Proteins with Glycosylphosphatidylinositol (GPI) Signal Sequences Have Divergent Fates during a GPI Deficiency GPIs ARE ESSENTIAL FOR NUCLEAR DIVISION IN TRYPANOSOMA CRUZI*

Glycosylphosphatidylinositols (GPIs) are membrane anchors for cell surface proteins of several major protozoan parasites of humans, including Trypanosoma cruzi, the causative agent of Chagas’ disease. To investigate the general role of GPIs in T. cruzi, we generated GPIdeficient parasites by heterologous expression of T. brucei GPI-phospholipase C. Putative protein-GPI intermediates were depleted, causing the biochemical equivalent of a dominant-negative loss of function mutation in the GPI pathway. Cell surface expression of major GPI-anchored proteins was diminished in GPI-deficient T. cruzi. Four proteins that are normally GPI-anchored in T. cruzi exhibited different fates during the GPI shortage; Ssp-4 and p75 were secreted prematurely, while protease gp50/55 and p60 were degraded intracellularly. These observations demonstrate that secretion and intracellular degradation of GPI-anchored proteins may occur in the same genetic background during a GPI deficiency. We postulate that the interaction between a protein-GPI transamidase and the COOH-terminal GPI signal sequence plays a pivotal role in determining the fate of these proteins. At a nonpermissive GPI deficiency, T. cruzi amastigotes inside mammalian cells replicated their single kinetoplast but failed at mitosis. Hence, in these protozoans, GPIs appear to be essential for nuclear division, but not for mitochondrial duplication.


Construction of pTEX.GPI-PLC and Transfection of T. cruzi
An EcoRI fragment of T. brucei GPI-PLC cDNA from plasmid pDH4 (8) was cloned into the EcoRI site of a T. cruzi episomal expression vector pTEX (14). Epimastigotes were transfected (15) with 25 g of pTEX.GPI-PLC. Following 48 h of incubation at 28°C, G418 was added to a final concentration of 100 g/ml for selection of stable transformants, which were eventually subcultured in medium containing 200 -800 g/ml G418. G418-resistant cells were studied 3 months after drug selection.

Triton X-114 Phase Partition of GPI-anchored Proteins
[ 35 S]Methionine-labeled wild-type T. cruzi amastigotes (5 ϫ 10 7 cells) were lysed at 4°C by incubation for 30 min in 100 l of Tris-buffered saline (0.1 M Tris-HCl, pH 7.5, 0.15 M NaCl) containing 2% precondensed Triton X-114 (18) and protease inhibitor mixture (11). Before lysis, parasites, when indicated, were treated with PI-PLC as described above. The lysate was incubated at 37°C for 5 min and centrifuged at 14,000 ϫ g for 1 min at room temperature to separate the aqueous (detergent-depleted) and detergent-enriched phases. Triton X-114 (12%) was added to the aqueous phase, while Tris-buffered saline was added to detergent phase to finally have 2% Triton X-114 concentration in both phases. The phase separation was repeated. Aqueous and detergent-enriched phases were pooled separately, and proteins therein precipitated (19) and resolved by SDS-PAGE/fluorography.
To isolate GPIs, [ 3 H]EtN-labeled glycolipids were resolved by high performance thin layer chromatography (HPTLC) in CMW on Silica Gel 60 plates. TcAmLP-3 (2 ϫ 10 9 cell eq) was scraped from the TLC plate and extracted thrice each time with 1 ml of CMW. The eluates were pooled, dried under a stream of nitrogen and redissolved in 100 l of n-butanol. Samples were stored at Ϫ20°C until use. [ 3 H]Man 1-3 GlcN-PI from T. brucei (ILTat at 1.3) bloodstream form were generated in a cell-free system in presence of 0.25 mM phenylmethylsulfonyl fluoride (20).
Products of cleavage were extracted with 300 l of water-saturated n-butanol, followed by two back extractions of the butanol phase with 300 l of water. Butanol-soluble products were dried under nitrogen and resuspended in 10 l of n-butanol. Products from various digestions were resolved by TLC on Silica Gel 60 in CMW (10:10:3) or CHCl 3 : MeOH:0.25% KCl (11:9:2) as indicated in figure legends and detected by fluorography.
Partial Acid Hydrolysis and Exoglycosidase Digestion-The reduced neutral [ 3 H]glycan (ϳ20,000 cpm) was resuspended in 100 l of 100 mM trifluoroacetic acid and heated at 100°C for 4 h. When needed, complete hydrolysis was carried out in 300 mM trifluoroacetic acid for 8 h. Additionally, the sealed reaction tube was monitored so that liquid which evaporated on to the cap was recovered repeatedly by microcentrifugation. After drying in a SpeedVac, the products were analyzed by HPTLC/fluorography. Reduced neutral [ 3 H]glycans were digested with jack bean ␣-mannosidase (JBAM; 2.5 units/ml, Boehringer Mannheim) in 100 l of 100 mM NaOAc, pH 5.0, 1.5 mM ZnSO 4 for 16 h at 37°C. The reaction was terminated and analyzed as described previously (23).

Flow Cytometric Analysis
T. cruzi cells were washed in PBS containing 0.1% bovine serum albumin (Sigma) and 0.1% sodium azide (PAB). For each assay, 1 ϫ 10 6 parasites were resuspended in 50 l of PAB containing purified antibody (100 g/ml) at either 1:1000 dilution or in undiluted hybridoma supernatants and kept for 30 min at 4°C. The parasites were washed with 1 ml of PAB and incubated for 30 min at 4°C in the dark in 50 l of PAB containing a 1:50 dilution of fluorescein isothiocyanate-labeled affinity-purified goat F(AbЈ) 2 anti-mouse or anti-rabbit Ig (IgGϩIgM) antibody (Southern Biotechnology Associates, Birmingham, AL). Parasites were washed, resuspended in 1 ml of PAB, and analyzed by flow cytometry on an EPICS 753 Elite cytofluorimeter (Coulter Electronics, Hialeah, FL). In some experiments, before staining, parasites (2 ϫ 10 7 ) were washed with PBS and then treated with 2 ϫ 10 Ϫ2 units of B. cereus PI-PLC.

Growth of T. cruzi Amastigotes in Mammalian Cells
CSWAE1A cells (a neomycin-resistant mouse fibroblast cell line transfected with E1A gene of adenovirus) were irradiated with 7500 Rad to stop their division, and allowed to attach to coverslips in 24-well flat-bottomed culture plates for 24 h at 37°C, 5% CO 2 in RPMI, 5% FBS containing 400 g/ml G418. The fibroblasts were infected with metacyclic trypomastigotes (20:1, parasites:CSWAE1A cells) of pTEX/T. cruzi or pTEX.GPI-PLC/T. cruzi previously cultured in 400 g/ml G418. Infected cells were stained with 1 mM SYTO 11 nucleic acid stain (Molecular Probes Inc., Eugene, OR) at 37°C for 10 min and visualized with a laser scanning confocal microscope (Bio-Rad, MRC-600).

VSG Cleavage Activities in pTEX.GPI-PLC/T. cruzi-Epi-
mastigotes (extracellular insect stage form) of T. cruzi were transfected with pTEX.GPI-PLC (see "Experimental Procedures") and selected in G418 to obtain stable transfectants, which were then converted to the different life cycle stages. Epimastigotes harboring pTEX.GPI-PLC and adapted to grow in 200 g/ml G418 expressed approximately 370 units of GPI-PLC activity/10 8 cells. Adapting the cells to grow in higher concentrations of G418 (400 or 800 g/ml) raised the level of GPI-PLC expression to 560 and 750 units (per 10 8 cells), respectively. The background level of VSG cleavage activity in control pTEX/T. cruzi epimastigotes was approximately 1.1 units/10 8 cells, representing 336 -682-fold less activity than that present in pTEX.GPI-PLC transfectants.
The possibility that decreased TcAmLP-3 and TcAmLP-4 in pTEX.GPI-PLC/T. cruzi is the result of a generalized reduction in cellular metabolism can be ruled out, because (i) the lipids migrating close to the solvent front (marked with asterisks) are present in approximately equal quantities in both wild-type and in pTEX.GPI-PLC/T. cruzi (Fig. 1A, compare lanes 1 and  2); (ii) no differences were observed between the wild-type and GPI-PLC transfectants in their [ 3 H]EtN-labeled neutral lipids extracted with chloroform/methanol (Fig. 1A, lanes 3 and 4); and (iii) approximately equal quantities of TcAmLP-5, TcAmLP-6, and TcAmLP-7, none of which was cleaved in vitro by GPI-PLC (data not shown), were detected in wild-type T. cruzi and pTEX.GPI-PLC/T. cruzi (Fig. 1). TcAmLP-7 is not identical to TcAmLP-3, because the former molecule is not cleaved by GPI-PLC in vitro under conditions when the latter is cleaved (data not presented). These observations suggest that pTEX.GPI-PLC/T. cruzi and pTEX/T. cruzi are equally competent in general utilization of [ 3 H]EtN. (We note that on occasion EtN labeling of TcAmLP-1 and TcAmLP-2 was greater than observed for TcAmLP-3 and TcAmLP-4 (see profiles in Fig. 1A and 1C). The data presented in Fig. 1A are more typical, and the occasional variability in the pattern of EtN-labeling does not alter our conclusions.) Partial Structure of TcAmLP-3-Information on the structure of TcAmLP-3 ( Fig. 1) was obtained by a combination of enzymatic and chemical cleavages. [ 3 H]EtN-labeled TcAmLP-3 was cleaved in vitro with purified PI-PLC from B. cereus and T. brucei GPI-PLC (Fig. 1D, lanes 1 and 5, respectively). TcAmLP-3 is resistant to JBAM digestion (Fig. 1D, lanes 7 and  8), indicating that the terminal mannosyl (see next sections) was blocked: Under identical conditions, the mannosyl residues of the control GPIs (Man 1-3 GlcN-PI) were cleaved by JBAM (Fig. 1D, lane 12). Treatment with phospholipase A 2 did not affect the mobility of TcAmLP-3 (Fig. 1D, lane 4). As a positive control, [ 3 H]mannose-labeled Man 1-3 GlcN-PI from T. brucei was digested with PLA 2 resulting in the loss of label (Fig. 1D, lane 10). The lyso species of Man 1-3 GlcN-PI was detected in the aqueous phase of the butanol extraction (data not shown). TcAmLP-3, as well as Man 1-3 GlcN-PI, was sensitive to base treatment (Fig. 1D, lanes 6 and 11, respectively). Thus, the inability of PLA 2 to cleave TcAmLP-3 suggests it contains a fatty acid only in the sn-1 position of a glycerolipid, or that the alkyl group is derived from ceramide. The latter possibility is excluded by the mild base sensitivity of TcAmLP-3. We infer from these properties that TcAmLP-3 is a GPI containing a phosphoglycerol backbone, which is acylated at the sn-1 position.
In summary, TcAmLP-3 consists predominantly of EtNphospho-Man 1 -GlcN-Ins-phospho-sn-1-glycerolipid. Although we did not specifically determine the location of the EtN, we presume that it is found on the single mannosyl substituent, since it is the only component of GPI-glycans known to accept the phospho-EtN (reviewed in Ref. 2).

GPI-deficient T. cruzi Express Less GPI-anchored Protein on
Their Plasma Membrane-The effect of a GPI deficiency (see Fig. 1A) on the expression of cell surface proteins was assessed by staining three developmental stages (epimastigotes, trypomastigotes, and amastigotes) of T. cruzi with antibodies raised against GPI-anchored proteins. Flow cytometric analysis showed that in all three stages, the cell surface expression of GPI-anchored proteins was decreased in the GPI-deficient strain (pTEX.GPI-PLC/T. cruzi) relative to control wild-type or pTEX/T. cruzi parasites (Fig. 3). In GPI-deficient epimastigotes stained with anti-gp50/55 (Fig. 3A) and in amastigotes stained with anti-Ssp-4 or anti-gp50/55 (Fig. 3, C and D, respectively), the decrease in expression of these GPI-anchored proteins was  4 -6), and NG-TcAmLP-3 (lanes 7-9) were subjected to partial acetolysis (lanes 2, 5, and 8, respectively), or digested with jack bean ␣ mannosidase (lanes 3, 6, and 9, respectively). Products were analyzed by HPTLC/fluorography. similar to that obtained when wild-type parasites were treated with an excess of extracellularly added phosphatidylinositolspecific phospholipase C (PI-PLC) from Bacillus cereus. Trypomastigotes of pTEX.GPI-PLC/T. cruzi also showed less cell surface expression of a trans-sialidase (evident by anti-SAPA (shed acute phase antigen) staining (Fig. 3B)). The general decrease in cell surface expression of proteins in pTEX.GPI-PLC/T. cruzi was specific for GPI-anchored molecules, as shown by staining with the anti-amastigote antibody IC10B 2 ; expression of the molecule recognized by this monoclonal antibody was similar in wild-type and GPI-deficient amastigotes (Fig. 3E).
A Select Group of Proteins Is Degraded Intracellularly in GPI-deficient T. cruzi-To investigate the fate of GPI-anchored proteins, amastigotes were metabolically labeled with [ 35 S]methionine and chased. For both cells and "chase medium," a profile of [ 35 S]methionine-labeled proteins was obtained after SDS-PAGE and fluorography.
Significant differences were observed between the major [ 35 S]methionine-labeled amastigote proteins of pTEX.GPI-PLC/T. cruzi and the wild-type control cells. Two proteins of Ϸ 60 kDa (p60 doublet, marked with arrows; Fig. 4A) present in wild-type cells at the beginning of the chase period, remained unchanged relative to other proteins (in the same lysate) during a 20-h chase (Fig. 4A, compare lanes 1, 3 and 5). The p60 doublet was absent from pTEX.GPI-PLC/T. cruzi even at the beginning of the chase (Fig. 4A, compare lanes 2, 4, and 6), and was not detected in amastigotes labeled for 15 min with [ 35 S]methionine (data not shown). Apparently, p60 is either not expressed at all, or is degraded very rapidly in the GPI-deficient cells. A 50-kDa protein, p50 (marked with asterisk, Fig.  4A) was expressed in wild-type T. cruzi at levels apparently similar to p60 and remained relatively unchanged during a 20-h chase. In pTEX.GPI-PLC/T. cruzi, p50 was present in slightly higher amount at 0 h than in wild-type cells (Fig. 4A,  lane 1 and 2). The level of this protein remained unchanged within the first 6 h (Fig. 3A, lane 4), but only Ϸ30% of the total remained after 20 h (Fig. 4A, compare lanes 2 and 6). Since the identities of p60 (Fig. 4A) and p50 were unknown and we have no antibodies specific to either protein, the possibility that they are GPI-anchored was checked by Triton X-114 phase partition before and after PI-PLC digestion. Membrane fractions from a hypotonic lysate of the parasites were studied for this purpose. The [ 35 S]methionine-labeled p60 doublet, similar to p50, was in the detergent phase prior to PI-PLC digestion (Fig. 4B, lanes 1 and 2). In contrast, both p60 and p50 partitioned into the aqueous phase after treatment with PI-PLC (Fig. 4B, lanes 3 and 4), indicating that p60 and p50 are GPI-anchored.
The chased medium was also examined for proteins not detected inside GPI-deficient T. cruzi (i.e. as compared with wild-type cells; Fig. 4A). The predominant radiolabeled proteins in the medium were 50, 56, 75, and 180 kDa (p50, p56, p75, and p180, respectively) (Fig. 5C). Interestingly, the predominant secreted proteins p75 and p180 were not the major polypeptides in the total cell lysate (compare Figs. 4A and 5C). The latter observation indicates that secretion of proteins did not occur en mass. Several other proteins of less than 45 kDa were secreted in the GPI-deficient cells (Fig. 5C, lanes 4 and 6).
In conclusion, two effects of a GPI deficiency are observed on amastigote proteins that are normally GPI-anchored, as exemplified by Ssp-4 and gp50/55: (i) a 5-fold acceleration of secretion (Fig. 5, A and B), and (ii) a 5-fold increase in the rate of intracellular degradation (Fig. 4, B and C).
GPI Deficiency Is Associated with Inhibition of Nuclear Division-The GPI deficiency in T. cruzi was associated with striking phenotypic changes. Metacyclic trypomastigotes obtained from epimastigotes that had been cultured in 200 g/ml G418 infected various mammalian fibroblast cell lines, differentiated into and replicated slowly as amastigotes, but still completed the life cycle. 3 However, transfectants cultured at 400 g/ml G418, thereby increasing the level of expression of GPI-PLC, infected mammalian cells and differentiated into amastigotes but divided only once.
GPI-deficient amastigotes in most cases replicated the kinetoplast, but failed to sustain replication of the cell nucleus, and were di-kinetoplastid (Fig. 6). Hence, GPI-deficient T. cruzi amastigotes are arrested in the cell cycle apparently in anaphase. In agreement with these observations, replication of GPI-deficient T. cruzi was severely inhibited such that a lawn of CSWAE1A cells, which is normally lysed in 4 days if infected with wild-type T. cruzi, remained intact after 10 days of infection with GPI-deficient parasites. 3

DISCUSSION
GPI Deficiency in T. cruzi Conferred by Heterologous Expression of a GPI-PLC-One of our major interests was to examine the effect of a GPI deficiency on T. cruzi. However, studies of GPI biosynthesis in T. cruzi have only recently been initiated (27)(28)(29). No genes involved in this pathway have been cloned. In the absence of genes for targeted mutagenesis of the GPI pathway, we generated cells with this desired deficiency by stable expression in T. cruzi of a GPI-PLC gene from the related protozoan parasite T. brucei. This approach was used previously to explore the topography of the protein and polysaccharide-GPI pathways in Leishmania (11).  3 and 6). 35 S incorporation into macromolecules was determined by trichloroacetic acid precipitation of a lysate of 10 6 cells to be 27,696 cpm and 29,323 cpm for wild-type and pTEX.GPI-PLC/T. cruzi, respectively.) Ssp-4 was immunoprecipitated from chase medium (5 ϫ 10 7 cell eq) with mAb 2C2. GPI-PLC requires GlcN(␣1-6)Ins for efficient substrate recognition (30) and is therefore highly specific for GPIs. 4 The enzyme cleaves most GPI biosynthetic intermediates in vivo (Fig. 1), thereby conferring on pTEX.GPI-PLC/T. cruzi a phenotype equivalent to a dominant-negative loss of function mutation in the GPI pathway. The severity of the phenotype was dependent upon the amount of GPI-PLC expressed, which was in turn driven by the concentration of G418 used in the culture medium. At a permissive G418 concentration (i.e. Ͻ200 g/ml), biochemical studies were possible (Figs. [1][2][3][4][5], since the parasite completed its life cycle. However, at a nonpermissive level of G418 (Ն400 g/ml), replication and differentiation of pTEX.GPI-PLC/T. cruzi amastigotes was inhibited. 3 These observations are reminiscent of the temperature-sensitive yeast mutants in GPI biosynthesis, which fail to grow at the nonpermissive temperature (5,6).
TcAmLP-3 is the first GPI to be characterized from the amastigote stage of T. cruzi (Figs. 1-2). Its depletion in GPI-PLC expressing cells (Fig. 1, A and B) is diagnostic of a GPI deficiency. The structure of TcAmLP-3 is EtN-P-Man-GlcN-Ins-phosphoglycero-sn-1-lipid (reminiscent of H5, a mammalian GPI; Refs. 22 and 31). This situation contrasts with GPIs from the epimastigote stage of T. cruzi, where only the Man 4 species contains EtN-P (32,33). Apparently, the pathway for GPI biosynthesis is developmentally regulated in T. cruzi (or stage-specific), as documented for T. brucei (2,34).
The Fate of Proteins with GPI-addition Signal Sequences during a GPI Shortage-The GPI deficiency in T. cruzi causes a decrease in cell surface expression of proteins that are normally GPI-anchored (Fig. 3). Some of these proteins are secreted constitutively (e.g. Ssp-4 (Fig. 5, A and B)), while others are degraded intracellularly (e.g. p60 (Fig. 4A) and protease gp50/55 (Fig. 4, B and C)).
In murine Thy-1-negative T cell lymphomas, degradation of proteins during a GPI deficiency occurs in class A, C, and H complementation groups (3). Secretion of Thy-1 occurs in class E and B GPI-deficient Thy-1-negative cells (3). However, our work with T. cruzi provides the first evidence that both secretion and degradation of proteins with GPI signal sequences can occur in an identical genetic background during a GPI deficiency.
How could proteins with functional GPI signal sequences acquire different fates although present in the same cell? We hypothesize entry into either route (i.e. intracellular degradation or secretion) is determined by the primary structure of the protein under consideration.
A protein-GPI transamidase (PGTase) normally transfers prefabricated GPI anchors to the cleavage/addition site of the COOH-terminal GPI signal sequences in the endoplasmic reticulum (ER) (reviewed in Ref. 35). During a GPI shortage, the absence of one substrate (i.e. GPIs) could affect the interaction of PGTase with its second substrate (i.e. COOH-terminal GPI signal sequence). For example, a protein with a high affinity GPI signal sequence might outcompete another protein with a weak GPI signal sequence for binding to PGTase. Binding to PGTase in the absence of GPIs can result in cleavage of the COOH-terminal GPI signal sequence, in a half-reaction of the normal transamidation, possibly followed by secretion of the cleaved protein. Cleavage of a GPI signal sequence without attachment of a GPI has been documented (36,37). Proteins with lower affinity GPI signal sequences (i.e. with regard to binding the PGTase), being outcompeted for binding to the transamidase, might not be cleaved and could be retained in the cell initially. Extended lingering in the ER, presumably attached to the lumenal membrane by a hydrophobic COOH terminus, coupled with exposure of degradation signals (38), might result in proteolysis. In effect, during a GPI shortage, proteins with low affinity GPI signal sequences might be analogous to proteins with "uncleavable GPI signal sequences," which are degraded in an ER/post-ER compartment (39,40), while proteins with high affinity GPI signal sequence are secreted.
Possible Roles of GPIs in Nuclear Division-Unlike most eukaryotic cells, which contain numerous mitochondria, the Trypanosomatidae contain a single mitochondrion (kinetoplast) whose replication is coordinated with nuclear division. In T. brucei, initiation of kinetoplast replication precedes mitosis in S phase with completion of nuclear division after successful kinetoplast replication (41). That this temporal order of kinetoplast and nuclear division occurs in T. cruzi as well is supported by our data. In GPI-deficient T. cruzi kinetoplast division can be uncoupled from division of the cell nucleus.
We speculate that GPIs might act by one of the two general mechanisms to influence nuclear division. First, lack of some GPI-anchored proteins or free GPIs unattached to proteins might lead to the inhibition of mitosis. Both these classes of putative mediators (i.e. free GPIs and GPI-anchored proteins), however, are expected to be found mainly on the plasma membrane of these intracellular amastigotes, so how could they function? First, they could be receptors for signals originating from the cytoplasm of the mammalian cells to trigger amastigote proliferation. Second, it is possible that free GPIs (or a GPI-anchored protein) control nuclear division in T. cruzi. Here, it is worth recalling that the ER, where GPIs are synthesized (11,42), is continuous with the outer membrane of the cell nucleus. This implies that free GPIs and/or GPI-anchored proteins may be present on nuclear membranes from where they could influence mitosis.