Plasmodium falciparum stomatin-like protein forms a putative complex with a metalloprotease in distinct mitochondrial loci

TL;DR

This study shows that the STOML protein in malaria parasites is important for asexual growth and localizes with a protease in the mitochondrion.

Contribution

The study identifies a novel STOML-FtsH complex in P. falciparum mitochondria and its role in asexual blood-stage development.

Findings

Deletion of STOML in P. falciparum causes slower asexual blood-stage development but does not affect sexual-stage development.

STOML localizes to mitochondrial branching points and forms a complex with the metalloprotease FtsH.

Structural modeling suggests STOML may form a scaffold similar to bacterial HflK/C proteins.

Abstract

Members of the Stomatin, Prohibitin, Flotillin and HflK/C (SPFH) protein family form large membrane anchored or spanning complexes and are involved in various functions in different organelles. The human malaria causing parasite Plasmodium falciparum harbors four SPFH proteins, including prohibitin 1 and 2, prohibitin-like protein (PHBL), and stomatin-like protein (STOML), which all localize to the parasite mitochondrion. In the murine model parasite Plasmodium berghei, STOML appears essential for asexual blood-stage (ABS) development and is localized to puncta on mitochondrial branching points in oocyst stages. In this study, we show that deletion of P. falciparum
STOML causes a significant growth defect and slower ABS development, while sexual-stage development remains unaffected. Parasites lacking STOML were not more sensitive to respiratory chain targeting drugs, rendering a function of STOML in respiratory chain assembly unlikely. Epitope tagging of endogenous STOML revealed a distinct punctate localization on branching points and endings of the ABS mitochondrial network. STOML resides in a large protein complex and pulldown experiments identified a zinc dependent metalloprotease, FtsH, as a likely interaction partner. The predicted AlphaFold2 structure of STOML shows high similarity with the bacterial HflK/C, which has been shown to form a large vault-like structure around bacterial FtsH hexamers. Combined, our results suggest that a similar STOML-FtsH complex localized to specific loci of P. falciparum mitochondria facilitates the parasite’s ABS development.

Asexual blood-stage malaria parasites depend on a single mitochondrion that is essential for their growth and survival. This organelle differs from the mitochondria in human cells, making it an attractive drug target. Proteins of the SPFH family are conserved membrane scaffolds that can organize into large protein complexes in the mitochondrion but their roles in malaria parasites remain poorly understood. In this study, we investigate the function of the stomatin-like protein (STOML) in the human malaria parasite Plasmodium falciparum. We show that parasites lacking this protein grow significantly slower in the disease causing asexual blood stage, while sexual-stage development required for transmission to the mosquito remains unaffected. Using fluorescent microscopy, we show that STOML localizes to specific sites on the mitochondrial branching points and branch endings. Biochemical analysis shows that STOML is part of a large protein complex that includes the mitochondrial metalloprotease FtsH. Structural modeling suggests that STOML may assemble into a barrel-like scaffold similar to bacterial complexes that regulate protease activity. Together, our findings shed more light on STOML’s role in parasite mitochondrial function and highlight this protein family for future research.

Figures

ABS development of stoml-mito parasites is delayed.A) Schematic overview of the competition growth assay. PfSTOML knockout or control parasites expressing cytosolic GFP (green) are mixed with wild-type (WT) parasites expressing cytosolic mScarlet (red) in an approximate 1:1 ratio. Overtime, the distribution of red/green parasites is measured using flow cytometry. If PfSTOML knockout causes a growth defect, the WT population will grow faster and the ratio red/green parasites will shift. B) Bar graph showing distribution of red/green parasites in the competition growth assay at day 0, day 7, and day 14. The graph shows one representative example of three independent experiments. Cyto-mScarlet parasites are mixed with green cyto-NG (control), stoml-, or stoml-mito parasites to create mixed cultures. C) Giemsa-stained thin blood smears and fluorescent images showing the mitochondrial mScarlet marker (magenta) and DNA (blue) in MitoRed (WT) and stoml-mito parasites over time. Fluorescence microcopy images are maximum intensity projections of Z-stack confocal Airyscan images. D-E) Growth curve of parasitemia of MitoRed and stoml-mito over time in hours (D) and days (E). To visualize continuous growth across culture dilutions, parasitemia values were corrected for dilution factors, resulting in cumulative (corrected) parasitemias exceeding 100%. The graphs show one representative experiments of two independent experiments.

Localization of STOML-3HA-NG in ABS parasites.A) Live imaging of stoml-NG with PfSTOML-3HA-NG (green), mito-mScarlet mitochondrial marker (magenta), Hoechst for DNA visualization (blue), and DIC through the asexual replication cycle. Images are maximum intensity projections of Z-stack confocal Airyscan images. Arrowheads indicate PfSTOML-3HA-NG signal at mitochondrial branching points (white) or mitochondrial branch endings (yellow). Scale bars, 2 µm. B) 3D visualization of Z-stack confocal Airyscan images using Arivis 4D vision software. Fluorescent signal is segmented by manual thresholding. Arrowheads indicate STOML-NG signal at mitochondrial branching points (white) or mitochondrial branch endings (yellow). Scale bars, 1 µm. C) Boxplot indicating number of PfSTOML-3HA-NG foci per parasite in ring (n = 7), trophozoite (n = 12) and schizont (n = 9) stages with a total of 28 parasites.

Identification and characterization of STOML protein complex.A) Heatmaps and line graphs based on previously published complexome profiling data [39] showing migration patterns and relative protein abundance of STOML (red), FtsH (blue), and PF3D7_1306200 (green) in asexual blood stages (ABS) and gametocytes (GTC). Co-migration on blue native gel within the same molecular weight (MW) range (x-axis) indicates complex formation. Their relative abundance was normalized with the highest iBAQ value for a given protein set to 1 (shown red in the heatmap). B) Anti-mNG co-immunoprecipitation of PfSTOML-3HA-NG containing complexes. The volcano plot showing mean log2 fold changes (FC) and -log10 false discovery rate (FDR) for anti-HA pulldown in comparison with control pulldown. Horizontal and vertical dotted lines indicate log2 FC > 3 (FC > 8) and -log10 FDR > 1.301 (FDR > 0.05) respectively. Dark dots represent proteins that are highly enriched or reduced in the anti-mNG pulldown compared to the control pulldown. C) Overlap of enriched proteins in both anti-HA pulldown on STOML-HA and anti-NG pulldown on STOML-NG.

AlphaFold2 predictions of PfSTOML complex structure and comparison with the bacterial HflK/C supercomplex structures.A) Protein domains of EcHflK EcHflC, and PfSTOML, including the N-terminal domain (NT), transmembrane domain (TM), SPFH1 and SPHF2 domains, long coiled-coil domain 1 (CC1), coiled-coil domain 2 (CC2), and C-terminal region (CT). B) Protein structures of EcHflK and EcHflC determined by Ma et al. [21]. C) Predicted AlphaFold2 structure of PfSTOML. D) Overlay of EcHflK and EcHflC structures with PfSTOML AlphaFold2 structure. E) Cryo-EM structure of HflK/C – FtsH complex with the side view (top) and bottom view (bottom). Purple/blue-colored proteins are HflC, green/yellow-colored proteins are HflK, red/pink-colored proteins are part of the FtsH hexamers [21]. PDB ID: 7VHP. F) Predicted AlphaFold Multimer structure of PfSTOML 22-multimer with side view (top) and top view (bottom), coloring according to model confidence with very high confidence (pLDDT > 90) dark blue, high confidence (pLDDT > 70) light blue, and low confidence (pLDDT > 50) yellow. G) Overlay of predicted AlphaFold Multimer PfSTOML 22-multimer structure (green) with HflK/C – FtsH complex (light blue) with side view (top) and top view (bottom).

Full text

Introduction

Malaria is an infectious disease caused by Plasmodium parasites, which takes more than 600,000 mostly young lives annually [1]. Plasmodium falciparum is the most virulent malaria causing species. Resistance to current antimalarial drugs is spreading fast, emphasizing the need for the continuous development of novel antimalarial compounds. Plasmodium parasites harbor a unique mitochondrion that differs considerably from human mitochondria, which makes it a suitable drug target of antimalarial compounds such as atovaquone, DSM265, ELQ300, and proguanil [2,3].

The mitochondrion consists of an inner and outer membrane, which are both rich in large protein complexes. Indeed, the inner mitochondrial membrane (IMM) is considered one of the most protein-rich membranes in any cell-type and contains large multiprotein complexes, such as respiratory chain complexes, ATP synthase, and the mitochondrial contact site and cristae organizing system (MICOS). Similarly to all other biological membranes, the mitochondrial membranes are organized into domains of distinct protein and lipid composition [4,5]. These membrane microdomains are important for the spatial and temporal control of membrane protein complex assembly and regulation [4]. SPFH (Stomatin, Prohibitin, Flotillin and HflK/C) family proteins are enriched in eukaryotic and prokaryotic membrane microdomains of various organelles, such as plasma membrane, nucleus, endoplasmic reticulum (ER), and mitochondria [6]. The common feature of SPFH proteins is the presence of the highly conserved SPFH or Band-7 protein domain [7]. These proteins form large self-oligomerizing membrane-spanning or membrane-anchored complexes and have been indicated in a diverse set of functions [6]. In human mitochondria, a subset of SPFH proteins, including two prohibitins (PHB1 and PHB2) and stomatin-like protein 2 (SLP2), localizes to the IMM. PHB1 and PHB2 form a large protein complex together, which has been indicated to play a role in mitochondrial protein degradation, cristae formation, mitochondrial dynamics, cell cycle regulation, and apoptosis [8–11]. SLP2 localizes to cardiolipin enriched membrane microdomains, where it interacts with and controls stability of the PHB complex [12,13]. The PHB and SLP2 complexes are both important for the formation and stability of the respiratory chain complex and mitochondrial translation [8,13–17]. They reside in large supercomplexes with metalloproteases and assert their proteolytic function through regulation of metalloprotease activity, similar to their bacterial family member HflIK/C [8,11,12,18–21].

Plasmodium parasites harbor three conserved SPFH proteins: PHB1, PHB2, and stomatin-like protein (STOML), as well as an unusual prohibitin-like protein (PHBL). PHBL is specific to the unicellular Myzozoa, a clade that includes apicomplexan parasites and free-living dinoflagellates [22]. Attempts to delete the four genes using both genome-wide screens in P. falciparum and the murine malaria model parasite Plasmodium berghei, and targeted approaches in the latter, resulted in conflicting results [22–25]. Localization studies through fluorescent tagging of the endogenous P. berghei genes revealed a mitochondrial localization of three SPFH proteins throughout the P. berghei life cycle [22]. Although tagging of PHB1 was unsuccessful, PHB1/2 heterodimerization is evolutionary well-conserved [15,26] and PfPHB1 ranks 84^th^ on the validated list of predicted Plasmodium mitochondrial proteins [27]. Functional complementation of yeast PHB mutants provided further support for prohibitin heterodimerization in P. falciparum [28]. PfPHBs were shown to be involved in stabilizing mitochondrial DNA, maintaining mitochondrial integrity, and rescuing yeast cell growth [28]. PHBL-deficient parasites failed to colonize Anopheles mosquitos as they arrest during ookinete development, which is correlated with depolarization of the mitochondrial membrane potential [22].

The role and importance of STOML remains unclear. Genetic screens in P. falciparum and P. berghei both suggested a dispensable role, yet targeted approaches did never yield a pure isogenic or clonal line free of wild-type (WT) parasites, indicating possible developmental issues [22–24]. Interestingly, PbSTOML localizes to punctate foci at the parasite mitochondrion during oocyst growth, often at organellar branching points [22]. This specific mitochondrial localization combined with the uncertainty about its importance and function drove us to further investigate the role of STOML in the human malaria causing P. falciparum.

In this study, we show that deletion of STOML in P. falciparum causes a significant growth delay of asexual blood stages (ABS), while sexual-stage development is not affected. PfSTOML localizes to punctate foci at mitochondrial branch endings and at branching points throughout ABS development. PfSTOML resides in a large protein complex and pulldown experiments identified the metalloprotease FtsH as a likely interaction partner. We also show that the predicted AlphaFold Multimer structure of PfSTOML is highly similar to its bacterial family member HflK/C, which has recently been shown to form a large, oligomerized, vault structure around FtsH hexameres, this way regulating their accessibility. This suggests that a similar scenario might apply to the STOML-FtsH complex in P. falciparum. These results provide novel insights into the function of STOML in P. falciparum and pave the way for future studies into the function of SPFH proteins and their potential as antimalarial drug targets.

Results

Knockout of PfSTOML results in a significant growth defect

To study the function of PfSTOML (PF3D7_0318100) during ABS development, we aimed to generate PfSTOML knockout (KO) parasites using a targeted replacement strategy (S1A Fig). Although the first three transfection attempts were unsuccessful, we managed to generate two PfSTOML KO parasite lines in NF54 (stoml^-^) and the MitoRed background (stoml^-^mito), the latter harboring a fluorescent mitochondrial marker (mito-mScarlet) [29]. Correct integration and the absence of unaltered wild-type (WT) NF54 or MitoRed parasite contaminations were verified by diagnostic PCR (S1B Fig). To demonstrate if PfSTOML KO causes a growth defect, we set up a new competition growth assay analogous to the protocol used with Plasmodium berghei [30]. In both PfSTOML KO lines, STOML is replaced by GFP under the constitutive P. falciparum histone 2B (PfH2B, PF3D7_1105100) promotor, making them green fluorescent. By mixing these with WT parasites harboring a constitutively expressed mScarlet, the relative abundance of red-only and green fluorescent parasites can be determined by flow cytometry and followed over time (Fig 1A). The average factor by which the red/green ratio changed from the first to the second timepoint in three independent experiments was defined as fr. We included a control condition in which mNeonGreen expressing WT parasites (cyto-NG) are mixed with mScarlet expressing WT parasites (cyto-mScarlet) (S2 Fig). We confirmed that the ratio of red and green parasites in this control culture was stable over time (fr = 1.0) (Fig 1B). However, when co-culturing either of our PfSTOML KO lines with cyto-mScarlet, we found that the red/green distributions shift significantly over time (f_r_ = 20.9, p < 0.0001). In one representative experiment, the ratio of cyto-mScarlet versus stoml^-^mito parasites changes from 47:53–94:6 after one week and 99:1 after two weeks, indicating that stoml^-^mito grows approximately 12 times slower in a week period compared to WT in mixed culture conditions. We observed a very similar trend in stoml^-^ mixed cultures. To test whether this growth defect is caused by lower number of viable offspring per parasite, reduced invasion, or delayed development throughout the ABS cycle, we quantified growth and analyzed stage development microscopically in two independent experiments, either every 8–16 h over an 88-h period or daily for eight days (Fig 1C-E). These experiments showed that stoml^-^mito develops slower throughout the ABS replication cycle compared to MitoRed WT parasites. At the end of the first replication cycle (40 h), MitoRed WT cultures contained mostly segmented schizonts and rings, and the parasitemia has increased 6.0-fold compared to the 24 h timepoint (Fig 1D). stoml^-^mito cultures contained mostly early schizonts and very few rings at 40 h, and the parasitemia has only increased 1.6-fold. However, at 48 h, stoml^-^mito parasitemia has almost “caught up” with WT parasitemia with a 4.9-fold increase compared to the 24 h timepoint. This trend of delayed ABS development continues in the next replication cycles (Fig 1E). These results indicate that stoml^-^mito has a growth defect that is mainly caused by slower and prolonged development throughout the asexual replication cycle.

ABS development of stoml-mito parasites is delayed.A) Schematic overview of the competition growth assay. PfSTOML knockout or control parasites expressing cytosolic GFP (green) are mixed with wild-type (WT) parasites expressing cytosolic mScarlet (red) in an approximate 1:1 ratio. Overtime, the distribution of red/green parasites is measured using flow cytometry. If PfSTOML knockout causes a growth defect, the WT population will grow faster and the ratio red/green parasites will shift. B) Bar graph showing distribution of red/green parasites in the competition growth assay at day 0, day 7, and day 14. The graph shows one representative example of three independent experiments. Cyto-mScarlet parasites are mixed with green cyto-NG (control), stoml-, or stoml-mito parasites to create mixed cultures. C) Giemsa-stained thin blood smears and fluorescent images showing the mitochondrial mScarlet marker (magenta) and DNA (blue) in MitoRed (WT) and stoml-mito parasites over time. Fluorescence microcopy images are maximum intensity projections of Z-stack confocal Airyscan images. D-E) Growth curve of parasitemia of MitoRed and stoml-mito over time in hours (D) and days (E). To visualize continuous growth across culture dilutions, parasitemia values were corrected for dilution factors, resulting in cumulative (corrected) parasitemias exceeding 100%. The graphs show one representative experiments of two independent experiments.

PfSTOML is unlikely to be involved in assembly of the respiratory chain

In other eukaryotes, stomatin-like proteins are thought to be involved in a variety of mitochondrial functions [13,20,31–33]. To explore if PfSTOML has a similar mitochondrial function in P. falciparum, we were curious to see if PfSTOML KO would alter mitochondrial dynamics. We compared mitochondrial morphology of stoml^-^mito with MitoRed parasites and found no obvious differences throughout different stages of ABS development in two independent experiments (S3 Fig). Mature stoml^-^mito schizonts showed divided and segregated mitochondria, similarly to MitoRed WT parasites.

SLP2, the human STOML homolog, has been indicated to play an essential role in the assembly of the respiratory chain [12,31]. To test if STOML has a similar function in P. falciparum, we investigated if stoml^-^ parasites would have an increased sensitivity to drugs targeting the respiratory chain as demonstrated successfully in the past for other mitochondrial proteins [34–36]. We performed drug assays with different mitochondrial drugs (including DSM1, DSM265, atovaquone, ELQ300, and proguanil) and non-mitochondrial drugs (chloroquine, DHA, and MMV183) and found no difference in drug sensitivity between stoml^-^ and WT parasites in two independent experiments (S4 Fig). Energy metabolism in P. falciparum ABS relies heavily on glycolysis and oxidative phosphorylation (OXPHOS) is only essential for ubiquinone recycling for pyrimidine synthesis [37]. However, in gametocytes, there is an increased TCA cycle utilization and presumably respiration [38,39]. To our surprise, stoml^-^mito parasites develop to healthy-looking, mature gametocytes within a comparable time frame to WT parasites in three independent experiments. Furthermore, we found no obvious aberrations in mitochondrial morphology of mature male and female stoml^-^mito gametocytes (S5A Fig). stoml^-^mito parasites were still able to exflagellate and had dispersed mitochondria after activation, similarly to what we described for the MitoRed WT line [29] (S5B Fig). Based on these results, it is unlikely that STOML is directly involved in maintaining mitochondrial morphology or respiratory chain assembly in P. falciparum.

PfSTOML localizes to specific foci at the mitochondrion during ABS development

To learn more about the function of PfSTOML, we analyzed its subcellular localization. To do this, we generated a transgenic parasite line, stoml-NG, in which STOML is fused with a 3HA-mNG-GlmS tag. We also integrated a mitochondrial marker cassette mito-mScarlet for protein co-localization (S1A Fig). Correct integration and the absence of WT parasite contaminations were verified by diagnostic PCR (S1B Fig). Western blot analysis of ABS parasite extract confirmed expression of the full length PfSTOML-3HA-NG (S1C Fig). Live fluorescent microscopy of stoml-NG in three independent experiments showed localization of PfSTOML-3HA-NG to punctate foci during ABS development (Fig 2). In ring stages, PfSTOML-3HA-NG localized to a single spot, close to the mitochondrion (Fig 2A). As the parasites progress to late rings and the mitochondrion elongates, PfSTOML-3HA-NG is consistently found in two foci that reside at both endings of the mitochondrion (Fig 2A and 2B and S1 Movie). In trophozoites, the mitochondrion starts to form a branched structure and the number of PfSTOML-3HA-NG foci per parasite increases (Fig 2C). PfSTOML-3HA-NG foci are found both at endings of mitochondrial branches, as well as branching points (Fig 2A). A similar localization is found in early schizont stages, where the mitochondrion forms a complex, branched network (Fig 2A and 2B and S2 Movie). The number of foci per parasite increases during schizont maturation. In late schizonts, when the mitochondrial branches orient in a radial fashion prior to division, the PfSTOML-3HA-NG foci are found at the endings of most mitochondrial branches, although PfSTOML-3HA-NG signal can also be observed along mitochondrial branches (Fig 2A and 2B and S3 Movie). In a fully segmented parasite that seems to have egressed form the RBC, only few PfSTOML-3HA-NG foci were found. The PfSTOML-3HA-NG foci are largely but not completely overlapping with the mito-mScarlet signal in trophozoite and schizont stages (Fig 2B), suggesting localization to the mitochondrial membrane, although we lack the resolution to distinguish between IMM and OMM. This unique localization pattern is different from the homogeneous mitochondrial localization of PbSTOML observed in P. berghei ABS [22].

Localization of STOML-3HA-NG in ABS parasites.A) Live imaging of stoml-NG with PfSTOML-3HA-NG (green), mito-mScarlet mitochondrial marker (magenta), Hoechst for DNA visualization (blue), and DIC through the asexual replication cycle. Images are maximum intensity projections of Z-stack confocal Airyscan images. Arrowheads indicate PfSTOML-3HA-NG signal at mitochondrial branching points (white) or mitochondrial branch endings (yellow). Scale bars, 2 µm. B) 3D visualization of Z-stack confocal Airyscan images using Arivis 4D vision software. Fluorescent signal is segmented by manual thresholding. Arrowheads indicate STOML-NG signal at mitochondrial branching points (white) or mitochondrial branch endings (yellow). Scale bars, 1 µm. C) Boxplot indicating number of PfSTOML-3HA-NG foci per parasite in ring (n = 7), trophozoite (n = 12) and schizont (n = 9) stages with a total of 28 parasites.

PfSTOML resides in a large protein complex with PfFtsH

PHBs and STOML are found in large hetero- and homo-oligomers at the IMM in other eukaryotes, such as humans and yeast [15,20]. P. falciparum mitochondrial complexome profiling data showed that PfSTOML migrates across a broad size range of ~1.5-3.5 MDa on a native gel indicating that it forms part of one or potentially multiple large protein complexes (Fig 3A) [39]. In order to identify the proteins in complex with PfSTOML, we performed two independent pulldown experiments (Figs 3A, 3B and S7). In the first experiment, late-ABS stoml-NG parasites (24–40 h.p.i.) were lysed through saponin lysis and nitrogen cavitation, and the organelle fraction was used as input for co-immunoprecipitation with anti-mNG coated magnetic beads. As a control, the same fraction was loaded on uncoated beads. Mass spectrometry revealed that 27 proteins were significantly enriched after pulldown with mNG beads, of which PfSTOML was the most significantly enriched (Fig 3B and S2 Table). For the second pulldown experiment, we generated a transgenic parasite line in which STOML is fused with an 3HA-GlmS tag, which we termed stoml-HA. Correct integration and the absence of WT parasite contaminations were verified by diagnostic PCR (S1B Fig) and western blot analysis confirmed expression of PfSTOML-3HA (S1C Fig). Fluorescence microscopy confirmed that PfSTOML-3HA localizes to mitochondrial branching points and branch endings, consistent with the pattern observed in live imaging of PfSTOML-3HA–mNG (S1C-S2C Fig). This indicated that the addition of the large mNeonGreen tag does not affect the native mitochondrial localization of PfSTOML. The organelle fraction of late-ABS stoml-HA parasites was used as input for pulldown with anti-HA coated magnetic beads and empty protein G beads were used as control. In the second experiment, 122 proteins were significantly enriched after HA pulldown (S7 Fig and S3 Table). Three proteins were significant hits in both pulldown experiments: STOML, an ATP-dependent zinc metalloprotease FtsH (PF3D7_1464900), and a conserved protein of unknown function (PF3D7_1306200) (Fig 3C). In our published complexome profiling experiments, PF3D7_1306200 did not comigrate with STOML in either ABS or gametocyte stages (Figs 3A and S6) [39]. PF3D7_1306200 is predicted to be an essential protein and is expressed in late schizonts [23,29,40]. It contains an AB-hydrolase domain and is thought to localize to the apicoplast [41]. FtsH, on the other hand, is a predicted mitochondrial protein (ranking 265^th^) [27] and phylogenetic analysis shows clustering with i-AAA proteases in the IMM [42]. Both in ABS and gametocytes, FtsH migrates corresponding to a mass of ~2.5 MDa, which in gametocytes corresponds with the most dominant STOML migration peak (Figs 3A and S6) [39]. In ABS, these normalized migration profiles show weaker apparent comigration, as FtsH overlaps with a minor peak of STOML. However, when considering absolute protein abundance (iBAQ values), the FtsH peak comigrates with a similar quantity of STOML, compatible with a putative complex (S6 Fig). As STOML is substantially more abundant than FtsH (≈173-fold in gametocytes and ≈130-fold in ABS), this can obscure comigration in normalized profiles. The broad STOML migration pattern may therefore reflect multiple (sub)assemblies, with only a small subset of STOML associating with FtsH.

Identification and characterization of STOML protein complex.A) Heatmaps and line graphs based on previously published complexome profiling data [39] showing migration patterns and relative protein abundance of STOML (red), FtsH (blue), and PF3D7_1306200 (green) in asexual blood stages (ABS) and gametocytes (GTC). Co-migration on blue native gel within the same molecular weight (MW) range (x-axis) indicates complex formation. Their relative abundance was normalized with the highest iBAQ value for a given protein set to 1 (shown red in the heatmap). B) Anti-mNG co-immunoprecipitation of PfSTOML-3HA-NG containing complexes. The volcano plot showing mean log2 fold changes (FC) and -log10 false discovery rate (FDR) for anti-HA pulldown in comparison with control pulldown. Horizontal and vertical dotted lines indicate log2 FC > 3 (FC > 8) and -log10 FDR > 1.301 (FDR > 0.05) respectively. Dark dots represent proteins that are highly enriched or reduced in the anti-mNG pulldown compared to the control pulldown. C) Overlap of enriched proteins in both anti-HA pulldown on STOML-HA and anti-NG pulldown on STOML-NG.

SPFH proteins are known to form large protein complexes with metalloproteases, regulating their protease activity [8,18,20,21,43]. The human STOML homolog, SLP2, forms a large proteolytic complex termed the SPY complex at the inner mitochondrial membrane with rhomboid protease PARL and i-AAA metalloprotease YME1L [20]. SLP2 regulates the activity of YME1L, which forms homo-hexamers and is involved in degradation of unfolded or excess mitochondrial proteins [44]. In Trypanosoma brucei, another unicellular protozoan parasite, SLP2 can also be found in a complex with the Yme1L homolog, TbYme1 [45]. A BLAST search of hYME1L identified P. falciparum FtsH as top hit with 41% identity.

Cryo-electron microscopy revealed that the bacterial HflK and HflC form a large, hetero-oligomeric vault structure around four membrane-anchored FtsH hexamers (Fig 4E) [21]. We compared the predicted AlphaFold2 [46] structure of PfSTOML with the bacterial HflK/C complex (Fig 4A-D). Although PfSTOML is not predicted to contain a transmembrane domain [47] the overall predicted structure of the protein is highly similar to its bacterial family members. To further explore if PfSTOML could form a similar multimer barrel structure, we used AlphaFold Multimer [48] to predict the PfSTOML 24-multimer structure (S8A Fig). We used the SPFH2 and long alpha helix domains of PfSTOML, as these are the best predicted parts of the PfSTOML structure (pLDDT>90). The PfSTOML 24-multimer structure was predicted to form a distorted circular barrel structure with the endings of the multimer structure not joining together to close the structure, which visually seemed to include too many PfSTOML proteins (S8A Fig). In order to roughly estimate the correct amount of PfSTOML proteins in the barrel complex, we used AlphaFold Multimer to predict the structure of PfSTOML 8-multimer structure with the SPFH1, SPFH2, and long alpha helix domains (S8B Fig). We then measured the angle between the SPFH domains in the 8-multimer complex to estimate the curvature of the barrel (S8C Fig). We found an angle of 16.5 degrees between SPFH domains, suggesting that the PfSTOML complex might consist of approximately 22 PfSTOML monomers. AlphaFold Multimer predicts an intact, slightly oval barrel structure for a PfSTOML 22-multimer with SPFH2 and long alpha helix domains (Figs 4F and S8D). Considering the co-immunoprecipitation evidence and the high similarity between the predicted PfSTOML multimer structure and the HflK/C-FtsH complex, we hypothesize that STOML forms a similar supercomplex with FtsH likely in the IMM in P. falciparum.

AlphaFold2 predictions of PfSTOML complex structure and comparison with the bacterial HflK/C supercomplex structures.A) Protein domains of EcHflK EcHflC, and PfSTOML, including the N-terminal domain (NT), transmembrane domain (TM), SPFH1 and SPHF2 domains, long coiled-coil domain 1 (CC1), coiled-coil domain 2 (CC2), and C-terminal region (CT). B) Protein structures of EcHflK and EcHflC determined by Ma et al. [21]. C) Predicted AlphaFold2 structure of PfSTOML. D) Overlay of EcHflK and EcHflC structures with PfSTOML AlphaFold2 structure. E) Cryo-EM structure of HflK/C – FtsH complex with the side view (top) and bottom view (bottom). Purple/blue-colored proteins are HflC, green/yellow-colored proteins are HflK, red/pink-colored proteins are part of the FtsH hexamers [21]. PDB ID: 7VHP. F) Predicted AlphaFold Multimer structure of PfSTOML 22-multimer with side view (top) and top view (bottom), coloring according to model confidence with very high confidence (pLDDT > 90) dark blue, high confidence (pLDDT > 70) light blue, and low confidence (pLDDT > 50) yellow. G) Overlay of predicted AlphaFold Multimer PfSTOML 22-multimer structure (green) with HflK/C – FtsH complex (light blue) with side view (top) and top view (bottom).

Discussion

The SPFH protein family is a highly conserved family involved in the formation of microdomains in membranes of various organelles. Prohibitins and stomatin-like proteins (STOML) are localized to the IMM and have been implicated in several mitochondrial functions, including cristae formation and assembly of the respiratory chain [11,12,14,31]. Plasmodium harbors four SPFH family members, including two prohibitins (PHB1 and PHB2), a prohibitin-like protein (PHBL), and STOML, which all localize to the mitochondrion [22]. STOML is likely essential in P. berghei and localizes to foci on mitochondrial branching points during oocyst stages. In this study, we investigated the function of STOML in the human malaria causing parasite P. falciparum.

In contrast to the in vivo murine model parasite P. berghei in which STOML could not be deleted, we were able to generate two STOML KO lines in P. falciparum. stoml^-^ parasites developed slower throughout the ABS replication cycle compared to WT parasites, indicating an important but non-essential function for PfSTOML under in vitro culture conditions. In trophozoite and schizont stages, PfSTOML localizes to specific foci at the mitochondrial branching points and endings of mitochondrial branches. In late schizonts, when mitochondrial branches are oriented in a radial fashion prior to division [29], PfSTOML has a punctate localization at the endings of mitochondrial branches. While STOML appears to have a more uniform mitochondrial distribution in most P. berghei life-cycle stages, PbSTOML localizes to foci on mitochondrial branching points in oocyst stages [22]. This specific localization suggests a potential function for PfSTOML in mitochondrial dynamics and/or segregation. On the other hand, PfSTOML knockout does not seem to affect mitochondrial division, and segregation (S3 Fig). Therefore, the exact function of PfSTOML at this specific mitochondrial localization remains to be elucidated.

In human, yeast and plants, STOML has been indicated to play a role in the assembly of the respiratory chain [12,13,31]. We hypothesized that STOML might have a similar function in P. falciparum and that PfSTOML knockout would lead to an increased sensitivity to drugs targeting the respiratory chain, such as atovaquone and ELQ300. Such an approach has been used successfully in several studies of mitochondrial proteins in P. falciparum highlighting potential roles in respiratory chain assembly or functioning of PfRieske, PfPNPLA2, and PfOPA3 [34–36]. However, our data show no changes in drug sensitivity upon PfSTOML knockout. Additionally, stoml^-^mito parasites were able to form healthy gametocytes and undergo male gametogenesis, during which mitochondrial ATP synthesis is thought to be essential [49]. This indicates that in contrast with orthologs in other eukaryotes, PfSTOML appears not directly involved in respiratory chain assembly in P. falciparum.

Our mitochondrial complexome profiling data show that PfSTOML consistently resides in one or more large ~1.5-3.5 MDa protein complexes [39]. SPFH proteins are known to form large homo- or heteromeric complexes, often together with proteases [8,18]. In order to further characterize STOML complexes in P. falciparum, we performed two complementary STOML-pulldown experiments to identify potential interactors. Three proteins, including PfSTOML, PfFtsH metalloprotease, and a protein of unknown function (PF3D7_1306200), were identified as significantly enriched in both pulldown experiments. PF3D7_1306200 contains an AB-hydrolase domain and is predicted to be essential [23]. Similarly to PfSTOML, it is mostly expressed in schizont stages [40], yet the putative hydrolase is predicted to localize to the apicoplast [41,50] and did not show comigration with PfSTOML on a native gel in our complexomics data [39]. Of note, for the co-immunoprecipitation experiments we used synchronized late-stages samples whereas the complexome profiling was performed on mixed ABS. Based on its predicted apicoplast localization, and the lack of comigration with PfSTOML on a native gel, it seems unlikely that PF3D7_1306200 forms a complex with PfSTOML at the IMM, yet, the consistent pull down with the PfSTOML is remarkable [41,50]. Reciprocal pull-down experiments could confirm interaction between PF3D7_1306200 and PfSTOML. During asexual and sexual blood-stage development there are plentiful of close appositions of the mitochondrion and apicoplast, the nature of which remains enigmatic [29]. It might be possible that the PfSTOML - PF3D7_1306200 interaction plays a role in these organelle contact sites. A more detailed microscopic analysis of contact sites with (conditional) knockout of both proteins could shed more light on this hypothesis.

PfFtsH belongs to the AAA (ATPases Associated with various cellular Activities) metalloprotease family at the IMM, which play a role in protein surveillance by degrading non-native integral membrane proteins and membrane associated proteins such as unassembled units of the respiratory chain [51,52]. P. falciparum harbors three FtsH homologs [42]. Two of the P. falciparum FtsH homologs, including PF3D7_1464900, which we identified in our PfSTOML pull-down, locate to the IMM, have a single transmembrane domain, and cluster with i-AAA FtsH homologs, which are exposed to the intermembrane space. The third homolog, PfFtsH1 (PF3D7_1239700), has two transmembrane domains and clusters with m-AAA FtsH homologs at the IMM that are exposed to the mitochondrial matrix. PfFtsH1 forms oligomeric complexes and has a punctate distribution on the mitochondrial branching points in late trophozoite and early schizont stages [42], which is similar to PfSTOML distribution in these stages. Contradictory, another study showed that actinonin, a small molecule inhibitor, targets PfFtsH1 and disrupts apicoplast biogenesis [53] and its homolog in T. gondii is also localized to the apicoplast [54]. Expression of PfFtsH1 in E. coli causes defective cytokinesis, implying a potential role in organelle division. Unfortunately, Amberg-Johnsen and colleagues were unable to generate endogenously tagged knockdown parasite lines of the two i-AAA proteases [53]. Therefore, the exact function and substrates of FtsH in P. falciparum remain to be elucidated.

The interaction between PfSTOML and PfFtsH is well-supported by evidence. Their respective human homologs, SLP2 and Yme1L, form the SPY complex [20], which is essential for the proteolytic regulation of proteins involved in mitochondrial dynamics and quality control. Yme1L also contributes to OPA1 cleavage, a mitochondrial GTPase which is involved in mitochondrial fusion and cristae formation [55,56], however, no Plasmodium OPA1 homolog has been identified to date. In Trypanosoma brucei, TbSLP2 forms a large mitochondrial complex with TbYme1, which is involved in mitochondrial stress resistance [45]. Furthermore, in the filamentous fungus Neurospora crassa, STOML2 has been found in a large complex with an i-AAA protease (IAP1) [57]. Other SPFH family members are also known to form large complexes with AAA+ proteases, such prohibitins with Yta10/Yta12 in yeast [8] or HflK/C with FtsH in bacteria [18]. Characterization of the HflK/C-FtsH supercomplex structure showed that HflK/C forms a 24-heteromer vault structure around four hexameric FtsH complexes at the bacterial membrane [21] (Fig 4E). Here, we show that the PfSTOML AlphaFold2 predicted structure shows high similarity with that of bacterial HflK/C. Our AlphaFold Multimer predictions suggest that PfSTOML might form a 22-multimer barrel structure that is highly similar to the vault structure of HflK/C-FtsH supercomplex in bacteria. Although these predictions are based on modelling and therefore need to be interpreted with caution, the high structural similarity of the PfSTOML 22-multimer complex with HflK/C-FtsH supercomplex, combined with our co-immunoprecipitation data, suggests that PfSTOML might form a similar supercomplex structure with PfFtsH, possibly regulating PfFtsH accessibility.

Taken together, knockout of PfSTOML causes a significantly delayed ABS development, while gametocytes develop normally. PfSTOML has a punctate distribution to mitochondrial branching points and endings of mitochondrial branches but knockout of PfSTOML does not affect mitochondrial morphology. Knockout of PfSTOML did not affect sensitivity to drugs targeting the respiratory chain, suggesting that PfSTOML is not directly involved in respiratory chain assembly. PfSTOML resides in a large supercomplex with PfFtsH, likely forming a large, multimeric barrel structure that regulates the accessibility of PfFtsH, similar to its bacterial family members. Although the exact function of the STOML-FtsH complex in P. falciparum remains to be elucidated, these results could pave the way for future studies into this highly conserved protein family and their role in proteolytic processes and membrane organization.

Materials and methods

P.

falciparum culture and transfections

P. falciparum NF54 and mutant parasites lines were cultured in RPMI1640 medium supplemented with 25 mM HEPES, 10% human type A serum (Sanquin, The Netherlands) and 25 mM NaHCO_3_ (complete medium). Parasites were cultured in 5% human RBCs type O (Sanquin, The Netherlands) at 37°C with 3% O_2_ and 4% CO_2_. For transfection, 60 μg of HDR plasmid was linearized by overnight digestion, precipitated, and transfected with 60 μg Cas9 plasmid using either RBC loading or ring transfection [58]. For RBC loading, plasmids were loaded into RBCs by electroporation (310 V, 950 μF) and a trophozoite parasite culture was added to the transfected RBCs. One day after transfection, parasites were treated with 2.5 nM WR99210 (Jacobus Pharmaceutical) for five days. For ring transfection, a ring-stage sorbitol synchronized parasite culture was transfected with the plasmids by electroporation (310 V, 950 μF). Five hours after transfection, parasites were treated with 2.5 nM WR99210 for five days. Success of transfection was assessed by diagnostic PCR (S1 Fig). Gametocyte cultures were maintained in a semi-automatic culturing system with media changes twice a day [59]. Gametocytes were stress-induced through asexual overgrowing. A mixed asexual culture of 1% was set up and cultured for up to 2 weeks.

Plasmid constructs

To generate the STOML KO repair plasmid, the pGK plasmid was used, which contains a pBAT backbone [60] with H2B promotor, GFP and PBANKA_142660 bidirectional 3’UTR, flanked by multiple cloning sites. The 5’ and 3’ homology regions (HRs) were amplified and cloned into pGK sequentially, using XmaI + XhoI and NcoI + EcoRI restriction sites, respectively, generating pRF0038 STOML KO repair plasmid. CRISPR-Cas9 guide plasmids targeting two different sites in STOML were generated. Guide oligonucleotides were annealed and cloned into pMLB626 plasmid [61] (a kind gift from Marcus Lee) using BbsI restriction enzyme, generating the pRF0039 and pRF0040 final guide plasmids (S1 Table).

To generate STOML tagging repair plasmid, pRF0079 empty tagging plasmid was used, containing 3HA-NG-GlmS, PBANKA_142660 bidirectional 3’UTR, and the mito-mScarlet mitochondrial marker [29]. 5’ HR was generated by overlap PCR, harboring a shield mutation that prevents cutting of CRISPR-Cas9 when the construct is integrated. 5’ and 3’ HRs were cloned into pRF0079 using KpnI + BamHI and EcoRI + NgoMIV restriction enzymes, respectively, generating pRF0166 STOML tagging plasmid. Because this plasmid was unsuccessful in generating mutant parasite line after three transfection attempts, we decided to clone the DHFR selection marker in the repair plasmid and remove it from the guide plasmid. By addition of WR after transfection, we will then directly select for parasites with integration of the DHFR cassette, instead of selection on the guide plasmid. The DHFR cassette was removed from pRF0040 guide plasmid by digestion with EcoRI and ApaI, followed by blunt end generation with DNA polymerase I (klenow), following manufacturer’s instructions, and ligation. The new guide plasmid without DHFR was termed pRF0210. The DHFR cassette cloned from MLB626 plasmid into pRF0166 with SphI and EcoRI restriction enzymes, generating pRF0213 3HA-NG-GlmS tagging plasmid with mito-mScarlet and DHFR selection marker. Since a big fluorescent tag might interfere with protein function, we also generated a STOML tagging repair plasmid by removing mNG from pRF0213, generating pRF0266 3HA-GlmS tagging plasmid, using BamHI and NheI restriction enzymes.

For generation of the repair plasmids for cyto-mScarlet and cyto-mNG parasite lines (used for the competition growth assay), SIL7 reporter plasmid (pRF0057) was used [29]. mScarlet was amplified from p1.2RhopH3-HA-mScarlet [62] (a kind gift from Prof. Alan Cowman) (S1 Table) and cloned into pRF0057 using AfeI and NheI restriction enzymes, generating pRF0278 cyto-mScarlet repair plasmid. mNeonGreen was amplified from pRF0079 plasmid and cloned into pRF0278 with AflII and NheI restriction sites to generate pRF0290, the cyto-mNG repair plasmid. CRISPR-Cas9 guide plasmids targeting SIL7 were used [29]. All enzymes used were obtained via New England Biolabs.

Competition growth assay

For the competition growth assay, parasite lines harboring a cytosolic mScarlet or cytosolic mNG were generated by integration of cyto-mScarlet and cyto-mNG constructs in SIL7 integration site [29]. Cyto-mScarlet, Cyto-mNG, stoml^-^ and stoml^-^mito were synchronized by a 63% Percoll centrifugation. Late-stage parasites were isolated from the Percoll gradient and added to fresh RBCs. Four hours later, 5% sorbitol synchronization was performed, which allowed only young rings that just invaded a new RBC to survive. Ring-stage parasites were counted and diluted to each have 0.4% final parasitemia in Cyto-mScarlet + Cyto-mNG, Cyto-mScarlet + stoml^-^, and Cyto-mScarlet + stoml^-^mito mixed cultures. Samples for flow cytometry analysis were taken directly after set-up, at day 7 and at day 14. Samples from each mixed culture were taken and stained with 0.5 μg/ml Hoechst 33342 (Invitrogen, H3570) for 30 min at 37°C. Samples were directly analyzed on BD FACSAria III Cell Sorter and number of red and green parasites were counted. Data was analyzed in FlowJo (version 10.10).

Growth assay

For this growth assay, MitoRed (WT), stoml^-^, and stoml^-^mito parasites were synchronized by a 63% Percoll centrifugation. Late-stage parasites were isolated from the Percoll gradient and added to fresh RBCs. Four hours later, 5% sorbitol synchronization was performed, which allowed only young rings that just invaded a new RBC to survive. Ring-stage parasites were counted and diluted to 0.05% parasitemia. Samples for flow cytometry analysis and fluorescent microscopy were taken directly after setup (t = 0), and then every 8, 16 or 24 h over a period of 8 days. To prevent overgrowth, parasite cultures were cut back 1/100 on day 3, and 1/50 on day 6. For flow cytometry, samples were taken and fixed in 0.25% glutaraldehyde. All samples from different time points were processed at the same time, by staining with Hoechst 33342 for 30 min at 37°C and then analyzed by flow cytometry (Beckman Coulter Cytoflex) to determine parasitemia using the 405 nm laser. Data was analyzed in FlowJo (version 10.10). Final parasitemia was adjusted for the dilution factor, explaining why final parasitemia can reach more than 100%. For fluorescent microscopy, parasite samples were processed as described in “fixed imaging” paragraph below.

Live imaging

Stoml-NG parasites were stained with Hoechst 33342 for 30 min at 37°C and settled in an 8-well imaging chamber (Ibidi) in complete media without phenol red. Parasites were imaged on a Zeiss LSM880 or LSM900 Airyscan microscope with 63x oil objective and 37°C heated stage, using 405, 488, and 561 nm excitation lasers. Images were Airyscan processed before analysis with FIJI software [63] and Arivis vision4D (Zeiss).

Fixed imaging

For fixed immunofluorescence microscopy of the asexual and sexual blood stages of the stoml^-^mito lines, parasites were settled on a poly-L-lysine coated coverslip for 20 min at 37°C. Parasites were fixed using 4% EM-grade paraformaldehyde and 0.0075% EM-grade glutaraldehyde in PBS for 20 min at room temperature. Cells were permeabilized with 0.1% Triton X-100 for 10 min, before staining using 1 μM DAPI in PBS for 1 h. For imaging of stoml^-^mito gametocytes, parasites were stained for 1 h with primary mouse anti-alfa-tubulin antibody (A11126, ThermoFisher, 1:500) and secondary goat anti-mouse Alexa Fluor 647 (A21247, ThermoFisher, 1:200) antibody in 3% BSA in PBS. For experiments using the stoml-HA parasite line, asexual blood stage cultures were stained with MitoBrilliant 646 (Tocris, 1:10,000 in complete medium) for 20 min at 37°C. Afterwards, cells were fixed and permeabilized as described above. Parasites were stained for 1 h with primary rat anti-HA (3F10, Roche, 1:100) and secondary goat anti-rat Alexa Fluor 488 (A48262, ThermoFisher, 1:200) antibodies diluted in 3% BSA in PBS, followed by a DAPI staining as described above. All slides were mounted with Vectashield (Vector Laboratories). Samples were imaged with a Zeiss LSM880 or LSM900 Airyscan microscope with 63x oil objective and 405, 488, 561, and 633 nm excitation lasers. Images were Airyscan processed using Zeiss Zen Blue Software, before analysis with FIJI software [63] and Arivis vision4D (Zeiss).

Co-immunoprecipitation assay

Stoml-HA and stoml-NG parasites were synchronized with 5% sorbitol and harvested 22 h later to obtain late-stage parasites. Parasites were treated with 0.06% saponin, snap-frozen in liquid nitrogen, and stored at -80°C until further processed. Nitrogen cavitation was used for cell disruption as described [39]. On the day of the experiment, 18 pellets of 30-ml cultures per parasite line were resuspended and pooled in 25 ml ice-cold MESH-buffer (250 mM sucrose, 10 mM HEPES, 1 mM EDTA, 1 × cOmplete EDTA-free Protease Inhibitor Cocktail (Sigma), pH 7.4). The sample was added to the pre-chilled cell disruption vessel (#4639 Parr Instrument Company) and pressurized with nitrogen gas at 1500 psi for 45 min on ice. The parasites were then sheared through slow release. The organelle-enriched fraction was obtained by differential centrifugation as described [39]. Protein concentrations were determined by Pierce BCA Protein Assay Kit (Thermo Scientific). Samples were solubilized with n-dodecyl-β-D-maltoside (DDM) (Sigma), using 3:1 detergent:protein (w/w) ratios. Solubilized samples were spun down at 22,000 x g at 4°C. Supernatant derived from stoml-NG samples were applied on ChromoTek mNeonGreen-Trap magnetic agarose beads (ChromoTek), or empty binding control agarose beads (ChromoTek). Supernatant from stoml-HA samples were applied on Pierce HA-tag magnetic beads (Thermofisher) or empty protein G binding control beads (Thermofisher). Both pulldowns were carried out with three technical replicates. Beads were incubated at 4°C for 30 minutes with gentle agitation and then washed twice with washing buffer (PBS, 1mM EDTA, 1 × cOmplete EDTA-free Protease Inhibitor Cocktail, 0.05% DDM) and three times with ice-cold PBS, using a magnetic stand. After washes, on bead digestion was performed as follows: beads were resuspended in 50 μl elution buffer (2M urea, 100 mM Tris-HCl pH 8.0, 10 mM DTT) and incubated for 20 minutes at 25°C while shaking. To alkylate cysteines, iodoacetamide was added to a final concentration of 50 mM. Samples were kept in the dark for 10 min at 25°C. Subsequently, 0.25 μg of sequencing grade tryspin (Promega) was added to digest the proteins. The samples were shaken at 25°C for 2 h. The supernatants, containing the partially digested proteins, were collected and 50 μl of fresh elution buffer was added to the beads and shaken for another 5 min. Next, these supernatants were collected and combined with the first supernatant. Another 0.1 μg of trypsin was added, to stimulate overnight digestion at 25°C. The next day, samples were concentrated and purified on C18 stagetips [64]. Samples were analyzed on a Thermo Exploris 480 mass spectrometer, operated with an online Easy-nLC 1000. A gradient of buffer B (80% acetonitrile, 0.1% formic acid) was applied for 60 min. The mass spectrometer was ran in Top20 mode, while dynamic exclusion was enabled for 45 sec. Raw data was analyzed using Maxquant version 1.6.6.0 [65] with a Plasmodium database (strain 3D7, version August 5th 2022, obtained from plasmodb.org [66]). LFQ, iBAQ and match between runs were enabled, and deamidation (NQ) was added as additional variable modification. The output was filtered using Perseus 1.5.0.15 [67]. Proteins marked as potential contaminants, reverse hits, and proteins with less than 2 peptides were removed. Samples were grouped into triplicates, and proteins with less than 3 valid values in at least 1 group were removed, after which missing values were imputed using the default settings. A t-test was performed to identify specific outliers. Data was visualized using R. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [68] partner repository with the dataset identifier PXD039772.

Drug sensitivity assay

NF54 and stoml^-^ parasites were used in a replication assay as described [69] to determine sensitivity to anti-malarial compounds. Briefly, parasites were diluted to 0.83% parasitemia and 3% hematocrit. Thirty microliters of diluted parasites were combined with 30 µl of compound serially diluted in dimethyl sulfoxide (DMSO) and RPMI 1640 medium to reach a final DMSO concentration of 0.1% in a total assay volume of 60 µl. Parasites were incubated at 37°C for 72 hours with mixed gas (3% O_2_ and 4% CO_2_). Then 30 µl of lysis buffer containing 1:15,000 SYBR Green reagent (Life Technologies), 13.3 mM Tris-HCl, 3.3 mM EDTA, 0.067% TritonX-100 and 0.0053% saponin was added and fluorescence intensity was quantified using BioTek Synergy 2 plate reader. GraphPad Prism was used for data analysis and inhibitory dose-response curves were determined with a variable slope model, in which the curve is generated with the following formula: $[eqn]$

DALI search and AlphaFold2 structure predictions

AlphaFold Multimer [48] predictions were performed using the COSMIC^2^ platform (https://cosmic-cryoem.org/tools/alphafoldmultimer/). All predicted protein and protein complex structures and alignments were visualized using PyMOL Molecular Graphics System (version 2.5.2. Schrödinger, LLC).

Supporting information

S1 FigGeneration of STOML tagging and KO parasite lines.A) Schematic overview of STOML tagging and KO strategy. For tagging of STOML with 3HA-NG-glmS or 3HA-glmS, CRISPR-Cas9 (indicated by scissors) is used to introduce a double-strand break to facilitate integration of the linear repair constructs 3HA(-NG)-glmS tag directly after STOML before the stop codon, while at the same time integrating a mito-mScarlet mitochondrial marker and a DHFR drug selection cassette. For STOML KO, two CRISPR-Cas9 introduced DNA breaks at the 5’ and 3’ of the gene will be repaired by the linearized HDR plasmid. After integration, STOML will be replaced by GFP under the control of the H2B promotor. B) Diagnostic PCR of stoml-NG, stoml-HA, stoml^-^ and stoml_-(mito)_ parasite lines with integration specific primer combinations (indicated in panel A), demonstrating successful 5’ and 3’ integration and the absence of WT parasites (T = total). C) Western blot analysis showing expression of STOML-3HA-NG (73 kDa) and STOML-3HA (47 kDa) at expected sizes using anti-HA antibody and anti-HSP70 for loading control. D) Fluorescence microscopy of stoml-HA with anti-HA antibody (green), mito-mScarlet mitochondrial marker (magenta), DAPI for DNA visualization (blue), and DIC. Images are maximum intensity projections of Z-stack confocal Airyscan images, except images in right panel, which are 3D visualizations generated with Imaris analysis software. Arrowheads indicate PfSTOML-3HA signal at mitochondrial branching points (white) or mitochondrial branch endings (yellow). Scale bars are 2 µm for maximum intensity projections and 1 µm for 3D visualizations.(TIF)

S2 FigGeneration of cyto-mScarlet and cyto-mNG parasite lines.A) Schematic overview of transfection strategy to generate cyto-mScarlet and cyto-mNG. CRISPR-Cas9 and two guides were used to generate double stranded breaks in a silent intergenic locus (SIL7), characterized in Verhoef et al. [29] (indicated by scissors). DNA breaks are repaired by double homologous recombination with a repair plasmid containing 5’ and 3’ homology regions (HRs) and a fluorescent protein (FP, mScarlet or mNeonGreen) under the control of the H2B promotor and PBANKA_142660 bidirectional 3’UTR. B) Diagnostic PCR of cyto-mScarlet and cyto-mNG parasite lines with integration-specific primer combinations (indicated in panel A), demonstrating successful 5’ and 3’ integration and the absence of WT parasites (T = total).(TIF)

S3 FigMitochondrial morphology in stoml_-mito_ ABS parasites.Fluorescent microscopy of stoml_-mito_ and MitoRed (WT) parasites during ring, trophozoite, early and late schizont stages. The mito-mScarlet signal is preserved after fixation and can be observed without antibody staining. DNA was stained using DAPI. Images are maximum intensity projections of Z-stack confocal Airyscan images. Scale bars, 2 µm.(TIF)

S4 FigSensitivity of stoml^-^ parasites to anti-malarial compounds.Drug sensitivity assay for P. falciparum NF54 and stoml^-^ parasites. The graphs show average values for mean parasite density relative to controls for asexual blood-stage replication assay and represent one of the two independent replicates. Error bars indicate SEM determined from two technical replicates per experiment. The data were analyzed using nonlinear regression in GraphPad Prism. Proguanil, DSM1, DSM265, ELQ300, and Atovaquone are compounds targeting the parasite mitochondrion, while DHA, chloroquine, and MMV183 are non-mitochondrial compounds.(TIF)

S5 FigStoml_-mito_ parasites develop to healthy gametocytes that exflagellate.A) Fluorescent microscopy on male (M) and female (F) stoml_-mito_ stage V gametocytes. Parasites were stained for tubulin (yellow) to distinguish male (high α-tubulin signal) from female (low α-tubulin signal) gametocytes. B) fluorescent microscopy on exflagellating stoml_-mito_ male gametes at 20 minutes after activation. Parasites were stained with tubulin to visualize axonemes. A-B) Visualization of mito-mScarlet mitochondrial marker (magenta), cytosolic GFP, DAPI for DNA visualization (blue), and DIC. Images are maximum intensity projections of Z-stack confocal Airyscan images. Scale bars, 2 µm.(TIF)

S6 FigAbsolute protein abundance of STOML, FtsH and PF3D7_1306200 across protein complex migration on native PAGE gel.Line graphs based on previously published complexome profiling data [39] showing migration patterns and absolute protein abundance (iBAQ value, logarithmic scale) of STOML (red), FtsH (blue), and PF3D7_1306200 (green) in asexual blood stages (ABS) and gametocytes (GTC). Co-migration on blue native gel within the same molecular weight (MW) range (x-axis) indicates complex formation.(TIF)

S7 FigIdentification of STOML interacting proteins with co-immunoprecipitation.Anti-HA immunoprecipitation of PfSTOML-HA containing complexes. The volcano plot showing mean log_2_ fold changes (FC) and -log_10_ false discovery rate (FDR) for anti-HA pulldown in comparison with control pulldown. Horizontal and vertical dotted lines indicate log_2_ FC > 2.5 (FC > 5.66) and -log_10_ FDR > 1.301 (FDR > 0.05) respectively. Dark dots represent proteins that are highly enriched or reduced in the anti-HA pulldown compared to the control pulldown.(TIF)

S8 FigAlphaFold2 structure prediction of PfSTOML multimers.A) AlphaFold2 prediction of PfSTOML 24-multimer with side view (top) and top view (bottom). B) AlphaFold2 prediction of PfSTOML 8-multimer with side view (left) and top view (right). Coloring in A and B represent model confidence as indicated by the color legend in B. C) Top view of predicted PfSTOML 8-multimer structure, indicating the angles measured between SPFH domains of different STOML proteins in the complex. D) Graphs with pLDDT scores representing model confidence of predicted PfSTOML 24, 8, and 22 multimer structures.(TIF)

S1 TablePrimer and guide sequences for generation of repair and guide plasmids.Used abbreviations: HR = homology region, F = forward primer, R = reverse primer. Overhang for restriction sites are red, restriction sites are underlined, and gRNA sequences are blue. The same primers are used for the integration PCR of cyto-mScarlet and cyto-mNG.(XLSX)

S2 TableProteins detected in STOML-3HA-NG pulldown.This table includes all proteins detected in the STOML-3HA-NG pulldown, indicated by their gene ID in the first column and the annotation in the second column. T-test significance, unique peptides, log p-value and t-test difference are indicated in the following columns. Mitochondrial prediction score is indicated in the last column as “Mito score”, which refers to the ranking of the mitochondrial proteome as shown by Esveld et al. [27].(XLSX)

S3 TableProteins detected in STOML-3HA pulldown.This table includes all proteins detected in the STOML-3HA pulldown, indicated by their gene ID in the first column and the annotation in the second column. T-test significance, unique peptides, log p-value and t-test difference are indicated in the following columns. Mitochondrial prediction score is indicated in the last column as “Mito score”, which refers to the ranking of the mitochondrial proteome as shown by Esveld et al. [27].(XLSX)

S1 Movie3D visualization of STOML-3HA-NG in late ring stage.3D visualization of STOML-3HA-NG (green) and mito-mScarlet (magenta) in ring stages with live confocal Airyscan microscopy, using Arivis 4D vision software. Fluorescent signal is segmented by manual thresholding.(MP4)

S2 Movie3D visualization of STOML-3HA-NG in early schizont stage.3D visualization of STOML-3HA-NG (green) and mito-mScarlet (magenta) in early schizonts with live confocal Airyscan microscopy, using Arivis 4D vision software. Fluorescent signal is segmented by manual thresholding.(MP4)

S3 Movie3D visualization of STOML-3HA-NG in late schizont stage.3D visualization of STOML-3HA-NG (green) and mito-mScarlet (magenta) in late schizonts with live confocal Airyscan microscopy, using Arivis 4D vision software. Fluorescent signal is segmented by manual thresholding.(MP4)