Molecular Cloning and Characterization of a Novel Repeat-containing Leishmania major Gene,ppg1, That Encodes a Membrane-associated Form of Proteophosphoglycan with a Putative Glycosylphosphatidylinositol Anchor*

Leishmania parasites secrete a variety of proteins that are modified by phosphoglycan chains structurally similar to those of the cell surface glycolipid lipophosphoglycan. These proteins are collectively called proteophosphoglycans. We report here the cloning and sequencing of a novel Leishmania major proteophosphoglycan gene,ppg1. It encodes a large polypeptide of approximately 2300 amino acids. The N-terminal domain of approximately 70 kDa exhibits 11 imperfect amino acid repeats that show some homology to promastigote surface glycoproteins of the psa2/gp46 complex. The large central domain apparently consists exclusively of approximately 100 repetitive peptides of the sequence APSASSSSA(P/S)SSSSS(±S). Gene fusion experiments demonstrate that these peptide repeats are the targets of phosphoglycosylation in Leishmania and that they form extended filamentous structures reminiscent of mammalian mucins. The C-terminal domain contains a functional glycosylphosphatidylinositol anchor addition signal sequence, which confers cell surface localization to a normally secreted Leishmania acid phosphatase, when fused to its C terminus. Antibody binding studies show that the ppg1 gene product is phosphoglycosylated by phosphoglycan repeats and cap oligosaccharides. In contrast to previously characterized proteophosphoglycans, the ppg1gene product is predominantly membrane-associated and it is expressed on the promastigote cell surface. Therefore this membrane-bound proteophosphoglycan may be important for direct host-parasite interactions.

displayed on their surface (2) or secreted (3). These glycoconjugates, which are thought to have crucial functions for parasite survival, development, and virulence in the sandfly vector and the mammalian macrophage, include a family of phosphoglycan-modified molecules, which comprises lipid-linked, protein-linked, and unlinked forms (4 -6). Lipophosphoglycan (LPG) 1 is the best studied representative of this family: it consists of 1-alkyl-2-lyso-phosphatidylinositol that is linked via a diphosphoheptasaccharide core to a linear polymer of up to 40 phosphodiester-linked disaccharides that are modified by species-, strain-, and stage-specific side chains. The polymer chain terminates with variable neutral cap oligosaccharides (2). LPG is expressed in high copy number on the surface of promastigotes (7)(8)(9)(10), but its synthesis is strongly down-regulated in amastigotes (11)(12)(13). The lipid anchor and in particular the phosphoglycan chain of LPG exhibit a variety of biological activities, which may be important for Leishmania survival in the vector sandfly and the mammalian host. These activities include receptor functions, protection against complement and hydrolases, and the modulation of macrophage killing mechanisms (14 -22).
More recently, it has been found that the biologically active phosphoglycan structures of LPG are also present on several highly glycosylated Leishmania proteins, collectively termed proteophosphoglycans (PPGs, Refs. 5 and 6). This group of molecules includes secreted acid phosphatase synthesized by promastigotes of all Leishmania species except Leishmania major (23,24), non-filamentous proteophosphoglycan of Leishmania mexicana amastigotes (25), and a filamentous proteophosphoglycan (fPPG) secreted by promastigotes of all species investigated so far (5,26). Capped phosphoglycan chains are linked to the polypeptide backbone of these proteins via phosphodiester linkages to serine, an unusual type of protein modification called phosphoglycosylation (27)(28)(29). Among the PPGs, fPPG exhibits some unique features; the contour length of filaments observed by electron microscopy in purified preparations may exceed 6 m. The fPPG protein backbone is largely formed by Ser, Ala, and Pro (together Ͼ87 mol%), and more than 40 mol% of its amino acids are phosphoglycosylated Ser residues. This extensive glycosylation leads to high protease resistance of fPPG and may also be responsible for its filamentous ultrastructure that is reminiscent of mammalian mucins (28). In a recent study, it has been shown that fPPG, initially observed in promastigote cultures (26), is also abundantly present in Leishmania-infected sandflies, where it forms a gel of three-dimensional filamentous meshworks that immobilizes the parasites and occludes the lumen of the insect digestive tract (30). This fPPG/parasite plug may be important for efficient transmission of metacyclic promastigotes to the mammalian host during the bite of the vector insect (see Refs. 30 and 31 and references therein).
In this paper we describe the identification of a L. major gene family whose members share common repetitive sequences encoding exclusively Ser, Ala, and Pro in ratios very similar to those found in fPPG. One of the genes (ppg1) was cloned and sequenced, and its expression and functional elements were analyzed. Our results suggest that ppg1 encodes a novel membrane-bound and possibly glycosylphosphatidylinositol (GPI)anchored PPG that is present on the surface of L. major promastigotes.

EXPERIMENTAL PROCEDURES
Parasites-Promastigotes of L. major LRC-L137, clone 121, were grown in vitro in semidefined medium 79 (32) supplemented with 5% heat-inactivated fetal bovine serum. Cells were harvested at densities of 5 ϫ 10 7 to 10 8 /ml. DNA and RNA Techniques-Agarose gel electrophoresis, isolation of -phage DNA, purification of genomic Leishmania DNA and total RNA, Southern and Northern blotting, -phage and bacterial colony lifts, antibody screening of a gt11 expression gene library, restriction enzyme digests, DNA ligations, and transformation of Escherichia coli were performed according to standard methods (33,34). Plasmid DNA and DNA fragments from agarose gels were isolated using commercial kits according to the manufacturers' instructions (Qiagen, Hilden, Germany; Gene-Clean, Bio101, Vista, CA). Digoxygenin (DIG) labeling of DNA or RNA fragments was performed by random priming and by the polymerase chain reaction (PCR) or by in vitro transcription, respectively (Roche Molecular Biochemicals, Mannheim, Germany). DIG-labeled DNA or RNA on blots was detected using anti-DIG-F ab fragments coupled to alkaline phosphatase (Roche) and CDP-Star TM (Tropix, Bedford/MA) as chemiluminescent substrate according to the manufacturers instructions. PCR was performed using the Expand TM high fidelity PCR system (Roche). Reverse transcriptase PCR (RT-PCR) was performed from total L. major promastigote RNA using the Masteramp RT-PCR kit (Epicentre Technologies, Madison, WI).
Cloning and Sequencing of ppg1-A gt11 expression library constructed from sheared genomic L. major DNA (36) was propagated in E. coli Y1090 and plated on agar dishes. Nitrocellulose filter lifts of gt11 phage plaques were used for antibody screening for the PPG genes using a rabbit antiserum against HF-dephosphorylated secreted L. major fPPG (28) that had been preadsorbed four times with E. coli lysate (33). L. major DNA inserts were excised from DNA of positive gt11 phages and cloned into NcoI/NdeI-cut pGEM5z ϩ (37). Isolated inserts were labeled with DIG and used for Southern blot analyses or hybridization on bacterial filter lifts (see below). For the construction of dedicated genomic plasmid libraries L. major DNA was digested with XhoI or SphI and separated on 0.7% agarose gels. DNA fragments that corresponded to positive bands on Southern blots using the DIG-labeled -phage inserts as probes, were cloned into XhoI-cut pBSK ϩ or SphI-cut pGEM5z ϩ , respectively. Filter lifts from transformed E. coli XL1-Blue colonies were hybridized with DIG-labeled -phage inserts and the clones pBSK ϩ ppg1/XhoI and pGEM5z ϩ ppg1/SphI containing overlapping inserts were isolated (compare Fig. 3A). To obtain the entire ppg1 gene on one plasmid the XbaI/XhoI fragment of pGEM5z ϩ ppg1/SphI (Fig. 3A) was ligated into XbaI/XhoI-cut pBSK ϩ . This construct was cut with XhoI and ligated with the XhoI fragment from pBSK ϩ ppg1/XhoI (Fig. 3A), yielding pBSK ϩ ppg1/ORF. The correct orientation of the insert was confirmed by restriction digests and sequencing. Sequence analysis of the clones was performed by the dideoxy chain termination method (38). Both DNA strands were sequenced in nonrepetitive regions. For sequencing of the large repetitive region, the central 4.75-kb Sau3AI fragment was isolated from pBSK ϩ ppg1/XhoI and ligated into BamHI-cut pBSK ϩ . The resulting construct (pBSK ϩ ppg1/Sau3AI) was sequenced from both ends. Internal regions of the repeats were characterized by constructing exonuclease III/mung bean nuclease nested deletion clones (New England Biolabs, Schwalbach, Germany) of pBSK ϩ ppg1/Sau3AI followed by sequencing. RT-PCR of the ppg1 3Јregion used oligonucleotides flanking this sequence (5Ј-GACTCTAGA-GCTTCAAGTCGACATCAC; 3Ј-CATGGATCCCGATAGACTCTATCT-TC) with an initial reverse transcription step of 60°C for 20 min, followed by 1 min of denaturation at 95°C, 1 min of annealing at 55°C, and 1 min of extension at 72°C for 35 cycles. Determination of the ppg1 splice site involved an initial RT-PCR reaction using an oligonucleotide complementary to the L. major miniexon consensus sequence (35) and the oligonucleotide TGCAGTCGACTACCTTCTCTATCAGA located within ppg1 coding region. Conditions were as described above except that the extension was done for 25 cycles and was followed by a second round of nested PCR using the same miniexon primer and an oligonucleotide situated immediately downstream of the ppg1 start codon (GC-GTCTAGACATGATCCTAAAACGTTTCTG). The product was cloned into pBSKϩ and its primary structure determined by DNA sequencing.
Expression Constructs-The XbaI/ScaI fragment from pBSK ϩ ppg1/ ORF that includes part of the ampicillin gene was ligated with the XbaI/ScaI-cut Leishmania expression vector pX (39) yielding pXppg1/ ORF. For the generation of fusion constructs of the repetitive ppg1 domains to the 3Ј-region of a L. mexicana membrane-bound acid phosphatase (lmmbap; Ref. 40) whose transmembrane anchor sequence had been removed (modified lmmbap ϭ modMAP; Ref. 41), the expression construct pXmodMAP (41) was cut with BglII and ligated in frame with the large Sau3AI fragment (4.75 kb, compare Fig. 3B) of ppg1 yielding pXmodMAP-repeat. Correct orientation was confirmed by restriction digests with NcoI. The C-terminal region of ppg1 was amplified using the primers TCTAGATCTCCCAGGATTCTTCCGTGGATG and TC-CGGATCCCATGCCTGCATGCCTCAAGT to create a BglII and a BamHI site for in frame ligation with BglII-cut pXmodMAP yielding pXmodMAP-GPI. The correct orientation was confirmed by restriction digests and sequencing. L. major promastigotes were transformed by electroporation as described previously (42) and grown in semidefined medium 79 plus 5% heat-inactivated fetal bovine serum that was supplemented with 20 -100 g/ml G418 to select for plasmid-harboring parasites.
Production of Rabbit Antisera against Recombinant PPG1 Protein-A 1026-bp SalI/HindIII DNA fragment of the ppg1 5Ј-region (Ndomain) and a 303-bp SphI/HpaI fragment of the ppg1 3Ј-region (Cdomain) (Fig. 3A) were ligated in frame into SalI/HindIII-cut and SphI/SmaI-cut pQE31, respectively. Hexa-His-tagged recombinant proteins expressed in E. coli M15 were purified using Ni-NTA-agarose columns according to the manufacturer's instructions (Qiagen). Rabbits were immunized with 200 g of purified recombinant protein by standard protocols (28). For immunoblotting studies, anti-N-domain antibodies as well as anti-C-domain antibodies were affinity-purified on recombinant protein that had been electrotransfered to PVDF membranes after SDS-PAGE as described previously (40).
Immunofluorescence and Electron Microscopy-Late log phase L. major promastigotes were washed with PBS, fixed with an equal volume of 4% para-formaldehyde in PBS, and incubated on poly-L-lysinecoated glass microscope slides for 30 min at 25°C (26) followed by two washings with PBS. After incubation with 2% BSA in PBS for 30 min at 25°C, dilutions of mouse mAbs (1:10 for mAb LT6, 1:500 -1:2000 for mAb AP4) or rabbit antisera/preimmune sera (1:100 -1:500) in PBS plus 2% BSA were added to the fixed promastigotes for 30 -60 min at 25°C. After three washings with PBS, the cells were incubated with either goat anti-mouse IgG or goat anti-rabbit IgG coupled to Cy3 (Dianova) diluted 1:500 in 2% BSA in PBS for 30 -60 min at 25°C. The cells were then washed three times with PBS, incubated with 1 g/ml 4,6-diamidino-2-phenylindole in PBS for 10 min, and washed again three times with PBS, followed by embedding in Mowiol and inspection by fluorescence microscopy.
Glycerol spraying and rotary metal shadowing of purified modMAP and modMAP-repeat protein with and without addition of mAb AP4 for electron microscopic inspection was performed as described earlier (28).

Isolation of Gene Fragments from L. major Genomic DNA
Encoding for Ser-, Ala-, and Pro-rich Peptide Repeats of PPG-Multiple attempts to generate peptide sequence information from L. major fPPG were unsuccessful (28), which precluded a gene cloning strategy using degenerate oligonucleotides. Therefore, a gt11 expression library constructed from sheared genomic L. major DNA was screened with a rabbit antiserum directed against the 40% HF-deglycosylated protein backbone of fPPG (anti-HFP serum; Ref. 28). Three positive gt11 clones were identified, subcloned into plasmid vectors, mapped by restriction enzyme digests and partially sequenced. These clones appeared to consist exclusively of repetitive DNA sequence potentially encoding Ser-, Ala-, and Pro-rich peptide repeats (Fig. 1). The average amino acid composition of these predicted peptide repeats was 67.7 mol% Ser, 19.4 mol% Ala and 12.9 mol% Pro, which was highly reminiscent of the amino acid composition of the glycosylated protease-resistant fPPG domains analyzed in an earlier study (59 mol% Ser, 23.1 mol% Ala, 15.3 mol% Pro, 2.6 mol% other amino acids; Ref. 28). This result strongly suggested that the identified gt11 clones contained fragments of DNA related to or identical with the fPPG gene(s). Southern blot analysis using the repeat-containing gt11 phage inserts as a probe revealed multiple bands of very large (Ͼ23 kb), intermediate (20 -5 kb), and small size (3-0.8 kb) with most restriction enzymes ( Fig. 2A), indicative of several fPPG genes or genes encoding for fPPG-related parasite proteins.
Isolation of ppg1 and Characterization of Its Nonrepetitive Domains-For the isolation of one of the fPPG gene candidates we focused on the fragments of intermediate size that were positive on Southern blots with the gt11 repeat probe ( Fig.  2A). To this end, two dedicated plasmid libraries of size-fractionated L. major genomic DNA containing a 5.9-kb XhoI fragment ( Fig. 2A, lane 1) and a 20-kb SphI fragment (data not shown), respectively, were constructed and screened with the gt11 repeat probe. Overlapping XhoI-and SphI clones apparently containing a complete open reading frame were identified and a detailed restriction map was constructed (Fig. 3, A and  B). One peculiar feature of the gene-containing Eco47 III/XhoI L. major DNA fragment (Fig. 3B) was the presence of a central Sau3AI/NcoI fragment (Ͼ4.6 kb), which was the smallest mapped fragment to react with the repeat DNA probe on Southern blots (data not shown; compare Figs. 2A and 3A, and see below). The nonrepetitive DNA in the 5Ј-region from this central fragment starting from the Eco47 III site and in the 3Ј-region from the central fragment ending at a XhoI site was sequenced completely on both strands (Figs. 3B and 4). An open reading frame was identified upstream and just before of the central repeats that encompassed 1926 bp. The splice addition site of its mRNA was determined after sequencing a fragment obtained by reverse transcriptase PCR using an oligonucleotide derived from the Leishmania splice leader sequence and oligonucleotides derived from the region near the putative start codon (Fig. 4). The open reading frame upstream of the central repeats encoded a potential signal sequence for import into the endoplasmic reticulum (Ref. 46; see also Figs. 4 and 5A) and three potential N-glycosylation sites. Remarkably, the regions between the amino acids 153 and 600 contained 11 imperfect 24 amino acid repeats that showed some homology to members of the psa2/gp46 gene family of Leishmania (Fig. 5B). The rest of the DNA sequence upstream of the repeats and the derived amino acid sequence showed no significant homology to any sequences in the data base. The open reading frame in the region 3Ј of the central repeats extended for 456 bp and showed also no significant homology to known DNA or protein sequences. Importantly, however, the derived C-terminal peptide sequence was highly reminiscent of an addition signal for GPI membrane anchor addition (Figs. 4 and 5, A and C).
Characterization of the Central Repetitive ppg1 Domain-The central Sau3AI/NcoI DNA fragment of ppg1 (Fig. 3B) was most unusual in being devoid of any palindromic hexanucleotide restriction sites (data not shown) and most palindromic tetranucleotide restriction sites, except for those of the enzymes AluI, HinPII (Fig. 6A), and BstUI (data not shown). By sequencing both ends of the DNA fragment and by sequencing of exonuclease III deletion clones, more than 90% (Ͼ4.2 kb) of the very GC-rich (Ͼ73%) primary structure of the central repeat domain could be determined, but the perfection of the repeat sequence precluded the accurate identification of overlaps between individual sequencing runs. The data suggest, however, that the central repeat domain consists exclusively of an estimated 100 closely related 48-and 45-bp repeats, like those identified in the gt11 expression clones of the initial screen ( Figs. 1 and 4). Digestion of the central repeat region (Fig. 3B) by the restriction enzyme AluI that cuts once per repeat leads to the exclusive appearance of fragments whose size is consistent with this assumption (Fig. 5B). The few variations that occur in the repeats mainly involve the third position of some codons and do not lead to amino acid changes, except for an occasional transition at a first codon position (C to T) leading to a change of Pro to Ser (Fig. 1).
Genomic Structure and Transcription of ppg1 and Related Genes-Southern blots of L. major genomic DNA digested with a selection of restriction enzymes were hybridized with a DIGlabeled probe of the ppg1 sequence downstream (3Ј) of the central repeats (NcoI/NcoI-DIG, compare Fig. 3B). These blots yielded a band pattern consistent with the presence of a single gene copy (Figs. 2C and 3, A and B). However, using a DIGlabeled probe of the domain upstream (5Ј) from the central repeats, only about 50% of the observed fragments could be identified as being derived from ppg1 (Fig. 2B, marked with  asterisks). In this Southern blot, the very large fragments (Ͼ23 kb), except for those derived from XhoI and from NcoI digests, are most likely derived from related but different genes, as are some smaller fragments in AatII, PstI, and XhoI digests (Figs. 2B and 3A). With a DIG-labeled repeat probe, only a minority of the fragments could be identified as derived from ppg1 ( Fig.  2A), while most fragments, in particular those of very large size (Ͼ23 kb, except for NcoI digests), as well as those of smaller size (1-3 kb) must also be derived from genes related but different from ppg1 ( Figs. 2A and Fig. 3, A and B). Most remarkably, on Southern blots of L. major DNA digested with palindromic tetranucleotide-cutting restriction enzymes probed with DIG-labeled ppg1 repeat DNA, the very large bands (Ͼ23 kb) were still present, except for those enzymes cutting frequently in the repeats, which only gave rise to very small fragments (Fig. 6C). These results suggest that in L. major a multigene family is present whose members contain ppg1-type repeat sequences. These repeats may, in some of the genes, be longer than 23 kb. However, ppg1 is a distinct member of this multigene family in possessing a unique 3Ј-region (Fig. 2C). To assess the expression of the ppg genes on the RNA level, a DIG-labeled RNA probe of the ppg1 repeats was used on a Northern blot of RNA from L. major promastigotes and mouse lesion-derived amastigotes (Fig. 7, A and B). This probe does not distinguish between different members of the ppg gene family. A broad smear showing maximal intensity in the size region larger than the 9.5-kb marker was observed in both samples, indicating the very large size of the ppg gene transcripts, which is in agreement with the Southern blot data ( Fig.  7B; see also Fig. 2). A distinct transcript of the size predicted for the ppg1 gene (6.85 kb) was not observed, which may be due to the presence of a long 3Ј-untranslated region. To investigate specifically the ppg1 gene expression, RT-PCR was performed on L. major promastigote and amastigote total RNA using oligonucleotide primers derived from the specific 3Ј-region of the ppg1 gene. Fragments of the predicted size and sequence were amplified in both RNA preparations indicating that ppg1 gene transcripts are present in both L. major promastigotes and amastigotes (Fig. 7C).
Functional Analysis of the Putative GPI Membrane Anchor Addition Sequence-To determine whether the putative GPI anchor addition sequence of ppg1 was functional, a DNA fragment encompassing this region was ligated in frame to the 3Ј-region of a modified L. mexicana membrane-bound acid phosphatase gene, which lacks the sequence encoding the  (Fig. 8A). L. major promastigotes transfected with a pX control vector contained, like untransfected cells, only a small amount of acid phosphatase and did not secrete any enzyme activity into the medium (Fig. 8B). Parasites transfected with pXmodMAP produced large amounts of this enzyme activity (Ͼ600 milliunits/ml culture, 5 ϫ 10 7 cells/ml) and secreted most of it (Ͼ97%) into the culture medium (Fig. 8B). L. major promastigotes transfected with pXmodMAP-GPI also synthesized considerable amounts of enzyme (Ͼ250 milliunits/ml culture, 5 ϫ 10 7 cells/ml), which was, however, largely associated with the cells (ϳ80%) (Fig. 8B). In immunofluorescence experiments using a mAb directed against the modMAP tag (41, 45), these pXmodMAP-GPI transfected parasites showed a strong surface labeling (Fig. 9A), while wild type or pXmodMAP-transfected promastigotes did not show any signal on their cell surface (data not shown). All of the cell-associated modMAP-GPI protein was found in the membrane-containing pellet after ultracentrifugation of disrupted promastigotes (data not shown) and Ͼ90% of the activity partitioned into the detergent-rich phase in a Triton X-114 phase separation experiment (Fig. 8C). After incubation with GPIspecific phospholipase C from Trypanosoma brucei, however, the majority (Ͼ90%) of the acid phosphatase activity partitioned into the upper detergent-poor phase, which was a strong indication for the presence of a GPI anchor on modMAP-GPI (Fig. 8C).
Functional Analysis of the Central ppg1 Repeat Sequence-The central 4.75-kb Sau3AI fragment of ppg1 (Fig. 3B), which consists largely of repetitive sequence, was ligated in frame to modMAP gene and cloned into pX yielding pXmodMAP-repeat (Fig. 8A). L. major promastigotes transfected with this gene fusion construct synthesized acid phosphatase activity (Ͼ150 milliunits/ml culture, 5 ϫ 10 7 cells/ml). Approximately 70% of the enzyme was detected in the culture supernatant (Fig. 8B). Immunofluorescence microscopy of transfected promastigotes using mAb AP4 revealed a strong flagellar pocket labeling (Fig.  9, B and C) indicating secretion of the enzyme via this parasite organelle. It appeared likely that the Ser/Ala/Pro-rich peptide repeats encoded by the ppg1 Sau3AI DNA fragment were target peptide sequences for the extensive phosphoglycosylation observed on fPPG (28). To test this hypothesis, similar amounts of the modMAP and modMAP-repeat proteins from promastigote culture supernatants were captured by the mAb AP4 onto microtiter plates and then analyzed for their binding of antiphosphoglycan repeat (WIC79.3 and WIC108.3) and cap antibodies (L7.25) in an ELISA. While modMAP showed only marginal binding of L7.25 and no binding of the mAbs WIC79.3 and WIC108.3, all three antibodies strongly bound to modMAPrepeat, which demonstrated its modification by capped phosphoglycan chains (Fig. 8D). The enzymes modMAP and mod-MAP-repeat were purified by affinity chromatography on a column of mAb AP4 coupled to Sepharose CL-4B and inspected by electron microscopy. While modMAP consisted only of small globular particles of approximately 4 nm in diameter (Fig. 10a), modMAP-repeat also exhibited these globular particles, but they were attached to filamentous structures with a contour length of up to 250 nm (Fig. 10b). The identity of these molecules was confirmed by labeling with mAb AP4, which bound to the globular domain of modMAP-repeat (Fig. 10c).
Membrane Association and Phosphoglycosylation of ppg1-To investigate the expression of ppg1 in L. major promastigotes at the protein level, an antiserum was raised against a recombinant peptide from the C-terminal non-repetitive domain of ppg1 that was not shared with other members of the PPG multigene family (anti-C-domain serum, see Figs. 2C and 3B). Using affinity-purified antibodies isolated from this serum, SDS-PAGE/immunoblots of total cell lysates from PBS-washed L. major promastigotes showed a strong signal in the region corresponding to the stacking gel and the top of the separating gel (Fig. 11, lane 1), a position typical for PPGs (25,28,29). This cell-bound PPG1 protein remained associated with the membrane pellet after disruption of the cells and ultracentrifugation (Fig. 11A, lanes 2 and 3). Further washing of the membranes with 10 mM EDTA-containing solutions and 10 mM EDTA plus 250 mM NaCl-containing solutions solubilized some of the compound (Fig. 11A, lanes 4 and 6, respectively), but most PPG1 protein remained membrane-bound (Fig. 11A, lanes  5 and 7). Importantly, the anti-C-domain serum did not recognize the secreted fPPG from L. major supernatant (Fig. 11A,  lane 8). Therefore, we propose the term mPPG for this novel membrane-bound PPG1 from L. major. Using the anti-phosphoglycan mAb LT6, which does recognize secreted fPPG but not cell surface LPG of L. major procyclic promastigotes (Fig.  11B, lane 8; compare also Refs. 28 and 30), the typical PPG signal pattern is also obtained on SDS-PAGE/immunoblots of L. major lysates (Fig. 11B, lanes 1-7). A significant proportion of this PPG remains associated with the membrane pellet throughout the different washing steps (Fig. 11B, lanes 3, 5,  and 7), which indicates that it may be identical to mPPG. In immunofluorescence microscopy of fixed L. major promastigotes, most cells showed a strong signal on their cell surface using anti-C-domain serum (Fig. 9E) compared with preimmune serum (Fig. 9D), indicating that at least part of membrane-bound mPPG is localized on the parasite cell surface. In labelings with the mAb LT6, most cells showed the expected surface fluorescence, which was, however, highly heterogeneous in its intensity (Fig. 9F). Cell lysates of L. major promastigotes that were transfected with a pX expression vector carrying a copy of ppg1 (pXppg1-ORF) showed a much stronger signal in the stacking gel region on SDS-PAGE/immunoblots using anti-C-domain antibodies than wild type cells (Fig. 11,  compare A and C, lanes 1), which confirmed that ppg1 encodes   FIG. 6. Analysis of the central repeat domain of ppg1 by agarose gel electrophoresis and Southern blot. A, 1.8% agarose gel electrophoresis of the repeat-containing 4.75-kb Sau3AI ppg1 DNA fragment digested with various restriction enzymes cutting palindromic tetranucleotide restriction sites. 1-3, DNA markers, whose size in bp is indicated: 1, -DNA/HindIII; 2, pEMBL/TaqI; 3, pBR322/MspI. 4 -12, Sau3AI ppg1 DNA fragment digested with the following restriction enzymes: 4, RsaI; 5, NlaIII; 6, MseI, 7, HpaII; 8, HinPII; 9, HaeIII; 10, BfaI; 11, AluI; 12, undigested fragment. B, 2.5% agarose gel electrophoresis of the repeat-containing 4.75-kb Sau3AI ppg1 DNA fragment. 1, undigested; 2, AluI-digested; 3, DNA marker fragments (VIII, Roche Molecular Biochemicals) whose size in bp is indicated. C, Southern blot of L. major genomic DNA incubated with PstI and restriction enzymes cutting palindromic tetranucleotide sites separated by 0.7% agarose gel electrophoresis. The blot was probed with DIG-labeled SacI/NcoI fragment (Fig. 3B) 7. Analysis of ppg1 expression by Northern blotting and RT-PCR. a, 2.5 g of L. major promastigote (P) and amastigote (A) total RNA was separated by gel electrophoresis using a 1% formaldehyde-containing agarose gel and stained with ethidium bromide. b, the agarose gel from a was blotted onto a nylon membrane and probed with a DIG-labeled ppg1 RNA repeat fragment. c, RT-PCR was performed using L. major promastigote and amastigote total RNA and primers specific for the unique ppg1 3Ј-region region. The products were separated by 1.2% agarose gel electrophoresis and visualized with ethidium bromide. The size of DNA marker fragments is indicated in bp.
mPPG. Most of the mPPG in the overexpressing L. major strain remained also associated with the membrane fraction after repeated washing and ultracentrifugation (Fig. 11C, lanes 3, 5,  and 7), but, more pronounced than in wild type cells, a signif-icant proportion could be extracted by EDTA-containing buffer (Fig. 11C, lanes 4 and 6). In addition to a signal in the stacking gel region, some bands reactive with anti-C-domain antibodies were visible that migrated in SDS-PAGE with an apparent molecular mass lower than 30 kDa (Fig. 11C) and most likely represented partial proteolysis of the mPPG C-terminal part.
The phosphoglycosylation of mPPG with LPG/fPPG-type glycans was demonstrated by detection of mPPG immunoprecipi- tated from promastigote detergent lysates with anti-C-domain serum and antisera raised against a recombinant polypeptide from the N-terminal region of mPPG (anti-N-domain serum), by the anti-phosphoglycan cap mAb L7.25 and the anti-phosphoglycan repeat mAbs WIC79.3 and LT6. The region on the blots corresponding to the stacking gel and the upper separating gel showed for both immunoprecipitates a signal with all three mAbs, while no specific signal was present with a preimmune serum control (Fig. 12). DISCUSSION Promastigotes of most Leishmania species secrete a network-forming filamentous fPPG that accumulates in the center of cell aggregates in culture (26). In infected sandflies, fPPG is present in the lumen of the digestive tract, where it may play a crucial role for parasite transmission from the insect to the mammalian host (30). In the present study we have attempted to clone the fPPG gene by antibody screening of a L. major genomic expression library. This approach led to the identification of repetitive DNA fragments that encode exclusively for Ser, Ala, and Pro in ratios in line with the known composition of fPPG. However, Southern blot analysis indicated that such repeats are present in several L. major genes. One of these genes was cloned and characterized in this study; it encodes a large protein of approximately 2300 amino acids with three distinct domains. The N-terminal polypeptide chain of 642 amino acids contains a putative signal sequence for import into the endoplasmic reticulum with 5 positive charges in its nregion (46), which is typical for signal peptides of Leishmania proteins (47). It also exhibits 11 imperfect 24 amino acid repeats located between amino acid 153 and amino acid 600. These peptide repeats show some homology to repeats of the psa2/gp46 gene family of GPI-anchored promastigote cell surface proteins (48,49). The N-terminal domain continues into a large central domain of approximately 1500 amino acids, which is encoded by approximately 4.5-kb GC-rich repetitive DNA. The perfection of the DNA repeats precluded the determination of overlaps and the generation of a contig during DNA sequencing. However, the sequence information of more than 90% of the repeats and restriction enzyme digests suggest that the central domain consists exclusively of approximately 100 closely related 45-and 48-bp repeats encoding exclusively for Ser, Ala, and Pro ( Figs. 1 and 4). Remarkably, less than half of the codons in these repeats show variations, and the variation is confined to transitions of the third position (G to A; T to C) and do not cause amino acid changes, except for an occasional first position transition (C to T) leading to a change of Pro to Ser (Fig. 1). Southern blots of genomic DNA purified from one L. major strain over a period of 4 years show no evidence of rearrangements in the central ppg1 domain. How and why L. major maintains this long stretch of almost perfect repeats is currently unknown.
The presumed role of the derived Ser-, Ala-, and Pro-containing peptide repeats (Fig. 1) as attachment sites for phosphoglycosylation was confirmed by fusing the central repetitive ppg1 domain to the gene encoding a modified acid phosphatase (modMAP) of L. mexicana (40,41) resulting in the chimeric construct modMAP-repeat. Expression of modMAP-repeat in L. major yields a phosphoglycosylated product and antibody binding data suggest that the modification pattern resembles that of fPPG (28). In addition our study shows that the phosphoglycosylated APSASSSSA(P/S)SSSSS(ϮS) repeats give rise to filamentous structures that are similar to the extended phosphoglycosylated domains of SAP2 (50) and fPPG (28) as well as the highly glycosylated domains of mammalian mucins and proteoglycans (51). Southern blot analysis suggests that the DNA sequence encoding the short C-terminal PPG1 domain of approximately 16 kDa is unique. Antibodies directed against this domain do not recognize the network-forming secreted fPPG, which suggests that this polypeptide is not encoded by ppg1, but most likely by other as yet unidentified members of the ppg gene family. The PPG1-specific anti-C domain antibodies react on immunoblots with a predominantly membrane-associated high molecular weight product of L. major promastigote lysates, and they bind to the surface membrane of fixed promastigotes. These results suggest that the ppg1 gene encodes a novel surface membranebound PPG species, for which we propose the name mPPG. There is evidence that the membrane association of mPPG is via a GPI anchor. The C-terminal hydrophobic peptide sequence of mPPG reassembles known GPI anchor addition sequences. Second, fusion of this C-terminal mPPG peptide to a secreted version of a L. mexicana membrane-bound acid phosphatase (41) and its expression in L. major promastigotes lead to GPI-phospholipase C-sensitive membrane association and cell surface localization of this enzyme. Attempts to directly demonstrate of a GPI anchor on mPPG by GPI-phospholipase C digestion and Triton X-114 phase separation of promastigote membranes led to inconclusive results, possibly due to poor solubility of membrane-bound mPPG in this detergent. Although we have evidence that mPPG can be biosynthetically labeled with [ 3 H]ethanolamine, 2 unequivocal confirmation of the GPI anchor on mPPG will most likely require purification of this molecule for structure analysis.
A variety of studies over the last 15 years have established LPG as the dominant cell surface glycoconjugate of Leishmania promastigotes and many functions have been proposed for this complex glycolipid (2,52). The identification of a membranebound form of PPG, which shares several phosphoglycan repeat and cap structures with LPG, raises the question whether it may also exhibit some of its biological and pharmacological activities. In particular, mPPG may be, due to its unusual structure, an ideal molecule as a complement acceptor and as a ligand for macrophage and sandfly midgut receptors. A conservative estimate of a contour length of 200 pm per amino acid in a phosphoglycosylated polypeptide domain (50,51) suggests that mPPG may extend more than 300 nm above the plasma membrane making it highly accessible for binding partners, whereas even the longest forms of LPG with 40 repeat units cannot extend further than 40 nm, more common estimates being 9 -17 nm (Refs. 53 and 54; compare Fig. 12D). Second, in comparison with LPG, mPPG most likely provides many more potential receptor and complement binding sites, since it may be modified with up to 800 phosphoserine-linked phosphoglycan chains, while LPG carries just one. We have recently shown that promastigotes of the LPG-deficient mutants Leishmania donovani R2D2 (55) and L. major LRC-L119 (56) still express phosphoglycan repeat or cap structures, or both, on their surface. The carrier molecule for these glycans is still unknown, but mPPG is a likely candidate. 3 Like several other Leishmania promastigote cell surface proteins such as gp63/leishmanolysin (57,58) and psa2/gp46 (48,49), mPPG belongs to a family of related genes. Preliminary sequence data of other members of the ppg gene family containing repeat domains larger than 23 kb suggest that, in contrast to ppg1, their gene products are not membrane-anchored. They may, therefore, encode secreted fPPG. 4 The predicted mPPG structure model (Fig. 12D) shows some remarkable similarities to the GPI-anchored mucin-like glycoproteins of the parasitic protozoon Trypanosoma cruzi (59) and the human tumor antigen variably called episialin or sialomucin (51,60,61). Both types of molecules are known to play a role in cell-cell interactions. The challenge for the future is to determine the precise biological function of mPPG and its relative the fPPG.