Inverted topology of the Toxoplasma gondii ROP5 rhoptry protein provides new insights into the association of the ROP2 protein family with the parasitophorous vacuole membrane

Authors


*E-mail dubremet@univ-montp2.fr; Tel. (+33) 467 143 455; Fax (+33) 467 144 286.

Summary

Toxoplasma gondii, as many intracellular parasites, is separated from the cytosol of its host cell by a parasitophorous vacuole membrane (PVM). This vacuole forms during host cell invasion and parasite apical organelles named rhoptries discharge proteins that associate with its membrane during this process. We report here the characterization of the rhoptry protein ROP5, which is a new member of the ROP2 family. Contrasting with what is known for other ROP2 family proteins, ROP5 is not processed during trafficking to rhoptries. We show here that ROP5 is secreted during invasion and associates with the PVM. Using differential permeabilization of infected cells, we have shown that ROP5 exposes its C-terminus towards the host cell cytoplasm, which corresponds to a reverse topology compared with ROP2 and ROP4. Taken together with recent modelling data suggesting that the C-terminal hydrophobic domain hitherto described as transmembrane may correspond to a hydrophobic helix buried in the catalytic domain of kinase-related proteins, these findings call for a reappraisal of the current view of ROP2 family proteins association with the PVM.

Introduction

The protozoan phylum Apicomplexa includes a large number of human and animal parasites, responsible for a wide variety of diseases such as malaria, toxoplasmosis, coccidiosis, cryptosporidiosis, etc. The common feature of its members is the presence of an apical complex that plays an essential role in parasite entry into the host cell. Toxoplasma gondii is a worldwidely distributed Apicomplexa remarkable for its ability to invade almost any nucleated cell of warm-blooded animals, including men (Dubey and Beattie, 1988). While the infection in humans is usually asymptomatic in healthy adults, it may cause neurological diseases upon congenital infection or in immunocompromised individuals (Luft et al., 1993; Martin, 2001).

Host cell invasion by T. gondii is a rapid event (< 10 s) during which the parasite forcibly indents and modifies the host cell plasma membrane, thereby creating the parasitophorous vacuole (PV) in which it develops. The molecular mechanisms of host cell entry remain poorly understood. Three types of organelles named micronemes, rhoptries and dense granules discharge their contents during host cell invasion (Dubremetz et al., 1993; Carruthers and Sibley, 1997). The initial discharge of micronemal proteins from the anterior end of the parasite is associated with gliding and attachment to the host cell (Fourmaux et al., 1996; Carruthers et al., 2000; Brecht et al., 2001). Then microneme and rhoptry neck proteins form a moving junction with the host cell plasma membrane that propels the parasite within the developing parasitophorous vacuole (Alexander et al., 2005; Lebrun et al., 2005). Subsequently, rhoptry bulb proteins (ROP proteins) become associated with the parasitophorous vacuole membrane (PVM) (Saffer et al., 1992; Beckers et al., 1994; Carey et al., 2004; El Hajj et al., 2005). Once the parasite is within the vacuole, GRA proteins are released in the PV (Dubremetz et al., 1993; Carruthers and Sibley, 1997).

Among rhoptry proteins, the ROP2 family (Sadak et al., 1988; Leriche and Dubremetz, 1991; H. El Hajj, submitted) was defined after a prototype, the ROP2 protein, that was shown to be inserted in the PVM during invasion, exposing its N-terminal (Nt) end in the host cell cytosol (Beckers et al., 1994). The Nt domain of ROP2 was later shown to interact with the mitochondrial import machinery and to mediate the tight association between host mitochondria and PVM (Sinai and Joiner, 2001). Targeted depletion of ROP2 using a ribozyme-modified antisense RNA strategy resulted in disruption of rhoptry biogenesis,impairment of cytokinesis, reduction in the association of host cell mitochondria with the PVM, reduced capacity of parasites to invade and replicate in human fibroblasts, and attenuation of virulence in mice (Nakaar et al., 2003).

The importance of ROP2 and the fact that the parasite is synthesizing simultaneously several related proteins suggest that ROP2 family proteins serve crucial functions; yet, the apparent indispensability of ROP2 suggests they may not complement one another and raises questions on their individual role. Several other members of the family have been characterized more recently. The 60 kDa ROP4 protein associates with the PVM where it becomes phosphorylated on multiple sites, likely partially resulting from the action of host cell protein kinase(s) (Carey et al., 2004). ROP7, a 57 kDa protein closely related to ROP4 (71% identity), is also translocated in the PVM (El Hajj et al., 2005). Besides, ROP2 family proteins possess a variably degenerated kinase domain that we have recently shown to be functional in some members of the family. Molecular modelling of the kinase-like domain suggests a folding burying into the core of the protein, a hydrophobic segment hitherto considered as a transmembrane domain (H. El Hajj, submitted).

ROP5 (Leriche and Dubremetz, 1991) has been identified as a new member of the family both by protein purification (H. El Hajj, submitted) and rhoptry proteomics (Bradley et al., 2005). We report here its fate during parasite development and upon host cell invasion and demonstrate original features that make it unique among the ROP2 family proteins described so far. Contrasting with ROP2, 4 and 7, ROP5 is not processed during biosynthesis, and although it is targeted to the PVM during invasion, its topology in the PVM appears inverted compared with ROP2. This latter result added to previous structure predictions further questions the transmembrane status of these proteins in the PVM.

Results

ROP5 sequence

The open reading frame deduced from EST Cluster 95059837 (APIDBest) corresponding to the protein purified with monoclonal antibody (mAb) T5 3E2 (H. El Hajj, submitted) was amplified from genomic DNA. This sequence has been cloned independently by Bradley et al. (Bradley et al., 2005) and deposited under accession number DQ116423. The ROP5 primary sequence contains several noteworthy features, including two hydrophobic regions: an Nt signal sequence, with a predicted cleavage site between residues 24 and 25 (Fig. 1) and a putative C-terminal (Ct) transmembrane domain between residues 449 and 470. A putative serine/threonine protein kinase domain in the Ct half of the ROP5 sequence was identified by a prosite search (Hanks and Hunter, 1995), but this domain lacks a conserved aspartic acid in the catalytic loop critical for phosphotransferase activity. ROP5 shares about 25% identity with the ROP2 protein, but its closest relative in the family is the ROP18 protein, which contains all the features requested for kinase activity. The ROP5 sequence has no YXXφ motif within the C-terminal tail, but a dileucine motif is present (515–516) corresponding to the one described as involved in trafficking the ROP2 protein to the rhoptries (Hoppe et al., 2000; Ngo et al., 2003). Molecular modelling of ROP5 suggests that the Ct hydrophobic stretch is an alpha helix buried within the protein (El Hajj et al., 2006).

Figure 1.

Alignment of the ROP5 sequence on ROP2. The putative signal sequences are in lower cases. The predicted cleavage site of ROP2 by SUB2 is italicized. The ROP5-Nt sequence used as a GST fusion to generate anti-ROP5-Nt antibodies is underlined. The mitochondrial targeting signal of ROP2 is underlined. The Ct hydrophobic stretches are in bold faces.

ROP5 is not processed in the secretory pathway

Most rhoptry proteins described so far in T. gondii including members of the ROP2 family are synthesized as pro-proteins that are subjected to proteolytic cleavage removing the Nt pro-region (Sadak et al., 1988; Soldati et al., 1998). The exact site of cleavage for ROP2 family proteins is unknown, but the sequence (SWLE/QE) present near the putative end of the pro-domain of ROP2, ROP4 and ROP8 is suspected to be cleaved by the maturase TgSUB2 (Miller et al., 2003). Interestingly, this sequence is absent in ROP5 (Fig. 1) raising the question of ROP5 processing. To answer this question, we studied the biosynthesis of ROP5 by pulse-chase metabolic labelling with [35S]methionine. After a 20 min pulse, one major labelled protein was immunoprecipitated by mAb T5 3E2 at 60 kDa (Fig. 2). An identical pattern was obtained when a 1 h chase was performed, suggesting that ROP5 does not undergo the post-Golgi proteolytic processing described for the other members of the ROP2 family studied so far, as shown using mAb T3 4A7 (anti-ROP2–ROP4) as a control.

Figure 2.

Pulse-chase analysis of ROP5 biosynthesis and immunoprecipitation. T. gondii-infected fibroblasts were labelled for 20 min with [35S] methionine/cysteine and either harvested (lane 1, 3) or chased for 1 h (lane 2, 4). Then the NP40 lysate was immunoprecipitated with mAb anti-ROP5 (T5 3E2, lane 1, 2) or mAb anti-ROP2-4 (T3 4A7, lane 3, 4). No difference was observed for ROP5 between pulse and chase, whereas the control with mAb T3 4A7 that immunoprecipitates ROP2 (55) and 4 (60) showed the processing of both proteins from higher molecular mass precursors.

ROP5 is secreted during invasion and associates with the parasitophorous vacuole membrane

We studied the fate of ROP5 during host cell invasion by RH tachyzoites. Cytochalasin-D (Cyt-D)-treated tachyzoites can attach to but not enter the host cell. In these conditions, the contents of rhoptries are discharged, leading to the accumulation of vesicles into the host cytosol called the evacuoles that contain rhoptry proteins including ROP1 (Hakansson et al., 2001), ROP4 (Carey et al., 2004) and ROP7 (El Hajj et al., 2005). We stained Cyt-D-arrested parasites with mAb T5 3E2. As previously shown for other ROPs, ROP5 is translocated into the host cell, and is found in the resulting evacuoles (Fig. 3A). Dual labelling showed that the ROP5 and ROP1 colocalized in evacuoles.

Figure 3.

ROP5 is translocated in the parasitophorous vacuole membrane upon invasion. ROP5 colocalizes with ROP2 in (A) evacuoles of cytochalasin-D-arrested tachyzoites, or (B) the parasitophorous vacuole membrane of freshly invaded parasites.

To follow the fate of ROP5 early after invasion, newly infected cells were permeabilized with saponin, which does not damage the parasite cell membrane and therefore allows for detecting selectively the protein exocytosed during invasion (Carruthers and Sibley, 1997). When freshly invaded parasites were reacted with mAb T5 3E2, ROP5 was found in the PVM and colocalized with ROP2 (Fig. 3B).

ROP5 exposes its C-terminus towards host cell cytoplasm, in contrast to ROP2 and ROP4

In order to study the topology of ROP5 in the PVM, we developed tools to detect specifically the Nt or Ct of the ROP5 protein. A GST-NtROP5 recombinant protein encompassing AA 89–284 (Fig. 1) was used to generate mouse polyclonal antibodies against the ROP5 Nt. On Western blot of RH tachyzoites, these antibodies recognized a 59 kDa band, comigrating with the one detected by mAb T5 3E2, specific for ROP5 (Fig. 4A). By immunofluorescence assay (IFA), both antibodies reacted with the apical end of permeabilized tachyzoites and colocalized with ROP2 (Fig. 4B). On freshly invaded and on Cyt-D-arrested parasites, using saponin permeabilization,anti-NtROP5 produced the same patterns as mAb T5 3E2 (not shown).

Figure 4.

Anti-ROP5 Nt antibodies.
A. Antibodies raised against the recombinant ROP5 Nt (lane 1) react on Western blot with a band of 59 kDa comigrating with ROP5, recognized by mAb T5 3E2 (lane 2).
B. By IFA, they colocalize with anti-ROP2 antibodies.

All attempts at identifying the mAb T5 3E2 epitope or at generating antibodies against the Ct of ROP5 failed, which prompted us to create a stable transgenic parasite expressing a Ct myc-tagged ROP5. Western blot analysis of these transgenic parasites using mAb T5 3E2 detected two bands (Fig. 5A) likely corresponding to native ROP5 and ROP5-myc protein. The anti-myc tag mAb recognized only a band comigrating with the upper band (Fig. 5A) whereas anti-NtROP5 serum recognized the same bands as mAb T5 3E2 (Fig. 5A). As a control, we created a stable T. gondii transfectant expressing ROP2 with a Ct myc tag. In this mutant, mAb T3 4A7 (Sadak et al., 1988) recognized three major bands, the upper band corresponding to ROP4 (60 kDa), the lower band at 55 kDa corresponding to native ROP2 and an intermediate band at 57 kDa, corresponding likely to ROP2-myc (Fig. 5A). This latter was indeed the only band detected by the anti-myc antibody (Fig. 5A). By IFA on both ROP2-myc and ROP5-myc tachyzoites, the anti-myc mAb reacted with the apical end of parasites, with perfect colocalization with ROP2. (Fig. 5B). On freshly invaded and on Cyt-D-arrested ROP2-myc or ROP5-myc parasites, using saponin permeabilization, the anti-myc mAb produced the same patterns as mAb T5 3E2 or anti-ROP1 (data not shown). All these results indicated that anti-NtROP5 antibodies and the transgenic T. gondii expressing Ct myc-tagged ROP5 were suitable to study the topology of the protein after secretion in the PVM upon invasion. In streptolysin-O (SLO)-treated infected cells, the host cell plasma membrane was selectively permeabilized without affecting the PVM (Beckers et al., 1994), allowing selective detection of exposed cytosolic domain of PVM-associated protein. In all these experiments, anti-SAG1 antibodies were used to control the integrity of the PVM and to distinguish intracellular parasites (IFA-negative) from the extracellular ones that remained attached on the cells (IFA-positive); polyclonal anti-tachyzoites antibodies that label the external side of the PVM (Beckers et al., 1994) were used as positive control (Fig. 6A). In SLO permeabilized cells, infected by the transgenic ROP5-myc strain, and using mAb anti-myc, we detected a PVM label at 25 min (Fig. 6B), 2 h and 4 h post invasion (not shown). This label was no longer found at 24 h after invasion (not shown). This result strongly suggested that the Ct of ROP5 is exposed to the host cell cytoplasm. No labelling was obtained using anti-NtROP5 at any time post invasion, on RHΔhx or ROP5-myc tachyzoites (Fig. 6B). We used as a control the transgenic ROP2-myc strain. Indeed, ROP2 exposes its Nt to the host cell cytoplasm (Beckers et al., 1994), and this Nt contains the epitope of mAb T3 4A7, in the mitochondrion targeting sequence of ROP2 (Sinai and Joiner, 2001). On samples fixed 20 min after invasion, we obtained a PVM label with mAb T3 4A7 on SLO permeabilized infected cells, but no label with mAb anti-myc (Fig. 6C), confirming the previously reported data. No label was obtained after 2 h, 4 h or 24 h post invasion on SLO permeabilized infected cells with either T3 4A7 or mAb anti-myc (not shown). The disappearance of T3 4A7 label is likely due to the masking of the epitope when the ROP2-Nt is interacting with host cell mitochondrion.

Figure 5.

ROP5 myc and ROP2 myc are targeted to the rhoptries where they colocalize with ROP2.
A. On Western blot, on ROP5-myc transfected parasites, mAb T5 3E2 and α ROP5-Nt recognize both ROP5 (5) and the additional band above native ROP5 (5myc), this latter being the myc-tagged ROP5 detected by anti-myc antibodies. In ROP2-myc transfected parasites, anti-myc antibodies recognize a 57 kDa band corresponding to tagged ROP2 (2myc), whereas native ROP2 (2), tagged ROP2 (2myc) and ROP4 (4) are detected by mAb T3 4A7.
B. By IFA, both ROP5-myc and ROP2-myc colocalize with anti-ROP2 antibodies.

Figure 6.

ROP5 exposes its Ct towards host cell cytoplasm. In cells treated with SLO after invasion, anti-T. gondii polyclonal antibodies detect parasite proteins exposed on the cytoplasmic side of the PVM (A, anti-toxo), whereas anti-T. gondii SAG1 surface protein cannot reach their target (A, anti-SAG1). On ROP5-myc vacuoles (B), anti-myc antibodies detect the Ct of ROP5myc and anti-NtROP5 antibodies fail to react, whereas on ROP2-myc vacuoles (C) mAb T3 4A7 detects the Nt of ROP2 and anti-myc antibodies fail to react.

Discussion

ROP5 is a newly described member of the ROP2 family, characterized at the sequence level by being more distantly related to the ROP2 prototype than the ROP4, 7 and 8 (El Hajj et al., 2006). Yet, the characteristic features of the family (rhoptry location, protein size, kinase homology in the C-terminus, hydrophobic stretch near Ct, basic residues stretch near the Nt) are perfectly conserved in ROP5, and translate into 25% identity with ROP2. What we show here is that ROP5 shows additional original features that make it unique among the family and raises several questions on its role and on the general understanding of the fate and function of the ROP2 family proteins.

Indeed, ROP5 differs from all other members of the ROP2 family studied so far in being synthesized as a mature form, whereas ROP2, 4 and 7, are processed in a post-Golgi compartment (Sadak et al., 1988; Carey et al., 2004; El Hajj et al., 2005). Two rhoptry proteinases have been described that are likely candidates for this cleavage (Que et al., 2002; Miller et al., 2003). ROP5 does not possess the SWLE/QE motif that is the putative TgSUB2 cleavage site present in ROP2, 4 and 7, but instead, there is a 10 AA gap at this very location in the ROP5 sequence. This would be in favour of TgSUB2 being the ROP2 family proteins maturase.

The ROP5 protein is translocated in the PVM at invasion, which was expected as most rhoptry proteins follow this route. What was surprising, however, was the topology of the protein as defined by our differential permeabilization experiments, suggesting that only the Ct of ROP5 is accessible to antibodies on the cytoplasmic side of the PVM, whereas the reverse, i.e. the Nt is found for ROP2. Such a major difference between ROP5 and the other members of the ROP2 family was unexpected. Indeed what has been so far considered as a putative transmembrane stretch is found at the same location, about 100 AA from the Ct (AA 449–470). In addition, the 529–530 dileucine motif of ROP2 that has been shown to be involved in rhoptry targeting (Ngo et al., 2003) is well conserved in ROP5, suggesting a similar topology in the biosynthesis pathway. Therefore, our finding of an inverted topology concerning two proteins showing high homology and likely to be synthesized and trafficked with the same membrane orientation raises questions on what happens upon organelle storage and/or translocation to the vacuole membrane during invasion. So far, there has been no clear explanation for the paradox encountered for many T. gondii organellar proteins that are suggested to be synthesized as integral proteins and to end up after invasion in the vacuole membrane, such as shown for ROP2 and GRA5 (Beckers et al., 1994; Lecordier et al., 1999). In addition, when dealing with the ROP2 family of proteins, the significance of the hydrophobic stretch found c. 100 AA from the C-terminus is questionable, as it is found in the part of the proteins that have kinase homology, and is most likely buried in the tertiary structure of the protein rather than being transmembrane (El Hajj et al., 2006).

A classical way of understanding rhoptry proteins targeting to the PV follows the hypothesis according to which rhoptries may derive from multivesicular endosomes (Ngo et al., 2003). Then a rhoptry protein synthesized as an integral protein could end up in the equivalent of an exosome, and rhoptry exocytosis would be equivalent to exosome release and fusion with the developing PVM. In this case, the topology of the protein in the PVM would be analogous to the one in the biosynthetic pathway, i.e. a Ct towards cytoplasm in the tachyzoite would become a Ct in the host cell cytoplasm. This would fit well with ROP5, but would not with ROP2, for which the mitochondrial interaction signal is in the Nt. The transmembrane hypothesis is also supported by the Ct tyrosine or dileucine residues that would be involved in rhoptry targeting. However, Bradley et al. (Bradley et al., 2004) have shown rhoptry targeting of a green fluorescent protein-tagged truncated ROP4 that had lost the Ct motifs and the Ct putative transmembrane. In addition, we have found that ROP18, another member of the family, also lacks the tyrosine and dileucine Ct residues (H. El Hajj, unpublished). These findings strongly suggest that neither these residues nor the transmembrane topology is needed for ROP2 family protein targeting to the rhoptries.

One possible explanation for the difference of fate between ROP2 and ROP5 might be that the removal of the propeptide has implications in the topology after exocytosis, whatever the targeting mechanism. However, the remarkable conservation of the residues critical for the kinase folding (H. El Hajj, submitted) leads us to propose another hypothesis where what we observe by using differential permeabilization procedures is not a transmembrane topology, but a differential accessibility of part of these proteins that would be translocated in the host cell cytoplasm and would reassociate with the external face of the PVM by some way that may hide part of the protein to antibody binding. A tentative schematic drawing illustrating this hypothesis is proposed in Fig. 7. Experiments such as protease protection would help solving this question, but they have failed so far in our hands. The translocation of ROP5 in evacuoles does not help solving this question, as all rhoptry proteins studied so far are found in these structures, but no data on their topology have been obtained. Therefore, a reappraisal of the ROP2 family protein topology seems to be needed.

Figure 7.

Schematic drawing of two possible explanations of the observations related in this paper. On the right is the transmembrane hypothesis in which our results suggest an inverted topology for ROP2 and ROP5. On the left is the alternate hypothesis where antibody accessibility would make the distinction between the proteins, both of which would be associated on the cytoplasmic side of the PVM.

ROP5 knockouts seem to be lethal, as we could not obtain stable ROP5 KO, and whenever we observed ROP5-deficient vacuoles during these KO experiments, they had a degenerated phenotype suggesting that the parasites were not viable. Conditional mutagenesis should be performed to further characterize this ROP5 requirement.

Host mitochondria and endoplasmic reticulum are recruited and physically tethered to the PVM; ROP2 interact with host cell mitochondria (Sinai and Joiner, 2001), ROP4 is phosphorylated on several serine/threonine residues suggesting that this protein may be a target for host cell and/or parasite kinases (Carey et al., 2004). While host mitochondria are recruited by interacting with ROP2, the process of recruitment of the endoplasmic reticulum around the PVM surrounding internalised parasites remains unknown. In addition, the host cell intermediate filament network is reorganized around the PVM (Halonen and Weidner, 1994) and may therefore interact with cytoplasm-exposed molecules such as ROP5. Last, the intracellular parasite is capable of inhibiting host cell apoptosis (Goebel et al., 1999) and does so by sequestrating IkBalpha at the PVM (Molestina and Sinai, 2005). ROP5 may therefore contribute one or more of these functions at the interface between the parasite and the cytoplasm of its host cell, and further investigations are needed to understand its role in the biology of the parasite.

Experimental procedures

Host cell and parasite cultures

Tachyzoites of the RH strain of T. gondii (Sabin, 1941) and the T. gondii RH hxgprt (RHΔhx) deleted for hypoxanthine guanine phosphorybosyl transferase (HXGPRT) (Donald et al., 1996) were used throughout the study. All parasites were maintained by serial passage in primary human foreskin fibroblasts (HFF) grown in DMEM (GibcoBRL) supplemented with 10% fetal calf serum (FCS) and 2 mM glutamine.

Antibodies

Antibodies used in this study for immunoprecipitation, IFA and/or immunoblot analysis included mAb T5 3E2 (Leriche and Dubremetz, 1991), an anti-myc tag monoclonal antibody (9E10) (Evan et al., 1985), a rabbit anti-T. gondii serum obtained by rabbit infection, an anti-recombinant SAG1 rabbit serum (Harning et al., 1996) (kindly donated by Dr Eskild Petersen) and an anti-recombinant ROP1 and ROP2 rabbit serum (J.F. Dubremetz and O. Mercereau-Puijalon, unpublished).

Cloning procedures and plasmids construct

The ROP5 cloning was based on the EST cluster found in ToxoDB-APIBest (95059837). The ROP5 gene was polymerase chain reaction (PCR)-amplified from genomic DNA with primers HH1 (5′-ATGCAATTGATGGCGACGAAGCTCGCTAGA-3′; MfeI site underlined) and reverse primer HH2 (5′-GCATGCATAGCGACTGAGGGCGCAGCA-3′, NsiI site underlined), and subcloned as a MfeI/NsiI fragment into pUni/V5-His-TOPO (Invitrogen), to give pROP5.

The plasmid pROP5myc was designed to express a Ct myc-tagged ROP5 protein in RHΔhx tachyzoites. It was constructed by inserting the coding sequence of ROP5 without signal peptide in pROP1myc (Soldati et al., 1995), to generate a fusion between the signal peptide of ROP1 and ROP5 under the control of ROP1 promoter. The ROP5 gene coding sequence between positions corresponding to AA 26 and 549 was PCR-amplified from pROP5 with forward primer HH5 (5′-TGCATGCATGGTACCGTTCAGCTCTCCGCCAAAC-3′; NsiI and KpnI sites underlined) and reverse primer HH6 (5′-TGCATGCATTAGCGACTGAGGGCGCAGCA-3′; NsiI site underlined), and subcloned as a NsiI/NsiI fragment into pROP1myc that had been dephosphorylated with shrimp alkaline phosphatase (SAP).

The plasmid pROP2myc was designed to express a Ct myc-tagged ROP2 protein in RH tachyzoites. It was constructed by inserting the coding sequence of ROP2 without signal peptide in pROP1myc, to generate a fusion between the signal peptide of ROP1 and ROP2 under the control of ROP1 promoter. The ROP2 gene coding sequence between positions corresponding to AA 30 and 561 was PCR-amplified from T. gondii genomic DNA with forward primer 5′-GCAATGCATCACGTACAGCAAGGCGCTGGC-3′ (NsiI underlined) and reverse primer 5′-GCCACCTGCAGCTGCCGGTTCTCCATCAG-3′ (PstI underlined). The resulting PCR was subcloned as an NsiI/PstI fragment into pROP1myc.

All constructs were verified by sequencing. Protein sequence alignments were carried out using ClustalW program (http://www.ebi.ac.uk/clustalw/). Signal peptide and transmembrane region were predicted by using PROSITE (http://www.expasy.org/prosite/) and TEMpred (http://www.ch.embnet.org/software/TMPRED_form.html).

Production of a recombinant GST-NtROP5 protein and of a specific antiserum

Antibodies specific for the Nt sequence of ROP5 (anti-NtROP5) were raised against a fusion between GST and peptide 89–284 of ROP5. The DNA sequence coding for amino acids 89–284 was amplified by PCR from the pROP5 plasmid using forward primer HH24 (5′-CGCGGATCCCCGCTGCTGGACCCTTCGTTT-3′; BamH1 site underlined) and reverse primer HH25 (5′-CCGCTCGAGTCTGTCCCGTGCCTCCTCTGG-3′; XhoI site underlined) and cloning into BamHI and XhoI sites of pGEX-4T3 vector (Pharmacia). The GST-NtROP5 recombinant protein was expressed in Plys Escherichia coli (Stratagene) at room temperature overnight. The GST-NtROP5 protein and control GST were purified on glutathione-agarose beads as described (Frangioni and Neel, 1993) Specific antibodies were obtained by subcutaneous immunizations of BALB/c mice with 20–50 μg of purified protein (first injection in Freund's complete adjuvant; two further injections performed at 3 week intervals in Freund's incomplete adjuvant). The mice were bled 1 week after the third immunization.

Parasite transfection and selection

Transgenic parasites expressing ROP5myc were obtained by cotransfection using 30 μg of pROP5myc linearized with SacI and 3 μg of circular pTUB/CAT plasmids into 107 RHΔhx tachyzoites and selection with 20 μM chloramphenicol (Kim et al., 1993). The ROP2-myc parasites were obtained by the same procedure but the ROP2myc construct was linearized with NotI.

Immunofluorescence assay

HFF were seeded on 12 mm circular coverslips in 24 wells tissue culture plates and infected with tachyzoites 24 h before performing the IFA. Infected cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 30 min at room temperature, washed and permeabilized with 0.1% Triton X-100 in PBS for 10 min, blocked with 10% FCS in PBS for 10 min. Coverslips were subsequently washed in PBS, then incubated with mAb T5 3E2 or the anti-NtROP5 serum diluted 1:200, and with rabbit anti-ROP1 at 1:500 for 30 min at room temperature, washed in PBS and then incubated with affinity-purified goat anti-mouse immunoglobulin G conjugated to FITC at 1:500 (Sigma) and with goat anti-rabbit IgG conjugated to RITC (Jackson ImmunoResearch) at 1:500 for 30 min. Finally, coverslips were washed and mounted onto microscope slides using Immunomount (Calbiochem).

Invasion was studied by incubating parasites with confluent monolayers of HFFs for 15 min at 4°C, then for 20 min at 37°C, and processing for IFA as described above except that Triton X-100 was replaced by 0.05% saponin.

For SLO permeabilization, HFFs were plated on coverslips as above. After 24 h, 5 × 105 parasites were added per well and incubated for 15 min at 4°C, followed by 20 min (or longer times depending on the experiments, see Results) at 37°C. All incubations were then performed on ice. The coverslips were washed three times with Hanks buffered saline solution (HBSS; Gibco-Invitrogen) and subsequently incubated with 10 IU ml−1 SLO reconstituted in the dilution buffer provided by the manufacturer (La Technique Biologique, Paris), for 30 min at 4°C followed by two washes in the SLO dilution buffer, followed by pore formation for 20 min at 20°C. Then the coverslips were washed twice with HBSS, blocked for 30 min with HBSS-10% FCS and incubated for 1 h with anti-myc (at 1:200) or anti-NtROP5 (at 1:200) or mAb T3 4A7 (at 1:50) and with rabbit anti-SAG1 (1:500) as control for PVM integrity. The coverslips were then washed four times with HBSS to remove unbound antibodies and fixed in 4% paraformaldehyde in PBS for 30 min. After one wash with PBS, cells were permeabilized for 10 min in 0.1% Triton X-100 in PBS, followed by a 30 min incubation with anti-mouse and anti-rabbit IgG conjugates and mounted as described above. Evacuoles were generated in the presence of Cyt-D as previously described (Hakansson et al., 2001).

All observations were performed on a Leica DMRA2 microscope equipped for epifluorescence; images were recorded with a COOLSNAP CCD camera (Photometrics, Tucson, AZ) driven by Metaview (Universal Imaging, Downington, PA) and processed using Adobe photoshop 7.0 (Adobe Systems, Mountain View).

SDS-polyacrylamide gel electrophoresis and Western blotting

SDS-PAGE was performed according to Laemmli (Laemmli, 1970). Freshly released tachyzoites were boiled in sample buffer and separated on 10% polyacrylamide gels. Mr markers (Biorad) were used for calibration.

Proteins were transferred electrophoretically to nitrocellulose membranes (Protran, Schleicher and Schuell) at 0.8 mA cm−2 for 90 min by semidry transfer after SDS-PAGE. The nitrocellulose strips were saturated for 1 h in 5% non-fat dry milk in 15 mM Tris-HCl pH 8, 150 mM NaCl, 0.05% Tween 20 (TNT). They were then incubated in mAb (mouse ascitic fluid) diluted 1:500 in TNT for 1 h. After washing, the strips were incubated with alkaline phosphatase conjugated anti-mouse diluted 1:1000 in TNT and stained with BCIP-NBT.

Metabolic labelling and pulse-chase analysis

Highly infected HFF monolayers were rinsed in methionine and cysteine-free Dulbecco's Modified Eagle's Medium (DMEM; Invitrogen) containing 1% dialysed FCS. They were incubated in the same medium for 30 min at 37°C in a 5% CO2 incubator prior to the addition of 50 μCi ml−1[35S] methionine/cysteine (700 Ci mM−1, MP Biomedicals, Vannes, France). For pulse-chase experiments, the infected monolayers were labelled for 30 min, rinsed with complete DMEM containing 10% FCS and either arrested or incubated for 2 h chase prior to immunoprecipitation. Infected monolayers were then solubilized in lysis buffer [Tris-HCl 50 mM pH 8.3, NaCl 150 mM, EDTA 4 mM, PMSF 1 mM, 1% Nonidet 40 (NP40)] for 1 h at 4°C. The lysate was centrifuged 1 h at 16 000 g and the supernatant was collected for immunosorption. The immunosorbents were prepared by incubating 20 μl of ascitic fluid with 20 μl of Protein G-sepharose for 1 h in 1 ml of PBS; they were then incubated with radiolabelled lysate at 4°C for 2 h under gentle agitation, washed four times with a buffer containing 1 M NaCl and 0.5% NP40 in 50 mM Tris-HCl pH 8.3 then in 5 mM Tris-HCl pH 6.8. Elution was then performed during 5 min at 95°C with electrophoresis sample buffer. After SDS-PAGE, the gel was impregnated with Amplify (Amersham), dried and exposed to X-O-mat film.

Acknowledgements

We thank Gilles Labesse and John Boothroyd for stimulating discussions and suggestions during the course of this study, and Dominique Soldati for providing the pROP1myc and Ptubcat plasmids.

Ancillary