Removal or Maintenance of Inositol-linked Acyl Chain in Glycosylphosphatidylinositol Is Critical in Trypanosome Life Cycle*

The protozoan parasite Trypanosoma brucei is coated by glycosylphosphatidylinositol (GPI)-anchored proteins. During GPI biosynthesis, inositol in phosphatidylinositol becomes acylated. Inositol is deacylated prior to attachment to variant surface glycoproteins in the bloodstream form, whereas it remains acylated in procyclins in the procyclic form. We have cloned a T. brucei GPI inositol deacylase (GPIdeAc2). In accordance with the acylation/deacylation profile, the level of GPIdeAc2 mRNA was 6-fold higher in the bloodstream form than in the procyclic form. Knockdown of GPIdeAc2 in the bloodstream form caused accumulation of an inositol-acylated GPI, a decreased VSG expression on the cell surface and slower growth, indicating that inositol-deacylation is essential for the growth of the bloodstream form. Overexpression of GPIdeAc2 in the procyclic form caused an accumulation of GPI biosynthetic intermediates lacking inositol-linked acyl chain and decreased cell surface procyclins because of release into the culture medium, indicating that overexpression of GPIdeAc2 is deleterious to the surface coat of the procyclic form. Therefore, the GPI inositol deacylase activity must be tightly regulated in trypanosome life cycle.

The protozoan parasite Trypanosoma brucei is coated by glycosylphosphatidylinositol (GPI)-anchored proteins. During GPI biosynthesis, inositol in phosphatidylinositol becomes acylated. Inositol is deacylated prior to attachment to variant surface glycoproteins in the bloodstream form, whereas it remains acylated in procyclins in the procyclic form. We have cloned a T. brucei GPI inositol deacylase (GPIdeAc2). In accordance with the acylation/deacylation profile, the level of GPIdeAc2 mRNA was 6-fold higher in the bloodstream form than in the procyclic form. Knockdown of GPIdeAc2 in the bloodstream form caused accumulation of an inositol-acylated GPI, a decreased VSG expression on the cell surface and slower growth, indicating that inositol-deacylation is essential for the growth of the bloodstream form. Overexpression of GPIdeAc2 in the procyclic form caused an accumulation of GPI biosynthetic intermediates lacking inositol-linked acyl chain and decreased cell surface procyclins because of release into the culture medium, indicating that overexpression of GPIdeAc2 is deleterious to the surface coat of the procyclic form. Therefore, the GPI inositol deacylase activity must be tightly regulated in trypanosome life cycle.
African trypanosomes (Trypanosoma brucei) are digenetic protozoan parasites, which have a dense coat of glycosylphosphatidylinositol (GPI) 5 -anchored proteins. The bloodstream form, which causes sleeping sickness in human, is covered by variant surface glycoproteins (VSG), whereas the procyclic form, which resides in the tsetse fly, is covered by procyclins (reviewed in Refs. [1][2][3][4]. T. brucei escapes the immune response of the mammalian host by sequentially expressing structurally different forms of VSG (reviewed in Refs. 1 and 4). Several lines of evidence indicate that VSG coat is critical even in culture. First, knock-out of TbGPI10 or TbGPI12 genes, which are required for the biosynthesis of GPI, can be achieved only when extra TbGPI10 or TbGPI12 genes, respectively, is introduced and those knock-out bloodstream forms cannot lose corresponding episomal genes (5,6). Second, RNAi-mediated inhibition of the expression of the TbGPI8 gene, which encodes a component of GPI transamidase required for the attachment of GPI to VSG, results in cell death with a block in cytokinesis (7). Finally, inhibition of VSG synthesis by RNAi triggers cell cycle checkpoint and causes blocking of cell division at a precytokinesis stage, indicating that VSG protein or transcript is monitored during cell cycle (8).
In the tsetse fly gut, procyclins are thought to protect the parasites from digestion (reviewed in Ref. 4). There are two classes of procyclins, glutamic acid-proline repeat (EP)-procyclins and pentapeptide repeat (GPEET)-procyclins. When EP-procyclin genes were knocked out, the ability of the procyclic form to establish a heavy infection in the tsetse fly was severely affected (9). The infectivity to tsetse fly was almost completely lost when the TbGPI8 gene was disrupted (7,10). These results indicate that procyclins are important during the infection of the tsetse fly. On the surface of procyclic forms, in which all procyclin genes or TbGPI8 gene are disrupted, free GPI molecules are expressed (7,10,11). These free GPI molecules may be important for the viability of these knock-out strains in culture (7,11). Thus, GPI-anchored proteins are important for the infection and survival of trypanosomes in both life cycle stages (5,(12)(13)(14).
The structures of GPI anchors are different between the bloodstream and procyclic forms (reviewed in Refs. 15 and 16) (see Fig. 1). One prominent difference between the GPI structures of procyclins and VSG is that the inositol moiety of procyclin GPI is modified by a fatty acid, whereas inositol is not acylated in the VSG GPIs (17)(18)(19)(20). Consistent with these structurally different GPIs, the GPI biosynthesis pathways in bloodstream and procyclic forms are different. In both stages, inositol can be acylated during GPI biosynthesis once the first mannose is attached. The inositol acylation is reversible such that the acylated and non-acylated GPI intermediates are in dynamic equilibrium and is a prerequisite for the addition of phosphoethanolamine to the third mannose (19, 21) (see Fig. 1). In the procyclic form, the inositol remains acylated down to PP1, the final precursor to be attached to proteins (see Fig. 1). In contrast, in the bloodstream form, inositol is acylated up to CЈ, but the fatty acid is then removed prior to the attachment of GPI to a VSG (19) (see Fig. 1). The significance of inositol acylation and deacylation during GPI biosynthesis and the physiological significance of the structural difference between procyclins and VSG are not well understood.
A GPI inositol deacylase (GPIdeAc) gene has been cloned, and the enzyme activity of an affinity-purified recombinant GPIdeAc has been * This work was supported by grants from the Ministry of Education, Culture, Sports, Science and Technology of Japan and the Core Research for Evolutional Science and Technology, Japan Science and Technology Agency. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Fig. 1 demonstrated (22). Furthermore, the GPIdeAc knock-out resulted in accumulation of inositol-acylated GPI biosynthetic intermediates. However, some GPI inositol deacylase activity remained in the knock-out mutant, suggesting that at least one other inositol deacylase is present (22,23). Here, we have identified a second T. brucei GPI inositol deacylase (GPIdeAc2) based on sequence homology to mammalian GPI inositol deacylase (PGAP1) (24) and shown that GPIdeAc2 is a major GPI inositol deacylase. Our data suggested that the regulation of GPIdeAc2 expression is critical in the life cycle of T. brucei.

EXPERIMENTAL PROCEDURES
Cell Culture-Strain 427 procyclic and bloodstream clone 221a T. brucei were used. Single marker bloodstream-form cell (a gift from Dr. G. A. M. Cross, Rockefeller University) (25), co-expressing T7 polymerase, and the Tet repressor, were used for conditional expression of the GPIdeAc2 double-stranded RNA. Bloodstreamform and procyclic-form cells were cultured in HMI-9 medium and SDM-79 medium, respectively, with appropriate drug selection.
RNAi and Transfection-To generate the plasmid p2T7 TAblue / GPIdeAc2, a 536-bp fragment, corresponding to the C-terminal onefifth of the GPIdeAc2 gene was PCR-amplified using the primers 5Ј-GGATAATACCGGTCACTGTGTTTGG-3Ј and 5Ј-CAGCACCC-AAACAATACGAAAAGAC-3Ј and cloned into Eam1105I site of the tetracycline-inducible RNAi vector p2T7 TAblue (26) (a gift from Dr. D. Horn, London School of Hygiene and Tropical Medicine). The sequence of the 536-bp region had no significant homology with other sequences in T. brucei genome data base. For transfection into the bloodstream form via integration into an rDNA spacer region, 2 ϫ 10 7 log phase cells were electroporated with 50 g of NotI-linearized p2T7 TAblue /GPIdeAc2 plasmid in a total volume of 0.4 ml of cytomix buffer (27). The transfected cells were then selected in a medium containing hygromycin (2.5 g/ml) exactly following a published procedure (26). RNAi was induced by incubation with doxycycline (1 g/ml) for 48 h. To express GPIdeAc2 in the procyclic form, we cloned GPIdeAc2 into the HindIII-and BamHI-cut expression vector pPPMCS, which was constructed from pHD590 (28) by replacing its promoter and luciferase gene with the normal procyclic acidic repetitive protein promoter and the multiple cloning sites (10). After transfection, transfectant cell line was obtained by serial dilution after selection with hygromycin (50 g/ml).
Northern Blot Analysis and Real Time PCR-For the Northern blot analysis, total RNA was isolated using an RNeasy kit (Qiagen). Filters were hybridized with a probe corresponding to the 536-bp fragment used for the GPIdeAc2 RNAi construct. A fragment of the ␣-tubulin gene was used as a loading control. For real time PCR analysis, total RNA was isolated using TRIzol (Invitrogen), reverse-transcribed using SuperScript III RT (Invitrogen), and analyzed using a QuantiTect SYBR Green PCR kit (Qiagen) and ABI sequence detection systems (ABI PRISM 7000).
GPI Biosynthesis Analysis-Hypotonic cell lysate was prepared as described previously, except that the tunicamycin treatment was omitted (29). Frozen cell lysate was thawed and washed twice with 10 volumes of HKMTL buffer (50 mM HEPES (pH 7.4), 25 mM KCl, 5 mM MgCl 2 , 0.1 mM N␣-p-tosyl-L-lysine chloromethyl ketone, and 1.0 g/ml leupeptin) by centrifugation (6,000 ϫ g, 10 min, 4°C). The membrane pellet was suspended at 10 9 cell equivalents/ml with HKMTL buffer Steps from PI through M3-acyl to CЈ (PP3) are common in both forms. In the procyclic form, PP3 is converted to PP1, a lyso-form of PP3, and is transferred to procyclins. In the bloodstream form, CЈ (or lyso-CЈ) is converted to AЈ (or ) by inositol deacylation and then is subjected to fatty acid remodeling before transfer to VSG. supplemented with 5 mM MnCl 2 , 1 mM dithiothreitol, and 0.8 g/ml tunicamycin. To radiolabel GPI mannose residues, the membranes were incubated with GDP-[3,4-3 H]Man (DuPont, 20 Ci/ml) and 1 mM UDP-GlcNAc for 60 min at 37°C (bloodstream form) or 27°C (procyclic form). The reaction was terminated by adding CHCl 3 /CH 3 OH (1:1, v/v) to give a final ratio of CHCl 3 /CH 3 OH/H 2 O (10:10:3, v/v/v). The lipids were extracted for 1 h, and insoluble debris was removed by centrifugation. The supernatants were dried under a stream of nitrogen and lipids were resuspended by adding 150 l each of n-butanol-saturated water and water-saturated n-butanol. After centrifugation, the organic upper phase was saved, and the lower aqueous phase was re-extracted with 100 l of water-saturated n-butanol. The combined butanol phase was back-extracted with 100 l of n-butanol-saturated water, and the final butanol phase was then dried in a SpeedVac concentrator. Dried lipids were resuspended in 20 l of CHCl 3 /CH 3 OH/H 2 O (10:10:3, v/v/v) for application to TLC on high performance TLC silica gel 60 (Merck). TLC plates were developed with the same solvent. For autoradiography, the plates were exposed to an image plate and developed by an image analyzer (BAS-2500, Fuji).
Microscopy and Flow Cytometry-For Hoechst staining, 1 ϫ 10 6 cells were fixed overnight in methanol at Ϫ20°C. Cells were washed with phosphate-buffered saline and resuspended with 1 ml of phosphatebuffered saline containing 0.5% bovine serum albumin and 1 g/ml Hoechst 33258 (Molecular Probes). After centrifugation, cells were washed again with phosphate-buffered saline, placed on glass slides, and mounted with PermaFluor solution (Immunon, Pittsburg, PA). For cell cycle analysis using propidium iodide staining and FACS, we followed exactly the published method (30). FACS analysis of EP-procyclins was also performed as described previously (5).
Pulse-chase Analysis of Procyclins-Procyclic cells were washed in phosphate-buffered saline and were resuspended at 10 8 cells/ml in SDM-79 without proline supplemented with 10% (v/v) dialyzed fetal bovine serum. [ 14 C]Proline (Amersham Biosciences) was added (10 Ci/ml) and incubated for 30 min at 27°C. The chase was initiated by 10-fold dilution with prewarmed complete SDM-79 medium containing 10% (v/v) fetal bovine serum and continued for 19 h. At each time point, 1.0 ml of culture was centrifuged, and the supernatant and the cell pellet were separated. Supernatants were filtered through a 0.2-m pore membrane to eliminate residual parasites. Cell pellets were solubilized in 1.0 ml of buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 5 mM EDTA, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS, and protease inhibitor mixture (31). These samples were precleared by mixing with 20 l of 50% suspension of protein G-Sepharose and gently rotating for 1 h at 4°C. After centrifugation at 5,000 ϫ g for 10 min, EP procyclins were immunoprecipitated by incubating with 10 g of anti-EP procyclin monoclonal antibody (Cedarlane Lab-oratories, Ontario, Canada) or an isotype-matched control monoclonal antibody for 1 h at 4°C, followed by mixing with 20 l of a 50% suspension of protein G-Sepharose and gently rotating for 1 h at 4°C. EP procyclins were analyzed by 10 -20% gradient gel SDS-PAGE and autoradiography.
Trans-sialidase Assay-Trans-sialidase activity was assessed by measuring hydrolysis of a fluorogenic substrate, 2Ј-(4-methylumbelliferyl)-D-N-acetylneuraminic acid (Sigma) (32). In brief, the enzyme was incubated with 0.5 mM substrate in 20 mM HEPES (pH 7.2) at 28°C in a final volume of 20 l. The reaction was terminated by adding 200 l of 0.2 M Tris-HCl (pH 9.5), and the fluorescence was measured with a Fluoroskan II microplate reader (Labsystems, Flow Laboratories Inc.). One unit of enzyme is defined as the activity to hydrolyze 1 mol of the substrate in 1 min.

RESULTS
Cloning the GPIdeAc2 Gene Homologous to Mammalian GPI Inositol Deacylase-We used the amino acid sequence of rat PGAP1 to identify a gene encoding a GPI inositol deacylase in the T. brucei genome data base (BLAST). We found an open reading frame Tb927.3.2610 with BLAST search and amplified its homologue with 99% nucleotide sequence identity from T. brucei 427 genomic DNA. We named this gene GPIdeAc2 (DDBL/GenBank/EMBL accession number AB219559). The GPIdeAc2 open reading frame encoded a protein of 815 amino acids, which had an ϳ50-amino-acid extension at the N terminus compared with PGAP1 (supplemental Fig. 1). The yeast PGAP1 homologue Bst1p had a similar N-terminal extension compared with PGAP1 (24). There is a second methionine at position 35 within the extension in GPIdeAc2. We chose the first methionine rather than the second as the initiation site because the sequences of the extensions in GPIdeAc2 and Bst1p had some homology (sequence not shown). GPIdeAc2 had 20.3% amino acid identity to rat PGAP1 (supplemental Fig. 1) and 20.6% to Bst1p. An alignment of the PGAP1 and GPIdeAc2 sequences showed a conserved esterase/lipase/thioesterase motif with a catalytic serine (InterPro IPR00379) (supplemental Fig. 1). Hydropathy profiles of GPIdeAc2 and PGAP1 were similar, having a large hydrophilic region with the catalytic serine followed by multiple transmembrane regions (data not shown).
Transcript Levels of GPIdeAc2 in Procyclic and Bloodstream Forms-We compared the mRNA levels of GPIdeAc2 in procyclic and bloodstream forms using Northern blots and real time PCR. A 2.4-kb transcript, which corresponded to the length of the GPIdeAc2 gene, was expressed at a six times higher level in the bloodstream form than in the procyclic form ( Fig. 2A). Consistent with this result, the mRNA level of GPIdeAc2 determined using real time PCR was six times higher in the bloodstream form than in the procyclic form (Fig. 2B). The high expression of GPIdeAc2 mRNA in the bloodstream form coincides with the deacylated structure of VSG GPI in the bloodstream form, whereas the low expression of GPIdeAc2 mRNA in the procyclic form coincides with the usage of acylated inositol in procyclin GPI. Thus, the transcript profile of GPIdeAc2 is consistent with the inositol acylation and deacylation status of GPI.
Acylated GPI Intermediates Accumulated in the Presence of GPIdeAc2 RNAi in the Bloodstream Form-To determine whether GPIdeAc2 mediates GPI inositol deacylation in the bloodstream form, we used RNAi to reduce the expression of GPIdeAc2. After induction of RNAi by doxycycline for 48 h, we examined the mRNA levels of GPIdeAc2 and ␣-tubulin (a loading control) by Northern blot analysis. The induction of RNAi caused a ϳ70% decrease in the GPIdeAc2 mRNA level (Fig. 3A).
Then, we examined whether the suppression of GPIdeAc2 affected the synthesis of GPI precursors. After induction of RNAi for 48 h, cell lysates were prepared and incubated with GDP-[ 3 H]mannose to label GPI biosynthetic intermediates. Lipid extracts were treated with GPI-PLD and PI-PLC prior to TLC. GPI-PLD cleaves all GPI intermediates (data not shown). PI-PLC cleaves GPI intermediates without inositol acylation, whereas it does not cleave inositol-acylated GPI. One distinctive band was seen in GPIdeAc2 knockdown cells (Fig. 3B, lane 7, arrow) and its Rf value was consistent with glycolipid CЈ. This species was sensitive to GPI-PLD (lane 10) but resistant to PI-PLC (lane 8), indicating that it was inositol-acylated GPI. It was also resistant to JB␣M digestion, indicating that the non-reducing terminal mannose was protected by an ethanolamine phosphate group (Fig. 3C, lanes 7 and 8,  arrow). Based on these results, we concluded that the accumulated GPI was indeed glycolipid CЈ (EtNP-Man 3 GlcN-(acyl)-PI) (see Fig. 1). Concomitant with glycolipid CЈ accumulation, the amount of glycolipid AЈ (EtNP-Man 3 GlcN-PI) decreased to ϳ50% of control in the GPIdeAc2 knockdown cells (Fig. 3B, compare lane 5 with 7). Furthermore, the amount of glycolipid was also decreased, and the inositol-acylated form of (termed lyso-CЈ) (19, 33) accumulated (Fig. 3B, lanes 7 and 8,  arrowhead). These results suggested that knockdown of GPIdeAc2 resulted in slower conversion of glycolipid CЈ and lyso-CЈ to AЈ and , demonstrating that GPIdeAc2 is involved in GPI inositol deacylation in the bloodstream form.
It was also noted that RNAi of GPIdeAc2 caused a nearly complete loss of one glycolipid (double asterisk) and a significant decrease of the other (asterisk) (Fig. 3, B and C, lane 5 versus 7). These fast moving glycolipids may be non-inositol-acylated GPI bearing three or two mannoses.
GPIdeAc2 Is Important for the Surface Expression of VSG and Optimal Growth of the Bloodstream Form-We examined the expression of VSG on the surface of the GPIdeAc2 knockdown cells by flow cytometric analysis. Control cells (empty vector transfectants) maintained surface VSG expression after a 48-h induction, whereas GPIdeAc2 knockdown cells showed a 65 Ϯ 15% decrease in the surface expression of VSG, indicating GPIdeAc2 is important for the efficient surface expression of VSG (Fig. 4A).
We then examined the growth of the GPIdeAc2 knockdown cells. Cell growth was monitored in the absence or presence of doxycycline for 5 days. GPIdeAc2 knockdown caused a severe inhibition of cell growth, indicating that GPIdeAc2 was essential for optimal growth of the bloodstream form (Fig. 4B). Many morphologically abnormal cells were seen in the GPIdeAc2 knockdown population. They had multiple flagella and multiple nuclei (Fig. 4C). Cell cycle analysis showed that ϳ60% of cells were at a stage of 6n or higher (Fig. 4D).

Deacylated GPIs Were Produced in the GPIdeAc2
Overexpressing Procyclic Form-As shown in Fig. 2, the mRNA level of GPIdeAc2 was low in the procyclic form consistent with the structural characteristics of procyclin GPI anchors. We next asked whether maintenance of low GPIdeAc2 expression is required for inositol-acylated GPI anchors of procyclins. We stably overexpressed GPIdeAc2 in the procyclic form and examined GPI biosynthesis by incubating cell lysates with GDP-[ 3 H]mannose. If overexpressed GPIdeAc2 mediates GPI inositol deacy- lation, then the presence of normally inositol-acylated GPI species would be decreased, and deacylated GPI species, such as AЈ-and -like intermediates, which are not usually seen in procyclic cells, would be generated. As expected, the concentration of PP3, which is a major acylated species, was greatly decreased (Fig. 5A, lane 1 versus 3). In addition, the accumulation of two slow migrating glycolipids was seen (Fig. 5A, lane 3, arrow and arrowhead). To characterize these glycolipids, we treated them with PI-PLC (Fig. 5A), PLA 2 (Fig. 5B), JB␣M (Fig.  5C), and GPI-PLD (Fig. 5D). Both were sensitive to PI-PLC, indicating that these glycolipids were indeed deacylated (Fig. 5A, lane 4). The upper glycolipid (arrow) was sensitive to PLA 2 (Fig. 5B, lane 4) and GPI-PLD (Fig. 5D, lane 4) but not JB␣M (Fig. 5C, lane 4), indicating that it is an AЈ-like glycolipid. The other glycolipid (arrowhead) was also sensitive to PI-PLC and GPI-PLD (Fig. 5, A and D, lane 4) but resistant to JB␣M (Fig. 5C, lane 4) and PLA 2 (Fig. 5B, lanes 3 and 4, arrowhead). Therefore the lower glycolipid is -like. These results indicate that inositol deacylation occurred in the procyclic form overexpressing GPIdeAc2 and that maintaining a low expression level of GPIdeAc2 is required for inositol-acylated GPI in the procyclic form.

GPI-anchored Proteins
Were Released into the Medium from the GPIdeAc2 Overexpressing Procyclic Form-We examined whether the surface expression of procyclins is affected by overexpression of GPIdeAc2. If the GPI anchors, which contain only a single fatty acid (-like), are attached to procyclins such procyclins might be unstably anchored to the plasma membrane. If the GPI with only one fatty acid is not attached to procyclins, procyclins cannot be membrane-bound. Consistent with these possibilities, the surface EP-procyclin expression was significantly decreased in the GPIdeAc2 overexpressing procyclic form (Fig. 6A).
To examine the fate of procyclins in the GPIdeAc2-overexpressing cells, EP-procyclins in empty vector or GPIdeAc2 transfectants were pulse-labeled with [ 14 C]proline for 30 min and chased for various times up to 19 h. At each time point, we immunoprecipitated EP-procyclins from the cell lysates and the culture supernatants and analyzed them by SDS-PAGE. In the empty vector transfectant, radiolabeled EP-procyclins were efficiently chased into the mature 50-kDa band, and little was found in the culture supernatants (Fig. 6, B and D). In contrast, in the GPIdeAc2 overexpressing cells, more of the immature ER-form EPprocyclins were found in the cell lysates (Fig. 6C). Furthermore, EPprocyclins were released into the medium and degraded (Fig. 6E). These data together with flow cytometric analysis (Fig. 6A) suggested that the procyclic form lost cell surface EP-procyclins because of release into medium. The release was most likely due to an unstable association of the -like GPI anchor with the plasma membrane because EP-procyclins in the transfected cells matured at a similar rate to that of the non-transfected cells with only a minor accumulation of immature species. Taken together, overexpression of GPIdeAc2 is deleterious to the surface expression of EP-procyclins.
To see whether the expression of other GPI-anchored proteins was affected by overexpression of GPIdeAc2, we measured the transsialidase activity in the cell lysates and the culture supernatants. A higher trans-sialidase activity was found in the culture supernatant of GPIdeAc2-overexpressing cells than in that of wild-type cells (Fig.  6F, right). The secreted level of the enzyme was comparable to that of TbGPI10 knock-out cells, which secrete non-GPI-anchored transsialidase because of the hydrolysis of its GPI attachment signal by transamidase (10). TbGPI8 knock-out cells that are defective in GPI transamidase did not secrete trans-sialidase, as expected (10). These results are consistent with the pulse-chase experiments showing that EP-procyclins are unstably associated with the membrane of the procyclic form when GPIdeAc2 is overexpressed.

DISCUSSION
In the present study, we have identified a major GPI inositol deacylase, GPIdeAc2, in T. brucei. This finding enabled us to determine the roles of inositol acylation and deacylation in the trypanosome life cycle. Because the inositol-linked acyl chain is removed before fatty acid remodeling and attachment to VSG in bloodstream form, we knocked down GPIdeAc2 and analyzed its effect on the surface expression of VSG and growth of the parasite. The RNAi-mediated decrease of GPIdeAc2 mRNA by 70% caused accumulation of glycolipid CЈ and lyso-CЈ and concomitant decreases in glycolipids AЈ and (deacylated intermediates downstream to CЈ and lyso-CЈ), indicating that deacylation of inositol was significantly inhibited (Fig. 3, B and C). Under these conditions, the surface expression of VSG was severely decreased, and growth was nearly completely stopped (Fig. 4). RNAi with knockdown constructs corresponding to two other regions in GPIdeAc2 caused similar accumulation of glycolipid CЈ (data not shown). It appeared that the inhibition of inositol deacylation resulted in decreased fatty acidremodeled GPI that is competent for attachment to VSG and the decreased surface expression of VSG, which in turn caused cessation of growth. We, therefore, demonstrated that deacylation of inositol is critical for VSG-GPI anchor biosynthesis in the bloodstream form.
The effect of GPIdeAc2 knockdown on cell cycle progression (Fig. 4, C and D) was similar to that of TbGPI8 knockdown (7) in that cells became multinuclear, multikinetoplast, and multiflagellar. These knockdown phenotypes suggest that the defects in GPI biosynthesis result in the inhibition of cytokinesis while allowing cell cycle progression. In a striking contrast, VSG knockdown triggered a specific cell cycle checkpoint at 2N2K stage resulting in blocking of cell division (8).
In cells with defective GPI biosynthesis, transcription and translation of VSG continue and only post-translational modification by GPI is inhibited. It is, therefore, possible that monitoring system for VSG protein or transcript (8) cannot detect the abnormality.
In contrast to the bloodstream form, GPI anchors are inositol-acylated in the procyclic form (17,18). The mRNA level of GPIdeAc2 in the procyclic form is only ϳ15% of that in the bloodstream form (Fig. 2), suggesting that GPIdeAc2 is down-regulated during or after differentiation from the bloodstream form to the procyclic form. We forced the expression of GPIdeAc2 in the procyclic form by transfection and assessed its effect on the surface expression of EP-procyclins. Overexpression of GPIdeAc2 caused a nearly 90% decrease in the surface EP-procyclins because of secretion into culture  (Fig. 6). It was reported that the lipophosphoglycans of Leishmania major, containing lyso-alkyl-PI, are released from the cell surface (35). This result is consistent with our interpretation that the secretion of procyclins into the medium from the GPIdeAc2-overexpressing procyclic cells was because of an unstable association of procyclins bearing a single fatty acid chain. Therefore, down-regulation of GPIdeAc2 is critical for generation of the procyclin coat on the procyclic form. These findings together indicate that GPIdeAc2-mediated inositol deacylation must be tightly regulated in the trypanosome life cycle.
It was reported that knock-out of another inositol deacylase GPIdeAc led to reduced inositol deacylase activity (22). GPIdeAc consists of 558 amino acids, which is much smaller than GPIdeAc2 and PGAP1. GPIdeAc has an N-terminal signal peptide but no transmembrane domain, in contrast to GPIdeAc2 and PGAP1, which have multiple transmembrane domains. GPIdeAc is homologous to mammalian acyloxyacyl hydrolase that removes fatty acids from bacterial lipopolysaccharides. GPIdeAc has no significant sequence homology with GPIdeAc2 and PGAP1. GPIdeAc is a non-essential gene and partially accounts for inositol deacylase activity in T. brucei. If GPIdeAc also deacylates glycolipid CЈ, its activity alone is not sufficient to generate enough amount of glycolipid AЈ. Alternatively, GPIdeAc may act on other inositol-deacylation reactions. It was reported that there are two pools of glycolipid A, of which only one corresponding to 38% of total glycolipid A is competent for attachment to VSG. Disruption of the GPIdeAc gene resulted in the increase of the VSG-attachment competent pool to 75% of the total, suggesting the presence of complex regulation to VSG-anchor precursor (23).
In the GPI biosynthetic pathway of T. brucei, inositol acylation takes place only after the formation of mannosyl-glucosaminyl-phosphatidylinositol. The inositol-linked acyl chain can be removed from and added again to any of the three mannosylated intermediates, that is, GPI intermediates bearing one to three mannoses are in dynamic equilibrium between inositol-acylated and -non-acylated species (19,21,34). Overexpression of GPIdeAc2 in the procyclic form caused accumulation of two non-acylated GPI forms bearing two and three mannoses (Fig. 5A, lane 3). Conversely, knockdown of GPIdeAc2 in the bloodstream form caused a decrease in these GPI species (Fig. 3, B and C, lanes 7), indicating that these GPIs can be substrates of GPIdeAc2. It is suggested that GPIdeAc2 expressed at a low level in the procyclic form is involved in this equilibrium status.
Although GPIdeAc2 is homologous to mammalian PGAP1 sharing the same lipase motif, these two enzymes act with different timing. In trypanosomes, inositol deacylation occurs before attachment to proteins, whereas in mammalian cells, the acyl chain is removed from inositol after attachment to proteins (2,19,36). As discussed above, GPIdeAc2 acts on various GPI intermediates, suggesting that the biggest structure it recognizes is mannosyl-glucosaminyl-(acyl)-phosphatidylinositol. Because PGAP1 does not deacylate free GPIs, it may additionally recognize the amide linkage between ethanolamine and the C-terminal amino acid. In agreement with this possibility of a structurally larger recognition motif, PGAP1 has a significantly bigger luminal catalytic domain than GPIdeAc2. In conclusion, the present study revealed critical roles of GPIdeAc2 in the life cycle of African trypanosomes and demonstrated that expression of GPIdeAc2 must be tightly regulated.