The origin of parasitophorous vacuole membrane lipids in malaria-infected erythrocytes

Malaria parasites require both a vertebrate and an invertebrate host to complete their life cycle. In the vertebrate host, asexual-stage merozoites invade and multiply within erythrocytes. The parasite-induced modification and destruction of erythrocytes that occur during this stage of the life cycle give rise to the life-threatening aspects of the malaria infection. Consequently, it is important to understand and interdict the process whereby the merozoite invades and becomes established within the vertebrate host erythrocyte.

Morphological studies at the light and electron microscope levels have revealed that invasion is a sequential, multistep process (Dvorak et al., 1975; Bannister and Dluzewski, 1990; Aikawa et al., 1978; Miller et al., 1979). Upon contact with an erythrocyte, the merozoite attaches and orients its anterior end towards the erythrocyte. A region of tight apposition, or ‘junction’ (Aikawa et al., 1978), develops between the membranes of the two cells. Membrane-bound organelles at the anterior end of the merozoite, the rhoptries, discharge their contents onto the erythrocyte surface, which begins to indent at the point of contact. The junction transforms from a localized patch to an orifice, through which the merozoite penetrates into a progressively deepening, membrane-bound vacuole. The membrane surrounding the fully internalized parasite is known as the parasitophorous vacuole membrane (PVM).

The adult erythrocyte appears to be incapable of receptor-mediated endocytosis (Haberman et al., 1967; Schekman and Singer, 1976; Zweig and Singer, 1979), and the mechanism of formation of the PVM is unknown. It has been reported that erythrocyte proteins (McLaren et al., 1977; Aikawa et al., 1981; Atkinson et al., 1987; Dluzewski et al., 1989) and lipids (Dluzewski et al., 1992) are excluded from the PVM, leading to the suggestion that the PVM is formed from substances stored in the rhoptries and secreted into the erythrocyte membrane during invasion (Bannister and Dluzewski, 1990; Dluzewski et al., 1992; Joiner, 1991). To study the origin of the PVM, we labeled the surface of the erythrocyte with a variety of fluorescent probes, and followed the fate of these probes during PVM formation by low-light-level video fluorescence microscopy. We found that while erythrocyte membrane proteins are indeed excluded from the PVM, the lipids of the forming PVM are indistinguishable from those of the host cell membrane.

PKH2, PKH26, and Diluents A and C were purchased from Zynaxis Cell Science, Inc. (Malvern, PA). Bisbenzimide Hoe33342, N-2-hydroxyethylpiperazine-N’-2-ethanesulfonic acid (HEPES; Ultrol grade), cytochalasin B, adenosine diphosphate (ADP), and hypoxanthine were from Calbiochem (San Diego, CA). Octadecyl rhodamine B (R18), 1,1’-dihexadecyl-3-3’-3-3’-tetramethylindocarbocyanine (DiIC16), fluorescein-5-thiosemicarbazide, Texas Red sulfonyl chloride, and 5-(4,6-dichlorotriazinyl)aminofluorescein (DTAF) were from Molecular Probes, Inc. (Eugene, OR). 1-palmitoyl-2[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]caproyl]-sn-glycero-3-phosphoserine (NBD-PS) and 1-acyl-2[6-[(7-nitro-2-1,3-benzoxadiazol-4-yl)amino]caproyl]-sn-glycero-3-phosphoethanolamine (NBD-PE) were from Avanti Polar Lipids (Alabaster, AL). Leupeptin and Nutridoma-NS were from Boehringer Mannheim Corp. (Indianapolis, IN), NuSerum I was from Collaborative Biomedical Products (Bedford, MA), Versilube silicon oil was from General Electric Co. (Waterford, NY), citrate phosphate dextrose anticoagulant (CPD) was from Baxter Healthcare Corp. (Deerfield, IL), Percoll was from Pharmacia LKB Biotechnology Inc. (Piscataway, NJ), and prestained high molecular mass standards were from Bio-Rad Laboratories (Melville, NY). Phosphate-buffered saline (PBS) and horse serum were from BioFluids, Inc. (Rockville, MD). Gentamicin, fetal bovine serum, and RPMI-1640 (#430-1800EC) were from Gibco BRL Life Technologies, Inc. (Grand Island, NY). Acid-washed glass beads (<0.106 μm) and fatty acid-free (#A-6003) bovine serum albumin (BSA) were from Sigma Chemical Co. (St. Louis, MO).

Human Duffy-positive (Fya⁻b⁺) and -negative (Fya⁻b⁻) erythrocytes (Department of Transfusion Medicine, National Institutes of Health) and rhesus monkey (Macaca mulatta) erythrocytes were obtained by venipuncture into syringes containing 0.15 volumes CPD. Blood was stored at 4°C until use.

Human serum was from Interstate Blood Bank, Inc. (Memphis, TN). To prepare rhesus monkey serum, whole blood was gently rocked for 2.5 h at 24°C and centrifuged twice (5 min, 1300 g), saving the supernatant (serum) and discarding the pellet each time. All sera were heat inactivated (30 min at 56°C) and stored at -20°C until use.

Erythrocytes were washed 3 times prior to labeling by resuspension in approximately 30 packed cell volumes of PBS and centrifugation for 60 s at 3200 g. After labeling, an equal volume of fetal bovine serum was added to the cells, and the suspension was incubated for 2-5 min at 24°C to remove unincorporated dye. The labeled cells were then washed with PBS, pelleted through a 45% Percoll cushion (400 μl suspension in PBS layered over 700 μl Percoll and spun 60 s at 3200 g), and washed 3 times with RPMI-1640 supplemented with 11 mM glucose, 37 μM hypoxanthine, 29 mM HEPES, pH 7.2 (MED-1), changing tubes between each wash. The cells in the final pellet were resuspended in 3 volumes MED-1 and stored at 24°C until use. NBD-PS- and NBD-PE-labeled erythrocytes were processed differently after labeling, as described below.

Washed, packed cells (20 μl) were added (while vortexing) to 180 μl Diluent C, followed immediately by 200 μl of 7.5 μM PKH26 in Diluent C. The cells were labeled for 60 s at 24°C with intermittent vortexing. Identical results were obtained using cells labeled with 3.75 μM PKH26 (as above) or 1.0 μM PKH26.

Washed, packed cells (10 μl) were added (while vortexing) to 190 μl Diluent A, followed immediately by 200 μl of 6 μM PKH2 in Diluent A. The cells were labeled for 5 min at 24°C with intermittent vortexing.

Washed, packed erythrocytes were resuspended to 10% hematocrit in PBS. Freshly prepared R18 (1 mM in dimethylsulfoxide) was added to a concentration of 10 μM, and the suspension was gently agitated in the dark for 30 min at 24°C.

To measure the extent of R18 self-quenching, labeled and washed erythrocytes were lysed in 20 volumes 10% PBS (10 min, on ice), pelleted (5 min, 12,000 g), resuspended in 10% PBS, and repelleted. Fluorescence associated with the erythrocyte ghosts was measured in 10% PBS ± 1% volume/volume Triton X-100 using an Aminco-Bowman spectrofluorimeter/ratio photometer, and% self-quench was calculated as previously described (Hoekstra et al., 1984). At a labeling concentration of 10 μM, incorporated R18 was approximately 50% self-quenched. Self-quenching was eliminated when the erythrocytes were labeled with concentrations of R18 ≤ 250 nM.

Washed, packed erythrocytes were resuspended to 1% hematocrit in PBS. Freshly prepared diIC16 (100 μg/ml in ethanol) was added to a concentration of 1 μg/ml, and the suspension was gently agitated in the dark for 50 min at 24°C.

Washed, packed erythrocytes were resuspended to 25% hematocrit in PBS. Freshly prepared sodium m-periodate (5 mM in PBS) was added to a concentration of 0.5 mM, and the suspension was gently agitated for 30 min at 24°C. The cells were washed three times with PBS, resuspended to 25% hematocrit in PBS containing 0.66 mg/ml fluorescein thiosemicarbazide, and gently agitated in the dark for 45 min at 24°C.

Washed, packed erythrocytes were resuspended to 25% hematocrit in PBS-borate (1 volume PBS + 1 volume 200 mM boric acid, pH adjusted to 8.5 with NaOH). Freshly prepared DTAF (7.5 mg/ml stock in 200 mM sodium borate, pH 9.0) was added to a concentration of 0.5 mg/ml, and the suspension was gently agitated in the dark for 30 min at 24°C.

Erythrocytes were washed 3 times in PBS-borate, and resuspended to 10% hematocrit in PBS-borate. An equal volume of 0.5 mg/ml Texas Red sulfonyl chloride in PBS-borate was added, and the suspension was gently agitated in the dark for 30 min at 24°C. An equal volume of fetal bovine serum was added, and after 2 min at 24°C the cells were pelleted and washed 3 times with PBS. Then 10 μl of washed packed cells were stained with PKH2 as described above.

Washed, packed erythrocytes (50 μl) were resuspended in 450 μl PBS. NBD-PS (2 mg/ml in ethanol) was added (while vortexing) to a final concentration of 10 μg/ml, and the suspension was gently agitated in the dark for 45-60 min at 37°C (Connor et al., 1990). All subsequent manipulations were carried out at 4°C. The labeled cells were washed 4 times with 200 packed cell volumes of PBS, followed by 6 back extractions in defatted BSA to remove label in the outer leaflet of the membrane. Each back extraction consisted of a 2 min incubation in 200 packed cell volumes of MED-1 containing 1% (w/v) fatty acid-free BSA, followed by 4 min centrifugation at 3200 g. The cells were transferred to a new tube between each wash or back extraction. After the final back extraction, the cells were washed once in MED-1, resuspended in 3 volumes of MED-1, and used immediately.

Erythrocytes were labeled with NBD-PE as described (Dluzewski et al., 1992). Briefly, 5 μl of washed, packed erythrocytes were resuspended in 495 μl of MED-1 containing 0.22% (w/v) sodium bicarbonate (MED-1B). NBD-PE (5-8 mg/ml in ethanol) was added (while vortexing) to a final concentration of 50-80 μg/ml, and the suspension was gently agitated in the dark for 60 min at 37°C. The cells were washed, back-extracted, and resuspended to 25% hematocrit in MED-1 as described above for NBD-PS.

Erythrocytes containing ring-stage Plasmodium knowlesi parasites (Malaysian H strain, clone Pk1[A⁺]) from an infected rhesus monkey (typically at 20-50% parasitemia) were collected in heparin, passed through a column of glass beads in the presence of 0.25 mg/ml ADP (Miller et al., 1979), washed in MED-1, and cryopreserved as previously described (Aley et al., 1984).

For in vitro culture, cryopreserved parasitized erythrocytes were thawed as previously described (Aley et al., 1984), resuspended to 4×10⁷ parasites/ml in RPMI-1640 containing 10 μg/ml gentamicin, 360 μM hypoxanthine, 25 mM HEPES, pH 7.2, 0.24% (w/v) sodium bicarbonate (MED-2) supplemented with 10% human A-positive serum, and incubated for 19-24 hours at 37°C in an atmosphere of 5%CO₂:5%O₂:90%N₂. Parasite development was monitored with thin smears, which were air-dried, methanol-fixed, and stained with MED-1 containing 10 μg/ml bisbenzimide Hoe33342 to fluorescently label the nuclei of the parasites. When the majority of developing schizonts contained 4-8 nuclei, the cells were pelleted (5 min, 1300 g) and resuspended at 4×10⁷ parasites/ml in fresh MED-2 containing 10% human serum, 50 μg/ml chymostatin and 50 μg/ml leupeptin (David et al., 1984; Miller et al., 1983). After 3-5 hours at 37°C (5%CO₂:5%O₂:90%N₂), the cells were pelleted, washed and resuspended in MED-1, and passed 10 times through a 25 G needle to release the merozoites (Miller et al., 1983). Unsheared schizonts and uninfected erythrocytes were pelleted (60 s, 1100 g) and discarded. The supernatant was centrifuged again (60 s, 1100 g) and the pellet discarded; free merozoites were pelleted from the supernatant (90 s, 1900 g), resuspended in MED-1, and stored at room temperature until use. Infected erythrocytes cultured in MED-2 containing 10% human serum yielded approximately 0.1 merozoite per initial ring stage parasite; these yields were better than with MED-2 containing 2.5% human serum; 10% rhesus, fetal bovine, or horse serum; 10% NuSerum; 1% Nutridoma-NS ± 2.5% human serum; or 4% Nutridoma-NS + 1% bovine serum albumin.

In some experiments, splenectomized rhesus monkeys were infected by intravenous (i.v.) injection of the thawed inoculum. When the parasitemia reached 1-10% mid-to late-stage schizonts (4-10 nuclei), blood was collected in 0.15 volumes CPD and passed through a column of glass beads in the presence of 0.25 mg/ml ADP. The effluent from the column was diluted with 3 volumes MED-1; 30-ml portions of the suspension were layered over 20-ml cushions of 45% volume/volume Percoll in MED-1, and centrifuged 5 min at 1300 g. Mid-to late-stage schizonts remain on top of the Percoll cushion, while the earlier stages and uninfected erythrocytes pellet at the bottom of the tube. The schizont layer (typically 90-98% schizonts) was processed a second time by Percoll gradient centrifugation, as above, yielding suspensions containing 98-100% schizonts. The cells were diluted with 3 volumes MED-1, pelleted (5 min 1300 g), resuspended to 4×10⁷ parasites/ml in MED-1B containing 50 μg/ml chymostatin and 50 μg/ml leupeptin, and incubated 3-5 hours at 37°C (5%CO₂:5%O₂:90%N₂). Merozoites were recovered after washing and syringe release as described above. We typically recovered 1 merozoite per schizont by these methods. These merozoites and the merozoites prepared from in vitro cultures (above) gave identical results in the experiments presented here.

Plasmodium falciparum (HB3 strain) were grown in continuous culture as described (Trager and Jensen, 1976). P. falciparum invasion assays using gelatin-enriched schizonts (Jensen, 1978) were carried out as described (Dolan et al., 1990) using a ratio of 5 target erythrocytes per schizont.

Merozoites and erythrocytes were mixed at a ratio of approximately 5:1 and incubated at 37°C with occasional mixing. Rhesus erythrocytes were used unless otherwise indicated. To separate free merozoites from attached or internalized merozoites, 50-250 μl of invasion mixture were layered onto a 500 μl cushion of 45% Percoll and centrifuged for 60 s at 1720 g in a Fisher 59A swinging bucket microcentrifuge; free merozoites remain on top of the cushion, while attached or internalized merozoites pellet with the erythrocytes. For fluorescence microscopy, the pellet was resuspended in MED-1 and an equal volume of fetal bovine serum containing 20 μg/ml bisbenzimide Hoe33342 was added immediately prior to observation. To score% invasion (ring formation), the pellet was resuspended in fetal bovine serum, and used to make thin blood smears, which were air-dried, methanol-fixed, and stained with Giemsa.

Invasion was arrested at the attachment step by treating merozoites for 3 min with 10 μg/ml cytochalasin B prior to adding them to target erythrocytes (Miller et al., 1979). The 45% Percoll and resuspension solutions used to separate free and attached merozoites (see above) were also supplemented with 10 μg/ml cytochalasin B.

To visualize invasion in real time, 1 μl portions of the merozoite and erythrocyte suspensions were mixed with 2 μl of fetal bovine serum containing 20 μg/ml bisbenzimide Hoe33342 on a 22 mm × 40 mm coverglass. The mixture was surrounded by a thin ring of silicon oil, approximately 3 μl of the invasion mixture was removed, and the coverglass was inverted onto a glass microscope slide. The cells were visualized with an Axiophot microscope (Zeiss, Thornwood, NY) equipped with a vertical fluorescence illuminator and differential interference contrast (DIC) optics. A custom-built computer-controlled shutter (J. A. Dvorak, unpublished), a light attenuator, an infrared blocking filter, and beam-shaping optics were placed in the filter holder slots of the vertical fluorescence illuminator of the Axiophot to minimize radiation damage to the cells during visualization of the fluorescent probes. The Zeiss 487915 filter set was used to visualize PKH26, R18, diIC16, and Texas Red sulfonyl chloride fluorescence; the Zeiss 487909 filter set was used to visualize PKH2, fluorescein-5-thiosemicarbazide, NBD, and DTAF; the Zeiss 487905 filter set was used to visualize bisbenzimide Hoe33342. The microscope stage temperature was maintained at 33°C with an NPI Air Stream Incubator (Nicholson Precision Instruments, Gaithersburg, MD).

Video images were obtained from the microscope with a Westinghouse Model ETV 6006E low-light level video camera (Westinghouse Electric Corp., Horseheads, NY), or a Sony XC-77 CCD video camera (Sony Components Products Div., Cypress, CA) connected either directly to a Gen II Sys image intensifier (Dage-MTI, Inc., Michigan City, IN), or through a Hamamatsu C2400-60 control unit (Hamamatsu Photonic Systems Corp., Bridgewater, NJ) to a Hamamatsu Model C2400-68 image intensifier. The magnification of the video image was adjusted with projection optics placed between the microscope and the video camera. The automatic gain control circuitry was disabled on all video equipment; an oscilloscope was used to monitor adjustment of the video signal. The video signal was combined with time and date code and the composite signal was collected with a model DT2861 or DT2862 arithmetic frame grabber (Data Translation, Marlboro, MA) operating in an IBM PC-AT-compatible computer. The video images were stored in a digital format and analyzed with either ImagePro Plus (Media Cybernetics, Silver Spring, MD) or Aries (Skylord Designs, Bethesda, MD), prepared for publication with Halo f/x (Media Cybernetics) or Animator Pro (Autodesk, Sausalito, CA), and transferred to film with a Lasergraphics LFR film recorder (Lasergraphics, Irvine, CA).

A 60 μl portion of packed, fluorescently-labeled erythrocytes was added (while vortexing) to 3 ml of 5 mM sodium phosphate, pH 7.4, at 4°C. After 5 min on ice, the lysed cells were pelleted (5 min, 12,000 g) and washed twice with 1.5 ml 5 mM sodium phosphate (4°C). The final pellet was dissolved in sample buffer and analyzed by SDS-PAGE as described (Ward et al., 1983). Total ghost proteins were visualized with Coomassie Blue, and fluorescently-labeled proteins were visualized on a UVT 400-M transilluminator (International Biotechnologies, New Haven, CT).

When erythrocytes labeled with the fluorescent lipophilic probes PKH26, PKH2 or DiIC16 were invaded by P. knowlesi merozoites and observed 20 min after invasion, internalized fluorescence was seen in the region of the intraerythrocytic parasite (Fig. 1c). In contrast, when erythrocytes were labeled 20 min post-invasion, no internalized probe was seen (Fig. 1f), suggesting that the internal probe seen in prelabeled erythrocytes (Fig. 1c) was probably internalized at or near the time of invasion. Haldar and Uyetake (1992) have reported similar observations with P. falciparum and DiIC16-labeled erythrocytes.

When parasite entry into PKH26-labeled rhesus or human erythrocytes was blocked by cytochalasin B (Miller et al., 1979), no transfer of fluorescent probe from erythrocytes to attached merozoites was observed during the normal time course of an invasion assay (Fig. 1i). However, if the parasites were allowed to remain attached to either rhesus or human erythrocytes for prolonged periods (>60 min), they became weakly fluorescent (data not shown). Merozoites attached to human erythrocytes lacking the Duffy blood group antigen also became detectably fluorescent, but only after longer periods of attachment (>135 min). Free merozoites were not detectably labeled, even after 135 min.

We followed the kinetics of fluorescent probe internalization during invasion by low-light level video microscopy, using both DIC and fluorescence optics. The DIC images of a typical interaction are shown in Fig. 2 (upper row). After initial attachment of the merozoite (Fig. 2A) and apical reorientation, the surface of the erythrocyte begins to indent at the point of contact. As the merozoite enters, this indentation grows progressively deeper to form the invasion pit (Fig. 2B), the neck of which constricts and ultimately closes once the parasite is completely inside the erythrocyte, to form the PVM (Fig. 2C, D; see also Aikawa et al., 1978). During the process of invasion the erythrocyte often changes from a biconcave to a spherical shape (Fig. 2A-D; see also Dvorak et al., 1975). PKH26 fluorescence images corresponding to the DIC images are shown in the lower row of Fig. 2. The fluorescent probe is initially distributed uniformly throughout the erythrocyte membrane (Fig. 2A); the biconcave shape of the erythrocyte at this stage is characterized by the bright circle of fluorescence that appears at the steep transition in erythrocyte thickness. As the merozoite enters, the fluorescent probe demarcates the initial indentation of the erythrocyte membrane (not shown), the deepening invasion pit (Fig. 2B), and ultimately the fully formed PVM (Fig. 2D). These images demonstrate that the probe is indeed internalized at the time of invasion. Furthermore, since the posterior end of a partially internalized parasite is not fluorescent (e.g. see Fig. 2B), the probe is apparently internalized via the PVM, and not incorporated into the merozoite itself.

The video microscopy system had been adjusted to produce a linear intensity output response for all of the images collected. Consequently, we were able to quantify fluorescence intensity profiles of the images to determine the relative distribution of PKH26 in the erythrocyte membrane and PVM. Intensity profiles of the images presented in Fig. 2 are shown in Fig. 3; fluorescence intensity was determined along the axis shown in the inset of each profile. The mean fluorescence intensity in regions encompassing the membranes of both the erythrocyte and the PVM was approximately two-fold higher than the mean intensity in regions encompassing only the erythrocyte membrane. The two-fold difference in [erythrocyte + PVM] to erythrocyte fluorescence was highly reproducible: in 29 independent images of invading or recently internalized (<10 min) parasites, the mean [erythrocyte + PVM]/erythrocyte fluorescence intensity ratio was 1.92 (s.e.m. = 0.06). A Shapiro-Wilke test of the data demonstrated that it was normally distributed about the mean (P=0.65). As there are twice as many membrane surfaces present in regions containing the PVM, these results indicate that, within the experimental limits of the technique, the PVM and the erythrocyte membrane contain approximately equal concentrations of the fluorophore. The concentration of PKH26 in the PVM remains equal to the concentration in the erythrocyte membrane during the first 10 minutes post-invasion (Figs 3, 7 and data not shown).

In light of a recent report suggesting that host membrane phospholipids are excluded from the PVM (Dluzewski et al., 1992), we performed studies similar to those described above using the fluorescent phospholipid analog, NBD-PE.

Twenty minutes after mixing NBD-PE-labeled erythrocytes with unlabeled merozoites, we again observed internal fluorescence in the region of the intracellular parasite (Fig. 4c). However, in contrast to PKH26, PKH2 and DiIC16, NBD-PE did not remain confined to the labeled erythrocytes in the presence of other unlabeled cells. Within 5 minutes of mixing merozoites with NBD-PE-labeled erythrocytes, free merozoites and erythrocyte-attached merozoites became detectably fluorescent (Fig. 4f). Furthermore, when unlabeled erythrocytes were invaded by merozoites, NBD-PE added 20 minutes after invasion labeled the internal parasite and/or PVM (data not shown; see also Pouvelle et al., 1991), rendering interpretation of images such as those shown in Fig. 4c difficult. Bright internal fluorescence was also observed in newly invaded erythrocytes 7 hours after mixing P. falciparum schizonts with NBD-PE-labeled human erythrocytes (Fig. 4i).

A second phospholipid analog, NBD-PS, is known to be more efficiently translocated to the inner leaflet of erythrocytes than NBD-PE (Colleau et al., 1991) and would, therefore, be expected to show less cell-cell exchange than NBD-PE (Sandra and Pagano, 1979). After labeling erythrocytes with NBD-PS and back extracting with fatty acid-free BSA to remove probe present in the outer leaflet (Connor et al., 1990), we found that the amount of NBD-PS resistant to further extraction represented >95% of the total probe associated with the cells (Fig. 5A). When erythrocytes were depleted of ATP prior to labeling, to inhibit the pump responsible for translocating aminophospholipids from the outer to the inner leaflet (Daleke and Huestis, 1985; Devaux, 1991), less probe was resistant to back extraction (approximately 20% the level seen in control cells; Fig. 5A). After a 15 min incubation under the conditions necessary to visualize invasion, >90% of the probe remained resistant to back extraction (Fig. 5B).

In video microscope-based invasion studies of NBD-PS-labeled erythrocytes, NBD-PS fluorescence was seen to demarcate both the forming and the fully-formed PVM (Fig. 5C, upper row). Merozoites that were only partially internalized were not fluorescent (data not shown), again indicating that the fluorescent probe was initially incorporated into the PVM, and not the parasite. No detectable transfer of NBD-PS from labeled erythrocytes to attached merozoites was seen after 20 min at 37°C (Fig. 5C, t=20:38).

Quantitative analysis of the distribution of NBD-PS fluorescence during invasion showed that, like PKH26, the concentration of NBD-PS in the newly formed PVM was indistinguishable from the concentration in the erythrocyte membrane (Fig. 6A, B). Unlike PKH26, however, the amount of internalized NBD-PS steadily increased over the subsequent 10 minutes post-invasion (Fig. 6C, D); these differences in the post-invasion distribution of NBD-PS and PKH26 are shown graphically in Fig. 7. We were unable to determine by fluorescence microscopy whether the NBD-PS, which accumulated post-invasion was confined to the PVM, or was also incorporated into the parasite itself.

During the interaction between merozoites and PKH-labeled erythrocytes, we often observed patches of increased fluorescence on the erythrocyte surface at or near the point of merozoite attachment. This was most easily seen when cytochalasin B was used to inhibit internalization (Fig. 8 a/d, b/e; see also Fig. 1i), but it also occurred in the absence of cytochalasin B treatment (Fig. 8 c/f). Erythrocytes labeled with PKH2, PKH26, DiIC16 and NBD-PS all exhibited this behavior, as did erythrocytes labeled with the non-exchanging lipophilic probe, R18. The patches of increased fluorescence were present on erythrocytes labeled with both self-quenched and unquenched concentrations of R18 (data not shown), and they were seen with merozoites attached to Duffy-positive human erythrocytes, but not with merozoites attached to Duffy-negative human erythrocytes.

In contrast to our results with lipophilic membrane probes, it has been shown by freeze-fracture and immunoelectron microscopy that erythrocyte membrane proteins are excluded from the PVM (McLaren et al., 1977; Aikawa et al., 1981; Atkinson et al., 1987; Dluzewski et al., 1989). To confirm and extend these observations, we fluorescently labeled erythrocyte surface proteins and followed the fate of the labeled proteins during invasion, while simultaneously monitoring the fate of erythrocyte lipids using PKH2 or PKH26. Fluorescein-5-thiosemicarbazide was used to label sialic acid residues on the glycophorins (Fig. 9A, lane a’; see also Golan et al., 1986), and the amino-reactive reagents DTAF (Sheetz et al., 1980) and Texas Red sulfonyl chloride were used to label Band 3 (and several other minor proteins; Fig. 9A, lane b’). All three reagents gave qualitatively similar results: labeled erythrocyte surface proteins were largely excluded from the newly formed PVM (e.g. see Fig. 9B, d). Images of partially internalized merozoites (e.g. see Fig. 9B, e-h) reveal that the exclusion of proteins occurs during vacuole formation: the invasion pit, though stained with PKH2 (Fig. 9B, g), is largely free of labeled protein (Fig. 9B, h).

The PVM has not been isolated from malaria-infected erythrocytes and, as a consequence, little is known about its macromolecular composition. Immunoelectron microscopy and freeze-fracture studies have shown that erythrocyte membrane proteins are essentially absent from the PVM (McLaren et al., 1977; Aikawa et al., 1981; Atkinson et al., 1987; Dluzewski et al., 1989). It has been reported that the PVM is also largely depleted of host membrane lipids (Dluzewski et al., 1992). We studied the composition of the PVM by labeling the erythrocyte membrane with a variety of fluorophores and determining the distribution of these probes during invasion by fluorescence microscopy. In initial experiments with several fluorescent lipophilic probes, we found that the probes were associated with the internalized parasite shortly after invasion (Fig. 1c). A similar result was described by Haldar and Uyetake (1992).

It was not clear from these initial experiments how probe originally inserted into the erythrocyte membrane was internalized, i.e. through incorporation into the forming PVM or via transfer into the invading parasite itself. PKH2, PKH26, and DiIC16 reportedly undergo little or no intermembrane exchange once they have been incorporated into a membrane bilayer (Horan and Slezak, 1989; Slezak and Horan, 1989; Honig and Hume, 1989). However, we detected transfer of these probes from labeled rhesus or Duffy-positive human erythrocytes to attached merozoites after a 60 minute incubation at 37°C. A reduced rate of transfer was observed using labeled Duffy-negative human erythrocytes, and no detectable transfer was observed between labeled erythrocytes and free merozoites. A gap of approximately 10 nm separates the membrane of a Duffypositive erythrocyte from that of an attached merozoite, whereas the separation is approximately 150 nm for a Duffy-negative erythrocyte (Miller et al., 1979). These data suggest that while PKH2, PKH26, and DiIC16 are relatively non-exchanging, if two membranes are sufficiently tightly apposed some transfer can occur across the aqueous space between them. Furthermore, in many of our post-invasion images the internalized probe was not confined to a rim surrounding the newly invaded parasite, but exhibited a heterogeneous substructure. Haldar and Uyetake (1992) reported a similar observation, and demonstrated directly that at some time during or after invasion the probe was transferred to the parasite itself.

To resolve the issue of how the probes became internalized, we developed methods for following their distribution in real time during the process of invasion. The results showed that the probes are in fact internalized via the PVM during invasion (Fig. 2). In addition, we found that the concentration of probe in the erythrocyte membrane and in the forming PVM were essentially indistinguishable (Fig. 3). If the probes used accurately reflect the behaviour of endogenous erythrocyte lipids, these results demonstrate that the lipid composition of the PVM is similar to that of the erythrocyte membrane and suggest that the lipids of the PVM are derived directly from the host cell membrane during invasion.

It was recently reported that following invasion of NBD-PE-labeled human erythrocytes by P. falciparum mero-zoites, the PVM surrounding the young (<7 hour) intracellular trophozoite contains only 12% as much probe as the host erythrocyte membrane (Dluzewski et al., 1992). Furthermore, when the internalization of P. knowlesi merozoites was inhibited with cytochalasin B, an NBD-PE-depleted patch was seen in the erythrocyte membrane at the point of contact between the two cells. The distribution of NBD-PE was determined in these experiments by immunoelectron microscopy using an anti-NBD antibody. Since these data were at variance with our observations on the distribution of PKH26, PKH2, and DiIC16 during invasion, we attempted to follow, by video microscopy, the distribution of NBD-PE during invasion. We found that NBD-PE, originally inserted into the erythrocyte membrane, was associated with the young intracellular parasite shortly after invasion (Fig. 4c). We also observed marked transfer of probe from labeled erythrocytes to unlabeled merozoites in the suspension. Thus, in the case of an attached merozoite, we did not see the patches of erythrocyte membrane depleted of NBD-PE described by Dluzewski et al. (1992); we found that the attached merozoite was itself stained (Fig. 4f).

The reasons for the differences between our results and those of Dluzewski et al. are unknown; every attempt was made to reproduce exactly the labeling conditions used by them. The discrepancies are not due to differences between species: we observed internalized NBD-PE in P. falci - parum-infected human erythrocytes 7 hours after initiating infection (Fig. 4i), and transfer of probe from labeled erythrocytes to attached merozoites (Fig. 4f) was seen using either human or rhesus erythrocytes. It seems most likely that the discrepancies are due to differences in the methodologies used to visualize the probe. During the ethanol dehydration steps used to process the samples for immunoelectron microscopy, lipids are known to be extracted (Edwards et al., 1992); perhaps some difference between the (e.g. protein) composition of the PVM and the erythrocyte membrane results in more extensive extraction of NBD-PE from the PVM than from the erythrocyte membrane. The possible effects on probe distribution of incubation in iodoacetate and N-ethylmaleimide prior to fixation (used to reduce lipid extraction during sample processing; Dluzewski et al., 1992) are also unknown. Finally, the preparations of NBD-PE used in these studies are known to be heterogeneous in both the length of the non-fluorescent acyl chain and the distribution of the C6-NBD moiety (20% in the sn1 position, 80% sn2; Martin and Pagano, 1986). If one of the minor species present in the NBD-PE were more reactive with the anti-NBD antibody than the major fluorescent compound, the distribution of the probe as seen by immunoelectron microscopy could be different from that seen by fluorescence microscopy.

Our observation that NBD-PE could not be confined to labeled erythrocytes in the presence of other unlabeled cells was probably due to the inability of the erythrocyte to translocate this particular analog to the inner leaflet of its lipid bilayer (Colleau et al., 1991), thereby leaving it available for bilayer-bilayer exchange (Sandra and Pagano, 1979). In contrast, NBD-PS is efficiently translocated or ‘flipped’ to the inner leaflet of the erythrocyte bilayer (Col-leau et al., 1991; Connor et al., 1990; see Fig. 5A), and only slowly ‘flops’ back to the outer leaflet (Connor et al., 1990; see Fig. 5B). Consistent with these results, we found that transfer of probe from NBD-PS-labeled erythrocytes to unlabeled free merozoites or membrane vesicles was markedly reduced compared to the exchange from NBD-PE-labeled erythrocytes. Video microscopy studies of invasion using NBD-PS-labeled erythrocytes showed that the probe was present in both the forming and newly formed PVM, at a concentration similar to that present in the erythrocyte membrane (Fig. 6A, B). Thus, NBD-PS is not excluded from the forming PVM.

In contrast to PKH26, the intensity of NBD-PS fluorescence associated with the internalized parasite and/or PVM increased steadily for the first 10 minutes post-invasion (Fig. 7). Transfer of fluorescent probe from the erythrocyte membrane to the intracellular parasite and/or PVM could occur by selective pinocytosis (Haldar et al., 1989; Grellier et al., 1991), by monomer diffusion through the erythrocyte cytoplasm (Haldar et al., 1989), or by transport through the ‘tubovesicular network’ of the infected erythrocyte (Haldar, 1992). Both monomer diffusion and transfer into the tubovesicular network would be facilitated in the case of NBD-PS by its localization in the inner leaflet of the erythrocyte bilayer. Alternatively, NBD-PS internalization could occur along ‘parasitophorous ducts’ (Pouvelle et al., 1991), membrane-lined channels that have been proposed to directly connect the erythrocyte membrane to the PVM. The observation that PKH26, PKH2, and DiIC16 added to erythrocytes after invasion do not have access to young intraerythrocytic P. knowlesi (Fig. 1f, and data not shown) argues against both this interpretation and the existence of the parasitophorous duct. However, it is possible that these particular probes are differentially excluded from the duct, as PKH26 is also excluded from P. falciparum-infected erythrocytes under conditions that permit free access of other macromolecules to the intracellular parasite (T. Taraschi, personal communication).

The nature of the patches of increased fluorescence on the erythrocyte surface at or near the point of contact between a labeled erythrocyte and an attached merozoite (Fig. 8) is unknown. The phenomenon is not due to localized dequenching of probe fluorescence (Hoekstra et al., 1984), as it is observed in erythrocytes labeled with both quenched and unquenched concentrations of R18. Erythrocyte proteins concentrated in the junction between the erythrocyte and merozoite (Aikawa et al., 1981) may bring with them associated lipids (Sackmann et al., 1987), effectively increasing the local lipid concentration. Alternatively, secretions from the merozoite might create a microenvironment within the erythrocyte membrane that either locally ‘solubilizes’ more probe or changes its quantum efficiency. Finally, merozoites attached to erythrocytes induce within the erythrocyte cytoplasm membrane-lined tubes and vesicles, originating from the point of contact between the two cells (Miller et al., 1979); the localized patch of increased fluorescence may represent a low-resolution image of these parasite-induced structures.

Several models have been proposed for the mechanism by which the merozoite induces PVM formation (see Bannister and Dluzewski, 1990, and references therein). One model proposes that the PVM is derived from the parasite. In this model, the rhoptries contain a large store of lipid that is secreted in a controlled fashion and inserted into the erythrocyte plasma membrane during invasion; this newly inserted bilayer invaginates to form the PVM. This model is supported by ultrastructural observations of membranelike lamellar structures within the rhoptries and in rhoptry secretions (Bannister and Mitchell, 1989; Bannister et al., 1986; Stewart et al., 1986), the transfer of fluorescent lipid analogs (Mikkelsen et al., 1988) and rhoptry proteins (Sam-Yellowe et al., 1988) from the merozoite to the erythrocyte during invasion, and the absence in the PVM of erythrocyte membrane proteins (Fig. 9B; McLaren et al., 1977; Aikawa et al., 1981; Atkinson et al., 1987; Dluzewski et al., 1989).

Alternatively, the PVM may be formed during invasion by invagination of the erythrocyte membrane itself, rather than created de novo from material stored in the parasite. Endovesiculation can be induced in erythrocytes and erythrocyte ghosts by trypsin, MgATP, and a variety of amphipathic drugs (Schrier, 1987, and references therein). The parasite might similarly induce erythrocyte invagination by locally disrupting the erythrocyte cytoskeleton (Elgsaeter et al., 1976; Koslov et al., 1990; Saxton, 1992), by inserting small amounts of material into the inner leaflet of the erythrocyte lipid bilayer (Sheetz and Singer, 1974; Ferrell et al., 1985), or by some combination of the two. Without knowledge of the amount of lipid, protein, or lamellar material secreted from the rhoptries during invasion, the data mentioned above supporting a parasite-derived PVM are equally consistent with an erythrocyte-derived PVM, as secretion of some amount of material from the rhoptries might be necessary to disrupt the erythrocyte cytoskeleton or otherwise induce erythrocyte invagination. It is also important to note that in at least one other Apicomplexan parasite (Theileria parva), rhoptry discharge appears to occur after interiorization and PVM formation (Shaw et al., 1991).

Our observation that the lipid composition of the newly formed PVM is similar to that of the erythrocyte membrane would appear to argue against the idea that the PVM is formed via bulk insertion of rhoptry-derived lipids. However, since merozoite interiorization takes 10-20 seconds (Dvorak et al., 1975), it is possible that bulk insertion of lipids occurs but lateral diffusion of the fluorescent probe into the newly inserted bilayer occurs more quickly than new bilayer is inserted. This explanation would require free diffusion of lipids past the orifice of the invasion pit and into the forming PVM, which erythrocyte proteins are clearly unable to do (Fig. 9B;McLaren et al., 1977; Aikawa et al., 1981; Atkinson et al., 1987; Dluzewski et al., 1989). While plasma membrane barriers to protein diffusion, through which lipids pass freely, are not without precedent (Dragsten et al., 1981; van Meer and Simons, 1986), the simplest explanation for our data is that the PVM is formed by a parasite-induced invagination of the erythrocyte lipid bilayer and rearrangement of the erythrocyte cytoskeleton.

It has recently been suggested that bulk insertion of rhop-try-derived lipids into the host cell membrane during Tox - oplasma invasion might not only play a role in PVM formation, but also endow the PVM with properties favorable to survival of the parasite by giving it a lipid composition different from that of the host cell plasma membrane (Joiner, 1991). While our results can not definitively rule out bulk insertion of lipids as a possible mechanism of PVM formation, they do demonstrate that the lipid composition of the newly-formed PVM is similar to that of the erythrocyte plasma membrane, arguing against such a role for rhoptry-derived lipids in Plasmodium.

We thank Dr W. Gratzer for several helpful discussions and the sharing of unpublished data with us. We also thank Drs R. Pagano and T. Taraschi for helpful discussions, Dr S. Dolan for providing us with P. falciparum schizonts, and H. Cascio for providing assistance in television technology. We are grateful to Drs C. Plowe, Y. Raviv, I. Baines and S. Dolan for critical reading of the manuscript, and L. Lambert for assistance in animal care and handling.