washed in Hanks’ Balanced Salt Solution (Gibco). Samples were first normalized on parasite counts, but further adjustment was performed after parasite pellet resuspension in SDS lysis buffer (50 mm Tris-HCl pH8, 10 mm EDTA pH8, 1% SDS) and protein quantification with a bicinchoninic acid assay kit (Abcam). For each condition, 20 μg of total proteins were separated on a 12% SDS-PAGE run for 20 min at 100 V, stained with colloidal blue (Thermo Fisher Scientific), and each lane was cut in three identical fractions. Trypsin digestion and mass spectrometry analysis in the Q Exactive Plus mass spectrometer (Thermo Fisher Scientific) were carried out as described previously [125].

For peptide identification and quantification, the raw files were analyzed with MaxQuant version 1.6.10.43 using default settings. The minimal peptide length was set to 6. Carbamidomethylation of cysteine was selected as a fixed modification and oxidation of methionine, N-terminal-pyroglutamylation of glutamine and glutamate and acetylation (protein N terminus) as variable modifications. Up to two missed cleavages were allowed. The files were searched against the T. gondii proteome (March 2020 -https://www.uniprot.org/proteomes/UP000005641-8450entries). Identified proteins were filtered according to the following criteria: at least two different trypsin peptides with at least one unique peptide, an E value below 0.01 and a protein E value smaller than 0.01 were required. Using the above criteria, the rate of false peptide sequence assignment and false protein identification were lower than 1%. Peptide ion intensity values derived from MaxQuant were subjected for label-free quantitation. Unique and razor peptides were considered [126]. Statistical analyses were carried out using R package software. ANOVA test with threshold of 0.05 was applied to identify the significant differences in the protein abundance. Hits were retained if they were quantified in at least three of the four replicates in at least one experiment. Additional candidates that consistently showed absence or presence of LFQ values versus the control, and mean LFQ was only considered if peptides were detected in at least three out of the four biological replicates.

Statistical analysis for phenotypic assays

Unless specified, values are usually expressed as means ± standard error of the mean (SEM). Data were analysed for comparison using unpaired Student’s t-test with equal variance (homoscedastic) for different samples or paired Student’s t-test for similar samples before and after treatment. For comparisons with of ratios of groups of samples with a reference set to 1, analysis of variance (ANOVA) was used.