Streptolysin-O (SLO), a cholesterol-binding agent, was used for studies on the release of glycosylphosphatidylinositol (GPI)-anchored alkaline phosphatase (AP) from ROS cells. Treatment of cells with SLO resulted in a time- and concentration-dependent release of AP into the extracellular medium. This release was potentiated by Ca2+ and bovine serum, but not by GPI-specific phospholipase D (GPI-PLD) purified from bovine serum. The released AP distributed to the detergent phase after Triton X-114 phase separation. This result suggested that the released AP contained an intact GPI anchor, and thus both proteolysis and anchor degradation by anchor-specific hydrolases, including GPI-PLD, as the potential mechanisms for SLO-mediated AP release were ruled out. The released AP sedimented at 100,000 g. A substantial amount of lipids was detected in the 100,000 g pellet. Cholesterol and sphingomyelin were enriched in SLO-released material, compared with intact cells. These results were consistent with vesiculation as the mechanism for SLO induction of AP release. Two other cholesterol-binding agents, saponin and digitonin, were also able to release AP, possibly by a similar vesiculation mechanism, whereas others, including nystatin, filipin and beta-escin, failed to elicit any AP release. Eight GPI-anchored proteins were identified in ROS cells, and all were substantially enriched in the vesicles released by SLO. Taken together, these results do not provide any support for the hypothesis that the clustering of GPI-anchored proteins in the plasma membrane is responsible for their resistance to GPI-PLD cleavage.

Previous studies have shown that some cells (e.g. SKG3a) express human placental alkaline phosphatase (AP) in a form which can be released from the membrane by bacterial PtdIns-specific phospholipase C (PI-PLC) while others (e.g. HeLa) are relatively resistant to this enzyme. Chemical and enzymic degradation studies have suggested that the PI-PLC resistance of AP is due to inositol acylation of its glycosylphosphatidylinositol (GPI) anchor. In order to identify the biosynthetic origin of PI-PLC resistance we determined the PI-PLC sensitivity of AP in 35S-labelled cells (10 min pulse; 0-60 min chase) by Triton X-114 phase separation. At the beginning of the chase period, the majority of the AP synthesized was hydrophilic, indicating that it had not acquired a GPI anchor. The concentration of hydrophilic AP species decreased with a t1/2 of 30-60 min but was not processed to an endoglycosidase H-resistant species or secreted into the medium. In both SKG3a and HeLa cells all of the hydrophobic, GPI-anchored AP detectable at the beginning of the chase was PI-PLC sensitive. PI-PLC-resistant species of AP were only observed in HeLa cells and these only appeared after about 30 min. The delayed appearance of PI-PLC resistance was unexpected as previous studies have suggested that candidate GPI-anchor precursors are PI-PLC-resistant as a result of inositol acylation. This work reveals unanticipated complexities in the biosynthesis of AP and its GPI anchor.

Glycosylphosphatidylinositol (GPI)-specific phospholipase D (GPI-PLD) is abundant in mammalian plasma. It could potentially regulate the surface expression of GPI-anchored proteins, but it remains to be established which tissue(s) or cell type(s) are the principal sources of the circulating enzyme. Here we report that all the myeloid cell lines tested, including K562 (multipotential blast), KG-1 (human myeloblast), HL-60, NB4, PLB-985 (human promyelocyte), U937 (human promonocyte), THP-1 (human monocyte) and J774, RAW264.7 (mouse monocyte/macrophage), contained GPI-degrading activity. T.l.c. analysis of reaction products confirmed the activity as a phospholipase D. These cells also exhibited positive immunofluorescent staining with an anti-GPI-PLD monoclonal antibody. The expression of GPI-PLD activity was not substantially reduced when the cells were cultured in either serum-free medium or GPI-PLD-depleted regular medium. Both granulocytic and monocytic differentiation of myelomonoblastic lines (e.g. HL-60) induced by dimethyl sulphoxide or phorbol diester respectively was accompanied by a 2-3-fold increase in GPI-PLD activity. J774 and HL-60 cells secreted GPI-PLD into the medium constitutively. Taken together, these data suggest that myeloid cells are a potential contributor to the circulating GPI-PLD pool. As leucocytes express many important GPI-anchored surface antigens, these cells may prove to be a valuable model system for studying the physiological functions of GPI-PLD.

Mammalian serum and plasma contain high levels of glycosylphosphatidylinositol-specific phospholipase D (GPI-PLD). Previous studies with crude serum or partially purified GPI-PLD have shown that this enzyme is capable of degrading the GPI anchor of several purified detergent-solubilized cell surface proteins yet is unable to act on GPI-anchored proteins located in intact cells. Treatment of intact ROS17/2.8, WISH or HeLa cells (or membrane fractions prepared from them) with GPI-PLD purified from bovine serum by immunoaffinity chromatography gave no detectable release of alkaline phosphatase into the medium. However, when membranes were treated with GPI-PLD in the presence of 0.1% Nonidet P-40 substantial GPI anchor degradation (as measured by Triton X-114 phase separation) was observed. The mechanism of this stimulatory effect of detergent was further investigated using [3H]myristate-labelled variant surface glycoprotein and human placental alkaline phosphatase reconstituted into phospholipid vesicles. As with the cell membranes the reconstituted substrates exhibited marked resistance to the action of purified GPI-PLD which could be overcome by the inclusion of Nonidet P-40. Similar results were obtained when crude bovine serum was used as the source of GPI-PLD. These data indicate that the resistance of cell membranes to the action of GPI-PLD is not entirely due to the action of serum or membrane-associated inhibitory factors. A more likely explanation is that, in common with many other eukaryotic phospholipases, the action of GPI-PLD is restricted by the physical state of the phospholipid bilayer in which the substrates are embedded. These data may account for the ability of endothelial and blood cells to retain GPI-anchored proteins on their surfaces in spite of the high levels of GPI-PLD present in plasma.

We directly manipulated the levels of PtdIns, PtdInsP and PtdInsP2 in digitonin-treated adrenal chromaffin cells with a bacterial phospholipase C (PLC) from Bacillus thuringiensis and by removal of ATP. The PtdIns-PLC acted intracellularly to cause a large decrease in [3H]inositol- or [32P]phosphate-labelled PtdIns, but did not directly hydrolyse PtdInsP or PtdInsP2. [3H]PtdInsP and [3H]PtdInsP2 levels declined markedly, probably because of the action of phosphatases in the absence of synthesis. Removal of ATP also caused marked decreases in [3H]PtdInsP and [3H]PtdInsP2. The decrease in polyphosphoinositide levels by PtdIns-PLC treatment or ATP removal was reflected by the inhibition of the production of inositol phosphates upon subsequent activation of the endogenous PLC by Ca2(+)-dependent catecholamine secretion from permeabilized cells was strongly inhibited by PtdIns-PLC treatment and by ATP removal. Ca2(+)-dependent secretion was similarly correlated with the sum of PtdInsP and PtdInsP2 when the level of these lipids was changed by either manipulation. PtdIns-PLC inhibited only the ATP-dependent component of secretion and did not affect ATP-dependent secretion. Both PtdIns-PLC and ATP removal inhibited the late slow phase of secretion, but had little effect on the initial rapid phase. Although we found a tight correlation between polyphosphoinositide levels and secretion, endogenous phospholipase C activity (stimulated by Ca2+, guanine nucleotides and related agents) was not correlated with secretion. Additional experiments indicated that neither the products of the PtdIns-PLC reaction (diacylglycerol and InsP1) nor the inability to generate products by subsequent activation of the endogenous PLC is likely to account for the inhibition of secretion. Incubation of permeabilized cells with neomycin in the absence of ATP maintained the level of polyphosphoinositides and more than doubled subsequent Ca2(+)-dependent secretion. The data suggest that: (1) Ca2(+)-dependent secretion has a requirement for the presence of inositol phospholipids; (2) the enhancement of secretion by ATP results in part from increased polyphosphoinositide levels; and (3) the role for inositol phospholipids in secretion revealed in these experiments is independent of their being substrates for the generation of diacylglycerol and InsP3.

Phosphatidylinositol-specific phospholipase C (PI-PLC) produced by Bacillus thuringiensis has been used as a probe for the distribution of phosphatidylinositol in hepatocyte membranes. Approx. 50% of this phospholipid was hydrolysed in microsomal vesicles (endoplasmic reticulum) with no significant hydrolysis of the remaining membrane phospholipids. Latency of mannose-6-phosphatase was retained during treatment indicating that the vesicles remained impermeable. Stripping of the ribosomes did not increase hydrolysis of phosphatidylinositol; however, when the vesicles were opened using dilute sodium carbonate, hydrolysis increased to greater than 90%. Hydrolysis of phosphatidylinositol of Golgi membranes was 35% and of plasma membranes was 50%. After treatment with PI-PLC, radiolabelled secretory proteins were retained in Golgi membranes and trapped lactate dehydrogenase was retained in plasma-membrane preparations indicating that the vesicles remained closed. Hydrolysis of phosphatidylinositol increased to greater than 90% when the membranes were opened by treatment with dilute sodium carbonate. These observations indicate that PI-PLC of Bacillus thuringiensis is a suitable probe for the distribution of phosphatidylinositol in membranes, and that in liver membranes this phospholipid occurs on each side of the bilayer, a topography consistent with its diverse roles.

Myelin basic protein has been isolated from bovine central-nervous-system myelin by four methods, none of which exposes the protein to acid. After purification the inositol content of both hydrolysed and unhydrolysed protein was quantified by g.c.-m.s. Basic protein prepared by all methods contained less than 4 mol % of inositol. It is concluded, contrary to a previous proposal, that covalent binding to phosphoinositides does not represent a general mechanism for attachment of this cytoplasmically-oriented protein to its membrane.

Renal dipeptidase (dehydropeptidase-I, EC 3.4.13.11) was released from pig kidney membrane preparations by treatment with phosphatidylinositol-specific phospholipase C from Staphylococcus aureus and Bacillus thuringiensis and a phospholipase C preparation from Bacillus cereus to a similar extent as alkaline phosphatase. Endopeptidase-24.11 and aminopeptidase N were not released by this treatment. After treatment of the membrane fraction with the S. aureus phospholipase C the dipeptidase was converted from an amphipathic to a hydrophilic form, as deduced from phase-separation experiments in Triton X-114. It is concluded that renal dipeptidase is anchored to the microvillar membrane by covalently attached phosphatidylinositol.

Our earlier evidence suggested that both acetylcholinesterase and alkaline phosphatase are anchored to the cell surface via covalently-attached phosphatidylinositol [Low, Futerman, Ferguson & Silman (1986) Trends Biochem. Sci. 11, 212-215]. We now present chemical data, based upon a nitrous acid deamination reaction, showing that in both proteins the phosphatidylinositol moiety is attached through a glycosidic linkage to a sugar residue bearing a free amino group.

Alkaline phosphatase in a wide range of tissues has been shown to be anchored in the membrane by a specific interaction with the polar head group of phosphatidylinositol. It has previously been suggested that the production of low Mr alkaline phosphatase during the commonly used butanol extraction procedure may result from the activation of an endogenous phosphoinositide-specific phospholipase C which removes the 1,2-diacylglycerol responsible for membrane anchoring. This conversion process was investigated in greater detail with human placenta used as the source of alkaline phosphatase. Mr and hydrophobicity of the alkaline phosphatase were determined by gel filtration on TSK-250 and partitioning in Triton X-114, respectively. Alkaline phosphatase extracted from human placental particulate fraction with butanol at pH 5.4 or released by incubation with Staphylococcus aureus phosphatidylinositol-specific phospholipase C produced a form of alkaline phosphatase of Mr approx. 170,000 and relatively low hydrophobicity. By contrast, the butanol extract prepared at pH 8.3 was an aggregated form of Mr approx. 600,000 and was relatively hydrophobic. The effect of a variety of inhibitors and activators on the amount of low Mr alkaline phosphatase produced during butanol extraction revealed that it was a Ca2+- and thiol-dependent process. Proteinase inhibitors had no effect. [3H]Phosphatidylinositol hydrolysis by the particulate fraction, unlike low Mr alkaline phosphatase production, was relatively sensitive to heat inactivation, indicating that the phosphoinositide-specific phospholipases C from cytosol and lysosomes were unlikely to be responsible for conversion. A butanol-stimulated activity which removed the [3H]myristic acid from the variant surface glycoprotein ([3H]mfVSG) of Trypanosoma brucei was detectable in the human placental particulate fraction. Since this activity was acid active, Ca2+- and thiol-dependent and relatively heat stable, it may be the same as that responsible for production of low Mr alkaline phosphatase. The only 3H-labelled product identified was phosphatidic acid, suggesting that the [3H]mfVSG-cleaving activity is a phospholipase D. These data strongly support the proposal that production of low Mr alkaline phosphatase during butanol extraction is an autolytic process occurring as the result of an endogenous phospholipase. However, they also suggest that the lysosomal and cytosolic phosphoinositide-specific phospholipases C that have previously been described in many mammalian tissues are not responsible for this process.

The origin and physiological significance of the multiple Mr forms of phosphoinositide-specific phospholipase C in human platelets were investigated. The higher-Mr (400,000 and 270,000) forms of the phospholipase C were converted into the 100,000-Mr form without substantial loss of activity by incubation with a Ca2+-dependent proteinase partially purified from human platelets. These three forms of the phospholipase C were purified approx. 200-500-fold from outdated human platelet supernatants. SDS/polyacrylamide-gel electrophoresis and gel-filtration analysis suggested that the higher-Mr forms of phospholipase C were complexes of 140,000-Mr subunits, whereas the lower-Mr form consisted of a single 95,000-Mr subunit. The substrate specificity of the purified phospholipase C was investigated by using 32P-labelled polyphosphoinositide substrates purified from human platelets by a new method utilizing h.p.l.c. on an amino column. Activity against all three phosphoinositides was detected at micromolar concentrations of Ca2+; this hydrolysis was markedly stimulated by phosphatidylethanolamine and inhibited by phosphatidylcholine. Comparison of the different forms of purified phospholipase C revealed no major differences in Ca2+-sensitivity or substrate specificity. Thus, although the suggestion that the high-Mr forms of human platelet phosphoinositide-specific phospholipase C were converted into a lower-Mr form by a Ca2+-dependent proteinase has been substantiated, the physiological significance of this process remains to be determined.

Quantitative solubilization of the phospholipid-associated form of acetylcholinesterase (AChE) from Torpedo electric organ can be achieved in the absence of detergent by treatment with phosphatidylinositol-specific phospholipase C (PIPLC) from Staphylococcus aureus [Futerman, Low & Silman (1983) Neurosci. Lett. 40, 85-89]. The sedimentation coefficient on sucrose gradients of AChE solubilized in detergents (DSAChE) varies with the detergent employed. However, the coefficient of AChE directly solubilized by PIPLC is not changed by detergents. Furthermore, PIPLC can abolish the detergent-sensitivity of the sedimentation coefficient of DSAChE purified by affinity chromatography, suggesting that one or more molecules of phosphatidylinositol (PI) are co-solubilized with DSAChE and remain attached throughout purification. DSAChE binds to phospholipid liposomes, whereas PIPLC-solubilized AChE and DSAChE treated with PIPLC do not bind even to liposomes containing PI. Sodium dodecyl sulphate/polyacrylamide-gel electrophoresis shows that PIPLC-solubilized AChE, like unmodified DSAChE, is a catalytic subunit dimer; electrophoresis in the presence of reducing agent reveals no detectable difference in the Mr of the catalytic subunit of unmodified DSAChE, of AChE solubilized by PIPLC and of AChE solubilized by Proteinase K. The results presented suggest that DSAChE is anchored to the plasma membrane by one or more PI molecules which are tightly attached to a short amino acid sequence at one end of the catalytic subunit polypeptide.

The Mr distribution of phosphoinositide-specific phospholipase C in the supernatants isolated from a variety of animal tissues was analysed by high-performance gel-filtration chromatography. In most tissues, at least four peaks of activity were resolved. However, different tissues showed quite marked differences in the distribution of activity between these peaks. In rat heart, lung and kidney, the predominant form had Mr approx. 90000, whereas the predominant form in brain had Mr approx. 290000. In liver, the Mr-90000 form predominated, but this tissue also contained relatively large amounts of a form of Mr approx. 150000. Phospholipase C in these tissues from other animal species gave similar distributions of activity between the peaks. In supernatants prepared from platelets sonicated in the presence of leupeptin (0.5 mM) or EGTA (20 mM), the Mr-290000 form predominated. However, when leupeptin or EGTA (inhibitors of Ca2+-dependent proteinase) was omitted from the sonication buffer, the Mr-290000 form appeared to be replaced by a form of Mr 100000. Similar changes in Mr were not demonstrated with the other tissues. These results may be relevant to the intracellular regulation of phospholipase C, since Ca2+-dependent proteolysis has been reported to occur during platelet activation.

Phospholipase C activity capable of hydrolysing phosphatidylinositol in bovine heart was resolved into four forms (I-IV) by ion-exchange chromatography. Some of these forms could only be detected if the assay was performed at acidic pH (I and IV) or in the presence of deoxycholate (II). Gel-filtration chromatography indicated that the four forms had different molecular weights in the range 40000-120000. I, II and III all had pH optima in the range 4.5-5.5. However, the major form (III) also had substantial activity at pH 7.0 and above. The activities of I, II and III at pH 7.0 were stimulated by deoxycholate; this effect was most marked with I and II, which had very low activity at this pH. All forms of the enzyme were inhibited by EGTA and required 2-5 mM-CaCl2 for maximal activity. When the fractions eluted from the ion-exchange and gel-filtration columns were assayed with polyphosphoinositides as substrates there was a close correspondence to the elution profile obtained with phosphatidylinositol as substrate; there was no evidence for the existence in heart of phospholipase C activities specific for individual phosphoinositides.

Purified phosphatidylinositol-specific phospholipase C from Staphylococcus aureus released a substantial proportion of the total alkaline phosphatase activity from a wide range of tissues from several mammalian species. Co-purification of the phospholipase C and alkaline phosphatase-releasing activities and the inhibition of both these activities by iso-osmotic salt solutions suggested that the releasing effect was unlikely to be due to a contaminant.

A phosphatidylinositol-specific phospholipase C from Staphylococcus aureus was purified by a three-step procedure. The specific activity of the purified enzyme was approx. 6000 times that of the culture supernatant, with an overall recovery of approx. 10%. Estimation of the molecular weight by sodium dodecyl sulphate/polyacrylamide-gel electrophoresis and by gel filtration gave values of 33000 and 20000 respectively. A thiol group appears to be necessary for the activity of the enzyme. The purified enzyme had no detectable delta-haemolytic activity and was unable to hydrolyse S. aureus phospholipids. Phosphatidyl-inositol in erythrocyte ‘ghosts’ was readily hydrolysed by the purified phospholipase C. However, in contrast with our previous preliminary observations, phosphatidylinositol in intact erythrocytes was not significantly hydrolysed. These results suggest that at least 75-80% of the phosphatidylinositol is located at the inner leaflet of the membrane.

A phospholipase C prepared from lymphocytes readily hydrolysed pure phosphatidyl-inositol but was relatively ineffective against phosphatidylinositol in erythrocyte “ghosts” and rat liver microsomal fraction and also against sonicated lipid extracts from these membranes. In contrast, a phospholipase C prepared from Staphylcoccus aureus readily hydrolysed phosphatidylinositol in sonicated lipid extracts but had only low activity against purified phosphatidylinositol. Unlike the enzyme from lymphocytes, the S. aureus phospholipase C did not require Ca2+ for its activity and was inhibited by cations. The previously reported specificity of this enzyme was confirmed by our observation of hydrolysis of approx. 75% of the phosphatidylinositol in ox, sheep and cat erythrocyte “ghosts” together with no detectable effect on the major erythrocyte membrane phospholipids. The phosphatidylinositol of rat liver microsomal fraction was hydrolysed only to a maximum of 15%. Some preliminary experiments showed that approx. 60% of the phosphatidylinositol of ox or sheep erythrocytes could be hydrolysed without causing substantial haemolysis.