INTRODUCTION

Malaria, a life-threatening disease caused by parasitic protozoa of the genus Plasmodium, is a major health problem throughout the tropical and subtropical regions of the world. Among the four species that infect humans, Plasmodium falciparum is the most virulent. New approaches such as vaccine development and novel therapeutic agents are urgently needed due to the emergence of parasite strains resistant to chloroquine and other commonly used drugs (1).

Vaccines based on antigens of the blood stage parasite are under intensive study. A major focus has been on synthetic peptides or recombinant proteins of cell surface antigens (2-6). However, this approach has not been highly effective, although immunization with native cell surface proteins purified from the erythrocytic stage parasite is known to confer significant protective immunity (7-10). It is plausible that post-translational modifications play an important role in antigenicity. Accordingly, a basic understanding of such modifications may assist in the development of effective vaccines.

Glycosylation is an often extensive post-translational modification of eukaryotic proteins. Carbohydrate moieties of glycoproteins perform a variety of functions including modulation of immunological properties, receptor-ligand interactions, sorting and localization of proteins, cell adhesion, and cell-cell communication (11). In addition, they contribute to protein conformation and, thus, to proteolytic processing (11). Finally, the carbohydrate moieties can be highly antigenic and may contribute to disease pathology.

Several proteins of erythrocytic stage P. falciparum are known to contain carbohydrates (12-23). The deduced amino acid sequences of the parasite proteins have potential N-glycosylation sites (7, 24-26). However, Dieckmann-Schuppert et al. (27) have reported that P. falciparum has no N-glycosylation capability based on their finding that dolichol phosphate-linked oligosaccharides, the obligatory biosynthetic intermediates, and dolichol phosphate-oligosaccharyltransferase activity were not detectable in the parasite. However, while this work was in review, Kimura et al. (28) reported the existence of N-linked glycans in P. falciparum.

Two laboratories have previously reported that O-glycosylation is the major carbohydrate modification in proteins of the intraerythrocytic asexual stage of P. falciparum (21, 23) and that the O-linked carbohydrates are mainly single residues of GlcNAc with the remainder being oligosaccharides, some containing GlcNAc at the reducing end. While one group reported that the parasite proteins also contain O-linked GalNAc (19, 21), another group showed the absence of this sugar moiety and the lack of its biosynthesis (23). Kimura et al. reported the presence of O-glycanase-releasable oligosaccharides as the major carbohydrates in the proteins of late trophozoite and schizont stage P. falciparum (28).

Previous studies reported the presence of glycosylphosphatidylinositol (GPI)¹ in several proteins of erythrocytic stage P. falciparum including merozoite surface protein 1 (MSP-1), merozoite surface protein 2 (MSP-2), 72-kDa heat shock protein (HSP-72), 102-kDa transferrin receptor, and 75-kDa serine protease (29-32). Recently, Gerold et al. (33) studied the GPI lipids (not anchored to proteins) of P. falciparum. Two putative GPI anchor precursors, ethanolamine-phosphate-(Man1-2)6Man1-2Man1-6Man1-4GlcN-PI and ethanolamine-phosphate-6Man1-2Man1-6Man1-4GlcN-PI, were identified (33). While the current study was in progress, Gerold et al. (34) reported the structures of the GPI anchors of P. falciparum MSP-1 and MSP-2. Both were shown to consist of ethanolamine-phosphate-(Man1-2)6Man1-2Man1-6Man 1-4Glc4GlcN1-phosphatidylinositol, bearing a myristic acid substitution on the inositol ring and predominantly palmitic acid in the diacylglycerol moiety (34).

In this study, using several P. falciparum strains, we clearly demonstrate that O-linked carbohydrates are either absent or present only at very low levels in intraerythrocytic P. falciparum proteins. In agreement with Kimura et al. (28), we show that P. falciparum apparently contains N-linked carbohydrates. However, in contrast to the former study (23), our results establish that N-linked oligosaccharides are minor constituents, and that GPI anchors represent the major carbohydrate modification in intraerythrocytic P. falciparum proteins. Furthermore, we establish that GPI anchors of the parasite proteins have Man₄-GlcN cores with substituents that are susceptible to the conditions of nitrous acid deamination on the terminal Man residues. These unusual structures may contribute to the reported cellular and immunological functions of P. falciparum GPI anchors (35, 36).

EXPERIMENTAL PROCEDURES

Aspergillus saitoi -mannosidase (400 milliunits/mg), jack bean -mannosidase (30 units/mg), bovine milk GalT, and a mixture of Glc oligomers were from Oxford Glycosystems (Rosedale, NY); PMSF and (4-amidinophenyl)-methylsulfonyl fluoride hydrochloride were from Boehringer Mannheim; N-glycanase was from New England Biolabs (Beverly, MA); [1-³H]ethanolamine hydrochloride (10-30 mmol), L-[6-³H]Fuc (60-90 Ci/mmol), D-[6-³H]GlcN hydrochloride (40-60 Ci/mmol), D-[6-³H]Gal (40-60 Ci/mmol), D-[6-³H]Man (20-30 Ci/mmol), UDP-[6-³H]Gal (40-60 Ci/mmol), and UDP-[6-³H]GlcNAc (40-60/mmol) were from American Radiolabeled Chemicals (St. Louis, MO); rainbow protein molecular weight markers, rainbow ¹⁴C-methylated protein molecular weight markers, and Amplify^TM fluorographic solution were from Amersham Corp.; Bio-Gel P-4 (fine and extra fine), AG 50W-X12 (H⁺), AG 50W-X16 (H⁺), and AG 4-X4 (base) were from Bio-Rad; HPTLC plates coated with silica gel 60 on glass plates and aqueous HF (52%) were from EM Science (Gibbstown, NJ); chicken ovalbumin (grade V), leupeptin, pepstatin A, chymostatin, antipain, TLCK, TPCK, AMP, benzamidine, iodoacetamide, and sodium borohydride were from Sigma; EN³HANCE^TM fluorographic spray was from DuPont NEN; Percoll was from Pharmacia Biotech Inc.; human erythrocytes (blood group O) were from the American Red Cross (Baltimore, MD); human serum was from Interstate Blood Bank Inc. (Memphis, TN); RPMI 1640 medium was from Life Technologies, Inc. Man1-2Man1-2Man1-6Man1-4AHM (Man₄-AHM) and a mixture containing 2,5-anhydromannitol (AHM), Man1-4AHM (Man-AHM), Man1-6Man1-4AHM (Man₂-AHM), Man1-2Man1-6Man1-4AHM (Man₃-AHM) and Man₄-AHM were a generous gift from Dr. Alvaro A. Serrano and were derived from the GPI anchors of mucin-like glycoproteins of Trypanosoma cruzi (37).

Asexual blood stage P. falciparum parasite clone FCR-3 (obtained from Gambian isolates) was provided by Dr. Isabella Quakyi (Department of Biology, Georgetown University). Clone D6 (isolated from Sierra Leone I/CDC isolate) and clone W2 (from a 50:50 mixed culture of Indochina III/CDC and Sierra Leone I/CDC isolates) (38) were provided by Dr. Dennis E. Dyke (Walter Reed Army Institute of Research, Washington, D. C.). Parasite strain NF54 (isolated from a patient in Amsterdam infected with an African strain; Ref. 39) was provided by Dr. Stephen Hoffman (Naval Medical Research Institute, Bethesda, MD).

Parasite strains were maintained as asynchronous continuous cultures in RPMI 1640 medium supplemented with 25 mM HEPES, 25 mM sodium bicarbonate, 0.5% hypoxanthine, and 10% human serum at 3-4% hematocrit. Cultures were incubated at 37 °C in an atmosphere of 90% N₂, 5% O₂, and 5% CO₂. Parasite growth was monitored by counting the infected erythrocytes in Giemsa-stained thin blood smears under light microscopy.

Parasite cultures at 12-14% parasitemia were synchronized to the ring stage by treating with 5% sorbitol for 5 min (40) and then washed and suspended in complete medium. Parasitemia was adjusted to 10% by the addition of fresh erythrocytes, and the cells were incubated at 3% hematocrit for 16-18 h. The parasites were then metabolically labeled with ³H-sugars (50 µCi/ml) in medium (10 ml) containing 5 mM D-Glc and 10% human serum for 6-8 h. To label the parasites at different developmental stages, cultures were treated with [³H]GlcN for 16 h after synchronization with sorbitol and for 6-8 h each after the synchronized cultures were maintained in complete medium for 18, 24, 30, and 36 h. Labeling with [³H]ethanolamine (50 µCi/ml) was carried out in complete medium.

After metabolic labeling, erythrocytes were harvested by centrifugation at 1500 × g and washed with RPMI 1640 medium. The cells were suspended in RPMI 1640 medium, layered onto 70% isotonic Percoll, and centrifuged at 2700 × g for 20 min (41). The infected erythrocytes in the Percoll layer were recovered, washed, and then treated with 50 mM Tris-HCl, 100 mM NaCl, 5 mM EDTA, pH 8.0, containing 2% Nonidet P-40, 1 mM (4-amidinophenyl)-methylsulfonyl fluoride hydrochloride, 0.2 mM leupeptin, 0.2 mM chymostatin, 0.2 mM TPCK, 0.2 mM TLCK (infected erythrocyte pellet:lysis buffer, 1:20 (v/v)) at 4 °C for 20 min. The cell lysates (Percoll-enriched infected erythrocytes from 10 ml of culture) were centrifuged at 3500 × g at 4 °C for 10 min and then at 100,000 × g at 4 °C for 30 min. The supernatants were concentrated to 200-250 µl in a Speed-Vac and stored at 70 °C.

SDS-PAGE (42) was performed with 5-20%, 1.5-mm-thick, polyacrylamide gradient gels. The ³H-labeled parasite cell lysates were mixed with an equal volume of 125 mM Tris-HCl, pH 6.8, containing 4% SDS and 20% glycerol, heated in a boiling water bath for 5 min, and electrophoresed. After the run, the gels were treated with MeOH, water, glacial HOAc (50:40:10, v/v/v) for 1 h, washed with water for 5 min, and soaked in Amplify^TM fluorographic solution for 1 h. The gels were dried under vacuum at 65 °C and exposed to x-ray film at 70 °C.

The ³H-labeled parasite protein bands in the SDS-polyacrylamide gels, visualized by salicylic acid-enhanced fluorography (43), were excised, and the filter paper backing was scraped off. The procedure outlined here is for a single protein band. For large scale isolation, corresponding protein bands from different lanes were combined, and reagents were appropriately scaled up. The gels were cut into ~1-mm pieces and suspended in water (1 ml). The excess water was removed, and the gel was washed with water (3 × 1 ml) to remove the radioactivity enhancer (44). In some instances, water-swollen gel pieces were washed with MeOH (3 × 1 ml), which also removes the enhancer effectively. The water-swollen gel pieces were treated with an equal volume of 100 mM NaOH, 1 M NaBH₄, diluted to 2 ml with 50 mM NaOH, 0.5 M NaBH₄, and incubated at 44 °C for 24 h (45). The reaction mixture was cooled in an ice bath and neutralized with 2 M HOAc, and the clear solution was removed. The gel pieces were washed with water (5 × 2 ml) and then with MeOH (3 × 2 ml). The combined reaction solution and washes were centrifuged at 10,000 × g to remove insoluble particles. The supernatant was then dried in a rotary evaporator at 35 °C. Boric acid was removed by evaporation with 0.1% HOAc in MeOH (3 × 4 ml). The residue was dissolved in water and chromatographed on Bio-Gel P-4 (see below). Typically 20,000-30,000 cpm of [³H]GlcN-labeled carbohydrates were recovered from the 200-215-, 82-, and 75-kDa protein bands.

The protein bands in the SDS-polyacrylamide gels were cut into small pieces, washed with water and MeOH, and then dried in a Speed-Vac. The gels were suspended in water (0.5 ml) and homogenized to a fine paste using a glass minihomogenizer. The pastes (from two bands of the same protein) were suspended in 4 ml of 100 mM Tris-HCl, 1 mM CaCl₂, pH 8.0, containing 0.05% SDS and 0.5% Nonidet P-40, and incubated with Pronase (5 mg; 1-mg aliquots were added at intervals of 8-12 h) at 55 °C for 60 h. The enzyme digests were centrifuged, and the gels were washed with water (3 × 2 ml) and then with MeOH (2 × 2 ml). The combined supernatants and washings were dried in a rotary evaporator at 30 °C. The residues were dissolved in water (0.5 ml) and then either chromatographed on Bio-Gel P-4 (1 × 90 cm) in 100 mM pyridine, 100 mM HOAc, pH 5.2, or extracted with water-saturated 1-butanol (4 × 0.5 ml) (46-49). The 1-butanol layers were extracted with water (2 × 0.5 ml), and the organic layers containing the GPI anchor were dried. About 20,000-26,000 cpm of GPI anchors were obtained from a major [³H]GlcN-labeled protein band.

The parasite cultures at early and late trophozoite stages were labeled with [³H]GlcN for 10-18 h, and the infected erythrocytes were enriched by centrifugation on a Percoll cushion as described above. The cells were extracted three times with CHCl₃, MeOH, water (10:10:3, v/v/v) to remove glycolipids including free GPI-anchors. The pellet was dissolved in 25 mM Tris-HCl, pH 7.5, containing 1% SDS, 10 mM benzamidine, 5 mM EDTA, 1 mM PMSF, 1 µg/ml pepstatin A, 1 µg/ml leupeptin, 2 µg/ml antipain, and 1 µg/ml chymostatin, and dialyzed (M_r cut-off of 6,000-8,000) extensively for 48 h against the above buffer followed by dialysis for 1 h against distilled water containing protease inhibitors. The lysates were lyophilized, washed with 90% MeOH, and then subjected to 1) alkaline -elimination, 2) digestion with Pronase, or 3) digestion with N-glycanase.

The parasite cell lysate (~120,000 cpm) was dissolved in 50 mM NaOH, 0.5 M NaBH₄ (2 ml), incubated at 44 °C for 22 h, and then neutralized with 2 M HOAc in an ice bath. The solution was dried, boric acid was removed with MeOH, the residue was dissolved in water, and then the solution was chromatographed on Bio-Gel P-4 (1 × 90 cm) in 0.1 M pyridine, 0.1 M HOAc, pH 5.2. Fractions containing radioactivity were pooled and lyophilized.

The [³H]GlcN-labeled parasite lysate (~110,000 cpm) was suspended in 300 µl of 0.5% SDS, 1% -mercaptoethanol and then heated in a boiling water bath for 10 min and cooled. Then 0.5 M sodium phosphate, pH 7.5 (35 µl), and 10% Nonidet P-40 (35 µl) were added. The suspension was sonicated in an ultrasonic bath for 10 min to obtain a clear solution, and then N-glycanase (400 IUB milliunits) was added and incubated at 37 °C for 36 h. Four volumes of MeOH were added, and the solution was chilled on dry ice for 10 min and then centrifuged. The supernatant was dried in a Speed-Vac, and the residue was dissolved in water and extracted with water-saturated 1-butanol to remove the detergents. The aqueous phase was chromatographed on Bio-Gel P-4 (1 × 90 cm) in 0.1 M pyridine, 0.1 M HOAc, pH 5.2. The radioactive fractions were pooled and lyophilized, and their carbohydrate compositions were determined before and after reduction with NaBH₄.

The [³H]GlcN-labeled parasite lysate (~110,000 cpm) was suspended in 100 mM Tris-HCl, 1 mM CaCl₂, pH 8.0 (0.5 ml) and incubated with Pronase (4 × 1 mg, added at 10-12-h intervals) at 37 °C for 48 h. The solution was extracted with water-saturated 1-butanol (3 × 0.5 ml), and the two phases were separately dried in a Speed-Vac. The residue from the 1-butanol phase was analyzed for radiolabeled carbohydrates by HPAEC before and after treatment with HN0₂. The residue from the aqueous phase was dissolved in water and chromatographed on Bio-Gel P-4 (1 × 90 cm) in 0.1 M pyridine, 0.1 M HOAc, pH 5.2. The eluted radioactive components were analyzed for carbohydrates.

Erythrocytes harboring trophozoite and schizont stage parasites were enriched by centrifugation on a Percoll cushion and lysed with 0.015% saponin (18). The released parasites were washed with PBS and then lysed with 25 mM HEPES, pH 7.3, containing 0.05% SDS, 1% Nonidet P-40, 0.1 mM PMSF, 1 µg/ml leupeptin, 1 µg/ml pepstatin A, 1 µg/ml chymostatin, 2 µg/ml antipain, and 5 mM benzamidine. To the parasite cell lysate containing 100 µg of protein was added 20 µl of 100 mM HEPES, pH 7.3, containing 100 mM Gal, and 50 mM MnCl₂, 100 milliunits of GalT, 2 µl of 100 × protease inhibitor stock solutions. The solution was diluted with water (180 µl), 5 µCi of UDP-[³H]Gal in water (20 µl) containing 25 mM AMP was added, and the solution was incubated at 37 °C for 2 h (50). The reaction was stopped by the addition of 100 mM EDTA, 10% SDS (20 µl). Aliquots of the reaction mixture containing 30-40 µg of parasite proteins were analyzed by SDS-PAGE fluorography using a 7-20% SDS-polyacrylamide gradient gel (42).

Crude erythrocyte ghosts were prepared by the lysis of human red cells with 1:20 diluted phosphate-buffered saline for 30 min, centrifuged, and washed with the same buffer. The pellet was suspended in 25 mM HEPES, 1% Nonidet P-40, pH 7.3. The detergent-solubilized human erythrocyte ghosts and ovalbumin were galactosylated in parallel as positive controls and analyzed by SDS-PAGE fluorography.

The analysis was performed as described by Dieckmann-Schuppert et al. (23). The saponin-released parasites were lysed in ice-cold water containing 1 µg/ml leupeptin, 2 µg/ml antipain, 1 µg/ml pepstatin A, 1 µg/ml chymostatin, 1 mM PMSF, and 5 mM iodoacetamide. To the lysate was added an equal volume of 25 mM Tris-HCl, pH 7.2, containing 5 mM MgCl₂, 5 mM MnCl₂, 0.8% Triton X-100. The parasite cell lysate (200 µl) corresponding to 10⁸ parasites was incubated with 2-5 µCi of UDP-[³H]GlcNAc and 1 mM synthetic peptide Pro-Tyr-Thr-Val-Val at 37 °C for 30-60 min. Control incubations not containing the synthetic peptide were carried out in parallel. The reaction mixture was diluted with 0.9 ml of ice-cold water and then deionized on a column of Ag 1-X 8 (Cl). The effluents and washings were concentrated, extracted with 1-butanol to remove the detergent, and then chromatographed on Bio-Gel P-2 (0.9 × 50 cm) in 0.1 M pyridine, 0.1 M HOAc, pH 5.2. Fractions (0.5 ml) were collected, and the radioactivity was measured by liquid scintillation counting.

The peptide Pro-Tyr-Thr-Val-Val was custom synthesized by Research Genetics (Huntsville, AL). The crude peptide was over 95% pure by HPLC and mass spectral analysis; it was further purified by gel filtration on Bio-Gel P-2 (0.9 × 50 cm). Fast atom bombardment mass spectrometry analysis indicated that the purified peptide was over 98% pure.

Gel filtration of radioactive components released from proteins by Pronase digestion or treatment with NaOH/NaBH₄ was performed on columns (1 × 90 cm) of Bio-Gel P-4 (fine) equilibrated with 100 mM pyridine, 100 mM HOAc, pH 5.2. Fractions (1.1 ml) were collected, and aliquots were assessed for ³H by liquid scintillation counting. The glycan cores, isolated by treating the parasite protein GPI anchors with 50% aqueous HF, HNO₂, and NaBH₄, were similarly chromatographed. Fractions containing radiolabeled carbohydrates were pooled, lyophilized, and deionized with AG 50W-X12 (H⁺) and AG 4-X4 (base).

The GPI anchors, isolated by Pronase digestion (20,000-30,000 cpm) of protein gel bands, were treated with 100 µl of 25 mM NaOH overnight at 37 °C. The solutions were neutralized with an equal volume of 100 mM HOAc and analyzed on columns (1 × 90 cm) of Bio-Gel P-4 in 100 mM pyridine, 100 mM HOAc, pH 5.2.

The [³H]GlcN-labeled GPI anchors (~10,000 cpm) were dissolved in 100 µl of aqueous sodium bicarbonate (100 mg/ml) and cooled in an ice bath. Acetic anhydride (4.5 µl) was added in three equal aliquots at 10-min intervals, and the solutions were warmed to room temperature. After 30 min, the reaction mixtures were deionized with AG 50W-X16 (H⁺) and then dried in a rotary evaporator.

The [³H]GlcN-labeled GPI anchors obtained by the Pronase digestion and extraction with 1-butanol were treated, before and after N-acetylation, with 25 mM sodium periodate in 50 mM NaOAc, pH 5.5 (200 µl) for 16 h at room temperature in the dark. Excess periodate was destroyed by the addition of glycerol (10 µl), and the products were chromatographed on a column of Bio-Gel P-4 in 100 mM pyridine, 100 mM HOAc, pH 5.2. Fractions containing radioactivity were pooled, lyophilized, and then hydrolyzed with 4 M HCl for hexosamine analysis.

The ³H-labeled GPI anchors (30,000-40,000 cpm), isolated from protein bands either by NaOH/NaBH₄ treatment or digestion with Pronase, were treated with 50% aqueous HF (50 µl) in an ice bath for 48 h (46-49). The acid was neutralized with frozen saturated LiOH and centrifuged, and the supernatant was chromatographed on Bio-Gel P-4 columns (1 × 90 cm) in 100 mM pyridine, 100 mM HOAc, pH 5.2.

The [³H]GlcN-labeled GPI anchors (30,000-150,000 cpm), isolated from protein bands either by NaOH/NaBH₄ treatment or digestion with Pronase, were treated with 0.2 M NaOAc, pH 3.8 (75 µl) and 1 M NaNO₂ (75 µl) (46-49). After 18 h at room temperature, saturated boric acid (60 µl) was added, and the pH of the solution was adjusted to 10-11 with 2 M NaOH. Immediately, 1 M NaBH₄ in 100 mM NaOH (100 µl) was added and allowed to react at room temperature for 5-6 h. Excess NaBH₄ was destroyed by acidifying to pH 5 with cold 2 M HOAc in an ice bath, the solution was dried in a rotary evaporator, and boric acid was removed by repeated evaporation with MeOH. The samples were chromatographed on columns of Bio-Gel P-4 (1 × 90 cm) in 100 mM pyridine, 100 mM HOAc, pH 5.2.

The GPI anchors (30,000-40,000 cpm) isolated from gel bands were first dephosphorylated and then deaminated as above (46-49). The samples were desalted using AG 50W-X16 (H⁺) and AG 4-X4 (base) resins.

The GPI anchor glycan cores from 200-215- and 82-kDa proteins (20,000-30,000 cpm) isolated as above were dried in reactive vials and peracetylated with 40 µl of pyridine, acetic anhydride (1:1, v/v) at room temperature for 18 h (47-49). The solutions were dried in a Speed-Vac and then treated with 30 µl of acetic anhydride, HOAc, concentrated sulfuric acid (10:10:1, v/v/v) for 8 h at 37 °C (47-49). To each reaction mixture, 10 µl of pyridine and 500 µl of water were added and then extracted with CHCl₃ (250 µl). The organic layers were washed with water, dried, and treated with 30% ammonia (200 µl), MeOH (1:1, v/v) at 37 °C for 24 h. The solutions were dried in a Speed-Vac, and the residues were dissolved in water and then chromatographed on Bio-Gel P-4 in 100 mM pyridine, 100 mM HOAc, pH 5.2.

The [³H]GlcN-labeled GPI anchors (60,000-100,000 cpm) isolated by NaOH/NaBH₄ treatment of 200-215- and 82-kDa protein bands were deaminated with HNO₂ and reduced with NaBH₄. The samples were dissolved in 0.1 M trifluoroacetic acid (400 µl), heated at 100 °C for 4 h (46, 49), and then dried in a Speed-Vac. About half of the partial hydrolysates were treated with 25 µl (30 units/ml) of jack bean -mannosidase, deionized with AG 50W-X16 (H⁺) and then dried. The enzyme-treated and untreated partial hydrolysates were dephosphorylated with 50% aqueous HF and then deionized with AG 50W-X16 (H⁺) and AG 4-X4 (base) resins.

For hexosamines and hexosaminitols analysis, the [³H]GlcN-labeled parasite carbohydrate moieties (2000-4000 cpm) were hydrolyzed with 3 M HCl at 100 °C for 6 h and then dried in a Speed-Vac. For sialic acid analysis, samples were hydrolyzed with 50 mM sulfuric acid at 80 °C for 50 min, neutralized with 0.2 M NaOH. The hydrolysates were mixed with appropriate standard sugars and analyzed with a Dionex BioLC HPLC system coupled to pulsed amperometric detection using a CarboPac PA1 high pH anion exchange column (4 × 250 mm) or a CarboPac MA1 high pH anion exchange column (4 × 250 mm) (51). The eluents were as follows: 1) 20 mM NaOH at a flow rate of 0.8 ml/min, for hexosamines on CarboPac PA1; 2) 100 mM NaOH, 150 mM NaOAc at a flow rate of 0.8 ml/min, for sialic acids on CarboPac PA1; and 3) 100 mM NaOH for 10 min and then a gradient elution to increase the concentration of NaOH to 600 mM over a period of 10 min (flow rate, 0.4 ml/min), for hexosaminitols on CarboPac MA1. Elution of radioactive sugars was monitored by liquid scintillation counting. ³H-Sugars in samples were identified by either coelution or comparison of elution time with standard sugars. Nonradioactive standard sugars were detected by pulsed amperometric detection.

The neutral glycans of the parasite GPI anchors, their partial hydrolysates, and the products of exoglycosidase digestion and partial acetolysis were analyzed on columns of Bio-Gel P-4 (extra fine, 1 × 115 cm) in water at 60 °C at a flow rate of 2 ml/h (46-48, 52). Fractions (0.5 ml) were collected, and radioactivity was measured by liquid scintillation counting. The column was calibrated with ³H-labeled Glc oligomers, standard Man₄-AHM, and a mixture containing AHM, Man-AHM, Man₂-AHM, Man₃-AHM, and Man₄-AHM. The GPI glycans were characterized by size comparison with Glc oligomers and by comparison of retention times with those of standards. Glc was routinely used as an internal standard, and the elution was monitored with the phenol sulfuric acid reagent.

The neutral glycans of the GPI anchors and the products of exoglycosidase digestion were also analyzed on CarboPac PA1 (4 × 250 mm) using the Dionex HPLC system (46-48). Elution was with 100% of 100 mM NaOH and 0% of 100 mM NaOH, 250 mM NaOAc at sample injection and then a linear gradient to 40% of 100 mM NaOH, 250 mM NaOAc at 35 min and then maintenance of 40% of 100 mM NaOH, 250 mM NaOAc for 10 min. Fractions (0.4 ml) were collected, and the radioactivity was measured. Glc oligomers were added to samples as internal standards, and were monitored by pulsed amperometric detection. The [³H]GlcN-labeled parasite glycans were identified by comparison of elution time with those of Glc oligomers and standard AHM, Man-AHM, Man₂-AHM, Man₃-AHM, and Man₄-AHM.

The glycan cores of the parasite GPI anchors, their partial hydrolysates, and the products of exoglycosidase digestion and partial acetolysis were also analyzed by HPTLC (46, 49). The plates were developed with 1-propanol, acetone, water (9:6:4, v/v/v), dried, sprayed with EN³HANCE, and then exposed to x-ray film at 70 °C. The parasite GPI glycans were identified by comparison of retention times with those of Man₄-AHM standard and an AHM to Man₄-AHM ladder.

Treatment with Jack Bean -Mannosidase

GPI glycans were treated with 25-50 µl of jack bean -mannosidase (30 units/ml) in 100 mM NaOAc, pH 5.0, containing 2 mM Zn²⁺ at room temperature for 2 h and then at 37 °C for 22 h (46, 47, 49). The solutions were heated in a boiling water bath for 5 min and desalted either with AG 50W-X16 (H⁺) resin or by gel filtration on Bio-Gel P-4 columns (1 × 90 cm) in 100 mM pyridine, 100 mM HOAc, pH 5.2.

Treatment with A. saitoi -Mannosidase

The GPI glycans were incubated with 20 µl of A. saitoi -mannosidase (1 milliunit/ml) in 100 mM NaOAc, pH 5.0, at 37 °C for 22 h (46, 47, 49). The solutions were heated in a boiling water bath for 5 min and desalted with AG 50W-X16 (H⁺) and AG 4-X4 (base) resins.

RESULTS

Intraerythrocytic stage P. falciparum was metabolically labeled with [³H]GlcN, [³H]Man, [³H]Gal, or [³H]Fuc in medium containing 5 mM D-Glc and 10% human serum. SDS-PAGE analysis of the parasitized erythrocyte cell lysates and fluorographic identification of the protein bands demonstrate that [³H]GlcN was incorporated into more than 15 proteins (size range, 14-260 kDa) (Fig. 1). [³H]Man gave a similar labeling pattern, but the incorporation of radioactivity was only about 5-10% of that observed with [³H]GlcN (data not shown). [³H]Gal and [³H]Fuc were not significantly incorporated into the parasite proteins (data not shown). Since the incorporation of [³H]Man to the parasite proteins was low, subsequent metabolic labeling was routinely performed with [³H]GlcN.

Metabolic labeling with [³H]GlcN was performed at different developmental stages of the parasite (rings undergoing transformation to schizonts) (Fig. 1). Among the radiolabeled parasite proteins (Fig. 1A), 200-215-, 82-, and 75-kDa proteins were predominant. Two proteins (36 and 53 kDa) were labeled relatively intensely by the ring stage parasite, and their intensity drastically decreased in the trophozoites and schizonts. The incorporation of [³H]GlcN by the schizonts was significantly lower (Fig. 1A, lanes 6 and 7) compared with the rings and trophozoites (Fig. 1A, lanes 1-5). At least four radiolabeled proteins (38, 43, 46, and 56 kDa, some not well resolved on the gel) appear to be synthesized only at the trophozoite and schizont stages (Fig. 1A, compare lanes 2-7 with lane 1). A 260-kDa protein band was labeled to a significant level only by the ring stage parasites (Fig. 1A, lane 1). Several minor proteins with molecular weights ranging from 130,000 to 200,000 were also labeled with [³H]GlcN. In contrast to a previous report (23), noninfected red cells did not incorporate radiolabeled precursors (Fig. 1B, lane 4); >99% of the incorporated radioactivity was parasite-dependent. Previously, Udeinya and Van Dyke (18) and others (28) reported that noninfected erythrocytes do not incorporate [³H]GlcN into proteins.

To determine whether the carbohydrate modification of P. falciparum proteins differs between strains, three other parasite strains, W2, D6, and NF54, were also metabolically labeled with [³H]GlcN. SDS-PAGE of the cell lysates and fluorography revealed similar protein labeling patterns to that observed with the FCR-3 strain, with the exception of the altered mobility of the 75- and 82-kDa proteins (Fig. 1, compare panel A, lane 1, with panel B). In the W2 strain, the mobility of these two proteins was slower than in the FCR-3 strain, and they electrophoresed as two distinct, widely separated bands (Fig. 1B, lane 2). In the D6 strain, the mobility of the 75- and 82-kDa proteins (Fig. 1B, lane 1) was comparable with the respective proteins of the FCR-3 strain. These proteins appear as a single band in the NF54 strain (Fig. 1B, lane 3). The difference in mobility of 75- and 82-kDa proteins from different parasite strains is not due to altered glycosylation, since they have similar carbohydrate moieties (see below).

After SDS-PAGE of the parasite lysates and fluorography, the individual [³H]GlcN-labeled protein gel bands were separately excised and washed with water to remove the radioactive enhancer. More than 98% of the radiolabeled proteins remained in the gel slices. Pronase digestion (200-215-, 82-, and 75-kDa proteins from the FCR-3 strain) or alkaline borohydride treatment (all radiolabeled proteins from four parasite strains) released more than 95% of the radiolabeled carbohydrate moieties from the gel slices. These were isolated by gel filtration using Bio-Gel P-4 (Figs. 2 and 3). Approximately 95% of the Pronase-released carbohydrate eluted in the void volume along with the detergents from the reaction mixture, and the remainder was eluted at a 1200-1600-Da range (Fig. 2). However, after treatment with mild alkali or HNO₂, all of the radioactivity was eluted at a 1200-1600-Da range (Fig. 2).

The radiolabeled carbohydrates released by NaOH/NaBH₄ treatment of 200-215-, 43-46-, and 14-kDa protein bands eluted on Bio-Gel P-4 as single symmetrical peaks corresponding to a molecular weight of 1200-1600 (Fig. 3). In contrast, the radiolabeled carbohydrates from the 36-38-, 53-56-, 75-, and 82-kDa protein bands (FCR-3 strain) eluted as broad (in some cases heterogeneous) peaks at a 1200-2400-Da range (Fig. 3C). Upon digestion with Pronase, however, all of the samples were eluted as symmetrical peaks at a molecular weight range similar to that from the 200-215-kDa protein (Fig. 3D), suggesting that the observed higher sizes were due to peptide moieties associated with the carbohydrates. Therefore, the release of radiolabeled carbohydrates from parasite proteins by NaOH/NaBH₄ is not due to classical -elimination of Ser/Thr-linked O-glycosidic bonds but apparently to alkaline degradation of protein moieties. The [³H]GlcN-labeled peptide-carbohydrate moieties obtained from various proteins of the D6, W2, and NF54 strains were also analyzed (data not shown). In all cases, the results were similar to those from the FCR-3 strain. In no case did significant radioactivity elute at the hydrodynamic volume corresponding to hexosamines or hexosaminitols.

The [³H]GlcN-labeled carbohydrate moieties released by Pronase digestion of the parasite protein bands (200-215, 82, and 75 kDa from the FCR-3 strain at different erythrocytic stages) were partitioned between water and water-saturated 1-butanol. Approximately 95% of the radioactivity was extracted into the organic phase, indicating that the carbohydrate moieties contain hydrophobic substituents. However, after mild saponification, the Pronase-released carbohydrates remained in the aqueous phase, suggesting that the hydrophobic moieties are acyl esters. Further, the carbohydrate moieties released from the protein bands (by treatment with Pronase or alkaline borohydride) bind almost quantitatively to anion exchange resin (data not shown). This binding was abolished upon treatment with 50% aqueous HF, which removes phosphate ester groups. Thus, phosphate ester group(s) are presumed to be present. The parasite carbohydrate moieties also quantitatively bind to a cation exchange resin. This binding was abolished upon N-acetylation or deamination with HNO₂, indicating the presence of free amino groups.

Dionex-HPAEC analysis of the acid hydrolysates of the carbohydrates (Table I), isolated from the [³H]GlcN-labeled 200-215-, 82-, and 75-kDa parasite proteins (separately from each of the FCR-3, D6, W2, and NF54 strains) after Pronase digestion, showed the presence of GlcN but neither sialic acid nor galactosamine. Compositional analysis of the carbohydrate moieties isolated by alkaline borohydride treatment of the major [³H]GlcN-labeled parasite proteins (from the FCR-3, D6, W2, and NF54 strains) also gave only GlcN; glucosaminitol and galactosaminitol were not detected (Table I). In contrast to previous reports (19-23, 28), these results demonstrate that the previously reported O-linked carbohydrates are either absent or present at very low levels in the parasite proteins.

Table I. Hexosamine composition and the nature of carbohydrate moieties of SDS-polyacrylamide-separated P. falciparum proteins P. falciparum-infected erythrocytes were metabolically labeled with [3H]GlcN, and the parasite-infected erythrocytes were enriched by centrifugation on Percoll and then lysed. The lysates were separated on 5-20% SDS-polyacrylamide gradient gels, and the radiolabeled proteins were visualized by fluorography. The gel bands were excised, and the carbohydrate moieties were isolated as described under "Experimental Procedures." Protein band Relative proportions of the [3H]GlcN-labeled carbohydrate moietiesa Hexosamine detectedb Nature of the [3H]GlcN-labeled carbohydrate moiety detected 200-215 kDa 28.5 GlcN GPIc 82 kDa 33.8 GlcN GPIc 75 kDa 26.6 GlcN GPId 53-56 kDa 3.1 GlcN GPId 43-46 kDa 2.2 GlcN GPId 36-38 kDa 3.3 GlcN GPId 14 kDa 2.5 GlcN GPId a Approximate relative proportions of the radioactive carbohydrate moieties recovered after treatment of the gel bands with NaOH/NaBH4 and Bio-Gel P-4 chromatography. b By HPAEC analysis of the acid hydrolysates of the carbohydrates isolated from SDS-polyacrylamide gel bands (from FCR-3, D6, W2, and NF54 strains) and by sensitivity to HNO2. c The microsequencing of the glycan core was carried out in detail using all the three analytical techniques described under "Results." The locations of the ethanolamine-phosphate and the unidentified substituents were also studied. d The sugar sequence of the glycan core was studied only using jack bean -mannosidase and A. saitoi -mannosidase.

The [³H]GlcN-labeled carbohydrate moieties of the parasite proteins (14, 36-38, 43-46, 53-56, 75, 82, and 200-215 kDa from the FCR-3 strain) released by alkaline borohydride were treated with HNO₂ and then analyzed for hexosamines (Table I). More than 90% of the GlcN was sensitive to HNO₂, indicating that the carbohydrate moieties of parasite proteins contain nonacetylated GlcN (Table I). Although unlikely, it is possible that the GlcN could have been derived from an N-acetylglucosaminyl moiety by de-N-acetylation during the treatment with alkaline borohydride. Therefore, the carbohydrate moieties of the parasite proteins released by the Pronase digestion were similarly deaminated and then partitioned between water and water-saturated 1-butanol. After deamination, ~90% of the radioactivity remained in the aqueous phase, and the remainder partitioned into the 1-butanol layer. A retreatment of the radioactivity in the organic phase with HNO₂ partitioned almost all of the carbohydrate moieties into the aqueous phase. GlcN was barely detectable after deamination of the carbohydrate moieties of parasite proteins. In all parasite proteins analyzed, GlcN was converted into 2,5-AHM (identified as 2,5-anhydromannitol; see below) after treatment with HNO₂. These results indicate that the carbohydrate moieties of parasite proteins contain predominantly GlcN. This finding was confirmed for all P. falciparum strains used in this study.

The presence of GlcN in the carbohydrates of parasite proteins was further supported by periodate oxidation studies. Dionex-HPAEC analysis of the acid hydrolysates of periodate-oxidized products revealed that the GlcN residues were almost quantitatively destroyed. However, upon N-acetylation, the GlcN residues were completely resistant to periodate oxidation.

The predominance of GlcN with little or no N-acetylhexosamines in all of the predominantly radiolabeled parasite proteins (from FCR-3, D6, W2, and NF54 strains) suggests that GPI anchors are the major carbohydrate moieties of the parasite proteins. To verify whether this is the case, the parasites were metabolically labeled with [³H]ethanolamine. SDS-PAGE of the cell lysates and fluorography gave a labeling pattern of proteins similar to that observed with [³H]GlcN labeling (data not shown). The Bio-Gel P-4 chromatography of the [³H]ethanolamine-labeled components released from the protein gel bands, either by Pronase or alkaline borohydride, gave elution patterns similar to the carbohydrate moieties obtained from [³H]GlcN-labeled parasite proteins (data not shown).

Although N- and O-linked carbohydrates were not detected in the individual parasite proteins separated by SDS-PAGE, it is possible that they are present in very low proportions compared with GPI moieties. Therefore, entire P. falciparum cell lysates were analyzed for these carbohydrate moieties. The [³H]GlcN-labeled parasite lysates were extensively dialyzed to remove all free [³H]GlcN and then precipitated with MeOH. The later step ensures the removal of detergents and any free [³H]GlcN remaining in the lysate. The total parasite proteins were then analyzed for GPI anchors and N- and O-linked carbohydrates by the following three procedures.

Treatment of the whole parasite proteins with alkaline borohydride quantitatively converted the metabolically labeled radioactivity into lower molecular weight components. Upon chromatography on Bio-Gel P-4, these were eluted as broad heterogeneous peaks (Fig. 4A) with ~90% of the radioactivity eluting at volumes similar to the elution of GPI anchors obtained from the SDS-polyacrylamide gel bands. The eluted radioactivity was pooled together as shown in Fig. 4A and analyzed for carbohydrates. [³H]GlcNAcol or N-[³H]acetylgalactosaminitol were not detected in fractions A₁, A₂, and A₃, by HPAEC analysis, either before or after acid hydrolysis and re-N-acetylation. However, after acid hydrolysis, ~20-30% and <5% of radioactivity in fractions A₂ and A₃, respectively was recovered as GlcN. The remainder of the radioactivity in fractions A₂ and A₃ appears to be noncarbohydrate (see below). Carbohydrate analysis, before and after treatment with HNO₂, indicated that GlcN accounts for ~95% the radioactivity in fraction A₁, and ~5% appears to be GlcNAc.

Approximately 6% of the radioactivity was released from the [³H]GlcN-labeled, delipidated, whole parasite proteins by digestion with N-glycanase. Upon Bio-Gel P-4 chromatography, the released material eluted as three distinct peaks (Fig. 4B), B₁, B₂, and B₃ representing, respectively, 1.2, 2.3, and 2.5% of the total [³H]GlcN incorporated into the parasite proteins. The eluted radioactivity was pooled as indicated in Fig. 4B and analyzed for carbohydrates. All three fractions (B₁, B₂, and B₃) gave GlcN after acid hydrolysis. However HPAEC analysis before acid hydrolysis, showed the absence of hexosamines and N-acetylhexosamines in fraction B₃. Fraction B₃ does not bind to Ag 1-X 12 (H⁺) resin, suggesting that it is not free GlcN. However, almost all radioactivity in fraction B₃ was bound to Ag 1-X8 (carbonate) and Ag 4-X4 (base) resins. These results suggest that fraction B₃ is a negatively charged (possibly phosphate) derivative of GlcNAc. Fractions B₂ and B₃ gave both GlcN and GlcNAcol on sequential NaBH₄ reduction, acid hydrolysis, and N-acetylation suggesting that they contain reducing end GlcNAc. Fraction B₂ appears to release 1 or 2 residues of Man on treatment with jack bean -mannosidase. On Bio-Gel P-4, the elution position of fraction B₂ corresponds to Man₃GlcNAc₂. Together, these results suggest that parasite proteins contain a low level of low molecular weight N-linked oligosaccharides. Further characterization of these carbohydrates could not be carried out because of their low abundance.

Digestion of the [³H]GlcN-labeled parasite proteins with Pronase and partitioning of the digest between water and 1-butanol gave ~90% of the radioactivity in the organic phase and the remainder in the aqueous layer. After acid hydrolysis, the materials in both the organic and aqueous phases gave only [³H]GlcN as the radioactive sugar. The GlcN residues of the carbohydrates in the 1-butanol phase were almost quantitatively sensitive to HNO₂, suggesting that this fraction represents predominantly GPI anchors. The Bio-Gel P-4 elution profile of the radioactivity in the aqueous phase is shown in Fig. 4C; the eluted radioactivity was pooled as shown. Acid hydrolysis of the fractions C₁ and C₂, before and after treatment with HNO₂, and HPAEC analysis indicated that both fractions contained only [³H]GlcNAc; no other radiolabeled sugars were detected. Only 40-50% of the radioactivity in fraction C₂ is accounted for by [³H]GlcNAc; the remainder appears to be noncarbohydrate, possibly amino acids or peptides formed from nonspecifically radiolabeled proteins due to the entry of GlcN into glycolytic pathway (53). Treatment of Fraction C₁ with jack bean -mannosidase shifted the elution position on Bio-Gel P-4 to a slightly lower molecular weight region, corresponding to the removal of two or three Man residues. These results, taken together with those from alkaline -elimination and N-glycanase treatment, suggest that parasite proteins contain a low level of N-linked oligosaccharides.

To determine whether P. falciparum also contains unsubstituted terminal residues of GlcNAc, parasites at the trophozoite and schizont stages were isolated from the Percoll-enriched infected erythrocytes by saponin treatment as described (18). In a separate experiment, [³H]GlcN-labeled parasites were released by treating the infected erythrocytes with saponin as above. SDS-PAGE and fluorography showed the presence of radiolabeled proteins in the parasites but not in the supernatant (data not shown), suggesting that the parasites were intact. The parasite lysates were galactosylated using UDP-[³H]Gal and bovine milk GalT (50). SDS-PAGE fluorography (Fig. 5) demonstrated that whereas ovalbumin and several erythrocyte proteins ranging from 20 to 300 kDa were galactosylated, none of the proteins with molecular weight >25 kDa were galactosylated in the parasite lysate. However, two distinct proteins (14.5 and 18 kDa) and a smear at 20-25 kDa were labeled in the parasite lysate. Galactosylatable proteins with comparable molecular weights were also present in the lysates of erythrocyte ghosts, albeit in low proportions (Fig. 5). Considering that the parasite feeds on erythrocyte components and that many erythrocyte glycoproteins contain high levels of terminal GlcNAc residues, it is distinctly possible that the observed labeling is due to erythrocyte proteins internalized and degraded by the parasite.

P. falciparum was reported to contain a peptide O-glycosidic GlcNAcT that can transfer GlcNAc onto Thr residues of a synthetic peptide, Pro-Tyr-Thr-Val-Val (23). This was investigated by incubating the above synthetic peptide and UDP-[³H]GlcNAc with the parasite lysate as described (23). Product analysis on Bio-Gel P-2 (23), demonstrated that the peptide was not glycosylated to a detectable extent. Several preparations of the parasite lysates and parallel negative controls (without the peptide, and only UDP-[³H]GlcNAc in buffer) were analyzed. In each case, product analysis (23) showed the presence of a radioactive peak corresponding to GlcNAc (at an elution volume higher than the peptide). Approximately the same amount of radioactivity was also eluted from control incubations (uninfected red cell lysates and buffer only), and it appeared to correspond to ~0.5% of free [³H]GlcNAc contaminant present in UDP-[³H]GlcNAc. These results suggest that P. falciparum either does not express peptide O-glycosidic GlcNAcT activity or contains only a very low level of such enzyme activity.

The [³H]GlcN-labeled carbohydrate moieties, isolated by the alkaline borohydride treatment and Bio-Gel P-4 chromatography, of various parasite protein bands were separately dephosphorylated with aqueous HF and then deaminated and reduced with NaBH₄. In each case, Man₄-AHM was identified as the major product (>95%) by size exclusion chromatography on Bio-Gel P-4. The neutral glycans derived from the 200-215-, 82-, and 75-kDa proteins were each microsequenced by analyzing the products of exoglycosidase digestion and partial acetolysis using three different chromatographic techniques (see below). The glycans of the GPI anchors of 14-, 30-38-, 43-56-kDa proteins were characterized by analyzing on a Bio-Gel P-4 column before and after treatment with A. saitoi -mannosidase and/or jack bean -mannosidase (Table I).

Gel filtration of neutral glycans of individual parasite protein GPI anchors on calibrated Bio-Gel P-4 columns using Glc as an internal standard, gave, in all cases, a major peak corresponding to 5.3 GU with a small amount (<5%) of higher molecular weight components; the elution positions of the major peaks were similar to the standard Man₄-AHM (data not shown), suggesting that the glycan moieties consist of four hexoses and one GlcN residue. Digestion of the neutral glycans with A. saitoi, the 1,2-linkage-specific -mannosidase, shifted the elution position to that of standard Man₂-AHM (3.3 GU), indicating the removal of two Man residues (data not shown). Acetolysis of the neutral glycans under conditions that preferentially hydrolyzes 1,6-glycosidic bonds, gave a major and a minor peak (2.4 and 1.7 GU) corresponding to that of standard Man-AHM and AHM, respectively (not shown). Treatment with jack bean -mannosidase shifted the elution volume to that of standard AHM (not shown). Taken together, these results demonstrate that the glycan moieties of the parasite protein GPI anchors consist of four Man residues and one GlcN residue.

On Dionex-HPAEC, the neutral glycans of the 200-215- and 82-kDa protein GPI anchors eluted as single peaks corresponding to the Man₄-AHM standard (data not shown). Treatment of the neutral glycans with A. saitoi -mannosidase shifted their retention times to a value comparable to that of standard Man₂-AHM (data not shown). The products of jack bean -mannosidase coeluted with AHM (not shown).

HPTLC analysis of the GPI anchor glycan cores, isolated from 200-215- and 82-kDa proteins, showed one major band at an R_f value identical to Man₄-AHM standard (Fig. 6, lane 4; see also Fig. 7, A and B, lanes 2). Several slow migrating minor bands representing <5% of the total radioactivity were also present (see Fig. 7, A and B, lanes 2). These may represent glycans with unidentified substituents, and they were not further characterized because of their low abundance. Treatment of the glycans with A. saitoi -mannosidase increased the R_f values to that of authentic Man₂-AHM (Fig. 6, lanes 1 and 5). Partial acetolysis gave one major (Man-AHM) and two minor products (Man₂-AHM and AHM) (Fig. 6, lanes 2 and 6). Digestion with jack bean -mannosidase gave one major product with the same R_f value as that of authentic AHM (Fig. 6, lanes 3 and 7).

The linkage position between the Man residue and the GlcN was determined by periodate oxidation. Compositional analysis using Dionex HPAEC indicated that the GlcN residues of the GPI anchors are quantitatively oxidized by periodate (data not shown). However, upon N-acetylation, the GlcN residues were completely resistant to periodate. These results indicate that GlcN is substituted either at C-4 or at both C-4 and C-6, but not at C-3. However, the stability of this glycosidic bond to partial acetolysis excludes the possibility of a 1,6-linked glycosidic bond between the first Man residue and the GlcN residue. Thus, this linkage position should be 1,4.

The results of the above analyses establish the sequence Man1-2Man1-2Man1-6Man1-4AHM for the glycan cores of the parasite protein GPI anchors.

The carbohydrate moieties of the parasite proteins (200-215 and 82 kDa) were deaminated, reduced with sodium borohydride, and then subjected to partial hydrolysis using conditions that do not affect the ethanolamine phosphate linkage (46, 49). The partial hydrolysates were divided into two equal parts. In each case, one part was directly dephosphorylated with aqueous HF and the other part was digested with jack bean -mannosidase before dephosphorylation. The products were analyzed by gel filtration on Bio-Gel P-4 (data not shown) and by HPTLC (Fig. 7).

The partial hydrolysates that were directly dephosphorylated gave five peaks that correspond to Man₄-AHM, Man₃-AHM, Man₂-AHM, Man-AHM, and AHM on Bio-Gel P-4 chromatography (data not shown). HPTLC analysis gave a ladder of Man₄-AHM, Man₃-AHM, Man₂-AHM, Man-AHM, and AHM (Fig. 7, A and B, lanes 3). Digestion of the partial hydrolysates with jack bean -mannosidase before treatment with aqueous HF gave a mixture of Man₃-AHM and AHM (Fig. 7, A and B, lanes 4). Man₃-AHM is formed from GPI glycan cores that were unaffected and those in which the fourth Man was cleaved during the partial hydrolysis, whereas AHM is formed from glycans in which the partial hydrolysis cleaved two or more Man residues. These results demonstrate the presence of a phosphate ester substituent on the third Man residue from AHM. This substituent is likely to be the conserved protein-anchoring ethanolamine-phosphate moiety attached to the O-6 position of the third Man residue (for review see Ref. 54).

Jack bean -mannosidase, which removes only the unsubstituted -Man residues from the nonreducing end, was used to identify substituents on the terminal Man residue. The GPI anchors of 200-215- and 82-kDa parasite proteins isolated by mild alkaline saponification of the Pronase-digested product or those obtained by treatment with alkaline borohydride were digested with jack bean -mannosidase before and after treatment with HNO₂ and NaBH₄. The products were dephosphorylated with aqueous HF, and those not already deaminated were then treated with HNO₂ and NaBH₄. The GPI anchors that were treated with -mannosidase after nitrous acid deamination gave Man₃-AHM on Bio-Gel P-4 chromatography (data not shown) and on HPTLC (Fig. 8, lanes 2 and 5). However, the GPI anchors that were digested with -mannosidase prior to nitrous acid deamination and sodium borohydride reduction gave a mixture of Man₄-AHM (50-60%) and Man₃-AHM (40-50%) (data not shown, and Fig. 8, lanes 3 and 6). These results indicate that at least 50-60% of the nonreducing end Man residues are substituted and that these substituents were almost quantitatively removed under the conditions of nitrous acid treatment. The results were reproducible for several purified preparations of GPI anchors from both 200-215- and 82-kDa proteins. The GPI anchor samples, those treated and not treated with HNO₂/NaBH₄ prior to the -mannosidase digestion, were similarly purified on Bio-Gel P-4. Therefore, the observed results were not due to incomplete removal of the terminal Man residues by jack bean -mannosidase, caused by contaminants in samples that were not treated with HNO₂/NaBH₄. Thus, the glycan cores of the parasite protein GPI anchors have the following structure. where X represents an unidentified substituent.

DISCUSSION

In this study, metabolic labeling with [³H]GlcN established that about 15 proteins of the intraerythrocytic stage P. falciparum are dominantly modified with carbohydrate moieties. Carbohydrate compositional analysis, partitioning of the products of Pronase digestion between water and water-saturated 1-butanol, and structural studies demonstrate that GPI anchors represent the major carbohydrate modification in these parasite proteins.

P. falciparum proteins contain, besides GPI anchor moieties, low levels of N-glycanase-releasable high Man type and/or incompletely processed N-linked oligosaccharides and novel negatively charged GlcNAc residues; the latter was not released by the reductive alkaline -elimination. These structures together account for about 6% of the total [³H]GlcN incorporated into the parasite proteins. Previously, Dieckmann-Shuppert et al. (23) reported the presence of a small proportion (7-10%) of N-glycanase-releasable carbohydrates in radiolabeled parasite proteins. Although the structures were not studied (23), these were likely to be N-linked carbohydrates. Recently, Kimura et al. (28) reported the existence of N-linked carbohydrates in the parasites. Considering that GPI anchors represent the major carbohydrate modification in P. falciparum and that proteins are modified with GPI anchors on a mol/mol basis, the level of N-glycosylation in the parasite proteins is very low. This explains why the presence of N-glycosylation was evident only on examination of whole cell lysates and was not readily detectable in individual parasite proteins. The low content of N-linked carbohydrates in P. falciparum is also in agreement with the previously reported undetectable levels of dolichol pyrophosphate-oligosaccharide intermediates, and peptide N-glycosidic oligosaccharyltransferase activity in the parasite (27).

The observed low content of N-linked carbohydrates in P. falciparum proteins is not due to the incomplete release by N-glycanase. The amount of radiolabeled carbohydrates released by N-glycanase is comparable with the amounts of glycopeptides recovered in the water phase (<10% of the total), after partitioning of the Pronase digests of the parasite proteins between water and water-saturated 1-butanol. The carbohydrates remaining in the water phase should correspond to N-linked structures, because the GPI anchor moieties are the only other carbohydrates detectable in the parasite proteins (~90%).

In contrast to the results presented here, Kimura et al. (28) reported that the N-linked carbohydrates represent as much as 70% of the radiolabeled carbohydrates of ring and early trophozoite stage P. falciparum, and about 30% in old trophozoites. These investigators have not considered the presence of GPI anchor moieties in the parasite proteins. Therefore, N-linked carbohydrates and other trace amounts of glycans may have been viewed as the major constituents of the parasite proteins (28). Despite this difference, our results are in partial agreement with those of Kimura et al. regarding the structural features of N-linked carbohydrates (28). In this study, compositional analysis of the N-glycanase-released carbohydrates, after treatment with NaBH₄, gave both GlcN and glucosaminitol, suggesting that these glycans contain at least two GlcN residues, one of which is at the reducing end. Based on their size and susceptibility to jack bean -mannosidase, these oligosaccharides appear to be Man₃-GlcNAc₂ and larger high Man type and/or hybrid type structures. Kimura et al. (28) have also found similar structures in addition to N-glycanase-released chitobiose and single residues of GlcNAc in the parasite proteins. The latter were not detected in this study. However, novel, negatively charged residues of GlcNAc apparently linked to Asn were found in this study.

The observed low content of N-glycosylation in intraerythrocytic P. falciparum is not due to the low abundance of potential N-glycosylation sites in the parasite proteins. The deduced amino acid sequences of erythrocytic stage P. falciparum proteins contain several potential N-glycosylation sites (7, 24, 26). For example, MSP-1 contains as many as 15 potential N-glycosylation sites (7), and the heat shock protein HSP-72 contains four such sites (26). Moreover, a 72-kDa C-terminal peptide of P. falciparum MSP-1, expressed in mammalian cells, contains one to four N-linked oligosaccharide chains per molecule.² Therefore, it is likely that the low content of N-linked carbohydrates is due to very low N-glycosylation capacity of the parasite. This conclusion is consistent with the previously reported undetectable levels of GDP-Man, dolichol pyrophosphate-oligosaccharide intermediates, and peptide N-glycosidic oligosaccharyltransferase activity (27).

In contrast to the presence of GPI anchors and N-linked carbohydrates, O-linked carbohydrates were not detectable either in the individual protein gel bands or whole P. falciparum lysates. In support of this conclusion, neither free N-acetylhexosaminitols nor oligosaccharides with N-acetylhexosaminitols at the reducing ends were detected after NaOH/NaBH₄ treatment of the parasite proteins radiolabeled at different stages of intraerythrocytic development. Galactosylation of lysates of P. falciparum free of red cells using GalT and UDP-[³H]Gal, and SDS-PAGE analysis also did not provide evidence for the abundant presence of O-linked GlcNAc residues in the parasite proteins. In contrast to these results, it was previously reported that O-glycosylation is the major carbohydrate modification in parasite proteins (20-23). The reported carbohydrate structures included single residues of O-GlcNAc (20-23) and oligosaccharides with terminal GlcNAc residues (23). While some of the discrepancies may be due to the lack of consideration given to GPI moieties (20-23, 28), others (28) appear to be due to problems with the technical approaches used (see below).

Previously, Dieckmann-Schuppert et al. (23) used total proteins from P. falciparum cultures containing 90% uninfected and 10% parasitized red cells for the release of O-linked carbohydrates with NaOH/NaB³H₄ and for the galactosylation of terminal GlcNAc with GalT and UDP-[³H]Gal. These procedures may have radiolabeled predominantly the carbohydrates of the native erythrocyte glycoproteins (23) because of the abundance of O-linked GlcNAc and oligosaccharides bearing terminal GlcNAc in human erythrocyte glycoproteins (55). This would explain several of the previously reported contradictory results (23). First, the size and structural features of the presumed O-linked carbohydrates of the parasite proteins were strikingly similar to those obtained from control erythrocyte proteins (23). Second, the 2.5-3 times higher level of radiolabeling of carbohydrates might have been due to batch-to-batch variations in radioactivity incorporation into the carbohydrates of the erythrocytic glycoproteins rather than radiolabeling of the parasite carbohydrates. Third, the amount of presumed Gal1-4GlcNAcol, obtained after galactosylation, appeared to be higher for proteins from control red cells compared with those from red cells with 10% parasitemia. Fourth, the reported effective size of Gal1-4GlcNAcol on Bio-Gel P-4 was 2 GU or less (23), a figure not in agreement with the reported effective sizes of galactitol, N-acetylgalactosaminitol, and Hex1-4GlcNAcol of 1.5, 2.6, and 3.2, respectively (52), raising doubts about the authenticity of the characterized structure. Finally, whereas we and others (18, 28) found that uninfected erythrocytes do not metabolically incorporate sugars into their proteins to significant levels, Dieckmann-Shuppert et al. (23) reported that the level of ¹⁴GlcN incorporation into proteins of control red cells was about half the level of incorporation into proteins of red cells with 10% parasitemia (27). The latter is not plausible, considering that UDP-GlcNAc is not synthesized by uninfected red cells (27).

To avoid high background interference due to the abundance of O-linked carbohydrates in erythrocyte glycoproteins, we performed biochemical analysis with the metabolically radiolabeled P. falciparum proteins separated on SDS-polyacrylamide gels and with lysates of Percoll-enriched parasitized erythrocytes. Galactosylation was carried out with lysates of the isolated parasites. Despite these precautions, interference from red cell glycoproteins was apparent in galactosylation studies (see Fig. 5, lane 2); this is likely due to glycoproteins internalized and degraded by the parasite. The results of metabolic labeling argue against these being parasite proteins, because similar size glycoproteins bearing N- or O-linked carbohydrates were not apparent in the parasite.

It was reported that P. falciparum contains peptide O-glycosidic GlcNAcT activity, based on the use of a synthetic peptide, Pro-Tyr-Thr-Val-Val, as an O-GlcNAc acceptor (23). However, the results of this study demonstrate that neither the lysates of Percoll-enriched parasitized erythrocytes nor the lysates of the parasite isolated by mild saponin lysis of the infected erythrocytes carry out detectable glycosylation of Pro-Tyr-Thr-Val-Val. This discrepancy may be due to very low levels of O-glycosidic GlcNAcT activity in the parasite, because the results of biochemical analysis demonstrate that O-GlcNAc residues are undetectable in the P. falciparum proteins.

Kimura et al. (28) have reported that P. falciparum proteins contain O-glycanase-releasable carbohydrate(s). Although the structure(s) was not determined, the released carbohydrate(s) was suggested to be Gal1-3GalNAc (28) in view of the strict substrate specificity of the enzyme (56, 57). In contrast to this, the present study demonstrates that GalNAc is not detectable in the parasite proteins. This result is in agreement with the previous reports that P. falciparum neither synthesizes GalNAc nor metabolically utilizes exogenous GalNAc for glycosylation (23) and that the Golgi-resident GalT is undetectable in the parasite (58). Thus, the occurrence of O-glycanase-releasable carbohydrates in P. falciparum remains to be established.

Although the results of this study indicate that O-linked carbohydrates are undetectable in P. falciparum proteins, O-GlcNAc may be present in the parasite proteins at very low levels, i.e. below the detection limits of the methods used. O-Glycosylation of cytoplasmic and nucleoplasmic proteins with single residues of GlcNAc has been shown to be generally present in almost all eukaryotic cells, and this kind of glycosylation is reported to be a dynamic process, responsive to cellular metabolism (50, 59-61). Therefore, single residues of O-GlcNAc may be present in P. falciparum proteins at very low levels, i.e. <1-2% compared with the total GPI anchors of the parasite proteins, since N-linked carbohydrates representing ~1-2% (<1 chain/~50 molecules of proteins) were distinctly detected. However, single residues of O-GlcNAc are not likely to be present in GPI-anchored cell surface parasite proteins, because plasma membrane proteins, in general, do not undergo cytoplasmic O-glycosylation (61).

In this study, we establish that the GPI glycan cores of multiple proteins from four different strains of P. falciparum consist of Man1-2Man1-2Man1-6Man1-4AHM. We also show that the glycan cores of the GPI anchors contain substituents attached to the first Man residue from the nonreducing end, and they are susceptible to the conditions of nitrous acid deamination. The nature and location of the substituents appear to be unusual and have not been previously reported in GPI moieties of proteins from other sources.

The GPI anchors of P. falciparum proteins have been reported to be involved in several critical functional roles (28-30, 62). In agreement with these reports, our recent studies suggest that mannosamine, a known inhibitor of GPI biosynthesis (63-65), is lethal to the erythrocytic stage parasite.³

The P. falciparum GPI anchors can induce cytokine release in host macrophages and cause pathological conditions in mice that include transient pyrexia, hypoglycemia, and lethal cachexia (32). Recently, it has been reported that both the GPI anchors MSP-1 and MSP-2 induce nitric-oxide synthase expression in macrophages and vascular endothelial cells by a protein-tyrosine kinase-dependent and protein kinase C-dependent signaling pathway (36). The GPI anchors of these proteins are also reported to up-regulate the levels of intercellular adhesion molecule 1, vascular cell adhesion molecule 1, and E-selectin expression in vascular endothelial cells as well as to cause increased leukocyte and parasite cytoadherence to vascular endothelial cells via tyrosine kinase-dependent signal transduction (35). The structural basis for this broad bioactivity of the GPI anchors of P. falciparum proteins is not known. However, it has been suggested that the phosphatidylinositol moiety of the parasite GPI alone is not sufficient for protein-tyrosine kinase-induced cell signaling and that the glycan moiety is also involved in this activity (36).