S-Myristoylation of a Glycosylphosphatidylinositol-specific Phospholipase C inTrypanosoma brucei *

Covalent modification with lipid can target cytosolic proteins to biological membranes. With intrinsic membrane proteins, the role of acylation can be elusive. Herein, we describe covalent lipid modification of an integral membrane glycosylphosphatidylinositol-specific phospholipase C (GPI-PLC) from the kinetoplastid Trypanosoma brucei. Myristic acid was detected on cysteine residue(s) (i.e. thiomyristoylation). Thiomyristoylation occurred both co- and post-translationally. Acylated GPI-PLC was active against variant surface glycoprotein (VSG). The half-life of fatty acid on GPI-PLC was 45 min, signifying the dynamic nature of the modification. Deacylation in vitro decreased activity of GPI-PLC 18–30-fold. Thioacylation, from kinetic analysis, activated GPI-PLC by accelerating the conversion of a GPI-PLC·VSG complex to product. Reversible thioacylation is a novel mechanism for regulating the activity of a phospholipase C.

Lipid modification of proteins appears to occur in most eukaryotes. Three general classes of covalent modification have been documented (i.e. N-acylation, isoprenylation, and thioacylation) (reviewed in Ref. 1). Fatty acids, depending on their length, are linked to proteins by one of two bonds; long chain fatty acids (e.g. palmitate) are esterified to cysteines (i.e. thiopalmitoylation) in varying positions along the polypeptide (see Refs. 2 and 3 for reviews). (On occasion, long chain fatty acids occur in amide linkage to lysine (4).) On the contrary, shorter chain fatty acids (e.g. myristate) have been found exclusively in amide linkage to N-terminal glycine (N-acylation) (5).
Trypanosoma brucei, the causative agent of African trypanosomiasis (sleeping sickness), is a protozoan parasite, which in the bloodstream of a mammal is covered with about 10 7 molecules of a variant surface glycoprotein (VSG), 1 a GPI-anchored molecule. A 39-kDa GPI-specific phospholipase C (GPI-PLC) that can cleave diacylglycerol from VSG and GPI biosynthetic intermediates is expressed in bloodstream form T. brucei (6 -8).
The parasite contains approximately 4 ϫ 10 4 molecules of GPI-PLC per cell (6), about one enzyme molecule per 250 molecules of VSG. With a turnover number (k cat ) of 144 min Ϫ1 (9), there appears to be sufficient enzyme intracellularly to cleave all of the VSGs within a few minutes. However, cleavage of VSG in living T. brucei is not a major catabolic pathway, although purified GPI-PLC cleaves VSG efficiently in vitro (9). These observations suggest, among other possibilities, that (i) GPI-PLC may not have access to VSG in a living cell and/or (ii) GPI-PLC is not enzymatically active in vivo. GPI-PLC has been immunolocalized to the cytoplasmic leaflet of intracellular vesicles (10), apparently sequestered, topologically, from VSG which is attached to the outer (exoplasmic) leaflet of the plasma membrane. Such localization studies support the former hypothesis, although the latter cannot be ruled out completely. Biosynthesis of GPIs is initiated on the cytoplasmic side of the endoplasmic reticulum (11), implying that GPI intermediates probably co-localize with GPI-PLC on the cytoplasmic side of cellular membranes in T. brucei. This assertion is supported by two studies involving the stable transfection of a cDNA encoding T. brucei GPI-PLC into the related protozoan parasites Leishmania major and Trypanosoma cruzi. First, L. major expressing GPI-PLC acquires a GPI-negative phenotype resulting apparently from cleavage of protein-GPI intermediates by the enzyme (12). Second, in T. cruzi GPI-PLC causes a GPI deficiency that is associated with failure of the parasites to sustain division of the cell nucleus (13). Given these observations, it is imperative to examine why GPI-PLC in its native environment (i.e. in T. brucei) does not cause a GPI deficiency.
As part of efforts aimed at unraveling the mechanisms that regulate activity of GPI-PLC in T. brucei, we investigated the possibility that the enzyme, itself an integral membrane protein, had a lipid modification. Herein, we demonstrate that GPI-PLC is covalently modified with myristic acid in a thioester linkage. The observation is the first example of acylation of a GPI-PLC in vivo. The esterified fatty acids modulated activity of this PLC, proving that phospholipases may be regulated by covalently attached lipids.

Cell Types/Strain
T. brucei ILTat 1.3 was harvested from the blood of infected rats by cardiac puncture. Buffy coats were prepared and parasites purified by DE52 chromatography (14). [9,

Cycloheximide Treatment
After metabolic labeling of 10 (20 nM).) Cells were held on ice for 20 min, after which the lysate was centrifuged at 14,000 ϫ g (4°C) for 10 min, and the supernatant was discarded. This lysis step was repeated.

Immunoprecipitation of GPI-PLC
A membranous pellet, from the second lysis step (above), was resuspended in 1 ml of immunoprecipitation dilution buffer (1% (v/v) Triton X-100, 0.2 M NaCl, 60 mM Tris-HCl, pH 7.5, 6 mM EDTA, 10 units/ml aprotinin, and PIC). The lysate was incubated on ice for 30 min. One microliter of mc2A6-6 ascites fluid was added to the detergent-solubilized pellet, which was incubated (4°C) with continuous inversion for 2 h. To get rid of membrane-associated VSG, samples were centrifuged at 14,000 ϫ g for 5 min at 4°C and the pellet discarded. To the supernatant, 50 l of a suspension (1:1) of protein A-Sepharose:water was added followed by incubation at 4°C with repeated inversion for 18 h. Samples were centrifuged at 14,000 ϫ g at 4°C for 10 s, and the supernatants were discarded. Recovered beads were washed as follows: once with 1 ml of PBS; twice, each time with 750 l of buffer KGKOG (3.17 M potassium glutamate, 0.59 M KCl, and 0.1% (w/v) n-octyl glucoside); twice, 1 ml each, with PBS containing 2% Nonidet P-40 (15). The beads received a final wash in 1 ml of PBS. Protease inhibitors were added to all buffers before the washes, performed by repeated inversion on a laboratory rotator at 4°C for 10 min. Beads were recovered in between washes by centrifugation at 14,000 ϫ g for 10 s. To help recovery of the protein A-Sepharose beads after washing with KGKOG, 750 l of deionized water was added, to reduce the viscosity of the solution prior to the centrifugation step.

SDS-PAGE and Fluorography
Immunoadsorbed 3 H-acylated GPI-PLC (approximately 50 l) was eluted by heating the protein A-Sepharose-antibody complex in 25 l of a modified 2.5ϫ SDS-PAGE sample buffer. A SDS-PAGE buffer lacking Tris was used. We found that presence of Tris base in SDS-PAGE sample buffer deacylated GPI-PLC when heat (90°C, 5 min) was applied. Eluted samples were centrifuged for 3 min at 14,000 ϫ g (room temperature). A 20-l aliquot of the supernatant was analyzed by SDS-PAGE (14% minigel) (Bio-Rad). The gel was soaked in Entensify (DuPont), dried, and exposed to pre-flashed Hyperfilm-MP at Ϫ80°C. Alternatively, proteins were transferred to Immobilon P with a Trans-Blot semi-dry cell (Bio-Rad). It was operated for 2 h at 20 V at 25°C and 0.5 A in 48 mM Tris, 39 mM glycine, 20% (v/v) methanol, pH 9.2, containing 1.3 mM SDS. The Immobilon P membrane was coated with molten MeltiLex wax, air-dried, and exposed to pre-flashed Hyperfilm-MP at Ϫ80°C.

Western Blots
The Immobilon P membrane was first soaked in blocking solution (1% (v/v) Tween 20, 10% (v/v) newborn calf serum, 13% (w/v) glycerol, 18% (w/v) D-glucose in PBS) for 1 h at room temperature with shaking. This was followed by 1 h incubation in blocking solution containing a rabbit polyclonal antibody raised against GPI-PLC polypeptide (R18B3), at a 1:3000 dilution. The membrane was then washed three times (10 min each) with PBS followed by incubation for another hour with 1:1000 dilution of alkaline phosphatase-conjugated goat anti-rabbit IgG. The PBS wash was repeated. The membrane was then washed twice (10 min each) in alkaline phosphatase reaction buffer (75 mM NaCl, 75 mM Tris-HCl, pH 9.5, and 3.8 mM MgCl 2 ). Antigens were visualized in alkaline phosphatase reaction buffer, using 5-bromo-4chloroindolyl phosphate and p-nitro blue tetrazolium chloride as substrates.

Deacylation of GPI-PLC
Deacylation on Immobilon P Membrane-Immunoadsorbed 3 H-acylated GPI-PLC or proteins from [ 3 H]myristate-labeled T. brucei were resolved by SDS-PAGE and transferred to Immobilon P. The membrane was soaked in 200 ml of one of the following solutions: 1 M NH 2 OH (adjusted to pH 7 with 10 N NaOH); 0.2 M KOH; and 1 M Tris-HCl, pH 7. Each solution was incubated at 30°C with shaking. These solutions were replaced three times (every 30 min), after which the membranes were incubated overnight in 200 ml of the same solution. The membranes were air-dried, coated with MeltiLex wax, and radiolabeled proteins detected by fluorography.
In Gel Deacylation of GPI-PLC-Immunopurified [ 3 H]GPI-PLC was electrophoresed on an SDS-PAGE. Gel slices containing the radiolabeled polypeptide were excised, sliced, and placed in a 1.5-ml microcentrifuge tube (approximate volume of gel was 400 l). After rinsing with 1 ml of deionized water, [ 3 H]GPI-PLC was deacylated in situ with 1 ml of neutral hydroxylamine by incubation at 30°C. At 20-h intervals, the gel suspension was centrifuged at 14,000 ϫ g for 1 min, and the supernatant was withdrawn. The incubation and centrifugation steps were repeated five more times, and the supernatants were pooled and dried in vacuo. Released 3 H-fatty acids were identified as described below.
Deacylation of Native 3 H-Acylated GPI-PLC and 3 H-Myristoylated mfVSG-A complex of 3 H-acylated GPI-PLC (from 4 ϫ 10 9 cells) adsorbed to monoclonal antibody 2A6-6, which was in turn bound to protein A-Sepharose (in 100 l), was resuspended in 250 l of 1 M NH 2 OH, pH 7.0. The suspension was incubated at 30°C for 20 min with shaking, followed by centrifugation at 14,000 ϫ g for 1 min, and withdrawal of the supernatant. Two more hydroxylamine treatments were performed. Eluates were pooled, and 54 l of concentrated HCl added, to acidify the solution. [ 3 H]mfVSG was deacylated to provide cell-derived [ 3 H]myristate as a control in the TLC analysis. Approximately 300,000 dpm of purified protein was incubated at 30°C for 1 h with 900 l of 0.2 M KOH, and the solution was acidified with 20 l of concentrated HCl.

Deacylation of GPI-PLC and Determination of Phospholipase C Activity
GPI-PLC (from 2 ϫ 10 10 cells of a hypotonic lysate of T. brucei) was adsorbed to monoclonal antibody 2A6-6 (mc2A6-6) and washed as described above. The immunoadsorbent was incubated in 5 ml of 1 M neutral hydroxylamine at 30°C for 40 min with shaking. One-ml aliquots (approximately 2 g of GPI-PLC) were withdrawn after 0, 10, 20, 30, and 40 min and centrifuged at 14,000 ϫ g for 1 min. Recovered beads were washed three times in 3 ml of PBS. GPI-PLC was eluted from the column with 100 l of 50 mM Tris-HCl, pH 12, containing 0.1% Nonidet P-40, into an equal volume of chilled (0°C) 1 M Tris-HCl, pH 6.0, containing 1.0% Nonidet P-40 to neutralize the elution buffer.
Enzyme activity in the eluate was assayed as follows. A reaction mixture containing 2 g of [ 3 H]myristate-labeled VSG in 25 l of AB (50 mM Tris-HCl, 5 mM EDTA, 1% Nonidet P-40) was assembled on ice (in a 1.5-ml microcentrifuge tube). To this mixture, several dilutions (in AB) of the eluate from the mc2A6-6⅐protein A-Sepharose column was added (the objective being to obtain values within the linear range of the assay (9)). This reaction mixture was incubated at 37°C for 30 min and terminated by vortex mixing with 500 l of water-saturated nbutanol (at room temperature). Phases were separated by centrifugation (12,000 ϫ g, 1 min, 25°C), and enzyme activity was quantified by measuring the amount of [ 3 H]dimyristoylglycerol released into the upper butanol phase (9).

Kinetic Analysis of Acylated and Deacylated GPI-PLC
To obtain GPI-PLC for kinetic analysis, the immunoadsorption protocol was modified to reduce contamination of the purified enzyme with antibody leaching from the monoclonal antibody column. GPI-PLC (from 2.5 ϫ 10 9 parasites) was adsorbed to a 100-l suspension of mc2A6-6 that had been chemically cross-linked to protein A-Sepharose (9). The complex was treated with 1 M neutral hydroxylamine or PBS (see above, "Deacylation of Native 3 H-Acylated GPI-PLC and 3 H-Myristoylated mfVSG") for 40 min. GPI-PLC was eluted from the immune complex as described above. Amount of GPI-PLC that cleaved approximately 20% of [ 3 H]myristate-labeled VSG was determined empirically and used in the kinetic analysis.
Reaction mixtures were assembled on ice as described above with the exception that the AB buffer contained 0.1% Nonidet P-40. VSG cleavage was allowed to proceed for 15 min. Lineweaver-Burk plots were used to determine the Michaelis constant (K m ) (Table I). VSG was assumed to be a dimer in the calculation of substrate concentration. For purposes of determining the turnover number (k cat ), protein concentration of eluates from the mc2A6-6 column was determined by quantitative Western blots using known amounts of purified recombinant GPI-PLC as standards.  12 ) led to incorporation of radiolabel into GPI-PLC detected after SDS-PAGE (Fig. 1). This observation suggested that a fatty acid might be covalently bound to GPI-PLC. Lipids are covalently attached to proteins by one of three general mechanisms as follows: (i) through an amide bond (e.g. N-acylation), (ii) by esterification, as in S-palmitoylation; or (iii) in an ether linkage (e.g. Sisoprenylation). To define the link between the acyl group and GPI-PLC, attempts were made at deacylation after electrotransfer of purified 3 H-acylated GPI-PLC to Immobilon P membrane. Reagents tested were 0.2 M KOH (pH ϳ13), which cleaves both oxyesters and thioesters, 1 M NH 2 OH, pH 7 (neutral hydroxylamine), which can specifically cleave thioesters (17), and 1 M Tris-HCl, pH 7, a control. Cleavage (by de-Omyristoylation) of a GPI anchor (on VSG) was studied as a control for the effectiveness and specificity of these reagents.
Metabolic labeling of GPI-PLC with three fatty acids (i.e. [9,10-3 H]palmitate (C 16 ), [9,10-3 H]myristate (C 14 ), and [11, H]laurate (C 12 )) ( Fig. 1) raised the possibility that the enzyme was heterogeneously acylated. To address this issue, the fatty acid on GPI-PLC was characterized. Cells were metabolically labeled with [9,10-3 H]myristate; GPI-PLC was immunoadsorbed to protein A-Sepharose and deacylated with neutral hydroxylamine. Methyl esters of the released fatty acids were generated and analyzed by reversed phase-high performance thin layer chromatography (RP-HPTLC)/fluorography (Fig. 3,  lane 2). Using fatty acid methyl ester standards (Fig. 3, lane 1), myristic and palmitic acids were detected (Fig. 3, lane 2). The ratio of the myristate:palmitate was 3:1, as determined by laser scanning densitometry of the fluorographs. When the parasites were metabolically labeled with [ 3 H]palmitate, myristate and palmitate were detected on GPI-PLC, although the efficiency of labeling the enzyme was reduced significantly (data not presented). Together these data suggest that T. brucei metabolizes fatty acids before their incorporation into GPI-PLC.
Because this report is the first in the literature of myristate in a thioester bond to an enzyme, a second approach was adopted to check whether the assignment of one of the fatty
acids on GPI-PLC as myristic acid was in error. For this purpose, purified VSG was deacylated with KOH, and the released fatty acid was converted to the methyl ester and compared with the FAMES from lipids of GPI-PLC. The VSG fatty acids were correctly identified as myristic acid (Fig. 3, lane 3) (18). More importantly, the VSG fatty acid comigrated with the myristate from GPI-PLC (Fig. 3, lane 2). Thus, the assignment of the faster migrating fatty acid species in the GPI-PLC sample as myristate is not an artifact caused by spurious migration of a longer chain fatty acid on the HPTLC plate.
Co-and Post-translational Acylation of GPI-PLC-To investigate the temporal order of GPI-PLC acylation with regard to protein synthesis, cycloheximide (CHX), an inhibitor of T. brucei protein synthesis, was used in experiments in which metabolic labeling with [ 3 H]myristate was attempted before or after addition of CHX. When the trypanosomes were labeled with [ 3 H]myristate prior to CHX treatment (CHX-II), the inhibitor appeared to have no effect on acylation of GPI-PLC within that time frame studied (Fig. 4A, compare lane 1 to lane 3). How-ever, if the parasites were treated with cycloheximide (CHX-I) prior to addition of [ 3 H]myristate, acylation of GPI-PLC was inhibited partially (Fig. 4A, lane 2), compared with the control (Fig. 4A, lane 1). Approximately 50% of the lipid modification FIG. 2. GPI-PLC is thioacylated. GPI-PLC immunoprecipitated from [ 3 H]myristate-labeled cells (5 ϫ 10 8 ) and total acylated membrane proteins (10 7 cell eq) from T. brucei were resolved by SDS-PAGE (14% minigel). The proteins were electro-transferred to Immobilon P. The membranes were treated with 1 M Tris-HCl, pH 7 (lanes 1 and 2), 0.2 M KOH (lanes 3 and 4), or 1 M NH 2 OH, pH 7 (lanes 5 and 6). Lanes 1, 3, and 5 contain total acylated membrane proteins, and lanes 2, 4, and 6 contain GPI-PLC immunoprecipitates. Hyperfilm TM (Amersham Pharmacia Biotech) was developed after gels were exposed for either 18 h (lanes 1 and 5) or 7 days (lanes 2-4, and 6). on GPI-PLC (as estimated by laser scanning densitometry of fluorographs) occurs in the absence of new protein synthesis. We infer that 50% of the acylation occurred on nascent GPI-PLC, possibly on ribosomes. These observations indicate that acylation of GPI-PLC occurs both co-translationally and post-translationally.
In control experiments, the effect of CHX on the GPI-PLC polypeptide backbone was studied in [ 35 S]methionine-labeled T. brucei (Fig. 4B). Here, CHX blocked synthesis of the enzyme if pre-incubated with the cells (Fig. 4B, lane 2), but no effect on GPI-PLC was apparent when added after metabolic labeling (Fig. 4B, lane 3). These observations confirm our previous conclusion that a significant proportion of the S-acylation of GPI-PLC occurs post-translationally. When a gel identical to that depicted in Fig. 4A was analyzed by Western blotting with an anti-GPI-PLC antibody, similar amounts of GPI-PLC were detected in all three lanes (Fig. 4C). We conclude that loss of 3 H-lipid (Fig. 4A, lane 2) was due to a decrease in fatty acid addition instead of protein degradation.
To investigate whether myristate or palmitate was preferentially added to GPI-PLC post-translationally, trypanosomes were pretreated with CHX (CHX-I) prior to metabolic labeling with [ 3 H]myristate. [ 3 H]Acyl-GPI-PLC was immunoadsorbed and deacylated (see "Experimental Procedures"). The released fatty acid(s) were analyzed by RP-HPTLC/fluorography (see "Experimental Procedures") (Fig. 5). Both myristate and palmi-tate were added to GPI-PLC post-translationally, but myristate addition dropped about 50% (Fig. 5, compare lanes 2 and 3). The ratio of myristate:palmitate (recovered from GPI-PLC) decreased to 2:1 (i.e. from 3:1). As compared with the control lane (Fig. 5, lane 2), the amount of palmitate added posttranslationally to GPI-PLC did not change. Apparently, most of the post-translational acylation of GPI-PLC is due to myristate incorporation. These experiments also ruled out the possibility that the palmitate and myristate detected on GPI-PLC when the enzyme was adsorbed to a mc2A6-6⅐protein A-Sepharose column (Fig. 3) (Fig. 5, lane 2), it is clear that both fatty acids are covalently attached to GPI-PLC. The ratio of [ 3 H]myristate: [ 3 H]palmitate on GPI-PLC was again 3:1, as determined by laser scanning densitometry of fluorographs from the HPTLC plates.
Acylated GPI-PLC Is Enzymatically Active-Lipid modification has a potential for regulating the activity of GPI-PLC. For example, acylation of the protein could inactivate the enzymatic activity. Alternatively, deacylation might then be required to impart catalytic activity to the polypeptide. These ideas were tested as described below.
GPI-PLC cleaves the [ 3 H]myristate-labeled GPI of VSG during hypotonic lysis of T. brucei, releasing the radioactive lipid as di-[ 3 H]myristoylglycerol (19). If deacylation of GPI-PLC was necessary for enzyme activity, loss of the fatty acid from GPI-PLC might be detectable under conditions where cleavage of VSG is documented, that is loss of [ 3 H]acyl group from GPI-PLC might precede or be coincident with the loss of [ 3 H]myristate from the VSG GPI.
We attempted to determine whether any changes in the extent of GPI-PLC acylation occurred during VSG cleavage. For this purpose, [ 3 H]myristate-labeled trypanosomes were lysed hypotonically and incubated either at 4°C (Fig. 6B, lanes  1-4) or 37°C (Fig. 6B, lanes 5-8) in an attempt to initiate hydrolysis of the VSG GPI. GPI-PLC cleaves approximately 70% of [ 3 H]myristate-labeled VSG during hypotonic lysis of T. brucei (20). Residual VSG that is not cleaved initially is the substrate of interest in these experiments because it is lumenal (21,22) and therefore inaccessible to GPI-PLC which is cytoplasmically oriented (10,12). Detergent is needed to promote vesicle mixing which brings the enzyme and substrate together.
Cleavage of VSG occurred at both 4 and 37°C in the presence of detergent (Fig. 6B, lanes 4 and 8). However, there was no discernible loss of the [ 3 H]acyl group from GPI-PLC (Fig. 6B,  lanes 3 and 7). That loss of the [ 3 H]myristate from VSG (i.e. GPI cleavage) was not the result of degradation of VSG was confirmed by Western blotting; full-length VSG was present in the lysate after loss of the [ 3 H]myristate (data not presented).
We conclude that deacylation is not required for GPI-PLC activity. In support of this assertion, acylated GPI-PLC is catalytically more proficient than the deacylated form of the enzyme (Table I). Thus, the possibility that a minute fraction of unacylated GPI-PLC is responsible for cleavage of VSG in these lysates is very remote.
Acylation of GPI-PLC Is Reversible in Vivo-GPI-PLC in T. brucei and recombinant GPI-PLC expressed in Escherichia coli are both integral membrane proteins (6,9), suggesting that membrane binding is not the result of an eukaryote-specific chemical modification(s). Therefore, thioacylation (which has been demonstrated only in eukaryotes) seemed unlikely to be responsible for the ability of the enzyme to bind to membranes. Given these circumstances, the possibility that thioacylation played a regulatory role on the activity of the enzyme was entertained. Regulatory modifications on proteins (e.g. phosphorylation) are frequently reversible. Accordingly, we tested the possibility that acylation of GPI-PLC was dynamic.
Acylation of T. brucei GPI-PLC was reversible. When trypanosomes were labeled with [ 3 H]myristate and then "chased," lipid detected on the polypeptide decreased with a half-life of 45 min (Fig. 7). After a 60-min "chase" only ϳ30% of the acylation signal was detectable on GPI-PLC.
Reversibility of the acylation reaction and the ability of T. brucei to add the lipids post-translationally indicate that thioacylation of GPI-PLC in T. brucei is dynamic.
Deacylation Reduces Activity of GPI-PLC-An hypothesis that thioacylation activated GPI-PLC was next tested, by determining whether removal of the fatty acid from GPI-PLC would affect the kinetics of GPI cleavage. For this purpose, a complex consisting of GPI-PLC immunoadsorbed to protein A-Sepharose (i.e. a GPI-PLC⅐mcAb2A6-6⅐protein A-Sepharose) (see "Experimental Procedures") was treated either with neutral NH 2 OH or PBS (as a control) for varying lengths of time. GPI-PLC was then eluted from the column and its enzymatic activity assayed (9).
Deacylation of GPI-PLC under these conditions reduced activity of the enzyme 4-fold (Fig. 8A). This decrease in activity was not due to loss of GPI-PLC, since the full-length polypeptide was detectable by Western blotting following the hydroxylamine treatment (Fig. 8B). By following the loss of the [ 3 H] acyl signal from 3 H-acylated GPI-PLC during a similar deacylation experiment, a 4-fold decrease in signal was obtained after 40 min of neutral hydroxylamine treatment (Fig. 8C). There is a correlation between decrease in acylation (Fig. 8C) and loss of enzyme activity (Fig. 8A), suggesting that 100% deacylation might lead to complete loss of GPI-PLC activity. Exposure of GPI-PLC to phosphate-buffered saline neither deacylated nor inhibited the enzyme (data not presented). Finally, when recombinant GPI-PLC expressed in E. coli was subjected to the above deacylation conditions, no inhibition of enzyme activity was observed (data not presented). This result FIG. 6. Acyl-GPI-PLC is active. A, T. brucei (10 9 ) were metabolically labeled with [ 3 H]myristate and lysed hypotonically. Cell lysates were processed as illustrated. B, a membranous pellet of a lysate from [ 3 H]myristate-labeled cells was resuspended in lysis buffer and incubated for 15 min either at 4°C (lanes 1-4) or 37°C (lanes 5-8). In some experiments, Nonidet P-40 was added to 1% (final concentration) before incubation (lanes 3, 4, 7, and 8). Lanes 1, 3, 5, and 7 depict immunoprecipitates of GPI-PLC (from 5 ϫ 10 8 cell eq), and lanes 2, 4, 6, and 8 contain 10 7 cell eq of total proteins. SDS-PAGE (14% minigel) was used to resolve the proteins, which were then detected by fluorography. The dried gel was exposed to Hyperfilm TM (Amersham Pharmacia Biotech) for 5 days.  7. Acylation of GPI-PLC is dynamic. [9,10-3 H]Myristate-labeled trypanosomes (7 ϫ 10 9 cells) were resuspended in 14 ml of labeling medium and incubated for 60 min (37°C). Two-ml aliquots (10 9 cells) were harvested every 10 min, immunoprecipitated, and the eluted GPI-PLC analyzed by SDS-PAGE/fluorography. Dried gels were exposed to Hyperfilm TM (Amersham Pharmacia Biotech) for 12 days. Images were quantitated using an IS-1000 Digital imaging system. The value obtained at t ϭ 0 was assigned a value of 100%.
is consistent with our inability to detect acylation of GPI-PLC in the bacterium. Furthermore, the result indicates that the deacylation conditions per se do not inhibit GPI-PLC.
We conclude that deacylation reduces the enzymatic activity of GPI-PLC significantly and infer that thioacylation activates the enzyme.
Turnover Number of GPI-PLC Is Reduced by Deacylation-A kinetic basis for the reduced rate of GPI cleavage by deacylated GPI-PLC (Fig. 8A) was sought. Rate of the reaction could be decreased by altering either k cat or K m ; therefore, both parameters were determined for acylated GPI-PLC and the deacylated form of the enzyme (Table I). During enzyme-catalyzed reactions, product formation is preceded by assembly of an enzyme⅐substrate complex (E⅐S). At steady state, the Michaelis constant (K m ) equals (k Ϫ1 ϩ k 2 )/k 1 , where k 1 is the kinetic constant for formation of an E⅐S complex. k Ϫ1 is the reaction rate constant for breakdown of the E⅐S complex into free enzyme (E) and substrate (S) (i.e. GPI-PLC and VSG, respectively). k 2 is the rate constant for conversion of an E⅐S complex to product (P) and free enzyme; k 2 equals k cat when k Ϫ1 Ͻ Ͻ k 2 .
Deacylation of GPI-PLC did not have a profound effect on VSG binding by the enzyme, since K m was not altered significantly (Table I). However, when the maximal reaction velocity was corrected for amount of enzyme added (to obtain the turnover number, k cat ), the two enzymes had strikingly different properties. Turnover number decreased 18 -30-fold for deacylated GPI-PLC, as compared with the acylated enzyme (Table I). The properties of deacylated T. brucei GPI-PLC appear to be similar to E. coli-expressed recombinant GPI-PLC which is not acylated and whose turnover number was 10-fold less than GPI-PLC isolated from T. brucei (15).
In essence, acylation alters the properties of the E⅐S complex. The reduced reaction velocity of the deacylated enzyme is traceable to the diminished rate of converting the GPI-PLC⅐VSG complex to products. We surmise that thioacylation of GPI-PLC promotes rapid conversion of an E⅐S complex to product (P) and free enzyme (E). DISCUSSION Thiomyristoylation, a Novel Modification of a Phospholipase C-The use of myristate in thioacylation (i.e. S-myristoylation) of GPI-PLC is unique. Frequently, palmitate is the fatty acid attached to cysteines on proteins (reviewed in Ref. 1). Myristate is typically linked to polypeptides through an amide bond (N-acylation) (1). In T. brucei, myristate is a rare fatty acid (23,FIG. 8. Deacylation reduces activity of GPI-PLC. A, GPI-PLC was immunoprecipitated from 2 ϫ 10 9 trypanosomes with 100 l of mc2A6-6⅐protein A-Sepharose complex, which was then treated with neutral hydroxylamine (see "Deacylation of GPI-PLC and Determination of Phospholipase C Activity" under "Experimental Procedures") for the indicated periods. GPI-PLC was eluted, and activity of the enzyme was determined. Activity obtained at time 0 was assigned a 100% value. B, decrease in activity of GPI-PLC (A) is not due to loss of the polypeptide. Western blot analysis was performed on the deacylated samples using R18B3, a polyclonal anti-GPI-PLC antibody. In general, these enzymes are soluble, only making transient contact with membranes to hydrolyze their membrane-bound substrates, usually in response to extracellular signals (reviewed in Ref. 28). Lipid modification could be used by phosphatidylinositol-specific PLCs as a means of gaining access to membranes. However, no acylation of a phosphatidylinositolspecific PLC has been reported yet.
T. brucei GPI-PLC is the only eukaryotic PLC known to be stably integrated into a lipid bilayer (6 -8). Could modification with lipid be responsible for the membrane association of this enzyme? Current evidence suggests otherwise. 2 When expressed in E. coli, GPI-PLC remains tightly associated with membranes (15), but attempts to label the enzyme metabolically with fatty acids have yielded negative results. Hence, membrane binding by GPI-PLC does not appear to require an eukaryote-specific modification.