INTRODUCTION

Parasitic protozoa of the genus Leishmania are the causative agent of a wide range of human and animal diseases, which are transmitted by an insect vector, the sandfly. The digenetic life cycle of the parasite encompasses the infective metacyclic and several forms of procyclic, extracellular promastigotes in the digestive tract of the insect and the intracellular amastigotes residing in parasitophorous vacuoles of the mammalian host cells, the macrophages. Expression of complex and unique glycoconjugates appears to be crucial for the survival and development of the parasites in the sandfly vector and the mammalian host (reviewed in Refs. 1, 2, 3). The dominant cell surface glycolipid of Leishmania promastigotes is lipophosphoglycan (LPG).¹ In the sandfly, LPG serves as a ligand for the attachment of procyclic promastigotes to the midgut wall lining and may protect the parasites against the hydrolytic environment of the insect's digestive tract. After transmission of metacyclic promastigotes to the mammalian host, LPG confers complement resistance; it has been shown to act as a receptor for invasion of macrophages and may protect the parasites against the microbicidal response of the host cell (reviewed in Refs. 1, 2).

The structure of LPG from five different Leishmania species has been described (4, 5, 6, 7, 8, 9, 10), including life stage- and strain-specific features (8, 11, 12). Conserved structural elements are a lyso-alkylphosphatidylinositol membrane anchor, a phosphohexasaccharide core structure, and a backbone of up to 40 phosphodiester-linked disaccharides of the structure PO₄-6-Gal1-4Man. Species-, strain-, and stage-specific modifications include glucose-1-PO₄ linked to the core structures of some LPGs, carbohydrate side chains on the phosphodisaccharide repeats, and the structures of the terminating neutral (cap) glycans.

More recently it has been reported that acid phosphatases secreted by promastigotes of various Leishmania species are modified by LPG-related glycans (13, 14, 15, 16). Secreted acid phosphatase (sAP) from Leishmania mexicana is a highly glycosylated and phosphorylated enzyme, organized as an unusual filamentous polymer (17). Structural analysis of the L. mexicana sAP glycans revealed the modification of Ser(P) by unsubstituted and glucosylated phosphodisaccharides and/or mannooligosaccharides, representing a novel type of protein glycosylation (18). Specific serine/threonine-rich sequence motifs clustered near the C terminus of the sAP polypeptides appear to be the sites for this phosphoglycan modification (19).

Ultrastructural studies on the release of polymeric sAP and other macromolecular compounds from the flagellar pocket of L. mexicana led to the identification of a second, distinct, type of phosphoglycan-modified filamentous secretory product. This material forms fibrous networks in the center of promastigote aggregates and in the culture medium (20). Based on metabolic labeling and immunoprecipitation studies, a novel phosphate-containing secreted Leishmania antigen with very low electrophoretic mobility in SDS-polyacrylamide gels has been proposed as the candidate molecule. The secretion of network-forming filamentous phosphoglycan antigens by promastigotes appears to be ubiquitous in the genus Leishmania (3, 20), implying an important function for these macromolecular structures.

In the present study, we describe the purification of this novel filamentous phosphoglycan-modified secretory compound from Leishmania major promastigotes. Structural and ultrastructural studies characterize this antigen as a proteophosphoglycan (PPG) with organization and properties similar to the mammalian mucins.

EXPERIMENTAL PROCEDURES

Promastigotes of L. major LRC-L137, clone V121, were grown in vitro in semi-defined medium 79 (21) supplemented with 5% heat-inactivated fetal calf serum (Flow).

Two liters of spent culture medium of late log/early stationary phase promastigotes (1-1.5 × 10⁸ cells/ml) were passed at a flow rate of 120 ml/h over a DE52-cellulose column (12 × 3 cm; Whatman) equilibrated with 100 m NaCl, 20 m Tris-HCl, pH 7.5. After washing with equilibration buffer (200 ml), bound PPG was eluted with a linear gradient of 100-500 m NaCl in 20 m Tris-HCl, pH 7.5 (150 ml), at a flow rate of 60 ml/h. PPG-containing fractions were collected, adjusted to 1  (NH₄)₂SO₄, and applied onto an octyl-Sepharose column (15 × 1.5 cm; Pharmacia Biotech Inc.) equilibrated with 1  (NH₄)₂SO₄, 5 m Na₂EDTA, 20 m Tris-HCl, pH 7.5. The column was consecutively washed with equilibration buffer, 20 m Tris-HCl, pH 7.5, and 50% 1-propanol (40 ml, respectively). PPG-containing flow-through fractions were subjected to ultracentrifugation (3 h, 144,000 × g, Ti 60 rotor, Beckman). The pellet was resuspended in 3.56  CsCl, 5 m Na₂EDTA, pH 7.5 ( = 1.45 g/ml) and ultracentrifuged for 3 h at 215,000 × g. The PPG pellet was resuspended in H₂O, again subjected to ultracentrifugation (3 h, 144,000 × g) to remove residual salt, and then dissolved in H₂O.

Alternatively, the concentrated octyl-Sepharose eluate (Centricon 30, Amicon) was applied onto a Superose 6 gel filtration column (30 × 1 cm, Pharmacia) equilibrated with 250 m ammonium acetate, pH 7, at a flow rate of 0.5 ml/min. Both PPG- and PG-containing fractions were collected and concentrated by Centricon 30 ultrafiltration. PPG was identified throughout the purification procedure by phosphate determination, two-site ELISA, and SDS-PAGE (see below). Culture supernatant and DE52-cellulose flow-through were treated with proteinase K (100 µg/ml, 3 h, 55 °C, Boehringer Mannheim) to degrade the serum proteins present in these samples before loading onto an SDS-polyacrylamide gel. The electrophoretic mobility of PPG was not affected by this treatment (see below).

Protein and phosphate were determined as described previously (22, 23). In the case of L. major culture supernatant, free phosphate and phosphate-containing low molecular weight compounds were removed by extensive ultrafiltration (Centricon 3, Amicon) prior to analysis. The values obtained were corrected for the phosphate content of identically treated fresh culture medium. Quantification of purified PPG was based on a PO₄ content of 20% (w/w). Two-site ELISA of L. major PPG was performed as described previously for L. mexicana sAP (18) using mAb AP3 (IgM) as capture antibody and biotinylated mAbs AP3, WIC79.3, or WIC108.3 or LT6 hybridoma supernatant as detecting antibodies (see Refs. 16, 24 for antigen specificities). Discontinuous SDS-PAGE (25) was performed on 4% stacking gels of either 0.2- or 1.5-2-cm length over 7.5-20% separating gels. For Western blots (26) and dot blots (1-µl aliquots of column fractions) either nitrocellulose membranes (0.2 µm, Schleicher & Schuell) or cationized nylon membranes (Zeta probe, Bio-Rad) were used. Stains-all (Bio-Rad) staining of phosphoglycan antigens in SDS-polyacrylamide gels and their immunodetection on membrane supports has been previously described (16). Silver staining of Stains-all-stained polyacrylamide gels was performed as described in Ref. 27. PPG was also analyzed on agarose gels (0.6% agarose in 40 m Tris acetate, 4 m Na₂EDTA, pH 7.8; 5 V/cm, 2-3 h) after the addition of 0.2 volume of sample buffer (0.25% bromphenol blue, 0.25% xylencyanol, 15% Ficoll). HindIII-cut -phage DNA (Promega) was used as molecular weight marker and the gels were stained with Stains-all.

Extensive biotinylation of PPG (1 mg in 750 µl of 50 m NaHCO₃, pH 9.6, 37 °C) was performed by adding 3 × 25 µl of 75 m NHS-LC-Biotin (Bio-Rad) dissolved in dimethyl sulfoxide at 2-h intervals. Biotinylated PPG was recovered by ultracentrifugation as described above.

PPG was reduced by incubation in 6  guanidinium chloride (GdmCl), 250 m Tris-HCl, pH 8.5, 5 m Na₂EDTA containing 20 m dithiothreitol for 2 h at 50 °C. Alkylation was either performed by the addition of a 5-fold molar excess of iodoacetamide over dithiothreitol and incubation in the dark for 1 h at room temperature or by adding 0.2 volume of vinylpyridine and incubation for 45 min at room temperature. The reaction was quenched by 0.2 volume of 2-mercaptoethanol. Reduced and alkylated PPG was concentrated by ultrafiltration (Centricon 30, Amicon) and applied onto a Superose 6 column equilibrated with 6  GdmCl, 50 m Tris-HCl, pH 8.0, 5 m Na₂EDTA.

PPG (~50 µg) in 20 µl of 100 m Tris-HCl, pH 8.0, 2 m CaCl₂ was incubated with either trypsin (10 µg, Sigma), proteinase K (10 µg) at 37 °C or with thermolysin (10 µg, Boehringer Mannheim) at 55 °C for 1 h. The proteinases were inactivated by heating to 100 °C for 10 min, and the samples were analyzed by SDS-PAGE and agarose gel electrophoresis. To obtain larger amounts of thermolysin-cleaved antigen, PPG (500 µg) was dissolved in 100 m Tris-HCl, pH 8.0, 2 m CaCl₂ containing proteinase (20 µg) and incubated for 5 h at 55 °C. The samples were adjusted to 6  GdmCl, 5 m Na₂EDTA and subjected to Superose 6 chromatography in the same buffer.

PPG was subjected to mild acid hydrolysis (20 m HCl, 15 min, 100 °C (5, 6, 28)), neutralized with 250 m ammonium acetate, lyophilized, and dissolved in H₂O. The acid-released neutral and phosphorylated oligosaccharides were separated by high pH anion exchange HPLC on a Dionex BioLC Carbohydrate analyzer (Dionex Corp., Sunnyvale, Ca) using a Carbo-Pac PA1 column and pulsed amperometric detection using three sequential linear gradients of sodium acetate (0 m for 6 min, raised to 50 m over 18 min, to 175 m over 1 min, and to 250 m over 30 min). Peak fractions were neutralized immediately with 2  acetic acid, desalted by passage over AG50X12 (H⁺) (Bio-Rad), and lyophilized. In some experiments the protein backbone of PPG was separated from the released glycans by repeated ultrafiltration (Centricon 3, Amicon) of the hydrolysate. The filtrate and the retentate were lyophilized and subjected to monosaccharide analysis.

Phosphoglycans were dephosphorylated by incubation with calf intestine alkaline phosphatase (400 units/ml, Sigma) in 50 m NH₄HCO₃, pH 8.5, for 16 h at 37 °C. Neutral glycans were treated with jack bean -mannosidase (10 units/ml, Sigma) or bovine testes -galactosidase (0.2 units/ml, Boehringer Mannheim) in 50 m sodium acetate pH 5.0, for 16 h at 37 °C. Subsequently, the samples were desalted by passage over a tandem column of AG3X4 (OH) (Bio-Rad) over AG50X12 (H⁺), lyophilized, and rechromatographed by HPLC on a Carbo Pac PA1 column.

Combined GC-MS was performed using a Finnigan MAT1020B GC-MS (Finnigan, Sunnyvale, Ca), fitted with either a 25-m × 0.3-mm CPSil5 low polarity column (Chrompack, Middleburg, The Netherlands) for trimethylsilyl derivatives and permethylated alditol acetates (PMAAs) or a 25-m × 0.22-mm BPX-70 high polarity column (SGE, Australia) for alditol acetates and PMAAs as described previously (12, 29).

Native PPG (20 µg) containing scyllo-inositol as an internal standard and monosaccharide standards were subjected to methanolysis, re-N-acetylated, and following trimethylsilylation analyzed by GC-MS either directly or after methylation with diazomethane (30) as described previously (9, 12). Alternatively, neutral monosaccharides of native, mild acid-treated, and HF-dephosphorylated PPG (20 µg, respectively) were also analyzed as alditol acetates, prepared and analyzed by standard methods (31). Methylation linkage analysis of dephosphorylated glycans (0.5-20 µg) was performed as described previously (7, 12, 18).

All NMR spectra were recorded on a Bruker ARX 500 spectrometer at a temperature of 300 K. One-dimensional ¹H NMR spectra were recorded at 500.13 MHz with a spectral width of 5050 Hz, a pulse length of 7.5 µs (60^o), accumulation of typically 16 scans (up to 512 scans), and with a relaxation delay between scans of 3 s. Spectra were processed using an exponential line broadening function of 0.3 Hz. Some spectra were processed using a Gaussian window function for resolution enhancement to assist in the measuring of coupling constants. Chemical shifts were referenced to 2,2-dimethyl-2-silapentane-5-sulfonate at 0.0 ppm. One-dimensional ³¹P NMR spectra were recorded at 202 MHz with a spectral width of 5000 Hz, a pulse width of 5 µs (45^o), and a relaxation delay between scans of 2 s. Typically 500 scans were acquired prior to Fourier transformation. Spectra were processed using an exponential line broadening function of 5-10 Hz. Chemical shifts were referenced to an external capillary of 85% H₃PO₄ at 0.0 ppm. A pH titration was carried out by recording ³¹P spectra over the pH range 3.4-9.2. Two-dimensional-DQFCOSY and TOCSY (80-ms mixing time) spectra were recorded over a spectral width of 3030 Hz in both dimensions, with 4096 data points in F2 and 256 experiments in F1, each of 64 scans. HMBC spectra were recorded using 2048 complex data points in F2 over a spectral width of 2000 Hz centered at the frequency of the residual HOD signal. This signal was suppressed by mild presaturation (75 db) during the relaxation delay. A total of 32 scans was recorded for each 128 slices over a spectral width of 800 Hz in F1. A relaxation delay of 1 s was used between scans, with an evolution delay of 50 ms in the HMBC sequence. While this is not tuned precisely to the expected size of the CH₂-O-P coupling, a series of trial experiments showed that it resulted in optimum signal to noise, presumably due to a compromise between the optimum evolution time for coupling (1/2J) and the loss of signal due to relaxation for longer delays. The two-dimensional data were processed as a 2048 × 1024 matrix, with a squared sine bell window function applied in both dimensions.

PPG (50-1000 µg) was lyophilized, cooled, and mixed with pre-cooled (20 °C) 40% aqueous HF (50-200 µl, BDH Chemicals, Australia). After 48-60 h at 0 °C, the samples were frozen in liquid N₂ and lyophilized over NaOH, redissolved in 50 m NH₄HCO₃, and lyophilized again. HF-treated PPG was dissolved in 6  GdmCl, 50 m Tris-HCl, pH 8.5, 5 m Na₂EDTA and applied at 0.5 ml/min to a Superose 6 column equilibrated in the same buffer. Fractions were analyzed for optical density at 280 nm, phosphate and by immunodot blot. Bovine serum albumin and phosvitin were used as controls for detection of peptide bond cleavage and complete protein dephosphorylation, respectively. In some experiments the protein backbone of PPG was separated from the released glycans by repeated ultrafiltration (Centricon 3, Amicon) of the hydrolysate as described above.

Amino acid determinations were performed after hydrolysis of PPG (20-100 µg) with 6  HCl, 0.1% phenol (500 µl) at 110 °C for 16-20 h in vacuo. After evaporation of the HCl, free amino acids were analyzed on a Beckman amino acid analyzer (model 6300).

To detect phosphorylated amino acids, PPG (100 µg) was subjected to partial acid hydrolysis (6  HCl, 0.1% phenol, 100 µl, 90 min at 110 °C). Dried hydrolysates were converted to phenylthiocarbamyl derivatives and analyzed by reversed-phase HPLC (Picotag, Waters) as described previously (32) except for using 300 m sodium acetate at pH 5.5 instead of pH 6.5. Phosvitin, phosphoserine (Ser(P)), phosphothreonine, and phosphotyrosine and an amino acid standard mixture (Sigma) were used as controls. Norleucine (Sigma) was added as an internal standard to all samples. In some analyses, PPG, phosvitin, and Ser(P) were subjected to various treatments prior to hydrolysis. The samples were incubated at 22 and 100 °C in 20 m ammonium acetate or at 100 °C in 20 m HCl for 15 min. Subsequently, the samples were adjusted to 100 m NH₄HCO₃, pH 8.5, and incubated at 37 °C for 16 h with and without the addition of calf intestine alkaline phosphatase (100 units/ml). After lyophilization, the samples were subjected to phosphoamino acid analysis as described above.

Rabbits were immunized subcutaneously with antigen resuspended in phosphate-buffered saline, emulsified in 50% complete Freund's adjuvant for the first immunization, and in 50% incomplete Freund's adjuvant in all subsequent booster immunizations. Synthetic polyserine (Sigma), or the insoluble peptide pellet of 0.5 mg of mild acid-treated L. major PPG, or 0.33 mg of HF-dephosphorylated PPG repurified by Superose 6 chromatography in the presence of 6  GdmCl were used as antigens for immunization. Serum was obtained 10-14 days after each booster immunization yielding anti-polyserine (anti-PS), anti-mild acid-insoluble peptide (anti-MIP), and anti-HF-peptide (anti-HFP) antibodies. The presence of specific antibodies was assessed by dot blot and Western blot using the respective antigens and native PPG.

Purified PPG was dissolved in 40% glycerol (50 µg/ml) and sprayed onto freshly cleaved mica. Mica pieces were transferred to a Balzers BAF 300 evaporation unit, rotary-shadowed with platinum/carbon at an angle of 8^o, and backed with carbon. Replicas were floated onto H₂O and mounted on 400-mesh grids (33). PPG was also absorbed to pioloform and carbon-coated grids and negatively stained with Nano-W (Nanoprobes, Brookhaven, NY). For negative staining immunoelectron microscopy, adsorbed PPG was incubated with the mAbs LT6, WIC79.3, AP3, and LT8.2 (16) and stained with 1% aqueous uranyl acetate as described previously (20). In the case of LT6 labeling, the signal was enhanced by a subsequent incubation with ProteinA-6 nm gold.

RESULTS

It has been previously shown that the candidate antigen for the fibrous phosphoglycan-modified networks formed among L. major promastigotes, identified in the current study as a proteophosphoglycan (PPG), remains at the top of the separating gel in discontinuous SDS-PAGE (3, 20). In this study, the 4% acrylamide stacking gel was enlarged (from 0.2 to ~1.5-2 cm). This modification resulted in almost complete retention of PPG in this part of the SDS-polyacrylamide gel and allowed assessment of impurities migrating in the separating gel (Fig. 1).

PPG present in L. major promastigote culture supernatant could be detected by SDS-PAGE followed by staining with Stains-all. When the spent culture medium was subjected to anion exchange chromatography on DE52-cellulose, PPG bound quantitatively to the column (Fig. 1A, lanes 1-3). A large proportion of the LPG and phosphoglycan (PG) also present in promastigote culture supernatant (3, 34, 35) did not bind. Elution of the bound material by a NaCl gradient and analysis of the eluate by phosphate determination and by two-site ELISA yielded broad, overlapping peaks (not shown) containing PPG, LPG/PG, and various proteins, as judged by SDS-PAGE/Stains-all staining (Fig. 1A, lane 3) and silver staining (not shown). The pooled DE52-cellulose fractions were then passed over octyl-Sepharose under high salt conditions and eluted stepwise with a high salt buffer, a low salt buffer, and 50% 1-propanol. The majority of PPG (>90%) and PG were recovered in the high salt wash (Fig. 1A, lane 4), whereas most proteins and LPG remained bound. Ultracentrifugation of the PPG-containing octyl-Sepharose flow-through led to the quantitative recovery of the antigen in a glassy, voluminous pellet, as judged by SDS-PAGE (not shown). PPG was purified further by resuspending the pellet in a CsCl solution. The high density of the solvent ( = 1.45 g/ml) ensured flotation of most proteins and glycoproteins (36) in a second ultracentrifugation, whereas PPG was quantitatively recovered in the pellet. PPG was desalted by resuspension in H₂O followed by ultracentrifugation and analyzed by SDS-PAGE (Fig. 1A, lane 6).

In an alternative purification procedure, octyl-Sepharose flow-through containing small amounts of PPG (<200 µg) were concentrated and applied onto a Superose 6 gel filtration column. Two phosphate-containing peaks were observed (Fig. 2, peaks A and B). Peak A eluted in the void volume, earlier than the 2000-kDa marker, and overlapped with the two-site ELISA signal obtained with three anti-LPG mAbs of different specificity, AP3, LT6, and WIC79.3. The binding of these mAbs indicates the presence of terminal mannooligosaccharides as well as unsubstituted and galactosylated phosphodisaccharides in PPG. Fractions of peak A were pooled, and PPG was recovered by ultrafiltration (Fig. 1A, lane 5). A second phosphate peak (Fig. 2, peak B) not reactive in two-site ELISA was found near the elution position of the 80-kDa marker protein. On SDS-PAGE, pooled fractions of peak B showed a broad Stains-all-stained area between 50 and 10 kDa apparent molecular mass (Fig. 1A, lane 8) which is typical for LPG and PG. Monosaccharide analysis of these fractions yielded Man, Gal, Gal-6-PO₄, and Ara (not shown). It is therefore likely that peak B corresponds to PG (34) recently characterized in Leishmania donovani (35), and it was not further analyzed. After ultracentrifugation of PPG-containing octyl-Sepharose eluates in the presence of CsCl (see above), PG remained in the supernatant and peak B was not observed in subsequent Superose 6 chromatography (not shown, compare Fig. 9A). Quantities of PPG larger than 200 µg could not be processed by Superose 6 chromatography, because the high viscosity of PPG solutions exceeding about 1 mg/ml led to loss of resolution and extensive backpressure. PPG at a concentration above 10 mg/ml formed a gel, a property not observed for either PG or LPG at even higher concentrations.² Therefore, ultracentrifugation in the presence of CsCl was the method of choice for the purification of larger amounts of PPG. One liter of culture supernatant from densely grown L. major promastigotes (>10⁸/ml) yielded 5-10 mg of PPG (compare Table I).

Table I. Purification of L. major PPG Total protein Total PO4a PO4/protein Purification factorb mg mg Culture supernatant 3366.0 17.6 0.005 1 DE52-cellulose eluate 23.25 7.20 0.31 62 Octyl-Sepharose eluate 0.412 6.27 15.2 3004 Ultracentrifugation in CsCl 0.075 3.13 41.7 8347 a  PO4-associated with Leishmania-derived macromolecules >3 kDa. b  Based on the PO4/protein ratio.

The different stages of PPG purification were also monitored on immunoblots with the anti-LPG-mAb WIC108.3, which binds to the network-forming antigens secreted by L. major promastigotes (37). The Stains-all-stained material in the SDS-PAGE stacking gel, which was enriched during the purification procedure (Fig. 1A, lanes 3-6), strongly reacted with this mAb (Fig. 1B, lanes 1-4).

Both purification procedures resulted in a final product free from protein and LPG/PG contamination, as judged by several criteria. 1) Silver staining of overloaded SDS-polyacrylamide gels did not reveal any bands other than the broad area occupied by PPG in the stacking gel (Fig. 1A, lane 7). 2) Extensive biotinylation under alkaline conditions favoring lysine modification, pH 9.5, followed by SDS-PAGE, blotting, and detection of biotinylated products by enzyme-labeled avidin led only to the detection of PPG (Fig. 1B, lane 6). Since large amounts of biotinylated PPG with a very low lysine content (less than 0.8 mol %, see Table II) in comparison with potential contaminating proteins (average lysine content of 6-7 mol % (38)) were loaded, a high level of protein contamination is unlikely. 3) Both Stains-all staining after SDS-PAGE and immunodetection on blots by WIC108.3 failed to detect LPG- or PG-like compounds in the final product (Fig. 1A, compare lanes 5-8; Fig. 1B, compare lanes 3-5).

Table II. Monosaccharide analysis of L. major PPG Monosaccharide HF/ trifluoroacetic acid/alditol acetates Methanolysis/ trimethylsilylation diazomethane treatment High pH anion exchange HPLC mol/mol mol/mol mol/mola Mannose 62b 62b 62b Galactose-6-PO4 +c 52 Galactose 68 27 19 Glucose 1 2 Arabinose 1 1.5 1 a  Calculated from the relative abundance of PPG glycans (Table V). b  Normalized to 62 mannose. c  Present but not quantitated.

Purified PPG contains large amounts of phosphate, whereas its protein content determined by the Coomassie Blue dye binding assay is very low (Table I). Based on the PO₄/protein ratio, an overall purification factor for PPG of more than 8,000 was calculated (Table I).

L. major PPG contains 4.4% amino acids, 20% phosphate (PO₄), and 75.6% monosaccharides (w/w). Monosaccharide analysis of native PPG yielded Man, Gal, and Gal-6-phosphate as well as small amounts of Ara and Glc (Table II). To quantitate the ratio of hexoses in PPG, it was dephosphorylated by HF prior to sugar analysis. The ratio of Man to Gal was 1:1.1. PPG did not contain detectable amounts of GlcNAc, GalNAc, or myo-inositol.

To confirm the presence of a protein backbone, PPG was incubated with various proteinases, and the products were subjected to SDS-PAGE and agarose gel electrophoresis. Stains-all staining patterns of the PPG proteinase digestion products on SDS-PAGE were indistinguishable from native PPG even on 4% acrylamide gels (not shown). However, on agarose gels, proteinase K and thermolysin digestion led to an increased mobility of PPG (Fig. 3, compare lanes 1-3). To a lesser degree, a mobility shift could also be detected after trypsin (Fig. 3, lane 4), papain, and chymotrypsin digestion (not shown).

Amino acid analysis of PPG revealed a highly unusual composition dominated by three amino acids, Ser (>50 mol %), Ala, and Pro, which together represent more than 87 mol % of the protein backbone(s) residues of L. major PPG (Table III).

Table III. Amino acid analysis (mol %) of L. major PPG Amino acid Native PPGa PPG after reduction and alkylationd PPG after thermolysin digestionb,c PPG/mild acid, soluble fractionc PPG/mild acid, insoluble fractionc Asx 1.7  ± 0.3 1.7 0.75  ± 0.1 0.5  ± 0 9.2  ± 0.2 Thr 1.4  ± 0.1 1.2 0.5  ± 0 0.5  ± 0 6.5  ± 0.1 Ser 51.8  ± 1.4 51.0 58.95  ± 0.5 58.6  ± 0.1 11.6  ± 0.2 Glx 1.7  ± 0.3 1.55 0.4 0.0 10.1  ± 0.5 Pro 14.4  ± 0.5 14.6 15.3  ± 0.1 16.0  ± 0 6.2  ± 0 Gly 1.5  ± 0.2 1.4 0.6  ± 0.3 0.6  ± 0.1 8.0  ± 0.3 Ala 21.25  ± 0.7 21.6 23.15  ± 0.2 23.4  ± 0 9.8  ± 0.2 Val 1.2  ± 0.2 1.1 0.0 0.0 5.6  ± 0.6 Met 0.25  ± 0.2e 0.3 0.0 0.0 2.1  ± 0 Ile 0.6  ± 0.1 0.8 0.0 0.0 3.35  ± 0.5 Leu 1.9  ± 0.2 2.2 0.35  ± 0.1 0.4  ± 0 9.9  ± 0.3 Tyr 0.4  ± 0.3e 0 0.0 0.0 3.7  ± 0.3 Phe 0.4  ± 0.4e 0.6 0.0 0.0 3.9  ± 0.3 His 0.2  ± 0.2e 0.4 0.0 0.0 1.65  ± 0 Lys 0.7  ± 0.1 (0.85)c 0.0 0.0 4.2  ± 0.2 Arg 0.5  ± 0.5e 0.7 0.0 0.0 4.3  ± 0.1 a  Average of three determinations. b  PPG was repurified by Superose 6 chromatography in the presence of 6  GdmCl. c  Average of two determinations. The standard deviations are indicated. d  Coelution of lysine and S-pyridylethylocysteine. e  High standard deviation caused by 0 readings in one of the analyses, respectively.

PPG was hydrolyzed by HCl under mild conditions that preferentially hydrolyze hexose-1-PO₄ linkages, whereas glycosidic linkages remain intact (5, 6, 28). Under these conditions, the carbohydrate was quantitatively (>99.8%) released as low molecular mass (<3000 Da) neutral and phosphorylated oligosaccharides. Alternatively, PPG was subjected to HF dephosphorylation that also released >99.5% of the carbohydrates as neutral glycans with HPLC peak patterns identical to that of mild acid-released and enzymatically dephosphorylated PPG glycans (not shown). The mild acid-released glycans of L. major PPG were separated by high pH anion exchange HPLC by a salt program that resolves both neutral and phosphorylated oligosaccharides. Three neutral (N2, N2, N3) and four phosphorylated glycans (P2, P3, P4a, P4b) were detected (Fig. 4A). N2 and N2 co-eluted on anion exchange HPLC with Gal1-4Man and Man1-2Man, respectively, which had been previously identified as cap structures in various Leishmania LPGs (8, 9, 10, 11, 12) and L. mexicana sAP (18). Methylation analysis (Table IV) and compositional analysis (not shown) were consistent with these structures. The anomeric configurations were confirmed by their susceptibility to -galactosidase and -mannosidase treatment, respectively. N3 contained Gal and Man in the ratio of 2:1. Methylation analysis (Table IV) and the degradation of N3 by -galactosidase to monosaccharides suggests the structure Gal1-3Gal1-4Man. This is corroborated by the co-elution of N3 with dephosphorylated P3 isolated from L. major LPG. In the case of the phosphorylated oligosaccharides, the detection of Gal-6-PO₄ as the only phosphorylated monosaccharide, the results of methylation analysis of P2, P3, P4a, and P4b after enzymatic dephosphorylation, and co-elution on HPLC with their previously characterized counterparts isolated from L. major LPG (Fig. 4, A and B) (7, 8) indicate the structures PO₄-6-Gal1-4Man (P2), PO₄-6(Gal1-3)Gal1-4Man (P3), PO₄-6(Ara1-2Gal1-3)Gal1-4Man (P4a), and PO₄-6(Gal1-3Gal1-3)Gal1-4Man (P4b), respectively. The hydrolysis of dephosphorylated P2, P3, and P4b to monosaccharides by -galactosidase confirms the anomeric configuration of their Gal residues and the carbohydrate sequence, whereas the resistance of P4a to this enzyme is in agreement with a terminal Ara capping the oligosaccharide.

Table IV. Methylation analysis of the L. major PPG oligosaccharides Deduced glycosidic linkage N2b N2 (+N2)a,b N3c P2a P3c P4a (+P2)a,b P4bc t-Arap 0 0 0 0 0 1.0 0 t-Galp 1.1 1.1 1.2 1.1 1.0 (1.1) (0.6) t-Manp 0 0.6 0 0 0 0 0 2-Manp 0 0.6 0 0 0 0 0 4-Manp 1.0 1.0 1.0 1.0 1.4 1.0 (1.0) (0.6) 2-Galp 0 0 0 0 0 1.5 0 3-Galp 0 0 0.8 0 1.05 1.2 1.9 a  GC-MS on a BP×70 column. b  N2- and P4a isolated by high pH anion exchange HPLC contained small amounts of N2 and P2, respectively. The PMAAs derived from these contaminants are depicted in italics. c  GC-MS on a cpsil5 column.

Methylation analysis of HF-treated unfractionated PPG glycans yielded the same derivatives as shown in Table IV. No 3,6-linked Man was detected, which together with the absence of GlcNAc indicates that N-linked glycans, if present, are of low abundance. The small amount of Glc detected in the compositional analysis (Table II) is most likely due to contaminants, since no corresponding glycan structure could be identified from high pH anion exchange HPLC.

The anomeric proton region of the one-dimensional ¹H NMR spectrum of native PPG is shown in Fig. 5A. The various signals were assigned on the basis of J_1, 2 coupling constants, heteronuclear J_HP couplings, chemical shifts, and comparisons with previously reported NMR data (5, 7, 18, 39). The doublet at 5.68 ppm (J_HP ~8 Hz) corresponds to the shift for the H-1 proton of Man1-PO₄ substituted at the 2-position with -Man (18). The anomeric signal of the terminal -Man linked to this residue appears as a well resolved peak at 5.05 ppm with a matching integral (Fig. 5A) to the peak at 5.68 ppm. The assignment of both residues has been confirmed by two-dimensional TOCSY. The intense overlapping doublet signals at 5.44 ppm correspond to 4-linked Man1-PO₄ (5, 7). The small doublet upfield of this peak is assigned as H-1 of terminal Ara_p1-2 based on the similarity of the H-1 (5.36 ppm) and H-2 (3.86 ppm, determined from a TOCSY spectrum, not shown) chemical shifts to those of the corresponding dephosphorylated oligosaccharide from L. major LPG (7, 8). The doublet at 4.64 ppm shows a chemical shift and large coupling constants (J_1, 2 ~7.8 Hz) corresponding to Gal in a -configuration. Its corresponding H-2 shift was determined to be 3.63 ppm from a DQFCOSY spectrum (not shown), and it is assigned as arising from a terminal Gal1-linked to the 3-position of an adjacent sugar residue. Linkage of Gal- to the 4-position rather than the 3-position of the adjacent residue results in an upfield shift for both H-1 and H-2. A cross-peak in the DQFCOSY spectrum between 4.46 and 3.55 ppm thus allowed the H-1 doublet (J_1, 2 ~7.8 Hz) at 4.46 ppm to be assigned to a terminal Gal1-4. The overlapped multiplet at 4.54 ppm was assigned to 3-linked PO₄-6-Gal1-4 since the DQFCOSY spectrum showed that the shifts of H-2 (3.71 ppm), H-3 (3.87 ppm), and H-4 (4.25 ppm) were almost identical to those of the corresponding residues in LPG (7). Finally, the doublet at 4.48 ppm (partially overlapped with the doublet at 4.46 ppm) was assigned to H-1 of the PO₄-6-Gal1-4 moiety. These results are consistent with the proposed structures for the mild acid-released glycans (Table V). ³¹P NMR spectroscopy of PPG in D₂O at pH 7.5 resulted in a single peak envelope (Fig. 5B). A non-Lorentzian peak shape, however, suggested a superposition of several different ³¹P nuclei. The chemical shift of the peak envelope is consistent with PO₄ in phosphodiester bonds. Titration experiments (pH 3.1-9.2) resulted in virtually no change (within 0.05 ppm) in chemical shift, corroborating a phosphodiester environment for the ³¹P nuclei. In contrast, ³¹P resonance signals of phosphomonoesters in proteins are expected to have different and pH-dependent chemical shifts (40), as demonstrated on the phosphoprotein phosvitin (not shown). Based on the intensity of the phosphodiester signal and the base-line noise level, phosphomonoester, if present, would account for less than 2% of the PO₄ detected in PPG.

Table V. Structure and relative abundance of oligosaccharides released from L. major PPG and LPG by mild acid hydrolysis Oligosaccharide Structure PPG LPGa mol %  N2 Gal1-4Man 13.0 1.0 N2 Man1-2Man 9.5 2.7 N3 Gal1-3Gal1-4Man 2.5 1.4 P2 PO4-6Gal1-4Man 43.0 8.6 P3 PO4-6[Gal1-3]Gal1-4Man 28.5 39.0 P4a PO4-6[Ara1-2Gal1-3]Gal1-4Man 2.0 16.2 P4b PO4-6[Gal1-3]2Gal1-4Man 1.5 20.9 P4c PO4-6[Glc1-3Gal1-3]Gal1-4Man NDa 0.2 P5a PO4-6[Ara1-2Gal1-3Gal1-3]Gal1-4Man ND a 4.4 P5b PO4-6[Gal1-3]3Gal1-4Man ND a 4.8 P6 PO4-6[Gal1-3]4Gal1-4Man ND a 0.8 Ratio of phosphorylated glycans to neutral glycans 3.0:1 18.6:1 a  For the structural analysis of LPG cap oligosaccharides and phosphooligosaccharides and the assignment of their peaks see McConville et al. (7, 8, 11): ND = not detected.

To resolve the overlapping ³¹P NMR peaks a HMBC spectrum (41) of PPG in D₂O was recorded. This shows only protons having a long range spin-spin coupling interaction with ³¹P nuclei. A major advantage of this two-dimensional technique is the resolution of partially overlapping ³¹P NMR signals by differing chemical shifts of ¹H resonance signals. Three signals are detected in the two-dimensional contour plot (Fig. 5C). The two well resolved signals in the downfield region correspond to a 2-linked Man1-O-P and a 4-linked Man1-O-P with ¹H chemical shifts of 5.68 and 5.44 ppm, respectively. The signal at 4.06 ppm falls in a crowded region of the ¹H spectrum. Initially, it was believed that it corresponded to the linkages -O-6-C₂-Gal and -O-C₂-Ser, which are both abundant in PPG (Table II and see below Fig. 6A). HMBC experiments with authentic Ser(P) and Gal-6-PO₄ confirmed that both types of CH₂/P-correlations were detectable and that the ¹H chemical shifts of the CH₂ groups of both compounds occurred in this region of the ¹H spectrum. However, subsequent HMBC spectra on phosvitin, which contains abundant P-O-CH₂-Ser linkages, did not produce cross-peaks in HMBC spectra, despite their detection in Ser(P) itself. The difficulty in detecting the latter linkage in the intact PPG (and phosvitin) is likely due to the relatively high degree of immobilization of the P-O-CH₂-Ser moiety in these proteins. The resultant fast T₂ relaxation leads to loss of magnetization in the HMBC experiment for this linkage, whereas the more mobile P-O-6-CH₂-Gal linkage is readily detected. It therefore appears that the upfield signal at 4.06 ppm in Fig. 5C may be due to P-O-6-CH₂-Gal alone.³ These data provide direct evidence for the presence of the sequence 4-Man1-PO₄-6Gal in PPG. Interestingly, no Gal-6-PO₄ peak appears at the ³¹P shift of the 2-linked Man1-PO₄. A possible explanation could be that Man1-2Man1-PO₄ is predominantly directly linked to the ``HMBC-invisible'' Ser and not to 6-Gal.

The ratios of the different neutral and phosphorylated PPG glycans were determined by high pH anion exchange HPLC (11, 18). The molar ratios of Man, Gal, Gal-6-PO₄, and Ara in L. major PPG calculated from peak integration of the PPG oligosaccharides are consistent with the results obtained by monosaccharide analysis (see Table II). The dominating structures in PPG are P2 and P3 followed by N2 and N2, whereas N3, P4a, and P4b are only minor components (Table V). In LPG purified from the parasite cell pellet grown in the culture supernatant used for PPG isolation, all the PPG oligosaccharides are also detected. However, P3 and P4b followed by P4a are the dominant structures in LPG, whereas P2, by far the most abundant structure in PPG, is only a minor component in LPG (Fig. 4B and Table V). LPG contains additional, more highly substituted phospho-oligosaccharides (P5a, P5b, P6, and P4c, (7), see Fig. 4B), which were not detected in PPG. The average ratio of phosphorylated oligosaccharides to neutral oligosaccharides is 18.6:1 in the case of LPG and 3:1 in PPG (Table V).

Mild acid hydrolysis and aqueous HF treatment released almost quantitatively the PPG glycans from the protein backbone (see above). Since these hydrolysis conditions do not cleave glycosidic bonds (5, 6, 29, 42, 44), it is likely that all the glycans were linked to the protein by mild acid- and HF-labile phosphodiesters similar to the main glycans in L. mexicana sAP (18) and not by N- or O-glycosidic linkages. Phosphoamino acid analysis of the PPG protein backbone indicated the presence of large amounts of Ser(P) (Fig. 6A). No phosphothreonine or phosphotyrosine was detected. Ser(P) in PPG was not susceptible to alkaline phosphatase treatment (Fig. 7). Heating prior to alkaline phosphatase addition or removal of all the glycans by mild acid hydrolysis alone did not have an impact on the Ser(P) content of PPG. However, a combination of mild acid hydrolysis and alkaline phosphatase treatment resulted in the complete enzymatic dephosphorylation of PPG-associated Ser(P) (Fig. 7). Phosvitin, a highly phosphorylated egg yolk protein containing up to 100 Ser(P) residues (43), was completely dephosphorylated by alkaline phosphatase without any pretreatment, which demonstrates that protein-bound, clustered Ser(P) is not inherently resistant to enzymatic dephosphorylation (Fig. 7). These experiments indicate that all the Ser(P) residues in PPG are in mild acid-labile phosphodiester bonds with glycans, consistent with the results of the ³¹P NMR experiments on intact PPG (see above). Based on the degree of Ser(P) hydrolysis to Ser and phosphate under the partial acid hydrolysis conditions (30.6%, Fig. 6B) and the relative response factors of Ser(P) and serine as phenylthiocarbamyl derivatives (1:1.09), the degree of Ser phosphorylation in native PPG was estimated to be 84%. Therefore, Ser(P) linked to phosphoglycans and/or neutral glycans accounts for at least 43 mol % of the amino acids in the polypeptide backbone of PPG.

The selectivity of mild acid hydrolysis for hexose 1-phosphate linkages in phosphoglycans has been well established (5, 6, 7, 28, 42). In contrast, its effect on peptide bonds has not been described. Bovine serum albumin (BSA) and phosvitin were used as model proteins and subjected to mild acid hydrolysis. Both BSA and especially phosvitin were cleaved into a large array or smear of smaller polypeptides as indicated by their higher mobilities in SDS-PAGE (Fig. 8, lanes 1-4). Similarly, mild acid hydrolysis of PPG led to a broad area of hydrolysis products spanning an area from an apparent molecular mass of 90 kDa to the dye front of the gel (Fig. 8, lane 5). This broad staining pattern is not due to the released phosphoglycans, which migrate at the front of the gel (37). Protein sequencing of intact PPG did not yield any amino acids indicating a blocked N terminus, whereas after mild acid hydrolysis Ser, Ala, and Pro were detected as N-terminal residues. This result provides further evidence for peptide bond cleavage in PPG during mild acid deglycosylation. When large amounts of PPG (>500 µg) were subjected to mild acid hydrolysis, the formation of a precipitate was observed. This precipitate was insoluble in nondenaturing aqueous buffers and could be recovered as a pellet after centrifugation. The washed pellet fraction and the combined supernatants were subjected to amino acid analysis (Table III). The pellet fraction contained only about 5% of the amino acids; it was greatly depleted for Ser, Ala, and Pro and showed increased readings for all other amino acids proportional to their previous abundance in intact PPG. On SDS-polyacrylamide gels, this minor peptide fraction gave rise to a broad area spanning from approximately 90 to 40 kDa apparent molecular mass (Fig. 8, lane 6). About 95% of the amino acids were recovered in the mild acid hydrolysis supernatant. This soluble fraction was strongly enriched for Ser, Ala, and Pro, which accounted for 98 mol % of its amino acids (Table III). On SDS-PAGE this fraction was indistinguishable from the total hydrolysate (Fig. 8, compare lanes 5 and 7).

Based on these observations, two different approaches were chosen to obtain antisera against the deglycosylated protein backbone of PPG. 1) The insoluble peptide pellet of mild acid-hydrolyzed PPG was used for the immunization of rabbits, which yielded an antiserum directed against mild acid-insoluble peptides of PPG (anti-MIP serum). 2) Since the soluble fraction contains more than 58% serine (Table III), the presence of long polyserine stretches in the polypeptide backbone was considered likely, and antisera against synthetic polyserine were raised in rabbits (anti-PS serum).

PPG chromatographed on Superose 6 equilibrated in 6  GdmCl to minimize aggregation showed a co-eluting peak of phosphate and material absorbing at 280 nm. In dot blots the PPG peak fractions were strongly recognized by the anti-MIP and anti-PS antisera (Fig. 9A). The elution position of PPG was similar to that observed in 250 m ammonium acetate and suggested an apparent molecular mass of more than 2000 kDa also under dissociating conditions. Reduction and alkylation of PPG with either iodoacetamide or vinylpyridine did not result in any change in its elution position on Superose 6 (not shown), and its amino acid composition was nearly identical to that of native PPG (Table III). Thermolysin treatment leads to higher mobility of PPG on agarose gels indicative of some peptide bond cleavage (Fig. 3). On Superose 6, however, the majority of the phosphate (>90%) of thermolysin-digested PPG still eluted near the void volume, but the absorbance at 280 nm of these fractions was greatly diminished compared with the native PPG. Peak fractions reacted on dot blots strongly with the anti-phosphoglycan mAb WIC108.3 but more weakly with anti-MIP serum (Fig. 9B). Amino acid analysis of pooled peak fractions showed a composition highly enriched for Ser, Ala, and Pro (together >97 mol %), virtually identical to the composition of the soluble peptide fraction of mild acid-hydrolyzed PPG (Table III). These results indicate that while parts of PPG may be susceptible to thermolysin, the molecule also contains large highly glycosylated and phosphorylated polypeptide regions consisting almost exclusively of Ser, Ala, and Pro, which are resistant to proteinases.

PPG was dephosphorylated and deglycosylated by aqueous HF treatment. The impact of aqueous HF on peptide bonds was tested on BSA, which was included as a control in each set of incubations. In contrast to mild acid hydrolysis, HF did not result in peptide bond cleavage detectable on SDS-PAGE (Fig. 10, lanes 1 and 2). HF treatment of PPG resulted in the formation of a protein precipitate. After solubilization by the addition of 6  GdmCl and Superose 6 gel filtration, a single phosphate peak was detected near the inclusion volume, the expected position of free phosphate as a reaction product of HF dephosphorylation (Fig. 9C). In addition, a protein peak eluting earlier than the 2000-kDa marker compound was observed. Fractions collected over this protein peak strongly reacted with the anti-PS and the anti-MIP sera on dot blots, whereas reaction with anti-phosphoglycan mAb WIC108.3 was very weak or absent, as expected for a deglycosylated product (Fig. 9C). Therefore, it is likely that this protein peak corresponds to (part of) the polypeptide backbone of PPG. This is corroborated by amino acid analysis of peak fractions desalted by filtration onto polyvinylidene difluoride membrane supports, which showed a similar amino acid composition (>39% Ser, 15% Ala, 10% Pro) as native PPG. However, the elution close to the void volume, which was not altered by prior reduction (not shown), may indicate aggregation, even in the presence of GdmCl. The GdmCl-containing Superose 6 protein peak fractions of HF-treated PPG were used to raise polyclonal antibodies against the dephosphorylated and deglycosylated polypeptide backbone of PPG (anti-HFP serum).

Native and HF-treated PPG were subjected to SDS-PAGE and immunoblotting using the PPG/LPG glycan-specific mAb WIC108.3 as well as the anti-PS, anti-MIP, and anti-HFP serum as detecting reagents. Native PPG migrating mainly in the stacking region of SDS-polyacrylamide gels was strongly recognized by mAb WIC108.3, the anti-MIP, and the anti-HFP serum (Fig. 10, lanes 3, 9, and 12), whereas its reaction with anti-PS serum was weak (Fig. 10, lane 6). Some additional weak bands between the 94- and the 43-kDa marker protein (Fig. 10, lanes 3, 9, and 12) may correspond to small amounts of degradation products. After HF treatment, PPG did not react with WIC108.3 (Fig. 10, lane 4), whereas the anti-PS, the anti-MIP, and the anti-HFP serum recognized protein bands near the position of the 67-kDa and above the 94-kDa marker protein (Fig. 10, lanes 7, 10, and 13). In addition all three antisera, especially the anti-HFP serum, reacted with a smear in the stacking gel and the top of the separating gel, possibly aggregated PPG polypeptide(s). None of the antisera reacted with LPG (Fig. 10, lanes 8, 11, and 14), which is strongly recognized by mAb WIC108.3 (Fig. 10, lane 5). This result indicates that the three antisera are predominantly directed against peptide epitopes of PPG and not against glycan determinants. The stronger reaction of PPG with anti-PS serum after HF dephosphorylation may reflect the unmasking of Ser-rich polypeptide regions previously covered by phosphoglycans.

Inspection of purified L. major PPG by electron microscopy revealed long thread-like filaments with similar appearance after glycerol spraying and rotary shadowing (Fig. 11, a and b) as well as negative staining (Fig. 11c). The length of extended filaments is variable and can exceed 6 µm. The diameter of the majority of the PPG molecules appears to be homogeneous and may be around 3-6 nm. Occasionally thicker filaments with approximately twice this diameter were observed (Fig. 11b). Whether this subfraction is due to side by side-arranged filaments or corresponds to a distinct population of molecules remains to be determined. No evidence for branched structures has been found. The filaments exhibit high flexibility; they tend to curl up (Fig. 1, A-C) and can form large clumps on grids (not shown). These clumps were especially prevalent after labeling with monoclonal antibodies on grids, possibly due to cross-linking of flexible parts of the molecules not stably attached to the support (Fig. 11, e and f). PPG strands could be heavily labeled with the mAbs LT6, WIC79.3, AP3 (Fig. 11, d-f), and WIC108.3 (not shown), whereas a control antibody (LT8.2) showed very weak binding (Fig. 11g). These immunoelectron microscopy experiments confirm the modification of filamentous PPG by P2, P3/P4b, and N2 at the ultrastructural level, since these glycans are part of the epitopes recognized by the mAbs.

DISCUSSION

Previous studies have shown that Leishmania promastigotes secrete network-forming filamentous material, which strongly reacts with anti-phosphoglycan antibodies (3, 20). Leishmania also secrete a phosphoglycan-modified acid phosphatase, which is organized as filaments in some species (17). In contrast to all other Leishmania species investigated so far, L. major does not secrete this enzyme (46), the presence of which c