Phosphatidylinositol synthesis and exchange of the inositol head are catalysed by the single phosphatidylinositol synthase 1 from Arabidopsis Anne-Marie Justin, Jean-Claude Kader and Sylvie Collin Universite ´ Pierre et Marie Curie and CNRS, Laboratoire de Physiologie Cellulaire et Mole ´ culaire, Paris, France In order to study some of its enzymatic properties, phosphatidylinositol synthase 1 (AtPIS1) from the plant Arabidopsis thaliana was expressed in Escherichia coli, a host naturally devoid of phosphatidylinositol (PtdIns). In the context of the bacterial membrane and in addition to de novo synthesis, the plant enzyme is capable of catalysing the exchange of the inositol polar head for another inositol. Our data clearly show that the CDP-diacylglycerol-independent exchange reaction can occur using endogenous PtdIns molecular species or PtdIns molecular species from soybean added exogenously. Exchange has been observed in the absence of cytidine monophosphate (CMP), but is greatly enhanced in the presence of 4 l M CMP. Our data also show that AtPIS1 catalyses the removal of the polar head in the presence of much higher concentrations of CMP, in a manner that suggests a reverse of synthesis. All of the PtdIns metabolizing activities require free manganese ions. EDTA, in the presence of low Mn 2+ concentrations, also has an enhancing effect. Keywords: Arabidopsis; exchange reaction; phosphatidy- linositol; phospholipid synthesis; reverse reaction; synthase. Phosphatidylinositol (PtdIns) labelling has long been known to occur via two possible mechanisms: a de novo synthesis, catalysed by phosphatidylinositol synthase (EC 2.7.8.11), also known as CDP-diacylglycerol (DAG)/myo- inositol 3-phosphatidyltransferase, and an exchange reac- tion whereby the sugar head is exchanged between pre-existing PtdIns molecules and free inositol, leading in a test tube and in the absence of CDP-diacylglycerol (CDP- DAG) to the synthesis of labelled PtdIns when radioactive inositol is used [1]. In animal [2–6] as well as in plant tissues [7,8] and in the green algae Chlamydomonas reinhardtii [9], this exchange reaction has been associated to the endoplas- mic reticulum (ER) or to microsomal fractions rich in ER, where de novo synthesis of PtdIns is most active. Thorough enzymatic characterization of synthesis and exchange has attemptedtounderstandifbothreactionsaremediatedby the same enzyme, but only one study has shown that one gene product, phosphatidylinositol synthase from the yeast Saccharomyces cerevisiae, carries both activities [10]. In plants, for 20 years, the lack of cloned sequences had prevented the investigations necessary to determine whether the situation is the same as in yeast; however, the cloning and expression of a cDNA encoding PtdIns synthase in Arabidopsis thaliana [11] has enabled us to do so. The results presented here clearly show that AtPIS1 is able, when expressed in a bacterial system, to catalyse both de novo synthesis of PtdIns as well as exchange of the inositol polar head. In addition, we also suggest that placed in the appropriate conditions, the enzyme is able to catalyse the reaction reverse of synthesis. The substrates for exchange can be PtdIns molecular species made by the enzyme or exogenous molecular species differing in their fatty acid content. The presence of a chelating agent of manganese, which is indispensable for de novo synthesis and exchange activities [1,6,11], had the same effect on both, suggesting that the enzymatic active site used could be the same in both cases. MATERIALS AND METHODS Genetic nomenclature The cDNA used in this work corresponds to EMBL accession number H36646 [11] and gene AtPIS1. Growth conditions of the bacterial transformants Escherichia coli cells expressing the AtPIS1 cDNA enco- ding phosphatidylinositol synthase 1 from Arabidopsis thaliana (AtPIS1) were obtained in the same way and are the same as those described [11]. Two bacterial strains were used, one expressing the plant cDNA (previously called strain 2, now called strain +PIS), and the nonexpressing control strain (previously called strain 3a and now called strain –PIS). The cells were grown at 37 °CinLuria– Bertani medium (Miller, Difco) supplemented with 1 m M myo-inositol and 100 lgÆmL )1 ampicillin. In the case of Correspondence to S. Collin, Universite ´ Pierre et Marie Curie, Laboratoire de Physiologie Cellulaire et Mole ´ culaire, UMR 7632 CNRS, Paris 6, Tour 53, 4 e tage, Case Courrier 154, 4, Place Jussieu, 75252 Paris Cedex 05, France. Fax: + 33 1 44 27 61 51, Tel.: + 33 1 44 27 59 13, E-mail: scollin@snv.jussieu.fr Abbreviations: PtdIns, phosphatidylinositol; PtdOH, phosphatidic acid; CDP-DAG, CDP-diacylglycerol; CMP, cytidine monophosphate; CTP, cytidine triphosphate. Enzymes: phosphatidylinositol synthase or CDP-diacylglycerol/ myo-inositol 3-phosphatidyltransferase (EC 2.7.8.11). (Received 4 October 2001, revised 13 March 2002, accepted 20 March 2002) Eur. J. Biochem. 269, 2347–2352 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.02893.x experiments requiring bacterial membranes devoid of PtdIns, the cells were grown in M9 minimal medium [12] supplemented with 2 gÆL )1 vitamin-free casein (Sigma) and 100 lgÆmL )1 ampicillin. The water used was ultrapure, filtered on a Millipak Filter Unit 0.22 lm rated (Millipore). Expression of the cDNA was induced in fresh transform- ants as described previously [11] before the cells were washed in 50 m M Tris/HCl pH 8.0 and stored as pellets at )80 °C. These pellets were used for lipid analyses or as a source of membrane proteins. Enzymatic activities Membrane purifications were carried out at 5 °C. E. coli cells were resuspended at a density of 100 lg fresh cellsÆmL )1 in sonication buffer containing 50 m M Tris/ HCl pH 8.0, 8% (v/v) glycerol and 8 m M 2-mercaptoeth- anol. After sonication, intact cells were eliminated by centrifugation at 4500 g for 10 min and washed once in sonication buffer. The two supernatants were pooled and centrifuged at 100 000 g for 1 h. The membrane pellet was resuspended in 20 m M Tris/HCl pH 8.0, 20% (v/v) glycerol, 8m M 2-mercaptoethanol, aliquoted and stored at )80 °C. The protein concentration was determined according to Lowry et al. [13] using BSA as standard. De novo synthesis. Unless stated otherwise, the incubation conditions for de novo PtdIns synthase activity were 50 m M Tris/HCl pH 8.0, 0.3 m M CDP-DAG dipalmitoyl (Sigma), 2.4 m M Triton X-100, 0.5 m M myo-inositol (Amersham, [ 3 H]-labelled, diluted with unlabelled myo-inositol to a final activity of 500 BqÆnmol )1 ), 2.5 m M MnCl 2 and 10–50 lg membrane proteins. Samples were incubated in a final volume of 200 lLat30°C for 20 min The reaction was stopped on ice by the addition of 3 mL ice-cold methanol/ chloroform (2 : 1). Lipids were then extracted according to Sambrook et al. [12]. The chloroform phase was washed with methanol and 1% (w/v) sodium chloride in the proportion 1 : 1 : 1. Radioactivity was measured on a 100-lL aliquot mixed with 6 mL of Emulsifier Safe TM (Packard) using a scintillation counter. Exchange reaction. The exchange reaction was assayed by incubating 50–200 lg membrane proteins in 50 m M Tris/ HCl pH 8.0, 0.36 m M PtdIns from soybean (Sigma), 0.5 m M myo-inositol containing tritium-labelled myo-inos- itol for a final activity of 500 BqÆnmol )1 ,2.4m M Triton X100, 2.5 m M MnCl 2 in the presence or in the absence of 4 l M cytidine monophosphate (CMP), but without CDP- DAG. The incubation time at 30 °C was 20–30 min. CMP-dependent PtdIns hydrolysis. This reaction was followed by incubating 50–200 lg membrane proteins in 50 m M Tris/HCl pH 8.0, 3 m M CMP, 2.4 m M Triton X100, 2.5 m M MnCl 2 and 42 l M PtdIns [soybean PtdIns from Sigma diluted with [myo-inositol-2- 3 H(N)] PtdIns, (NEN) to a final activity of 240 BqÆnmol )1 ]at30°C for 20–40 min. After extraction of total lipids as described above, the reverse activity was measured on the lower phase after concentration under nitrogen and mixing with 6 mL Emulsifier Safe TM or on the upper phase washed with chloroform (2 : 1), concentrated and mixed with 12 mL Emulsifier Safe TM . Effect of EDTA on enzymatic reactions All buffers and aqueous solutions were prepared in distilled water (Autostill TM Autofour, Jencons). To the incubation medium containing the above-mentioned concentrations of Tris/HCl pH 8.0, Triton X100, MnCl 2 and EDTA were added the appropriate missing components according to the reaction studied. Each reaction was started by the addition of membrane proteins. Labelling of endogenous PtdIns molecular species Microsomes from germinating soybean (Glycine max L. cv. Weber) plantlets were prepared from 5 g seeds as described [14]. They were resuspended in the same buffer as E. coli membranes and stored at )80 °C. Membranes (1.5 mg membrane proteins for E. coli, 3 mg membrane proteins for soybean microsomes) were incubated in 50 m M Tris/HCl pH 8.0, 2.4 m M Triton X100, 5 m M EDTA, 0.1 m M cytidine triphosphate (CTP), 7.5 m M MnCl 2 and 2- 3 H(N)-myo-inositol (NEN, 0.46 MBq, 85 · 10 4 MBqÆmmol )1 ) in a final volume of 1.6 mL for 30 min at 30 °C. The reaction was stopped by the addition of 4.8 mL methanol/chloroform (2 : 1) as before and lipids were extracted. 3 H-Labelled PtdIns molecular species were identified by radio-HPLC. Analysis of radioactive PtdIns molecular species After separation by TLC according to Lepage [15], the PtdIns spot was scrapped and PtdIns eluted at 8 °C overnight from the silica in 5 mL methanol containing one drop of glacial acetic acid. After re-extraction of the sample according to Bligh & Dyer [16], PtdIns molecular species were analysed by radio-HPLC as described previ- ously [17]. RESULTS Enzymatic activities catalysed by AtPIS1 Previously published results indicate that the yeast PtdIns synthase placed in a microbial environment is capable of catalysing several enzymatic reactions: PtdIns de novo synthesis, using CDP-DAG and free inositol as substrates and producing PtdIns and CMP; a CMP-independent exchange reaction whereby the inositol molecule of PtdIns can be replaced by another inositol polar head without net synthesis of PtdIns; a CMP-dependent exchange reaction, far greater in quantitative terms than the CMP-independent one; and, finally, a reaction which is the reverse of synthesis, dependent on higher concentra- tions of CMP [10]. As the exchange reaction has also been suggested to occur in plants [7,8], we tested it using isolated bacterial membranes as a source of enzyme. The transformation of E. coli, a host naturally devoid of PtdIns synthase with a single cDNA ensured that the recombinant protein was the only candidate for the activities tested. E. coli cells expressing the AtPIS1 cDNA were cultivated on M9 minimal medium supplemented with vitamin-free amino acids to allow isolation of bacterial membranes lacking endogenous PtdIns, which would interfere with 2348 A M. Justin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 exogenous PtdIns. De novo synthesis was measured in the incubation conditions described above, in the presence of CDP-DAG and myo-inositol (Table 1). The exchange reaction was assayed by incubating bacterial membranes with 3 H-labelled free myo-inositol and cold soybean PtdIns, no exogenous CDP-DAG, in the presence or absence of 4 l M CMP. The PtdIns concentration used was 0.36 m M as in Klezovitch et al. [10]. At the end of the incubation period total lipids were extracted and the amount of labelled PtdIns was determined (Table 1). In the absence of CMP, the presence of AtPIS1 allows the incorporation of labelled inositol into PtdIns, according to the enzymatic reaction called exchange. This reaction is greatly enhanced by the presence of 4 l M CMP, about ninefold under our conditions. The assay for a reverse reaction was carried out by measuring the CMP-dependent release of label from 3 H-PtdIns in the aqueous phase of total lipid extractions as in Bligh & Dyer [16]. Radioactivity was measured in the lower chloroform and upper aqueous phases. In the absence of CMP and in both +PIS and –PIS strains, a release of label in the aqueous phase was observed after incubation. To calculate the liberation of radioactivity due to a putative reverse reaction, the amount of label released in the absence of CMP was substracted from that released in the presence of 3 m M CMP (Table 1). Mem- brane samples containing PtdIns synthase catalyse the liberation of label in the upper phase (which would correspond to 130 pmol inositolÆmin )1 Æmg )1 ), counterbal- anced by an equivalent disappearance of PtdIns in the lower phase ()108 pmolÆmin )1 Æmg )1 ). On the other hand, mem- branes lacking AtPIS1 do not induce any variation in the radioactivity partitioned between each phase during the incubation time, indicating that the appearance of 3 H-label in the upper phase is specific to AtPIS1. Several 5¢-CMP concentrations ranging between 0 and 6 m M were tested to assess the CMP dependence of PtdIns hydrolysis. The results (data not shown) do indicate that this is the case. Nevertheless, the appearance of CDP-DAGs could not be detected (data not shown). This protein-specific release and CMP-dependent release of label from PtdIns could be explained by a reaction which is the reverse of synthesis. The exchange reaction studied by analysis of radioactive PtdIns molecular species The exchange reaction was also studied by analysis of the radiolabelled PtdIns molecular species produced using various PtdIns substrates (Fig. 1). This experiment is based on the fact that the Ptdns molecular species synthesized in E. coli from endogenous CDP-DAG are very different from those found in plants, in particular soybean, whose PtdIns composition has been described previously [14]. PtdIns species made from endogenous CDP-DAG were first Table 1. Enzymatic reactions catalysed by AtPIS1 expressed in E. coli. Transformant E. coli were cultivated on a medium lacking inositol, membranes were purified as described in Materials and methods and incubated under various conditions with different substrates. The values indicated and in pmol PtdInsÆmin )1 Æmg )1 protein. In the upper phase they are given as pmol inositolÆmin )1 Æmg )1 .Eachvalueisthe mean ± SD of four (+PIS) or three (–PIS) experimental points using the same protein sample. Enzymatic reactions carried out with mem- branes isolated from different cultures of transformant E. coli cells gave very similar results. Conditions for: reverse reaction, + 3 H-PtdIns + 3 m M CMP; exchange reaction, + 3 H-Ins + 0.36 m M PtdIns. +PIS –PIS + CDP-DAG De novo synthesis 2019 ± 71 0 – CDP-DAG Reverse reaction Lower phase )108 ± 5 )1 Upper phase 103 ± 13 0 Exchange reaction – CMP 125 ± 5 0 +4l M CMP 1109 ± 138 0 Fig. 1. Radio-HPLC separation of PtdIns molecular species synthes- ized: (A) by AtPIS1 present in E. coli membranes incubated with CTP and myo-inositol, but no exogenous CDP-DAGs; (B) by germinating soybean microsomes, under the same substrate conditions as in (A); (C) in exchange conditions in the presence of 4 l M CMP and soybean PtdIns by membranes from E. coli +PIS grown on a medium supplemented with myo-inositol or (D) lacking myo-inositol. The incubation conditions are as described in Materials and methods. Ó FEBS 2002 Arabidopsis phosphatidylinositol synthase 1 (Eur. J. Biochem. 269) 2349 studied by incubating membranes from E. coli +PIS cultivated on medium lacking inositol or microsomes from germinating soybean in the presence of CTP and 3 H-labelled inositol as in Justin et al. [14]. No exogenous CDP-DAG was added so that PtdIns could only arise from an endogenous production following the reactions: phos- phatidic acid (PtdOH) + CTP fi CDP-DAG + PPi (pyrophosphate) followed by CDP-DAG + 3 H-inosi- tol fi 3 H-PtdIns + CMP. PtdIns molecular species were then separated by radio-HPLC. In E. coli (Fig. 1A), C14:0/ C16:0 PtdIns elutes almost at the same time as the C16:0/ C18:3 molecular species in soybean (Fig. 1B), but other molecular species are eluted at times which allow a perfect distinction between the two sets of PtdIns molecules: the C16:0/C17c + C18:1/C17c and C16:0/C19c peaks are characteristic of E. coli (A M. Justin, unpublished data) whereas the C16:0/C18:2 and C18:0/C18:2 PtdIns are characteristic of soybean [14]. In a second experiment, membranes isolated from E. coli + PIS grown on Luria–Bertani medium and inos- itol (Fig. 1C) or M9 medium lacking inositol (Fig. 1D) were incubated in conditions of exchange in the presence of PtdIns isolated from soybean. The radio-HPLC elution profile of PtdIns shows that when endogenous PtdIns is present in the bacterial membrane, molecular species typical of E. coli and others of soybean become labelled (Fig. 1C). The bacterial membrane contains endogenous PtdIns that could be used as a substrate for exchange of the inositol head but also endogenous CDP-DAG molecules which can be used for de novo synthesis of PtdIns. Nevertheless, when bacterial membranes lacking endo- genous PtdIns were incubated in the same exchange conditions, the only labelled PtdIns molecular species detected were of soybean origin (Fig. 1D), showing: (a) that exogenous soybean PtdIns is used for exchange; (b) that the endogenous bacterial CDP-DAGs are not present in sufficient amount to give rise to detectable de novo synthesized PtdIns molecular species. In Fig. 1C, the labelled PtdIns of bacterial origin therefore arises from an exchange reaction. Effect of EDTA on the enzymatic activities of PtdIns synthase The net activity of PtdIns synthase as an exchange enzyme stimulated by low concentrations of CMP is far from being negligible when compared to de novo synthesis (Table 1). For evaluation of net synthesis, one might wish to inhibit the exchange and one laboratory reported, to this end, that the addition of 5 m M EDTA was sufficient to abolish the exchange ability of PtdIns synthase when incubated in conditions of de novo synthesis [8]. In our own experiments, we have observed that adding 5 m M EDTA leads to an enhancement of inositol incorporation provided that the concentration of manganese is correspondingly increased to 7.5 m M to compensate the chelating effect of EDTA (unpublished results). We therefore investigated the effect of EDTA on the reactions catalysed by PtdIns synthase in the presence of 7.5 m M MnCl 2 . The results are shown in Fig. 2A and B. At 0 m M EDTA, i.e., 7.5 m M MnCl 2 , de novo synthesis reached 0.83 nmol PtdInsÆmg )1 Æmin )1 (Fig. 2A). As the EDTA concentration increased, synthesis increased up to a maximum activity of 2.0 nmol PtdInsÆmg )1 Æmin )1 between 2.5 and 5 m M EDTA. At 10 m M ,synthesis decreased to 0. In the conditions used here we detected two exchange activities, one stimulated by CMP and one independent of CMP. The CMP-dependent activity showed an incorporation profile of labelled inositol that closely paralleled that of the de novo synthesis activity (Fig. 2A), with values close to 0.8 nmol PtdInsÆmg )1 Æmin )1 at 0 m M EDTA, increasing to 2 nmol PtdInsÆmg )1 Æmin )1 between 2.5 and 5 m M EDTA and a drop to 0 at 10 m M . Abolition of all exchange activity at 10 m M EDTA shows that the process requires manganese ions. On the other hand, the CMP-independent activity was much lower, with a value close to 0.15 nmol PtdInsÆmg )1 Æmin )1 at 0 m M EDTA which peaked 1 at 0.35 nmol PtdInsÆmg )1 Æmin )1 between 2.5 and 5 m M . Comparatively, the CMP-depend- ent PtdIns hydrolysing activity was the lowest of all in our conditions. To evaluate better the effect of the chelator on each reaction, each activity was normalized to that obtained at 5m M EDTA and plotted as a function of the free Mn 2+ concentration (Fig. 2B). The data show that synthesis and exchange with or without CMP behave in the same way according to the concentration in available Mn 2+ . Fig. 2. Effect of EDTA on the enzymatic reactions catalysed by PtdIns synthase. The results shown here are typical of those observed for at least two independent cultures. (A) All activities were measured in the presence of 7.5 m M MnCl 2 . Fifty lg microsomal proteins were incu- batedfor20minasdescribedinMethods.d, De novo synthesis; j, exchange reaction in the presence of 4 l M CMP; h, exchange reaction with no CMP added; m, CMP-dependent PtdIns hydrolysis. (B) De novo synthesis (light grey), exchange activities (–CMP, striped bar; +CMP, medium grey) and CMP-dependent PtdIns hydrolysis (dark grey) normalized to their respective values obtained at 5 m M EDTA, plotted as a function of EDTA concentration in the medium. (C) Comparison of de novo synthesis activities obtained with different concentrations of EDTA at constant manganese concentrations: 7.5 m M (d)or2.5m M (s). (D) Comparison of de novo synthesis activities as a function of the calculated free Mn 2+ concentration available in the medium at 7.5 m M MnCl 2 and different concentrations of EDTA (black circles), 2.5 m M MnCl 2 and various amounts of EDTA (s)ordifferentconcentrationsofMnCl 2 [11] (m). At 0 m M Mn 2+ , all enzyme activities are null. Because of the x axis chosen, activities obtained at submicromolar Mn 2+ concentrations in the experiments involving fixed concentrations of manganese chloride and varying amounts of EDTA appear on the y axis just above 0. 2350 A M. Justin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Effect of EDTA as a function of the concentration in free manganese ions Theconcentrationoffreemanganeseionswascalculatedat each EDTA concentration using an apparent EDTA stability constant K 1 of 5 · 10 11 Æ M )1 for manganese at pH 8.0 [18]. At 7.5 m M EDTA, when the calculated con- centration of free manganese ions is in the submicromolar range, a synthesis activity of 1.53 nmol PtdInsÆmg )1 Æmin )1 could still be measured (Fig. 2C). When this activity was tested at a lower MnCl 2 concentration with various amounts of EDTA, the highest activity was not observed at 0 m M EDTA and 2.5 m M MnCl 2 as seen before [11], but at 2.5 m M EDTAwhentheconcentrationinfreemanganeseionsis close to 0.1 l M . An alignment on the same plot of the values found for de novo synthesis in the presence of 7.5 or 2.5 m M MnCl 2 with various amounts of EDTA, or in the presence of MnCl 2 only [11], as a function of the concentration in free manganese (Fig. 2D), shows that in the presence of EDTA concentrations < 2.5 m M free Mn 2+ give higher activities than those observed with Mn 2+ alone, suggesting an activating effect of EDTA on enzymatic activity. DISCUSSION The exchange reaction Using the same incubation conditions for PtdIns and CMP as those published by Klezovitch et al. [10] for examination of the PtdIns synthase from yeast expressed in a bacterial system, we found that the Arabidopsis enzyme expressed in E. coli is able to catalyse not only de novo synthesis of PtdIns but also the exchange reactions in the absence or in the presence of CMP. In plants, the exchange reaction is known to occur [7,8] but it has never been shown to be catalysed by the same enzyme as that responsible for de novo synthesis. The experimental conditions used reproduce the parameters defined as optimum by Sexton & Moore [7] and Sandelius & Morre ´ [8] with a pH of 8.0 and the presence of manganese ions. In the range of concentrations tested, we found an optimum free manganese ion concentration of 2.5 m M , a value that is in the same range as the concen- trations found by the authors cited above. In a preliminary work (data not shown), we have observed that when E. coli membranes prepared from cells grown in inositol-containing medium are incubated with labelled inositol but no CDP-DAG and no PtdIns, no labelled PtdIns appears, showing that the endogenous level of CDP-DAG is too low to allow detectable synthesis. The level of PtdIns in the membranes does not allow synthesis of detectable PtdIns by exchange of the polar head. As in Fig. 1D the only source of PtdIns molecular species characteristic of E. coli that we detect can only come from the bacterial membrane, we explain this result by a global stimulation of the exchange reaction by the exogenous soybean PtdIns. A question that arises is the relevance of the exchange of polar head when de novo synthesis is studied. We calculated that with a linear rate of synthesis of 2 nmol PtdInsÆmin )1 Æmg )1 (Fig. 2), the concentration of CMP in a reaction mixture of 200 lLis4l M after 1 min. with 50 lg protein. As our incubation conditions use 50 lg protein incubated for 20 min, although the corresponding PtdIns concentration does not reach the value used to study exchange, it is possible that this latter reaction takes place, possibly interfering with the net de novo synthesis capacity of PtdIns synthase, especially if the membranes used as a source of enzyme are rich in PtdIns. Another question is the reason why CMP stimulates the exchange reaction and what the exact mechanism for exchange is. Our data does not allow us to distinguish between a reverse reaction followed by re-synthesis, as suggested by Paulus & Kennedy [19], and a real exchange. The fact that EDTA has the same effect, in particular a stimulation between 2.5 and 5 m M , on all reactions catalysed by PtdIns synthase seems to suggest that the chemical reactions involved are not very different, but that the enzymatic parameters in favour of synthesis, reverse reaction or exchange are quite different in terms of concentrations of the reactants in each case; a simple reverse reaction followed by synthesis is puzzling or in any case very difficult to analyse in terms of kinetics. From a functional point of view, if a reaction reverse of synthesis followed by synthesis using CDP-DAGs different from those liberated was responsible for the exchange activity we detected, it could be a mechanism whereby in defined conditions cells could specifically destroy particular PtdIns molecular species to replace them by others more adapted to new physiological or environmental parameters. Removal of the polar head The data presented here are not an absolute proof that PtdIns synthase catalyses a true reverse reaction. The enzymatic activity we detected is nevertheless strictly dependent on CMP, and strictly associated with the AtPIS1 protein. The dependence of the PtdIns synthase reverse reaction has been extensively studied in rat pituitary GH3 cells, where the authors have identified the reverse products and shown that the CMP activation of the reverse of synthesis is by a different mechanism from base exchange [20]. Further analysis of the products released from PtdIns by the Arabidopsis enzyme will allow us to carry out a deeper study of this particular activity of the protein. Effect of EDTA on PtdIns synthase reactions Our data show that at free manganase ion concentrations far lower than those found to be the optimum when manganese is used on its own, there are still de novo synthesis and exchange reactions the activities of which are close to those seen when no chelating agent is used. This effect has been studied particularly in the case of de novo synthesis, where a comparison between data presented here and data published previously by us clearly show that EDTA enhances the activity. The pH remained unchanged between each condition of the EDTA plot, which rules out a simple pH effect to explain the variations in enzymatic activity. One possibility therefore is that EDTA could chelate inhibitory ions, whose identity remains unclear, maybe with a higher affinity than for Mn 2+ . If it is clear, for the first time, that a PtdIns synthase from Arabidopsis is able to catalyse both de novo synthesis of PtdIns and the exchange of the inositol moiety, but it is still uncertain by what mechanism this exchange is carried out. In terms of kinetics, the enzymatic conditions used do not Ó FEBS 2002 Arabidopsis phosphatidylinositol synthase 1 (Eur. J. Biochem. 269) 2351 favour a reverse reaction followed by re-synthesis. Never- theless, EDTA seems to have the same effect on both synthesis and exchange, so further characterization of each activity is needed to see whether PtdIns synthase uses two different active sites, as made possible by the existence of two hydrophilic pockets in the protein [11], whether the same active site can adopt different conformations accord- ing to substrate concentrations or physico-chemical condi- tions or, finally, whether exchange is, mechanistically, an inverse synthesis followed by re-synthesis. The work carried out on the rabbit lung and rat liver microsomal enzyme seemed to suggest different enzymes as the pH and divalent cation requirements were different for the two activities [4,21] but their results could perhaps be explained by two different catalytic sites. Sexton & Moore also suggested for the castor bean endosperm enzyme that exchange is not a reversal of the transferase activity [7]. No obvious sequence similarities with other proteins catalysing an exchange of phospholipid head, such as phosphatidylserine synthase (CMP-PtdOH/ L -serine 3-phosphatidyltransferase), came up when alignments were made (data not shown) so the question remains open until further data, currently being compiled in our laboratory, are obtained, with the aim of understanding how PtdIns synthase functions and is regulated. ACKNOWLEDGEMENTS The authors thank A. Zachowski, E. Ruelland and A. Jolliot-Croquin for helpful discussions and advice. This project was funded by the French Ministries of Research and Education via CNRS and the University Pierre et Marie Curie (Unite ´ Mixte de Recherche 7632). REFERENCES 1. Moore, T.S. (1990) Biosynthesis of phosphatidylinositol. In Inos- itol Metabolism in Plants (Morre ´ , D.J., Boss, W.F. & Loewus, F.A., eds), pp. 107–112. Wiley-Liss, New York. 2. Holub, B.J. (1974) The Mn 2+ -activated incorporation of inositol into molecular species of phosphatidylinositol in rat liver micro- somes. Biochim. Biophys. Acta 369, 111–122. 3. Takenawa, T. & Egawa, K. (1980) Phosphatidylinositol:myo- inositol exchange enzyme from rat liver: partial purification and characterization. Arch. Biochem. Biophys. 202, 601–607. 4. Bleadsdale, J.E. & Wallis, P. (1981) Phosphatidylinositol-inositol exchange in rabbit lung. Biochim. Biophys. Acta 664, 428–440. 5. McPhee, F., Lowe, G., Vaziri, C. & Downes, C.P. (1991) Phosphatidylinositol synthase and phosphatidylinositol/inositol exchange reactions in turkey erythrocyte membranes. Biochem. J. 275, 187–192. 6. Irvine, R.F. (1998) Manganese-stimulated phosphatidylinositol headgroup exchange in rat liver microsomes. Biochim. Biophys. Acta 1393, 292–298. 7. Sexton, J.C. & Moore, T.S. Jr (1981) Phosphatidylinositol synthesis by a Mn 2+ -dependent exchange enzyme in castor bean endosperm. Plant Physiol. 68, 18–22. 8. Sandelius, A.S. & Morre ´ , D.J. (1987) Characteristics of a phos- phatidylinositol exchange activity of soybean microsomes. Plant Physiol. 84, 1022–1027. 9. Moore, T.S. Jr & Blouin, A.M. (2001) Membrane lipid biosynthesis in Chlamydomonas reinhardtii. Identification and partial characterization of CDP-diacylglycerol: inositol transferase and myo-inositol exchange reactions. Quadrennial Joint Meetings of the American Society of Plant Biologists and the Canadian Society of Plant Physiologists, Plant Biology, Providence, Rhode Island, USA 21–25 July 2001. Poster, 460, 401–402. 10. Klezovitch, O., Brandenburger, Y., Geindre, M. & Deshusses, J. (1993) Characterization of reactions catalysed by yeast phospha- tidylinositol synthase. FEBS Lett. 320, 256–260. 11. Collin, S., Justin, A M., Cantrel, C., Arondel, V. & Kader, J C. (1999) Identification of AtPIS, a phosphatidylinositol synthase from Arabidopsis. Eur. J. Biochem. 262, 652–658. 12. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, 2nd edn. Cold Spring. Harbor Laboratory Press, Cold Spring Harbor, New York. 13. Lowry, O.H., Rosebrough, N.J., Farr, A.L. & Rendall, R.J. (1952) Protein measurements with the Folin phenol reagent. J. Biol. Chem. 193, 265–275. 14. Justin, A M., Demandre, C. & Mazliak, P. (1989) Molecular species synthesized by phosphatidylinositol synthases from potato tuber, pea leaf and soya bean. Biochim. Biophys. Acta 1005, 51–55. 15. Lepage, M. (1967) Identification and composition of turnip root lipid. Lipids 2, 244–250. 16. Bligh, E.G. & Dyer, W.J. (1959) A rapid method for total lipid extraction and purification. Can. J. Biochem. Physiol. 37, 911–917. 17. Justin, A M., Hmyene, A., Kader, J C. & Mazliak, P. (1994) Compared selectivities of the phosphatidylinositol synthase from maize coleoptiles either in microsomal membranes or after solu- bilization. Biochim. Biophys. Acta 1255, 161–166. 18. Dawson, R.M.C., Elliot, D.C., Elliot, W.H. & Jones, K.M. (1969) Data for Biochemical Research, 2nd edn, pp. 426–427. Oxford University Press, Oxford. 19. Paulus, H. & Kennedy, E.P. (1960) J. Biol. Chem. 235, 1303–1311. 20. Cubitt, A.B. & Gershengorn, M.C. (1990) CMP activates reversal of phosphatidylinositol synthase and base exchange by distinct mechanisms in rat pituitary GH 3 cells. Biochem. J. 272, 813–816. 21. Takenawa, T., Saito, M., Nagai, Y. & Egawa, K. (1977) Solubi- lization of the enzyme catalyzing CDP-DAG-independent incorporation of myo-inositol into phosphatidylinositol and its comparison to CDP-diglyceride: inositol transferase. Arch. Bio- chem. Biophys. 182, 244–250. 2352 A M. Justin et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Phosphatidylinositol synthesis and exchange of the inositol head are catalysed by the single phosphatidylinositol synthase 1 from Arabidopsis Anne-Marie. novo synthesis, catalysed by phosphatidylinositol synthase (EC 2.7.8 .11 ), also known as CDP-diacylglycerol (DAG)/myo- inositol 3-phosphatidyltransferase, and