Overexpression of Phosphatidylinositol Transfer Protein α in NIH3T3 Cells Activates a Phospholipase A*

In order to investigate the cellular function of the mammalian phosphatidylinositol transfer protein α (PI-TPα), NIH3T3 fibroblast cells were transfected with the cDNA encoding mouse PI-TPα. Two stable cell lines, i.e. SPI6 and SPI8, were isolated, which showed a 2- and 3-fold increase, respectively, in the level of PI-TPα. Overexpression of PI-TPα resulted in a decrease in the duration of the cell cycle from 21 h for the wild type (nontransfected) NIH3T3 (wtNIH3T3) cells and mock-transfected cells to 13–14 h for SPI6 and SPI8 cells. Analysis of exponentially growing cultures by fluorescence-activated cell sorting showed that a shorter G1 phase is mainly responsible for this decrease. The saturation density of the cells increased from 0.20 × 105 cells/cm2 for wtNIH3T3 cells to 0.53 × 105 cells/cm2 for SPI6 and SPI8 cells. However, anchorage-dependent growth was maintained as shown by the inability of the cells to grow in soft agar. Upon equilibrium labeling of the cells withmyo-[3H] inositol, the relative incorporation of radioactivity in the total inositol phosphate fraction was 2–3-fold increased in SPI6 and SPI8 cells when compared with wtNIH3T3 cells. A detailed analysis of the inositol metabolites showed increased levels of glycerophosphoinositol, Ins(1)P, Ins(2)P, and lysophosphatidylinositol (lyso-PtdIns) in SPI8 cells, whereas the levels of phosphatidylinositol (PtdIns) and phosphatidylinositol 4,5-bisphosphate were the same as those in control cells. The addition of PI-TPα to a total lysate ofmyo-[3H]inositol-labeled wtNIH3T3 cells stimulated the formation of lyso-PtdIns. The addition of Ca2+ further increased this formation. Based on these observations, we propose that PI-TPα is involved in the production of lyso-PtdIns by activating a phospholipase A acting on PtdIns. The increased level of lyso-PtdIns that is produced in this reaction could be responsible for the increased growth rate and the partial loss of contact inhibition in SPI8 and SPI6 cells. The addition of growth factors (platelet-derived growth factor, bombesin) to these overexpressers did not activate the phospholipase C-dependent degradation of phosphatidylinositol 4,5-bisphosphate.

Phospholipid transfer proteins are proteins that are able to transfer phospholipids between membranes in vitro. A major phospholipid transfer protein in mammalian tissues is the phosphatidylinositol transfer protein (PI-TP) 1 (1). Recently, two isoforms of PI-TP have been identified (i.e. PI-TP␣ and PI-TP␤) that demonstrate differences in cellular localization and in specific lipid transfer activity (2)(3)(4)(5).
PI-TP␣ has been purified from both rat and bovine brain (6,7). Cloning of the cDNA encoding rat brain PI-TP␣ showed that the protein consists of 271 amino acid residues (8). The subsequent isolation of the cDNAs encoding mouse and human PI-TP␣ revealed a high homology between the different mammalian PI-TPs (about 99% amino acid sequence identity) (9,10). Furthermore, the cross-reactivity of the antibodies raised against bovine PI-TP␣ with a 35-kDa protein from other animals (e.g. rat, mouse, chicken, frog, and lizard) indicates an extensive conservation of the amino acid sequence between species (11). An exception is PI-TP from yeast (i.e. SEC14p) that has the same molecular weight as mammalian PI-TP and comparable phospholipid transfer activities yet shows no homology in the amino acid sequence (12)(13)(14).
So far, very little is known about the precise cellular role of mammalian PI-TP␣. Since PI-TP␣ is able to transfer in vitro PtdIns between membranes in exchange for phosphatidylcholine, it was proposed that PI-TP␣ has a function in the transfer of PtdIns from its site of synthesis in the endoplasmic reticulum to other cellular membranes in order to maintain the level of PtdIns upon metabolism (15)(16)(17). PtdIns is a precursor molecule for several intracellular (and possibly also extracellular) lipid messengers, the best characterized of which are 1,2-diacylglycerol and inositol 1,4,5-trisphosphate (Ins(1,4,5)P 3 ) (18,19). These messengers are formed when PtdIns is phosphorylated by PtdIns 4-kinase and phosphatidylinositol 4-phosphate 5-kinase to phosphatidylinositol 4,5-bisphosphate (PtdIns (4,5)P 2 ), which subsequently is degraded by PLC. Other PtdIns derivatives of potential biological significance include those formed in the PtdIns 3-kinase pathway (20,21), the inositol polyphosphates (22), the cyclic inositolphosphates (23), the glycerophosphoinositols (24 -28), and lysophosphatidylinositol (lyso-PtdIns) (29 -32). A number of recent studies suggest a role of PI-TP␣ in the production of several of these derivatives. Thomas et al. (33) showed that PI-TP␣ is an essential cytosolic factor to stimulate PLC␤ activity in permeabilized HL60 cells. Furthermore, Cunningham et al. (34) showed that PI-TP␣ promotes the synthesis of PtdIns(4,5)P 2 . Recently, it was shown in permeabilized human neutrophils that PI-TP␣ stimulates the formylmethionyl leucylphenylanaline-dependent production of phosphatidylinositol 3,4,5-trisphosphate in the presence of Pt-dIns 3-kinase ␥ (35). Moreover, in permeabilized PC12 cells, PI-TP␣ was found to be one of the three essential factors needed for the ATP-dependent, Ca 2ϩ -regulated fusion of secretory granules with the plasma membrane (36). An additional effect on secretion was shown in permeabilized HL60 cells, where PI-TP␣ and PI-TP␤ were able to restore GTP␥S-stimulated protein secretion in the presence of ADP-ribosylation factor (37). In a cell-free system containing trans-Golgi membranes it was shown that PI-TP␣ (as well as PI-TP␤) stimulates the formation of constitutive secretory vesicles and immature secretory granules (38). These results indicate that PI-TP has a function in intracellular membrane traffic from the Golgi to the plasma membrane that may be linked to the production of intracellular lipid messengers.
The above studies, using semi-intact cells and in vitro systems, would indicate that PI-TP acts in different compartments of the cell, in particular at the plasma membrane and at the Golgi membranes. Localization studies by indirect immunofluorescence and by microinjection of fluorescently labeled purified PI-TP␣ and PI-TP␤ into intact mammalian cells have shown that PI-TP␣ is mainly localized in the nucleus and in cluster-like structures in the cytosol and that PI-TP␤ is mainly associated with the Golgi membranes (3,4,39,40). However, upon stimulation of the cells by different growth factors (bombesin, PDGF) that stimulate the phospholipase C-dependent degradation of PtdIns(4,5)P 2 , accumulation of PI-TP␣ near the plasma membrane was not observed. Thus, no correlation was found between the cellular localization of PI-TP␣ and its proposed sites of action.
In order to gain further insight in the function and the mechanism of action of PI-TP␣, we have established stable cell lines that overexpress PI-TP␣. In this paper, we show that overexpression of PI-TP␣ in NIH3T3 cells affects the phenotype, the growth characteristics, and the inositol lipid metabolism of these cells.

Materials
The pBluescript vector SKϩ was from Stratagene (La Jolla, CA). The anti-PI-TP antibodies were raised in rabbits against synthetic peptides representing the amino acid sequence of predicted epitopes in rat brain PI-TP␣ (39). Geneticin G418 and goat anti-rabbit IgG conjugated with alkaline phosphatase were obtained from Sigma. Goat anti-rabbit IgG conjugated with fluorescein isothiocyanate was from Nordic Immunological Laboratories (Tilburg, The Netherlands). Nitrocellulose membranes were from Schleicher and Schuell. Agar-agar (Agar Noble) was obtained from Difco, RNase A from Roche Molecular Biochemicals, and myo-[2-3 H]inositol from Amersham Pharmacia Biotech.

pSG5-PI-TP␣ Construct
The cDNA encoding mouse PI-TP␣ was isolated and cloned into the pBluescript vector (9). The PI-TP␣ cDNA contains a NcoI restriction site around the translational start codon and an EcoRI and XhoI site downstream of the translational stop codon. The NcoI-XhoI fragment was isolated (including the EcoRI site) and ligated into the cloning vector pUC21 (41) in the corresponding restriction sites in order to introduce an extra EcoRI site upstream of the PI-TP␣ cDNA. The resulting EcoRI fragment (containing the complete coding cDNA) was cloned into the unique EcoRI site of the pSG5 expression vector (42). A construct was selected with the cDNA encoding PI-TP␣ in the sense direction. This construct will be denoted as pSG5-PI-TP␣. The expression of PI-TP␣ will be regulated by the SV40 early promoter, and polyadenylation will be provided by the SV40 poly(A)-adenylation signal (42).

Transfections
wtNIH3T3 fibroblast cells were seeded 5 h prior to transfection at a density of 1.3 ϫ 10 4 cells/cm 2 . Cells were co-transfected with 30 g of pSG5-PI-TP␣ and 10 g of pSV2-neo (43) using a modified calcium phosphate precipitation technique at a CO 2 concentration of 7.5% (44). Fresh medium was added 20 h after transfection, and the next day the cells were seeded in new flasks at a density of 2500 cells/10 cm 2 . After 24 h, neomycin (400 g/ml Geneticin G418) was added for the selection of neomycin-resistant cells. Fresh medium containing neomycin was added every 4 days, and resistant clones were identified after 2 or 3 weeks of growth.

Gel Electrophoresis and Immunoblotting
The PI-TP␣ content of several neomycin-resistant clones was analyzed by immunoblotting with anti-PI-TP antibodies. Confluent cell cultures were washed twice with PBS0 (phosphate-buffered saline without Ca 2ϩ and Mg 2ϩ ) and removed from the dish by incubation with 8 mM EGTA in PBS0 for 5 min at 37°C. The cells were centrifuged, and the pellet was stored at Ϫ20°C. A cell homogenate in 0.1 ml of SET buffer (10 mM Tris-HCl, pH 7.4, 1 mM EDTA, and 0.25 M sucrose) was prepared in a Dounce homogenizer, followed by sonication (1 min at 50 watts). The homogenate was centrifuged for 10 min at 17,000 ϫ g, and the A 280 of the supernatant was used to calculate the protein content. 17.5 g of supernatant protein was loaded on an SDS-polyacrylmide gel, and gel electrophoresis was performed as described (39). The proteins were electrophoretically transferred to a nitrocellulose sheet in a Multiphor II Nova Blot electrophoretic transfer unit (Amersham Pharmacia Biotech) at room temperature applying 1 mA/cm 2 of gel for 2 h, and PI-TP␣ was detected as described (39). Quantification of the PI-TP␣ levels on an immunoblot was performed by scanning with a Bio-Rad GS 700 imaging densitometer equipped with an integrating program, with known PI-TP␣ concentrations as a standard.

Growth Assay
Cells were seeded at a density of 5 ϫ 10 4 cells/dish (9 cm 2 ) in DMEM containing 10% NCS. Cell growth was determined by counting the cells every day for 9 days (in duplicate). The saturation density of the different cell lines was determined by seeding 10 5 cells/dish (9 cm 2 ), and the number of cells was determined after 7 days (in triplicate). The medium was changed every 3 days in both assays.
To determine the ability of the different cell lines to grow in soft agar, 2 ϫ 10 4 cells were suspended in 0.3% agar in DMEM containing 10% NCS and layered on 0.5% agar in the same medium. Fresh medium was added every 5 days. The colony growth was determined after 3 weeks.

Analysis of the Cell Cycle by Fluorescence-activated Cell Sorting
Cells were resuspended in 10 mM EDTA in PBS and washed once with PBS. The cells were fixed in PBS/methanol (3:7, v/v) for 20 min at 4°C. A 9-fold excess of PBS was added, and the suspension was centrifuged for 5 min at 2500 rpm. The cell pellet was resuspended in 100 l of PBS containing RNase A (1 mg/ml). Propidium iodide (50 g/ml) was added, and the cells were incubated for 30 min at 37°C. The samples were diluted 10-fold with PBS before analysis by fluorescence-activated cell sorting.

Labeling of the Cells: Extraction and Analysis of Inositol Metabolites
Two methods were used to analyze the inositol metabolites. The first method was used to obtain a quantitative preparation of the watersoluble total inositol phosphate (IP n ) fraction. The second method was used to analyze the composition of the inositol phosphate fraction and of the inositol phospholipids.
Method 1-The cells were grown in a six-well plate. 60 -70% confluent cell cultures were incubated for 48 h with 1 Ci of myo-[ 3 H]inositol in HEPES-buffered DF medium without inositol, containing 2% dialyzed NCS. Cultures were washed twice with PBS0 and scraped in 20 mM Tris buffer, pH 7.4, containing 0.25 M sucrose, 1 mM EDTA, 0.1% Nonidet P-40, and 10 mM LiCl. The cells were sonicated for 1 min in a sonication bath (Branson 1200), and a small sample was removed for protein determination. The cells were extracted by a modified Bligh and Dyer method (45). Upon phase separation, the organic phase was washed twice with MeOH/CHCl 3 , and the water phases were combined and loaded on a Seppak column (Waters Accell TM Plus QMA Cartridges) Free myo-[ 3 H]inositol was eluted from the column by water, and the total IP n fraction was eluted with 500 mM triethyl ammonium hydrogen carbonate buffer. Radioactivity was determined by liquid scintillation counting. The organic phase, including the protein layer, was acidified by the addition of 0.03 N HCl and washed twice with H 2 O/MeOH/0.03 N HCl. The inositol phospholipids were separated by thin layer chromatography.
Method 2-The cells were grown in a 12-well plate and labeled as in method 1 except that 5 Ci of myo-[ 3 H]inositol was used. After labeling, the cells were stimulated for 10 min with PDGF (20 ng/ml) or bombesin (10 nM). Prior to stimulation, 0.5 ml of DF medium without inositol, containing 0.3% bovine serum albumin and 10 mM LiCl, was added to the cells. After 10 min at 37°C, the incubation was continued for 15 min in the absence or presence of the growth factors. Cultures were washed twice with PBS0 and harvested by scraping in 1 ml of Ϫ20°C MeOH. myo-[ 3 H]Inositol, the inositol phosphate fraction, and the inositol phospholipids were extracted and analyzed as described previously (29,30).
In Vitro Assay of PI-specific Phospholipase Activity wtNIH3T3 cells were labeled to equilibrium with myo-[ 3 H]inositol (10 Ci) for 48 h in 5 ml of DF medium without inositol containing 2% dialyzed NCS/50-cm 2 dish. The cells were scraped in 1.5 ml of 50 mM HEPES, pH 7.4, containing 350 mM sucrose, 154 mM NaCl, 1 mM EDTA, 10 g/ml leupeptin, 10 g/ml aprotinin, 10 g/ml phenylmethanesulfonyl fluoride and homogenized by 10 strokes in a Dounce homogenizer. The homogenate was centrifuged at 400 rpm for 2 min in an Eppendorf centrifuge at 4°C to remove unbroken cells and cell debris, and 0.4 ml of the supernatant, containing about 600,000 dpm in [ 3 H]inositol derivatives, was incubated for 30 min at 37°C in the absence or presence of Ca 2ϩ (5 mM) and PI-TP␣. Lipid extraction and analysis were performed as described in method 2 (see above).

Mouse NIH3T3 Fibroblast Cells
Stably Transfected with the DNA Encoding PI-TP␣-The cDNA encoding mouse PI-TP␣ (9) was cloned into the expression vector pSG5, and the vector with PI-TP␣ in the sense orientation is denoted as pSG5-PI-TP␣.
Mouse NIH3T3 fibroblast cells were co-transfected with both the pSG5-PI-TP␣ and pSV2-neo vectors or with only the pSV2neo vector (control) by a modified calcium phosphate precipitation technique (44). Stable clones were selected by using the antibiotic Geneticin G418. Several hundred positive neomycinresistant clones appeared after 2 weeks. From these clones, we selected three stable clones transfected with only the control pSV2-neo vector and 15 clones co-transfected with both the pSG5-PI-TP␣ and the pSV2-neo vector.
The stable clones transfected with the control vector (pSV2neo) are denoted as OPIx (control vector, clone x), and the clones co-transfected with both pSG5-PI-TP␣ and pSV2-neo are denoted as SPIx (sense PI-TP␣, clone x). The level of PI-TP␣ expression in these clones was estimated by immunoblotting of the cytosolic fractions of the cells. Two cell lines, SPI6 and SPI8, were selected because they express an increased level of PI-TP␣ as compared with OPI3 and wtNIH3T3 cells. Scanning of the immunostained PI-TP␣ bands indicated that the transfected cell lines SPI6 and SPI8 show a 2-and 3-fold increase in the PI-TP␣ level, respectively, as compared with the wtNIH3T3 and OPI3 cells (Fig. 1A).
Morphology,GrowthRate,DensitySaturation,andAnchoragedependent Growth of the Transfected Cell Lines-Clones that expressed increased amounts of PI-TP␣ displayed an altered morphology (Fig. 1B). These cells (panels 2 and 3) were somewhat smaller and rounder than wtNIH3T3 fibroblasts (panel 1). The stably transfected cell lines displayed an increased growth rate and a higher cell density at confluency (Table I).
The doubling time decreased from 21 h for wtNIH3T3 and OPI3 to 13-14 h for SPI6 and SPI8. In addition, the maximal cell density when the cultures are fully confluent increased from 0.20 ϫ 10 5 cells/cm 2 for wtNIH3T3 cells (or OPI3 cells, 0.16 ϫ 10 5 cells/cm 2 ) to 0.53 ϫ 10 5 cells/cm 2 for SPI6 and SPI8. Despite the difference in expression of PI-TP␣ between SPI6 and SPI8 cells, no significant difference in growth rate or saturation density was observed.
Analysis of the cellular DNA content in exponentially growing cell cultures by fluorescence-activated cell sorting showed that in wtNIH3T3 cells and in OPI3 cells, respectively, 38 and 35% of the cells were in the S phase/mitosis. For SPI6 cells and  SPI8 cells, this percentage was 43 and 46%, respectively (Table  I). From these values, and from the duration of the full cell cycle, it was calculated that the G 1 phase is significantly shorter in SPI6 cells and SPI8 cells (7-8 h) than the G 1 phase in wtNIH3T3 cells or in OPI3 cells (13 h, Table I). Confluent monolayers of wtNIH3T3 and SPI6/SPI8 cells were different. At full confluency, wtNIH3T3 cultures shed dead cells, while SPI6/SPI8 monolayers started to curl up from the edges of the culture dish, forming a "solid" piece of tissue after some time, indicating a different interaction between the cells.
To investigate whether increased expression of PI-TP␣ led to anchorage-independent growth of the cells, the capacity of the cells to grow in soft agar was investigated. However, none of the cell lines were able to form colonies in soft agar. Therefore, increased expression of PI-TP␣ does not lead to a loss of contact inhibition or to transformation.
Incorporation of myo-[ 3 H]Inositol-In order to investigate whether overexpression of PI-TP␣ affects the metabolism of the inositol phospholipids, the SPI8/SPI6 cells and wtNIH3T3 cells were labeled with myo-[ 3 H]inositol, and the relative incorporation in the inositol derivatives was determined as described in method 1. After equilibrium labeling of the cells (experimentally established by comparing various periods of labeling) the 3 H label in the total IP n fraction was determined (Fig. 2). In wtNIH3T3 cells, 4.4% of the myo-[ 3 H]inositol was incorporated in the IP n fraction. The addition of LiCl had no significant effect on the incorporation. However, incubation with bombesin, in the presence of LiCl, led to a 2-3-fold increase in the incorporation of myo-[ 3 H]inositol in the IP n fraction (Fig. 2). In the SPI6 and SPI8 cells, the percentage of incorporation in the IP n fraction was 2-3-fold higher as compared with wtNIH3T3 cells. Incubation with LiCl or with bombesin (in the presence of LiCl) did not further increase the level of incorporation. This indicates that the stimulation of the PLC-mediated degradation of PtdIns(4,5)P 2 is impaired in the cells overexpressing PI-TP␣.
TLC analysis of the inositol phospholipid fraction showed that the relative incorporation of myo-[ 3 H]inositol in PtdIns, PtdIns(4)P or PtdIns(4,5)P 2 was similar for wtNIH3T3, SPI6, and SPI8 cells (data not shown).
Analysis of Inositol Metabolites in wtNIH3T3 and SPI8 Cells-Since the water-soluble IP n fraction consists of a great number of inositol-containing metabolites, it is very possible that the composition of the IP n fraction from the SPI6 and SPI8 cells is different from that of the growth factor-stimulated wtNIH3T3 cells. This possibility was investigated by analysis of the water-soluble inositol metabolites by a high pressure liquid chromatograph connected to an on-line scintillation counter. The labeling and extraction procedures were adapted so as to increase the incorporation of myo-[ 3 H]inositol into the metabolites and to optimize the recovery of lyso-PtdIns, which tends to disappear from the organic phase by frequent wash steps. In agreement with Fig. 2, the initial analyses of metabolites from SPI6 and SPI8 cells gave comparable results. We therefore restricted the detailed analysis to the wtNIH3T3 and SPI8 cells.
As shown in Table II, the total incorporation of myo-[ 3 H] inositol in SPI8 cells was about 70% of that observed in wtNIH3T3 cells, whereas the total amount of protein per well, reflecting the number of cells, was twice as high. Despite this lower total incorporation, the absolute amount of label in the water-soluble inositol phosphates from the SPI8 cells was about twice as high, while the absolute amount of label in the inositol phospholipids was similar to that in the wtNIH3T3 cells.
The relative incorporation of myo-[ 3 H]inositol in the inositol phosphate and inositol phospholipid derivatives is shown in Table III. The analysis of the water-soluble inositol phosphates indicates that in SPI8 cells the levels of Ins(1)P and Ins(2)P are significantly (p Ͻ 0.05) increased. There was also a 2-fold increase in the level of GroPIns and a small change in the level of Ins(4)P. Low levels of labeled Ins(1,4)P 2 and Ins(1,4,5)P 3 were also detected, showing no significant difference between wtNIH3T3 and SPI8 cells. Analysis of the inositol phospholipids showed that in SPI8 cells the relative incorporation of myo-[ 3 H]inositol in lyso-PtdIns was clearly increased (p ϭ 0.06). In these cells, the incorporation in PtdIns(4)P was significantly decreased; no changes were observed in the relative labeling of PtdIns and PtdIns(4,5)P 2 .
As shown in Fig. 2, the stimulation of the overexpressers with bombesin (10 nM) did not result in an increased incorporation of myo-[ 3 H]inositol in the total IP n fraction, in contrast to what was observed with the wild type cells. In order to investigate whether the PLC-mediated degradation of PtdIns(4,5)P 2 is operative in the overexpressers, SPI8 and wild type cells were stimulated with bombesin (10 nM) or with PDGF (20 ng/ml). Analysis of the inositol phosphate fractions showed that stimulation of the wtNIH3T3 cells mainly resulted in an increased incorporation of myo-[ 3 H]inositol in Ins(1,4)P 2 and Ins(4)P; there was no significant effect on Ins(1)P and Ins(2)P (Table IV). However, no effect was seen on the level of Ins(4)P or Ins(1,4)P 2 in SPI8 cells, indicating the loss of growth factorstimulated PtdIns(4,5)P 2 degradation in these cells.
The Effect of PI-TP␣ on the Formation of Lyso-PtdIns in Vitro-The increased level of lyso-PtdIns in the SPI8 cells suggests the activation of a PLA 1 and/or PLA 2 . In order to investigate whether PI-TP␣ was able to stimulate the formation of lyso-PtdIns in vitro, a homogenate of myo-[ 3 H]inositollabeled wtNIH3T3 cells was incubated with different amounts of this protein in the absence and presence of Ca 2ϩ (5 mM). Each incubation contained about 0.1 g of endogenous PI-TP␣. As shown in Fig. 3, incubation with 0.5 g of PI-TP␣ in the absence of Ca 2ϩ led to a 2-fold increase in the level of lyso-PtdIns. The formation of lyso-PtdIns was further enhanced at a higher PI-TP␣ concentration (2.0 g). In the presence of Ca 2ϩ , the stimulatory effect of PI-TP␣ was more pronounced. The increased levels of lyso-PtdIns were accompanied by a significant decrease in the level of PtdIns (Fig. 3). Under the assay conditions, there was no change in the absolute levels of reflected the level observed in intact cells (compare Fig. 3 and Table III). These results indicate that wtNIH3T3 cells contain a PLA 1 /PLA 2 activity acting on PtdIns that can be activated by PI-TP␣ in a Ca 2ϩ -sensitive fashion.

DISCUSSION
The cellular function of PI-TP␣ has been extensively investigated using permeabilized cells and cell-free systems (33,36,38,46). From these studies it was inferred that PI-TP␣ is involved in the synthesis of PtdIns(4,5)P 2 , possibly by delivering PtdIns to PtdIns 4-kinase. A more direct approach is to study the PtdIns metabolism in cell lines in which the expression level of PI-TP␣ is changed. In the present study we have established stable mouse NIH3T3 fibroblast cell lines that express a 2-3-fold increase in the levels of PI-TP␣. These cells were chosen for the transfection experiments because they have a well defined PtdIns metabolism, several well known growth factor receptors, and a significant level of endogenous PI-TP␣.
As shown in Table I, enhanced levels of PI-TP␣ lead to a dramatic increase in the growth rate of the cells. The cause of the increased growth rate can be manyfold. However, since PI-TP␣ is involved, we have investigated the production of PtdIns metabolites. Two well known PtdIns-derived mitogenic signals are Ins(1,4,5)P 3 and 1,2-diacylglycerol, which are formed when a PtdIns(4,5)P 2 -specific PLC is activated by binding of growth factors to their receptors (18,19). Another PtdIns derivative with mitogenic activity is lyso-PtdIns that is produced by the PLA 2 -dependent pathway (24,25,29,30). In the present study, we show that the incorporation of myo-[ 3 H]inositol in the water-soluble inositol phosphate fraction was increased in the SPI8 cells when compared with wtNIH3T3 cells. An increase of 3 H-labeled inositol phosphates was also observed in permeabilized PC12 cells upon the addition of purified PI-TP␣ (33,34,37). However, in a number of these studies the exact composition of the IP n fraction was not established; an increased "IP n " fraction may include Ins(1)P and Ins(2)P as well as glycerophosphoinositol (products of PLA/lyso-PLA activation) and does not necessarily indicate that levels of Ins(4)P (and hence PLC activity) have increased. Hence, detailed analysis of the inositol phosphate fraction indicated that in SPI8 cells the levels of Ins(1)P and Ins(2)P were significantly increased, whereas the levels of Ins(4)P, Ins(1,4)P 2 , and Ins(1,4,5)P 3 were similar to that in wtNIH3T3 cells. This indicates that overexpression of PI-TP␣ in intact wtNIH3T3 cells has no effect on PtdIns(4,5)P 2 -specific PLC. Rather, the identified inositol phosphate derivatives are characteristic for the degradation of PtdIns by PLA 2 (24,25,27). In the latter studies, it was shown that the lyso-PtdIns that is produced upon activation of PLA 2 can be degraded by a lysophospholipase to GroPIns. Alternatively, lyso-PtdIns can be degraded by a PLC to Ins(1:2 cyc)P, which, due to the acidic extraction conditions used, may be converted into Ins(1)P and Ins(2)P (23,47). The activation of a potentially PtdIns-specific PLA in SPI8 cells was confirmed by the analysis of the inositol phospholipid fraction, showing that the level of lyso-PtdIns was 2-3-fold increased as   2؉ . The 3 H-labeled phosphoinositides were separated by TLC, and the distribution of 3 H label was determined. The percentages of 3 H label in PtdIns (E, q) and in lyso-PtdIns (Ⅺ, f) are presented. These data are representative for the results of three independent experiments. compared with control cells. In line with the enhanced level of lyso-PtdIns, the level of GroPIns was also increased in SPI8 cells. As for the other inositol phospholipids in SPI8 cells, the relative incorporation of myo-[ 3 H]inositol was significantly decreased in PtdIns(4)P, while there were no changes in PtdIns and PtdIns(4,5)P 2 . Furthermore, the addition of purified PI-TP␣ to a crude lysate of [ 3 H]inositol-labeled wtNIH3T3 cells induced a 2-3-fold increase in the level of lyso-PtdIns most probably derived from PtdIns, since the level of 3 H-labeled PtdIns decreased. No significant change was observed in the relative labeling of PtdIns(4)P and PtdIns(4,5)P 2 . These data strongly suggest that in cells with an increased expression of PI-TP␣, a potentially PtdIns-specific PLA is constitutively activated. Based on former studies, the activation of a PLA 2 is most likely (25,26). However, activation of a PtdIns-specific PLA 1 cannot as yet be excluded. The procedures generally used to extract inositol phospholipids may lead to a loss of the rather water-soluble lyso-PtdIns. This may explain why the PLAmediated signal transduction pathway has not been detected in the semi-intact cells or in isolated membrane systems upon the addition of PI-TP (33)(34)(35)(36)(37). Furthermore, and as indicated above, the detailed analysis of inositol phosphates is required to ascertain whether PLC and/or PLA is activated.

FIG. 3. The formation of lyso-PtdIns in lysates of myo-[ 3 H]inositol-labeled wtNIH3T3 cells incubated with PI-TP␣ in the absence (open symbols) and presence (closed symbols) of 5 mM Ca
Lyso-PtdIns has been shown to be a signaling molecule itself and could therefore be responsible for the increased growth rate of SPI6 and SPI8 cells. The mechanism by which this molecule acts is not yet fully clear. It has been proposed that lyso-PtdIns can either be released and act by binding to a membrane receptor analogous to the reported membrane receptor for lysophosphatidic acid or act intracellularly by interacting with target proteins (29,30,47,48). Furthermore, lyso-PLA activity on lyso-PtdIns can produce GroPIns, which can itself be phosphorylated to glycerophosphoinositol 4-phosphate, which has been reported to be a novel intracellular messenger of the Ras pathway (28). The increased levels of Ins(1)P and Ins(2)P in extracts of SPI8 cells may represent an increased level of Ins(1:2 cyc)P in intact cells. Increased levels of Ins(1:2 cyc)P have been correlated with a decreased level or activity of the enzyme Ins-1:2-cyc-2 phosphohydrolase (49). An increase in the level of Ins(1:2 cyc)P was thought to be correlated with the loss of contact inhibition (50) and, therefore, could be the reason for the higher cell density observed at confluency of SPI6 and SPI8 cells. Whether the level or activity of Ins(1:2 cyc)P 2-phosphohydrolase actually is changed in SPI6 and SPI8 cells will a the subject of future investigations. Upon activation of PLA 2 , arachidonic acid may also be released, a precursor of the eicosanoids, which have been shown to participate in cell regulation, such as control of mitogenesis (25,51).
The growth factors bombesin and PDGF were not able to activate the PLC-mediated degradation of PtdIns(4,5)P 2 in SPI6 and SPI8 cells. While in wtNIH3T3 cells, the Ins(4)P production is increased 4 -10-fold by bombesin or PDGF, incubation of both SPI6 and SPI8 cells with these growth factors had no effect on either the production of any inositol phosphate or any of the phosphoinositides. The desensitization of the PLC-dependent pathway could be explained by the SPI6 and SPI8 cells expressing either no receptors or impaired receptors for bombesin or PDGF. On the other hand, it has been shown that an increased level of lyso-PtdIns inhibits in vitro a GTPase-activating protein (29). Furthermore, the antibiotic neomycin that is used to select successfully transfected cells could also interfere with the PLC-mediated signal transduction pathway. However, NIH3T3 cells that were transfected with cDNA encoding PLC␥1 and selected by neomycin resistance were fully able to respond to PDGF stimulation with increased IP n production (52). 2 The activation of PLA 2 and the simultaneous desensitization of PLC as observed in the SPI6 and SPI8 cells has also been described for cells transfected with ras or other cytoplasmic (mos, raf) or membrane-associated (src, met, trk) oncogenes but not with nuclear (myc, fos) oncogenes (24). This may suggest that the intracellular mechanisms of action of PI-TP␣ and the above oncogenes have certain steps in common. It is also possible that PI-TP␣ is part of the mechanism used by the oncogene proteins to activate PLA 2 . On the other hand, activation of PLA 2 has also been observed during normal differentiation of neonatal liver cells (53), indicating that increased levels of GroPIns, lyso-PtdIns, and Ins(1)P could also be associated with different stages of differentiation rather than being characteristic of the malignant transformation process.