Plasmodium falciparum merozoite surface protein 1 Glycosylation and localization to low-density, detergent-resistant membranes in the parasitized erythrocyte Daniel C. Hoessli 1 , Monique Poincelet 1 , Ramneek Gupta 2 , Subburaj Ilangumaran 3 and Nasir-ud-Din 4 1 Department of Pathology, Centre me ´ dical universitaire, Geneva, Switzerland; 2 Center for Biological Sequence Analysis, Technical University of Denmark, Lyngby, Denmark; 3 Department of Experimental Therapeutics, Ontario Cancer Institute, Toronto, Canada; 4 Institute of Biomedical Sciences, Pakistan and HEJ Institute of Chemistry, University of Karachi, Lahore, Pakistan In addition to the major carbohydrate moieties of the glycosylphosphatidylinositol (GPI) anchor, we report that Plasmodium falciparum merozoite surface protein 1 (MSP-1) bears O-GlcNAc modifications predominantly in b-ano- meric configuration, in both the C- and N-terminal portions of the protein. Subcellular fractionation of parasitized erythrocytes in the late trophozoite/schizont stage reveals that GPI-anchored C-terminal fragments of MSP-1 are recovered in Triton X-100 resistant, low-density membrane fractions. Our results suggest that O-GlcNAc-modified MSP-1 N-terminal fragments tend to localize within the parasitophorous vacuolar membrane while GPI-anchored MSP-1 C-terminal fragments associate with low-density, Triton X-100 resistant membrane domains (rafts), redis- tribute in the parasitized erythrocyte and are eventually shed as membrane vesicles that also contain the endogenous, GPI-linked CD59. Keywords: detergent-resistant membranes; malaria; mer- ozoite surface protein; O-GlcNAc modification; vesicles. In the blood-stage forms of the malarial parasite Plasmodium falciparum, the merozoite surface protein 1 (MSP-1) is a major surface component [1] that undergoes selective proteolytic processing and reassembly in preparation for erythrocyte invasion [1–4]. MSP-1 is linked to the parasite plasma membrane via a glycosylphosphatidylinositol (GPI) anchor [5], but the functional consequences of this mode of anchoring for the merozoite to interact with the erythrocyte have not been fully evaluated [6]. In addition to the GPI- anchor modification, MSP-1 also contains mono- or oligo- saccharides in O-linkage to serines or threonines [7–10]. N-linked carbohydrates have also been described in association with asparagines on MSP-1 [9], despite the reported lack of N-glycosylating machinery in P. falciparum parasites [11]. As P. falciparum merozoite maturation takes place within an intraerythrocytic network of modified (parasitophorous vacuolar membrane) and newly made (tubo-vesicular network) membranes [12,13], it is possible that parasite surface proteins also constitute substrates for carbohydrate-modifying enzymes of the erythrocyte. In normal erythrocytes, O-GlcNAc modifications of serines/ threonines in intracellular proteins occur in a manner reciprocal to phosphorylation [14] and O-GlcNAc addition is considered a widespread and general mechanism for protein modification [15]. In this study, we have analysed MSP-1 for the presence of O-GlcNAc-modified serines and threonines, using specific antibodies to map the biosynthe- tically labelled modifications to the N and C terminus of the MSP-1 protein [16]. The presence of O-GlcNAc on both the C- and N-terminal ends of MSP-1 was confirmed by exogalactosylation and two-thirds of the [ 3 H]GlcN label incorporated into MSP-1 was sensitive to Jack Bean b-N-acetylglucosaminidase, suggesting the presence of O-GlcNAc moieties in b-anomeric linkage [15]. Predictions for a-andb-anomeric O-GlcNAc sites in five known MSP-1 sequences were made using methods based on artificial neural networks which are competent in recognizing fuzzy sequence motifs, and two distinct sets of a-and b-O-GlcNAc sites have been predicted. The GPI-anchored 19-kDa C-terminal fragment was found associated with detergent-resistant, low-density membranes of the parasi- tized erythrocyte, suggesting that GPI-linked MSP-1 prod- ucts redistribute within the membrane network of the parasitized host cell aboard detergent-resistant membrane domains. Infected erythrocytes were also found to release membrane vesicles containing parasitic 19-kDa MSP-1 fragments and endogenous CD59, both GPI-linked proteins. Materials and methods Materials Anti-MSP-1 mAbs reactive with the C (3B10) and N terminus (7B2) [16], were obtained from J.A. Lyon (Walter Reed Army Institute of Research, Washington, USA). A human immune serum against blood stage antigens was used to detect all MSP-1 epitopes [10]. Anti-CD59 mAb Correspondence to D. C. Hoessli, Department of Pathology, Centre me ´ dical universitaire, 1, rue Michel-Servet, 1211 Geneva 4, Switzerland. Fax: +41 22 7025746, Tel.: +41 22 7025893, E-mail: Daniel.Hoessli@medecine.unige.ch Abbreviations: GPI, glycosylphosphatidylinositol; MSP-1, merozoite surface protein 1. (Received 9 September 2002, revised 13 November 2002, accepted 26 November 2002) Eur. J. Biochem. 270, 366–375 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03397.x MEM-43 was supplied by V. Horejsi (Academy of Sciences of the Czech Republic, Prague). Recombinant, bovine b-1,4-galactosyltransferase was from Calbiochem. b-galac- tosidase (bovine brain) and b-N-acetylglucosaminidase (Jack Bean) were from Sigma. Metabolic radiolabelling of M25 Zaire P. falciparum parasites Asexual blood stage parasites were cultured in the asyn- chronous mode in 10-mL cultures at 37 °C in a candle jar. The culture medium contained 5% v/v human group A+ erythrocytes, in RPMI medium supplemented with 10% ARh+ human serum (HD Supplies, Aylesbury, Bucks, UK) and 0.1% glucose. Labelling was carried out for 16 h with 50 lCiÆmL )1 D -[6- 3 H]glucosamine hydrochloride ([ 3 H]GlcN; 40 CiÆmmol )1 , American Radiolabeled Chemi- cals, supplied by Anawa, Wangen, Switzerland), as previ- ously carried out to show carbohydrate modification of plasmodial proteins [9,17]. At the end of the labelling time, culture supernatants were collected and centrifuged at low speed (2000 r.p.m., 5 min) to remove uninfected and parasitized erythrocytes. The supernatant was centrifuged sequentially at 15000 g for 10 min at 4 °C and the 15000 g supernatant again at 100 000 g for 30 min at 4 °C (Beckman SW41 rotor) to collect the released membrane vesicles. The 100 000 g pellet was analysed as a source of parasite-free membrane nanovesicles (Fig. 1A). To study the distribution of MSP-1 and its fragments, the [ 3 H]GlcN-labelled, parasitized erythrocytes at the late trophozoite/schizont stage were collected by sedimentation over a 70% Percoll (Pharmacia) gradient [18] and hemolysedinH 2 O in the presence of protease inhibitors (0.19 m M leupeptin, 0.17 m M chymostatin, 2 m M N-p-tosyl- L -lysine chloromethyl ketone (TLCK), 2 m M N-p-tosyl- L -phenylalanine chloromethyl ketone (TPCK), 1 m M phenylmethanesulfonyl fluoride and 1 m M ortho-phenanthro- line; Fig. 1B). A small aliquot (5%) of the [ 3 H]GlcN-labelled parasites and haemoglobin-free erythrocyte membranes was directly solubilized in SDS/PAGE sample buffer (see Fig. 2, [ 3 H]GlcN; tot), resolved on SDS/PAGE and the labelled proteins revealed by fluorography. The gel was soaked in Enlightning (NEN), dried and exposed to Hyperfilms (Amersham). The bulk of this material was resuspended in TKM buffer (50 m M Tris/HCl pH 7.4, 25 m M KCl, 5 m M MgCl 2 ,1m M EGTA) containing 1% Triton X-100, 36% sucrose and protease inhibitors, and centrifuged at 250 000 g in a SW 50 rotor for 16 h at 4 °C. This procedure allows the parasites to be pelleted with remnants of the parasitophorous vacuolar membrane [19] (P, Fig. 1B), and to be separated from Triton X-100 soluble and Triton X-100 resistant components of the parasitized erythrocyte membranes that are recovered in the supernatant (S, Fig. 1B). SDS/PAGE, immunoprecipitation and Western blotting In 10% of the supernatant material, the proteins were precipitated with chloroform/methanol for Western blotting analysis [20] with anti-(C-ter) or anti-(N-ter) mAbs. The precipitated proteins were separated on a 10% minigel Fig. 1. Subcellular fractionation protocols utilized to generate extracts containing parasite proteins from [ 3 H]GlcN-labelled and unlabelled, P. falciparum-infected erythrocytes. (A) Isolation of microvesicles and nanovesicles from [ 3 H]GlcN-labelled, infected erythrocytes. (B) Isola- tion of parasite extracts (P) and Triton X-100 extracted proteins (S) from [ 3 H]GlcN-labelled, infected erythrocytes. (C) Isolation of Triton X-100 insoluble membranes and Triton X-100 soluble membrane and cytosolic proteins from infected erythrocytes. Ó FEBS 2003 Glycosylation and membrane localization of MSP-1 (Eur. J. Biochem. 270) 367 and transferred to nitrocellulose (Hybond-C, Amersham Pharmacia Biotech) with a semidry blotting apparatus (Bio-Rad). After 2 h of blocking at room temperature in NaCl/Tris/Tween (10 m M Tris/HCl pH 7.4, 100 m M NaCl, 0.05% Tween 20) containing 5% low-fat, dry milk powder (NaCl/Tris/Tween/5% MP), the filters were incubated with antibodies in NaCl/Tris/Tween/5% MP overnight at 4 °C. Thoroughly washed filters were incubated with horseradish- peroxidase-conjugated secondary antibodies for 1 h at room temperature. Chemiluminescence development was carried out with the Immun-Star Pack reagents (Bio-Rad) and the filters exposed to X-Omat Kodak films. The bulk (90%) of the supernatant (S, Fig. 1B) was dialysed to remove sucrose and sequentially immunopre- cipitated with Sepharose 4B-coupled anti-(N-ter), followed by anti-(C-ter) mAb. The antibodies were covalently coupled to CNBr-activated Sepharose 4B beads according to the manufacturer’s instructions. Incubation with each mAb was carried out for 6–10 h at 4 °C on a rotating wheel, the antibody beads were washed in TKM/Triton X-100 containing protease inhibitors, and the bound antigens extracted with SDS/PAGE sample buffer. The [ 3 H]GlcN- labelled, immunoprecipitated proteins were revealed by fluorography, as described above. The parasite pellet (P, Fig. 1B) was extracted with 10% SDS (1 h at room temperature), 10% of the extract sampled for Western blotting and the remainder diluted with Triton X-100 and BSA to obtain a final concentration of 0.05% SDS, 0.5% Triton X-100, 10 lgÆmL )1 BSA, suitable for immunoprecipitation. Sequential immunoprecipitation with Sepharose-coupled anti-N followed by anti-(C-ter) mAb was carried out as with the supernatant (S) material and the [ 3 H]GlcN-labelled, immunoprecipitated proteins visualized by fluorography as described above. Fig. 3. Vesicular release of C-terminal MSP-1 fragments by parasitized erythrocytes. The [ 3 H]GlcN-labelled, parasitized erythrocytes in the culture were purified over Percoll and processed for immunoprecipi- tation as outlined in Materials and methods. Immunoprecipitation was carried out with Sepharose-coupled 3B10 mAb and human malaria- immune serum coupled to protein A/G beads, and the immunopre- cipitated, labelled bands revealed by fluorography (Whole extract: IP anti C-ter; IP immune serum). The supernatant of [ 3 H]GlcN-labelled cultures at the late trophozoite/schizont stage were allowed to settle and the supernatant collected. Remaining erythrocytes were sedi- mented at 2000 r.p.m. for 10 min. The resulting supernatant was centrifuged once at 16 000 g to remove pelletable parasites, parasitized erythrocytes and uninfected erythrocytes. The last supernatant was ultracentrifuged at 100 000 g and the membrane pellet (nanovesicles) wasresuspendedinTKM.One-tenthwasextractedwith10%SDS sample buffer (vesicles: unselected) and the [ 3 H]GlcN- labelled bands revealed by fluorography. The remaining 90% was divided into two aliquots: one aliquot was extracted with 10% SDS and diluted to 0.5% Triton X-100, 0.05% SDS, 10 lgÆmL )1 BSA for immunoprecipitation with Sepharose-coupled anti-MSP-1 C terminus (vesicles: IP anti-C-ter) and the labelled bands revealed by fluorogra- phy. The other aliquot was kept in TKM and incubated for 6–10 h with Sepharose-coupled anti-(C-ter). The immunoselected vesicles were washed in TKM and extracted with SDS/PAGE buffer, trans- ferred to nitrocellulose and probed with MEM-43 anti-CD59 mAb (vesicles: WB anti CD59) and revealed by chemiluminescence. Fig. 2. Both C-terminal and N-terminal fragments of MSP-1 biosyn- thetically incorporate [ 3 H]GlcN. The parasitized erythrocytes were isolated by Percoll gradient centrifugation and lysed in hypotonic buffer. The resulting parasites and membrane ghosts were extracted in TKM-1% Triton X-100/35% sucrose and ultracentrifuged to yield a pellet (P) of parasites and a supernatant (S) containing Triton X-100- resistant complexes and Triton X-100-soluble membrane proteins. [ 3 H]GlcN: fluorogram of [ 3 H]GlcN-labelled P. falciparum proteins in a total (tot), SDS extract of Percoll-purified parasitized erythrocytes following hemolysis. C-ter and N-ter: probing with the 3B10 (C-ter) or 7B2 (N-ter) mAb following transfer to nitrocellulose and detection by chemiluminescence (WB), or immunoprecipitation with Sepharose- bound mAb and fluorography (IP). Results shown are representative of three separate experiments. 368 D. C. Hoessli et al. (Eur. J. Biochem. 270) Ó FEBS 2003 One-tenth of the nanovesicle pellet (Fig. 1A) was directly solubilized in SDS/PAGE sample buffer, resolved on a 10% minigel and processed for fluorography (unselected, Fig. 3). The remainder of the nanovesicle pellet was divided into two aliquots. One aliquot was solubilized in 10% SDS and subsequently diluted with TKM/Triton X-100 and BSA to a final concentration of 0.5% Triton X-100, 0.05% SDS, 10 lgÆmL )1 BSA. The MSP-1 19 kDa C-terminal fragment was immunoprecipitated with Sepharose-coupled anti- (C-ter) mAb as described above. The bound MSP-1 fragments were eluted in SDS/PAGE sample buffer, resolved on a 10% minigel and processed for fluorography. The other aliquot was incubated in TKM with Sepharose 4B-coupled anti-(C-ter) mAb for 6–10 h at 4 °C(rotating wheel). The antibody-bound membranes were extracted in SDS/PAGE sample buffer, resolved on a 10% minigel, transferred to Hybond, probed with MEM-43 anti-CD59 mAb and visualized by chemiluminescence. Exogalactosylation and deglycosylation of MSP-1 Auto-galactosylated, recombinant b-1,4-galactosyltrans- ferase (20 mU, Calbiochem) was used to probe nitrocel- lulose-immobilized parasite proteins for nonreducing terminal GlcNAc residues [21], using UDP-[6- 3 H]galac- tose (40 CiÆmmol )1 , American Radiolabeled Chemicals) as galactose donor. The MSP-1 proteins were specifically immunoprecipitated from a 10% SDS extract of Percoll- purified parasitized erythrocytes. This SDS extract was diluted with Triton X-100 as described above for the parasite pellets, and incubated with Sepharose-coupled antibodies. Affinity-purified C- or N-terminal MSP-1 proteins were eluted from the solid-phase antibodies with SDS/PAGE sample buffer, electrophoretically separated and transferred to nitrocellulose. The presence of the C- and N-terminal fragments was confirmed by probing a parallel lane containing identically immunoprecipitated MSP-1 proteins with anti-C and anti-(N-ter) mAbs and the adjacent nitrocellulose lane containing the appropri- ate protein was cut and subjected to exogalactosylation. Cut nitrocellulose pieces corresponding to the 195 (whole MSP-1), 56 (C-terminal) and 86 (N-terminal) kDa MSP- 1 proteins (marked by asterisks in Fig. 2) were incubated with 20 mU autogalactosylated, recombinant galactosyl- transferase overnight at 37 °Cin0.1 M cacodylate buffer pH 7.2, with 100 l M MnCl 2 ,with1lCi UDP-[6- 3 H] galactose as galactose donor. After washing in 0.1 M citrate-phosphate buffer pH 4.3, degalactosylation was carried out on the labelled proteins with 10 mU b-galactosidase in the same buffer. Radioactivity in the exogalactosylated and the degalactosylated bands was counted in a liquid scintillation counter. Control exoga- lactosylation reactions included anti-C or anti-(N-ter), sham selected material from lysates of uninfected eryth- rocytes, and nitrocellulose-transferred BSA, that does not bear the O-GlcNAc modification and thus cannot be exogalactosylated. b-N-acetylglucosaminidase from Jack Bean (Sigma) was utilized to remove biosynthetically incorporated [ 3 H]GlcN on MSP-1 retained on Sepharose-coupled anti-(N-ter) or anti-(C-ter) mAbs. b-N-acetylglucosaminidase treatment of MSP-1 bound to antibodies was carried out as described [22]. The radioactivity remaining on the beads after b-N-acetylglucosaminidase or control b-galactosidase treat- ments was counted. Prediction of O-GlcNAc addition sites on MSP-1 protein Sequences for the Ghana-RO33, Png-MAD20, Uganda, Thai-K1 and Wellcome (Swiss-Prot accession no. P19598, P08569, P50495, P04932, P04933) isolates were aligned as recommended in [23] using the sequence editor Jalview (M. Clamp, unpublished data). Alignment for the C-ter- minal third of the Ghana-RO33 isolate was missing in [23], and this was performed manually. O-GlcNAc modified sites in the a-anomeric configuration were predicted using the DICTYOGLYC 1.1 prediction server http://www.cbs.dtu.dk/ services/ DICTYOGLYC /[24]andb-anomeric O-GlcNAc sites were predicted using the YinOYang 1.2 prediction server (R. Gupta, S. Brunak & J. Hansen, unpublished data) available at http://www.cbs.dtu.dk/services/YinOYang/. Both prediction methods are based on neural networks and incorporate a surface-accessibility derived threshold which makes it more probable for a predicted site to be on a surface exposed Ser/Thr in the protein. The design of these methods is similar to NetOGlyc, a successful predictor for O-GalNAc mucin type glycosylation sites [25]. The methods have been rigorously cross-validated and have at least one experimental verification for prediction of each type of linkage. DictyOGlyc, the O-a-GlcNAc predictor, was trainedonanin vivo set of secreted and membrane proteins of Dictyostelium discoideum,andtheO-b-GlcNAc predictor was trained on a set of intracellular eukaryotic (mostly mammalian) proteins. Predictions from the servers were then mapped onto the alignment. Equilibrium sucrose density gradient centrifugation of P. falciparum -parasitized erythrocytes Lysates of Percoll-purified, late trophozoites/schizonts in TKM/1% Triton X-100 were adjusted to 40% sucrose, placed at the bottom of a Beckman SW41 tube, overlaid with 6 mL 36% and 3.5 mL 5% sucrose in TKM buffer (Fig. 1C). Following centrifugation at 250 000 g for 16 h at 4 °C, 1-mL fractions were collected from the top. Equal volumes (50 lL) of the floating, detergent-resistant mem- branes containing GPI-linked proteins (fractions 3 and 4) and the Triton X-100 soluble proteins (fractions 5–10) were concentrated and analysed by Western blotting [20] as described above. The parasite pellet containing remnants of the parasitophorous membrane (fraction 11) was solubilized in SDS/PAGE sample buffer and a matching amount subjected to Western blotting. MSP-1 was detected with the anti-(C-ter) mAb and the erythrocyte surface molecule CD59, a GPI-linked complement defence protein, was detected with the MEM-43 mAb. Results MSP-1 is O-GlcNAc-modified in the N and C termini Fig. 2 compares the MSP-1 protein and fragments detected in extracts of [ 3 H]GlcN-labelled, parasitized erythrocytes by immunoprecipitation or Western blotting. The parasites Ó FEBS 2003 Glycosylation and membrane localization of MSP-1 (Eur. J. Biochem. 270) 369 recovered in the pelletable material of the Triton X-100/ 36% sucrose extract were contained within residual para- sitophorous vacuolar membranes [19]. The supernatant (S) contained both the Triton X-100 solubilized proteins and the Triton X-100 resistant complexes emanating from the membranes of the parasitized erythrocyte. The fluorograph- ic pattern of [ 3 H]GlcN-labelled proteins from the total SDS extract of the parasitized, hemolysed erythrocytes is shown for reference ([ 3 H]GlcN; tot). The 195-kDa MSP-1 protein was labelled in the total [ 3 H]GlcN extract and immunoprecipitated by both anti- (C-ter) and anti-(N-ter) mAbs in the parasite pellet as well as in the supernatant. Western blotting with the anti-(C-ter) mAb showed a higher ratio of intact MSP-1 to the 19-kDa fragment in the parasite pellet than in the supernatant suggesting that MSP-1 C-terminal 19-kDa peptides were preferentially found in the membrane network of the parasitized erythrocyte. One 100-kDa peptide bearing both N- and C-terminal epitopes was detected by both antibodies on Western blots of the pellet and supernatant. The 86-kDa, N-terminal specific peptide was detected by Western blotting and immunoprecipitated as a [ 3 H]GlcN-labelled fragment only in the parasite pellet. Likewise a further N-specific and [ 3 H]GlcN-labelled peptide of 40 kDa was also immunoprecipitated from the parasite pellet. The C-terminal specific peptides consisted of one group of three bands between 48 and 58 kDa, detectable by Western blotting and immunoprecipitated as [ 3 H]GlcN- labelled peptides. The other C-terminal peptide of 19 kDa formed a heterogeneous group of peptides between 10 and 19 kDa (on Western blot) and predominated in the supernatant. The electrophoretic heterogeneity of these C-terminal fragments is compatible with their being modi- fied by GPI anchors [26]. The immunoprecipitated 19 kDa protein was detectable only as a single 19-kDa band. The only strong [ 3 H]GlcN-labelledbandinthetotalextract matching the Western blotted material was a 17-kDa band. The majority of the [ 3 H]GlcN-labelled material ( 70% of the total label) ran between 5 and 10 kDa and did not comigrate with either Western blotted or immunoprecipi- tated material. It is likely that this fast-moving [ 3 H]GlcN- labelled material corresponds to the GPI-anchored peptides no longer associated with MSP-1 C-terminal epitopes. The incorporated 3 H-label in this 5–10 kDa material ran as glucosamine by paper chromatography (data not shown), indicating that [ 3 H]GlcN had not been chemically trans- formed. With both antibodies, immunoprecipitation of intact MSP-1195 kDa was more efficient than that of the fragments. On the contrary, Western blotting detected the fragments more efficiently. This probably reflects conform- ational differences between MSP-1 intact protein and its fragments in solution and adsorbed onto nitrocellulose. The detectability of [ 3 H]GlcN-labelled, immunoprecipitated fragments is therefore likely to be suboptimal. Low M r C-terminal fragments are released in membrane vesicles by parasitized erythrocytes The 5- to 17-kDa [ 3 H]GlcN-labelled material pelleted with in vitro released membrane vesicles (Fig. 3, unselected) corresponding to the nanovesicles released from normal erythrocytes following Ca ++ exposure [27]. This high-speed pellet of released vesicles contained labelled 5- to 10- and 17-kDa fragments, as well as a labelled 19-kDa fragment immunoprecipitable with the anti-(C-ter) mAb (Fig. 2; IP anti C-ter). The [ 3 H]GlcN-labelled MSP-1 fragments detectable in the released vesicles were predominantly of low M r . No intact MSP-1 protein and no other C- or N- terminal fragments were detected by immunoprecipitation in the released vesicles. Importantly, endogenous CD59 was detected by Western blotting in the released membrane vesicles immunoselected with solid-phase anti-MSP-1 C-ter mAb (Fig. 3, WB anti-CD59). The parasite extract from the same culture (whole extract) contained the full spectrum of MSP-1 protein and fragments, including [ 3 H]GlcN-labelled 86 and 40 kDa N-terminal fragments detected by a polyclonal antibody (immune serum) directed against MSP-1. The absence of intact MSP-1 in the vesicles strongly suggested that they were free of parasites (merozoites) and consisted only of membranes emanating from the parasi- tized erythrocyte. Moreover, the coexistence of MSP-1 19 kDa and CD59 in nanovesicle membranes selected with anti-(C-ter) mAb further suggests that MSP-1 and CD59 proteins are released in the same membrane vesicles from P. falciparum-infected erythrocytes. Analysis of the non-GPI-anchored carbohydrate moieties of MSP-1 It is therefore possible that part of the remaining protein- bound, non GPI-anchored [ 3 H]GlcN label could be incor- porated on the surface of the molecule. This contention was further supported by the observation that the 86- and 40-kDa N-terminal fragments which cannot carry the GPI anchor were strongly labelled fragments compared to the C-terminal ones. These results were confirmed by exo- galactosylation of the affinity-purified MSP-1 proteins transferred to nitrocellulose membranes. The specifically immunoprecipitated 195-kDa MSP-1, and 56-kDa C-ter- minal and 86-kDa N-terminal fragments (marked with asterisks in Fig. 2) were exogalactosylated with 3 H-UDP- Gal at levels significantly above control labelling of the non O-GlcNAc-modified BSA (Table 1). Sham immunopreci- pitations with uninfected erythrocyte lysates did not yield exogalactosylated material above the BSA control at 195, 86 and 56 kDa (data not shown). Specificity of the b-1,4- galactosyl transferase-mediated labelling was confirmed by removal of the incorporated label following treatment with b-galactosidase. Further, the Jack Bean b-N-acetylglucosa- minidase, an enzyme that specifically cleaves O-GlcNAc residues in b-anomeric linkage [22], released 65% of the [ 3 H]GlcN label from biosynthetically labelled proteins im- munoprecipitated with either 3B10 or 7B2 mAbs (compare with C-ter and N-ter immunoprecipitates of pellet, Fig. 2). Prediction of potential O-glycosylation sites on MSP-1 The potential for O-GlcNAc modification of five known and verified MSP-1 sequences (Ghana [28]; Uganda-Palo Alto; Papua New Guinea MAD 20; Thai K1 and Wellcome isolates [23]) was evaluated using the DictyOGlyc 1.1 predictor [24]. Fig. 4 (left panel) shows that Thr at position 1278 and Ser at 1280 (single cross), and Ser at positions 1498 and 1506 (asterisk) in the MSP-1 sequences of Ghana, 370 D. C. Hoessli et al. (Eur. J. Biochem. 270) Ó FEBS 2003 PnGMAD20 and Uganda isolates had the potential to be modified by a-GlcNAc. The allelomorphic sequences of the Wellcome and Thai-K1 strains bear deletions at these positions [23] and thus could not be evaluated. One Ser however (1353, single cross) had a potential close to the threshold line in the Thai and Wellcome sequences. All of these sites are located within the C terminus and correspond to the allelomorphic block 5–16 of the MSP-1 sequence [23] but do not encompass the GPI-anchored 19-kDa fragment. In contrast to the C-terminal region, the N-terminal region does not contain any predictable a-GlcNAc sites. Interestingly, screening for potential b-O-GlcNAc modi- fication sites revealed a wider and different set of sites (Fig. 4, right panel). b-O-GlcNAc sites occured both in the N-terminal, nonpolymorphic (nonallelomorphic) part of MSP-1 (blocks 1–4), as well as in the polymorphic (allelo- morphic) partof MSP-1 (blocks 5–16). In block 1–4,multiple threonines between the aligned positions 80–135 were detectable in the Uganda sequence (filled triangles). In the other four sequences, one cluster of threonines between 75 and 80 and another cluster of serines (135–145) were also predicted. Inblocks 5–16, serines at aligned positions 931 and 957 were positive in the Ghana, Png and Uganda sequences (filled triangles). The next positively predicted O-GlcNAc sites occured at the aligned positions 1271 (Ser), 1278 (Thr) and 1283 (Ser) (filled circles). Position 1271 was positive in the Thai and Wellcome sequences while position 1283 was positive for the Ghana, Png and Uganda sequences. Position 1278 was the only position of the alignment where both a-andb-O-GlcNAchavebeenpredictedintheGhana,Png and Uganda sequences. The Thr1503 was predicted positive for b (Ghana, Png and Uganda: filled circles, right panel), while Ser1498 and 1506 were positive for a (double cross, left panel). Thr1693 in block 17 was found positive for b only in the Png sequence and suggests that the 19 kDa C-terminal fragment does not usually contain an O-GlcNAc site in MSP-1. The M25 Zaı ¨ re MSP-1 shows substantial labelling in its N terminus and thus fits the b-O-GlcNAc predictions made on the Ghana, Papua and Uganda sequences. Distribution of MSP-1 fragments in Triton X-100 resistant and soluble membranes of the parasitized erythrocyte GPI-anchored membrane proteins favour the environment of ordered lipids [29] and accumulate in the low density, detergent-resistant membranes recovered after equilibrium density centrifugation of Triton X-100 lysates of mamma- lian cells [20]. Parasitized erythrocytes were isolated by sedimentation in Percoll, washed and hypotonically lysed to remove haemoglobin (Fig. 1C). The resulting ghosts were extracted in Triton X-100 before equilibrium density centrifugation in sucrose. In this gradient system (Fig 1C and Fig. 5), the Triton X-100 resistant membranes floated to the 5–36% sucrose interface (fractions 3–4) while the membrane proteins that lack strong interactions with membrane lipids were solubilized (fractions 5–10) and the parasites were pelleted with remnants of the parasitopho- rous vesicular membrane (fraction 11). The floating mem- branes from such a Triton X-100 extract of parasitized erythrocytes were enriched in the 19-kDa C-terminal fragments of MSP-1 and in the erythrocytic, GPI-linked CD59. The Triton X-100 soluble fractions of the gradient contained the bulk of the proteins displayed in the gradient, but only small quantities of MSP-1 19 kDa and CD59, suggesting that the two GPI-linked proteins seek similar lipid-rich membrane environments. Discussion MSP-1 not only ensures adhesion of newly released merozoites to fresh erythrocytes, but its C-terminal GPI- linked fragments also appear to redistribute in the parasi- tized erythrocyte in detergent-resistant membrane domains. During the initial phase of this process, vesicle-borne MSP-1 fragments [30] could come in contact with glycosyltrans- ferases present in the erythrocyte cytosol [31,32], or in intracellular membranes [33,34]. The intracellular localiza- tion of the glycosyltransferases that catalyse O-GlcNAc addition remains undefined and the O-GlcNAc transferase activity has been found in membrane-free reticulocyte lysates [32], as well as membrane-associated. Most O-GlcNAc-modified proteins are indeed cytoplasmic or nuclear [15], but are also found at the cell surface [35]. This implies that GPI-linked MSP-1 fragments exposed to the lumenal side of intracellular membranes such as the tubo- vesicular network [13,36] could become O-GlcNAc-modi- fied similarly to the O-GlcNAc-modified proteins found at thecellsurface[35]. Our evidence for carbohydrate modifications of MSP-1 other than the GPI anchor is based on the following: (a) [ 3 H]GlcN biosynthetic labelling occurs in both the C- and N-terminal fragments; (b) exogalactosylation of terminal Table 1. Exogalactosylation of MSP-1 protein and fragments. From an SDS extract of parasitized erythrocytes (four 10-mL culture plates) obtained following hypotonic lysis of Percoll-isolated erythrocytes containing late trophozoites and schizonts (P in Fig. 1), C- and N-terminal MSP-1 fragments were immunoprecipitated with 3B10 (anti-C-ter) or 7B2 (anti-(N-ter)) mAbs coupled to Sepharose. The 195- kDa whole protein, and the 56-kDa C-terminal and the 86 kDa N-terminal fragments were identified by immunoblotting of an aliquot of the immunoprecipitate run in parallel. The corresponding 195-, 86- and 56-kDa proteins were subjected to exogalactosylation in duplicate. The data presented are representative of three different experiments. Substrate Galactosyl transferase b-galacto- sidase c.p.m. 195 kDa 0 0 283 MSP-1 20 mU 0 1469 20 mU 0 1365 20 mU 10 mU 198 86 kDa 0 0 264 N-ter 20 mU 0 1354 20 mU 0 1607 20 mU 10 mU 264 56 kDa 0 0 210 C-ter 20 mU 0 1263 20 mU 0 1277 20 mU 10 mU 200 BSA, 2 lg0 0 80 20 mU 0 158 Ó FEBS 2003 Glycosylation and membrane localization of MSP-1 (Eur. J. Biochem. 270) 371 O-GlcNAc also occurs in both termini of the MSP-1 molecule; and (c) 65% of the incorporated [ 3 H]GlcN associated with MSP-1 in an SDS parasite extract is removed by Jack Bean glucosaminidase, an enzyme that releases O-GlcNAc moieties in b-anomeric configuration [22]. The remaining 35% of incorporated [ 3 H]GlcN that is resistant to the Jack Bean hexosaminidase could be either GPI-linked, N-linked to asparagines or linked to the surface of the protein in a-O-GlcNAc configuration. Two a-O-GlcNAc sites are indeed predicted in the Ghana, PngMAD20 and Uganda strain MSP-1 proteins and it is remarkable that the three b-O-GlcNAc sites predicted in the allelomorphic portion of MSP-1 (blocks 5–16) are distinct from the predicted a-O-GlcNAc sites (with the exception of Thr1278), and distinguish the two dimorphic forms (Ghana, Png and Uganda vs. Thai-K1 and Wellcome). Such b-O-GlcNAc sites are not localized in the regions of homology (387–413 and 1100–1187) within the sequences of blocks 5–16 of the two dimorphic forms. The comparison could not be extended to the a-O-GlcNAc sites because the MSP-1 proteins of Thai-K1 and Wellcome strains contain deletions [23] in the regions of MSP-1 where a-O-GlcNAc sites have been predicted. Our previous findings indicated that [ 3 H]GlcN is linked to serines and [17,37,38] threonines [7,10] and we now show that O-GlcNAc addition takes place in both the C- and N-terminal ends of the protein, while the GPI-anchor remains the major carbohydrate modification of MSP-1, as established by others [17,37,38]. However, following SDS/ PAGE separation of MSP-1 fragments, the labelled GPI anchors appear dissociated from the 19-kDa fragment carrying the C-terminal epitope (compare Figs 2 and 3). The lower M r [ 3 H]GlcN-labelled material amounts to  80% of the biosynthetically incorporated GlcNAc Fig. 4. a-GlcNAc and b-GlcNAc predictions on the aligned sequences of MSP-1 from Ghana (RO-33), Papua New Guinea (MAD20), Uganda (Palo Alto), Thailand (K1) and Wellcome strains. Sequences extracted from SwissProt (accession no. P19598, P08569, P50495, P04932, P04933) were aligned according to [23] and designated as G (Ghana), P (Png MAD20), U (Uganda), T (Thai) and W (Wellcome). Alpha- and b-GlcNAc site predictions, made using methods based on neural networks, were marked on the alignment. The x-axis shows the position of the alignment, and the y-axis marks the predicted potentials. The horizontal wavy line is a surface-accessibility derived threshold. A vertical impulse crossing the threshold is said to represent a (predicted) glycosylated site. While the GlcNAc linkages in the N-terminal half of the protein are probably entirely the b form, the C-terminal half has a mix of a and b forms. (Left) ÔXÕ marks represent potential a-GlcNAc positions. For Ghana, PnG and Uganda strains, out of the four potential positions 1278, 1280, 1498, 1506, 1278 may be a b-O-GlcNAc. (Right) The triangles and circles represent N- and C-terminal predicted b-O-GlcNAc positions. Empty circles and triangles are other ÔpossiblesÕ (negative predictions but very close to the threshold). The positions in the figure are alignment positions. Exact sequence positions may vary slightly from strain to strain. The prediction methods are available at http://www.cbs.dtu.dk/services/. 372 D. C. Hoessli et al. (Eur. J. Biochem. 270) Ó FEBS 2003 (5–10, 17 and 19 kDa), but only the [ 3 H]GlcN-labelled 19-kDa band is immunoprecipitated by the anti-(C-ter) mAb. The same mAb however, detects the 17-kDa fragment andpartofthelowerM r fragments by Western blotting, but still does not recognize the bulk of the [ 3 H]GlcN-labelled peptides between 5 and 10 kDa. Those [ 3 H]GlcN-labelled, lipophilic peptides are contained in sedimentable vesicular membranes (nanovesicles, see [27]) released by parasitized erythrocytes in culture. This lipophilic behaviour strongly suggest that MSP-1 C-terminal peptides carry [ 3 H]GlcN- labelled GPI anchors. Vesicular membranes enriched in GPI-anchored peptides could be a vehicle for the bioactive inositol glycan moieties released by P. falciparum parasites [39]. As the tyrosine kinase activity of macrophages was shown to respond to the hydrophilic, carbohydrate moieties of GPI molecules, released vesicles should exert their biological effect by simple contact, whereas protein kinase C enzymes were modulated only by acylated inositol glycans and would thus require fusion of the vesicles with the target cell membrane [39]. The low M r C-terminal MSP-1 fragments, like the endogenous GPI-linked CD59, are selectively enriched in low density, detergent-resistant membranes of the para- sitized erythrocytes, suggesting that the parasite GPI- anchored proteins seek a similar environment as the endogenous ones in the membranes of the parasitized erythrocyte. The presence of CD59 in nanovesicles immu- noselected with anti-(C-ter) mAb strongly suggests that both GPI-linked proteins are inserted in membrane subdo- mains of similar properties that vesiculate as a unit in form of nanovesicles. The Triton X-100 resistant membranes described in this study are most probably derived from erythrocytes and not from the parasite membranes, as intact parasites were removed by centrifugation prior to sucrose gradient float- ation. Sphingomyelin is synthesized de novo in the parasi- tized erythrocyte under the control of P. falciparum [40] and may also contribute to the formation of detergent-resistant membrane domains in the newly made membranes. The Triton X-100 resistant membranes originally described in erythrocyte ghosts were characteristically rich in sphingo- lipids and cytoskeletal proteins spectrin, actin and band 4.1 [41]. Using standard equilibrium sucrose density gradients, Civenni et al. have shown that GPI-linked surface proteins such as acetylcholinesterase, CD55 and CD59 are also included in the detergent-resistant membranes of normal erythrocytes [42] and released as vesicles by the stressed or aging erythrocytes [43]. The considerable remodelling of the cytoskeleton–membrane interface taking place in the para- sitized erythrocyte [44] makes it difficult to define precisely the relationship of the GPI-rich, detergent-resistant mem- branes we describe with the detergent-resistant membranes of normal erythrocytes. However, a recent study proposes that vacuolar uptake of erythrocyte components (CD59, Duffy antigen) could be carried out by membranes with detergent-resistant properties in the parasitized erythrocyte [19]. We further show in this study that membranes containing GPI-linked MSP-1 C-terminal 19-kDa fragment and endogenous CD59 are released in vesicular form by the cultured parasitized erythrocyte. The in vivo implication of this finding is that MSP-1 C-terminal antigens may disperse in the bloodstream and possibly integrate other cellular membranes [45]. Acknowledgements This work was supported by the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases Grant ID 970604 and Swiss National Science Foundation Grant 31–57696.99. R. G. thanks J. Hansen for useful discussions, and the Danish National Research Foundation for funding. We are grateful to Drs J. A. Lyon and V. Horejsi for their kind gifts of antibodies. References 1. Holder, A.A. (1988) The precursor to major surface antigens: structure and role in immunity. Progr. Allergy. 41, 72–97. 2. Blackman, M.J., Whittle, H. & Holder, A.A. (1991) Processing of the Plasmodium falciparum merozoite surface protein-1: identifi- cation of a secondary processing product which is shed prior to erythrocyte invasion. Mol. Biochem. Parasitol. 49, 35–44. 3. Blackman, M.J. & Holder, A.A. (1992) Secondary processing of the Plasmodium falciparum merozoite surface protein (MSP-1) by a calcium-dependent membrane-bound serine pro- tease: shedding of MSP-1 (33) as a noncovalently associated complex with other fragments of the MSP-1. Mol. Biochem. Parasitol. 50, 307–316. 4. Blackman, M.J., Chappel, J.A., Shai, S. & Holder, A.A. (1993) A conserved parasite serine protease processes the Plasmodium fal- ciparum merozoite surface protein-1. Mol. Biochem. Parasitol. 62, 103–114. Fig. 5. The MSP-1 C-terminal 19-kDa fragment and the endogenous CD59 distribute to low-density, detergent-resistant membranes. Percoll- purified parasitized erythrocytes were washed once in TKM and lysed in TKM containing 1% Triton X-100 and protease inhibitors. The lysate was adjusted to 40% sucrose and subjected to equilibrium gradient centrifugation as described in Methods. At equilibrium, 11 1-mL fractions were collected from the top. Each fraction was con- centrated and run on SDS/PAGE. After transfer to nitrocellulose the blots were probed with MEM-43 mAb against erythrocyte CD59 (top panel) or 3B10 mAb against the MSP-1 C terminus (bottom panel). Fractions 3 and 4 correspond to the 5–36% sucrose interface where detergent-resistant membranes accumulate. Fractions 5–10 contain solubilized proteins. Fraction 11 is the gradient pellet containing whole parasites and remnants of the parasitophorous membranes. Ó FEBS 2003 Glycosylation and membrane localization of MSP-1 (Eur. J. Biochem. 270) 373 5. Gerold, P., Schofield, L., Blackman, M.J., Holder, A.A. & Schwarz, R.T. (1996) Structural analysis of the glycosyl-phos- phatidylinositol membrane anchor of the merozoite surface pro- teins-1 and – 2 of Plasmodium falciparum. Mol. Biochem. Parasitol. 75, 131–143. 6. Gratzer, W.B. & Dluzewski, A.R. (1993) The red blood cell and malaria parasite invasion. Semin. Hematol. 30, 232–247. 7. Dayal-Drager, R., Hoessli, D.C., Decrind, C., Del Guidice, G., Lambert, P H. & Nasir-ud-Din (1991) Presence of O-glycosylated glycoproteins in the Plasmodium falciparum parasite, Carbohydr. Res. 209, c5–c8. 8. Dieckmann-Schuppert, A., Bause, E. & Schwarz, R.T. (1994) Glycosylation reactions in Plasmodium falciparum, Toxoplasma gondii,andTrypanosome brucei brucei probed by the use of syn- thetic peptides. Biochim. Biophys. Acta. 1199, 37–44. 9. Kimura,E.A.,Couto,A.S.,Peres,V.J.,Casal,O.L.&Katzin, A.M. (1996) N-linked glycoproteins are related to schizogony of the intraerythrocytic stage in Plasmodium falciparum. J. Biol. Chem. 271, 14452–14461. 10. Nasir-ud-Din, Drager-Dayal, R., Decrind, C., Hu, B.H., Del Giudice, G. & Hoessli, D.C. (1992) Plasmodium falciparum synthesizes O-glycosylated glycoproteins containing O-linked N-acetylglucosamine. Biochem. Int. 27, 55–64. 11. Dieckmann-Schuppert, A., Bender, S., Odenthal-Schnittler, M., Bause,E.&Schwarz,R.T.(1992)ApparentlackofN-glycosyla- tion in the asexual intraerythrocytic stage of Plasmodium falci- parum. Eur. J. Biochem. 205, 815–825. 12. Lingelbach, K. & Joiner, K.A. (1998) The parasitophorous vacuole membrane surrounding Plasmodium and Toxoplasma:an unusual compartment in infected cells. J. Cell Sci. 111, 1467–1475. 13. Haldar, K. (1996) Sphingolipid synthesis and membrane forma- tion by Plasmodium. Trends Cell Biol. 6, 398–405. 14. Holt, G.D., Haltiwanger, R.S., Torres, C R. & Hart, G.W. (1987) Erythrocytes contain cytoplasmic glycoproteins. J. Biol. Chem. 262, 14847–14850. 15. Hart, G.W. (1997) Dynamic O-linked glycosylation of nuclear and cytoskeletal proteins. Ann.Rev.Biochem.66, 315–335. 16. Lyon, J.A., Haynes, J.D., Diggs, C.L., Chulay, J.D., Haidaris, C.G. & Pratt-Rossiter, J. (1987) Monoclonal antibody char- acterization of the 195-kilodalton major surface glycoprotein of Plasmodium falciparum malaria schizonts and merozoites: identi- fication of additional processed products and a serotype-restricted repetitive epitope. J. Immunol. 138, 895–901. 17. Gowda, D.C., Gupta, P. & Davidson, E.A. (1997) Glycosylphos- phatidylinositol anchors represent the major carbohydrate modification in proteins of intraerythrocytic stage Plasmodium falciparum. J. Biol. Chem. 272, 6428–6439. 18. Dluzewski, A.R., Ling, I.T., Rangachari, K., Bates, P.A. & Wilson, R.J. (1984) A simple method for isolating viable mature parasites of Plasmodium falciparum from cultures. Trans. R. Soc. Trop. Med. Hyg. 78, 622–624. 19. Lauer,S.,VanWye,J.,Harrison,T.,McManus,H.,Samuel,B.U., Hiller, N.L., Mohandas, N. & Haldar, K. (2000) Vacuolar uptake of host components, and a role for cholesterol and sphingomyelin in malarial infection. EMBO J. 19, 3556–3564. 20. Ilangumaran, S., Briol, A. & Hoessli, D. (1997) Distinct interac- tions among GPI-anchored, transmembrane and membrane associated intracellular proteins, and sphingolipids in lymphocyte and endothelial cell plasma membranes. Biochim. Biophys. Acta. 1328, 227–236. 21. Parchment,R.E.,Ewing,C.M.&Shaper,J.H.(1986)Theuseof galactosyltransferase to probe nitrocellulose-immobilized glyco- proteins for non-reducing terminal N-acetylglucosamine residues. Anal. Biochem. 154, 460–469. 22. Previato, J.O., Sola-Penna, M., Agrellos, O.A., Jones, C., Oeltmann, T., Travassos, L.Z. & Mendonca-Previato, L. (1998) Biosynthesis of O-N-acetylglucosamine-linked glycans in Trypa- nosoma cruzi. J. Biol. Chem. 273, 14982–14988. 23. Miller, L.H., Roberts, T., Shahabuddin, M. & McCutchan, T.F. (1993) Analysis of sequence diversity in the Plasmodium falciparum merozoite surface protein-1 (MSP-1). Mol. Biochem. Parasitol. 59, 1–14. 24. Gupta, R., Jung, E., Gooley, A.A., Williams, K.L., Brunak, S. & Hansen, J. (1999) Scanning the Dictyostelium discoideum pro- teome for O-linked GlcNAc glycosylation sites using neural net- works. Glycobiology 9, 1009–1022. 25. Hansen, J., Lund, O., Tolstrup, N., Gooley, A.A., Williams, K.L. & Brunak, S. (1998) NetOGly: prediction of mucin type O-gly- cosylation sites based on sequence context and surface accessi- bility. Glycoconjugate J. 15, 115–130. 26. Hooper, N.M. & Turner, A.J. (1988) Ectoenzymes of the kidney microvillar membrane: differential solubilization by detergents can predict a glycosylphosphatidylinositol membrane anchor. Biochem. J. 250, 865–869. 27. Salzer, U., Hinterdorfer, P., Hunger, U., Borken, C. & Prohaska, R. (2002) Ca ++ -dependent vesicle release from erythrocytes involves stomatin-specific lipid rafts, synexin (annexin VII), and sorcin. Blood 99, 2569–2577. 28. Certa, U., Rotmann, D., Matile, H. & Reber-Liske, R. (1987) A naturally-occurring gene encoding the major surface antigen precursor P190 of Plasmodium falciparum lacks tripeptide repeats. EMBO J. 6, 4137–4142. 29. Brown, D.A. & London, E. (1998) Structure and origin of ordered lipid domains in biological membranes. J. Membrane Biol. 164, 103–114. 30. Hibbs, A.R. & Saul, A.J. (1994) Plasmodium falciparum:highly mobile small vesicles in the malaria-infected red blood cell cyto- plasm. Exp. Parasitol. 79, 260–269. 31. Chakraborty, A., Saha, D., Bose, A., Chatterjee, M. & Gupta, N.K. (1994) regulation of eIF-2 alpha-subunit phosphorylation in reticulocyte lysate. Biochemistry 33, 6700–6706. 32. Starr, C. & Hanover, J.A. (1990) Glycosylation of nuclear pore protein p62. J. Biol. Chem. 265, 6868–6873. 33. Haltiwanger, R.S., Holt, G.D. & Hart, G.W. (1990) Enzymatic addition of O-GlcNAc to nuclear and cytoplasmic proteins. Identification of a uridine diphospho-N-acetyglucosamine: peptide beta-N-acetylglucosaminyltransferase. J. Biol. Chem. 265, 2563– 2568. 34. Capasso, J.M., Abeijon, C. & Hirschberg, C.B. (1988) An intrinsic membrane glycoprotein of the Golgi apparatus with O-linked N-acetylglucosamine facing the cytosol. J. Biol. Chem. 263, 19778– 19782. 35. Torres, C R. & Hart, G.W. (1984) Topography and polypeptide distribution of terminal N-acetylglucosamine residues on the surfaces of intact lymphocytes. J. Biol. Chem. 259, 3308–3317. 36. Lauer, S.A., Rathod, P.K., Ghori, N. & Haldar, K. (1997) A membrane network for nutrient import in red cells infected with the malaria parasite. Science. 276, 1122–1125. 37. Gowda, D.C. & Davidson, E.A. (1999) Protein glycosylation in the malaria parasite. Parasitol. Today. 15, 147–152. 38. Kimura, E.A., Katzin, A.M. & Couto, A.S. (2000) More on protein glycosylation in the malaria parasite. Parasitol. Today 16, 38–39. 39. Tachado, S.D., Gerold, P., Schwarz, R., Novakovic, S., McCon- ville, M. & Schofield, L. (1997) Signal transduction in macro- phages by glycosylphosphatidylinositols of Plasmodium, Trypanosoma,andLeishmania: Activation of protein tyrosine kinases and protein kinase C by inositolglycan and diacylglycerol moieties. Proc.NatlAcad.Sci.USA94, 4022–4027. 40. Lauer, S.A., Ghori, N. & Haldar, K. (1995) Sphingolipid synthesis as a target for chemotherapy against malaria parasites. Proc. Natl Acad. Sci. USA 92, 9181–9185. 374 D. C. Hoessli et al. (Eur. J. Biochem. 270) Ó FEBS 2003 41. Yu, J., Fischman, D.A. & Steck, T.L. (1973) Selective solubli- zation of proteins and phospholipids from red cell membranes by nonionic detergents. J. Supramol. Struc. 1, 233–247. 42. Civenni, G., Test, S.T., Brodbeck, U. & Butikofer, P. (1998) In vitro incorporation of GPI-anchored proteins into human erythrocytes and their fate in the membrane. Blood 91, 1784–1792. 43. Butikofer,P.,Kuypers,F.A.,Xu,C.M.,Chiu,D.T.Y.&Lubin,B. (1989) Enrichment of two glycosyl-phosphatidylinositol-anchored proteins, acetylcholinesterase and decay accelerating factor, in vesicles released from human red blood cells. Blood. 74, 1481– 1485. 44. Deitsch, K.W. & Wellems, T.E. (1996) Membrane modifications in erythrocytes parasitized by Plasmodium falciparum. Mol. Bio- chem. Parasitol. 76, 1–10. 45. Ilangumaran, S., Robinson, P.J. & Hoessli, D.C. (1996) Transfer of exogenous glycosylphosphatidylinositol (GPI)-linked molecules to plasma membranes. Trends Cell Biol. 6, 163–167. Ó FEBS 2003 Glycosylation and membrane localization of MSP-1 (Eur. J. Biochem. 270) 375 . Plasmodium falciparum merozoite surface protein 1 Glycosylation and localization to low-density, detergent-resistant membranes in the parasitized erythrocyte Daniel. from the top. Equal volumes (50 lL) of the floating, detergent-resistant mem- branes containing GPI-linked proteins (fractions 3 and 4) and the Triton X -10 0