A surface antigen of Giardia lamblia with a glycosylphosphatidylinositol anchor.

Since Giardia lamblia trophozoites are exposed to high concentrations of fatty acids in their human small intestinal milieu, we determined the pattern of incorporation of [3H]palmitic acid and myristic acid into G. lamblia proteins. The pattern of fatty acylation was unusually simple since greater than 90% of the Giardia protein biosynthetically labeled with either [3H]palmitate or myristate migrated at approximately 49 kDa (GP49) in reducing sodium dodecyl sulfate-polyacrylamide gel electrophoresis during both growth and differentiation. GP49, which partitions into the Triton X-114 detergent phase, is localized on the cell surface since it is 125I-surface-labeled. GP49 was also biosynthetically labeled with [14C]ethanolamine and [3H]myoinositol, suggesting that it has a glycosylphosphatidylinositol (GPI) anchor. Moreover, phospholipase A2 (PLA2) or mild alkaline treatment released free fatty acids, indicating a diacylglycerol moiety with ester linkages. Finally, a 3H- and 14C-labeled species was released by nitrous acid deamination from [14C]palmitate- and [3H]myoinositol-labeled GP49. The GPI anchor of GP49 is unusual, however, because purified GP49 was cleaved by Bacillus cereus phosphatidylinositol (PI)-specific PLC, but not by Staphylococcus aureus PI-PLC, or plasma PLD, and did not react with antibody against the variant surface glycoprotein cross-reactive determinant. Moreover, the double-labeled deaminated GP49 anchor migrated faster than authentic PI in TLC and produced [3H]glycerophosphoinositol after deacylation. In contrast to the variable cysteine-rich G. lamblia surface antigens described previously, GP49 was identified in Western blots of every isolate tested, as well as in subclones of a single isolate which differ in expression of a major cysteine-rich 85/66-kDa surface antigen, which does not appear to be GPI-anchored. These observations suggest that GP49, the first common surface antigen to be described in G. lamblia, may play an important role in the interaction of this parasite with its environment.

The parasitic protozoan Giardia lamblia is a major cause of waterborne enteric disease worldwide (Craun, 1984). This flagellate is of great biologic interest as well as medical importance because it belongs to the earliest identified lineage to diverge from the eukaryotic line of descent (Sogin et al., 1989). Giardia has two life cycle stages which are well adapted to survival in very different and hostile environments. The dormant cyst form, which is responsible for transmission, survives well in cold water (Bingham et al., 1979). In contrast, the flagellated trophozoite form, which causes disease, colonizes the human upper small intestine where it is exposed to fatty acids and bile, as well as digestive enzymes. Therefore, it is important to understand how the outer surface of this unique parasite enables it to survive in such a degradative milieu. Key surface antigens of many parasites have glycolipid anchors. However, little is known of how Giardia proteins are anchored in the plasma membrane.
Previously, we have shown that a mixture of biliary lipids (Gillin et al., 1986) or fatty acids complexed to BSA' (Wieder et al., 1983) could replace serum in supporting growth of G.
Lamblia in vitro. Moreover, exposure of cuItured G. lamblia trophozoites to bile salts and fatty acids at the slightly alkaline pH (pH 7.8) of the human small intestine triggered encystation (Gillin et ab, 1988). The importance of fatty acids for growth and differentiation stimulated us to assay the proteins of both life cycle stages of G. lamblia for fatty acylation.
In the present study, we show that the pattern of fatty acylation of G. lamblia proteins is very simple and does not change during differentiation. We found that the major fattyacylated species is a protein of -49 kDa which is linked to the outer face of the plasma membrane by a GPI anchor. In contrast, we show that the predominant 66/85-kDa cysteinerich surface antigen (Gillin et al., 1990), which is variably expressed among Giardia isolates and subclones, does not appear to be fatty-acylated. Moreover, we have observed apparently identical fatty acylation patterns and expression of GP49 among every isolate and subclone of G. lamblia tested, suggesting that it may be a common antigen with properties that are important for parasite survival. SDS-PAGE. Fig. 2 shows that GP49 (labeled with either palmitate or myristate) is an amphiphilic protein which partitions into the Triton X-114 detergent phase. After Triton X-114 phase separation, GP49 consistently migrates slightly faster, a t around 46 kDa. We do not yet know the relationship between these two species.
In order to determine the native molecular weight of this protein, the phase-separated ["Hlpalmitate-labeled GP49 was analyzed on nonreducing SDS-PAGE. Fig. 3 shows major fatty acylated species of M , = 90,000 and 49,000 and a minor band a t 46,000. We do not yet know whether the 90-kDa species represents a dimer of GP49 or GP49 complexed to another protein.
Biosynthetic Labeling of GP49 with GPI Components and Nitrous Acid Dearnination-Next we asked whether the fatty acid associated with GP49 is part of a phospholipid anchor. Fig. 4 shows the metabolic incorporation of GPI components into GP49, as analyzed by SDS-PAGE.
[3H]myoinositol and [14C]ethanolamine, as well as [3H]glucosamine (not shown), were each incorporated into GP49, but [''Hlcholine was not (data not shown). Fig. 5A shows immunoprecipitation of ""Isurface-labeled GP49 using monospecific rabbit antiserum against GP49, showing that this protein is exposed on the cell surface and supporting the specificity of this antibody.

RESULTS
Fatty Acylation Is Not Stage-specific-We have observed (Gillin et al., 1987(Gillin et al., , 1988) that exposure of cultured trophozoites to free fatty acids in the presence of a primary bile salt stimulates encystation. Therefore, we performed biosynthetic labeling experiments to determine if ["Hlpalmitate or myristate is incorporated into stage-specific proteins of growing or encysting trophozoites. The major protein labeled with [,"HI palmitate or myristate migrated a t -49 kDa (GP49) with a fainter band a t 46 kDa in SDS-PAGE (Fig. 1). Since the 49-kDa and minor labeled proteins were observed equally in both encysting and nonencysting trophozoites, fatty acylation did not appear to be altered during encystation. Surface iodination (see below) revealed that GP49 is exposed on the cell surface of nonencysting trophozoites. In contrast, the major 66-and 85-kDa surface-iodinated species of strain WB clone C6 (Gillin et al., 1990) did not appear to be fatty-acylated (Fig. 1). Differential Extraction of GP49 by Triton X-114-Since most fatty acylated proteins are membrane-associated, radiolabeled proteins were solubilized in Triton X-114, and the detergent-rich and detergent-poor phases were analyzed on Portions of this article (including "Materials and Methods" and Figs. 2, 3, 5, and 8-10) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.

G. lamblia Surface
Antigen et al. (1988) have shown that ethanolamine is incorporated into several proteins of the parasitic trematode Schistosoma rnansoni, which are not GPI-anchored. However, GP49 is the major protein which is labeled with fatty acids, myoinositol, and ethanolamine, suggesting that it is GPI-anchored. The free amino group of the glucosamine residue is susceptible to nitrous acid deamination which cleaves its glycosidic linkage (Shively and Conrad, 1976). Fig. 6 shows that the "Hand "C-labeled radioactive product released by nitrous acid treatment of immunopurified GP49 migrated faster (R, = 0.7-0.8) than the PI marker (R, = 0.5-0.6), suggesting that it is more hydrophobic. In comparison, the deaminated product of the well characterized GPI-anchored trypanosoma1 protein, VSG, migrated with authentic PI (Fig. 6). Recently, Walter et al. (1990) have demonstrated that human erythrocyte decay-accelerating factor contains an ester-linked fatty acid substitution on the inositol ring of its GPI anchor and that the product released by nitrous acid deamination migrates faster than the authentic PI in TLC. The nitrous acid deamination result, along with the ohservations that GP49 is not cleaved by S. aureus PI-PLC and is not recognized by anti-CRD antibody (see below), suggests that it may contain a substituted inositol ring (Roberts ct al., 1988a(Roberts ct al., , 1988bWalter et al., 1990). The observation that treatment of the nitrous acid-deaminated anchor with methylamine released all I T radioactivity supports this hypothesis. This procedure releases ester-linked fatty acids, hut not etherlinked fatty alcohols. Moreover, the water-soluble product of deacylation contained only 'H label and co-migrated with glycerophosphoinositol in HPLC (Fig. 7).
Hydrolysis of GP49 by Phospholipases-Bacterial PI-specific PLCs, which release 1,2-diacylglycerol from phosphoglycerides, are widely used to identify GPI-anchored molecules. These enzymes have little or no specificity for the acyl or alkyl group of the PI substrate, but differ in specificity toward the inositol head group (see Ferguson and Williams, 1988, for review). Therefore, gel-purified ["Hlpalmitate-labeled GP49 and ['HHImyristate-labeled mf VSG were treated with PI-PLC from various sources. The results in Table I  Ester-linked fatty acids were released from the nitrous acid-deaminated product hy methylamine treatment as descrihed under "Materials and Methods," and the water-soluhle product was analyzed hy HI'LC on a Whatman I'artisil 10 SAX column (0.45 X 25 c m ) (Auger et al., 1989). Samples were loaded in approximately 1 ml. Hadioactlvity was detected with a Reckman online 171 radioisotope detertor with a 1-ml flow cell. The column's flow rate WAS 1 ml/min. The scintillation fluid, Ready Flo Ill (Reckman), was pumped at a rate of 3 ml/min. A , glycerophosphoinositol; H . glvcerophosphoinositol-3-11; C, gIycerophosphoinositol-4-P; I), gl~cerophosphoinositol-~1,5-1'~.

Percent
show that 3H-fatty acid-labeled GP49 was cleaved by the PI-PLC from B. cereus. When the hydrophobic product was analyzed on TLC, approximately 65-70% of the radioactivity released migrated as 1,2-sn-diacylglycerol (Fig. 8). However, GP49 was not hydrolyzed by other phospholipase C (Table I) or released in soluble form from giardia1 membranes by B.
cereus PI-PLC (data not shown). PLD, which is abundant in mammalian plasma and hydrolyzes many GPI-anchored proteins (Low and Prasad, 1988), including human erythrocyte acetylcholinesterase (Roberts et al., 1988a(Roberts et al., , 1988b also was not effective against GP49 (Table I) for reasons which are not yet understood. In mixing experiments, we found that GP49 itself is not an inhibitor of the phospholipase C tested because [3H]mf VSG was hydrolyzed in the presence of ["C] GP49 (data not shown). This agrees with the above evidence (Fig. 6) that the GPI anchor of GP49 has some modification which prevents its hydrolysis by the phospholipases tested other than B. cereus PI-PLC. The unusual structure of GP49 was also supported by the observation (not shown) that in contrast to VSG, purified GP49 cleaved by B. cereus PI-PLC did not react in Western blots with antibody to the crossreacting determinant (CRD) of VSG which reacts with many GPI anchors after cleavage with PI-PLC.
Release of 3H-Fatty Acid from GP49 by Mild Alkali and Phospholipase AP Treatment-The release of radioactive fatty acid from metabolically labeled acyl proteins by treatment with nucleophilic hydroxylamine or mild alkali has been widely used to identify amide or ester-linked fatty acids. In GPI-anchored proteins, fatty acids are attached to the 1,2-sndiacylglycerol backbone via ester linkages. When GP49 was digested with mild alkali, more than 90% of its radioactivity was released as free and esterified palmitic acid (Fig. 9A), showing ester linkages (Ferguson and Cross, 1984).
PLAZ hydrolyzes the fatty acid ester bond in position 2 of phospholipids. As reported previously (Ferguson et al., 1985a), phospholipase A2 removed 50% of the myristate from ["HI myristic acid-labeled VSG. Similarly, when [3H]palmitatelabeled GP49 was treated with snake venom PLA,, approximately 40% of the fatty acid was released. This observation was confirmed by identifying the released product on TLC (Fig. 9B). Taken together, the results of alkali and PLA, treatments indicate a 1,2-sn-diacylglycero1 moiety in GP49 which is typical of GPI anchors.
GP49 Is a Common Surface Antigen-The G. lamblia surface antigens which have been well characterized to date are a group of extremely cysteine-rich proteins which vary both among isolates and between subclones of a single isolate (Adam et al., 1988;Aggarwal et al., 1989). Therefore, we asked whether GP49 is also a variable antigen. Recently, we (Gillin et al., 1990) cloned and sequenced the major cysteine-rich 66/ 85-kD surface antigen (called TSA 417) of G. lamblia WB clone C6. In the present study, we used a rabbit polyclonal serum directed against recombinant TSA 417 with complement to select clones which do not express this major antigen. Biosynthetic labeling with ['HJpalmitate showed GP49 to be the major fatty acylated protein in TSA 417-positive and -negative subclones (Fig. 10). Moreover, this band reacted with antibody to GP49 in Western blots (not shown). Since strain WB originated in Afghanistan, we asked whether GP49 is present in different G. lamblia isolates. Immunoblot analysis with the anti-GP49 antibody, which appears to recognize protein determinants: revealed GP49 in isolates of G. lamblia from Turkey, Portland (OR), Alaska, and Peru (Fig. 11). In every case, the minor 46-kDa species was also present. The autoradiogram of the same Western blot demonstrates co-S. Das   migration of the major fatty acid-labeled species. These results show that GP49 is present in all isolates tested and does not vary with the major cysteine-rich species.

DISCUSSION
The two life cycle stages of G. lamblia are remarkably well adapted to survival in very different and inhospitable environments. The dormant, egg-shaped cyst form which is responsible for transmission, survives for months in fresh water at 4 to 8 "C. After ingestion and passage through the stomach and into the small intestine, flagellated trophozoites emerge from the cysts to colonize the human upper small intestine, a complex environment containing everchanging concentrations of fatty acids, bile salts, hydrogen ions, and food and digestive enzymes as well their products. The structure and function of cell surface components of G. lamblia trophozoites which permit them to survive in this hostile environment are largely unknown. Therefore, it is important to identify the surface components of G. lamblia which play a key role in its adaptation and survival.
Farthing et al. (1985) proposed that in the small intestine, G. lamblia might obtain its lipids from bile. Similarly, we reported that defined mixtures of biliary lipids containing bile salts, phospholipids, and cholesterol support the growth of G. lamblia in uitro (Gillin et al., 1986). Moreover, Jarroll et al. (1981) showed that these parasites have a limited capacity for de novo lipid biosynthesis. Blair and Weller (1987) reported the preferential incorporation of polyunsaturated and saturated fatty acids into different phospholipids. Interestingly, palmitic acid was mainly incorporated into phosphatidylinositol. We have also shown that exposure of cultured G. lamblia trophozoites to bile salt and free fatty acid triggers encystation (Gillin et al., 1988). Since fatty acids are important for both growth and differentiation, we asked whether crucial proteins or antigens of G. lamblia are fatty acylated.
GPI anchors have been found in many proteins with diverse biological activities, including cell surface antigens, membrane-bound enzymes, and adherence proteins. In Trypanosoma brucei, VSG anchored through its GPI molecule is susceptible to cleavage by an endogenous phospholipase C and is released quickly during host-parasite interactions. This may also be true of the lipophosphoglycan antigen of Leishmania (Turco et al., 1984) or the membrane form of the P30 antigen of Toxoplasma (Nagel and Boothroyd, 1989), which are also GPI-anchored.
In the present study, the biosynthetic labeling experiments with [3H]palmitic and -myristic acid show that the 49-kDa major fatty-acylated protein (GP49) and a minor 46-kDa species are observed equally in both encysting and nonencysting trophozoites (Fig. 1). Therefore, the pattern of fatty acylation does not appear to change during encystation. GP49 is a membrane-associated, hydrophobic surface protein since it can be surface-labeled with lZsI ( Fig. 5A) and partitions into the detergent phase after Triton X-114 extraction (Fig.  2). The migration of GP49 appears to be influenced by detergents and reducing agents. After Triton X-114 phase separation, GP49 migrates slightly faster, around 46 kDa. Under nonreducing conditions, a substantial portion of GP49 migrates as a dimer, at -90 kDa. However, the major species is still 49 kDa and the 46-kDa minor species is apparent. We do not at present understand the relationship between these three protein species.
Since many surface antigens of protozoan and mammalian cells are anchored on the cell surface by a GPI anchor, it was of interest to determine whether this is true of GP49. Presence of a GPI anchor was supported by the metabolic incorporation of [l4C]ethano1amine, [3H]myoinositol, and fatty acids into GP49 (Fig. 4). This was confirmed by enzymatic and chemical cleavage experiments (Figs. 6-9). However, the GPI anchor of GP49 differed from that of the well characterized VSG in three respects. While GP49 is susceptible to cleavage by B. cereus PI-PLC, it was not hydrolyzed by S. aureus PI-PLC, T. brucei GPI-PLC, or rabbit or human plasma PLD (Table  I), suggesting a major difference between the GPI moiety of GP49 and VSG. Moreover, antibody to the VSG (CRD) which cross-reacts with many other GPI-anchored molecules did not recognize GP49 (not shown). Finally, GP49 reacts with wheat germ agglutinin, a lectin specific for N-acetylglucosamine or sialic acid residues (data not shown), which are usually not part of the GPI glycan. Further work is needed to determine whether N-acetylglucosamine and/or sialic acid are present in the GP49 GPI glycan or elsewhere on the molecule.
Recently, Roberts et al. (1988a, 1988b) proposed that the presence of palmitic acid in the 2-position of the inositol ring in human erythrocyte acetylcholinesterase is responsible for its resistance to cleavage by S. aureus PI-PLC. In the present study, we have shown that the purified GP49 of G. lamblia is resistant to PI-PLC from S. aureus but not from B. cereus. However, B. cereus PI-PLC did not release GP49 in soluble form from the intact membrane (data not shown). When palmitoylated GP49 was subjected to nitrous acid deamination, no radioactivity was detected either in the glycan moiety or in the protein part of the molecule (not shown). Sequential degradation (nitrous acid deamination followed by methylamine treatment) of purified GP49 double-labeled with ["C] palmitate and [3H]myoinositol supported the hypothesis that, like human erythrocyte acetylcholinesterase, the GP49 of G. lamblia may contain a fatty-acylated inositol ring. Although the chemical degradation studies of GP49 are consistent with an acylated GPI similar to that of human erythrocyte acetylcholinesterase, the results with B. cereus PI-PLC and plasma PLD are not, for reasons which are not yet understood. A detailed structural analysis of Giardia GP49 is necessary to locate the exact site of acylation in the inositol ring as well as to elucidate the total structure of the anchor molecule.
The G. lamblia surface antigens which have been characterized to date are a group of extremely cysteine-rich proteins which vary both among isolates and between subclones of a single isolate (Aggarwal et al., 1989). Adam et al. (1988) have demonstrated that a cloned isolate of G. lamblia expresses a cysteine-rich 170-kDa surface antigen (CRP 170) which undergoes antigenic variation. In their study (Adam et al., 1988), a monoclonal antibody against CRP 170 was used to select subclones in which the 170-kDa protein was replaced by other cysteine-rich proteins ranging from -50 to 170 kDa. Recently, we have cloned and sequenced the entire gene (TSA 417) which encodes the major 66-and 85-kDa cysteine-rich trophozoite surface antigen species of G. lamblia strain WB, clone C6 (Gillin et al., 1990). Although sequence analysis showed a hydrophilic protein with a hydrophobic C-terminal membrane-spanning region, the present studies show that this is probably not replaced by a GPI anchor since the 661 85-kDa protein does not appear to be metabolically labeled by GPI precursors. TSA 417 negative subclones were isolated with or without antibody selection. In contrast, GP49 did not appear to vary among TSA 417 positive and negative subclones (Fig. 10). Moreover, despite repeated selection with GP49 antisera and complement, surviving parasites still expressed GP49 as detected by Western blots (not shown). Furthermore, G. lamblia isolates from Turkey, Portland, Alaska, Peru, and Afghanistan, whose expression of TSA 417 varied from <OB1 to >80% of the population, showed identical patterns of fatty acylation and reactivity with anti-GP49 in Western blots (Fig. 1I.A). Thus, GP49 is a common, possibly invariant, antigen which could play a crucial role in the survival of G. lamblia and may be a target for vaccine development.