Processing and Trafficking of Leishmania mexicana GP63 ANALYSIS USING GPI8 MUTANTS DEFICIENT IN GLYCOSYLPHOSPHATIDYLINOSITOL PROTEIN ANCHORING*

GPI8 is a clan CD, family C13 cysteine protease and the catalytic core of the GPI-protein transamidase complex. In Leishmania mexicana, GPI8 is nonessential, and gpi8 mutants lack the GPI-anchored metalloprotease GP63, which is the major surface protein of promastigotes. We have identified the active site histidine and cysteine residues of leishmanial GPI8 and generated gpi8 lines expressing modified GPI8 proteins. This has allowed us to study the processing and trafficking of GP63 in wild type and gpi8 mutants. We show using pulse-chase labeling that in gpi8 non-GPI-anchored GP63 was glycosylated and secreted without further processing from the cell with a t1⁄2 of 120 min. This secretion was prevented by growth of cells in the presence of tunicamycin, indicating that glycosylation is necessary for secretion of non-GPI-anchored proteins. In contrast, in wild type cells the majority of GP63 was rapidly glycosylated, GPI-anchored, and trafficked to the surface with defined processing intermediate forms. Tunicamycin inhibited glycosylation but did not prevent GPI anchor addition or trafficking. These results show that GPI-anchored and unanchored GP63 are trafficked via different pathways. In addition, the balance between GPI anchor addition and secretion of GP63 in Leishmania can vary depending on the activity of the GPI-protein transamidase, which has implications for the hostparasite interaction.

Leishmania is a protozoan parasite that is the causative agent of a variety of human diseases collectively known as the leishmaniases. The parasite lives within the gut of the sandfly vector as motile proliferative procyclic promastigotes and in the mouth parts as motile nonproliferative metacyclic promastigotes. Within a mammalian host, the parasite lives as a nonmotile amastigote form that proliferates within macrophages. The surface of the promastigote is thought to contribute to survival within the sandfly and also invasion of and initial survival within a mammal (1). The surface is covered with a protective coat known as the glycocalyx, which is predominantly made up of glycosylphosphatidylinositol (GPI)-anchored 1 proteins, lipophosphoglycan, and glycoinositolphospholipids (2,3). A characteristic of Leishmania and other related trypanosomatids is the great abundance of GPI-anchored molecules on the cell surface, which is in contrast to higher eukaryotes in which the majority of surface proteins are attached via a transmembrane domain.
The major surface protein of Leishmania promastigotes is the 63-kDa GPI-anchored protein known as GP63. It is present at about 5 ϫ 10 5 molecules/parasite (4). GP63 is present on the amastigote at a greatly reduced level (5,6). GP63 is a zinc metalloprotease active in a neutral to alkaline pH range (pH 7-10) and is site-specific in its proteolytic activity (7,8). It is synthesized as an inactive precursor, which is activated via a cysteine switch mechanism (9). Latency is maintained by obstruction of the active site by the pro-region of the protein, with disruption of this complex resulting in an active enzyme. The precise mechanism and timing of this activation is not known, but it is not thought to be autocatalytic (9). The nascent protein has an endoplasmic reticulum (ER)-signaling sequence at the N terminus and adjacent to the regulatory pro-region. A GPI anchor addition site and a hydrophobic tail are present at the C terminus. These three domains are cleaved during the trafficking and processing of the protein.
The addition of a complete GPI anchor to a precursor protein occurs in the ER lumen by the simultaneous cleavage of the protein at the GPI anchor attachment site (near the C terminus) and addition of the GPI anchor in a transamidation reaction (10). The GPI anchor is attached to the protein by amide linkage between the terminal ethanolamine phosphate group of the anchor and the nascent C-terminal carboxyl group of the protein (11). The cleavage and anchor addition occur in one reaction catalyzed by a GPI-protein transamidase complex (12,13). The complex has been well characterized in mammalian cells and yeast and contains at least four components; GAA1, GPI8, PIG-S (GPI16), and PIG-T (GPI17) (14 -19). Studies using photo-cross-linking methods suggest that the complex may indeed contain more proteins (20). Recent studies demonstrated that GPI8 in mammalian cells associates directly with the protein to be anchored (20,21). GPI8 is thought to be the catalytic subunit of the GPI-protein transamidase complex and belongs to family C13 of the clan CD cysteine proteases (22), which are characterized as having a cysteine nucleophile with a catalytic dyad in the order histidine, cysteine (22).
The GPI-protein transamidase complex of trypanosomatids has not been characterized fully, although a cell-free assay for GPI anchoring in trypanosomes has been used to establish a reaction mechanism for GPI anchor addition (23). The Leishmania mexicana GPI8, unlike the mammalian and yeast homologues, has no transmembrane domain and appears to be a soluble homologue of the yeast and mammalian enzymes (24). A L. mexicana mutant lacking GPI8 (⌬gpi8) was deficient of GPI-anchored proteins, including GP63, yet remained viable in culture and was capable of infecting macrophages and also a mammalian host (24). The production of this L. mexicana ⌬gpi8 line provided an opportunity to study the effect that GPI anchor deficiency has on the processing and trafficking of a GPI protein. Secretory transport in trypanosomatids is thought to follow the general pathway found in higher eukaryotes (25,26), although endocytosis and exocytosis to the cell surface occurs via the flagellar pocket (27; reviewed in Ref. 28). In this study, we have defined the active site residues of L. mexicana GPI8 and demonstrated that GPI8 interacts with other proteins to carry out the transamidation reaction. The trafficking of GP63 in wild type and ⌬gpi8 cell lines was compared, and the role that the GPI anchor plays in the forward transport of GPIanchored proteins was examined.

EXPERIMENTAL PROCEDURES
Parasites-L. mexicana wild type (MNYC/BZ/M379) and ⌬gpi8 promastigotes were maintained in culture at 25°C in modified Eagle's medium containing 10% (v/v) heat-inactivated fetal calf serum. Neomycin (G418; Invitrogen) was added at 25 g ml Ϫ1 typically and up to 500 g ml Ϫ1 as required. Where necessary tunicamycin (Sigma) was added at 5 g ml Ϫ1 . Transfection of promastigotes with an episome was as described previously (24).
Metabolic Labeling-L. mexicana promastigotes were grown to midlog phase and washed twice in phosphate-buffered saline, and 6 ϫ 10 7 cells were resuspended in 1 ml of labeling medium (1ϫ minimum essential medium (ICN), 2 mM L-glutamine, 10% (v/v) dialyzed fetal calf serum), and 100 Ci of Expre 35 S 35 S ([ 35 S]methionine/cysteine protein labeling mix; NEN). The cells were grown at 25°C for 6 h and washed three times in ice-cold phosphate-buffered saline, and the cell pellets and medium fractions were stored at Ϫ80°C prior to analysis. For pulse-chase labeling experiments, 3.6 ϫ 10 7 cells were resuspended in 100 l of labeling medium containing 100 Ci of Expre 35 S 35 S/time point. The cells were labeled for 12 min at 25°C, washed three times in ice-cold phosphate-buffered saline, and resuspended in an equivalent volume of modified Eagle's medium containing 10% (v/v) fetal calf serum at 25°C. 100-l aliquots of cells were removed at appropriate time points, and the cells and medium were stored at Ϫ80°C in the presence of protease inhibitors (1 mM EDTA, 200 g ml Ϫ1 , Pefabloc SC, 5 g ml Ϫ1 pepstatin A, 40 g ml Ϫ1 leupeptin, 200 M phenylmethylsulfonyl fluoride, 1 mM phenanthroline).
Immunoprecipitation-Immunoprecipitation was carried out with GPI8 antibody as described previously (10). Briefly, the cell pellets were resuspended in 1ϫ solubilization buffer (50 mM Tris-HCl, pH 7.5,150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, and protease inhibitors) to 1 ml and centrifuged at 14,000 rpm for 15 min to preclear. 50 l of protein A/G-Sepharose (resuspended to a concentration of 0.5 mg/ml in solubilization buffer) with 6 l of ␣GPI8 were added to the supernatant, and the samples were mixed at 4°C for 12 h. The samples were then washed three times in TEN-D (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, 0.1% SDS, 0.5% deoxycholate) and once in TEN buffer (TEN-D buffer in the absence of detergent). 25 l of 2ϫ SDS loading buffer was added, and the samples were boiled prior to analysis by SDS-PAGE. For immunoprecipitation of GP63 antibody L3.8 was used (6). The cells were lysed in 1 ml of IDB buffer (1.25% Triton X-100, 190 mM NaCl, 60 mM Tris-HCl, pH 7.5, 6 mM EDTA, and protease inhibitors) and precleared. The medium samples had an equivalent volume of 2ϫ IDB buffer added and were then made up to 1 ml with 1ϫ IDB, 100 l of protein G-Sepharose (resuspended to 0.5 mg ml Ϫ1 in IDB), and 10 l of L3.8 antibody were added to the samples and incubated for 12 h at 4°C, and the beads subsequently washed three times in GP63 wash buffer (0.1% Triton X-100, 0.02% SDS, 150 mM Tris-HCl, pH 7.5) and once in TEN. 40 l of 2ϫ SDS-PAGE loading buffer was added to the samples prior to further analysis.
Triton X-114 Fractionation-Triton X-114 fractionation was performed following the method of Bordier (29). The cells were lysed in 200 l of Triton X-114 buffer (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.5% precondensed Triton X-114) on ice for 15 min and precleared by centrifugation at 10,000 ϫ g for 10 min at 4°C, and the supernatant was overlaid onto a sucrose cushion (10 mM Tris-HCl, pH 7.4, 150 mM NaCl, 6% w/v sucrose, 0.06% precondensed Triton X-114) in a fresh Eppendorf tube. The samples were incubated at 30°C for 3 min and centrifuged at 300 ϫ g for 3 min at room temperature, and the upper aqueous layer was removed to a fresh tube. 0.5% precondensed Triton X-114 was added to this sample and incubated at 4°C for 10 min, and then this upper layer was overlaid back on the original sucrose cushion. The sample was incubated at 30°C for 3 min and centrifuged for 3 min at 300 ϫ g at room temperature. The whole aqueous phase was removed to a fresh tube, leaving a small Triton X-114 pellet containing the membrane fraction. The aqueous phase was treated with 2% precondensed Triton X-114, incubated at 4°C for 10 min, transferred to 30°C for 3 min, and centrifuged at 300 ϫ g at room temperature for 3 min. The aqueous phase containing the soluble cell fraction was then transferred to a fresh tube, and the soluble and membrane fractions were subjected to further analysis.
Purification of GP63 on Concanavalin A-The cells were grown to stationary phase, the cells were harvested, and the medium was filtered and retained. The cells were lysed in 1ϫ ConA binding buffer (10 mM Tris-HCl, pH 7.4, 0.5 M NaCl, 0.5% Triton X-100, 1 mM CaCl 2 , 1 mM MnCl 2 ) for 30 min and centrifuged to preclear, and 500 l of ConA-Sepharose (resuspended to 4 mg ml Ϫ1 in ConA binding buffer) and protease inhibitors (200 g ml Ϫ1 Pefabloc SC, 5 g ml Ϫ1 pepstatin A, 40 g ml Ϫ1 leupeptin) were added. To the medium 1 volume of 2ϫ ConA binding buffer, 500 l of ConA-Sepharose, and protease inhibitors were added. The samples were mixed for 12 h at 4°C and washed four times in ConA binding buffer, and glycosylated proteins were eluted with an appropriate volume of elution buffer (1 M methyl ␣-D-mannopyranoside in ConA binding buffer).
PI-PLC Treatment-The cells were pulse-chase labeled, and two samples were collected for each time point, washed three times in ice-cold phosphate-buffered saline, and snap frozen. These samples were lysed in 200 l of Triton X-114 buffer but with the absence of Triton X-114 and the addition of 0.05% Triton X-100, incubated at room temperature for 10 min, and centrifuged to preclear. The samples were then transferred to a fresh tube, 4 l of PI-PLC (Sigma) was added to one sample from each time point, and all of the samples were incubated at 37°C for 1 h. 0.5% precondensed Triton X-114 was added to each sample, and the samples were incubated at 4°C subsequent to Triton X-114 fractionation as described previously.

Identification of Active Site Residues of L. mexicana
GPI8 -We demonstrated previously that L. mexicana GPI8 is a member of the C13 family of cysteine endopeptidases and a component of the GPI-protein transamidase of the parasite (24). We identified two cysteine (Cys 94 and Cys 216 ) and two histidine (His 63 and His 174 ) residues of the L. mexicana GPI8 that potentially could comprise the active site catalytic dyad from sequence alignment of yeast and human GPI8 homologues (24). To test whether these residues are essential for GPI-protein transamidase activity, each of the four residues was mutated individually in GPI8 by site-directed mutagenesis (H63A, C94G, H174A, and C216G). The mutated GPI8 genes, cloned in the shuttle vector pX, were transfected into the L. mexicana GPI8 deletion mutant (⌬gpi8) to give lines The ⌬gpi8 line itself is viable in culture and has been shown to be deficient in the major GPIanchored surface protein of L. mexicana, GP63 (24). Re-expression of GPI8 in ⌬gpi8 (the ⌬gpi8[pXGPI8] line) restored GPIanchored GP63 to the cell surface (24). Western blot analysis with an antibody specific to GP63 was used to assess the effect of the GPI8 mutations on the ability of the GPI-protein transamidase complex to add GPI anchors onto GP63 (Fig. 1A). GP63 was present in wild type promastigotes (lane 1) but not detectable in ⌬gpi8 (lane 2). Re-expression of the native gene restored GP63 (lane 3). In ⌬gpi8 expressing GPI8 H174A or GPI8 C216G , GP63 was not detected (lanes 6 and 7, respectively), whereas GP63 was present in cell lines expressing GPI8 H63A or GPI8 C94G (lanes 4 and 5). Immunofluorescence analysis of the different cell lines using an antibody that detects native GP63 showed surface expression of GP63 in ⌬gpi8[pXgpi8 H63A ] and ⌬gpi8[pXgpi8 C96G ] but not in ⌬gpi8[pXgpi8 H174A ] or ⌬gpi8[pXgpi8 C216G ] (data not shown).
To confirm that GPI8 was expressed in each of the cell lines transfected with mutated copies of GPI8, metabolic labeling of cells with Expre 35 S 35 S and subsequent immunoprecipitation with an antibody raised against recombinant L. mexicana GPI8 (10) was performed (Fig. 1B). A 42-kDa protein, which is the predicted size of GPI8, was detected in cell lines expressing either GPI8 (lane 2) or one of the four GPI8 mutants (lanes 3-6) but not in ⌬gpi8 (lane 1). These data demonstrate that although each of the four mutated GPI8 proteins is expressed in ⌬gpi8, only two of the mutated GPI8 proteins, GPI8 H63A and GPI8 C94G , can form part of a functional GPI-protein transamidase. The other two mutated proteins, GPI8 H174A and GPI8 C216G , result in inactive complexes. These findings provide convincing evidence that His 174 and Cys 216 are the catalytic active site dyad of L. mexicana GPI8.
Leishmanial GPI8 Is Essential for Transamidase Activity-In yeast and mammalian cells GPI8 has been shown to form a complex with other proteins to form an active GPIprotein transamidase (20,30). In an attempt to define the role of GPI8 in L. mexicana, the gene encoding nonfunctional GPI8 C216G was expressed from an episome in wild type cells (to give cell line WT[pXgpi8 C216G ]). This cell line was compared with wild type promastigotes expressing episomal GPI8 (cell line WT[pXGPI8]). Equal expression of GPI8 was verified in the two cell lines by metabolic labeling with Expre 35 S 35 S and subsequent immunoprecipitation with an ␣-GPI8 antibody ( Fig. 2A). The expression of GPI-anchored GP63 was examined by Western blot analysis (Fig. 2B). WT cells expressing the functional episomal copy of GPI8 (Fig. 2B, lanes 3-5) were found to express GP63 at levels similar to the wild type parasites (lane 1). An increase in the concentration of G418, and hence GPI8 levels, did not alter the levels of GP63 (compare lanes [3][4][5]. However, cells expressing the mutated form of GPI8 showed decreased levels of GP63 in the cells (compare lanes 1 and 6), and the amount of GP63 decreased as GPI8 C216G expression increased (compare lanes 6 -8). Overexpression of GPI8 did not appear to affect GP63 surface expression as examined by immunofluorescence, whereas expression of GPI8 C216G in wild type cells drastically reduced the amount of GP63 on the cell surface (data not shown). At the highest concentration of antibiotic (500 g ml Ϫ1 ), surface expression of GP63 was almost undetectable. Thus expression of GPI8 C216G in wild type promastigotes produced a pronounced dominantnegative effect, which provides compelling evidence that GPI8 is required for transamidation activity and is likely to be part of a GPI-protein transamidase complex in Leishmania.
Non-GPI-anchored GP63 Is Secreted-To determine whether GP63 was being degraded or secreted in the ⌬gpi8 mutant, WT promastigotes, ⌬gpi8, and lines containing mutated GPI8 were labeled for 6 h with Expre 35 S 35 S and GP63 immunoprecipitated from their respective culture supernatants. With WT parasites (Fig. 3A, lane 1), a small amount of three isoforms of GP63 were detected (of 63, 64, and 65 kDa, which are therefore designated 63s, 64s, and 65s (for secreted)), with the 63s isoform being the most abundant. In contrast, a large amount of the 65s isoform was detected in the medium in which ⌬gpi8 cells had been grown (lane 2). This 65s form secreted from ⌬gpi8 cells was subsequently found to be an isoform of GP63 The presence of GP63 was also analyzed in the cell lysates (Fig. 3B) by immunoprecipitation with the same anti-GP63 antibody. Three isoforms of GP63 were detected in WT cells (of 63, 64, and 65 kDa, which are therefore designated 63c, 64c, and 65c (for cell-associated)), with the 63c isoform being most abundant. In the ⌬gpi8 cell line, only the 65c isoform was detected (lane 2). The two smaller isoforms of GP63 (63c and 64c) were present in ⌬gpi8[pXGPI8] lysates (lane 3) at approximately equal levels. ⌬gpi8 cell lines expressing GPI8 H174A (lane 6) or GPI8 C216G (lane 7) were similar in cell-associated profiles to ⌬gpi8 with the 65c isoform predominating, whereas the ⌬gpi8 cell line expressing GPI8 H63A (lane 4) was the same as wild type with abundant 63c isoform. The ⌬gpi8 cell line expressing GPI8 C94G (lane 5) gave an intermediate pattern, with all three isoforms of GP63 present. These data confirm that GPI8 C216G and GPI8 H174A are nonfunctional enzymes, whereas GPI8 H63A is fully functional and GPI8 C94G is partially functional.
Soluble GP63 Is Secreted Rapidly from ⌬gpi8 -Clearly, lack of an active GPI8 affects the processing and trafficking of the GPI-anchored protein GP63 when compared with WT cells. To examine these variations in more detail, wild type promastigotes and ⌬gpi8 and ⌬gpi8[pXgpi8 C216G ] cells were labeled with Expre 35 S 35 S for 12 min and then chased in cold medium for a period up to 300 min. GP63 was immunoprecipitated from the culture medium (Fig. 4A) or cells (Fig. 4B) and analyzed by SDS-PAGE. Only a very low level of 35 S-labeled GP63 was secreted from wild type parasites, whereas high levels of secreted GP63 could be detected over a 180-min chase period for both ⌬gpi8 and ⌬gpi8[pXgpi8 C216G ]. After 300 min, the level of GP63 secretion from WT cells was estimated to be only 13% of that secreted from ⌬gpi8 cells, with the t1 ⁄2 for ⌬gpi8 GP63 secretion estimated to be 120 min (Fig. 4C). This clearly demonstrates the higher rate of secretion of newly synthesized GP63 in cells with a nonfunctional GPI8. Two forms of GP63 were secreted from wild type cells, a 65s isoform and a 63s isoform (Fig. 4A, WT, time 180).
The processing of GP63 within the cells was also examined by partitioning the cells into soluble and membrane-associated fractions by either extraction with NaCO 3 (data not shown) or Triton X-114 (Fig. 4B). Both methods of fractionation produced similar results. In wild type cells, GP63 partitioned exclusively into the membrane-associated fraction at all time points. At the start of the chase, the major isoform of GP63 was 65c. After 20 min, the majority of 65c had been chased into 64c, and by 40 min 64c was in the process of being chased into 63c. By 180 min all detectable label was in the 63c isoform. In contrast to wild type cells, only the 65c isoform was detected in the ⌬gpi8 and ⌬gpi8[pXgpi8 C216G ] lines (Fig. 4B). Moreover, most of 65c remained in the soluble phase in ⌬gpi8 and ⌬gpi8[pXgpi8 C216G ] cells.
Nascent GP63 Is Rapidly GPI-anchored-The timing of GPI anchor addition during the processing of GP63 was examined by determining which forms of GP63 have a GPI anchor. WT promastigotes were labeled with Expre 35 S 35 S for 12 min and then chased in cold medium for a period up to 180 min. The samples were taken at 0, 40, and 180 min, and the lysates were treated with or without PI-PLC followed by Triton X-114 fractionation and GP63 immunoprecipitation (Fig. 5). At the start of the chase, the major 65c band was present in the membraneassociated fraction (lane 2), but after PI-PLC treatment it was found in the soluble fraction (lane 3), consistent with the GPI anchor having been removed. The samples from time 40 min and time 180 min showed the same pattern for the maturing GP63 following PI-PLC treatment. All forms of GP63 identified were found only in the soluble fraction following treatment. This demonstrates that GP63 GPI addition occurred very rapidly subsequent to translation and translocation to the ER. It   FIG. 3. Secretion of GP63 from ⌬gpi8. Leishmania promastigotes were cultured with Expre 35 S 35 S for 6 h. Samples from the medium (A) or cell lysates (B) were collected, immunoprecipitated with anti-GP63 antibody, and electrophoresed on a 12% PAGE gel. The gels were scanned with a PhosphorImager. Secreted GP63 (65s, 64s, and 63s) and cellular GP63 (65c, 64c, and 63c)  should also be noted that a fourth minor isoform of GP63 could occasionally be detected (lanes 6 and 7).
Glycosylation of GP63 Occurs Prior to Addition of the GPI Anchor-Glycosylation of GP63 was examined by testing whether the protein bound to concanavalin A, which interacts with N-linked glycans (Fig. 6A). GP63 did not bind to Sepharose beads nonspecifically (data not shown). A much higher level of glycosylated GP63 bound to ConA from WT promastigotes (lane 1) than ⌬gpi8 cells (lane 2). In contrast glycosylated GP63 was present at a higher level in the medium of ⌬gpi8 cells (lane 3) compared with that from WT cells (lane 4), and the size of GP63 precipitated was larger in ⌬gpi8. This demonstrates that GP63 is N-glycosylated in both wild type and ⌬gpi8 cells and that N-linked glycosylation can occur in the absence of GPI-protein transamidase activity.
The two cell lines were grown in medium containing 5 g ml Ϫ1 tunicamycin to inhibit N-linked glycan formation. Pulsechase labeling was used to examine the processing of GP63 under these conditions (Fig. 6, B and C). GP63 was processed differently in WT cells grown in the presence or absence of tunicamycin. A smaller form (of ϳ63 kDa, which is therefore designated 63ct (for cellular material with tunicamycin)) was present at time 0, and this was chased into a 60-kDa form (60ct) after 180 min. Only a single minor intermediate form of GP63 was identified in cells grown in the presence of tunicamycin, compared with the one major (64c) and one minor band identified in normally grown WT cells. The ⌬gpi8 cells grown in the presence of tunicamycin also expressed a smaller GP63 (63ct). However, this was not chased to a 60-kDa form. The size difference between the proteins isolated at time 0 from cells grown in the presence and absence of tunicamycin correlates with the lack of N-glycosylation. The lower number of detectable intermediate forms present in WT cells grown in the presence of tunicamycin indicates that one of the isoforms detected during GP63 maturation may be a result of N-glycan processing. GP63 could not be detected in the medium of either ⌬gpi8 or wild type cells when grown in the presence of tunicamycin (Fig. 6C), showing that N-linked glycosylation is important for the secretion of GP63 from both WT and ⌬gpi8. DISCUSSION We have identified His 174 and Cys 216 as the L. mexicana GPI8 catalytic dyad characteristic of the C13 family of clan CD cysteine proteases. Mutation of either residue leads to the loss of GPI-anchored GP63 on the surface of the parasite, demonstrating that these residues are essential for activity of GPI8 and therefore GPI anchoring in Leishmania. These findings complement previous results demonstrating that a sulfhydryl group acts as the active site residue for the GPI-protein transamidase of Trypanosoma brucei (23) and that recombinant L. mexicana GPI8 can be inactivated by thiol alkylating agents in a cell-free system (10). Moreover, His 174 and Cys 216 of L. mexicana GPI8 align with those identified recently as the active site histidine and cysteine residues in yeast and human GPI8 (30,31). Removal of GPI-protein transamidase activity from trypanosome membranes by a high pH wash and reconstitution of that activity with recombinant GPI8 suggested that in trypanosomatids GPI8 is part of a complex and that this complex may be dynamic (10). The findings reported here that expression of the active site mutant GPI8 C216G in wild type parasites produced a pronounced dominant-negative effect, with GPI-protein transamidase activity being severely downregulated, provide further evidence that L. mexicana GPI8 is part of a larger protein complex. In yeast and higher eukaryotes, GPI8 is thought to form a stable but dynamic complex with at least three other components, GAA1, GPI16/PIG-S, and GPI17/PIG-T (16,19,31). No homologue of other transamidase complex members has yet been cloned from protozoa, although analysis of sequence data has identified a possible GAA1 homologue in Leishmania major (GenBank TM accession number CAB86709 (32)) and a similar protein in T. brucei (GenBank TM accession number AC087702).
We reported previously that GP63 could not be detected in the culture medium of ⌬gpi8 cells (24). In the current study we used pulse-chase labeling and immunoprecipitation assays that are considerably more sensitive than the previously employed methods, and this enabled us to show that GP63 was indeed secreted (Fig. 4). Thus Leishmania differ from higher eukaryotes in which non-GPI-anchored proteins accumulate in the ER (33,34) before being exported to the cytosol for degradation (35,36). Interestingly, a variant surface glycoprotein lacking the C-terminal GPI signal attachment sequence accumulated in the ER of trypanosomes (37). The variable surface glycoprotein was correctly folded and dimerized, and so it was concluded that retention in the ER was due to the lack of the GPI anchor itself, which plays an active role in forward trans-  (63ct and 60ct). C, medium fractions were immunoprecipitated with anti-GP63 antibody and electrophoresed on a 12% PAGE gel. port (38). In addition, replacement of the hydrophobic tail of the variable surface glycoprotein with a transmembrane domain resulted in ER accumulation followed by lysosomal degradation (39). Our results clearly demonstrate that retention in the ER of the unanchored form of GP63 does not occur in Leishmania. These findings are in agreement with previous studies utilizing expressed forms of GP63 that had been mutated to prevent GPI anchor addition (40).
The situation in Leishmania appears relatively complex, as evidenced by the observation that GP63 was processed and trafficked differently in wild type cells and the mutants containing modified GPI8 or entirely lacking the protein (Fig. 4) (Fig. 3). This suggests that this mutated GPI-protein transamidase is dysfunctional, being active but less so than the wild type protein. This could reflect differing abilities to form the functional complexes, perhaps because the Cys 94 residue is necessary for optimal folding of the GPI8 and/or is directly involved in its interaction with other proteins of the complex. Mutation of His 54 in yeast GPI8 and Cys 92 in human GPI8 also led to a partial loss of GPI-protein transamidase function (30,31), suggesting that protein-protein interactions are affected directly or indirectly by these residues. Our results with Leishmania suggest that the relative amounts of GP63 (or other GPI-anchored proteins such as the proteophosphoglycan (41)) directed to the surface of the parasite or secreted could be controlled by regulation of GPI8 activity. This may have an influence on the ability of the parasite to survive in the variety of environmental conditions that it encounters.
Characterization of the processing of GP63 in wild type promastigotes by pulse-chase labeling revealed that the first detectable cell-associated product was a 65-kDa form (Fig. 4B,  65c) that was GPI-anchored and glycosylated. This finding shows that addition of the GPI anchor to GP63 must occur very rapidly following translation and translocation into the ER and subsequent to processing of the N-terminal signal sequence. We have shown by immunofluorescence microscopy that GPI8 co-localizes with QM, a 60 S ribosomal protein, which suggests the GPI-protein transamidase has a close association with the rough ER (42). In eukaryotes the translocon contains the machinery for N-linked glycosylation, so it is likely that this event occurs during translocation into the ER. Our results support the proposal that the GPI-protein transamidase is closely associated with the translocon (20) and that GPI anchor processing occurs very soon post-translation (21).
The GPI-anchored 65c form was processed further to a 64c protein within 20 min and a mature 63c form within 40 min (Fig. 5). By 180 min, all of the label had been chased into the 63c form. These events can be accounted for by secondary glycosylation events (which are inhibited by tunicamycin; Fig.  6) and the cleavage of the pro-region to activate the enzyme. The processing events in the ⌬gpi8 line differed from those occurring in wild type parasites in that the cell soluble associated 65c protein present at time 0 did not undergo any processing events throughout the chase, as assessed by size variation. Thus although the GP63 in the ⌬gpi8 cell line was glycosylated during translocation (Fig. 6), no modifications could be detected thereafter.
L. amazonensis GP63 has three potential N-glycosylation sites (43). Previous studies have demonstrated that N-glycans are not required for either the activation or cell surface expression of GPI-anchored GP63 (40,44). This study demonstrates that GPI anchoring is not required for the addition of N-linked glycans to GP63 (Fig. 6A). We conclude that GP63 is N-glycosylated in both the wild type and ⌬gpi8 cell lines, and this process is independent of GPI anchoring. Inhibition of N-linked glycosylation by growth of cells in tunicamycin produced a smaller sized GP63 protein (63 kDa chased to a size of 60 kDa). In wild type cells incubated in the presence of tunicamycin, this protein underwent two size modifications, compared with three in wild type cells, showing that some of the size changes that GP63 normally undergoes are a result of glycosylation, possibly because of modifications to the N-linked glycans during trafficking. The final size change is likely to be due to the cleavage of the pro-region associated with metalloprotease activation. Because GP63 did not alter in size during the chase period in ⌬gpi8 cells, this indicates that the precursor enzyme remains unprocessed and therefore in an inactive state. 65-and 63-kDa forms of GP63 were found to be in the culture supernatant of wild type cells (Fig. 4). McGwire et al. (45) recently reported two extracellular forms of gp63, one secreted and one released from the cell surface. Proteolytically active, extracellular GP63 may influence the survival of the parasite in its host (45,46).
These results show that GPI-anchored GP63 and unan- FIG. 7. Model of GP63 processing in wild type and ⌬gpi8 cells. The model proposes that in WT cells the majority of GP63 is N-glycosylated and GPI-anchored rapidly in the ER (to give a 65-kDa form, the number in parentheses indicates the estimated size of the GP63 isoforms). N-Glycan processing (not defined in this model) and pro-domain removal give a mature, active 63-kDa isoform that is transported to the cell surface. In wild type cells some GP63 is N-glycosylated and secreted from the cell without further modification. In ⌬gpi8 all GP63 destined for secretion is N-glycosylated and transported without further modification. chored GP63 are trafficked via different pathways: a classical pathway whereby GP63 is GPI-anchored, processed, and trafficked to the cell surface and a direct secretion pathway whereby non-GPI-anchored GP63 is rapidly exported from the cell without further modification (Fig. 7). The secretion pathway is directed by N-glycans, because the loss of glycosylation results in loss of secretion. In contrast, trafficking of the GPIanchored GP63 to the cell surface appears to be regulated exclusively by the presence of a GPI anchor, because the loss of glycosylation does not affect forward trafficking. This study does not address whether GPI anchors act as a signal for the forward transport of proteins or whether they play an active role in protein transport. GPI-anchored proteins become insoluble to detergent extraction during trafficking through the secretory pathway, forming detergent-resistant membranes. Lipid microdomains are thought to occur in protozoa (47)(48)(49), and detergent-resistant membranes of Leishmania are enriched in components characteristic of eukaryotic lipid rafts: inositol phosphorylceramide, sterols such as ergosterol, and GPI-anchored molecules (both GP63 and LPG). Lipid rafts may be relevant to the forward trafficking and processing of GPI proteins in Leishmania, and studies are ongoing to investigate these in wild type and GPI8-deficient parasites.