Main

Apicomplexan parasites are invasive and defined by the presence of an apical complex used to recognize and gain entry into host cells. The apical complex includes two types of secretory organelle: micronemes and rhoptries3. Microneme proteins are secreted to the parasite surface and mediate motility, host cell recognition and invasion4. Rhoptry proteins are injected directly into the host cell2, where they anchor the machinery propelling the parasite into the host cell5, facilitate nutrient import6,7,8, interfere with the immune response and modulate gene expression to promote infection9. Rhoptry secretion requires a trigger in the form of microneme proteins binding to host cell receptors10,11. Importantly, rhoptry proteins not only cross the plasma membrane of the parasite (exocytosis) but also that of the host1,12. Rhoptry exocytosis factors identified to date11,13,14 are mostly specific to Apicomplexa, suggesting that these cells depend on unique lineage-restricted secretory mechanisms. Consistent with this, no eukaryotic SNAREs—the main drivers for fusing vesicles to the target membrane in eukaryotic systems—have so far been associated with rhoptry exocytosis. Apicomplexa also lack genes encoding prokaryotic secretion systems15. Thus, how rhoptry effectors are delivered into the host cytoplasm remains unclear.

To explore the mechanisms of rhoptry secretion, we searched for examples of regulated secretion in organisms phylogenetically closely related to Apicomplexa16. All Alveolata bear alveolar sacs beneath the plasma membrane, which give the superphylum its name. They contain elaborate membrane-bound secretory organelles with shared evolutionary origin6 but different morphologies and functions. Known as trichocysts in the Ciliata Paramecium, they function in defence, docking at the plasma membrane and discharging in response to predation17. Exocytic membrane fusion occurs at plasma membrane sites consisting of rosettes of eight to nine intramembranous particles (IMPs)18,19. Previous electron microscopy studies have described similar rosettes at the apex of several apicomplexan parasites20,21,22,23. Although it has been suggested to be involved in exocytic fusion20, the function of the IMP rosette has never been experimentally addressed in Apicomplexa. Analysis of Paramecium tetraurelia mutants defective in both trichocyst exocytosis and rosette assembly24,25 led to the identification of non-discharge (nd) genes26,27,28,29 (Supplementary Table 1). To determine whether similar factors could be involved in exocytosis of secretory organelles in Apicomplexa, we first searched for Nd homologues in the tree of life. Genome mining for nd genes and phylogenetic analyses revealed that nd6 and nd9 are conserved in Ciliata, but also in Dinoflagellata, Chromerida and Apicomplexa (Extended Data Fig. 1). We also found an Nd9 homologue in the Perkinsozoa, a sister group of Dinoflagellata. Altogether, this analysis suggested a conserved evolution and common function of nd genes across Alveolata.

To define the parasite localization of Nd proteins, we tagged nd6 (TGGT1_248640) and nd9 (TGGT1_249730) at the endogenous loci in Toxoplasma gondii, an experimentally tractable apicomplexan (Extended Data Fig. 2). Both proteins displayed a punctuate signal throughout the cytoplasm, but in addition, TgNd6 accumulated at the apical tip of the parasite (Fig. 1a,b). A closer look at the apical tip by immunoelectron microscopy localized TgNd6 at the site of rhoptry exocytosis, in association with the parasite plasma membrane and an underlying membranous spheroid known as the apical vesicle (Fig. 1c). Despite being previously observed by electron microscopy in Toxoplasma21,30 and indications that it is linked to rhoptries22,30, the composition and function of the apical vesicle remains unknown.

Fig. 1: Rhoptry secretion is dependent on rosette formation.
figure 1

a, IFAs of the control (Ctrl) untagged line and endogenously HA3-tagged TgNd6 and HA3-tagged TgNd9 tachyzoites. The green arrows point to TgNd6 apical puncta. ARO is a marker for rhoptries. Shown are the maximum-intensity projections of confocal z-stacks of fixed parasites. The results are representative of five experiments. DIC, differential interference contrast. b, Super-resolution microscopy of the apical dot of TgNd6. Top: schematic of the apical end of a tachyzoite. The images show maximum-intensity projections of z-stacks of TgNd6–HA3 parasites transiently expressing ring-1–GFP (RNG1–GFP) (middle) and centrin 2–Ty2 (CEN2–Ty2) (bottom). The protein RNG1–GFP marks the apical polar ring (APR) and CEN2–Ty2 marks pre-conoidal rings (PCR). Higher magnifications of the areas indicated by a dashed box (right) show that TgNd6–HA3 (red arrow) localizes above the apical polar ring (green arrow) and co-localizes partially with CEN2. The results are representative of two experiments. c, Left: Immunogold labelling of TgNd6–HA3. Right: TgNd6–HA3 on the apical vesicle (AV). Inset: higher-magnification view of the apical vesicle. Micronemes (m) and rhoptries (Rh) are visible in a transverse section of the conoid (Co). The results are representative of two experiments. d, Quantification of invasion after depletion of TgNd6 (left) and TgNd9 (right). The data represent means ± s.e.m. (n = 3 independent experiments). NS, not significant. e, Immunoblot showing microneme secretion assessed by the release of proteolytically cleaved AMA1 (arrow = processed/secreted TgAMA1) in Tgnd6 iKD ± IAA at 24 h (left) and Tgnd9 iKD ± ATc at 72 h (right). P, pellet; Sup, supernatant; Sup ind, propanolol-induced supernatant. Dense granule protein 3 (GRA3) was used as a loading control. The results are representative of three experiments for Tgnd6 iKD and four experiments for Tgnd9 iKD. f, Rhoptry secretion assay by SeCrEt. Shown are quantification of the rhoptry secretion of Tgnd6 iKD ± IAA (left) and Tgnd9 iKD ± ATc (right). The data represent means ± s.e.m. (n = 3 independent experiments). g, Left: Freeze-fracture electron microscopy image of a T. gondii tachyzoite (P-face) showing a rosette of IMPs (white arrow) at the middle of the apex. Middle: higher-magnification view of the images shown in the left panel. The white arrowheads point to the eight IMPs of the rosette. Right: quantification of rosettes of IMPs in Tgnd6 iKD ± IAA at 24 h (left) and Tgnd9 iKD ± ATc at 72 h (right) using the freeze-fracture method. The results are representative of one experiment for Tgnd6 iKD and two experiments for Tgnd9 iKD. Statistical significance in d and f was determined by unpaired two-tailed Student’s t-test.

Source data

Full size image

Because Tgnd6 and Tgnd9 were predicted to be fitness-conferring genes31, to investigate their function, we generated inducible knockdown mutants using an auxin-inducible degron for TgNd6 (ref. 32) and tetracycline-induced repression for TgNd9 (refs. 33,34) (Extended Data Fig. 2). Parasites conditionally depleted of TgNd6 or TgNd9 showed reduced plaque formation on fibroblast monolayers, indicating the inability of both mutants to efficiently complete the lytic cycle (Extended Data Fig. 3a). Tgnd6 and Tgnd9 inducible knockdown (iKD) mutants showed no detectible defects in their intracellular replication, egress, conoid protrusion, motility or host cell attachment (Extended Data Fig. 3b–g). In contrast, invasion was severely impaired for both mutants (Fig. 1d).

To understand the mechanistic basis of this invasion defect, we evaluated microneme and rhoptry secretion in the Tgnd6 iKD and Tgnd9 iKD mutants. Release of the micronemal protein apical membrane antigen-1 (AMA1) into the supernatant was unimpeded by the absence of TgNd6 or TgNd9 (Fig. 1e). In contrast, rhoptry secretion was remarkably impaired by the loss of TgNd6 or TgNd9, as revealed by quantifying the release of the rhoptry protein 1 (ROP1) into host cells using immunofluorescence assays (IFAs)35 (Extended Data Fig. 3h), or the delivery of Cre recombinase fused to the rhoptry protein toxofilin into the nucleus of a suitable reporter cell36 (Fig. 1f). This was not due to defects in rhoptry biogenesis, as the mutants showed normal organelle formation and apical positioning, as judged by IFA and electron microscopy (Extended Data Fig. 4). Analysis by freeze-fracture electron microscopy of the apex of Tgnd6 iKD mutants showed no consequential reduction in the presence of the apical rosette (Fig. 1g). In contrast, T. gondii depleted of TgNd9 displayed a strong decrease in cells with a rosette at the parasite apex (Fig. 1g). This last result links rosette formation with rhoptry secretion and supports an evolutionarily conserved mechanism of regulated exocytosis in Alveolata.

To determine whether other Apicomplexa require nd genes for rhoptry function, we analysed Nd9 in Plasmodium falciparum, the causative agent of the deadliest form of malaria. We confirmed that P. falciparum intracellular merozoites possess a fusion rosette of eight IMPs (Fig. 2a). As nd9 is also predicted to have a fitness cost in P. falciparum36, we used the rapamycin-inducible dimerizable Cre recombinase (DiCre) system37 to conditionally excise the low-expressed Pfnd9 gene (PF3D7_1232700) (Extended Data Fig. 5a–d). DiCre-mediated ablation of Pfnd9 resulted in substantial reduction in parasite proliferation (Fig. 2b), which was due to the inability of Pfnd9 inducible gene knockout (iKO) mutants to reinvade host cells while their intracellular development and egress were unaffected (Fig. 2b,c and Extended Data Fig. 5e). As in T. gondii, microneme secretion was unaltered in Pfnd9 iKO parasites (Fig. 2d) but rhoptry secretion was affected (Fig. 2e).

Fig. 2: PfNd9 is essential for rhoptry secretion in P. falciparum.
figure 2

a, Top: freeze-fracture electron microscopy of a P. falciparum merozoite (P-face) showing a rosette of IMPs (white arrow). Bottom: higher-magnification view of the image shown above. Scale bars, 100 nm. The results are representative of two experiments. b, Left: growth curves (parasitaemias) of p230p DiCre (Ctrl) and the Pfnd9 iKO mutant ± rapamycin showing that PfNd9-depleted parasites have a growth defect. The means of three technical replicates are shown. The results are representative of three independent experiments. Right: Giemsa staining of the representative growth experiment illustrating development and reinvasion of p230p DiCre (Ctrl) and Pfnd9 iKO asexual parasites (along two cycles) ± rapamycin treatment. hpi, hours post-invasion. c, Quantification of egress of Pfnd9 iKO ± rapamycin schizonts. The data were collected from eight videos of Pfnd9 iKO ± rapamycin. The means of two independent experiments are shown. d, Left: IFA images illustrating AMA1 protein stored in micronemes (top) or secreted and translocated at the surface of the parasite before egress (bottom). Middle: Magnified images of the areas indicated by a dashed box to the left. Right: quantification of the results. The means of two independent experiments are shown. e, Quantification of rhoptry secretion events in Pfnd9 iKO ± rapamycin-treated schizonts using anti-PfRAP2 antibodies to visualize rhoptry secretion events (spits of RAP2 exported into the RBC). The means of two independent experiments are shown.

Source data

Full size image

To gain a more complete understanding of the molecular composition of the rhoptry secretion machinery, we searched for Nd interacting proteins in T. gondii. TgNd9 displays two Armadillo repeats while TgNd6 shows homology with GDP/GTP exchange factors known to activate GTPases (Fig. 3a). We used TgNd9 for immunoprecipitation experiments as we found TgNd6 to be largely insoluble (Extended Data Fig. 2g). Mass spectrometry analysis revealed robust interaction of TgNd9 with TgNd6 and with TgFER2 (TGGT1_260470), a member of the ferlin calcium sensor family that is known to be essential for Toxoplasma rhoptry secretion13 (Fig. 3a and Supplementary Table 2). The TgNd9 immunoprecipitation also enriched TGGT1_222660, a protein harbouring Armadillo repeats and leucine-rich repeats, named hereafter TgNdP1 (Nd partner 1), and TGGT1_316730 (TgNdP2 (Nd partner 2)), a protein with a C2 calcium lipid-binding domain (Fig. 3a). The genes transcribing both proteins are broadly shared among Alveolata (Extended Data Fig. 1c,d) and are predicted to be fitness conferring in Toxoplasma31 and Plasmodium38. In contrast, the TgNd9 interacting partner TGGT1_277840, a GTPase, is restricted to Apicomplexa and Dinoflagellata, and TGGT1_253570 is only found in the apicomplexan subgroup Coccidia. Importantly, all of these proteins are also recovered when using reverse immunoprecipitation with tagged TgNdP1 protein (Supplementary Table 3).

Fig. 3: Nd6 and Nd9 are part of an Alveolata complex essential for organelle secretion in T. gondii (Apicomplexa) and T. thermophila (Ciliata).
figure 3

a, Mass spectrometry analysis of immuno-isolated TgNd9–HA3. Left: volcano plot of the proteins differentially enriched in TgNd9 versus control immunoprecipitation. This plot presents the fold change (difference) and significance (−log[P]) obtained from an unpaired two-tail Student’s t-test of three independent immunoprecipitations using LFQ intensity values. Right: schematic of TgNd proteins using SUPERFAMILY63. ARM, Armadillo-repeat domain; C2, calcium lipid-binding-like domain; RCC1, regulator of chromosome condensation 1-like domains (a versatile domain that performs many different functions, including guanine nucleotide exchange on small GTP-binding proteins64); LRR, leucine-rich-repeat domain. b, Immunofluorescence of endogenously HA3-tagged TgNdP1 and TgNdP2 tachyzoites. The white arrows point to TgNdP2 apical dots of two adjacent parasites, which are magnified on the right. ARO was used as a rhoptry marker and 4′,6-diamidino-2-phenylindole (DAPI) was used as a DNA marker. The results are representative of three experiments. c, Quantification of invasion after depletion of TgNdP1 (left; using ATc) and TgNdP2 (right; using IAA) along with negative control strains. The data represent means ± s.e.m. (n = 3 independent experiments). d, Rhoptry secretion quantification of TgndP1 iKD ± ATc (left) and TgndP2 iKD ± IAA (right). The data represent means ± s.e.m. (n = 3 independent experiments). e, Quantification of rosettes of IMPs in Tgndp1 iKD ± ATc at 72 h and Tgndp2 iKD ± IAA at 24 h using the freeze-fracture technique. The results are representative of one experiment. f, Quantification of mucocyst exocytosis by dibucaine assay. The data represent means ± s.e.m. (n = 3 independent experiments). Statistical significance in c, d and f was determined by unpaired two-tailed Student’s t-test.

Source data

Full size image

To learn more about the conserved partners of TgNd9, we generated T. gondii lines in which TgNdP1 or TgNdP2 was tagged by an epitope and could be ablated conditionally (Extended Data Fig. 6). Both proteins appeared as punctate cytoplasmic staining (Fig. 3b) but TgNdP2—like TgNd6—also appeared as a dot at the apical tip of the parasite, although consistently with lower intensity. Depletion of TgNdP1 or TgNdP2 resulted in a profound growth defect (Extended Data Fig. 7a) that we linked to impairment of host cell invasion and rhoptry secretion (Fig. 3c,d and Extended Data Fig. 7). Again, loss of rhoptry secretion went hand in hand with loss of the rosette in Tgndp1 iKD and Tgndp2 iKD parasites (Fig. 3e).

To further validate our conservation data and broaden our discoveries back to Ciliata, we generated knockouts for orthologues of both NdP1 (TTHERM_01287970; Ttn∆ndp1) and NdP2 (TTHERM_00498010; Ttn∆ndp2) in the free-living Ciliata Tetrahymena thermophila (Extended Data Fig. 8a,b). The homologous organelles to Paramecium trichocysts in Tetrahymena are called mucocysts and are non-essential for laboratory growth. We found that Ttn∆ndp1 and Ttn∆ndp2 cells were defective in mucocyst secretion, which was triggered by exposure to dibucaine39 (Fig. 3f). We further showed that the impairment of exocytosis was not due to defects in mucocyst biogenesis, given that mucocyst maturation (as measured by processing of mucocyst pro-proteins) and trafficking (monitored by IFA) remained unaltered (Extended Data Fig. 8c,d). Taken together, we identified a complex of proteins essential for organellar exocytosis and rosette assembly, conserved across Alveolata.

The exact position of the rosette relative to the apical tip of the rhoptry and the enigmatic apical vesicle remained elusive since the freeze-fracture technique—used to image the rosette on the membrane—does not capture the internal structures at the same time. To overcome this limitation, we imaged the Toxoplasma apex by cryo-electron tomography (cryo-ET)—a technique combining the advantages of three-dimensional imaging with molecular resolution to reveal the ultrastructure in situ in its native biological context. We were able to simultaneously visualize and define three linked elements: (1) the rosette (dark blue); (2) the apical vesicle (magenta); and (3) the apical tip (cyan) of the rhoptry (orange) (Fig. 4a,b and Extended Data Fig. 9). The rosette showed an eightfold rotational symmetry around a central axis (Fig. 4c,d) and extended under the parasite plasma membrane (light blue) to interact with the apical vesicle (Fig. 4c). Thus, the rosette is tightly sandwiched between the apical vesicle and the plasma membrane and extensively interacts with both membranes. The apical vesicle in turn sits over the rhoptry tip (Fig. 4a,b). The contiguity of all elements from the rhoptry tip to the plasma membrane supports the idea that the rosette and apical vesicle are integral parts of the rhoptry secretion machinery. Interestingly, the connection between the apical vesicle and rosette is sometimes observed even in the absence of a docked rhoptry (Fig. 4e), suggesting that they assemble independent of rhoptry docking. Supporting this hypothesis, when we imaged the Toxoplasma armadillo repeats only protein (ARO) mutant in which rhoptries fail to dock and are dispersed in the cytoplasm40, both the rosette (seen using the freeze-fracture technique; Fig. 4f) and apical vesicle (seen by electron microscopy; Fig. 4g) were still present at the apex of the parasite. Altogether, these results positioned the rosette and apical vesicle at the heart of the exocytic machinery in Apicomplexa and implied that rhoptries do not make contact and fuse directly with the plasma membrane.

Fig. 4: The rhoptry secretion machinery includes the apical vesicle and rosette.
figure 4

a, Left: tomogram slice showing a side view of the apical complex, including the conoid (brown), pre-conoidal apical rings (grey), micronemes (yellow), plasma membrane (light blue) and rhoptry secretion system (consisting of the rosette (dark blue), apical vesicle (magenta), rhoptry (orange) and rhoptry tip density (cyan)). Right: original image without annotation or colour overlays. b, Left: magnified images (with and without colour overlays) of the boxed region in a, showing the connections between the rhoptry, rhoptry tip density, apical vesicle, rosette and plasma membrane. Right: three-dimensional segmentations of the left image. In the left of the two images, the plasma membrane is rendered transparent to enable a view of the rosette. c, Magnified images (with and without colour overlays) of the boxed region in b, showing the side view of the rosette. d, Top views (with and without colour overlays and annotation) of the rosette in c, showing an eightfold rotational symmetry. e, Images (with and without colour overlays) of the apical vesicle connected with the plasma membrane via a rosette in the absence of docked rhoptry. All of the measurements were made in three dimensions. The images in c and d are from two different cells oriented differently on the electron microscopy grid. The results in a–e are representative of 20 tomograms produced from two independent samples. f, Quantification of apical rosettes in Tgaro iKD mutants ± ATc at 72 h. The results are representative of two experiments. g, Left: ultrastructure of wild-type tachyzoites with the apical vesicle positioned beneath the plasma membrane and above the tip of the rhoptry neck. Right: in Tgaro iKD ATc-treated tachyzoites, the apical vesicle is still properly positioned at the apex (presumably under the rosette), while rhoptries are not docked on the apical vesicle. Inset: magnified images of the vesicles. The results are representative of one experiment. ICM, pair of intraconoidal microtubules. h, Schematic of exocytosis in Alveolata, outlined by the presence of a rosette at the exocytosis site. In Ciliata, organelle discharge can have a defensive or predatory function. Rhoptry exocytosis in Apicomplexa is critical for host cell invasion. In Apicomplexa, but not in Ciliata, the apical vesicle, present between the tip of the rhoptry and the plasmalemma, is plausibly involved in the injection of rhoptry proteins into the host cell. The alveolar sac (AS) in Ciliata is homologous to the inner membrane complex (IMC) of Apicomplexa. PM, plasma membrane.

Full size image

Our work breaks ground on the molecular and structural mechanisms for rhoptry exocytosis in Apicomplexa. We defined the apical rosette of IMPs as the site for rhoptry exocytosis in Apicomplexa, and we characterized an Alveolata-specific Nd complex necessary for the assembly of the rosette. In Paramecium, physiological studies predicted PtNd6 to be active at the plasma membrane24 while PtNd9 appeared to be a diffusible cytoplasmic component interacting with both trichocyst and plasma membranes41,42. However, both the localization and identity of partners in Ciliata remained unknown. We localized TgNd9 to the cytoplasm and TgNd6 to the site of exocytosis in Toxoplasma. We found both proteins to form a complex that included TgNdP1 and TgNdP2, which we demonstrated to be essential for organelle exocytosis in both Ciliata and Apicomplexa. This complex further includes proteins with C2 domains (TgFER2 and TgNdP2), a homologue of the membrane fusion ferlin family (FER2)13, a GTPase (TGME49_277840) and a putative GDP/GTP exchange factor protein (TgNd6). These proteins and their domains yield a regulatory model in which calcium signalling and nucleotide binding and hydrolysis constitute key steps that control rosette assembly and/or organelle discharge (for more discussion, see the Supplementary Text).

In this study, we also shed light on the enigmatic apical vesicle. Described in Toxoplasma22,30, it is also visible in early micrographs from other apicomplexan parasites20,43,44 (Extended Data Fig. 10). We have shown that the apical vesicle is on one side tightly connected with the plasma membrane via the rosette, and from the other sits on the tip of the rhoptry, precluding a direct link and fusion between the rhoptry and the plasma membrane. In Ciliata, where no comparable apical vesicle is present, the docking and fusion of the trichocysts to the plasma membrane shapes the rosette. Diverging from this scenario, we have also shown that the docking of the rhoptries in Toxoplasma is dispensable for rosette formation (Fig. 4e–g), which suggests that the assembly of the rosette in Toxoplasma might instead be induced by docking of the apical vesicle.

While Paramecium trichocysts are discharged into the environment to thwart predators, apicomplexan rhoptries translocate their contents directly into the cytoplasm of host organisms to infect and parasitize. The difference in terms of the presence or absence of the apical vesicle (Fig. 4h) suggests that the apical vesicle is an adaptation in Apicomplexa for parasitism and cell invasion. Its presence may reflect additional complexity of the secretory machinery in apicomplexans, in which exocytosis must be coupled with injection of rhoptry content through the barrier of a host cell membrane. In support of this hypothesis, a similar vesicle is present at the apex of Perkinsus marinus (previously named Dermocystidium marinum)45. This organism possesses rhoptries and invades and parasitizes the cells of oysters. Phylogenetically, Perkinsus falls between Dinoflagellata and Apicomplexa and is viewed as a basal taxon and an example of early adaptation to parasitism.

The rosette and the set of Nd proteins highlight a common ancestry for the fusion machinery linked to secretory organelles in two groups of protists that diverged hundreds of millions of years ago and have adopted radically different lifestyles. The architecture and molecular composition of the rhoptry system that enables the intracellular parasitism of Apicomplexa is now revealed. Future studies combining high-resolution structural biology approaches with the genetics and cell biology of Toxoplasma will allow a full understanding of how this fascinating mechanism unfolds and is regulated. Such an understanding may find application in designing strategies against malaria, cryptosporidiosis and toxoplasmosis.

Methods

Parasite immunofluorescence microscopy

IFAs on intracellular T. gondii parasites were conducted as previously described46. Briefly, cell monolayers were washed and fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) for 20 min. After three washes with PBS, the cells were permeabilized with 0.1% (vol/vol) Triton X-100 or with saponin 0.1% (vol/vol) (for invading parasites) in PBS for 5 min, blocked with 10% foetal bovine serum in PBS for 45 min, incubated with primary antibodies diluted in 2% foetal bovine serum in PBS, washed three times and then incubated with secondary antibody. The coverslips were mounted onto microscope slides using Immu-Mount (Calbiochem).

For IFAs on P. falciparum, thin blood smears of highly synchronized dimethyl sulfoxise (DMSO) and rapamycin-treated Pfnd9 iKO schizonts were air-dried, fixed in 4% (wt/vol) PFA and permeabilized in 0.1% (vol/vol) Triton X-100. After blocking in 1.5% bovine serum albumin (BSA), samples were probed for 1 h with the suitable antibodies: rabbit anti-PfAMA1 antibody (1:1,000)47; mouse anti-PfMSP1 antibody (1:1,000)48; and mouse anti-PfRAP2 antibody (1:500)49. The secondary antibodies used were Alexa Fluor 488 and 594-conjugated antibodies against mouse or rabbit immunoglobulin (highly cross-adsorbed), both diluted 1:4,000 (Molecular Probes). Samples were then counterstained with Hoechst and slides were mounted with Vectashield antifade mounting medium.

Except when specified, observations were performed with a Zeiss Axioimager Z1 epifluorescence microscope equipped with a Zeiss AxioCam MRm CCD camera and a 100×/1.4 Oil Plan Apochromat objective. Images were processed using ZEN Blue 2.3 Pro (Zeiss) software. Imaging of TgNd6, TgNd9, TgNdP1 and TgNdP2 co-localization with the rhoptry protein TgARO (Figs. 1 and 3) was performed on a ZEISS confocal LSM 880 equipped with an Airyscan detector and a 63×/1.4 Oil Plan Apochromat objective. ZEN Black (Zeiss) was used for image Airyscan processing. Z-stack images and z-projection images were denoised, adjusted in brightness and contrast and coloured using the program Fiji50.

Adjustments for brightness and contrast were applied uniformly on the entire images and, when required, matching pairs of images were recorded with the same exposure time and processed identically. All optical images were collected at the Montpellier Ressources Imagerie (MRI) facility of the University of Montpellier (www.mri.cnrs.fr).

Invasion assays

Toxoplasma plaque assays, replication assays and two-colour invasion assays were performed as previously described35,51. Briefly, to synchronize invasion, freshly egressed tachyzoites (5 × 106), pre-treated for 72 h ± ATc (tetracycline) (Tgnd9 iKD and Tgndp1 iKD) or 24 h ± IAA (indole-3-acetic acid or auxin) (Tgnd6 iKD and Tgndp2 iKD), were harvested and settled on ice for 20 min on a monolayer of human foreskin fibroblasts grown on coverslips in 24-well plates. Invasion was allowed for 5 min at 38 °C and stopped by fixation with 4% PFA in Hank’s Balanced Salt Solution (HBSS) for 20 min at room temperature. To detect extracellular parasites, immunodetection was performed with the mouse monoclonal antibody T4 1E5 anti-SAG1 (dilution 1:1,000) in 2% foetal calf serum/HBSS, without previous permeabilization. After permeabilization with 0.01% saponin for 15 min, a second IFA was performed using rabbit anti-ROP1 antibody (dilution 1:1,000) to label the parasitophorous vacuole of intracellular parasites. Extracellular and intracellular parasites were counted by microscopic examination of at least 20 fields per coverslip (n = 3 coverslips). The graphs (Figs. 1d and 3c) show the mean of three independent invasion assays.

Microneme secretion assay

Microneme secretion in T. gondii was assayed by monitoring release into the culture medium. Freshly egressed T. gondii tachyzoites, pre-treated for 72 h ± ATc (Tgnd9 iKD and Tgndp1 iKD) or 24 h ± IAA (Tgnd6 iKD and Tgndp2 iKD), were harvested by centrifugation at 600g and washed twice in intracellular buffer (5 mM NaCl, 142 mM KCl, 1 mM MgCl2, 2 mM EGTA, 5.6 mM glucose and 25 mM HEPES (pH 7.2)) prewarmed to 37 °C. Parasites were resuspended in Dulbecco’s Modified Eagle’s Medium (supplemented with 2 mM glutamine) ± 500 µM propranolol and incubated at 37 °C for 20 min to induce microneme secretion. Parasites were pelleted at 1,000g for 5 min at 4 °C, washed once in PBS and stored at −20 °C. Supernatants were centrifuged at 2,000g for 5 min at 4 °C and used as excreted/secreted antigen. Pellets and excreted/secreted antigen samples were analysed for micronemal protein (AMA1) by western blot. Microneme secretion in P. falciparum was assayed by monitoring the surface translocation of AMA1. Highly synchronized DMSO and rapamycin-treated Pfnd9 iKO schizonts were exposed to 10 µM E64 to block egress. IFAs were then performed with anti-PfAMA1 antibodies to analyse AMA1 translocation to the merozoite surface.

Rhoptry secretion assays

To evaluate the efficiency of rhoptry secretion in T. gondii, we expressed in the Tgnd iKD and Tgndp iKD lines a rhoptry secretion reporter protein consisting of the rhoptry protein toxofilin fused with Cre recombinase, a nuclear localization signal and a myc tag, collectively called secreted Cre, epitope-tagged (SeCrEt)36. The reporter strains generated were called Tgnd9 iKD_SeCrEtUPRT, Tgndp1 iKD_SeCrEtUPRT and Tgnd6 iKD_SeCrEtHXGPRT. DsRed cells, constitutively expressing DsRed and able to switch to enhanced green fluorescent protein (GFP) expression upon Cre-mediated recombination, were used as reporter cells36. DsRed cells were maintained at <50% confluence (regardless of the flask size) and plated the day before parasite infection at a density of 2 × 105 cells per ml (in T25 flasks). Tachyzoites pre-treated for 72 h ± ATc (Tgnd9 iKD and Tgndp1 iKD) or 24 h ± IAA (Tgnd6 iKD and Tgndp2 iKD) were collected and used to infect the DsRed cells at a multiplicity of infection of 3. At 24 h post-invasion, infected DsRed cells were trypsinized and examined by fluorescence-activated cell sorting using FACSDiva software (BD Biosciences) to assess the percentage of cells expressing DsRed (rhoptry secretion impairment) and GFP (successful rhoptry secretion). The number of GFP-positive cells was normalized as a percentage compared with 100% in the control (−IAA or −ATc).

Additionally, rhoptry secretion has also been quantified by classical e-vacuole assay35. Freshly egressed parasites, pre-treated for 72 h ± ATc (Tgnd9 iKD and Tgndp1 iKD) or 24 h ± IAA (Tgnd6 iKD), were preincubated with 1 µM cytochalasin D (cytD) for 10 min and then incubated with human foreskin fibroblast cells in the presence of cytD for 15 min. IFAs were then performed with anti-ROP1 (rhoptry secretion) and anti-SAG1 (parasite surface), and the number of ROP1 stainings per field was determined by microscopic examination of at least 20 fields per coverslip (n = 3 coverslips).

To quantify rhoptry secretion in P. falciparum, the same number of DMSO- and rapamycin-treated Pfnd9 iKO purified schizonts were arrested with 1.5 µM C2 for 4 h before being washed twice to allow them to egress. Parasites were then incubated for 30 min in complete medium with 1 µM cytD in the presence of red blood cells (RBCs). IFAs were then performed with anti-PfRAP2 antibodies to visualize rhoptry secretion events (spits of RAP2 export into the RBC). The number of secretion events was counted over the total RBCs by microscopic examination of 11 fields (~3,000 events analysed).

Freeze-fracture analysis

Cells used for the freeze-fracture analysis were harvested, pelleted and fixed in 2.5% glutaraldehyde in 0.1 M PBS at room temperature for 2 h. Following primary fixation, the samples were rinsed in the same buffer and kept overnight in 30% glycerol/0.1 M PBS. The cells were then quickly frozen by immersion in liquid nitrogen under a vacuum. The frozen samples were fractured in a Bal-Tec BAF060 apparatus. The fractured surface was shadowed by evaporating platinum 45° (∼3.2 nm thick) and carbon 90° (∼25 nm thick). The replicas were then washed in 6.5% sodium hypochlorite, rinsed in a chloroform solution (2:1, v/v), rinsed in distilled water and finally mounted on copper grids. Two different surfaces were taken in consideration when analysing the freeze-fracture technique replicas: the E-face (ectoplasmic) and the P-face (protoplasmic)52. The E-face is the portion of the lipid bilayer associated with the exterior of the cell or extracellular face. The P-face is the portion associated with the interior of the cell.

In vitro growth assay of P. falciparum parasite asexual development

The growth capability of Pfnd9 iKO mutant parasites was assessed by comparing it with the p230p DiCre parental line53 ± rapamycin in biological triplicates. DMSO (vehicle control) and rapamycin-treated cultures were synchronized at ring stages by sorbitol synchronization and parasitaemia adjusted to 0.2%, to follow growth over three cycles. Parasites around 30 h post-invasion were collected at each cycle to determine parasitaemia by counting Giemsa-stained parasites on at least 1,000 RBCs per culture.

P. falciparum-induced egress and time-lapse microscopy

P. falciparum egress was imaged using C2 compound to tightly synchronize egress. DMSO- and rapamycin-treated Pfnd9 iKO parasites were highly synchronized and blocked 40 h post-invasion (mature schizonts) with 1.5 µM C2 compound to prevent egress. After 40 h, the parasites were washed twice with warm RPMI 1640 medium (Gibco) to remove C2 and placed into a 35 mm dish (MatTek). The parasite suspension was sealed by adhering a 22 mm × 22 mm square coverslip to the dish and the preparation was introduced on a temperature-controlled microscope stage at 37 °C. Brightfield images were collected 5 min after washing off the C2 at 1 s intervals over a total of 15 min using a ZEISS Axio Observer microscope fitted with a CoolSNAP HQ2 digital camera and 63×/1.4 Oil Plan Apochromat objective. Images were exported to MOV videos using ZEN 2 (blue edition) software. The number of egress events in 15 min was normalized as a percentage compared with 100% in the control.

Tetrahymena dibucaine assay to quantify mucocyst secretion

Tetrahymena cells were grown to stationary phase (106 cells per ml) in 25 ml SPP (2% proteose peptone, 0.1% yeast extract, 0.2% dextrose and 0.003% ferric-EDTA supplemented with 250 μg ml−1 penicillin G, 250 μg ml−1 streptomycin sulfate and 0.25 μg ml−1 amphotericin B fungizone) for 48 h, then concentrated by centrifugation into a loose 1.5 ml pellet. Cells were stimulated with 2.5 mM dibucaine, vigorously mixed for 30 s and diluted to 15 ml with 10 mM HEPES (pH 7.5) and 5 mM CaCl2. After gently mixing, the culture was centrifuged at 1,200g for 2 min, resulting in the formation of a cell pellet/flocculent bilayer. Quantification of mucocyst secretion was done by weighing the flocculent layer overlying the pellet of cells.

Transmission electron microscopy

Tgnd9 iKD and Tgaro iKD, together with the Δku80-TATi parental strain, were treated with ATc for 72 h. Tgnd6 iKD parasites and the Δku80-Tir1 parental strain were treated with IAA for 24 h. Extracellular parasites were collected in the culture medium after natural egress and fixed by adding and equal volume of 0.1 M PBS containing 5% glutaraldehyde for 1 h at room temperature. Parasites were then centrifuged and resuspended in 1 ml fresh buffer with 2.5% glutaraldehyde for 2 h before being kept at 4 °C until further processing. Each of the following steps was performed in suspension followed by centrifugation steps in a tabletop microcentrifuge. Samples were post-fixed with 1% OsO4 and 1.5% potassium ferricyanide in 0.1 M PBS for 1 h at room temperature, washed in water and then incubated in 2% uranyl acetate in water overnight at 4 °C. Tachyzoites were then dehydrated in a growing concentration of acetonitrile, followed by impregnation in epon 118/acetonitrile at a ratio of 50:50 for 2 h, 90:10 for two additional hours and then pure epon overnight. Pellets were polymerized in fresh epon for 48 h at 60 °C. Ultrathin sections (70 nm) were cut with a Leica ultracut system (Leica Microsystems) and counterstained with uranyl acetate and lead citrate.

Immunoelectron microscopy

TgNd6–HA3-infected fibroblast monolayers were trypsinized and fixed with equal volumes of 8% formaldehyde in PBS overnight at 4 °C and resuspended in 4% fresh formaldehyde until further processing. Cells were then incubated with 0.1% glycine in PBS, pelleted and embedded in 12% gelatin, cut into small blocks (<1 mm) and infused in 2.3 M sucrose on a rotating wheel for 24 h at 4 °C. Gelatin blocks were mounted on specimen pins and frozen in liquid nitrogen. Cryo-sectioning was performed on a Leica UC7 cryo-ultramicrotome and 70 nm cryosections were picked-up in a 1:1 mixture of 2.3 M sucrose and 2% methylcellulose in water and stored at 4 °C. For on-grid immunodetection, grids were floated on 2% gelatin in PBS for 30 min at 37 °C to remove the methylcellulose/sucrose mixture, then blocked with 1% fish-skin gelatin (Sigma–Aldrich) in PBS for 5 min. Successive immunolabelling steps were performed on drops as follows: (1) rat monoclonal anti-HA antibodies (clone 3F10; Roche) in 1% BSA; (2) rabbit polyclonal anti-rat immunoglobulin antibody (Sigma–Aldrich) in 1% BSA; and (3) Protein A gold (UMC Utrecht) in 1% BSA. Four washes (2 min each) with 0.1% BSA were performed between steps. After Protein A treatment, the grids were washed four times for 2 min each with PBS, fixed for 5 min in 1% glutaraldehyde in water and washed six times for 2 min each with distilled water. The grids were then incubated with 2% methylcellulose:4% uranyl acetate at a ratio of 9:1 on ice in the dark for 15 min, picked-up on a wire loop and air-dried.

All chemicals were from Electron Microscopy Sciences and all solvents were from Sigma–Aldrich. Observations and image acquisition were performed on a Jeol 1200 EXII transmission electron microscope at the Electron Microscopy Platform of the University of Montpellier (http://mea.edu.umontpellier.fr). Transmission electron microscopy images were processed with Fiji for contrast optimization, and the Image J plugin was used to make the electron microscopy panels.

Cryo-ET

Freshly isolated T. gondii cells were suspended in HBSS along with fiducials (10 nm colloidal gold from Ted Pella for alignment of tilt series). A 4 µl drop of the suspension was applied onto electron microscopy grids, excess liquid was blotted away and the suspension was plunge frozen in a liquid ethane/propane mixture (pre-cooled using liquid nitrogen) using an electron microscopy GP2 automatic plunger (Leica Microsystems)54. The blotting chamber was set to 95–100% relative humidity at 37 °C and blotting was done either from the sample side of the grid or from the back using Whatman filter paper (No. 1). Plunge-frozen grids were subsequently loaded into AutoGrid cartridges (Thermo Fisher Scientific). Electron microscopy cartridges containing frozen grids were stored in liquid nitrogen and maintained at ≤−170 °C throughout storage, transfer and cryo-ET imaging.

Cryo-ET data collection was performed on a Thermo Fisher Scientific Krios G3i 300 keV field emission cryogenic transmission electron microscopy gun equipped with a 6k × 4k K3 direct electron detector (Gatan) at the Beckman Center for Cryo-Electron Microscopy in the Singh Center for Nanotechnology, University of Pennsylvania. The camera was operated in electron-counting mode to enable motion correction. An energy filter (Gatan Imaging Filter; Gatan) with a slit width of 20 eV was used to increase the contrast of the projection images. Additionally, a Volta phase plate was used to boost the image contrast at defocus values of −1 to –3 µm. A magnification of 33,000× with corresponding pixel sizes of 2.65 Å was used for imaging. SerialEM software55 was used for all imaging. Cells were first assessed at lower magnifications for suitability of ice thickness and outer membrane integrity (the parasites sometimes tend to vesiculate their outer membrane, possibly in response to blotting). Once the target cells were identified and marked, anchor maps were used to revisit these locations and collect tilt series in an automated fashion. Each tilt series was collected from −60° to +60° (bidirectionally from 0°), with increments of 2°, in an automated fashion using the low-dose functions of tracking and focusing. The cumulative dose of each tilt series ranged between 100 and 150 e− Å−2. Once acquired, tilt series were binned twice or four times before being aligned using the 10 nm colloidal gold as fiducials and reconstructed into tomograms by our in-house automated computation pipeline utilizing the IMOD software package56. For presentation purposes, tomograms were favourably oriented in three dimensions before averaging a few slices around the slice of interest (to enhance contrast) using the slicer window in IMOD. In a few cases (Fig. 4b–e), the contrast was further enhanced using the nonlinear anisotropic diffusion filter in IMOD.

Immunopurification and mass spectrometry analysis

Immunopurification was performed using anti-HA magnetic beads (Pierce; R88836) as previously described57. Purified proteins were loaded on a sodium dodecyl sulfate polyacrylamide gel electrophoresis system and digested in gel (two bands per sample) as previously described58. Samples were loaded onto a 25 cm reversed-phase column (75 mm inner diameter; Acclaim PepMap 100 C18; Thermo Fisher Scientific) and separated with an UltiMate 3000 RSLC system (Thermo Fisher Scientific) coupled to a Q Exactive HFX system (Thermo Fisher Scientific). Tandem mass spectrometry analyses were performed in a data-dependent mode. Full scans (375–1,500 m/z) were acquired in the Orbitrap mass analyzer with a resolution of 60,000 at 200 m/z. For the full scans, 3 × 106 ions were accumulated within a maximum injection time of 60 ms. The 12 most intense ions with charge states ≥2 were sequentially isolated (1 × 105) with a maximum injection time of 45 ms and fragmented by higher-energy collisional dissociation in the collision cell (normalized collision energy of 28%) and detected in the Orbitrap analyzer at a resolution of 30,000.

Raw spectra were processed with MaxQuant59 using standard parameters with label-free quantification (LFQ) and matched between runs using Andromeda60. Tandem mass spectrometry spectra were matched against the UniProt reference proteomes of T. gondii and humans (proteome IDs UP000001529 and UP000005640, respectively) and 250 frequently observed contaminants, as well as reversed sequences of all entries (MaxQuant contaminant database). Statistical analyses were done using Perseus on LFQ data61.

Amino acid sequence alignments and phylogenetic analyses, cell and parasite culture, parasite cloning strategies, parasite transfections, parasite immunoblots, T. gondii plaque assays, T. gondii intracellular growth assays, T. gondii immunofluorescence-based induced egress assays, T. gondii conoid extrusion assays, T. gondii gliding assays, T. gondii attachment assays, Ciliata culture conditions, Ciliata biolistic transformation, generation of the Tetrahymena knockout strain, Ciliata immunoblots, Ciliata immunofluorescence microscopy, statistical analysis, reagents and antibodies are found in the Supplementary Methods.

Reporting Summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.