Characterization of a Novel GDP-mannose:Serine-protein Mannose-1-phosphotransferase from Leishmania mexicana *

Protozoan parasites of the genusLeishmania secrete a number of glycoproteins and mucin-like proteoglycans that appear to be important parasite virulence factors. We have previously proposed that the polypeptide backbones of these molecules are extensively modified with a complex array of phosphoglycan chains that are linked to Ser/Thr-rich domains via a common Manα1-PO4-Ser linkage (Ilg, T., Overath, P., Ferguson, M. A. J., Rutherford, T., Campbell, D. G., and McConville, M. J. (1994) J. Biol. Chem. 269, 24073–24081). In this study, we show that Leishmania mexicana promastigotes contain a peptide-specific mannose-1-phosphotransferase (pep-MPT) activity that adds Manα1-P to serine residues in a range of defined peptides. The presence and location of the Manα1-PO4-Ser linkage in these peptides were determined by electrospray ionization mass spectrometry and chemical and enzymatic treatments. The pep-MPT activity was solubilized in non-ionic detergents, was dependent on Mn2+, utilized GDP-Man as the mannose donor, and was expressed in all developmental stages of the parasite. The pep-MPT activity was maximal against peptides containing Ser/Thr-rich domains of the endogenous acceptors and, based on competition assays with oligosaccharide acceptors, was distinct from other leishmanial MPTs involved in the initiation and elongation of lipid-linked phosphoglycan chains. In subcellular fractionation experiments, pep-MPT was resolved from the endoplasmic reticulum marker BiP, but had an overlapping distribution with thecis-Golgi marker Rab1. Although Man-PO4residues in the mature secreted glycoproteins are extensively modified with mannose oligosaccharides and phosphoglycan chains, similar modifications were not added to peptide-linked Man-PO4residues in the in vitro assays. Similarly, Man-PO4 residues on endogenous polypeptide acceptors were also poorly extended, although the elongating enzymes were still active, suggesting that the pep-MPT activity and elongating enzymes may be present in separate subcellular compartments.

Parasitic protozoa of the genus Leishmania cause a spectrum of human and animal diseases that are transmitted by a sand fly vector. During their development in the digestive tract of the sand fly, these parasites differentiate from non-infective procyclic promastigotes to infective metacyclic promastigotes that target mammalian macrophages when introduced into the host during the insect's blood meal. Following their internalization into the macrophage phagolysosome, metacyclic promastigotes differentiate into a replicative amastigote stage and eventually rupture the host cell and perpetuate disease by infecting other host cells. A number of cell-surface and secreted virulence factors are thought to be crucial for the survival of these different developmental stages in their respective host environments. These include an abundant glycosylphosphatidylinositol-anchored lipophosphoglycan (LPG) 1 (1)(2)(3) and a number of secreted glycoproteins and proteophosphoglycans (PPGs) (reviewed in Refs. 4 and 5). Strikingly, both the cellsurface LPGs and the secreted molecules are elaborated with structurally related phosphoglycan chains that are thought to be the major functional determinants of these molecules.
The secreted glycoproteins and proteoglycans of Leishmania have been shown to form distinct macromolecular complexes in the flagellar pocket and extracellular milieu (4). The most intensively characterized of these molecules are the secreted acid phosphatases (sAPs), which aggregate into large pearl-like filamentous polymers (6,7). Early studies indicated that sAPs were heavily glycosylated and phosphorylated enzymes that contained glycan epitopes characteristic of LPGs (6,8,9). We have recently shown that Leishmania mexicana sAP is extensively modified with Man␣1-PO 4 residues that are linked to serine residues in the polypeptide backbone (10). This unusual type of linkage, in which a monosaccharide is linked to protein via a phosphodiester bridge, has been termed phosphoglycosylation (11) and may be widespread in several lower eukaryotes (12). In L. mexicana, the Man␣1-PO 4 residues can be further elaborated with ␣1-2-linked mannose oligosaccharides or short chains of phosphorylated di-or trisaccharides (10), which are also found in the long phosphoglycan chains (as capping structures or internal repeat units, respectively) of L. mexicana LPG (see Fig. 1). Interestingly, the nature of these modifications may be influenced by the size of the Ser/Thr-rich repeat domains in the polypeptide backbone. For example, sAP-1 contains a relatively short Ser/Thr-rich domain and is modified primarily with mannose oligomers, whereas sAP-2, which contains a longer stretch of Ser/Thr-rich repeat sequences, is extensively modified with short phosphoglycan chains (7,10). In 1 The abbreviations used are: LPG, lipophosphoglycan; PPG, proteophosphoglycan; aPPG, amastigote proteophosphoglycan; sAP, secreted acid phosphatase; MPT, mannose-1-phosphotransferase; pep-MPT, peptide-specific mannose-1-phosphotransferase; CHAPS, 3-[(3-chloramidopropyl)dimethylammonio]-1-propanesulfonic acid; RP-HPLC, reversed-phase high performance liquid chromatography; HPTLC, high performance thin-layer chromatography; ESI-MS, electrospray ionization mass spectrometry. addition to sAP, two heavily phosphoglycosylated PPGs have been recently characterized from leishmanial promastigotes and amastigotes (13)(14)(15)(16). Promastigote PPGs are produced by all species of Leishmania and form a network of unbranched filaments in the center of aggregated promastigotes (14,16). Amastigote PPGs (aPPGs) are produced by most species and accumulate to very high levels in the lumen of the parasitophorous vacuole and in the extracellular space of lesions (13,15). The polypeptide backbones of PPGs differ from sAPs in being rich in proline and alanine as well as serine and threonine and are more extensively modified with phosphoglycan chains, which may account for as much as 90% of the molecule (13)(14)(15). Whereas the phosphoglycan chains of Leishmania major promastigote PPG are similar to those of the corresponding LPG, L. mexicana aPPGs are elaborated with novel stage-specific phosphoglycans that are distinct from those found on both promastigote sAPs and LPGs (15). However, all these glycans are thought to be linked exclusively to the polypeptide backbone via a Man␣1-PO 4 -Ser core sequence (14,15).
Recent functional studies have shown that these molecules may be important for parasite infectivity and survival in both the insect and the mammalian host. L. mexicana aPPG is a potent activator of the complement cascade via the mannanbinding lectin pathway and may effectively deplete the local supply of complement components needed for parasite lysis (17). The release of PPGs into the phagolysosome may also inhibit membrane fusion events and contribute to the formation of large parasitophorous vacuoles, which are characteristic of New World species such as L. mexicana (18). On the other hand, secretion of sAP and promastigote PPG filamentous networks by the promastigote stages may be responsible for the tendency of cells to aggregate into large clusters in stationary growth. Cell aggregation may improve the efficiency of transmission during the sand fly bite or afford cells within the cluster some protection from complement lysis during the early stages of infection (4).
As this type of protein glycosylation is unique to Leishmania, the enzymes involved in initiating and assembling these glycans may be potential targets for new anti-leishmanial drugs. At present, several enzyme activities have been detected in cell-free assays and detergent extracts that are involved in the synthesis of the phosphoglycan chains of LPG, but that may also be involved in the synthesis of the shorter phosphoglycan chains of sAP and PPGs. These include two elongating enzymes, a putative ␣-mannose-1-phosphotransferase (MPT) and a ␤1-4-galactosyltransferase, that transfer Man␣1-PO 4 and Gal from GDP-Man and UDP-Gal, respectively, to form the repeating Gal␤1-4Man␣1-PO 4 backbone of both the LPG and PPG glycans ( Fig. 1) (19,20). In addition, a ␤1-3-galactosyltransferase activity that adds the L. major-specific side chains to these repeat units has also been characterized (21). Interestingly, a separate MPT activity is thought to be involved in adding the first Man-PO 4 residue to the LPG anchor precursor and thus to be required to initiate LPG phosphoglycan biosynthesis (22). As one of the enzymes involved in synthesizing the LPG anchor is localized to the Golgi apparatus (23), it is likely that both the initiating MPT and phosphoglycan chain-elongating and -branching enzymes are also localized in either a Golgi or post-Golgi compartment. This is consistent with the finding that a GDP-Man transporter required by these MPTs is also localized to the Golgi apparatus (24) and that phosphoglycan chain elongation is inhibited by the Golgi-perturbing ionophore, monensin (25).
In this study, we have identified a peptide-specific MPT (pep-MPT) activity from L. mexicana promastigotes that is most likely involved in initiating protein phosphoglycosylation. Mass spectrometry of the glycopeptides containing the Man-PO 4 modification provides the first direct characterization of this novel linkage. We show that pep-MPT is most active against peptides containing the Ser/Thr-rich sequences of endogenous polypeptide acceptors and that it is not inhibited by glycan acceptors of the LPG phosphoglycan-initiating and -elongating MPTs. Furthermore, we provide evidence that this enzyme occurs in a distinct subcompartment of the Golgi anchor of LPG is modified with a single phosphoglycan chain comprising a long domain of phosphodi-and phosphotrisaccharide repeat units and a mannose oligosaccharide cap at the nonreducing terminus (2,3). In contrast, a Ser/Thr-rich domain in the secreted sAP glycoproteins is modified either with mannose oligosaccharides or short phosphoglycan chains that are capped with mannose oligosaccharides. Both classes of glycans are thought to be attached to serine residues via a Man␣1-PO 4 linkage (10). The more complex branched phosphoglycans of the L. mexicana aPPGs are not shown (15). apparatus from enzymes involved in phosphoglycan chain elongation.

EXPERIMENTAL PROCEDURES
Materials-Alkaline phosphatase, GDP-Man, UDP-Gal, and stachyose were from Sigma; jack bean ␣-mannosidase was from Boehringer Mannheim. The synthetic oligosaccharide acceptor L2 (Gal␤1-4Man␣1-PO 4 -(CH 2 ) 8 CHϭCH 2 ) was generously provided by Professor Michael Ferguson (University of Dundee). GDP-[ 3 H]Man was prepared using previously described methods (26,27). Briefly, [2-3 H]Man-6-PO 4 was synthesized enzymatically from [2-3 H]Man (NEN Life Science Products) using yeast hexose kinase and then converted to the sugar nucleotide using a mixture of yeast proteins supplemented with snake venom pyrophosphatase and glucose-1,6-diphosphate. The yeast proteins, corresponding to a 50 -70% ammonium sulfate cut, were prepared from protease A-deficient yeast strain SC295. GDP-[ 3 H]Man was purified on a column (1 ml) of concanavalin A-Sepharose (Amersham Pharmacia Biotech) that was washed with 15 mM ammonium acetate, 1 mM MgCl 2 , and 1 mM CaCl 2 and then eluted with the same buffer containing 50 mM ␣-methylmannoside. Fractions containing radioactivity were desalted on a column of Sephadex G-10 eluted in water. Polyclonal antibodies against L. major Rab1 protein (28) and Trypanosoma brucei BiP (29) were generously provided by Dr. Emanuela Handman (Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia) and Dr. James Bangs (Department of Biochemistry, University of Wisconsin, Madison, WI), respectively.
Synthesis of Peptide Substrates-Peptides were synthesized using Fmoc (N-(9-fluorenyl)methoxycarbonyl) chemistry on a Applied Biosystems Model 431A peptide synthesizer. Peptides were cleaved from the resin using 95% trifluoroacetic acid (3 h, 25°C), recovered by extraction with diethyl ether, and purified on a C 18 reversed-phase column (4.6, inner diameter, ϫ 100 mm; Brownlee) eluted with acetonitrile (0 -60%) in 0.1 M ammonium acetate over 20 min at a flow rate of 1 ml/min. Reductive methylation of sapp-1 was carried out in 0.2 M borate buffer, pH 9.0 (140 l), containing 16 mM formaldehyde and 0.1 mM NaB 3 H 4 (5 mCi) at 0°C for 10 min (30). The mixture was acidified with 1 M acetic acid to remove excess NaB 3 H 4 , and the radiolabeled peptides were recovered after application to a Sep-Pak C 18 cartridge (Waters) following elution with 40% acetonitrile.
Mannose-1-phosphotransferase Assays-pep-MPT activity was measured using two assays. In the first method, microsomal membranes (2 ϫ 10 7 cell eq, 20 -60 g of protein) were resuspended in buffer A (50 l) containing GDP-[ 3 H]Man (160,000 cpm, 0.1-0.5 mM) and 0.4 mM synthetic peptide and incubated at 32°C for 30 min. Triton X-100-solubilized membranes were assayed under identical conditions, except that the reaction mixture contained 0.1% Triton X-100 and assays were performed at 16°C. The reaction mixture was diluted with 0.1 M ammonium acetate (950 l), and the enzyme reaction was terminated by boiling (3 min). Precipitated proteins was removed by centrifugation (14,000 ϫ g, 5 min), and the supernatant was loaded onto Sep-Pak C 18 cartridges pre-equilibrated in 0.1 M ammonium acetate. After washing the cartridge with 0.1 M ammonium acetate (10 ml) to remove unincorporated GDP-[ 3 H]Man, [ 3 H]Man-labeled peptide was eluted with 40% acetonitrile (3 ml), and eluted radioactivity was measured by scintillation counting.
In the second method, Triton X-100-solubilized membranes (2 ϫ 10 7 cell eq, 20 -60 g of protein) were diluted in buffer A (15 l) containing 0.1% Triton X-100, GDP-Man (0.5 mM), and 3 H-labeled peptide (500,000 cpm, 0.16 mM). Assay mixtures were incubated at 16°C for 15 min. In the substrate competition experiments described in the legend to Fig. 6, the reaction mixtures contained unlabeled peptide or glycan acceptors at 0 -5 mM final concentration. Reactions were terminated by boiling (3 min), and the precipitated protein was removed by centrifugation at 14,000 ϫ g for 5 min. Underivatized and mannosylated 3 H-labeled peptides in the reaction mixture were resolved from each other by analytical thin-layer electrophoresis of 4 l of the reaction mixture. Alternatively, 3 H-labeled peptides were recovered on Sep-Pak C 18 cartridges as described above for subsequent chemical and enzymatic analysis.
Thin-layer Electrophoresis-Man-PO 4 -modified 3 H-labeled peptides were resolved from unmodified 3 H-labeled peptides by thin-layer electrophoresis on plastic-backed cellulose thin-layer sheets (20 ϫ 20 cm; Macherey Nagel). Aliquots of the reaction mixture (containing 100,000 cpm) were spotted onto the cellulose sheets, which were then saturated with pyridine/acetic acid/water (1:10:89, v/v) and subjected to electrophoresis (500 V, 30 mA, 75 min) under a layer of petroleum spirit and pyridine/acetic acid/water (1:10:89, v/v) in each electrophoresis chamber. Radioactivity on the dried cellulose sheets was detected using an automatic TLC scanner (Berthold) and/or by fluorography as described above. Incorporation of radioactivity into phosphoglycosylated peptides was quantitated by scraping the cellulose containing the bands using the fluorograph as template, extracting the cellulose with 30% acetonitrile (2 ϫ 250 l), and liquid scintillation counting of the extracts.
Electrospray Ionization Mass Spectrometry-Unmodified and phosphoglycosylated peptides were analyzed with a triple quadrupole mass spectrometer (Finnigan MAT Model TSQ-700) equipped with a Finnigan MAT electrospray ionization (ESI) source and rapid capillary RP-HPLC. The column used in this study (0.2, inner diameter, ϫ 150 mm; Vydac C 18 ) was fabricated and operated as described elsewhere (34). The column was developed with a linear 30-min gradient at 1.6 l/min from 0 to 100% solvent B, where solvent A was 0.1% trifluoroacetic acid and solvent B was 60% acetonitrile containing 0.1% trifluoroacetic acid (35). The ESI needle was operated at Ϫ4.5 kV. The sheath liquid was methoxyethanol delivered at 3 l/min. Nitrogen sheath and auxiliary gasses were supplied at 30 p.s.i. and 15 units (arbitrary units), respectively. The heated capillary was set at 150°C. Mass spectra were collected every 3 s in centroid mode. Peptide masses were calculated using Finnigan MAT BIOMASS TM software. For tandem mass spectrometric peptide sequence analysis, quadrupole Q1 was operated with a resolution of 2-2.5 Da and Q3 with a resolution of 1-1.5 Da (35). The collision cell offset voltage was calculated by multiplying the mass of the ion selected for collision-induced dissociation by 0.04. The daughter ion offset voltage was set at twice the collision cell offset value. The parent ion offset voltage was set at one-third the value of the daughter ion offset voltage. Argon was used as the collision gas at a pressure of 2-2.5 millitorr.
Characterization of Endogenous Polypeptide and LPG Acceptors-Sonicated microsomal membranes were incubated with GDP-[ 3 H]Man in the presence of exogenous peptide acceptors and UDP-Gal as described above. Membranes were recovered by centrifugation (14,000 ϫ g, 5 min), washed with water to remove residual peptide, and recentrifuged, and the supernatants were combined for recovery of the peptide on a Sep-Pak column as described previously. The membrane pellet was subsequently extracted twice in 1% Triton X-100 (200 l) for 10 min at 30°C to remove LPG and soluble phosphoglycan and proteoglycans. Insoluble material was recovered by centrifugation; the supernatants were combined and subjected to two-phase partitioning with 1-butanol; and the detergent-free aqueous phase was dried in a Speed-Vac concentrator. This material was resuspended in 0.1 M NH 4 OAc containing 5% 1-propanol and loaded onto a minicolumn of octyl-Sepharose equilibrated in the same buffer. The column was washed with 0.1 M NH 4 OAc containing 5% 1-propanol and then with 40% 1-propanol to elute bound LPG (36). The Sep-Pak-eluted peptide, octyl-Sepharose-purified LPG, and Triton X-100-insoluble material were hydrolyzed in 40 mM trifluoroacetic acid (12 min, 100°C), and the acid was removed under reduced pressure in a Speed-Vac concentrator (36). Acid-released glycans were dephosphorylated with alkaline phosphatase and desalted by passage over a column of AG 50-X12 (H ϩ ) over AG 3-X8 (OH Ϫ ) for analysis by HPTLC (3).
Subcellular Fractionation-L. mexicana promastigotes (8 ϫ 10 8 ) were harvested by centrifugation (750 ϫ g, 10 min), washed with phosphate-buffered saline, and then resuspended (2 ϫ 10 8 cells/ml) in assay buffer A supplemented with 0.25 M sucrose. The cell suspension was transferred to a prechilled nitrogen cavitation bomb (Kontes Glass Co.) equilibrated at 450 p.s.i. N 2 pressure for 15 min at 0°C, and cells were lysed by abruptly releasing the pressure. Equilibration and expulsion of the suspension were repeated twice more to achieve complete cell lysis. Cell debris and nuclei were removed by centrifugation of the lysate at 3000 ϫ g for 10 min, and the supernatant was layered on top of a linear sucrose gradient. The sucrose gradient (density ϭ 1.05-1.27 g/ml) was prepared by layering 10 0.8-ml fractions (0.25-2 M sucrose in 0.25 mM HEPES-NaOH, pH 7.4) over a sucrose cushion (2.5 M) in Ultraclear centrifugation tubes (Beckman Instruments) and centrifuging at 218,000 ϫ g for 1 h at 4°C in a Beckman L-80 ultracentrifuge using an SW 41 rotor. Organelles in the cell lysate were fractionated by equilibrium centrifugation at 218,000 ϫ g for 6 h at 4°C. Fractions (0.5 ml) were collected from the bottom of the tube, and densities were calculated by measuring refractive index. The distribution of leishmanial BiP and Rab1, markers for the bulk endoplasmic reticulum and cis-Golgi cisternae, respectively (28,29), were determined by immunoblotting. Proteins in each fraction were precipitated in biphasic mixtures of chloroform/methanol/water (32), measured by the BCA assay, separated by SDS-polyacrylamide (12%) gel electrophoresis, and transferred to nitrocellulose by electroblotting. Strips of nitrocellulose corresponding to regions containing the BiP (70 kDa) and Rab1 (25 kDa) proteins were cut and processed individually. For detection of BiP, the nitrocellulose was blocked with 0.05% Tween 20 in Tris-buffered saline, incubated with rabbit anti-BiP antibody (1 h), washed with 0.05% Tween 20 in Tris-buffered saline, and then probed with horseradish peroxidase-conjugated anti-rabbit antibody (1:10,000 dilution; Silenus Laboratories, Pty., Ltd.) in 0.05% Tween 20 in Tris-buffered saline containing 1% powdered skim milk (1 h) and visualized with the ECL Western detection system (Amersham Pharmacia Biotech). Nitrocellulose strips containing Rab1 protein were probed in the same way, except that the nitrocellulose was initially blocked with 0.05% Tween 20 in Tris-buffered saline with 1% powdered skim milk, the rabbit anti-L. major Rab1 antibody was used at 1:500 dilution, and incubation with the secondary antibody was done in 0.05% Tween 20 in Tris-buffered saline.

Phosphoglycosylation of Peptides
Containing the Ser/Thrrich Repeat Sequences of L. mexicana sAP-We have previously shown that the major secretory proteins of L. mexicana promastigotes are modified with phosphoglycan chains that are probably linked to the polypeptide backbone via a Man␣1-PO 4 -Ser linkage (10). To confirm this linkage by mass spectrometry and to develop an assay for the initiating enzyme, a series of peptides containing the Ser/Thr-rich peptide repeats from L. mexicana sAP-1 and sAP-2 (7) were synthesized (sapp-1, -2, -3, and -4 in Table I) Fig. 2, incubation of the membranes with sapp-1 (containing two Ser-rich motifs) generated at least five 3 H-labeled peaks that eluted earlier than the unmodified peptide on RP-HPLC. These peaks were not detected when membranes were incubated in the absence of either the exogenous peptide or GDP-Man (data not shown). Treatment of the total peptide mixture or the individual HPLCpurified peaks with either jack bean ␣-mannosidase (data not shown) or mild trifluoroacetic acid (to selectively cleave hexose 1-phosphate linkages) generated a single labeled component in each case that comigrated with mannose on HPTLC (Fig. 2C). These data suggested that sapp-1 was being variably modified with one or more Man␣1-PO 4 residues. This was confirmed by positive ion ESI-MS of the HPLC-purified phosphomannosylated peaks 1 and 2 in Fig. 2. Multiply charged ions were obtained that were in agreement with the calculated mass for sapp-1 modified with one or two hexose phosphate residues, respectively (Table II). As the ionization of these phosphomannosylated peptides was relatively poor, a smaller peptide (sapp-2) containing a single Ser/Thr-rich repeat unit and a C-terminal lysine residue was used as substrate. As with sapp-1, several [ 3 H]Man-labeled peaks were generated when this peptide was incubated with sonicated membranes (Fig.  3A). Positive ion ESI-MS of the two major peaks contained both [M ϩ H] ϩ and [M ϩ 2H] 2ϩ ions corresponding to sapp-2 with either one (Fig. 3B) or two (Fig. 3C) hexose-PO 4 residues (Table  II). Fragment ions corresponding to the loss of one or two hexose residues were also evident in these mass spectra, indicating that the Man-PO 4 linkage is relatively labile under the ionization conditions employed. Taken together, these data indicate that peptides containing the same Ser/Thr-rich domains as endogenous polypeptides are modified with one or more single Man␣1-PO 4 residues, providing the first evidence for the presence of a pep-MPT activity.
Collision-induced dissociation of the monophosphomannosylated sapp-1 and sapp-2 peptides provided negligible or incomplete sequence information on the site(s) of Man-PO 4 substitution, reflecting the relatively poor ionization of these peptides. However, complete sequence information was obtained for monophosphomannosylated sapp-3, a peptide with a single serine-rich motif and two C-terminal lysine residues (Tables I and  II). Tandem mass spectrometry of the monosubstituted sapp-3 [M ϩ 2H] 2ϩ ion (m/z 995.4) resulted in the loss of mannose and fragmentation along the peptide backbone to produce a series of N-terminal (b-type) and C-terminal (y-type) ions (Fig. 4, A  and B). The y series of fragment ions were particularly informative. As shown in Fig. 4, the y 1-8 fragment ions observed in the mass spectrum corresponded to the unmodified peptide, indicating that none of the C-terminal residues in the peptide, including Ser 15 and Ser 16 , were phosphorylated. In contrast, KSSSSSSSSSSS two series of y 9 -11 ions, differing in mass by 80 Da (corresponding to both unmodified and phosphorylated forms of Ser 8 -10 ), and one series of y 12-16 ions corresponding to the phosphoryl-ated peptide were observed in the mass spectrum (Fig. 4, A and  B). Additional fragment ions corresponding to loss of H 3 PO 4 from the phosphorylated peptides, diagnostic ions for serineand threonine-phosphorylated peptides (37), were also present (y n Ј in Fig. 4, A and B). The progressive decrease in ion intensity of the y 9 -12 ions for the unmodified peptide and the corresponding increase in ion intensity of the y 9 -12 ions for the phosphorylated peptide (Fig. 4C) suggest that all four Ser residues in the central domain of this peptide were partially phosphorylated. Whereas the b series of fragment ions was less complete, the identification of the expected b 2-5 series for the unmodified peptide indicates that neither Ser 1 nor Ser 2 is phosphorylated. These data indicate that sapp-3 is heterogeneously modified with Man␣1-PO 4 on Ser 7-10 and also suggest that none of the Thr residues are modified.
Properties of the pep-MPT Activity-pep-MPT activity in detergent-solubilized microsomes was not affected by inclusion of amphomycin (0.6 mg/ml), a potent inhibitor of dolichol-phosphate mannose synthesis, 2 indicating that the mannose phosphate residue was being directly transferred from GDP-Man (data not shown). Although pep-MPT activity was solubilized from microsomal membranes (Ͼ80%) with a range of non-ionic detergents (see "Experimental Procedures"), Triton X-100 was the most effective at retaining enzyme activity (Fig. 5A) (data not shown). However, enzyme activity was considerably more labile after detergent solubilization, as maximal activity occurred at 16°C, rather than at 32°C, when non-solubilized membranes were assayed (data not shown). The detergentsolubilized pep-MPT activity displayed a pH optimum at 7.4 ( Fig. 5B) and had an apparent K m for GDP-Man of 180 M (Fig.  5C). All activity was abolished when 5 mM EGTA was included in the reaction mixture in the absence of added divalent cations, but was restored to maximal levels with the addition of 10 mM MnCl 2 (i.e. 5 mM free Mn 2ϩ ) (Fig. 5D). MgCl 2 also stimulated activity, but was not as effective as MnCl 2 and did not significantly further stimulate the effect of adding MnCl 2 alone (Fig. 5D). Addition of 5 mM free Ca 2ϩ had no effect on the pep-MPT activity in the absence (Fig. 5D) or presence (data not shown) of MnCl 2 or MgCl 2 . Interestingly, the pep-MPT activity varied markedly in a growth-and stage-dependent manner (Fig. 5A). Rapidly growing (procyclic) promastigotes contained 5-or 10-fold higher activity than either late stationary (metacyclic) promastigotes or freshly isolated lesion amastigotes, respectively.   activity was due to two previously characterized leishmanial MPT activities that are thought to be required for initiation and elongation of the LPG chain biosynthesis (20,22). These oligosaccharide MPT activities add Man-PO 4 to terminal galactose residues on either the LPG anchor precursor or phosphodisaccharide repeat units on the growing LPG phosphoglycan chain, respectively (20,22). A second assay was developed to assess whether synthetic oligosaccharide acceptors for these MPTs competed with the peptide substrates for pep-MPT activity. In this assay, N-terminally 3 H-labeled sapp-1 peptide (at a concentration of 2 ϫ K m ) and unlabeled GDP-Man were added to microsomal membranes in the absence or presence of unlabeled oligosaccharide or peptide substrates. In the absence of competitive substrate, ϳ30% of the 3 H-labeled sapp-1 peptide was modified with Man-PO 4 residues over a 30-min incubation period, as shown by the appearance of additional labeled bands on thin-layer electrophoresis that migrated more rapidly toward the cathode (Fig. 6A). The additional bands corresponded to the sapp-1 peptide with one, two, or three Man␣1-PO 4 residues, as sequential treatment of the peptide mixture with jack bean ␣-mannosidase and alkaline phosphatase collapsed these bands back to a single labeled species that comigrated with unmodified sapp-1 (Fig. 6B). As expected, the addition of 1 mM unlabeled sapp-1 resulted in a 55% decrease in the extent to which the reductively methylated 3 H-labeled sapp-1 peptide was modified (Fig. 6C). In contrast, the oligosaccharide acceptors for the LPG-specific MPTs, L2 (Gal␤1-4Man␣1-PO 4 -(CH 2 ) 8 CHϭCH 2 ) and stachyose (Gal␣1-6Gal␣1-6Glc␣1-2Fru), did not inhibit the pep-MPT activity when present at a 6 -30-fold excess over the concentration of the sapp-1 peptide (Fig. 6C). The concentration of L2 used in these experiments was higher than that used previously (ϳ600 M) to achieve close to maximal saturation of the putative elongating MPT (20). Moreover, in separate experiments, unlabeled sapp-4 was found not to inhibit the phosphomannosylation of L2. As shown in Fig. 7, incubation of detergent-solubilized microsomal membranes with L2 (1 mM) and GDP-[ 3 H]Man generated a labeled product with a slower HPTLC mobility compared with unmodified L2 (compare lanes 1 and 3). This product has previously been shown to correspond to L2 modified with a single terminal Man-1-PO 4 residue (20). Significantly, the modification of L2 by the putative elongating MPT was not affected by the inclusion of sapp-4 (up to 0.5 mM) in the assay (Fig. 7, compare lanes 1 and 2). As the apparent K m of pep-MPT for sapp-4 is 0.05 mM (Table III), these data support the notion that the elongating MPT and pep-MPT are separate enzyme activities. Similar experiments were not performed with substrates of the initiating MPT (stachyose or LPG anchor) because high concentrations of GDP-Man (1 mM) (22) are needed to detect this activity, precluding the use of GDP-[ 3 H]Man.

The pep-MPT Activity Is Distinct from the LPG Phosphoglycan-initiating and -elongating MPTs-Although
Substrate Specificity of pep-MPT-The minimal requirements for pep-MPT activity were investigated using a number of peptide substrates. As shown in Table III, peptides with two Ser/Thr-rich repeats (i.e. sapp-1) were slightly better substrates than peptides with one Ser/Thr-rich repeat (i.e. sapp-4) (K m ϭ 20 and 50 M, respectively). Replacement of the four Ser residues in sapp-4 with Thr (sapp-5) virtually abrogated all substrate potential, confirming the specificity of MPT for Ser residues (Table III). Interestingly, replacement of the two acidic amino acids in the sequences flanking the serines (sapp-6) resulted in a 17-fold increase in the K m of the pep-MPT activity, suggesting that these residues may contribute to MPT substrate recognition (Table III). In the competition assay, unlabeled sapp-4 (1 mM) was slightly less effective than unlabeled sapp-1 (50% versus 55%) at reducing the rate of phosphoglycosylation of 3 H-labeled sapp-1 (Fig. 6C). However, both sapp-5 and sapp-6 and a peptide containing 11 serine residues (Ser11 peptide) inhibited this reaction by Ͻ10%. These data suggest that replacement of Ser residues with Thr or removal of the acidic flanking residues in the repeat sequences abrogates recognition of the sAP-based peptides by pep-MPT.
Subcellular Distribution of pep-MPT-Initial experiments showed that pep-MPT activity was sedimented in a 100,000 ϫ g microsomal pellet and that activity was detected only when intact microsomal membranes were disrupted by sonication or detergent lysis, indicating that pep-MPT is a membrane-associated enzyme and that the catalytic domain has a luminal orientation. To investigate the subcellular location of this enzyme and thus the site at which protein phosphoglycosylation is initiated, L. mexicana microsomal membranes were prepared by nitrogen cavitation and fractionated by equilibrium velocity sucrose gradient centrifugation. Previously character-

FIG. 4. Positive ion collision-induced dissociation mass spectrum of monophosphomannosylated sapp-3.
Shown is the collisioninduced dissociation mass spectrum of the monophosphomannosylated sapp-3 peptide [M ϩ 2H] 2ϩ ion at m/z 995.4. Shown in A are the predicted masses for b-type fragment ions (top level) and y-type fragment ions (bottom level) derived from either the unmodified peptides (y n ) or the phosphorylated peptides (y n * for y n ϩ80 and y n Ј resulting from loss of H 3 PO 4 from the y n * peptide ion). Ions observed in the mass spectrum in B are underlined. Because the mannose residue was readily lost during collision-induced dissociation, only the phosphorylated form of the modified peptide was observed. The relative intensities of the y 9 -12 and y 9 -12 ϩ80 ions (as a percent of the sum of the two ions) are indicated in C.
ized markers for the endoplasmic reticulum (BiP) (29) and cis-Golgi apparatus (Rab1) (28) were separated on this sucrose gradient (Fig. 8), whereas analysis for other organelle-specific markers (i.e. hexose kinase (glycosomes), acid phosphatase (plasma membrane), and HSP60 (mitochondria)) showed that organelle separation had been achieved (data not shown). The pep-MPT activity displayed an overlapping (but not coincident) sedimentation pattern with Rab1 and was reproducibly separated by two fractions from the endoplasmic reticulum marker BiP. These data suggest that the phosphomannosylation of polypeptide acceptors is probably initiated in a post-endoplasmic reticulum compartment.
Differential Phosphoglycosylation of Exogenous Peptides and Endogenous Acceptors in Cell-free Systems-Although the exogenous peptides are utilized by pep-MPT in in vitro assays, the addition of either unlabeled GDP-Man (data not shown) or UDP-Gal (Fig. 9, lane 1) to these assays did not result in further extension of the peptide-linked Man-PO 4 residues with either ␣1-2-linked mannose oligosaccharides or the phosphorylated Gal␤1-4Man repeat units as occurs on endogenous sAP polypeptide acceptors (10). This was not due to disruption of the enzymatic machinery for phosphoglycan biosynthesis, as endogenous membrane-bound LPG acceptors were elaborated with at least one phosphodiester-linked disaccharide unit (Fig.  9, lane 3). This repeat unit probably corresponds to the backbone repeat unit PO 4 -6Gal␤1-4Man, as the dephosphorylated species comigrated with authentic Gal␤1-4Man on the HPTLC (Fig. 9, lane 3), and the synthesis of this oligosaccharide was dependent on the inclusion of UDP-Gal in the assay mixture (data not shown). Interestingly, the Triton X-100-insoluble pellet retained a significant amount of [ 3 H]Man label (ϳ30% of the non-peptide-associated activity), which could be released by mild acid treatment (Fig. 9, lane 2). As Triton X-100 quantitatively extracts both LPG and hydrophilic (non-lipid-linked) phosphoglycans, 3 it is likely that most of the labeled material in the Triton X-100 pellet corresponds to tightly bound polypeptide acceptors. Unlike the endogenous LPG acceptors, the Triton X-100-insoluble acceptors were modified primarily with Man-PO 4 and only to a small extent with the Gal␤1-4Man disaccharide (Fig. 9, lane 2). Thus, this second pool of acceptors may have limited access to the enzymes involved in the synthesis of the disaccharide repeat units. DISCUSSION We have characterized a novel pep-MPT activity from L. mexicana promastigotes that is likely to be responsible for initiating the phosphoglycosylation of a major class of secreted glycoproteins and mucin-like molecules. The phosphoglycans may account for 20 -90% of these secreted molecules and are structurally variable, ranging from single Man-PO 4 residues to exceedingly complex and highly branched phosphoglycans, depending on the nature of the polypeptide backbone and parasite developmental stage (10,14,15). However, all these glycans are thought to be linked to the polypeptide backbone by a common Man␣1-PO 4 -Ser linker sequence (10,14,15). This linkage was indicated by the detection of O-phosphoserine in sAPs and PPGs and the resistance of these residues to alkaline phosphatase unless the masking glycans were removed with ␣-mannosidase or mild acid hydrolysis (10,14,15). However, it has been difficult to confirm the nature of this linkage by mass spectrometry or NMR, as the heavily glycosylated Ser/Thr-rich domains are essentially resistant to proteolysis, preventing the generation of small well defined glycopeptides that are amenable to such analyses. In this context, the synthesis of Man␣1-PO 4 -modified peptides by L. mexicana membranes and the characterization of these peptides by electrospray collisioninduced dissociation mass spectrometry and chemical and enzymatic treatments have provided unequivocal confirmation of this novel linkage. Although the Man␣1-PO 4 -Ser linkage is apparently unique to leishmanial glycoconjugates, it is related to phosphodiester-linked glycans from other lower eukaryotic glycoconjugates. For example, the cysteine proteases of Dictyostelium discoideum are modified with GlcNAc␣1-PO 4 residues 3 M. J. McConville, unpublished data. Specific activity was measured in sonicated membranes with or without solubilization in 1% Triton X-100 (TX-100). Total pep-MPT activity in the detergent extracts was ϳ80% of that in the sonicated membranes in the absence of detergent. B-D, the effect of pH, GDP-Man concentration, and various cations, respectively, on pep-MPT activity. The standard assay buffer was used in all experiments, except for the cation dependence experiment, where buffer A minus MgCl 2 and MnCl 2 was used and cations were supplemented as indicated.
linked to Ser, whereas secreted glycoproteins from Trypanosoma cruzi may be modified with structurally complex oligosaccharides containing a putative Xyl-1-PO 4 -Thr (Ser) linker sequence (11,12). The abundance of this type of modification in several lower eukaryotes suggests that the glycan-PO 4 -Ser (Thr) motif may be more common than previously suspected.
We have provided evidence that L. mexicana pep-MPT transfers Man␣1-PO 4 from GDP-Man to Ser-rich peptide sequences that are found in L. mexicana sAP. The transfer of Man␣1-PO 4 from GDP-Man was supported by the finding that product formation was not inhibited by amphomycin, suggesting that dolichol-phosphate mannose is not utilized as a sugar donor by this enzyme. In addition, there was no evidence for the modification of the peptide in the absence of GDP-Man 3 indicating that this linkage is not being assembled by the sequential action of a serine kinase followed by a novel mannosyltransferase. Based on ESI-MS product analysis and the use of various peptide acceptors, the following conclusions were made concerning the substrate specificity of pep-MPT. First, this enzyme only adds Man␣1-PO 4 to Ser residues; a peptide containing Thr instead of Ser was neither a substrate nor an FIG. 6. Substrate specificity of pep-MPT. The substrate specificity of pep-MPT was tested using a competition assay. Triton X-100solubilized microsomal membranes were incubated with 3 H-labeled sapp-1 and GDP-Man in the absence or presence of the indicated competitive substrate (1 mM), and the labeled products were analyzed by thin-layer electrophoresis. A, when products were analyzed after a 30-min incubation (30Ј) in the absence of competitive substrate, several additional labeled bands (not present in the zero (0Ј) time point) were generated that migrated toward the cathode. B, sequential digestion of the products in the 30-min time point with jack bean ␣-mannosidase (JBAM) and alkaline phosphatase (AP) collapsed these bands back to unmodified sapp-1, indicating that these bands were differentially modified with Man␣1-PO 4 . C, inhibition of 3 H-labeled sapp-1 phosphomannosylation by glycan acceptors of the LPG-specific MPTs (L2 and stachyose) and various unlabeled peptide substrates (sapp-1, sapp-4, sapp-5, sapp-6, and Ser11) is shown using standard incubation conditions. The structures of L2 and stachyose are Gal␤1-4Man␣1-PO 4 -(CH 2 ) 8 CHϭCH 2 and Gal␣1-6Gal␣1-6Glc␤1-2Fru, respectively. The mean values of three experiments are shown. O indicates the sample loading origin on the thin-layer electrophoresis sheets.  inhibitor of the enzyme. Second, the ESI-MS sequence analysis indicated that the Man-PO 4 residues are added to several Ser residues within the Ser-rich repeats. Moreover, the partial characterization of peptides with up to five Man-PO 4 residues shows that at least three of the four contiguous Ser residues in a single Ser-rich repeat sequence can be modified, consistent with the very high degree of substitution (ϳ80%) observed on endogenous sAPs and PPGs (10,14). Third, amino acids flanking the contiguous Ser sequences may be important for pep-MPT recognition and influence the pattern of substitution. For example, sAP peptides lacking the flanking Glu or Asp residues (i.e. sapp-6) were poorly utilized by pep-MPT in direct assays and were poor inhibitors in competition assays. It is possible that these residues may be required to maintain a suitable conformation or the solubility of the peptide, as polyserine peptides become increasingly insoluble above six residues in length.
The pep-MPT characterized here appears to be distinct from the two oligosaccharide-specific MPTs that are involved in adding Man-PO 4 to the LPG anchor precursor (initiating MPT) or preformed phosphoglycan chains on lipid or peptide acceptors (elongating MPT) (20,22). The initiating MPT can be assayed using either dephosphorylated LPG core or the structurally related oligosaccharide, stachyose, as acceptor (22). In contrast, the elongating MPT adds Man␣1-PO 4 to the minimal disaccharide unit Gal␤1-4Man␣1-PO 4 that is present in the synthetic hydrophobic oligosaccharide acceptor L2 (20). Neither stachyose nor L2 inhibited the pep-MPT activity when present at 6 -30-fold excess over the peptide acceptors (Fig. 6C). Moreover, unlabeled sapp-4 peptide did not inhibit the phosphomannosylation of L2 when present at concentrations that inhibit the phosphomannosylation of the 3 H-labeled sapp-1 peptide (Fig. 7). pep-MPT could also be distinguished from the other MPTs by its solubility and stability in various detergents (20,22). On the other hand, all three MPTs appear to utilize GDP-Man as the sugar donor and have a similar pH optimum and requirement for divalent cations. Although the substrate specificities of these MPTs appear to be distinct, it is notable that both pep-MPT and elongating MPTs are maximally active against substrates that contain negatively charged groups near the aglycon acceptor (i.e. acidic amino acids or phosphate, respectively) (this study and Ref. 20), suggesting that these activities may have arisen from a common progenitor enzyme.
Marked differences were observed in the levels of pep-MPT activity during promastigote growth, with 5-or 10-fold higher levels of activity being observed in rapidly dividing procyclic promastigotes compared with late stationary phase promastigotes and lesion-derived amastigotes, respectively. There is no evidence that the low pep-MPT activity of the metacyclic promastigotes or the intracellular amastigotes is associated with a reduced level of phosphoglycosylation of secreted polypeptides. Indeed, aPPGs are much more extensively modified with phosphoglycan chains than promastigote sAPs or promastigote PPGs (15). It is thus possible that the low pep-MPT activity in non-dividing promastigotes and the intracellular amastigotes may reflect a low level of secretory activity in these developmental stages or the presence of other pep-MPT activities that preferentially recognize distinct peptide sequences in aPPGs (14,15).
Subcellular fractionation studies suggested that pep-MPT was localized in a post-endoplasmic reticulum compartment and that it had an overlapping (but not coincident) distribution with the Golgi marker Rab1. Rab1 has been localized primarily to cis-Golgi cisternae in L. major (28), but may also be localized to endoplasmic reticulum-Golgi transport vesicles, where it is thought to be involved in regulating the organization of vesicle proteins (38). The broader distribution of the Rab1 marker may reflect the partial distribution of this protein in lighter transport vesicles from which pep-MPT is excluded. Localization of pep-MPT to the Golgi apparatus is consistent with the recent demonstration that the leishmanial Golgi apparatus contains a GDP-Man transporter that is required for phosphoglycan biosynthesis (24). Similarly, the initiating and elongating MPTs After incubation, membranes were sequentially extracted with water and Triton X-100 to remove exogenous peptides and Triton X-100-soluble LPG/phosphoglycans/glycoproteins, respectively. The 3 H-labeled glycopeptides and LPG were further purified as described under "Experimental Procedures." This procedure generated three [ 3 H]Man-labeled fractions containing exogenous peptide (lane 1), Triton X-100-insoluble glycoproteins (lane 2), and LPG (lane 3). Each of these fractions was subjected to mild acid hydrolysis (to selectively cleave phosphodiesterlinked glycans) and then dephosphorylated with alkaline phosphatase, and the neutral glycans were analyzed by HPTLC. involved in LPG biosynthesis are also thought to be localized to the Golgi apparatus based on the finding that one of the enzymes involved in LPG anchor biosynthesis has been localized to the Golgi apparatus by electron microscopy (23). In this respect, it is of interest that Man-PO 4 residues on either the peptide or a pool of endogenous polypeptide acceptors were negligibly or poorly elaborated with phosphoglycan repeat units. In contrast, Man-PO 4 residues on the endogenous LPG acceptors were quantitatively modified with at least one repeat unit. Although it is possible that the Man-PO 4 -modified peptide is a poor substrate for these elongating enzymes, the low rate of elongation of Man-PO 4 on the endogenous polypeptide pool suggests that pep-MPT and elongating enzymes (including the elongating MPT) are localized to functionally distinct subcompartments of the Golgi apparatus. Compartmentalization of these enzymes may provide a mechanism for generating phosphoglycan structures with markedly different chain lengths on different classes of glycosylphosphatidylinositol or polypeptide acceptors (see Fig. 1).
In summary, we have confirmed the nature of the peptide linkage of the unusual phosphoglycans that are added to the major secretory proteins of L. mexicana and identified a novel pep-MPT activity. Similar types of phosphoglycosylation occur in all developmental stages of the parasite, including the amastigotes that perpetuate disease in the mammalian host and that have been shown to be functionally important. Consequently, it is likely that pep-MPT and possibly other MPTs are potential targets for new anti-leishmanial drugs.