Lipid Analysis of the Glycoinositol Phospholipid Membrane Anchor of Human Erythrocyte Acetylcholinesterase PALMITOYLATION OF INOSITOL RESULTS IN RESISTANCE TO PHOSPHATIDYLINOSITOL-SPECIFIC PHOSPHOLIPASE C*

The glycoinositol phospholipid membrane anchor of human erythrocyte acetylcholinesterase (EC 3.1.1.7) contains a novel inositol phospholipid which in this and the accompanying paper (Roberts, L., S., Reinhold, N., and Rosenberry, T. L. (1988) J. Bio2. Chem 263, 18776-18784) is shown to be a plas- manylinositol that is palmitoylated on the inositol ring. The inositol phospholipid was radiolabeled with the photoactivated reagent 3-(trifluoromethyl)-3-(rn-[1z~I] iodopheny1)diazirine and characterized by various chemical and enzymatic cleavage procedures whose products were analyzed by thin layer chromatography and autoradiography or gas chromatography. Acidic methanolysis of human erythrocyte acetylcholinesterase (E““ AChE) revealed 18:O or on a

The inositol phospholipid is linked to the anchor by a glucosamine in glycosidic linkage, and deamination with nitrous acid cleaved the glycosidic linkage and released the phospholipid. The deamination and acetolysis products from EhU AChE were purified by high performance liquid chromatography, and fatty acid analysis following acidic methanolysis of the purified products revealed that 2 fatty acid residues were associated with the deamination product and only one with the alkylacylglycerol acetolysis product. The other fatty acid residue was primarily palmitate and was indicated to be in ester linkage to an inositol hydroxyl(s). This linkage was shown to be responsible for the resistance of the inositol phospholipid to cleav- ** To whom correspondence should be addressed. age by Staphylococcus aureus phosphatidylinositolspecific phospholipase. Deacylation of the inositol phospholipid deamination product by treatment with base removed this palmitoyl group and facilitated release of alkyl-and alkylacylglycerol species by phosphatidylinositol-specific phospholipase C with concomitant formation of inositol 1-phosphate. In contrast, digestion of EbUAChE with a recently reported anchorspecific phospholipase D resulted in release of plasmanic acids from the intact palmitoylated plasmanylinositol. Acetylcholinesterase (EC 3.1.1.7) exists in multiple molecular forms which are widely distributed in many tissues (for reviews, see MassouliB and Bon, 1982;Rosenberry 1985). These diverse molecular forms possess similar catalytic and antigenic properties but differ in their extent of oligomeric assembly and their mode of attachment to the cell surface. AChE' on the surface of mammalian erythrocytes is an amphipathic globular dimer (G2) (Ott and Brodbeck, 1978;Dutta-Choudhury and Rosenberry, 1984). These erythrocyte G2 forms as well as the Gz forms that are abundant in Torpedo electric organ and insect heads are anchored in the plasma membrane by a covalently attached glycoinositol phospholipid Futerman et at., 1985a;Gnagey et al., 1987;Haas et al., 1988). Similar glycolipid anchors have been identified in many proteins with an extracellular orientation (Low, 1987;Ferguson and Williams, 1988) including trypanosome variant surface glycoproteins (Ferguson et al., 1985a), rodent Thy-1 (Tse et al., 1985), alkaline phosphatase (Low et al., 1987), decay accelerating factor (Davitz et al., 1986;Medof et al., 1986), merozoite and schizont proteins from Plasmodium fakiparum (Braun-Breton et al., 1988;Haldar et al., 1985;Haldar et al., 1986), Sgpl and Sgp2 from squid brain , and scrapie prion protein (Stahl et al., 1987). Common structural features of glycoinositol phospholipid anchors are an ethanolamine residue in amide linkage to the C terminus of the protein, phosphate groups, a mannose-containing glycan, and a nonacetylated glucosamine res- The abbreviations used are: AChE, acetylcholinesterase; EbO, bovine erythrocyte; Ehu, human erythrocyte; Gz, a globular dimeric AChE form; GLC, gas-liquid chromatography; GC-MS, gas-liquid chromatography-mass spectrometry; HPLC, high performance liquid chromatography; [1261]TID, 3-(trifluoromethyl)-3-(rn-['261]iodo-pheny1)diazirine; PIPLC, phosphatidylinositol-specific phospholipase C; VSG, variant surface glycoprotein from T. brucei. PIPLC-resistant Inositol  idue in glycosidic linkage to an inositol phospholipid (Ferguson and Williams, 1988).
One characteristic of most glycoinositol phospholipid-anchored proteins is their conversion from membrane-bound amphipathic forms to soluble hydrophilic forms by treatment with purified PIPLCs from bacteria or anchor-specific phospholipase C from trypanosomes (Slein and Logan, 1965;Ikezawa et al., 1976;Low and Finean, 1977a;Bulow and Overath, 1986;Hereld et al., 1986;Fox et al., 1986). Many proteins have been classified as having such anchors solely on the basis of their susceptibility to this conversion. However, a subclass of glycoinositol phospholipid anchors is apparently resistant to the action of PIPLC. AChE GZ forms can be released from ox, pig, and rat erythrocytes by the action of PIPLC, but AChE on mouse and human erythrocytes is largely resistant to cleavage by this enzyme (Low and Finean, 197713;Futerman et al., 1985b). The structural basis for the resistance of Eh" AChE to PIPLC resides in the anchor structure itself since the highly purified protein is >90% resistant to the action of PIPLC . This conclusion was supported by analysis of anchor fragments labeled with the photoactivatable lipophilic reagent ['2SI]TID, a radiolabel selective for the lipid portion of Ebo and Eh" AChE membrane anchors which has little effect on lipid TLC mobilities (Roberts and Rosenberry, 1986;Roberts et al., 1987).
Treatment of anchored proteins with nitrous acid deaminates the anchor glucosamine and cleaves its glycosidic linkage to inositol (see Scheme 1; Ferguson et al., 1985a). Deamination of [''51]TID-labeled AChEs released a radiolabeled product from Ebo AChE which comigrated with phosphatidylinositol on TLC, but the corresponding product from Eh" AChE had a mobility much greater than phosphatidylinositol . This novel deamination product from Eh" AChE contained myo-inositol and fatty acids and therefore appeared to be a modified form of phosphatidylinositol that is resistant to PIPLC. In this paper the structural basis for the PIPLC resistance of Eh" AChE was pursued by several approaches which are summarized in Scheme 1. First, since alkylacylglycerols were identified in EbO AChE (Roberts et al., 1988a), acidic methanolysis was used to liberate alkylglycerols from Eh" AChE for analysis by TLC and GLC. Second, acetolysis, a procedure that cleaves the same phosphodiester bond as PIPLC digestion, was used to generate diradylglycerol' acetates from Eh" AChE for further analysis. Third, the fatty acid composition of the acetolysis and deamination products from Eh" AChE were compared. Fourth, the structure of the deamination product was investigated in more Diradylglycerol is glycerol with two 0-linked acyl, alkyl, or alkenyl substituents. Similarly, plasmanic acid is glycerol phosphate with one 0-linked alkyl and one 0-linked acyl group on glycerol, and plasmanylinositol is the corresponding analog of phosphatidylinositol. detail using a combination of chemical and enzymatic techniques. Finally, both Eh" and Eb AChE were examined for susceptibility to a recently reported anchor-specific phospholipase D (Davitz et al., 1987;Low and Prasad, 1988). In the accompanying paper (Roberts et al., 1988b), fast atom bombardment mass spectrometry was employed for further structural characterization of the Eh" AChE anchor and its deamination product.

EXPERIMENTAL PROCEDURES
Preparation of AChEs-Eb and Eh" AChE were prepared by affinity chromatography and, where required, enzyme was depleted of Triton X-100 detergent by a second cycle of affinity chromatography . Moles of catalytic subunit were determined either from the myo-inositol content, measured here as 1.02 mol of inositol/ mol subunit (also see Roberts et al., 1987), or from the enzyme activity in a modified Ellman assay, assuming 410 units/nmol of catalytic subunit (Rosenberry and Scoggin, 1984). Samples of Eb" and Eh" AChE were radiolabeled with ["51]TID (Amersham Corp.) as described (Roberts and Rosenberry, 1986).
Acidic Methanolysis for Fatty Acid and Alkylglycerol Analysis-To a dried sample was added 1 M anhydrous methanolic HCl (100 pl), and the reaction mixture was heated at 65 "C for 16 h. For analysis only of fatty acid methyl esters, samples were extracted with 2,2,4trimethylpentane (Roberts and Rosenberry, 1985). For combined analysis of fatty acid methyl esters and alkylglycerols, chloroform (200 pl) and water (75 pl) were added to the sample, and, after vortexing, the lower organic phase was recovered. Samples for subsequent analysis by TLC were dried at this point. For GLC analysis, the aqueous phase was reextracted with chloroform (100 pl), and the combined organic extracts were acetylated by incubation with acetic anhydride/pyridine (l:l, 10 pl) for 30 min at 80 "C, dried, and resuspended in 2,2,4-trimethylpentane (10 pl).
After drying the reaction mixtures in uacuo, chloroform/methanol (2:1,200 pl) and water (50 pl) were added, the samples were vortexed, and the lower phases were removed and dried. Samples of detergentdepleted ['Z51]TID-labeled Eh" AChE for subsequent GLC analysis (65-85 nmol) were dried and treated in similar fashion except on a 10-fold larger scale. The dried samples were resuspended in acetonitrile (1 ml), and the product was extracted with three portions of hexane (1 ml each). The combined hexane phases were dried, resuspended in hexane/2-propanol (HPLC grade, 99.85:0.15), and chromatographed on a 4.6 X 250-mm Supelcosil silica HPLC column (Supelco, Inc., Bellefonte, PA) in the same solvent at a flow rate of 1 ml/min. The UV absorbance at 206 nm was monitored, and 1-min fractions were collected for 1261 determination.
Base Methanolysis of Eh" AChE Acetolysis Fragments-Dried samples of the HPLC-purified products generated from EhU AChE by acetolysis were treated with 1 M sodium methoxide in methanol/ toluene (3:2, 100 pl) for 15 min at 25 "C. Water (50 pl) was added, and after vortexing the upper phase was recovered. The lower aqueous phase was extracted with chloroform (100 pl) and after vortexing the lower phase was removed, combined with the previously removed upper phase, and dried for analysis. Nitrous Acid Deamination of E"" AChE-Detergent-depleted samples of ['251]TID-labeled Eh" AChE (15-50 nmol) were dialyzed against water and reduced in volume in a Speedvac concentrator. To the enzyme in 1.2 ml of 0.1 M sodium acetate (pH 3.5) was added 0.3 ml of 1 M sodium nitrite, and the pH of the solution was adjusted to 4.0 by the addition of 6 N HCl(6 pl). After incubation of the mixture at 50 "C for 4 h, the phospholipid product was removed by extraction with 4 ml of chloroform/methanol (2:l) and two 3-ml portions of chloroform. The combined organic phases were dried and yielded 50-60% of the initial lz5I radioactivity. The deamination product was purified by isocratic normal phase HPLC on a Supelcosil silica column using 2-propanol/hexane/water (860.4) mobile phase (Hax and Geurts van Kessel, 1977) at a flow rate of 0.5 ml/min. The absorbance at 206 nm was monitored, and 0.5-ml fractions were collected. The major peak of radioactivity (retention time 23 min) contained 60-75% of the radioactivity applied to the column. PIPLC Digestion of the E"" AChE Inositol Phospholipid-Samples of the Eh" AChE deamination product purified by HPLC were divided equally into five aliquots for TLC or four aliquots for GC-MS analysis. Two aliquots from each group were treated with 100 mM KOH in methanol (50 pl) for 30 min at 25 "C. Two additional aliquots for TLC were treated with NH3-saturated methanol (100 pl) for 2 h at 65 "C. After drying, one KOH-treated sample, one NH3-treated sample and one untreated aliquot were digested with purified Staphylococcus aureus PIPLC (Low, 1981; 5 pg/ml final concentration) in 20 mM sodium phosphate and 0.1% sodium deoxycholate (500 pl) adjusted to pH 7.4 for 90 min at 37 "C. A fourth aliquot of deamination product was hydrolyzed in 6 N HCl for myo-inositol analysis. Lipids were extracted with chloroform/methanol (2:1, 2 ml) and dried prior to TLC analysis. Samples for subsequent GLC or GC-MS analysis of myo-inositol or inositol 1-phosphate were dried in vacuo and converted to trimethylsilyl ethers .

Phospholipase D Digestion ofAChEs-Samples of ['251]TID-labeled
Eb" and Eh" AChE in 50 mM Tris, 0.05% Triton X-100, and 2.5 mM CaClz (pH 7.4; 100 pl) were incubated at 37 "C for 1 h with 1 pl of phospholipase D from rabbit plasma (Low and Prasad, 1988). After incubation the samples were extracted with 300 pl of 50 mM HCl in methanol/chloroform (21), and the lower phase which contained lZ5Iradiolabeled products was removed and dried prior to analysis by TLC.
TLC-Thin-layer chromatography was performed on 10 X 20-cm Silica Gel GF plates (Fisher) without activation. Development was either with solvent A (hexane/2-propanol(964)), solvent B (hexanel diethyl ether/acetic acid (6030:1)), or solvent C (chloroform/pyridine/formic acid (50307)). The positions of radioactive components were identified by autoradiography with Kodak XAR-5 film. The positions of standards were located by exposure of dried plates to iodine vapor. The presence of ['251]TID label alters the TLC mobility of lipid components only slightly if at all .
GLC-Analyses were performed on Hewlett Packard 5880 or 5890A capillary column gas chromatographs equipped with a flame ionization detector. Fatty acid methyl esters and alkylglcyerol diacetates generated from Eh" AChE and its fragments by acidic methanolysis were analyzed either on a SP-2380 (polar) fused silica capillary column (15 m X 0.32 mm inner diameter, Supelco Inc.) (Method 1) or on a DB-1 (nonpolar) capillary column (15 m X 0.32 mm inner diameter, J & W Scientific, Folson, CA) (Method 2) employing temperature programing from 100 to 260 "C at 10 "C/min with splitless injection and helium carrier gas at 4 p.s.i. A molar response factor for 180 and 181 alkylglycerol diacetates of 1.16 relative to the 17:O fatty acid methyl ester internal standard was determined empirically. HPLC-purified a and p products from acetolysis were examined on an SE-54 (nonpolar) bonded phase column (8 m X 0.32 mm inner diameter) with on-column injection and hydrogen carrier gas at 6 p.s.i. The oven temperature was programed from 40 to 150 "C at 30 "C/min after 0.5 min, then to 230 "C at 20 "C/min, to 280 "C at 10 "C/min, and to 340 "C at 5 "C/min (Method 3) (Myher and Kuksis, 1984a). These products also were analyzed on a RTx-2330 (polar) column (15 m X 0.32 mm inner diameter, Restek Corp., Port Matilda, PA) operated isothermally at 260 "C with split injection (split ratio 79) and hydrogen carrier gas at 3 p.s.i. (Method 4) (Myher and Kuksis, 198413). Fatty acid methyl esters and alkylglycerol acetates derived from acetolysis products were analyzed on a RTx-2330 column that was temperature-programed from 100 to 180 "C at 20 "C/ min and then to 240 "C at 5 "C/min with split injection and hydrogen carrier gas (Method 5). Inositol was quantified in detergent-depleted samples as previously described using Method 2 .

RESULTS3
Alkylglycerols in Ehu AChE-The photoactivated reagent ['251]TID has been shown to label the lipid groups in the glycoinositol phospholipid anchor of Ebo AChE with high selectivity. Greater than 90% of the incorporated label could be released by purified S. aureus PIPLC  and was associated with alkylacylglycerols whose individual molecular species were quantified by GLC analysis (Roberts et al., 1988a). Since Eh" AChE is largely resistant to PIPLC, Portions of this paper (including part of "Results," Figs. 7-9, and Tables 4-7) 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. alternate methods of chemical cleavage and analysis of alkyland alkylacylglycerols were employed. Samples of Eb" and E' " AChEs were labeled with ['251]TID and subjected to acidic methanolysis under conditions which cleave oxyester, thioester, and phosphodiester bonds. The radiolabeled fragments were extracted and analyzed by TLC. We have previously shown that under these conditions fatty acid methyl esters are released from both AChEs and that Eh" AChE contains considerably more palmitic acid than Eb AChE (Roberts and Rosenberry, 1985;Roberts et al., 1987). In the current experiments the more polar solvent chloroform was used in place of 2,2,4-trimethylpentane to extract quantitatively alkylglycerols as well as fatty acid methyl esters from the reaction mixture. The TLC autoradiograph in Fig. 1 indicates that acidic methanolysis liberates both fatty acid methyl esters and alkylglycerols from Ebo and Eh" AChEs.
The alkylglycerol content of Eh" AChE was quantitated by GLC after acidic methanolysis as described under "Experimental Procedures.'' Two major alkylglycerol species, l-stearylglycerol and 1-oleylglycerol, were observed at levels of 0.55 and 0.20 mol/mol of myo-inositol, respectively. No dimethylacetals, products of acid hydrolysis of alkenylglycerols, were detected. A profile of the fatty acids and alkylglycerols released from Eh" by acidic methanolysis is shown in Fig. 2.
Acetolysis of f251]TID-labeled Ebo and EhU AChEs-The identification of alkylglycerol and fatty acid components in Eh" AChE suggested that the anchor of this enzyme, like that of Ebo AChE, contains alkylacylglycerols. To test this hypothesis, acetolysis was investigated as an alternative to PIPLC cleavage. Acetolysis cleaves the phosphodiester bond in glycerophospholipids between phosphate and glycerol and generates a diradylglycerol acetate (Renkonen, 1965). This cleavage site is identical to that of PIPLC. Acetolysis of ['251]TID-labeled Eb and Eh" AChE permitted extraction of 72 and 73% of the radiolabel, respectively, into the lower chloroform phase. TLC autoradiographic patterns of these acetolysis products from the two enzymes were identical (Fig.   3A). The mobility of the major radioactive spot (p), near that of the unlabeled 1,3-dioleoylglycerol acetate standard (e), was consistent with the migration of 1,2-alkylacylglycero1 acetates (Renkonen, 1965). An additional unidentified spot (a) with greater mobility was generated from both enzymes by acetolysis. This additional product has not been reported previously (Renkonen, 1965;Ferguson et ul., 1985b) nor was a corresponding spot observed when acetolysis was performed on unlabeled phosphatidylinositol (data not shown). HPLC analysis of the acetolysis products (see Fig. 7, Miniprint Supplement) resolved a from the major peak p, which comprised 60-75% of the total products. The separated products a and p were treated with sodium methoxide in methanol/ toluene, and the products of this reaction were extracted with chloroform and analyzed by TLC and autoradiography (Fig.   3B). Lunes 1 (a) and 2 (p) both contained major radioactive spots which migrated as fatty acid methyl esters. Lane 1 indicates that base methanolysis of a also gave an unidentified spot with a RF about 80% of that of the fatty acid methyl ester standard. In contrast, methanolysis of p in lane 2 revealed a radioactive spot which migrated with the alkylgly-cero1 standard. Fig. 3B supports the assignment of p as an alkylacylglycerol acetate and indicates that the unidentified acetolysis product a does contain fatty acids and another component which is not an alkylglycerol.
Samples of / 3 were analyzed directly by GLC on both nonpolar and polar capillary columns. A relatively simple profile was obtained on a nonpolar column (Fig. 4) where alkylacylglycerol acetate species were resolved solely on the basis of their carbon number. Most of the alkylacylglycerol acetates in p (83%) corresponded to species with 40 carbon atoms in their alkyl and acyl chains ( Table 4, Miniprint Supplement). However, several peaks were resolved on a polar GLC column (Fig. 9, Miniprint Supplement), indicating considerable heterogeneity in the alkylacylglycerol acetates. Details of the alkylacylglycerol acetate molecular species analysis are presented in the Miniprint Supplement, and a summary is pro- AChE alkylacylglycerol acetates isolated by HPLC. Samples of diradylglycerol acetates were released from Eh" AChE by acetolysis, and the alkylacylglycerol acetate fraction (0) isolated by HPLC was subjected to analysis by GLC Method 3. Peaks were identified by reference to standard alkylacylglycerols with combined alkyl and acyl chain carbon numbers of 36 and 40. The number above each peak indicates the total number of carbon atoms in the alkyl and acyl chains of the species in each peak. Retention times and mole percents are summarized in Table 4 (Miniprint Supplement).  (Roberts and Rosenberry, 1985;Roberts et al., 1987). It was unclear whether or not these fatty acids were associated exclusively with a glycoinositol phospholipid. To estimate the quantity of fatty acids located outside this portion of the enzyme, Eh AChE was cleaved with PIPLC to remove the diradylglycerols (Roberts et al., 1987), and Eh" AChE was digested with papain to remove the entire glycolipid with the C-terminal dipeptide Rosenberry, 1985, 1986). The resulting AChEs were repurified and analyzed for residual fatty acids (Table 11). Complete removal of the glycolipid from Eh and 226 (n-3)) 11.0

PIPLC-resistant Inositol
The alkyl chains, presumably in the sn-1 position, are indicated by a prime and shown on the left. The values for sn-1,2-and sn-1,3isomers which were generated by acetolysis have been summed.

180' 22.4 (n-6)
32.4 The fatty acid content of the PIPLC-digested catalytic subunit of Eb" AChE (3-7 nmol) was determined by GLC Method 2 after acidic methanolysis, and the moles of fatty acids per mol of AChE were calculated based on myo-inositol content. The results are the mean and standard error of three determinations performed on three distinct enzyme preparations.
* The fatty acid content of the papain-digested catalytic subunit of Eh" AChE (14 nmol) was measured by GLC after acidic methanolysis and the moles of fatty acids per mol of AChE were caIculated based on enzyme activity as previously described (Roberts and Rosenberry, 1985). ND, not detected. see Schmidt, 1983;Sefton and Buss, 1987). Other acyl groups including stearic and oleic acids in addition to palmitic and myristic acids can be involved in convalent linkage to protein (Schmidt et al., 1979), consistent with Table 11. After correcting the total fatty acids per mol of Eh" AChE monomer (2.02 mol) for fatty acids not associated with the anchor (0.26 mol), we conclude that about 1.8 mol of fatty acids are associated with the Eh" AChE glycoinositol phospholipid anchor. Since only about one fatty acid is accounted for in the alkylacylglycerol analysis above, roughly 0.8 fatty acid residues must be present at another site on the glycoinositol phospholipid anchor of Eh" AChE.
To identify and localize these additional fatty acids, fatty acid compositions of lipophilic fragments generated from Eh" AChE by chemical cleavage at two sites are compared in Table 111. Nitrous acid deamination, which cleaves the C-1 glycosidic bond of the anchor glucosamine to generate an inositol phospholipid (Roberts et al., 1987), released a product whose fatty acid composition very closely resembled that of intact Eh" AChE. In contrast, the fatty acids released in alkylacylgIycerols (p) by acetolysis lacked most of the palmitate and small amounts of 14:0,161,181, and 182 fatty acids found in the intact enzyme, but the quantities of polyunsaturated fatty acids in the released acetolysis product were equal to or greater than those in either the released deamination product or the intact enzyme. The results indicate that an inositol hydroxyl group(s) is acylated, primarily with palmitate but also with smaller amounts of other 14, 16, and 18 carbon-containing fatty acids.

Basic Methanolyssis Abolishes the Resistance of the Ehu AChE
Deamination Product to PPLC-We have noted previously that an endogenous modification of the inositol phospholipid in Eh" AChE correlates with its resistance to cleavage by PIPLC . In this and the accompanying paper (Roberts et al., 198813) we characterize this modification as fatty acid acylation of the inositol portion of the phospholipid. To test if this modification is sufficient to confer PIPLC

TABLE III
Fatty acid composition of Eh" AChE acetolysis and deamination products Acidic methanolysis was performed on the acetolysis product B and on the deamination product from Eh" AChE as described under "Experimental Procedures." The liberated fatty acid methyl esters were analyzed by GLC Method 1. Mean from five enzyme preparations (from Roberts and Rosenberry, 1985), expressed as moles of fatty acid per mol of catalytic subunit based on enzyme activity.
ND means not detected. The detection limit is 0.02 mol/mol Of sample.

PIPLC-resistant Inositol Phospholipid
Anchor of Eh" AChE 18771 resistance, the '251-labeled product generated from EhU AChE by nitrous acid deamination was first deacylated with methanolic KOH or methanolic NH3 to remove the putative acyl group from inositol and then digested with PIPLC. As controls either PIPLC digestion or base methanolysis alone were performed. The results of this experiment are shown in Fig.  5. Methanolic KOH generated fatty acid methyl esters from the two samples which received this treatment. No alkylglycerols ( a ) were detectable in the control sample (lane l ) , but digestion of the putative alkyllyso-plasmanylinositol generated by methanolic KOH with PIPLC did release alkylglycerols as shown in lane 2. Ammonia methanolysis liberated fatty acid methyl esters and a small quantity of alkylglcyerols from the deamination product (lane 3 ) but more importantly permitted PIPLC digestion to release alkylacylglycerols (lane 4 ) . PIPLC digestion alone produced no detectable cleavage of the deamination product since all radioactivity remained at the origin (lane 5 ) . These results indicate qualitatively that alkyl-or alkylacylglycerols can be released from the EhU AChE inositol phospholipid by PIPLC digestion only after the inositol residue has been deacylated. The other product generated from alkyllyso-plasmanylinos-it01 by PIPLC digestion is inositol 1-phosphate. A more quantitative estimate of the extent of PIPLC cleavage was obtained by using GC-MS to detect this product. Basic methanolysis and PIPLC digestion were conducted as in Fig. 5 Under these solvent conditions, 1,2-alkylacylglycerols migrate between 1,2-and 1,3-dioleoylglycerol (Snyder, 1973;Roberts et d., 1987). Residual radioactivity at the origin in lanes 2 and 4 may result from the degradation of ['2SI]TID. 1986). Treatment with methanolic KOH followed by PIPLC digestion produced 1.1 & 0.2 ( n = 2) mol of inositol 1-phosphate/mol of myo-inositol. The levels of inositol 1phosphate detected after base methanolysis or PIPLC digestion alone were less than 0.03 mol/mol myo-inositol. These results demonstrate inositol 1-phosphate in the deamination product and provide our strongest evidence that deacylation of a hydroxyl group on myo-inositol permits the quantitative cleavage of the remaining inositol phospholipid by PIPLC.

Anchor-specific Phospholipase D Digestion of Eb" and Ehu
AChE-Recently, an abundant anchor-specific phospholipase D enzyme activity has been detected in mammalian serum (Davitz et al., 1987;Low and Prasad, 1988). Aliquots of this enzyme partially purified from rabbit serum were mixed with ['251]TID-labeled Eb" and EhU AChE to assess cleavage of the anchors in these enzymes. After digestion the hydrophobic products were extracted and subjected to TLC and autoradiography. The release of ['251]TID label into the chloroform phase was 47 f 5% for Eh" AChE and 39 k 11% for EbO AChE. This partial release is consistent with that observed for other substrates of the anchor-specific phospholipase D (Davitz et al., 1987;Low and Prasad, 1988). TLC analysis of the released products are shown in Fig. 6. The major radiolabeled fragment from both enzymes had a mobility very similar to that of a phosphatidic acid standard (g). These results indicate that the plasmanylinositols in both EhU and Eb AChE anchors are substrates for this phospholipase D activity. Furthermore, the observation that plasmanic acids' with equivalent TLC mobilities are released from both AChEs is consistent with the conclusion above that palmitoylation of the inositol phospholipid in EhU AChE occurs proximal to the phosphate group linking inositol and the alkylglycerol. This palmitoylation, although conferring resistance to PIPLC, apparently has little or no effect on the action of this phospholipase D. [Acyl (160) 7 1 H, CO"Alkyl (180,181)

DISCUSSION
Although many proteins with glycoinositol phospholipid membrane anchors have been identified, the chemical structure of only one of these, the Trypanosoma brucei variant 117 VSG anchor, has been fully elucidated (Ferguson et ai., 1985a;Ferguson et al., 1988). Characterization of the phospholipid portion of the VSG anchor phospholipid revealed the presence of dimyristoylglycerol, a diacylglycerol (Ferguson et al., 1985b). In contrast, other anchor or anchor-like glycolipids contain primarily alkylor alkylacylglycerols instead of diacylglycerols. Examples include an insulin-sensitive glycolipid from H35 hepatoma cells (Mato et al., 1987), a lipophosphoglycan from Leishmania donouani (Orlandi and Turco, 1987) and the membrane anchor of E' " AChE (Roberts et al., , 1988a. In all three cases, the alkylglycerol-containing component could be released by digestion with PIPLC. The identification of alkylacylglycerols as the main diradylglycerol species in E' " AChE prompted us to examine Eh" AChE for similar constituents. Acidic methanolysis of Eh" AChE and GLC analysis revealed two major alkylglycerol species, 18:O' (0.55 mol/mol) and 18:l (0.20 mol/mol), in Eh" AChE. This procedure illustrates quantitative analysis of fatty acid and alkylglycerol anchor components directly on a purified protein sample.
Further analysis of the molecular species released from E' " AChE by acetolysis revealed that alkylacylglycerols, predominantly combinations of 1 8 0 alkylglycerol with 22:4,22:5, and 22:6 acyl groups but with additional heterogeneity, were the only type of diradylglycerols in the membrane anchor of Eh" AChE. These results, which account for only 1 mol of fatty acid/mol of anchor, raised a question of the location of the second mol of fatty acids previously shown to be associated with the Eh" AChE anchor (Roberts and Rosenberry, 1985). Three approaches illustrated in Scheme 2 were used to answer this question.
First, a comparison of the fatty acid compositions of the Eh" AChE anchor lipid products generated by acetolysis and nitrous acid deamination indicated that the second fatty acid residue, primarily palmitate, is localized to a region of the deamination product outside the diradylglycerol. Second, digestion of Eb" and E' " AChE with anchor-specific phospholipase D generated similar amounts of plasmanic acid from both enzymes. This result indicates that the second fatty acid residue on Ehu AChE is not attached to the plasmanic acid portion of the anchor and is consistent with a second fatty acid residue linked in some manner to inositol. Finally, data obtained by fast atom bombardment mass spectrometry and presented in the accompanying paper (Roberts et al., 1988b) confirmed direct palmitoylation of an inositol hydroxyl group as the site of attachment of a second fatty acid residue, although the technique provided no information about which inositol hydroxyl group is acylated. To our knowledge fatty acid acylation of inositol has not been documented previously in any biological system, although a triacylated glycerophosphorylinositol in Corynebacterium xeroxis has been reported (Brennan, 1968).

IbH I
Deamination PLD Acetolysis To address our previous suggestion  that a substituent on the inositol phospholipid could be responsible for the resistance of the Ehu AChE anchor to PIPLC, the anchor deamination product was deacylated by basic methanolysis and digested with PIPLC. Following methanolysis with KOH, alkylglycerol and myo-inositol 1-phosphate were produced quantitatively by PIPLC. Methanolysis with NH3 selectively removed the fatty acid from inositol with little release of fatty acid from glycerol, a result confirmed by fast atom bombardment mass spectrometry (Roberts et al., 198813). PIPLC treatment of the resulting plasmanylinositol released alkylacylglycerols. This result provides specific evidence that acylation of the inositol ring is solely responsible for PIPLC resistance and ruled out any effect of the polyunsaturated fatty acids at the 2-position of the alkylglycerol on resistance to PIPLC. The presence of palmitic acid on the 2hydroxyl of inositol would prevent the formation of a 1,2cyclic phosphate intermediate by PIPLC (Ferguson et al., 1985a) and provide a mechanistic basis for PIPLC resistance. Palmitoylation at this position is currently under investigation.
Although the functions of glycoinositol phospholipid anchors in proteins currently are unknown, speculation focuses on the susceptibility of these proteins to release from the cell surface on activation of endogenous anchor-specific phospholipases (Low and Saltiel, 1988). Resistance to phospholipase C cleavage conferred by palmitoylation of the anchor inositol thus could provide a cellular mechanism for the regulation of phospholipase-induced protein release. Other glycoinositol phospholipid-anchoredproteins are at least partially resistant to PIPLC release, including 5"nucleotidase (Low and Finean, 1978;Shukla et al., 1980), decay accelerating factor (Davitz et al., 1986;Medof et al., 1986), and surface antigens from Dictyostelium discoideum (Sadeghi et al., 1988). Furthermore, the TLC mobility of the nitrous acid deamination product from ['251]TID-labeled decay accelerating factor was identical with that of the corresponding product from Ehu AChE (Medof et al., 1986;Walter et al., 1987). These observations support the suggestion that fatty acylation of an inositol hydroxyl(s) may be a general modification of glycoinositol phospholipidanchored membrane proteins, particularly in Ehu, that confers resistance to release by PIPLC. It is noteworthy, however, that decay accelerating factor purified from Eh" also can be converted to a hydrophilic form by phospholipase D (Davitz et al., 1987). This observation supports the conclusion that palmitoylation of inositol does not interfere with digestion by anchor-specific phospholipase D.
Despite the shortcomings of acetolysis, which include production of an unidentified by-product a (Fig. 3A) and formation of 1,3-diradylglycerol acetates (Fig. 9, see Miniprint Supplement), diradylglycerol acetates released from Ehu AChE by acetolysis were chosen for molecular species analysis rather than phosphatidic acids released by phospholipase D for several reasons. First, cleavage of diradylglycerols during acetolysis is complete and nonselective (Renkonen, 1965), while phospholipase D from human serum removes phosphatidic acid from only about 70% of the apparently homogeneous VSG substrate molecules (Davitz et al., 1987). Second, it is more difficult to resolve a mixture of phosphatidic acids than the less polar, more volatile diradylglycerol derivatives by either GLC or HPLC methods (Snyder, 1973). Finally, the chemical reagents used in acetolysis are far less likely to introduce contaminants than the crude biological preparations of phospholipase D. The major drawbacks of the acetolysis procedure, the isomerization of 1,2-diradylglycerols to 1,3-species and the generation of a , both can be cicumvented. The 1,2-and 1,3-isomers were resolved and identified on a polar capillary column. The fatty acid composition of a was very similar to that of the alkylacylglycerol acetates so its production did not affect determination of the mole percentages of the alkylacylglycerol species. The exact nature of a remains unclear. The data suggest that the major difference between CY and p involves their glyceryl ether moieties. Since acetolysis of EbO AChE, which contains almost exclusively alkylacylglycerols in its membrane anchor (Roberts et al., 1988a), also produced an a species and since the components generated from CY by acidic methanolysis gave a GLC profile similar to the alkylglycerols from p but with much shorter retention times, we conclude that the a product results from degradation of alkylglycerols during acetolysis. One possible mechanism of degradation could involve dehydration.