The Role of Inositol Acylation and Inositol Deacylation in the Toxoplasma gondii Glycosylphosphatidylinositol Biosynthetic Pathway*

Toxoplasma gondii is a ubiquitous parasitic protozoan that invades nucleated cells in a process thought to be in part due to several surface glycosylphosphatidylinositol (GPI)-anchored proteins, like the major surface antigen SAG1 (P30), which dominates the plasma membrane. The serine protease inhibitors phenylmethylsulfonyl fluoride and diisopropyl fluoride were found to have a profound effect on the T. gondii GPI biosynthetic pathway, leading to the observation and characterization of novel inositol-acylated mannosylated GPI intermediates. This inositol acylation is acyl-CoA-dependent and takes place before mannosylation, but uniquely for this class of inositol-acyltransferase, it is inhibited by phenylmethylsulfonyl fluoride. The subsequent inositol deacylation of fully mannosylated GPI intermediates is inhibited by both phenylmethylsulfonyl fluoride and diisopropyl fluoride. The use of these serine protease inhibitors allows observations as to the timing of inositol acylation and subsequent inositol deacylation of the GPI intermediates. Inositol acylation of the non-mannosylated GPI intermediate d-GlcNα1–6-d-myo-inositol-1-HPO4-sn-lipid precedes mannosylation. Inositol deacylation of the fully mannosylated GPI intermediate allows further processing, i.e. addition of GalNAc side chain to the first mannose. Characterization of the phosphatidylinositol moieties present on both free GPIs and GPI-anchored proteins shows the presence of a diacylglycerol lipid, whose sn-2 position contains almost exclusively an C18:1 acyl chain. The data presented here identify key novel inositol-acylated mannosylated intermediates, allowing the formulation of an updated T. gondii GPI biosynthetic pathway along with identification of the putative genes involved.

Toxoplasma gondii is a highly successful parasite, causing the disease toxoplasmosis that affects humans and a wide variety of mammals. The infection is normally controlled by the immune system of healthy individuals, leading to an asymptomatic infection. Although human infections are usually relatively benign and lead to lifelong immunity, acute T. gondii infection during pregnancy in the non-immune host can cause congenital transmission resulting in serious birth defects (1,2). Furthermore, T. gondii represents a major opportunistic organism in immunocompromised individuals, causing fatal encephalitis; thus, T. gondii is a major contributor to AIDS-related disease (3,4). Toxoplasma as with other Apicomplexa has an obligatory intracellular developmental stage where it multiplies before bursting from a host cell, pre-adapted to infect other host cells. The ability to invade nearly all nucleated cells in an active multistep process (5-7) is thought to be in part due to several surface proteins, among them the major surface antigen SAG1 (P30), which establishes the first contact between the parasite and the host cell (8,9). As in the case with many other parasitic protozoa (10 -14), glycosylphosphatidylinositol (GPI) 3 -anchored proteins like SAG1 dominate the plasma membrane. Protozoa in general express significantly higher densities of cell-surface GPI-anchored proteins than do higher eukaryotes.
The sequence of events underlying GPI biosynthesis has been studied in several organisms including Trypanosoma brucei (21)(22)(23)(24)(25)(26)(27)(28)(29), Trypanosoma cruzi (30), Plasmodium falciparum (31), Leishmania (32,33), Saccharomyces cerevisiae (34,35), and mammalian cells (36 -38) and to a minor extent in T. gondii (16, 39 -41). In all cases, GPI biosynthesis involves the initial addition of GlcNAc from UDP-GlcNAc to phosphatidylinositol (GPIno) to form GlcNAc-PI, which is then de-N-acetylated to GlcN-PI (43)(44). De-N-acetylation is a prerequisite for mannosylation of GlcN-PI to form later GPI intermediates (45)(46). These early events occur on the cytoplasmic face of the endoplasmic reticulum, whereas the subsequent steps of mannosylation by three separate Dol-P-Man-dependent mannosyltransferases (23) occurs on the luminal face (10), this followed by the transfer of ethanolamine phosphate to the third mannose from phosphatidylethanolamine (48). A recent study on the topology of the Toxoplasma GPI biosynthesis suggests the transfer of GalNAc and Glc to the first mannose takes place on the cytoplasmic face of the ER (49). The pre-assembled glycolipid GPI precursor is then flipped inside the ER where it is transferred en bloc to a protein via a transamidase reaction involving the cleavage of a hydrophobic C-terminal GPI signal sequence (for review, see Refs. 50 and 51).
Nevertheless, significant differences in the timing of certain biosynthetic steps occur between organisms, including the common but sometimes transient addition of an acyl chain linked to the 2-hydroxyl of the myo-inositol of GPI-anchor precursors. In P. falciparum as well as mammalian and yeast cells this inositol acylation is dependent upon acyl-CoA and is required before the addition of the first mannose (52)(53). In contrast, T. brucei inositol acylation only occurs after the addition of the first mannose (24,27), where it is a prerequisite for the subsequent addition of an ethanolamine phosphate to the third mannose (24,27). Uniquely, T. brucei inositol acylation is not acyl-CoA-dependent, and both the inositol acylation and inositol deacylation are inhibited by the serine protease inhibitors phenylmethylsulfonyl fluoride (PMSF) and diisopropyl fluoride (DFP) respectively.
In this paper we show that the serine protease inhibitors PMSF and DFP have a drastic effect on the T. gondii GPI pathway. Using a cell-free system and in vivo labeling allowed the characterization of novel mannosylated inositol-acylated T. gondii GPI intermediates. These allowed the elucidation of the roles of inositol acylation and inositol deacylation in the T. gondii GPI biosynthetic pathway.

EXPERIMENTAL PROCEDURES
Materials-D- [ Parasites and Cell Culture-RH strain T. gondii tachyzoites were grown in Vero cells (20). Confluent cell cultures (175 cm 2 ) were infected with 5 ϫ 10 7 tachyzoites in Dulbecco's modified Eagle's medium supplemented with 1% (v/v) fetal calf serum. Tachyzoites were harvested after 72 h and set free from their host cells using the Mixer Mill homogenizer (Retsch), and the suspension was run through a 20-ml glass wool column to remove cellular debris. The purity of the tachyzoite suspension was monitored microscopically. Cell lines and parasites were routinely tested for Mycoplasma contamination.
Metabolic Labeling of Tachyzoites-After infection with T. gondii (72 h post-infection), cell cultures were washed twice with glucose-free culture medium containing 20 mM sodium pyruvate. Labeling was performed using the same medium supplemented with 0.5 mCi D-[6-3 H]glucosamine for 6 h at 37°C. After labeling, parasites were liberated from host cells and purified as described previously (39 -40). Those in vivo labelings involving PMSF (1 mM) or DFP (1 mM) were preincubated for 6 h with D-[6-3 H]glucosamine before the addition of inhibitor and incubated for a further 5 min. Radiolabeled glycolipids were extracted and analyzed as described below.
Extraction of Glycolipids-All in vivo and in vitro labeled glycolipid products were extracted three times in chloroform/ methanol/water (10:10:3, v/v), dried, and recovered from a butan-1-ol partitioning, as previously described (20).
HPTLC-Samples as well as glycolipid standards were applied to 10-cm aluminum-backed silica gel 60 HPTLC plates that were developed using solvent system A, hexane, chloroform, methanol, water, acetic acid (3:10:10:2:1, v/v), or solvent system B, chloroform, methanol, 1 M ammonium hydroxide (10:10:3, v/v), both before and after enzymatic and chemical digests. Radiolabeled components were detected using either a Berthold LB2842 or a Bioscan AR-200 linear analyzer or by fluorography at Ϫ70°C after soaking in EA-Wax and using Kodak XAR-5 film with an intensifying screen.
Enzymatic and Chemical Treatments of Radiolabeled Glycolipids-Digestions with JB␣M, GPI-PLD, PI-PLC, base hydrolysis, deamination, and N-acetylation were performed as previously described (Ref. 25 and references therein). jack bean ␤-N-acetylhexosaminidase and ␣-glucosidase were performed following the manufacturer's instructions.
Glycan Head-group Analysis-The HPTLC-purified radiolabeled glycolipids were delipidated, deaminated, reduced, dephosphorylated with aqueous HF, and desalted by passage through AG50X12 (H ϩ ) and AG3X4 (OH Ϫ ) ion-exchange resins. The resulting neutral glycan head groups were analyzed before and after various glycosidic digests each by Bio-Gel P4 gel filtration (Ref. 16 and references therein). For these analyzes the radiolabeled glycans were detected by scintillation counting and correlated with the elution positions of the co-injected individual glucose oligomer standards.
Exoglycosidase Digestions of Glycans-Glycans obtained from HPTLC-purified radiolabeled glycolipids were digested with either JB␣M, jack bean ␤-N-acetylhexosaminidase, or ␣-glucosidase. All enzymatic digests were terminated by heating at 100°C for 5 min. The samples were desalted by passing through 0.25 ml of AG50X12(H ϩ ), dried, and flash co-evaporated with toluene to remove residual acetic acid.
Identification of the Phosphatidylinositol Moieties from Purified Free GPIs and GPI-anchored Proteins-T. gondii free GPIs and GPI-anchored proteins were purified as previously described (54). Aliquots (2-10 nmol) were dried and dissolved in 15 l of sodium acetate (0.3 M, pH 4) followed by the addition of 7.5 l of freshly prepared sodium nitrite (1 M) and incubated for 1 h at room temperature. An additional 15 l of sodium acetate (0.3 M, pH 4) and 7.5 l of freshly prepared sodium nitrite (1 M) were added and incubated for a further 2 h at 37°C. The GPIno moiety released by deamination was partitioned into butan-1-ol (3 ϫ 100 l). The pooled butan-1-ol extracts were dried and suspended in chloroform/methanol (1:2) and analyzed by negative ion electrospray mass spectrometry (ES-MS) on a Quattro Ultima triple quadrapole instrument. Samples were introduced into the mass spectrometer using nanospray tips. GPIno species were observed in negative ion mode with a capillary voltage of 0.9 kV and a cone voltage of 40 -60 V. Daughter ion ES-MS-MS spectra were obtained with a collision voltage of 35-50 V using argon at 3 ϫ 10 Ϫ3 torr as the collision gas. MassLynx was used to record and process the data.
Identification of Toxoplasma GPI Biosynthetic Genes-Genes for GPI biosynthesis were identified by using sequences of homologue genes for the GPI biosynthesis of different species as probes to systematically search for the homologous genes in data provided by the Toxoplasma Genome resource. In all cases putative homologous genes were identified and analyzed as to their suitability to be true homologues. The homologues were checked for Pfam motifs and conserved residues as well as topology and hydrophobicity comparisons, after which they were submitted to EMBL and GenBank TM (see Table 2).   Table 1.

Formation of GlcN-(acyl)PI-
(data not shown) to the cell-free system causes the formation of a third [ 3 H]GlcN glycolipid, which is resistant to PI-PLC but sensitive to GPI-PLD, deamination, and base hydrolysis (Figs. 1, C-E, and Table 1), suggesting the formation of GlcN-(acyl)PI in T. gondii. This inositol acyltransferase activity, like that of the P. falciparum, S. cerevisiae, and mammalian GPI pathways, acts on GlcN-PI (14,33) and is acyl-CoA-dependent, preferring Pam-CoA over myristyl-CoA or stearoyl-CoA (data not shown).
The addition of sulfydryl alkylating agents such as N-ethyl maleimide or iodoacetamide and removal of dithiothreitol prevented the formation of any [ 3 H]GlcN glycolipids compared with controls, suggesting that like GPI biosynthetic pathways of other organisms, the UDP-GlcNAc:PI GlcNAc transferase complex is inhibited by sulfhydryl alkylating agents (45).  (Table 1). These results are consistent with all three glycans having a Man 3 -anhydromannitol head group, but two of the three [ 3 H]mannosylated glycolipids were resistant to JB␣M, suggesting a group on the nonreducing Man, probably an ethanolamine phosphate group. These data taken together with their R f suggest that they are glycolipids similar to the T. brucei GPI intermediates Man 3 GlcN-PI (M3), EtNP-Man 3 GlcN-PI (AЈ), and EtNP-Man 3 GlcN-(lyso)PI (lyso-AЈ).

Inositol Acylation of GlcN-PI Is Required for Mannosylation-
Replacement of UDP-GlcNAc with an exogenous synthetic acceptor d-GlcN␣1-6-d-myo-inositol-1-HPO 4 -sn-1,2-dipalmitoylglycerol (GlcN-PI) in the presence of Pam-CoA (Fig. 2C) resulted in the formation of [ 3 H]mannosylated glycolipids, whose R F and characterization are the same to those formed with UDP-GlcNAc (Fig. 2B), including an extra minor [ 3 H]mannosylated glycolipid identified as lyso-M3 ( Table 1). The incorporation of [ 3 H]mannose into the [ 3 H]glycolipids formed utilizing the exogenous acceptor GlcN-PI ( Fig. 2C) is about double that of the endogenously UDP-GlcNAc-primed (Fig. 2B), probably because GlcN-PI only has to be inositol-acylated before [ 3 H]mannosylation as opposed to endogenous priming UDP-GlcNAc, having to form GlcNAc-PI, which undergoes de-N-acetylation to form GlcN-PI. The R F values of the [ 3 H]mannosylated glycolipids from the exogenous acceptor (Fig. 2C) coincided with those of the endogenously primed glycolipids (Fig. 2B), suggesting that not only do they have the same mannosylated glycan head group but also the total lipid hydrophobicity of the endogenous glycolipids must be similar to those formed from the exogenous synthetic acceptor D-GlcN␣1-6-D-myoinositol-1-HPO 4 -sn-1,2-dipalmitoylglycerol.
Preincubation of the cell-free system with amphomycin and CaCl 2 prevented the formation of Dol-P-[ 3 H]Man from GDP-[ 3 H]Man and in turn prevented the formation of any [ 3 H]mannosylated GPI intermediates (data not shown). This suggests that at least the first mannosyltransferase is Dol-P-Man-dependent, consistent with other GPI pathways where the man-  nosyltransferases utilize Dol-P-Man as their mannose donor (30,50,52). The addition of UDP-GalNAc to the cell-free system containing GDP-[ 3 H]Man, GlcN-PI, and Pam-CoA gave rise to two further [ 3 H]mannosylated glycolipids, III and VI (Fig.  2D). Both of these glycolipids were sensitive to both GPI-PLD and PI-PLC (Table 1), showing that they are not inositol-acylated GPI intermediates; treatment with JB␣M showed glycolipid III was resistant, although glycolipid VI was sensitive. The desalted 2,5-anhydromannitol-containing glycan head groups obtained from the glycolipids III and IV were analyzed by Bio-Gel P4 gel filtration before and after ␣-mannosidase and ␤-N-acetylhexaminidase (Table 1). The neutral glycans obtained from both glycolipids III and VI had a size of 5.9/6.0 Gu and were sensitive to ␣-mannosidase and ␤-N-acetylhexaminase, giving glycan products of 4.4 and 4.2 Gu, respectively, corresponding to GalNAc␤1-4Man␣1-4anhydromannitol and Man␣1-2Man␣1-6Man␣1-4anhydromannitol. Thus, the original neutral glycan obtained from both glycolipids III and IV is Man␣1-2Man␣1-6(GalNAc␤1-4)Man␣1-4anhydromannitol. The difference in R F of these glycolipids and the resistance of III but sensitivity of VI to ␣-mannosidase suggest that the former structure has an ethanolamine phosphate group on the terminal mannose; thus the structures of glycolipid III and VI are EtNP-Man 2 (GalNAc) ManGlcN-PI and glycolipid VI Man 2 (GalNAc)ManGlcN-PI, respectively.

Inhibition of Inositol Acylation and Inositol Deacylation in the T. gondii Cell-free System-Washed
However, when the alternative serine protease inhibitor DFP was preincubated with the cell-free system, different [ 3 H]mannosylated intermediates are observed (Fig. 3B) Table 1 (Table 1). These results are consistent with all three glycolipids having a Man␣1-2Man␣1-6Man␣1-4anhydromannitol head group, but glycolipid CЈ was resistant to JB␣M, suggesting the presence of an ethanolamine-phosphate group on the non-reducing Man. These data taken together with their differences in R F suggest that they are glycolipids similar to T. brucei aM3, CЈ, and lyso-aM3. Because all of the mannosylated glycolipids were inositol-acylated, this implies that DFP does not inhibit inositol acylation but inhibits inositol deacylation of the mannosylated GPI intermediates. To investigate this further, a cell-free system experiment was carried out to see if the same inhibition of inositol deacylation would occur if the Toxoplasma membranes were preincubated with GDP-[ 3 H]Man, GlcN-PI, and Pam-CoA for 5 min before the addition of DFP (Fig. 4A). The same three [ 3 H]mannosylated glycolipids, aM3, CЈ, and lyso-aM3, were observed also when preincubated with DFP (Fig. 3B). When a similar experiment was conducted with PMSF instead of DFP, the same [ 3 H]mannosylated glycolipids were observed (Fig. 4B), indicating that both PMSF and DFP had prevented inositol deacylation of the glycolipids aM3 and CЈ to form the glycolipids M3 and AЈ.

4.1/4.2 Gu
Inhibition of Inositol deacylation Prevents Further Processing of Mannosylated Intermediates-A possible consequence of not being able to inositol-deacylate these intermediates, i.e. (aM3, CЈ) may prevent further processing, i.e. the addition of the Glc-GalNAc side chain. This hypothesis was confirmed by preincubating Toxoplasma membranes with GDP-[ 3 H]Man, GlcN-PI, and Pam-CoA for 5 min before the addition of either PMSF or DFP followed by the addition of either UDP-GalNAc or UDP-GalNAc and UDP-Glc (Fig. 4C). The resulting inositolacylated glycolipids formed (aM3 and CЈ) did not undergo further modification by the addition of GalNAc or GalNAc and Glc as compared with no PMSF present (Figs. 2, D and E).
In the presence of DFP (Fig. 5B), in vivo labeling with [ 3 H]GlcN showed no significant difference from the control (Fig. 5A). However, in the presence of PMSF, only one [ 3 H]glycolipid was observed (Fig. 5C). Characterization of this [ 3 H]glycolipid and its corresponding neutral glycan head group as well (Table 1) suggests that the glycolipid could be Man 3 GlcN-   (Table 1); see Fig. 3 for identity of bands. NOVEMBER 2, 2007 • VOLUME 282 • NUMBER 44 Lipid Structure of T. gondii Free GPIs and GPI-anchored Proteins-Purified samples of free GPIs and GPI-anchored proteins, previously shown to be free of phospholipids (54), were treated with nitrous acid causing deamination and the release of the GPIno portion of the GPI structures. The GPIno moieties were extracted into butan-1-ol and analyzed by negative ion ES-MS. Several GPIno species were clearly observed when analyzed by ES-MS-MS using parent ion scanning of the collisioninduced daughter ion, inositol 1-2 cyclic phosphate ion (m/z 241). The GPIno species released from both free GPIs and GPIanchored proteins (Figs. 6, A and B, respectively) were very similar and were diacylglycerol species. Collision-induced dissociation daughter ion spectrum of the two major ions at m/z 835 (Fig. 6C) and m/z 861 (Fig. 6D)
Synthesis of glucosylated glycolipids is mediated by uridinediphosphate-glucose (16). The direct donor for the GalNAc moiety has not been defined so far, although the involvement of a lipid intermediate has not been ruled out. 4 Using hypotonically permeabilized T. gondii tachyzoites, the topology of the 4 M. Hyams and R. T. Schwarz, unpublished information.  free GPIs within the endoplasmic reticulum membrane was recently investigated (49). They demonstrate that the higher mannosylated and side chain (Glc-GalNAc)-modified GPI intermediates are preferentially localized on the cytoplasmic leaflet of the ER (49). A new early intermediate with an acyl modification on the inositol was identified indicating that inositol acylation also occurs in T. gondii. However, many details of the biosynthesis of Toxoplasma GPIs remained unresolved.
Here we provide a detailed study of the biosynthesis of T. gondii GPIs with an emphasis on the roles of inositol acylation and subsequent inositol deacylation.
The data presented in this paper support the following conclusions about the GPI biosynthetic pathway in T. gondii. (a) Inositol-acyltransferase acts on GlcN-PI to form GlcN-(acyl)PI. (b) Inositol acylation is a prerequisite for mannosylation. (c) PMSF inhibits both the inositol acylation and the inositol deacylation. (d) DFP does not inhibit inositol acylation but does inhibit inositol deacylation. (e. Inositol deacylation is not a prerequisite for ethanolamine-phosphate addition; hence, the possible formation of glycolipid CЈ. (f) Inositol deacylation is a prerequisite for the addition of the GalNAc side chain to the first mannose of the GPI intermediates Man 3 or AЈ. (g) Inositol deacylation maintains an equilibrium that favors non inositolacylated over inositol-acylated tri-mannosylated GPI intermediates. (h) Formation of mature lyso-GPI intermediates is either an initial step of remodeling of the sn-2 position with a C18:1 fatty acid or the initial stage of catabolism.
These features (described in detail below) allow the formation of a model for the T. gondii GPI biosynthetic pathway (Fig.  7). Putative T. gondii GPI biosynthetic genes for steps 1 (PIG-A, PIG-C, and GPI1), 2 (PIG-L), 3 (PIG-W), 4 (PIG-M), 5 (PIG-V), 6 (PIG-B), 7 (PIG-O and PIG-F), and 11 (GPI8 and GAA-1) and for the generation of Dol-P-Man (DPM1) were identified by searching the Toxoplasma Genome for homologues of known P. falciparum, human, and yeast GPI biosynthetic genes ( Table  2). No homologous for the other mammalian genes implicated in the first step (PIG-H, PIG-P, and DPM2) could be found.
The T. gondii cell-free system utilizes the multiprotein Glc-NAc transferase (Step 1, Fig. 7 and Table 2) and the GlcNAc-PI de-N-acetylase (Step 2, Fig. 7 and Table 2) to form GlcNAc-PI and GlcN-PI products, respectively. The addition of either ATP and CoA or Pam-CoA allows inositol acylation of GlcN-PI, resulting in GlcN-(acyl)PI (Step 3, Fig. 7 and Table 2). This inositol acylation in the Toxoplasma cell-free system, as in P. falciparum, S. cerevisiae, and mammalian cells, is mediated by acyl-CoA, with a preference for a C16 chain length and precedes mannosylation (31,34,36,53). However, in contrast to other acyl-CoA-dependent inositol acyltransferases, the T. gondii inositol acyltransferase is inhibited by PMSF. In T. brucei, inositol acyltransferase is also inhibited by PMSF, although it is acyl-CoA-independent, and the acylation takes place postmannosylation (25,28). Thus, the T. gondii inositol acyltransferase has features in common with both the T. brucei inositol acyltransferase, i.e. PMSF-sensitive (25), and the mammalian, yeast, and Plasmodium inositol acyltransferases, i.e. acyl-CoAdependent and utilizes GlcN-PI (34,36,53). This suggests that T. brucei and T. gondii both have an activated serine that is absent in the other homologues. The putatively identified T. gondii inositol acyltransferase (PIG-W) encodes a 558amino acid protein and is more similar to homologous from Schizosaccharomyces pombe (ϳ60% similarity), Cryptococcus neoformans, and Candida albicans than the human PIG-W (supplemental Fig. 1). Unfortunately, no homologue could be identified in the T. brucei genome.
The observation that preincubation with PMSF inhibits T. gondii inositol acylation indicates that inositol acylation is required before mannosylation (Step 4, Fig. 7 and Table 2). This is in contrast to T. brucei where inhibition of inositol acylation does not prevent mannosylation but prevents further postmannosylation processing, i.e. the addition of the ethanolamine phosphate to the third mannose, to form glycolipid CЈ, which in turn is a prerequisite for inositol deacylation to glycolipid AЈ (28).
Unlike Plasmodium, the T. gondii mature cell-surface GPIs and GPI-anchored proteins are not inositol-acylated, suggesting that the inositol acylation is only transient. Thus, inositol deacylation must take place at some point after mannosylation (Steps 4 -6, Fig. 7 and Table 2). Thus, one could speculate that acylation and deacylation of the inositol ring of T. gondii GPI intermediates play a role in controlling further processing as well as ethanolamine-phosphate addition in T. brucei as described above (28). Preincubation of the T. gondii cell-free system with a different serine protease inhibitor DFP reveals the presence of previously unseen inositol-acylated and man-nosylated T. gondii GPI intermediates i.e. aM3, CЈ, suggesting that DFP does not inhibit inositol acylation but inhibits the inositol deacylation of mannosylated GPI intermediates. These mannosylated intermediates are similar to those described before in the T. gondii cell-free system (16, 39 -41), except they are inositol-acylated, i.e. M3 becomes aM3, and AЈ becomes CЈ.
These inositol-acylated and mannosylated intermediates also accumulate when the T. gondii cell-free system is preincubated with the necessary substrates before the addition of PMSF, causing inositol deacylation inhibition. This leads to the conclusion that inositol deacylation occurs after the addition of all three mannoses, but it is not a prerequisite for ethanolamine phosphate addition (Step 7, Fig. 7 and Table 2); hence, the formation of glycolipid CЈ.
Inositol deacylation of mannosylated GPI intermediates in T. brucei is a prerequisite for fatty acid remodeling and is also inhibited by DFP but not PMSF (25,28). Yeast and mammalian inositol deacylases normally act on mature GPIs or GPI-anchored proteins and are not inhibited by serine protease inhibitors, suggesting a similarity between the T. gondii and T. brucei inositol deacylases; however, no obvious homologue to the T. brucei inositol deacylase (55) could be identified from the T. gondii genome.
Further processing of the GPI intermediate was then investigated to ascertain if the side-chain addition of GalNAc and subsequent Glc (Steps 9 and 10, Fig. 7 and Table 2) could take place on the inositol-acylated species. After initiating the GPI cell-free system, PMSF was added to inhibit inositol acylation followed by the addition of either UDP-GalNAc or UDP-Gal-NAc and UDP-Glc. No further modification to the inositolacylated glycolipids was observed, suggesting the GalNAc transferase does not act on inositol-acylated species. The inhibition of inositol deacylation preventing further processing of GPI intermediates was also observed in vivo in the presence of PMSF, but not DFP, as determined by the accumu-lation of Man 3 GlcN-(acyl)PI. The reason why no inhibition was observed with DFP is unclear but may simply be due to insufficient accessibility or being targeted by other unrelated serine protease-like enzymes elsewhere in the cell. The accumulation of Man 3 GlcN-(acyl)PI in the presence of PMSF may suggest either inhibition of the ethanolamine phosphate transferase or more likely inositol deacylation is a prerequisite for ethanolamine-phosphate addition. This is not so defined in the cellfree system by the very nature of the somewhat scrambled and disrupted membranes. Similar results were observed where in vivo inhibition by PMSF of T. brucei inositol acylation prevents ethanolamine-phosphate addition leading to an accumulation of Man 3 GlcN-PI (25).

TABLE 2
Identification of the T. gondii genes for the GPI biosynthetic pathway using homologues from P. falciparum, human, and yeast has been observed in the GPI pathways of other organisms (20,21,33). Thus, the formation of lyso-mature GPI intermediates (Step 11, Fig. 7 and Table 2) may be an initial step in a catabolic pathway to remove/recycle excess GPI anchors, as suggested for T. brucei (25).
The characterization of the GPIno moieties of the GPI anchors explains the almost identical R F values observed for the endogenously (UDP-GlcNAc) primed GPI intermediates, with a total hydrophobicity of C34:1 or C36:2, compared with priming with the exogenous synthetic GlcN-PI containing sn-1,2dipalmitoylglycerol (C32:0) (compare Figs. 2, B and C).
This clearly shows for the first time that the previously observed differences in the R F of T. gondii GPI intermediates formed by a cell-free system and those formed by an in vivo labeling (40,44,49) are not due to a difference in the lipid moiety, leaving the obvious conclusion that in an in vivo labeling there is an extra, as yet uncharacterized labile component attached to the glycan core that is causing the observed lower R F . The identification of this elusive component is presently being investigated.
To conclude, the role of inositol acylation in the T. gondii GPI pathway (Fig. 7) is to ensure full mannosylation of GPI intermediates before the addition of the GalNAc side chain on the first mannose. Together with the recent T. gondii topology study (49) and similar topology models for GPI biosynthetic pathways in other organisms (Ref. 42; for review, see Refs. 10, 50, and 51), a clear topological pathway can be envisioned. Early GPI intermediates are formed on the cytosolic face of the ER, after which translocation to the luminal side of the ER and inositol acylation facilitates mannosylation via the three Dol-P-Man-dependent mannosyltransferases. Inositol deacylation and translocation of the trimannosylated GPI intermediate to the cytosolic face of the ER allows the addition of the GalNAc and subsequent Glc side chain via the UDP-GalNAc-and UDP-Glc-dependent transferases. The addition of ethanolamine phosphate to the third mannose seems to be independent of inositol acylation and may occur at either face of the ER. The mature GPI intermediates must then translocated to the luminal side of the ER for attachment to protein via the transamidase followed by translocation to the cell surface.