Rhoptries are club-shaped, regulated secretory organelles that cluster at the apical pole of apicomplexan parasites. Their discharge is essential for invasion and the establishment of an intracellular lifestyle. Little is known about rhoptry biogenesis and recycling during parasite division. In Toxoplasma gondii, positioning of rhoptries involves the armadillo repeats only protein (ARO) and myosin F (MyoF). Here, we show that two ARO partners, ARO-interacting protein (AIP) and adenylate cyclase β (ACβ) localize to a rhoptry subcompartment. In absence of AIP, ACβ disappears from the rhoptries. By assessing the contribution of each ARO armadillo (ARM) repeat, we provide evidence that ARO is multifunctional, participating not only in positioning but also in clustering of rhoptries. Structural analyses show that ARO resembles the myosin-binding domain of the Caenorhabditis elegans myosin chaperone UNC-45. A conserved patch of aromatic and acidic residues denotes the putative MyoF-binding site, and the overall arrangement of the ARM repeats explains the dramatic consequences of deleting each of them. Finally, Plasmodium falciparum ARO functionally complements ARO depletion and interacts with the same partners, highlighting the conservation of rhoptry biogenesis in Apicomplexa.
Unicellular pathogens belonging to the phylum Apicomplexa represent a major threat to both human and animal health. The most notorious member of this phylum is the causative agent of human malaria, Plasmodium falciparum. Other members of considerable medical and veterinary importance include Toxoplasma, Neospora, Cryptosporidium, Babesia, Eimeria and Sarcocystis. These organisms are unified by phylum-specific cytoskeletal structures and sets of specialized secretory organelles termed micronemes, rhoptries and dense granules, discharge of which is necessary for the establishment of an obligate intracellular life cycle (reviewed in Carruthers and Sibley, 1997; Sibley, 2010). This cycle is initiated by host cell penetration; an active process that leads to the formation of a replication-permissive niche inside a specialised membranous sac termed the parasitophorous vacuole membrane (PVM). Rhoptries are positioned at the apical end of the parasite and adopt an elongated, club-shaped morphology with a bulbous body and a narrow neck (Dubey et al., 1998). These organelles play a key role not only during invasion, but also contribute to the formation of the PVM as well as subsequent evasion and subversion of the host cell defence mechanisms (reviewed in Kemp et al., 2013). During parasite entry, only one of the 10–12 rhoptries injects its contents, including the rhoptry neck proteins (RONs), rhoptry bulb proteins (ROPs) and some membranous materials, into the host cell (Boothroyd and Dubremetz, 2008). Several RONs form a complex with apical membrane antigen 1 (AMA-1), a protein secreted from the micronemes, resulting in the formation of the so-called moving junction (Besteiro et al., 2011; Straub et al., 2009). The moving junction is a tight apposition between the parasite and the host cell plasma membrane, and ensures efficient propulsion of the parasite into the host cell, ultimately leading to the formation of the PVM (Besteiro et al., 2011; Straub et al., 2009). Several ROPs migrate either to the lumen of the nascent parasitophorous vacuole, the PVM, or into the cytosol or nucleus of the infected host cell where they modulate cellular functions (Butcher et al., 2011; El Hajj et al., 2007a; Etheridge et al., 2014; Fleckenstein et al., 2012; Yamamoto et al., 2009). The signalling events leading to rhoptry secretion are poorly understood. However, it is known that rhoptry discharge occurs shortly after microneme discharge, and the microneme protein MIC8 has been implicated in rhoptry exocytosis in Toxoplasma gondii (Kessler et al., 2008).
During parasite division, rhoptries are formed de novo, first appearing as globular vesicles in close proximity to the Golgi complex, and then subsequently maturing into club-shaped organelles localizing to the parasite apex (Fig. 1B) (Nishi et al., 2008; Shaw et al., 1998). The cellular and molecular mechanisms underlying rhoptry biogenesis, including their targeting to the apical pole, are currently not well understood. Previous studies have, however, shown that perturbation of T. gondii proteins involved in vesicular trafficking, such as the dynamin-related protein B (DrpB), sortilin-like receptor (SORTLR), clathrin heavy chain 1 (CHC1) or the RabGTPases Rab5A and Rab5C result either in a block of both rhoptry and microneme biogenesis and/or in mistargeting of the proteins to either of the two organelles (Breinich et al., 2009; Kremer et al., 2013; Pieperhoff et al., 2013; Sloves et al., 2012). The T. gondii armadillo repeats only protein (ARO) is a key factor ensuring the apical distribution of rhoptries (Beck et al., 2013; Mueller et al., 2013). Anchoring of ARO to the cytosolic face of the rhoptry membrane is dependent upon acylation (Cabrera et al., 2012). Of the 18 members of the protein acyltransferases (PATs) identified in T. gondii, only one, DHHC7, is localized to the rhoptries and found to be responsible for palmitoylating ARO (Beck et al., 2013; Frenal et al., 2013). It is currently unclear at which stage during rhoptry biogenesis this palmitoylation event takes place. Upon conditional depletion of ARO, rhoptries are randomly dispersed within the parasite cytosol, ablating rhoptry secretion and hampering host cell invasion (Beck et al., 2013; Mueller et al., 2013). Co-immunoprecipitation experiments and subsequent mass spectrometry analyses have revealed that ARO interacts with myosin F (MyoF). This myosin belongs to the alveolate-specific class XXII and participates in apicoplast inheritance and centrosome positioning (Jacot et al., 2013). The functional disruption of MyoF and the use of cytochalasin D, an inhibitor of actin polymerization, interfere with the positioning of rhoptries, supporting the view that this is an actomyosin-based process (Jacot et al., 2013; Mueller et al., 2013).
In analogy to the well-studied actomyosin-dependent movement of organelles in yeast and melanocytes, we anticipated that additional accessory proteins would assist MyoF functioning (Hume and Seabra, 2011; Weisman, 2006). In this context, two proteins were identified to co-immunoprecipitate with ARO; a hypothetical protein with unknown function named ARO-interacting protein (AIP) and adenylate cyclase β (ACβ). Although not highly conserved, AIP is present in several apicomplexan genomes. Two distinct adenylate cyclases (ACα and ACβ) have been described in Plasmodium falciparum and are present and conserved across the Apicomplexa phylum (Baker, 2004). Plasmodium berghei ACα has been reported to mediate apically-regulated exocytosis in Plasmodium sporozoites through cAMP signalling, whereas the function of the apparently essential ACβ is unknown (Ono et al., 2008).
Here, we show that T. gondii ACβ is recruited to the rhoptry surface through interactions with ARO and that, together with AIP, they represent the first identified markers of a third sub-compartment separating the rhoptry bulb and neck (Lemgruber et al., 2010). AIP is a prerequisite for the targeting of ACβ to the rhoptry.
Depletion of ARO prevents the apical translocation of rhoptries and hence hampers further assessment of its potential contribution to anchoring the organelles to the tip of the parasite or to rhoptry discharge. To circumvent this limitation, we assessed the importance of each predicted armadillo (ARM) repeat of ARO by exploiting the functional complementation of the inducible knockout strain (ΔAROe/AROi–Ty; hereafter termed ARO-iKO) with ARO deletion mutants. These non-functional mutants revealed a new role for ARO in holding rhoptries in bundles, and also led to the detection of an extended structure that is most likely of membranous origin given that it is only labelled with markers of the rhoptry membrane. Structural analyses show that ARO closely resembles the myosin-binding domain of Caenorhabditis elegans UNC-45 and has a highly conserved binding groove that would most likely accommodate an extended polypeptide chain. Finally, the cross-genera complementation of the functions assigned to ARO by P. falciparum ARO highlights the conservation of these aspects of rhoptry biogenesis across the phylum.
T. gondii ACβ is a rhoptry surface protein that binds to ARO
ACβ was previously identified, through mass spectrometric analyses, as a potential partner of ARO in co-immunoprecipitation experiments using a GFP-tagged copy of ARO (Mueller et al., 2013). In order to confirm this interaction, we first determined the subcellular distribution of ACβ by inserting three Ty epitope tags at the C-terminus of the protein through single homologous recombination at the endogenous locus in the ΔKU80 strain (ACβ–3Ty). In parallel, recombinant ACβ (amino acids 1246–1600) was used to generate polyclonal anti-ACβ-specific antibodies in rabbits (anti-ACβ). Western blot analyses of lysates from the wild-type RH strain and ACβ–3Ty-expressing strain using anti-ACβ sera detected a strong band approximating the predicted size of 220 kDa that was co-migrating with the band detected by anti-Ty antibodies in ACβ–3Ty parasites (Fig. 1A). Fainter lower bands were also detected that likely correspond to degradation products, as well as non-specific background bands. Indirect immunofluorescence assay using anti-Ty and anti-GAP45 antibodies revealed that ACβ is apically distributed and that the anti-ACβ staining was identical to the anti-Ty staining. Co-labelling with anti-ARO and anti-ROP2 markers identified ACβ as a rhoptry neck protein (Fig. 1B).
In order to validate the interaction of ACβ with ARO, we transiently transfected ACβ–3Ty parasites with ARO–GFP–Ty or a control construct expressing GFP. Parasites were harvested 24 h later and co-immunoprecipitation utilising the GFP-Trap system was performed. Western blot analyses with anti-Ty and anti-GFP antibodies showed that ARO–GFP–Ty and not GFP pulled down ACβ–3Ty, confirming that ACβ is indeed associated with ARO (Fig. 1C).
Previous studies have indicated that T. gondii ACβ is closely related to P. falciparum ACβ and both belong to the family of soluble adenylate cyclases (Baker, 2004). Fractionation experiments confirmed that T. gondii ACβ is a soluble protein (Fig. S1A), and a proteinase K protection assay showed that ACβ does not reside in the rhoptry lumen, but, like ARO, is present at the surface of the organelle facing the parasite cytosol (Fig. S1B). To determine whether the presence of ACβ at the rhoptries is dependent upon its association with ARO, ARO-iKO parasites were treated for 48 h with anhydrotetracycline (ATc) to deplete AROi–Ty, and the localization of ACβ was assessed by immunofluorescence using anti-ACβ antibodies. In the absence of ARO, ACβ no longer localized to the neck of the dispersed rhoptries (Fig. 1D).
AIP, which has been described previously as a rhoptry neck protein (Mueller et al., 2013), perfectly colocalized with ACβ (Fig. 1E). Upon closer examination, confocal images of ACβ revealed that the protein does not cover the entire neck portion but instead occupies a distinct region of the neck adjacent to the rhoptry bulb, as shown by an indirect immunofluorescence analysis carried out using the anti-ACβ and anti-RON9 antibodies (Lamarque et al., 2012) (Fig. 1F). Interestingly, it has been reported in previous transmission electron microscopy (TEM) studies that there are three sub-compartments of rhoptries delimited by a dark and electron-dense neck, an amorphous and less electron-dense bulb, and a region of intermediate electron density, which connects the bulb to the neck (Lemgruber et al., 2010). Taken together, these results demonstrate that ARO interacts with ACβ and that, together with AIP, these two ARO partners constitute the first known markers restricted to the surface of an intermediate sub-compartment of the rhoptries.
AIP and ACβ localize to the rhoptries only in the presence of ARO
More insight into the association of AIP and ACβ at the rhoptry surface was provided by a conventional disruption of the AIP gene (AIP-KO) generated by double homologous recombination within the regions flanking the coding sequence. The genotype of the clones was confirmed by genomic PCR (Fig. S1C; Table S1). Immunofluorescence analyses of this strain using anti-ARO, anti-ROP2 and anti-ACβ antibodies revealed that (1) ARO was still found at the rhoptries, (2) the organelles were properly localized to the parasite apex and (3) ACβ was no longer detected at the rhoptry neck (Fig. 2A). Moreover, western blot analysis revealed that ACβ was reduced to undetectable levels in AIP-KO lysates, indicating that ACβ is unstable in absence of AIP. In contrast, the level of ARO remained constant between AIP-KO and wild-type parasites (RHΔHX; Fig. 2B). Functional complementation of AIP-KO by transient transfection of a construct encoding Myc-tagged AIP with the tubulin promoter (T8) (T8AIP–Myc) returned ACβ to the rhoptries in the presence of T8AIP–Myc but not in the negative control transiently expressing T8ARO–Myc (Fig. 2C). These results confirm that AIP is necessary both for stabilization of ACβ and its targeting to the rhoptries. AIP does not appear to harbour any transmembrane domains (TMDs) or acylation motifs that would confer membrane association. Fractionation experiments with lysates containing AIP–3Ty confirmed that this protein, like ACβ was fully soluble in PBS (Fig. S1D).
To further investigate whether the presence of AIP and ultimately ACβ at the rhoptries is dependent upon the association with ARO, we stably introduced a second copy of C-terminally Myc-tagged AIP controlled by the tubulin promoter in the ARO-iKO strain (ARO-iKO/T8AIP–Myc). Immunofluorescence analyses revealed that the presence of T8AIP–Myc on the rhoptries was abolished upon AROi–Ty depletion. Furthermore, the expression of T8AIP–Myc in the absence of AROi–Ty failed to position ACβ on the dispersed rhoptries (Fig. 2D). Taken together, ARO likely interacts directly with AIP whereas ACβ requires AIP to localize to the rhoptry intermediate compartment.
Each predicted armadillo repeat contributes to ARO function in rhoptry positioning
Earlier studies predicted that ARO comprises two proper ARM repeats (ARM3 and ARM4 in Fig. 3A) and at least three ‘degenerate’ ARM repeats (ARM2, ARM5, ARM6 in Fig. 3A), preceded by an N-terminal region responsible for membrane association (Cabrera et al., 2012). To address the importance of each of the predicted ARM repeats for ARO function, we generated deletion mutants for each of the ARM repeats 2–6 individually as well as both of the ARM3 and ARM4 domains together (ΔARM3,4–Myc). The mutants were assessed for functional complementation by stable expression in the ARO-iKO strain (Fig. 3A). Immunofluorescence analyses carried out with ARO-iKO expressing ΔARM3,4–Myc (ARO-iKO/ΔARM3,4–Myc) grown with or without ATc for 48 h demonstrated that AROi–Ty remained regulatable by ATc, whereas ΔARM3,4–Myc was constitutively expressed (Fig. 3B). ΔARM3,4–Myc localized to the rhoptries in the absence of ATc but failed to complement the phenotype in the presence of ATc. Rather unexpectedly, this truncated form of ARO did not colocalize with the ROP2 marker in the absence of AROi–Ty. In other words, in the presence of ATc there was no colocalization between this truncated form and the dispersed organelles, but instead it stained an extended membranous structure that has not been previously observed (Fig. 3B,C; Fig. S2A). Of importance is the finding that upon ATc treatment of this strain, ACβ was no longer detectable at the rhoptry neck (Fig. 3C). Western blot analyses indicated that constitutive expression of ΔARM3,4–Myc did not significantly reduce the level of ACβ following ATc treatment suggesting that the protein became cytosolic (Fig. S2B). Similarly, all individual ARM deletion mutants labelled this unusual structure and failed to complement for the absence of ARO leading to the organelle dispersion phenotype, and with the exception of one mutant lacking ARM6, to the delocalisation of ACβ to the cytosol (data not shown). Importantly, the membranous structure was not only detected by expression of the different ARM-repeat-truncated proteins but also in the ATc-treated ‘parental’ ARO-iKO strain that lacks ARO. In this case, detection of the structure was achieved by transiently expressing a Myc-tagged form of T. gondii carbonic anhydrase 1 (CAH1–Myc). CAH1 was originally identified in the rhoptry proteome (Bradley et al., 2005) and used here as the only other marker beside ARO known to specifically label the rhoptry membrane. CAH1 possesses hydrophobic segments at each extremity and was tagged in the central region by single homologous recombination at the endogenous locus, which was confirmed by western blotting (Fig. S2C–E). CAH1–Myc colocalized with rhoptry markers in immunofluorescence analyses and the central region between the two hydrophobic termini was found to be exposed to the cytosolic face of the organelle similarly to ARO, based on proteinase K protection assay (Fig. S2F–G). The membranous structure was only detectable in the absence of ARO (i.e. in ATc-treated ARO-iKO parasites) using the rhoptry membrane marker CAH1 (Fig. 3D) and not by ER or mitochondrial markers (data not shown), thus, suggesting that it is of rhoptry origin.
In line with the misplacement of rhoptry organelles, plaque assays revealed that deletion of the ARM motifs results in non-viable parasites (Fig. 3E). There was, however, a subtle difference between the ARO-iKO/ΔARM3,4–Myc strain that formed no plaques when grown in the presence of ATc compared to another mutant, ARO-iKO/ΔARM6–Myc, which formed very tiny plaques.
ARO is implicated in the clustering of the rhoptry organelles
Only the deletion mutant of ARM6 showed a distinct phenotype compared to the other mutants. ΔARM6–Myc failed to bring the rhoptries to the apical pole; however, the phenotype strongly contrasted with that seen in ARO-iKO parasites or ARO-iKO parasites expressing ΔARM3,4–Myc. The organelles were not dispersed but commonly (up to 90%) remained in bundles (Fig. 4A–C). The preserved clustering of the rhoptries was more clearly evident from the TEM analyses performed on parasites expressing ΔARM6–Myc grown in the presence of ATc for 48 h (Fig. 4B). As observed with the other deletion mutants, ΔARM6–Myc localized to the membranous structure in the presence of ATc, while AROi–Ty was still tightly regulated by ATc (Fig. S3A,B). Furthermore, ΔARM6–Myc was also the only mutant wherein ACβ was still present at the intermediate compartment of the mislocalized organelles (Fig. 4C). Although this phenomenon was clearly and repeatedly observed by immunofluorescence, immunoprecipitation experiments with lysates from cells expressing ΔARM6–GFPTy and ACβ–3Ty did not pull down ACβ (Fig. 4D). Besides this discrepancy, ΔARM6–Myc, like the other mutants, failed to properly position the rhoptries, which we assumed to be caused by a lack of a productive interaction with MyoF. However, rather unexpectedly, ΔARM3,4–GFPTy and ΔARM6–GFPTy still interacted with MyoF–3Ty based on immunoprecipitation experiments, whereas GFPTy alone was unable to bind to MyoF (Fig. 4E). Taken together, the functional dissection of ARO revealed that each individual ARM repeat is required for correct functioning of ARO in bringing the rhoptries to the apical pole. It appears that only the conventional conformation of ARO allows the establishment of a functional unit with MyoF, whereas just binding to MyoF does not require the presence of all ARM motifs. In addition, the C-terminal ARM6 is dispensable for the separate function of ARO in the maintenance of rhoptry clustering.
Importantly, the expression of ΔARM3,4 and ΔARM6 as second copies in wild-type parasites (RHΔHX) or in the ARO-iKO strain not treated with ATc, exhibited a dominant-negative effect that resulted in a partial mislocalization of the rhoptry organelles (Fig. S3C–E). The fact that these parasites are viable indicates that endogenous ARO ensures that at least one rhoptry is properly positioned at the apical end to assist in invasion. This is indeed the case, as shown by TEM analyses of the ARO-iKO strain expressing ΔARM3,4–Myc or ΔARM6–Myc in the absence of ATc, in which some but not all of the rhoptries are clearly dispersed (Fig. S3F,G).
ARO shares a conserved myosin-binding fold with UNC-45
To elucidate the folding and organization of the predicted ARM repeats in ARO, we used a combination of small-angle X-ray scattering (SAXS) and homology modelling to reconstruct the 3D structure of ARO in solution. The independently constructed homology and SAXS models both show that ARO is a globular, monomeric protein with a maximum intramolecular distance of ∼9 nm, comprising at least five ARM repeats and an additional N-terminal segment that also resembles an ARM repeat and might prove to be such in the structure (Fig. 5A–C). The homology model and the SAXS ab initio model can be superimposed nearly perfectly (Fig. 5B), and both show very good fits (χ2 values 1.27 and 1.43, respectively) to the measured scattering curve (Fig. 5D,E).
The ARM repeats of ARO form a right-handed super-helix, and ARM repeats 2–6 superimpose very well on the five C-terminal ARM repeats of the C. elegans myosin chaperone UNC-45 (Gazda et al., 2013; Lee et al., 2011) (Fig. 5A,B). The third helices of ARM repeats 2–6 line up to form a shallow groove (Fig. 6A,B; Fig. S4A,B), which in UNC-45 is implicated in myosin binding (Gazda et al., 2013; Lee et al., 2011). This groove is ideally suited for binding an extended polypeptide chain, and is likely to do so in UNC-45 (Barral et al., 2002; Gazda et al., 2013), although it has also been suggested that a folded part of the myosin motor domain could bind this motif (Fratev et al., 2013). A similar binding groove is also present in human importin α7 where it accommodates the extended tail of the influenza PB2 nuclear localization domain (Fig. S4B) (Pumroy et al., 2015). The groove in ARO is lined by several aromatic residues (Fig. 6C), and its deepest part is highly negatively charged (Fig. 6D). The importance of this surface for ligand binding is also demonstrated by the high degree of conservation of the aromatic and acidic residues along the groove (Fig. 6E).
ARO has previously been shown to depend on acylation of the N-terminal glycine and two cysteine residues, as well as basic residues in the N-terminal α-helix, for membrane attachment (Cabrera et al., 2012). The solution model suggests that the N-terminal ARM-like repeat does not form a continuum of the superhelix with the five proper C-terminal ARM repeats (Figs 5B and 6). In this arrangement, all the suggested membrane-interacting residues point out of the core structure and would allow for sufficient degrees of freedom for simultaneous binding of multiple ligands to the five-ARM-repeat core (Fig. 6A).
Trans-genera complementation of ARO within the phylum Apicomplexa
ARO is uniquely conserved amongst Apicomplexa with 63% identity and 79% similarity at the amino acid level between P. falciparum and T. gondii (Fig. 5A). Given that P. falciparum ARO has been shown to localize to the rhoptries of intraerythrocytic schizonts after acylation, we postulated that the protein fulfils a shared function across the phylum (Cabrera et al., 2012). In light of the findings reported here, it became pertinent to determine whether P. falciparum ARO could interact with the same partners as T. gondii ARO and accomplish its identified functions. Transient transfection of a vector expressing a synthetic and codon-optimized P. falciparum ARO cDNA (Fig. S4C) in wild-type (RH) and ARO-iKO parasites confirmed that P. falciparum ARO localizes to the rhoptries of T. gondii (Fig. 7A,B). Upon ATc-mediated depletion of AROi–Ty, P. falciparum ARO complemented both the apical positioning of the organelles and the correct targeting of ACβ to the rhoptries (Fig. 7B). To determine whether P. falciparum ARO also rescues rhoptry secretion and complements the invasion defect, a stable line of ARO-iKO parasites expressing P. falciparum ARO was generated (ARO-iKO/T8PfARO). As observed in transient experiments, upon AROi–Ty depletion the rhoptries were no longer dispersed but correctly attached apically when P. falciparum ARO was stably expressed, and no dominant-negative effect was detected. Plaque assays confirmed that P. falciparum ARO fully rescues the invasion phenotype in the absence of AROi–Ty (Fig. 7C).
Invasion and subversion of host cellular functions are two events that crucially depend upon the release of the rhoptry contents (Carruthers and Boothroyd, 2007; Kemp et al., 2013). Although these two functions have been extensively studied, there is a plethora of rhoptry-related questions that are yet to be fully addressed, e.g. what are the steps in rhoptry biogenesis and how or why are they club-shaped? How do proteins traffic to the rhoptries and how are they segregated into the specifically delineated organelle sub-compartments? How is the membranous material accumulated and organized within the organelles? How are these organelles anchored at the parasite apex and maintained in clusters? What are the triggers stimulating rhoptry secretion and what are the physical changes allowing such a large organelle to inject its contents within seconds? And finally, how are the rhoptry organelles from the mother recycled during parasite division?
Isolation and characterization of the rhoptry proteome and lipidome has been reported for T. gondii, Plasmodium spp. and Eimeria tenella (Besteiro et al., 2008; Blackman and Bannister, 2001; Bradley et al., 2005; Etzion et al., 1991; Leriche and Dubremetz, 1991; Oakes et al., 2013; Sam-Yellowe et al., 2004). These studies have been instrumental in developing a more comprehensive understanding of the function of this highly specialized organelle and have also served in approaching the questions stated above with specific molecular tools. Although many RON and ROP proteins, especially ROP kinases and pseudokinases have been (and continue to be) identified and characterized (Peixoto et al., 2010), not many proteins present at the surface of the rhoptries have been reported to date. These proteins, which are not secreted during invasion, are likely to play important roles in morphology, signal sensing or attachment of the organelles to the parasite cytoskeleton. Identified proteins localizing to the surface of rhoptries through TMDs or acylation motifs include: (1) NHE2, a non-essential sodium hydrogen exchanger with 12 predicted TMDs and a potential role in pH regulation and osmotolerance, (2) ARO, which is anchored to the rhoptries by N-terminal myristoylation and palmitoylation and plays an essential role in rhoptry positioning and hence invasion, (3) DHHC7, an essential, four TMD-containing PAT that is responsible for the palmitoylation of ARO and (4) a putative carbonic anhydrase (CAH1; TGME49_297070) predicted to contain a TMD or GPI anchor signal and shown here to localise to the surface of the rhoptry organelle (Beck et al., 2013; Frenal et al., 2013; Karasov et al., 2005; Mueller et al., 2013).
This study reports a thorough dissection of ARO function, aiming to understand how this protein mediates apical rhoptry positioning. A detailed scheme recapitulating our findings is presented in Fig. 8. Based on combined biochemical and genetic evidence, we have established that ACβ interacts with AIP, which in turn binds directly or indirectly to ARO. Both proteins are the first known markers of a morphologically defined intermediate compartment of the rhoptries that separates the neck from the bulb (Lemgruber et al., 2010). It is unclear at this point what restricts the two proteins to this sub-compartment. The complementation experiment with ΔARM6–Myc suggests that the C-terminus of ARO might not be involved in the interaction with ACβ, given that the latter was still present on the mislocalized rhoptries in AROi–Ty-depleted parasites as assessed by immunofluorescence analysis. Albeit indirectly, this also indicates that AIP is still able to bind to the ΔARM6–Myc mutant on the rhoptry neck. Concordantly, the C-terminal ARM repeat is markedly less conserved than the preceding repeats (Figs 5A and 6E), and the aromatic patch that could mark a binding surface for protein ligands is located on a face remote from the C-terminal ARM repeat (Fig. 5C). However, immunoprecipitation experiments did not confirm the immunofluorescence observations, given that ΔARM6–Myc did not pull down ACβ. The different experimental outcomes might be explained by the fact that two different strains were used (ΔARM6–Myc versus ΔARM6–GFP–Ty), or that other settings such as lysis and immunoprecipitation conditions might have been too harsh to preserve this particular possibly suboptimal protein–protein interaction. Although the function of AIP is still unclear, the stability of ACβ is intimately associated to the presence of AIP because ACβ is undetectable in AIP-KO parasites.
The functional complementation utilised a series of deletion mutants for the five C-terminal ARM motifs that were aimed at dissecting the multiple putative roles of ARO. Importantly, none of the mutant constructs were able to mediate apical rhoptry positioning, suggesting that each domain contributes to the overall folding and structure of ARO, which is a prerequisite for the recruitment of MyoF. This is understandable from the folding of the five C-terminal ARM repeats into a superhelix containing a highly acidic and conserved groove important for ligand binding. In analogy to other homologous ARM-repeat proteins, such as importin α7 (Pumroy et al., 2015), it seems likely that this groove would bind an extended polypeptide chain. Notably, all ARO-binding partners identified so far (MyoF, AIP and ACβ) are large multi-domain proteins that contain long disordered or extended stretches of amino acids. By contrast, homology modelling suggests that the globular WD40 domain at the C-terminus of MyoF has a highly basic surface (data not shown). Such a positively charged folded protein could also be a putative ligand for the highly acidic ARO surface groove. In yeast, Vac8p is an armadillo-repeat containing acylated protein involved in vacuolar membrane dynamics, e.g. vacuole inheritance and vacuolar membrane fusion (Fleckenstein et al., 1998; Wang et al., 1998). No structure of Vac8p is known to date, but homology modelling suggests a structure similar to importin α7, the UNC-45–CRO1–She4p (UCS) domain of UNC-45 and ARO. Vac8p also interacts with the actin cytoskeleton (Wang et al., 1998). Thus, it seems that the molecular mechanisms behind vacuole inheritance in higher eukaryotes and rhoptry positioning in Apicomplexa might be conserved.
A striking observation that emerged from analysing the ARM deletion mutants was their localization to membranous structures. In fact, these structures were clearly apparent in both wild-type and ARO-iKO parasites expressing the non-functional ARM deletion mutants, but also and importantly in ARO-iKO parasites treated with ATc using a rhoptry membrane marker for their detection. These structures are likely of membranous origin given that they were only detectable with rhoptry membrane markers [non-functional ARM deletion mutants, 20ARO–GFP (Cabrera et al., 2012) and CAH1–Myc]. We envision that these rhoptry-derived membranes originate from the improper recycling of mislocalized mature rhoptries from the mother cell during daughter cell formation. If this holds true, this would also suggest that the lipid fraction and rhoptry contents might be degraded and/or recycled by distinct mechanisms. To date, the machinery involved in the fast recycling of rhoptries is unknown. Intriguingly, T. gondii does not possess bonafide lysosomes, and autophagy has not been associated with the recycling of organelles that are produced de novo during the cell cycle (Besteiro, 2012).
Deletion of the C-terminal motif in ΔARM6–Myc turned out to be informative, as this deletion does not prevent the remaining ARM units from folding into a stable structure that retains partial functionality. Although this mutant failed to rescue the apical rhoptry localization and invasion defect, the organelles were no longer dispersed. TEM analyses clearly showed that the 10–12 rhoptries remained neatly clustered together, as observed in wild-type parasites, but were not targeted apically and hence were still unable to discharge their contents during invasion. These findings strongly suggest that ARO not only positions the rhoptries to the apical pole, but also maintains them in bundles. It is intriguing that ΔARM6–Myc is also the only mutant wherein the localization of ACβ to the rhoptries is not affected, at least as judged by immunofluorescence. In light of this, it is tempting to speculate that the intermediate compartment is ideally placed to position the machinery required to maintain the organelles in clusters.
Given the panoply of ARO partners and its implication in positioning, clustering and possibly recycling of rhoptries, it became relevant to determine whether the same complexity was preserved across the phylum of Apicomplexa. When expressed in T. gondii, P. falciparum ARO–Myc localized to the rhoptries, indicating that the PAT DHHC7 is capable of recognizing this heterologous substrate. Additionally, this successful trans-genera functional complementation implies that T. gondii MyoF can functionally interact with P. falciparum ARO. Furthermore, P. falciparum ARO is also able to interact with T. gondii ACβ because it brings this protein back to the rhoptries and to the appropriate sub-compartment. These findings demonstrate that many processes related to rhoptry biogenesis and discharge are conserved across the phylum, and in consequence, we can capitalize on comparative analyses between members of the Apicomplexa to tackle the many remaining open questions regarding this fascinating organelle.
MATERIALS AND METHODS
Parasite transfection and selection of clonal stable lines
T. gondii RH strains lacking HXGPRT or KU80 (RHΔHX or ΔKU80) were used as parental strains (Huynh and Carruthers, 2009). Parasites with tetracycline-controlled gene expression were regulated by 0.5 μg ml−1 anhydrotetracycline (ATc) (Meissner et al., 2002). T. gondii transfections were performed as previously described (Soldati and Boothroyd, 1993). ACβ–3Ty parasites were generated by transfecting ΔKU80 parasites with 50 μg of linearized (BglII) ACβ–3Ty vector (Fox et al., 2009; Huynh and Carruthers, 2009). Selection of transgenic parasites was performed with mycophenolic acid (MPA) and xanthine for HXGPRT (HX) selection and cloning was through serial dilution (Donald et al., 1996; Soldati et al., 1995). The AIP-KO strain was generated by transfecting p2855-HXGPRT-5′3′TgAIP into ΔKU80, which was then subjected to MPA and xanthine selection, and cloned by serial dilution. The previously generated parasite line ARO-iKO was transfected with 50 μg of the NotI linearized plasmid pT8AIP-Myc-BLE and selected with 30 μg ml−1 of phleomycin and then cloned by serial dilution (Mueller et al., 2013). ARO-iKO was used for transfection of SacI linearized plasmids pT8ΔARM3-Myc-BLE, pT8ΔARM4-Myc-BLE, pT8ΔARM3,4-Myc-BLE, pT8ΔARM2-Myc-BLE, pT8ΔARM5-Myc-BLE and pT8ΔARM6-Myc-BLE (50 μg plasmid). Transgenic parasites were selected with 30 μg ml−1 of phleomycin and subsequently cloned by serial dilution (Mueller et al., 2013). RHΔHX was transfected with 50 μg of SacI-linearized pT8ΔARM3,4-Ty-HX or pT8ΔARM6-Ty-HX. After selection with MPA and xanthine, parasites were cloned by serial dilution. ARO-iKO strains expressing T8PfARO-Myc-BLE were generated by transfecting 50 μg of SacI-linearized plasmid pT8PfARO-Myc-BLE. After selection with 30 μg ml−1 of phleomycin, the transgenic parasites were cloned by serial dilution. Plaque assays were performed as previously described (Mueller et al., 2013).
Cloning of DNA constructs
TaKaRa Ex Taq DNA polymerase (Clontech) and Phusion high-fidelity DNA polymerase (NEB) were used for all PCRs, with the primers listed in Table S1. The ACβ–3Ty-HX plasmid was generated by amplification of tachyzoite genomic DNA using the primer set 4189 and 4190. The PCR product as well as pT8-TgMIC13-3Ty-HX was digested with KpnI and NsiI restriction enzymes and the PCR product was subsequently ligated into the digested vector (Friedrich et al., 2010). For recombinant ACβ antibody production, the C-terminal coding part of ACβ was amplified by PCR using the primers 4461 and 4358. The fragment was digested with NcoI and SpeI and cloned into the pETHTb 6xHis expression vector, which had also been digested with these restriction enzymes. The plasmid pT8AIP-Myc-BLE was generated by amplifying the AIP open reading frame by performing two PCR reactions using primer sets 4275 and 4276 (product 1), and 4277 and 4278 (product 2). Product 1 was digested with MfeI and NsiI and product 2 with NsiI. Both products were then ligated simultaneously into the EcoRI- and NsiI-digested plasmid pT8ARO-Myc-BLE (Mueller et al., 2013). P2855-HXGPRT was digested with KpnI and XhoI for insertion of AIP 5′ UTR, and amplified with primers 4433 and 4434 that were also digested with KpnI and XhoI. Following digestion of the resulting plasmid with XbaI and NotI and insertion of the AIP 3′ UTR, amplified with primer set 4435 and 4436 and also digested with XbaI and NotI, the plasmid p2855-HXGPRT-5′3′TgAIP was obtained. For deletion of the different armadillo motifs, full-length ARO cDNA was amplified with primers 3796 and 3740 and cloned into pGEM-T Easy Vector (Promega). The entire plasmid was amplified with following primer sets: (1) 4192 and 4193 for deletion of ARM3; (2) 4194 and 4195 for deletion of ARM4; (3) 4192 and 4195 for deletion of ARM3+4; (4) 4639 and 4640 for deletion of ARM2; and (5) 4641 and 4642 for deletion of ARM5. Resulting amplified plasmids were first digested with NheI, and afterwards ligated and then re-digested with MfeI and NsiI and ligated into EcoRI/NsiI-digested pT8ARO-Myc-BLE. For deletion of ARM6, pT8ARO-Myc-BLE was used as a template to amplify a PCR product with the primers 3796 and 4643 that was subsequently digested with MfeI and NsiI and ligated into the EcoRI- and NsiI-digested pT8ARO-Myc-BLE. PT8ΔARM3,4-Ty-BLE and pT8ΔARM6-Ty-BLE were digested with KpnI and NsiI and subcloned into the KpnI- and NsiI-digested vector pT8ARO-Ty-HX, thereby generating pT8ΔARM3,4-Ty-HX and pT8ΔARM6-Ty-HX. Synthesis of the codon-optimized P. falciparum ARO sequence was carried out by Invitrogen. The synthetic sequence was amplified with the primers 4931 and 4932 and the product, as well as the plasmid pT8ARO-Myc-BLE, was digested with EcoRI and NsiI, which allowed ligation of the product into the digested plasmid. Cloning of the GST–His–ARO in the bacterial expression vector was carried out using pET42aTEV into which the full-length cDNA of ARO was introduced using NcoI and XhoI restriction sites. ARO cDNA was amplified using the primers 4820 and 4821. The CAH1–Myc–HX plasmid was generated by amplification of tachyzoite gDNA using the primer sets 4586 and 4587, and 4588 and 4589. The PCR product from 4586 and 4587 was digested with KpnI and BglII restriction enzymes, the PCR product from 4588 and 4589 was digested with BglII and PacI, and the vector pT8-TgARO-GFP-Ty-HX was digested with KpnI and PacI. The two PCR products were subsequently ligated into the digested vector.
Recombinant 6xHis–CtACβ protein was purified from E. coli BL21 strain and used as antigen to raise polyclonal antibodies against ACβ protein in rabbits. This was done according to standard protocols by Eurogentec. The polyclonal ARO antibodies were as described previously, used at 1/1000 dilution (Mueller et al., 2013). The monoclonal antibodies against the Ty tag BB2 (1/5 dilution), Myc tag 9E10 (1/5 dilution), actin (1/5 dilution), SAG1 T4-1E5 (1/1000 dilution), RON9 2A7 (1/1000 dilution), ROP1 T5-1A3 (1/2000 dilution), ROP2 T3-4A7 (1/2000 dilution), ROP5 T5-3E2 (1/1000 dilution), as well as the polyclonal antibodies anti-GAP45 (1/3000 dilution), anti-CAT and the anti-recombinant ROP1 (1/2000 dilution) and ROP2 (1/2000 dilution) rabbit sera, were previously described (Brecht et al., 2001; El Hajj et al., 2007b,, 2008; Leriche and Dubremetz, 1991; Plattner et al., 2008; Lamarque et al., 2012). For western blot analysis, secondary peroxidase-conjugated goat anti-rabbit-IgG and anti-mouse-IgG antibodies (Molecular Probes, Paisley, UK) were used. For immunofluorescence analysis, the secondary antibodies Alexa-Fluor-488-conjugated goat anti-rabbit IgG antibodies and Alexa-Fluor-594-conjugated goat anti-mouse-IgG antibodies (Molecular Probes, Invitrogen) were used.
Freshly egressed parasites were pelleted after complete host cell lysis. Parasites harbouring a tet-inducible copy of ARO (AROi–Ty) were grown for 48 h with or without 0.5 μg ml−1 ATc before harvesting. SDS-PAGE, semi-dry transfer to nitrocellulose and proteins visualized using ECL system (Amersham Corp) were preformed as described previously (Mueller et al., 2013). Proteinase K protection assays were performed as previously described by Cabrera et al. (2012) and analysed by western blotting using anti-Myc, anti-ARO, anti-ROP2-4 and anti-profilin antibodies.
Immunofluorescence assay and confocal microscopy
Human foreskin fibroblasts (HFF) seeded on coverslips in 24-well plates were inoculated with freshly released parasites. After two to four rounds of parasite replication, cells were fixed with 4% paraformaldehyde (PFA) or 4% PFA and 0.05% glutaraldehyd in PBS depending on the antigen to be labelled, and processed as previously described (Plattner et al., 2008). Confocal images were generated with a Zeiss confocal laser scanning microscope (LSM700) using a Plan-Apochromat 63× objective with NA 1.4. All the images taken are maximal projections of confocal stacks spanning the entire parasites.
Transmission electron microscopy
ARO-iKO parasites expressing either ΔARM3,4–Myc or ΔARM6–Myc that had undergone or not a 24 h pre-treatment with ATc, were used to infect a monolayer of HFF and grown with or without ATc for 24 h. The infected host cells were fixed with 2.5% glutaraldehyde in 0.1 M phosphate buffer pH 7.4, post-fixed in osmium tetroxide, dehydrated in ethanol and treated with propylene oxide prior to embedding in Spurr's epoxy resin. Thin sections were stained with uranyl acetate and lead citrate prior to examination using a Tecnai 20 electron microscope (FEI Company). Multiple thin sections from each of the two sample preparations were examined by the electron microscope.
Freshly released ACβ–3Ty and AIP–3Ty parasites were harvested and their pellets resuspended in PBS or PBS containing 1% Triton X-100 or 1 M NaCl. Following five freeze-thaw cycles using liquid N2 and a 37°C water bath, the pellet (P) and soluble fractions (S1) were separated by centrifugation at 16,000 g for 45 min at 4°C. The pellet that was initially resuspended in PBS, and then resuspended in 0.1 M Na2CO3 (pH 11), incubated for 30 min at room temperature, and the pellet and soluble (S2) fraction were separated by centrifugation at 16,000 g at room temperature for 10 min. The different samples were subsequently analysed by western blotting. As control, the solubility of CAT and GAP45 were tested.
Freshly released tachyzoites were harvested, washed in PBS, and lysed in co-immunoprecipitation buffer [PBS containing 0.2% (v/v) Triton X-100, 150 mM NaCl, protease inhibitor cocktail (Roche)] and incubated on ice for 20 min. After centrifugation at 16,000 g for 45 min at 4°C, the supernatant was subjected to immunoprecipitation using anti-GFP lama antibodies (GFP-Trap_M Magnetic Agarose Beads from ChromoTek). Beads were washed in a co-immunoprecipitation buffer with 300 mM NaCl.
Small-angle X-ray scattering and homology modelling
GST–His–ARO was expressed in E. coli BL21(DE3) cells at 37°C for 4 h using 1 mM IPTG for induction. The cell pellet was resuspended in lysis buffer [20 mM HEPES pH 7.5, 300 mM NaCl, 20 mM imidazole, 10 mg ml−1 DNase I, 0.1 mg ml−1 lysozyme, an EDTA-free protease inhibitor cocktail (Roche) and 5 mM β-mercaptoethanol], and the cells were lysed using sonication. The clarified supernatant was loaded onto Ni-NTA matrix pre-equilibrated with the lysis buffer without DNase I, lysozyme and protease inhibitors, and binding was carried out for 1 h at 4°C. The column was subsequently washed three times with 20 ml of buffer containing 20 mM HEPES, 300 mM NaCl, 20 mM imidazole, and 5 mM β-mercaptoethanol, and the protein was eluted with 300 mM imidazole in the lysis buffer. Imidazole was removed from the sample using a PD10 desalting column, and the fusion protein was subjected to TEV digestion overnight at 4°C. The cleaved GST–His tag and any uncleaved fusion protein were removed by passing the sample through a second Ni-NTA column. The unbound ARO sample was concentrated and loaded onto a Hiload 16/60 Superdex 75 column pre-equilibrated with buffer containing 20 mM Tris-HCl pH 8.0, 300 mM NaCl, 5% glycerol and 1 mM tris (2-carboxyethyl)phosphine. Fractions containing pure ARO were concentrated, flash frozen in liquid nitrogen, and stored at −70°C until use.
Small-angle X-ray scattering (SAXS) data on ARO at concentrations of 6.1, 3.1, and 1.5 mg ml−1 were collected at the EMBL beamline P12 at PETRA III/DESY, Hamburg. Analysis of the data was carried out using the ATSAS package (Petoukhov et al., 2012). As there seemed to be no concentration-dependent effects on the scattering curves, the highest concentration data were used for further modelling. An ab initio model of ARO was built using GASBOR (Svergun et al., 2001). A homology model was constructed based on the ARO sequence using the Phyre2 server (Kelley et al., 2015). The ab initio and homology models were superimposed both using SUPCOMB (Kozin and Svergun, 2001) as well as manually in Pymol. The fit of the homology model to the experimental scattering curve was calculated using CRYSOL (Svergun et al., 1995). The electrostatic surface potential was calculated using APBS (Baker et al., 2001) in Chimera (Pettersen et al., 2004) and the conservation plot using ConSurf (Ashkenazy et al., 2010).
We thank Peter Bradley and Maryse Lebrun for providing the anti-RON and ROP antibodies, Susanne Meier for technical assistance, Jean-Baptiste Marq and Jean François Dubremetz for fruitful discussions and assistance with TEM. The pET42aTEV and purification expertise were provided by Dr Stephane Thore. Hayley Bullen edited the manuscript. We acknowledge the support from the beamline staff at the EMBL beamline P12 at PETRA III/DESY, Hamburg, Germany.
C.M., P.-M.H. and N.K. performed the experiments reported in Figs 1, 2, 3, 4, 7 and Figs S1–S4. A.S., J.P.K. and I.K. performed the experiments and structural analyses reported in Figs 5, 6 and Fig. S4. C.M., D.S.-F. and I.K. conceived and designed the experiments and wrote the manuscript.
This work was supported by the Swiss National Science Foundation [grant number FN3100A0-116722]; the German Ministry of Education and Research (BMBF) [grant number 0313927 (to I.K.)]; and the Academy of Finland [grant numbers 257537 and 265112 (to I.K.)]. D.S.-F. is an International Scholar of the Howard Hughes Medical Institute. C.M. was supported by the Japanese–Swiss Science and Technology Program and the Fondation Ernst et Lucie Schmidheiny. A.S. has been supported by the Higher Education Commission Pakistan (HEC) through the University of Sindh; and J.P.K. by the Paulo Foundation.
The authors declare no competing or financial interests.