A Phospholipase C from Trypanosoma brucei which Selectively Cleaves the Glycolipid on the Variant Surface Glycoprotein*

The surface coat of Trypanosoma brucei is composed of 10’ molecules of the variant surface glycoprotein (VSG). Each VSG molecule is tethered to the cell membrane by a glycolipid moiety which contains 1,Z-di-myristoyl-an-phosphatidylinositol (Ferguson, M. A. J., Low, M. G., and Cross, G. A. M. (1985) J. Biol. Chern. 260, 14547-14555). Following cell lysis, an endoge- nous phospholipase C cleaves dimyristoyl glycerol from the glycolipid, releasing soluble VSG. We have purified this enzyme, which we designate VSG lipase, by detergent extraction, (NH4)&04 frac- tionation, hydrophobic chromatography, and cation exchange chromatography. It is purified 2600-fold and is virtually homogeneous. Based on sodium dodecyl sulfate-polyacrylamide gel electrophoresis, the apparent molecular mass is 37 kDa. In solutions containing the detergent 3-[(3-cholamidopropyl)dimethylam- moniol-1-propanesulfonic acid (CHAPS), the Stokes radius (2.6 nm), szo,w (3.7 S), and V (0.77 cm3/g) of VSG lipase suggest a molecular mass for the native enzyme of about

host's immune response by antigenic variation. During this process, the trypanosome changes its surface coat to a new one composed of a VSG with a different amino acid sequence (reviewed in [2][3][4][5]. Early studies on VSG raised questions about the way this molecule is bound to the cell surface and the mechanisms by which it can be released. It was clear that the VSG binds tightly to the surface of the living trypanosome, yet after cell lysis it is released rapidly as a soluble protein which has no apparent affinity for membranes. Cardoso de Almeida and Turner (6) resolved this paradox when they discovered that boiling trypanosomes in SDS allows isolation of an amphiphilic or membrane form of VSG (mfVSG). In contrast, lysis under nondenaturing conditions causes release of a hydrophilic or soluble form of VSG (sVSG) which lacks a hydrophobic moiety present in the C-terminal region of mfVSG. Cardoso de Almeida and Turner presented evidence that conversion of mfVSG to sVSG is enzyme catalyzed.
Recent structural studies in several laboratories have revealed that mfVSG contains a glycolipid covalently linked to its C terminus which anchors the protein to the plasma membrane. This glycolipid is composed of myristic acid (7), glycerol (8, 9), phosphate (6, lo), inositol (ll), several sugars (12)(13)(14), and ethanolamine (15). Part of the glycolipid is in the form of 1,2-dimyristoyl-sn-phosphatidylinositol which is glycosidically bonded to non-acetylated glucosamine (11,12). Consequently, nitrous acid treatment liberates dimyristoyl phosphatidylinositol (11). The structure of the remainder of the moiety, which contains the sugars and ethanolamine, is not yet known. The glycolipid is linked to the VSG polypeptide through an amide bond between the ethanolamine and the &-carboxyl group of the C-terminal amino acid (15). Conversion of mfVSG to sVSG during cell lysis involves cleavage of a phosphodiester bond and release of 1,2-dimyristoyl-sn-glycerol from the glycolipid moiety (9,16). Therefore, the enzyme involved in this reaction has the specificity of a phospholipase C.
Little is known about this phospholipase. It is present in a particulate fraction of lysed cells (17), and it seems to be inhibited by mercurial compounds and Zn2+ (1,6,(18)(19)(20). The enzyme activity is absent in cultured trypanosomes which resemble the coatless procyclic forms found in the gut of the tsetse fly vector (1,18).
The biological role of the enzyme is not yet known, but Turner and co-workers have discussed several possibilities (1). The most exciting are: (a) the enzyme removes the surface coat when the parasite enters the gut of the tsetse; (b) it participates in antigenic variation by disposing of the old surface coat. In any event, it would seem likely that the enzyme is subject to regulation. It is relatively quiescent in living bloodstream trypanosomes as there appears to be little, if any, VSG release (21). However, the enzyme is present in a latent form as lysis of the cell results in conversion of all of its mfVSG t o sVSG i n a few minutes (22).
To understand the biological role and possible regulation of this enzyme, which we designate VSG lipase, it is essential to obtain it in pure form.
In this paper we describe the purification of VSG lipase and some preliminary characterization of its biochemical properties.

MATERIALS AND METHODS
Trypanosomes-T. brucei (ILTat 1.3; from R. 0. Williams) were isolated from the blood of Swiss mice or Wistar rats (parasitemia about 109/ml) by DEAE-cellulose chromatography (23) using 50 mM NaCl, 5 mM KCl, 55 mM D-ghCOSe, 50 mM Na-Bicine, pH 8.0 (BBS) (24). After centrifugation (10 min, 4 "C) using a Sorvall HB-4 rotor (2500 rpm) or GS-3 rotor (3000 rpm), the cells were washed in icecold BBS (about 3 X 10' cells/ml) and centrifuged again. pH]Myristate-lubeled mfVSG-['HIMyristate-labeled mfVSG, for assay of VSG lipase, was prepared by labeling trypanosomes (2 X lo9/ 10 ml) in vitro with [9,10-3H]myristic acid (12.9 Ci/mmol, 0.1 mCi/ ml, New England Nuclear) (7,24). After 1 h at 37 "C, the cell suspension was cooled to 0 "C and centrifuged (HB-4 rotor, 2,500 rpm, 10 min, 4 "C). The medium was saved as it could be reused without significant loss of labeling efficiency. The cell pellet was washed with 5 ml of BBS at 0 "C and centrifuged (HB-4 rotor, 2,500 rpm, 10 min, 4 "C). The cells were lysed osmotically by resuspending in 4 ml of 10 m M Napi, pH 7.0, containing 1 pg/ml leupeptin, 0.1 mM TLCK, and 5 mM sodium p-chloromercuriphenylsulfonate (added from a 100 mM stock solution in 0.1 M NaOH). Sodium p-chloromercuriphenylsulfonate inhibits VSG lipase (1,18) and sulfhydryl proteases. After 10 min at 0 "C, the lysate was centrifuged (HB-4 rotor, 6,000 rpm, 5 min, 4 "C), and the pellet was washed in 4 ml of the same buffer. The washed pellet, which contained mfVSG, was extracted with 8 ml of CHC13/CHJOH (2:l) at 20 "C and centrifuged (HB-4 rotor, 9,000 rpm, 5 min). The pellet, including the mfVSG, was dissolved in 4 ml of 1% SDS by heating (100 "C, 10 min) and vortexing. To remove remaining 'H-labeled lipids, the solution was extracted twice with 5-8 ml of n-butyl alcohol at 20 "C. For each extraction, the solution was vigorously homogenized and centrifuged (HB-4 rotor, 8,000 rpm, 5 min) to resolve the aqueous and organic phases; the latter was discarded. In a third extraction, 9 ml of H20saturated n-butyl alcohol was used to maintain the aqueous phase. In the fourth extraction, the aqueous phase was eliminated by using 9 ml of n-butyl alcohol (not HzO saturated). The resulting gummy precipitate was recovered by centrifugation (HB-4 rotor, 8,000 rpm, 5 min), washed with anhydrous ether, air dried, and dissolved in 1.5 ml of 1% SDS by mixing and heating (100 "C, 10 rnin). Insoluble material was eliminated by centrifugation (13,000 X g, 5 rnin). The product, as assayed by SDS-PAGE and Coomassie staining (not shown), was greater than 90% pure mfVSG. Fluorography of the gel showed that virtually all the radioactivity was incorporated in mfVSG. The mfVSG concentration was 2 mg/ml, and the specific radioactivity was about 2,800 cpmlpg. VSG Lipase Assay-In this assay the release of [3H]myristatelabeled dimyristoyl glycerol from mfVSG is measured by liquid scintillation counting of a n-butyl alcohol extract of the reaction mixture. This assay is similar to one used by Bulow and Overath (18). Reactions (25 pl) contained 2 pg (about 5600 cpm) of [3H]myristatelabeled mfVSG, 0.04% SDS (introduced with the mfVSG), 1% Nonidet P-40, 5 mM EDTA, and 50 mM Tris-HC1, pH 8.0. If necessary, enzyme was diluted in 1% Nonidet P-40, 5 mM EDTA, 50 mM Tris-HCl, pH 8.0. After addition of enzyme and incubation at 37 "C for 30 min, the mixture was thoroughly mixed with 0.5 ml of HzO-saturated n-butyl alcohol. The phases were separated hy brief centrifugation (Fisher Micro-Centrifuge 235B, 1 rnin), and 0.4 ml of the upper phase was counted with 6 ml of Liquiscint (National Diagnostics). One unit of enzyme is defined as the amount which hydrolyzes 0.5 pg of mfVSG under the standard conditions. At all stages of the purification, the production of n-butyl alcohol extractable 3H radioactivity is linear with enzyme concentration for reactions in which up to 0.5 pg of mfVSG is hydrolyzed; therefore, all assays were conducted in this range. The release of n-butyl alcohol extractable radioactivity correlates with the conversion of mfVSG to sVSG as analyzed by SDS-PAGE (not shown). The product of the reaction is [3H]dimyristoyl glycerol as judged by silica gel thin layer chromatography (see Fig. 5, lunes 9 and IO). The identity of dimyristoyl glycerol as the product was confirmed by chromatography using another solvent system (hexane/diethyl ether/glacial acetic acid, 70301). In that system both the reaction product and a 1,2-dimyristoyl-rac-glycerol standard (Sigma) comigrated as a doublet, presumably because of some isomerization to 1,3-dimyristoyl glycerol (not shown).
Gel Electrophoresis and Protein Determinations-SDS-PAGE (25) was performed using 7.5-15% linear gradient polyacrylamide gels. Gels were stained with Coomassie Blue and then, unless noted otherwise, with silver according to Morrissey (26). Using bovine serum albumin as a standard, protein was determined, after precipitation by 10% trichloroacetic acid, as described by Lowry et al. (27) unless noted otherwise. SDS (0.5%) was included to avoid interference by other detergents (28).

Purification of VSG Lipase
Buffers used throughout the purification included 1 gg/ml leupeptin and 0.1 mM TLCK, and operations were performed at 0-4 "C unless specified otherwise. After obtaining trypanosomes, the entire purification procedure can be completed within 3 days. The typical purification described here is summarized in Table I.
Preparation of VSG-depleted Membranes-Trypanosomes (7.2 x 10") were lysed osmotically by suspending in 70 ml of 1 mM EDTA, 10 m M Napi, pH 8.0 (Buffer A) for 15 min. This lysate was designated Fraction I. Following centrifugation (Sorvall HB-4 rotor, 6000 rpm, 5 min), the pellet was resuspended in 70 ml of Buffer A at 37 "C and maintained at this temperature for 5 min to promote the VSG lipase-catalyzed release of membrane-bound VSG. The VSG-depleted membranes were centrifuged (HB-4 rotor, 8500 rpm, 15 min), washed with 70 ml of Buffer A, and centrifuged again.
T h e column was then washed with 25 mM sodium succinate, p H 6.0, containing (NH4),S04 which was decreased linearly in concentration from 50 to 0% of saturation over 100 ml. After further washing with 50 ml of 25 mM sodium succinate, p H 6.0, VSG lipase was eluted with 1% CHAPS (Sigma), 25 mM sodium succinate, pH 6.0 (Buffer C). Fractions (5.3 ml) were collected at a rate of 30 ml/h. As in the typical example shown in Fig. 1, most protein eluted during the wash or in a sharp peak immediately after elution with CHAPS. In contrast, VSG lipase eluted in a broad peak after introduction of CHAPS. The trailing fractions of the activity peak, which contained about 70% of the activity but relatively little pro- fractions. Gel lanes correspond closely to the fraction scale directly above in Panel A. Samples (0.4 ml of selected fractions) were combined with 0.65 ml of water and precipitated (0 "C, 1 h) by addition of 0.15 ml of 50% (w/v) trichloroacetic acid containing 0.2% sodium deoxycholate (included as carrier). The precipitates were collected by centrifugation (Fisher Micro-Centrifuge, 10 min, 4 "C) and dissolved (100 "C, 5 min) in SDS-PAGE sample buffer containing bromphenol blue. If yellow, samples were neutralized with NHa vapor until they became blue. After electrophoresis, proteins were detected by silver staining. Some of the bands (54-61 kDa), found in all fractions and control lanes (not shown), are artifactual. Protein molecular mass markers used are myosin (205 kDa), pgalactosidase (116 kDa), bovine serum albumin (66 kDa), ovalbumin (45 kDa), glyceraldehyde-3-phosphate dehydrogenase (36 kDa), carbonic anhydrase (29 kDa), soybean trypsin inhibitor (20 kDa), alactalbumin (14 kDa); some of these are indicated. The horizontal arrow shows the 37-kDa protein believed to be VSG lipase. tein, were pooled (Fraction IV, 333 ml).
Carboxymethyl-Sephx Chromatography-Fraction IV was applied directly t o a column of carboxymethyl-Sephadex (C-25, Pharmacia, 0.4 cm2 x 12.5 cm) equilibrated with Buffer C. The column was washed with 25 ml of Buffer C and eluted with a 50-ml linear gradient of 0-100 mM NaCl in Buffer C. Fractions (1.5 ml) were collected a t a rate of 15 ml/h. A typical CM-Sephadex fractionation is depicted in Fig. 2. VSG lipase was eluted as a single peak at about 30 mM NaCI. The most active fractions were pooled (Fraction V, 20 ml Samples (0.4 ml of selected fractions) were prepared for electrophoresis as described in the legend of Fig. 1. Following electrophoresis, the gel was silver-stained. Some of the bands (54-61 kDa), detected in all lanes (including controls, not shown), are artifacts often seen in silver-stained gels (50). Other bands seen in most of the lanes (45-75 kDa) may be similar artifacts or minor contaminants of this preparation. Note that they are not apparent in a Coomassie-stained gel of a concentrated pool from a different preparation (Fig. 3, Vn. The molecular weights of marker proteins are indicated. The horizontal arrow shows the 37-kDa protein believed to be VSG lipase. shown in Fig. 3. Both Fraction VI and Fraction VI1 contain a single major 37-kDa polypeptide. No other Coomassie-stained proteins are detectable in either fraction, although trace 48and 64-kDa contaminants of Fraction VI were revealed by silver staining (not shown). The 37-kDa protein cofractionated with activity during chromatography on phenyl-Sepharose (Fig. 1) and carboxymethyl-Sephadex (Fig. 2 ) , sedimentation on a sucrose gradient (Fig. 4A), and gel filtration on Sephacryl S-200 (Fig. 4B). In SDS-PAGE analyses, the intensity of staining of the 37-kDa polypeptide and VSG lipase activity were approximately proportional in almost every case.
Silver staining (Figs. 1-3 ( l a n e IV), and 4 A ; note the amounts of each fraction analyzed) required about lo3 units for detection. (In Fig. 4B, a different, perhaps less sensitive, staining method was used.) Coomassie Blue (Fig. 3 (lanes VI and V I 0 and other data not shown) was sensitive to about IO4 units. We conclude that Fractions VI and VI1 (and also Fraction V which is an unconcentrated form of Fraction VI) are virtually homogeneous VSG lipase.

Stability of VSG Lipase
VSG lipase is very stable. We obtained nearly the same yield of enzyme activity when we used VSG-depleted membranes which had been stored 6 months at -70 "C as we did from freshly isolated trypanosomes. Fractions I1 and I11 could also be stored at -70 "C for several weeks without significant loss of activity. Fraction V could be stored for at least 2 months at -70 "C with little or no loss of activity. A 30% loss of activity occurred upon storage of this fraction for 5 days at 4 "C, but storage in 50% glycerol reduced this loss to 10%.
Incubation of Fraction V for 10 min at 37 "C resulted in little loss of activity, but incubation at 50 "C caused 99% inactivation.  (10% final concentration), washed with 90% acetone, dissolved in SDS-PAGE sample buffer, and electrophoresed. Fraction VII, concentrated by lyophilization, was dissolved in 2% SDS, precipitated with 8 volumes of acetone (-20 "C, overnight), and centrifuged to recover the precipitate. These steps were repeated twice more to eliminate an oily CHAPS residue. Fractions are indicated above the gel lanes. Fractions I, 11, 111, VI, and VI1 were from the purification summarized in Table I;

Hydrodynamic Properties of VSG Lipase
Hydrodynamic measurements were taken in the presence of CHAPS which may bind to VSG lipase. Since CHAPS has a partial specific volume (V) of 0.81 cm3/g (29), a VSG lipase-CHAPS complex could have a t that is higher than that of most soluble proteins (usually 0.73-0.74 cm3/g). Therefore, we determined t and s20,u, by the method of Neer (30). In this method, the sedimentation of VSG lipase and standard proteins of known szo,, and t is measured in sucrose gradients prepared in H 2 0 (Fig. 4 . 4 ) and D20 (not shown). The szo,w of VSG lipase is 3.7 k 0.1 S and the V is 0.77 k 0.01 cm3/g. This rather high value for t suggests that CHAPS does bind to VSG lipase.
We determined the Stokes radius of VSG lipase (in the presence of CHAPS) by gel filtration on Sephacryl S-200 (Fig.  4B). As described by Siege1 and Monty (31), the Stokes radius was determined to be 2.6 nm from a plot of uersus the radii of standards.
Based on these hydrodynamic parameters, the molecular mass of the native VSG 1ipase.CHAPS complex, calculated from the Svedberg equation as described (31), is 47 -+ 3 kDa.
The frictional ratio ( f/fo) of 1.1 suggests a globular conformation.
These results indicate that VSG lipase (Fraction V) is a single polypeptide of 37 kDa. It is probably complexed with CHAPS. A multimeric structure would be inconsistent with the calculated molecular mass of the native enzyme and also with its elution position on the gel filtration column (between ovalbumin, 45 kDa, and trypsin inhibitor, 20 kDa, Fig. 4B).

Other Properties of VSG Lipase
The effects of several reagents on VSG lipase activity are shown in Table 11. EDTA (5 mM), and EGTA (5 mM) to a lesser degree, stimulated the activity; ZnClz (5 mM) completely inhibited VSG lipase, whereas CaClz (5 mM) and MgClz (5 mM) were only slightly inhibitory. KC1, NaC1, and (NHJ2SO4, at 125 mM and above, were moderately inhibitory. pCMPS at 5 mM completely inactivated VSG lipase; at this same concentration, N-ethylmaleimide was moderately inhibitory and iodoacetamide had no effect. Dithiothreitol (25 mM) stimulated the reaction about 6-fold.

TABLE I1
Effect of various reagents on VSG lipase activity Assays were conducted under the standard conditions except that the indicated reagents were added. In Experiment 1, EDTA was omitted from the standard conditions. Enzyme (Fraction V, 0.1-0.7 units) was added last, and assays were initiated without preincubation. Control assays, without added reagent, were designated 100%. The enzyme (Fraction V) was optimally active between pH 7.5 and 8.5 using Tris-HC1, Napi, or Na-Bicine. Half of the maximum activity was obtained at pH 6.5 using either Napi or sodium succinate and at pH 9.5 using sodium glycinate. There was little or no activity between pH 4.5 and 6.0 using either sodium citrate or sodium succinate buffers. All buffers used in studies on the effect of pH were 50 mM. These results agree with those obtained by Turner et al. (1) with impure VSG lipase.
Substrate Specificity of VSG Lipase VSG lipase is a phospholipase C. To examine its specificity in more detail, we tested several different substrates and analyzed the reaction products by thin layer chromatography (Fig. 5). Of all of the [3H]myristi~ acid-labeled lipids in a CHC13/MeOH (2:l) extract of [3H]myri~tate-labeled trypanosomes, only one, lipid A, was cleaved by VSG lipase to yield a compound which comigrated with dimyristoyl glycerol (lanes  1 and 2). Lipid A is a biosynthetic precursor of the VSG glycolipid, and it has a structure closely related to that of the VSG glycolipid.' Although the identities of the other labeled compounds in the extract are not known, this result suggests that VSG lipase has a narrowly defined specificity. Furthermore, only a small amount of cleavage of 1,2-dimyristoyl-snphosphatidylinositol to produce dimyristoyl glycerol was detected (lanes 3 and 4 ) ; the amount produced was proportional to the amount of enzyme added (not shown). We detected no cleavage of 1-stearoyl-2-arachidonoyl-sn-phosphatidylinosi-to1 (lanes 5 and 6 ) . In reactions containing mixtures of mfVSG and either of the phosphatidylinositols, mfVSG was completely hydrolyzed while neither phosphatidylinositol was significantly hydrolyzed (compare lanes 7 and 8 with lanes 3 and 5, respectively). Identical results (not shown) were obtained when reaction mixtures were transferred to new tubes prior to n-butyl alcohol extraction, indicating that the substrates were in solution during the incubation with the enzyme. The tentative conclusion from these experiments is that VSG lipase is highly specific. It cleaves the VSG glycolipid and structurally related compounds including its glycolipid precursor (lipid A) and, to a lesser extent, dimyristoyl phosphatidylinositol. Nevertheless, this apparent specificity of VSG lipase could be dependent on the conditions we employed and could be affected by other factors such as the presence of detergents or other lipids (32). The possibility that other coextracted lipids stimulated lipid As cleavage (Fig. 5, lanes 1  and 2) is unlikely since lipid A purified by thin layer chromatography was also cleaved efficiently (not shown).

DISCUSSION
VSG lipase is a phospholipase C of the bloodstream form of T. brucei which catalyzes the hydrolysis of the VSG glycolipid. It converts mfVSG to sVSG and releases dimyristoyl glycerol. Based on SDS-PAGE, the nearly homogeneous enzyme has an apparent subunit molecular mass of 37 kDa. Its Stokes radius of about 2.6 nm and apparent s20,w of 3.7 S indicate a molecular mass for the native enzyme of about 47 kDa (this value could include bound CHAPS). Therefore, the purified VSG lipase appears to be monomeric.
The detergent CHAPS, a zwitterionic derivative of cholic acid, plays an important role in the purification. In early trials with Nonidet P-40, n-octyl glucoside, or no detergent, we were unable to bind VSG lipase to ion exchange resins over a broad pH range. Only CHAPS permitted binding to carboxymethyl-   (lanes 7-10). The amount of VSG lipase used completely hydrolyzed the mfVSG (lanes 7, 8, and IO) as confirmed by counting the aqueous phase after n-butyl alcohol extraction.
Sephadex at pH 6.0. In preparation for carboxymethyl-Sephadex chromatography, we therefore eluted VSG lipase from phenyl-Sepharose with 1% CHAPS at pH 6.0. To our surprise, VSG lipase was among only a small number of proteins which eluted in a broad peak after the major protein peak (Fig. 1). Perhaps these proteins, including VSG lipase, are integral membrane proteins. Some VSG lipase activity (25-35%) typically elutes with the majority of proteins as a sharp peak immediately following introduction of CHAPS (Fig. 1). This activity is probably the same as that which elutes in the broad main activity peak because it behaves identically on CM-Sephadex, coeluting precisely with appropriate amounts of the 37-kDa polypeptide. Dithiothreitol had a surprisingly large stimulatory effect on VSG lipase. At 1 mM it stimulated 2-fold and at 25 mM it stimulated more than 6-fold (Table 11). Dithiothreitol had the same effect on freshly prepared Fraction I and other fractions stored for prolonged periods at -70 "C (not shown). Thus, this stimulatory effect does not seem to reflect progressive reversible inactivation of VSG lipase during purification. We do not know if dithiothreitol acts directly on the enzyme; instead, it may convert mfVSG to a more effective substrate by reducing disulfide bonds likely to be near its C terminus (3,33).
From the specific activity of the homogeneous enzyme ( The cellular location of VSG lipase is not yet known. Its appearance in the particulate fraction of lysed trypanosomes and extractability by non-ionic detergent suggests that VSG lipase is associated with a membrane. Since VSG is found on the cell surface, VSG lipase could be associated with the plasma membrane. Support for this location came from recent studies in our laboratory on the biosynthesis of VSG (34). In pulse-chase experiments, we found that the transport of VSG to the cell surface and to a cellular compartment containing VSG lipase had comparable kinetics. Nevertheless, it remains possible that VSG lipase resides in an internal membrane which fuses with the plasma membrane only after a triggering event such as cell lysis. Immunoelectron microscopy, using VSG lipase-specific antibodies, should permit localization of the enzyme.
Phospholipase Cs with different specificities have been described in many cell types. VSG lipase is potentially related to those phospholipase Cs which specifically hydrolyze phosphatidylinositol (reviewed in 35) since this moiety is part of the VSG glycolipid. It is of great interest that PI-PLCs are believed to generate second messengers within many cells in response to a wide variety of extracellular stimuli (36). Eukaryotic PI-PLCs have been found mainly in cytosolic fractions, although there are lower levels in particulate fractions. Bacterial PI-PLCs have been purified from culture filtrates. VSG lipase differs from most of these other enzymes in that it is found exclusively in the particulate fraction. In addition, VSG lipase has a pH optimum of 7.5-8.5; most PI-PLCs are optimally active at neutral and acidic pH. Similar to the bacterial PI-PLCs but unlike the majority of mammalian enzymes, VSG lipase does not require Ca2+ for activity.
Is VSG the true substrate of the enzyme which we have purified, or is its hydrolysis in vitro without biological significance? We must consider the latter possibility since the action of VSG lipase in trypanosomes has been detected only under nonphysiological circumstances (e.g. cell lysis). However, our preliminary investigation of substrate specificity suggests that VSG lipase could be highly specific. VSG and its biosynthetic precursor, lipid A, are hydrolyzed; 1,2-dimyristoyl-sn-phosphatidylinositol is also hydrolyzed but much less efficiently. We could detect no cleavage of l-stearoyl-2arachidonoyl-sn-phosphatidylinositol or any of the [3H]myristate-labeled lipids (other than lipid A) in a CHCb/MeOH (2:1) extract of trypanosomes. It seems possible that VSG lipase recognizes features of its glycolipid substrate such as carbohydrate or fatty acid structure. The activity and specificity of phospholipases are often highly sensitive to assay conditions (e.g. detergent concentration and the presence of lipid activators), and therefore more studies will be needed to prove our tentative conclusion that VSG lipase is highly specific for VSG. If this conclusion is true, it implies that this enzyme plays an important role in VSG metabolism.
Several membrane proteins in other eukaryotes are probably anchored to membranes by phospholipid moieties. Among parasitic protozoa, these include a 195-kDa surface antigen of Plasmodium falciparum (37) and an abundant 63-kDa myristylated surface glycoprotein of Leishmania major (38, 39). In higher eukaryotes, acetylcholinesterase (41), alkaline phosphatase (42), 5"nucleotidase (43,44), Thy-1 glycoprotein (45,  46), and decay-accelerating factor (47) appear to be anchored to membranes by structures containing phosphatidylinositol and, thus, can be released by exogenously added PI-PLCs (reviewed in 40). The Leishmania glycoprotein (cited in 40) and acetylcholinesterases from Torpedo and humans (48) can be cleaved by VSG lipase. Remarkably, VSG lipase treatment of both acetylcholinesterases exposes an antigenic determinant which is also found on trypanosome VSG (48). In some of these cases endogenous activities similar to VSG lipase may exist (38, 49).
The biological significance of phosphatidylinositol-containing glycolipids on these diverse proteins is not yet understood, and in some cases investigation will be hindered by limiting amounts of material. Because massive amounts of VSG can be purified from trypanosomes, and because VSG lipase can now be readily obtained in homogeneous form, this parasite may be the system of choice for revealing the structure, function, and significance of the enzymatic cleavage of these protein-bound glycolipids.