African trypanosomiasis is a human disease caused by the protozoan parasite Trypanosoma brucei. In the mammalian host, T. brucei is protected by a surface coat composed of a variant surface glycoprotein (VSG). ()VSG is glycosylphosphatidylinositol (GPI)-anchored; its GPI contains EtN-phospho-6Man(12)Man(16)Man(14)GlcN(16)-myo-Ins 1-phosphodimyristoylglycerol linked to the -carboxyl of the COOH-terminal residue of VSG through an amide bond with EtN(1) . T. brucei contains a glycosylphosphatidylinositol phospholipase C (GPI-PLC) that can cleave dimyristoylglycerol from VSG GPI, leaving VSG containing a GlcN(16)Ins 1,2-cyclic phosphate attached to the residual GPI glycan components(2) .

GPI-PLC is a 39-kDa integral membrane protein(3, 4, 5, 6) . It efficiently cleaves VSG GPI (apparent K = 370 nM; k = 2920 min) (5, 6) and some GPI biosynthetic intermediates(7) . Phosphatidylinositol (PI) is a very poor substrate for the enzyme(3, 4, 5, 6) . Although the biological function is unclear (reviewed in (8) and (9) ), GPI-PLC activity is detectable in bloodstream-form T. brucei, where VSG is expressed, and down-regulated 1000-fold in procyclic (insect stage) T. brucei. Thus, the enzyme could be involved in catabolism either of the VSG GPI anchor or of GPI biosynthetic intermediates. Nothing has been reported on the catalytic mechanism of GPI-PLC.

Phosphatidylinositol phospholipase C (PI-PLC) from Bacillus cereus cleaves GPIs(10) ; however, unlike GPI-PLC, it cleaves PI efficiently(6, 11, 12) . PI-PLC has a region of protein sequence similarity to GPI-PLC(11) ; 80 residues beginning at positions 69 and 70 for the T. brucei and B. cereus enzymes, respectively, can be aligned with 19 matches in a region that is 27.6% identical and 51.3% similar(8, 9, 13) .

Detailed information on GPI recognition is not available either for GPI-PLC or PI-PLC. Interestingly, when VSG and PI are present in the same reaction mixture at identical concentrations, GPI-PLC selectively cleaves VSG(6) , suggesting that glycan constituents of GPIs might be important for substrate recognition. Accordingly, we tested synthetic glycan components of EtN-phospho-6Man(12)Man(16)Man(14) GlcN(16)-myo-Ins, the ``conserved protein-GPI core,'' as potential inhibitors of GPI-PLC. We report that GlcN(16)Ins is probably the major glycan determinant of GPI-PLC specificity. Similar studies with GlcN(16)Ins and its derivatives on B. cereus PI-PLC indicate that steps toward cleavage of the identical GPI phosphodiester are different between GPI-PLC and PI-PLC.

EXPERIMENTAL PROCEDURES

Materials

Enzymes and Substrate

PI-PLC from B. cereus (600 units/mg) was obtained from Boehringer Mannheim. [H]Myristate-labeled VSG was isolated from T. brucei (ILTat 1.3)(3, 14) .

Synthesis and Characterization of Glycans

()

Inhibitor Solutions

GPI-PLC Assays

GPI-PLC (19 units, 1.19 ng) was first added to 20 µl of assay buffer (1 assay buffer = 50 mM Tris-HCl, pH 8.0, 5 mM EDTA, 1% Nonidet P-40), following which inhibitors were added to their specified final concentrations (defined for 30 µl), and the mixture was incubated on ice for 10 min. [H]Myristate-labeled membrane-form VSG (2 µg in 10 µl of 1 assay buffer) was added, and the tubes were incubated at 37 °C for 15 min. The reaction was terminated by chilling the mixture on ice and vortex mixing with 500 µl of water-saturated 1-butanol (at room temperature). Phases were separated by centrifugation (12,000 g, 1 min, 25 °C). Enzyme activity was quantified by measuring the amount of [H]dimyristoylglycerol released into the upper butanol phase using a Beckman LS 6000TA scintillation counter(3, 6) . Radioactivity from a mock digest (no enzyme addition) of [H]myristate-labeled VSG using 30 µl of assay buffer was subtracted as background from all counts obtained. Activity of GPI-PLC obtained without the addition of potential inhibitors in a parallel assay was assigned a value of 100%. Compounds that inhibited [H]myristate-labeled membrane-form VSG cleavage by >70% at 5 mM were investigated further (see ``Interfacial Kinetic Analysis''). Average values from duplicate determinations performed in several independent experiments showed a variation of 10%.

PI-PLC was diluted into 50 mM Tris-HCl, pH 8.0, and added (1 10 units, 0.17 ng) to 20 µl of PI-PLC buffer (25 mM HEPES/KOH, pH 7.5, 0.1% sodium deoxycholate) (10) with or without inhibitor on ice in a 1.5-ml microcentrifuge tube. The enzyme assay protocol was similar to that described above for GPI-PLC.

Interfacial Kinetic Analysis

GPI-PLC is presumed to be bound at the surface of Nonidet P-40 micelles with its active site facing bulk medium, a notion supported by the enzyme's ability to cleave GPI biosynthetic intermediates in vivo on the cytoplasmic side of the endoplasmic reticulum(22) . The micelle concentration of 1% Nonidet P-40 is 118.4 µM, while the concentrations of GPI-PLC and [H]myristate-labeled VSG are 1 nM and 1.2 µM, respectively, in the assay. Effective inhibitor concentration at the micelle interface was expressed as a mole fraction (X)(20, 21) . X was determined as described above for PI-PLC.

For both GPI-PLC and PI-PLC analysis, the concentration of hydrophilic compounds was included in the denominator for calculation of X for two reasons. 1) It emphasizes the interfacial nature of the inhibition events under discussion. Assuming that the interaction of hydrophilic compounds with detergent micelles is transient, the compounds still have to bind enzyme-micelle-VSG complexes to exert their inhibitory effects. 2) It makes for consistency in comparison of data from Fig. 2with X values presented in Table 1. The alternative approach of excluding the concentration of hydrophilic compounds from the denominator in X determinations produced similar conclusions, except that in examining X values, one was restricted to comparing hydrophilic compounds with each other and, likewise, amphipathic inhibitors with each other. The approach used here eliminates this apparent limitation.

Figure 2: Structures of glycans and their effects on GPI-PLC activity. Compounds were tested as described under ``Experimental Procedures.'' Percent inhibition at 5 mM was determined by comparison of the amount of [H]dimyristoylglycerol cleaved from [H]myristate-labeled VSG by GPI-PLC in the presence or absence of inhibitor. Values varied <10% in two or more independent duplicate determinations.

RESULTS

Glycan Components of the Conserved Protein-GPI Core Inhibit GPI-PLC

GPI-PLC assay conditions used for these inhibitor studies (see ``Experimental Procedures'') were empirically chosen to be linear with respect to enzyme concentration (Fig. 1A) and time (Fig. 1B), thereby ensuring that effects of potential inhibitors were discernible, as illustrated for GlcN(16)Ins 1,2-cyclic phosphate (compound VP-601L) (Fig. 1C). The glycan 6-O-(2-amino-2-deoxy--D-glucopyranosyl)-D-myo-inositol (GlcN(16)Ins; compound VP-606L) (Fig. 2) inhibited GPI-PLC moderately (28.9%). (All percentage inhibitions are quoted at 5 mM glycan. Variability in sets of duplicate determinations performed on different occasions was <10%.) We then explored whether modifications of GlcN(16)Ins could produce better inhibitors.

Figure 1: GPI-PLC inhibition conditions. A, the enzyme dose-response curve is shown. Duplicate sets of a standard 30-µl reaction mixture containing 2 µg of [H]myristate-labeled VSG (10,000 dpm, 1 µM final concentration) in 1 assay buffer were assembled on ice. Varying amounts of GPI-PLC (1.6 10 units/mg) were added as indicated, and the reaction was incubated at 37 °C for 15 min. [H]Dimyristoylglycerol cleaved from VSG was quantitated after extraction into 500 µl of water-saturated 1-butanol as described under ``Experimental Procedures.'' Data points are averages from duplicate reactions. A background of 353 dpm from a blank reaction (i.e. no GPI-PLC addition) has been subtracted from the data. The plot was generated with ``UltraFit'' (BioSoft, Ferguson, MO). B, the time course is shown. To a standard reaction mixture (as described for A) was added 19 units of GPI-PLC. The mixture was maintained at 37 °C and terminated at the indicated intervals, followed by quantitation of the released [H]dimyristoylglycerol as described for A. C, GlcN(16)Ins 1,2-cyclic phosphate (VP-601L) inhibits GPI-PLC cleavage of [H]myristate-labeled VSG. A reaction mixture containing 19 units of GPI-PLC (as described for B) was assembled on ice with or without GlcN(16)Ins 1,2-cyclic phosphate (VP-601L) as detailed under ``Experimental Procedures.'' Following a 15-min incubation at 37 °C, released [H]dimyristoylglycerol was quantitated (see ``Experimental Procedures''). The mole fraction (X) is the ratio of inhibitor concentration to the sum of inhibitor and detergent concentrations (see ``Experimental Procedures''). A mole fraction of 0.3 is equivalent to 5 mM VP-601L under these conditions.

The addition of a phosphate group to the Ins 1-hydroxyl of VP-606L, resulting in GlcN(16)Ins 1-phosphate (VP-600L), did not increase inhibitory potency (Fig. 2). However, cyclization of the Ins 1-phosphate to the Ins 2-hydroxyl, forming GlcN(16)Ins 1,2-cyclic phosphate (VP-601L), increased potency 2.3-fold (over that observed for GlcN(16)Ins 1-phosphate) to 74.6% (Fig. 2). The glycan O-(-D-mannopyranosyl)-(1,4)-O-(2-amino-2-deoxy--D-glucopyranosyl)-(1,6)-D-myo-Ins (Man(14)GlcN(16)Ins), which extends the conserved protein-GPI glycan core components to three, was only as inhibitory as GlcN(16)Ins.

Replacing the Ins 1-phosphate of VP-600L (see above) with Ins 1-dodecyl phosphonate increased inhibition of GPI-PLC. GlcN(16)Ins 1-dodecyl phosphonate (VP-604L) was 2.6-fold more inhibitory (81.1%) than GlcN(16)Ins 1-phosphate. However, GlcN(16)Ins 1-hexyl phosphonate (VFT-2) was not significantly better than GlcN(16)Ins 1-phosphate (Fig. 2). Interestingly, Ins 1-dodecyl phosphonate (VP-602L) inhibited GPI-PLC (60.3%) (Fig. 2), even though Ins 1-phosphate had no effect on the enzyme (data not shown).

Modifications of the amino group of GlcN(16)Ins affected inhibitory properties. Acetylation to GlcNAc(16)Ins (VC-105B) caused a 4.5-fold drop in the inhibition originally observed with GlcN(16)Ins (6.4%) (Fig. 2). Inhibitory activity was partially restored when the methyl group of GlcNAc(16)Ins was replaced by the bulkier N,N-dimethylamino group to produce N-(N,N-dimethylcarbamyl)-GlcN(16)Ins (VC-109B) (19.3%) (Fig. 2). Nevertheless, N-(N,N-dimethylcarbamyl)-GlcN(16)Ins was less inhibitory than GlcN(16) Ins.

Inositol ring modifications were also examined. Elimination of the hydroxyl group at the Ins 2-position of GlcN(16)Ins 1-phosphate (VP-600L) to form GlcN(16)-2-deoxy-Ins 1-phosphate (VP-612L) abolished inhibitory activity (Fig. 2). Interestingly, removal of the phosphate group from VP-612L, forming GlcN(16)-2-deoxy-Ins (VP-615L), re-established inhibitory activity (80.5% inhibition) (Fig. 2). This 2-deoxyIns analog is more potent than the parent compound, GlcN(16)Ins (VP-606L). GlcN(16)-2-deoxy-L-myo-inositol (VP-614L), in contrast to GlcN(16)-2-deoxy-Ins, had little effect on GPI-PLC (1.6%) (Fig. 2).

Compounds that inhibited GPI-PLC by >70% at 5 mM (except GlcN(16)Ins and Ins 1-dodecyl phosphonate (VP-602L)) were analyzed further to determine their inhibitory potency (X) (see ``Experimental Procedures'' for rationale and approach)(21, 23) . GlcN(16)Ins 1,2-cyclic phosphate (VP-601L) and GlcN(16)-2-deoxy-Ins (VP-615L) had X values of 0.16 and 0.11, respectively (Table 1). The X of GlcN(16)Ins (VP-606L) was not approached under the conditions of our assay. GlcN(16)Ins 1-dodecyl phosphonate (VP-604L) had an X of 0.14 (Table 1). Thus, the inhibitory potency of GlcN(16)Ins 1-dodecyl phosphonate (VP-604L), which is amphipathic, is comparable to that of GlcN(16)-2-deoxy-Ins (VP-615L), a hydrophilic compound.

Effects of Lipid Components of a GPI on GPI-PLC

Inhibition of PI-PLC

In contrast to its moderate inhibition of GPI-PLC, GlcN(16)Ins (VP-606L) had no effect on PI-PLC (Table 1). However, GlcN(16)Ins 1-phosphate (VP-600L) inhibited PI-PLC competitively (66.9%) (Fig. 3A). Remarkably, cyclization of the Ins 1-phosphate to the Ins 2-hydroxyl reduced inhibitory potency 3.8-fold; GlcN(16)Ins 1,2-cyclic phosphate (VP-601L) inhibited PI-PLC by only 17.3% (Table 1). GlcN(16)Ins 1-dodecyl phosphonate (VP-604L) was 44.6% inhibitory (Table 1); the inhibition was competitive (Fig. 3B) with an X of 0.69. On the contrary, GlcN(16)Ins 1-hexyl phosphonate (VFT-2) did not inhibit PI-PLC (data not shown). Ins 1-dodecyl phosphonate (VP-602L) inhibited PI-PLC (66.9%).

Figure 3: Nature of PI-PLC inhibition by hydrophilic (A) and (B) amphipathic GlcN(16)Ins derivatives. Shown is the ratio of the initial PI-PLC reaction rate in the absence of inhibitors (V) to the initial reaction rate in the presence of inhibitors (V) plotted against X/(1 - X); see ``Experimental Procedures'' for details. A: , GlcN(16)-2-deoxy-Ins 1-phosphate (VP-612L); , GlcN(16)-2-deoxy-L-myo-inositol (VP-614L); , GlcN(16)-2-deoxy-Ins (VP-615L); , GlcN(16)Ins 1-phosphate (VP-600L). B: , GlcN(16)Ins 1-dodecyl phosphonate (VP-604L). The coefficient of determination and yaxis intercept for a linear regression are as follows: VP-612L, 0.914 and -0.49; VP-614L, 0.968 and -0.027; VP-615L, 0.982 and 0.787; VP-600L, 0.998 and 1.357; and VP-604L, 0.978 and 0.769. If the coefficient of determination was >0.90 and the intercept on the yaxis was 1 ± 0.5, then data were fit with a straight line in A and B.

Although GlcN(16)Ins (VP-606L) did not inhibit PI-PLC, modifications of the Ins and GlcN moieties produced better inhibitors, some of which were highly specific for PI-PLC. The enzyme was competitively inhibited by GlcN(16)-2-deoxy-Ins (VP-615L) (X = 0.44) (Fig. 3A). PI-PLC was inhibited 91.7% by GlcN(16)-2-deoxy-Ins 1-phosphate (VP-612L) (Table 1). The X of GlcN(16)-2-deoxy-Ins 1-phosphate was 0.39, although the inhibition was not competitive (Fig. 3A). Intriguingly, GlcN(16)-2-deoxy-L-myo-inositol (VP-614L) was nearly as effective as GlcN(16)-2-deoxy-Ins 1-phosphate (VP-612L) (89.1%) (Table 1). Similarly, the inhibition was not competitive, and the X was 0.46 (Fig. 3A and Table 1). Modification of the amino group of GlcN(16)Ins influenced inhibitory activity; GlcNAc(16)Ins (VC-105B) inhibited PI-PLC (60%) (Table 1).

PI was a weak inhibitor of PI-PLC (15.6%) (Table 1). Similarly, PG, PS, PE, and PC had little effect on PI-PLC (Table 1).

Geneticin (G418) Specifically Stimulates PI-PLC

Figure 4: G418 stimulates PI-PLC. G418 was incubated with PI-PLC (0.017 ng, 1 10 units) () or GPI-PLC (1.19 ng, 4 units) () and analyzed as described for the glycans (see ``Experimental Procedures''). The ratio of reaction velocity in the presence or absence of G418 is presented.

DISCUSSION

Substrate Requirements of GPI-PLC

Susceptibility of polysaccharide-GPIs (glycoinositol phospholipids) to GPI-PLC provided clues to the substrate requirements of the enzyme. Glycoinositol phospholipids found in protozoan parasites of Leishmania sp. have a GPI with a conserved ``tetrasaccharide glycan core'' of galactofuranosyl(13)Man(13)Man(14)GlcN (reviewed in (26) and (29) ) and are cleaved by GPI-PLC(22) . This observation and the knowledge that PI is a very poor substrate suggested that GPI-PLC recognizes a glycan motif consisting minimally of Man(14)GlcN(16)Ins. We therefore focused on Man(14)GlcN(16)Ins as potentially having the requisites for GPI-PLC binding. Individual components of the protein-GPI core (Man, GlcN, Ins, and EtN, in all possible combinations) did not inhibit GPI-PLC, indicating that specific glycosidic bonds between the GPI components might be necessary for GPI-PLC recognition.

Glucosaminylinositol and its analogs inhibited GPI-PLC. GlcN(16)Ins (VP-606) and GlcN(16)Ins 1-phosphate (VP-600L) were about equally inhibitory (Fig. 2), suggesting that 1) GlcN(16)Ins is the major glycan determinant of GPI-PLC specificity, and 2) recognition of the phosphoryl group at the Ins 1-position is not critical for substrate binding. Interestingly, GlcN(16)Ins 1,2-cyclic phosphate (VP-601L) was a better inhibitor than GlcN(16)Ins 1-phosphate (VP-600L) ( Fig. 2and Table 1). GlcN(16)Ins 1,2-cyclic phosphate is found at the terminus of the EtN-phospho-6Man(12) Man(16)Man(14)GlcN(16)-myo-Ins 1,2-cyclic phosphate group that is covalently linked to a cleaved protein after GPI-PLC action(2) . Thus, GlcN(16)Ins 1,2-cyclic phosphate (VP-601L) might be a product analog. Last, the innermost mannosyl residue of the conserved glycan core of protein-GPIs does not appear to play a critical role in GPI-PLC substrate recognition since Man(14)GlcN(16)Ins was only as effective as GlcN(16)Ins.

Phosphonate derivatives of GlcN(16)Ins 1-phosphate (VP-600L) were more potent inhibitors, most likely because they are noncleavable substrate analogs. GlcN(16)Ins 1-dodecyl phosphonate (VP-604L) was more inhibitory than Ins 1-dodecyl phosphonate (VP-602L) ( Fig. 2and Table 1), attesting to the importance of the GlcN moiety in substrate recognition. Additionally, since GPI-PLC is associated with Nonidet P-40 micelles in our assays, one would predict that hydrophobic (and amphipathic) compounds would be better inhibitors because they would gain access to the enzyme more easily by associating initially with Nonidet P-40 micelles. If the phosphonate alkyl chain length is used as an indicator of hydrophobicity, the prediction is borne out. GlcN(16)Ins 1-dodecyl phosphonate (VP-604L), whose alkyl chain length is twice that of GlcN(16)Ins 1-hexyl phosphonate (VFT-2), is more inhibitory than VFT-2 (Fig. 2). Nevertheless, since GlcN(16)-2-deoxy-Ins (VP-615L) was as effective as GlcN(16)Ins 1-dodecyl phosphonate (VP-604L), a compound need not have a hydrophobic moiety to be a good inhibitor.

An unmodified amino group on GlcN(16)Ins is optimal for substrate recognition. Acetylation of the GlcN(16)Ins amino group abolishes recognition because GlcNAc(16)Ins (VC-105B) did not inhibit GPI-PLC (Fig. 2). This result is consistent with GlcNAc(16)Ins 1-phosphodiacylglycerol being a poor substrate for the enzyme(27) . We rule out steric hindrance at the GPI-PLC active site, due to replacement of an amino hydrogen of GlcN(16)Ins (VP-606) with an acetyl group, as the cause of GlcNAc(16)Ins loss of inhibitory activity because VC-109B, which contains an N,N-dimethylcarbamyl group instead of the hydrogen on the amino group of GlcN, inhibits GPI-PLC better than GlcNAc(16)Ins (Fig. 2). We do not have a simple structure-activity explanation for the VC-109B results.

Lack of inhibition by GlcN(16)-2-deoxy-Ins 1-phosphate (VP-612L) as compared with GlcN(16)Ins 1-phosphate (VP-600L) raises the possibility that inhibition by GlcN(16)Ins 1-phosphate is enzyme-mediated. Possibly, GlcN(16)Ins 1-phosphate inhibits only after enzyme-catalyzed cyclization to GlcN(16)Ins 1,2-cyclic phosphate, the GPI-PLC product analog. In the absence of the hydroxyl at the Ins 2-position, cyclization is impossible; hence, GlcN(16)-2-deoxy-Ins 1-phosphate is ineffective. Cyclization (possibly needed to eliminate the negative charge on the phosphate group) is not essential for inhibition when the phosphate at the Ins 1-position is removed from GlcN(16)-2-deoxy-Ins 1-phosphate. Hence, GlcN(16)-2-deoxy-Ins (VP-615L) is a potent inhibitor ( Fig. 2and Table 1). Specificity of these compounds is underscored by the lack of inhibition by GlcN(16)-2-deoxy-L-myo-inositol (VP-614L), the enantiomer of GlcN(16)-2-deoxy-Ins (VP-615L) (Fig. 2).

The role of the Ins 2-hydroxyl and a requirement for the D-myo-inositol enantiomer in GPI-PLC substrate recognition are addressed by the effects of GlcN(16)-2-deoxyIns and GlcN(16)-2-deoxy-L-myo-inositol, respectively. GlcN(16)-2-deoxy-Ins (VP-615L) inhibited GPI-PLC (Table 1), but GlcN(16)-2-deoxy-L-myo-inositol (VP-614L) had no effect. These results indicate that the Ins 2-hydroxyl is not required for substrate recognition, even though it is presumably needed for catalysis. The ineffectiveness of GlcN(16)-2-deoxy-L-myo-inositol (VP-614L) as compared with GlcN(16)-2-deoxy-Ins establishes the ability of GPI-PLC to distinguish between L-myo-inositol and D-myo-inositol.

Inhibitory constants obtained for the Ins 1-phosphonates in this study were slightly better than, but in the same range as, those obtained in the inhibition of B. cereus PI-PLC by Ins 1-palmitoyl phosphonate(30) . To our knowledge, however, our work represents the first use of GlcN(16)Ins and its derivatives as inhibitors of GPI phospholipases. If tested against other membrane-bound GPI-specific enzymes (e.g. phospholipases and glycosyltransferases), the inhibitory constants obtained are likely to be in the same range as reported here, mainly because the inhibition occurs at the interface between a micelle-bound enzyme and an aqueous soluble inhibitor.

Acidic Phospholipids Inhibit GPI-PLC

Selective Inhibition and Stimulation of PI-PLC and GPI-PLC: Implications for Catalytic Mechanisms and Possible Allosteric Regulation

The Ins 1-phosphoryl group is very important for glycan recognition by PI-PLC. This inference is backed by the observation that GlcN(16)Ins 1-phosphate (VP-600L) competitively inhibits PI-PLC, while GlcN(16)Ins (VP-606L) is completely ineffective (Table 1). (The identical phosphoryl group is not critical for GPI-PLC substrate recognition (see discussion above).) Possibly, abstraction of a proton from the Ins 2-hydroxyl is not an early event in PI-PLC cleavage of GPIs. Instead, attack on the phosphoryl group by an active-site nucleophile might lead to formation of a pentacoordinate enzyme-phosphoinositol glycan intermediate, which collapses subsequently to GlcN(16)Ins 1,2-cyclic phosphate and finally to GlcN(16)Ins 1-phosphate. It has been suggested that one of the paths toward cleavage of PI by mammalian PI phospholipases C involves formation of an enzyme-phosphoinositol intermediate(33) , similar to that proposed here for PI-PLC.

Other differences in the effects of the GlcN(16)Ins derivatives include the following. 1) GlcN(16)-2-deoxy-L-myo-inositol (VP-614L) inhibits PI-PLC, but has little effect against GPI-PLC (Table 1). 2) GlcNAc(16)Ins (VC-105B) is effective only against PI-PLC (Table 1). 3) G418 stimulates PI-PLC only (Fig. 4). 4) Acidic phospholipids inhibit GPI-PLC without exerting a significant effect on PI-PLC.

We conclude that PI-PLC binds glycans containing either D-myo-Ins or L-myo-Ins, unlike GPI-PLC, which appears to bind glycans containing D-myo-Ins only. Furthermore, since the inhibition of PI-PLC by GlcN(16)-2-deoxy-Ins 1-phosphate (VP-612L) and GlcN(16)-2-deoxy-L-myo-inositol (VP-614L) was not competitive, our data raise the possibility of allosteric regulation of PI-PLC at a novel carbohydrate-binding site. In this regard, the aminoglycosides G418 and gentamicin could stimulate PI-PLC by binding to a regulatory site analogous to the proposed allosteric site occupied by hydrophilic glycans. Consistent with this latter hypothesis, G418 reverses a GPI-negative phenotype in some mammalian cells by uncharacterized mechanisms(34) . Our data on PI-PLC suggest that G418 could bind an enzyme in the GPI biosynthesis/regulatory pathway and cause reversal of the GPI-negative phenotype. Inhibition of PI-PLC by GlcNAc(16)Ins is consistent with GlcNAc(16)Ins 1-phosphodiacylglycerol being a substrate for the enzyme(35) .

Regulation of GPI-PLC by acidic phospholipids in vivo is an intriguing possibility. GPI-PLC is a cytoplasmic membrane protein that cleaves GPI precursors in vitro. Yet, in T. brucei, where the enzyme is endogenous, GPI-PLC does not cause a depletion of GPI biosynthetic intermediates, even though it appears to colocalize with GPI intermediates on the cytoplasmic side of intracellular membranes, where GPI biosynthesis is initiated(22, 36) . How could GPI-PLC be prevented from catabolizing GPI intermediates in T. brucei? If inhibition by PI occurred in vivo (presumably at a lower concentration of PI since detergent is absent), GPI-PLC might be prevented from cleaving GPI intermediates in PI-enriched intracellular membrane microdomains where GPI biosynthesis might be initiated.

In summary, details of GPI recognition differ between GPI-PLC and PI-PLC. The remarkable contrasts in sensitivity of the two enzymes to some GlcN(16)Ins derivatives and gentamicins during cleavage of the identical GPI phosphodiester suggest the existence of mechanistic subclasses of GPI phospholipases C, of which GPI-PLC and PI-PLC might be prototypes. Hence, GlcN(16)Ins derivatives seem likely to be powerful tools for analyzing the mode of action of GPI phospholipases, and possibly GPI glycosyltransferases.

FOOTNOTES

*

This work was supported in part by National Institutes of Health Grants AI 33383 (to K. M.-W.) and GM 47109 (to T.-Y. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore by hereby marked ``advertisement'' in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

§

Supported by National Institutes of Health Predoctoral Training Grant 1-T32-AIO-7322.

¶

Supported by a Burroughs Wellcome Fund New Investigator Award in Molecular Parasitology. To whom correspondence should be addressed: Dept. of Cellular Biology, 724 Biological Sciences Bldg., University of Georgia, Athens, GA 30602. Tel.: 706-542-3355; Fax: 706-542-4271; mensawil{at}zookeeper.zoo.uga.edu.

()

The abbreviations used are: VSG, variant surface glycoprotein; GPI, glycosylphosphatidylinositol; EtN, ethanolamine; Ins, D-myo-inositol; GPI-PLC, glycosylphosphatidylinositol phospholipase C; GlcN(16) Ins, glucosaminyl(16)-D-myo-inositol; PI, phosphatidylinositol; PC, phosphatidylcholine; PS, phosphatidylserine; PG, phosphatidylglycerol; PE, phosphatidylethanolamine; PI-PLC, phosphatidylinositol phospholipase C.

()

L. Ping-Sheng and T.-Y. Shen, manuscript in preparation.

ACKNOWLEDGEMENTS

REFERENCES

1. Ferguson, M. A. J., Homans, S. W., Dwek, R. A., and Rademacher, T. W. (1988) Science 239, 753-759 [Abstract/Free Full Text]
2. Ferguson, M. A. J., Low, M. G., and Cross, G. A. M. (1985) J. Biol. Chem. 260, 14547-14555 [Abstract/Free Full Text]
3. Hereld, D., Krakow, J. L., Bangs, J. D., Hart, G. W., and Englund, P. T. (1986) J. Biol. Chem. 261, 13813-13819 [Abstract/Free Full Text]
4. Fox, J. A., Duszenko, M., Ferguson, M. A., Low, M. G., and Cross, G. A. M. (1986) J. Biol. Chem. 261, 15767-15771 [Abstract/Free Full Text]
5. Bülow, R., and Overath, P. (1986) J. Biol. Chem. 261, 11918-11923 [Abstract/Free Full Text]
6. Mensa-Wilmot, K., and Englund, P. T. (1992) Mol. Biochem. Parasitol. 56, 311-322 [CrossRef][Medline] [Order article via Infotrieve]
7. Masterson, W. J., Doering, T. L., Hart, G. W., and Englund, P. T. (1989) Cell 56, 793-800 [CrossRef][Medline] [Order article via Infotrieve]
8. Carrington, M., Walters, D., and Webb, H. (1991) Cell Biol. Int. Rep. 15, 1101-1114 [CrossRef][Medline] [Order article via Infotrieve]
9. Mensa-Wilmot, K., Doering, T. L., Raper, J., Buxbaum, L., Hart, G. W., and Englund, P. T. (1991) in Molecular and Immunological Aspects of Parasitism (Wang, C. C., ed) pp. 43-54, American Association for the Advancement of Science, Washington, D. C.
10. Low, M. G. (1992) in Lipid Modification of Proteins: A Practical Approach (Hooper, N. M., and Turner, A. J., eds) pp. 117-154, IRL Press, Oxford
11. Henner, D. J., Yang, M., Chen, E., Hellmiss, R., Rodriguez, H., and Low, M. G. (1988) Nucleic Acids Res. 16, 10383 [Free Full Text]
12. Griffith, O. H., Volwerk, J. J., and Kuppe, A. (1991) Methods Enzymol. 197, 493-502 [Medline] [Order article via Infotrieve]
13. Kuppe, A., Evans, L. M., McMillen, D. A., and Griffith, O. H. (1989) J. Bacteriol. 171, 6077-6083 [Abstract/Free Full Text]
14. Mensa-Wilmot, K., Morris, J. C., Al-Qahtani, A., and Englund, P. T. (1995) Methods Enzymol. , in press
15. Plourde, R., d'Alarcao, M., and Saltiel, A. R. (1992) J. Org. Chem. 57, 2606-2610 [CrossRef]
16. Berlin, W. K., Zhang, W. S., and Shen, T. Y. (1991) Tetrahedron 47, 1-20
17. Vacca, J. P., deSolms, S. J., Huff, J. R., Bullington, D. C., Bayker, R., Kulagowski, J. J., and Mawer, I. M. (1989) Tetrahedron 45, 5679-5702 [CrossRef]
18. Ames, B. N. (1966) Methods Enzymol. 8, 115-118
19. Volwerk, J. J., Filthuth, E., Griffith, O. H., and Jain, M. K. (1994) Biochemistry 33, 3464-3474 [CrossRef][Medline] [Order article via Infotrieve]
20. Berg, O. G., Yu, B. Z., Rogers, J., and Jain, M. K. (1991) Biochemistry 30, 7283-7297 [CrossRef][Medline] [Order article via Infotrieve]
21. Jain, M. K., Yuan, W., and Gelb, M. H. (1989) Biochemistry 28, 4135-4139 [CrossRef][Medline] [Order article via Infotrieve]
22. Mensa-Wilmot, K., LeBowitz, J. H., Chang, K. P., Al-Qahtani, A., McGwire, B. S., Tucker, S., and Morris, J. C. (1994) J. Cell Biol. 124, 935-947 [Abstract/Free Full Text]
23. Jain, M. K., Rogers, J., Hendrickson, H. S., and Berg, O. G. (1993) Biochemistry 32, 8360-8367 [CrossRef][Medline] [Order article via Infotrieve]
24. Volwerk, J. J., Shashidar, M. S., Kuppe, A., and Griffith, O. H. (1990) Biochemistry 29, 8056-8062 [CrossRef][Medline] [Order article via Infotrieve]
25. Englund, P. T. (1993) Annu. Rev. Biochem. 62, 121-138 [CrossRef][Medline] [Order article via Infotrieve]
26. McConville, M. J., and Ferguson, M. A. J. (1993) Biochem. J. 294, 305-324
27. Doering, T. L., Masterson, W. J., Englund, P. T., and Hart, G. W. (1989) J. Biol. Chem. 264, 11168-11173 [Abstract/Free Full Text]
28. Krakow, J. L., Doering, T. L., Masterson, W. J., Hart, G. W., and Englund, P. T. (1989) Mol. Biochem. Parasitol. 36, 263-270 [CrossRef][Medline] [Order article via Infotrieve]
29. Turco, S. J., and Descoteaux, A. (1992) Annu. Rev. Microbiol. 46, 65-94 [CrossRef][Medline] [Order article via Infotrieve]
30. Shashidhar, M. S., Volwerk, J. J., Keana, J. F. W., and Griffith, O. H. (1990) Biochim. Biophys. Acta 1042, 410-412 [Medline] [Order article via Infotrieve]
31. Newton, A. C. (1993) Annu. Rev. Biophys. Biomol. Struct. 22, 1-25 [Medline] [Order article via Infotrieve]
32. Robson, R. J., and Dennis, E. A. (1978) Biochim. Biophys. Acta 508, 513-524 [Medline] [Order article via Infotrieve]
33. Bruzik, K. S., Morocho, A. M., Jhon, D. Y., Rhee, S. G., and Tsai, M. D. (1992) Biochemistry 31, 5183-5193 <