Origins of the parasitophorous vacuole membrane of the malaria parasite, Plasmodium falciparum, in human red blood cells

The sequence of events that characterises the entry of the malaria parasite into the red blood cell has been defined at the morphological level (Bannister et al., 1975; Aikawa et al., 1978; Miller et al., 1979; Aikawa et al., 1981). In essence the merozoite first makes random contact with the red cell surface; it then reorients, so as to bring its apical surface into apposition with the host cell membrane. An invagination develops in the host cell and deepens as the merozoite enters. An electron-dense junction, which forms during the attachment phase, remains at the contact zone as the parasite moves inwards; when the parasite is engulfed, the host cell membrane closes behind it. The internalised parasite is thus encapsulated in a so-called parasitophorous vacuole.

The appearance of the invasion process in the electron microscope strongly suggested that the parasitophorous vacuole membrane (PVM) is formed by eversion of the host cell membrane, and until recently this was generally assumed to be the case (see e.g. Holz, 1977, for a review). The PVM, moreover, is devoid of the major red cell membrane proteins (Atkinson et al., 1987 and this is true also of the membrane that bounds the internal vacuole, which represents the first step in formation of the cavity in the host cell that eventually receives the parasite (Dluzewski et al., 1989). Nevertheless, counter-indications have accumulated that have been interpreted as indicating a parasite-derived origin for the PVM; in particular, the rhoptries and micronemes, which are organelles located near the apex of the merozoite, were shown to contain lamellar deposits, consisting almost certainly of lipid (Bannister et al., 1986; Stewart et al., 1986). This material drains through ducts into the area of contact with the host cell at the time of entry (Bannister and Mitchell, 1989) or attachment of the parasite (Aikawa et al., 1981). It also appears that fluorescent lipids, metabolically introduced into the parasite, appear in the PVM after invasion, indicating that parasite-derived lipid makes at least some contribution to the PVM (Mikkelsen et al.,1988)

We have attempted to resolve the conflicting evidence about the origin of the PVM by immuno-electron microscopy, making use of a haptenic lipid, inserted into the red cell membrane before exposure to the parasite. Our results support the view that the PVM is derived partly or entirely from the parasite.

Plasmodium falciparum parasites were cultured in vitro (Trager and Jensen, 1976) and synchronised by the sorbitol method (Lambros and Vandenberg, 1979). The duration of invasion experiments was 7 h. P. knowlesi parasites had been cryopreserved as ring-stage trophozoites in Callithrix jacchus red cells. These were thawed, rehydrated and cultured overnight to schizogony as described earlier (Bannister and Mitchell, 1989). Invasion of human red cells was initiated by addition of purified schizont preparations (Dluzewski et al., 1984) to the cells in RPMI 1640 culture medium containing 10% human serum. Externally, irreversibly attached P. knowlesi result when cytochalasin B is added to the merozoites or rupturing schizonts before mixing with the red cells (Miller et al., 1979); the conditions of treatment were as described previously (Dluzewski et al., 1989). Incubations were for 90 min with 2 μg ml ⁻¹ cytochalasin B. Human target cells were used.

The labelled lipid, NBD-PE, i.e. l-acyl-2-[6-[(7-nitrobenz-2-oxa-l,3-diazol-4-yl) amino] caproyl] phosphatidylethanolamine, was obtained from Avanti Polar Lipids. It was dispersed in the form of an ethanolic solution at 80 μg ml ⁻¹ in 100 volumes of RPMI medium, to which packed, washed red cells were then added to give a haematocrit of 1% and allowed to incorporate into the membrane for 1 h at 37°C (Tanaka and Schroit, 1983). The cells were washed in RPMI. During the incubation with parasites this extraneous lipid is translocated to the inner membrane leaflet (Devaux, 1988), and this becomes largely resistant to extraction by albumin-containing medium (see below).

For preparation of anti-NBD antibodies, bovine immunoglobulin G (Sigma) was derivatised by reaction with NBD-C1 (7-chloro-4-nitrobenzo-2-oxa-l,3-diazole); 1 mg reagent, dissolved in ethanol, was added to 5 ml protein at a concentration of 2 mg ml ⁻¹ in 0.1 M sodium phosphate, pH 8.1, and the reaction was allowed to proceed at room temperature for 90 min. An excess of buffered Tris hydrochloride was added and the protein was dialysed against 0.1 M sodium chloride, 20 mM sodium phosphate, pH 7.4. A rabbit was inoculated subcutaneously and intramuscularly with the protein in complete Freund’s adjuvant, followed by two booster injections with the antigen in Freund’s incomplete adjuvant over a period of 4 months. Serum was taken 6 weeks after the last boost.

Parasitised fluorescent cells in wet films under coverslips were examined in a Zeiss microscope in epifluorescence in the presence of p-phenylenediamine as anti-bleaching agent. For electron microscopy, cells were metabolically depleted before fixation: this was found to reduce the extraction of the antigenic lipid in the course of ethanol-dehydration. Thus the parasitised cells were suspended in isotonic phosphate-buffered saline, containing 4 mM iodoacetate, pH 7.4; the supernatant was then replaced by the same buffer, containing 5 mM N-ethylmaleimide, pH 8.0. The cells were fixed in 0.5% glutaraldehyde in 0.1 M potassium phosphate, pH 7.4, followed by suspension in 50 mM ammonium chloride for 20 min, and finally washed twice with 0.1 M potassium phosphate, pH 7.4. The treated cells were dehydrated by successive transfers to five solutions of increasing ethanol concentration, up to 75% (v/v). The cells were equilibrated with LR White monomer, transferred to gelatin capsules and the resin was polymerised by warming at 50°C for 24 h. Sections were prepared and mounted on 200-mesh copper grids.

The grids were floated on 50 mM potassium phosphate buffer, pH 7.4, which contained 10 mg ml^-1 bovine serum albumin and 2.5 mg ml ⁻¹ Tween 20. They were then transferred to the surface of a drop of antibody solution, diluted 1:20 -1:40, or, in the case of controls, normal rabbit serum and antiserum, absorbed with isolated membranes from NBD-Cl-labelled ghosts, and left in place for 1 h. They were washed by flotation on three changes of buffer and then floated on a solution of protein A, labelled with colloidal gold by conjugation with chlorauric acid (BDH), following the method of Roth (1982). The grids were rinsed twice with buffer, then with distilled water, air-dried and stained with uranyl acetate and lead citrate. The sections were coated with carbon and examined in a Philips 300 electron microscope at 80 kV accelerating voltage. The same procedure was used for the detection of the transmembrane protein, band 3, as described earlier (Dluzewski et al., 1989).

If a labelled phospholipid is to be introduced into the red cell membrane bilayer, the derivatising group must be in the acyl chain and not the head-group (Struck and Pagano, 1980). Moreover, an aminophospholipid modified in this manner will be translocated to the inner membrane leaflet by the membrane-associated translocase (Devaux, 1988). NBD-PE fulfilled these requirements and gave rise to abundant fluorescence in the cell; a part of this remained in the membrane and was only slowly extracted on incubation with the culture medium, which contains a high concentration of serum albumin. Thus the cells remained brightly fluorescent over the 7 h period of incubation in the invasion experiments, and even after 20 h there was only a moderate reduction in intensity. The labelled cells were invaded by P. falciparum with comparable efficiency to control cells. Examination in the fluorescence microscope revealed no detectable excess fluorescence in the region of the intraerythrocytic parasite, suggesting that the label had not entered the PVM; limitations of contrast and resolution, however, precluded definitive or quantitative conclusions on the basis of this negative observation. We therefore had recourse to gold-labelling immuno-electron microscopy.

It is likely that a proportion of the lipid would be extracted in the course of dehydration and embedding, although the aminophospholipids may be stabilised by the fixation step with glutaraldehyde. The results show in any event that enough lipid remains to give satisfactory immuno-gold labelling in thin sections. Fig. 1 shows labelling of the red cell plasma membrane, which is absent from sections treated with antiserum that had been absorbed out with ghosts from NBD chloride-derivatised red cells. Equally the labelling has no (or only a minimal) counterpart around the periphery of the parasite or elsewhere in either cell. The number of attached antibody molecules per unit length of host cell membrane and PVM contour is given in Table 1. These results strongly suggest that the PVM is not in large degree derived from the host cell membrane.

We then examined red cells bearing attached P. knowlesi merozoites: in this system the invasion cycle of the parasites, treated with cytochalasin B, is arrested at the stage of irreversible attachment (Miller et al., 1979). With the human cells used here we did not observe the internal vacuole in the host cell, opposite the zone of attachment, that were formed in simian cells (Dlu-zewski et al., 1989). In Fig. 2 no haptenically labelled lipid is to be seen in the red cell membrane in the region of contact between the attached parasite and the host cell. To establish this as a general feature of the attached state, because of the sparsity of the label, it was necessary to determine the distribution of gold particles around the membrane contour, measured from the left-hand boundary of the zone of contact (Fig. 3). Here the first element on the left represents the part of the membrane in contact with the parasite, and is the only one in which there is no antibody in any of the 14 cells with attached parasites examined. To evaluate the significance of this observation we may calculate the probability that this has happened by chance: if r particles (summed over all cells examined) are distributed between n membrane elements the probability (P_k) of finding k of these r particles in the ith membrane element (n> i> 1) is: For 132 particles, distributed randomly among 18 membrane elements, the probability of finding no particles in any one element (here by definition element 1), P₀, is 5.3 ×10 ⁻⁴. Thus the probability that the host cell lipid marker is absent from the region of contact by chance can be disregarded. We infer that the parasite at the stage of incipient invasion generates membrane material, which appears to form a pool at the point of attachment, that then normally expands inwards to form the vacuole into which the parasite passes. Because of the rapidity of the invasion process it may be surmised that the substance of the new membrane is wholly or partly derived from the pre-existing rhoptry and microneme contents (Bannister et al., 1986; Stewart et al., 1986; Bannister and Mitchell, 1989), rather than freshly synthesised lipid.

The debate about the provenance of the PVM (Joiner, 1991) has focussed on the appearance presented by the invasion process in the electron microscope (Aikawa et al., 1978, 1981), which undeniably suggests encapsulation of the invading merozoite by the host cell membrane, and on the other hand on the following items of evidence, all indirect: (1) multiple invasion, which is not uncommon, would, on the above model, result in the elimination of a considerable proportion of the host cell membrane area (some 7% for each merozoite internalised in the case of P. knowlesi); (2) the PVM is much more resistant than the red cell to lysis by saponin (Sherman and Hull, 1960), which suggests a different lipid composition; (3) the rhoptries and micronemes contain abundant amounts (sufficient, as Bannister and Mitchell (1989) have pointed out, to generate a bilayer with the area of the PVM) of material with the appearance of multilamellar lipid bodies (Bannister et al., 1986; Stewart et al., 1986); this material drains from the organelles at the moment of invasion; (4) when fluorescent lipid precursors are introduced into parasites and metabolically incorporated into lipids, discernible fluorescence appears in the PVM following invasion (Mikkelsen et al., 1988); (5) We found that after incubation a definite, though evidently minor proportion of the NBD-PE became resistant to back-extraction and was by implication in the inner leaflet. Whether this represents a level of translocation lower than detected by Colleau et al. or possibly another isomer present in the preparation is not at this stage clear. the PVM in mature intracellular parasites (Atkinson et al., 1987) and in freshly internalised parasites, as well as the internal vesicle membrane in red cells bearing irreversibly attached parasites (Dluzewski et al., 1989, and Fig. 2c), is devoid of major red cell membrane proteins. Thus either the endogenous proteins are swept away as the parasite invaginates the host cell (assuming that they are not eliminated in their entirety by proteolysis even before invasion begins), or the PVM is, as our present results lead us to infer, derived wholly or partly from parasite material alone.

Haidar and Uyetake (1992) have recently reported that a fluorescent carbocyanine dye, introduced into the red cell membrane, is carried in by the invading parasite, as judged by the distribution of fluorescence. The fluorescent intensity appears much higher in the parasite than the host cell, and this is attributed to a “different membrane environment” or selective increase in dye concentration in the parasite membrane. In either event the implication would be that the lipid composition of the PVM is not that of the red cell. In the absence of quantitative data, there is not necessarily any incompatibility between these data and ours.

The exchange of lipid by endocytic pathways or even through the fluid phase between parasite and host cell cannot be excluded a priori, but the absence of the NBD-PE from the PVM appears to eliminate such a process, at least on the time scale of invasion. Later in development (trophozoite stage) PE is evidently transported from the host cell plasma membrane to the parasite (Haidar et al., 1989) and according to Pouvelle et al. (1991) bulk diffusion of extraneous solutes occurs at this stage by way of a duct. Some migration of PC between the two membranes, presumably by a different mechanism, has also been detected at an early stage of development (Haidar et al., 1989).

This work was supported by the UNDP/World Bank/ World Health Organization Special Programme for Research and Training in Tropical Diseases. We thank Dr Gary Ward for an illuminating discussion.

Colleau et al. (1991) have stated that the NBD-PE, bearing the substituent in the C-6 position of the caproyl chain, does not undergo enzymic translocation to the inner membrane leaflet, as measured by back-extraction with serum albumin.