Phosphatidylinositol transfer protein (cid:2) regulates growth and apoptosis of NIH3T3 cells: involvement of a cannabinoid 1-like receptor

Mouse ﬁbroblast cells overexpressing phosphatidylinositol transfer protein (cid:2) [PI-TP (cid:2) ; sense PI-TP (cid:2) (SPI (cid:2) ) cells] show a signiﬁcantly increased rate of proliferation and an extreme resistance toward ultraviolet- or tumor necrosis factor- (cid:2) -induced apoptosis. The conditioned medium (CM) from SPI (cid:2) cells or the neutral lipid extract from CM stimulated the proliferation of quiescent wild-type NIH3T3 cells. CM was also highly effective in increasing resistance toward induced apoptosis in both wild-type cells and the highly apoptosis-sensitive SPI (cid:3) cells (i.e., wild-type cells overexpressing PI-TP (cid:3) ). CM from SPI (cid:2) cells grown in the presence of NS398, a speciﬁc cyclooxygenase-2 (COX-2) inhibitor, expressed a diminished mitogenic and antiapoptotic activity. This strongly suggests that at least one of the bioactive factor(s) is an eicosanoid. In accordance, SPI cells express enhanced levels of COX-1 and COX-2. The antiapoptotic activity of CM from SPI (cid:2) cells tested on SPI (cid:3) cells was inhibited by (cid:2) 50% by pertussis toxin and suramin as well as by SR141716A, a speciﬁc antagonist of the cannabinoid 1 receptor. These inhibitors had virtually no effect on the COX-2-independent antiapoptotic activity of CM from SPI (cid:2) cells. The latter results imply that PI-TP (cid:2) me-diates the production of a COX-2-dependent eicosanoid that activates a G-protein-coupled receptor, most probably a cannabinoid 1-like receptor. Phosphatidylinositol transfer protein (cid:2) regulates growth and apoptosis of NIH3T3 cells: involvement of a cannabinoid 1-like receptor.

Phosphatidylinositol transfer proteins (PI-TPs) belong to a family of highly conserved proteins that in vitro can catalyze the transfer of phosphatidylinositol, phosphatidylcholine, and sphingomyelin between membranes (1,2). In mammalian tissues, at least two isoforms are identified: PI-TP ␣ , which is localized in the nucleus and cytosol, and PI-TP ␤ , which is associated with the Golgi system (3). Possible cellular functions of these proteins have been obtained from experiments with permeabilized cells (4), reconstituted Golgi membrane systems (5), and cells in which the expression of the proteins has been altered (6)(7)(8). In permeabilized, cytosol-depleted cells, both isoforms restored guanosine 5 Ј -[ ␥Ϫ thio]triphosphate-stimulated protein secretion as well as phospholipase C-mediated inositol lipid signaling (9)(10)(11). On the other hand, intact mouse fibroblast cells with increased expression of PI-TP ␣ [sense PI-TP ␣ (SPI ␣ ) cells] showed an enhanced phospholipase A 2 (PLA 2 )-mediated degradation of PI that was not observed in cells with an increased expression of PI-TP ␤ (SPI ␤ cells) (7,8). Convincing evidence for distinct cellular functions was obtained by genetic approaches. Murine embryonic stem cells deficient in PI-TP ␤ fail to develop, whereas the embryonic development of cells deficient in PI-TP ␣ proceeds normally (12,13). In the latter case, the mice die within 2 weeks of birth.
The rate of proliferation of SPI ␣ cells is significantly increased, with a cell cycle duration of 13 h compared with 21 h for wild-type cells, indicating that PI-TP ␣ may be involved in the production of a mitogenic factor (7). The observation that in SPI ␣ cells a PI-specific PLA 2 is activated implies that, in addition to lysoPI, a significant amount of arachidonic acid is produced because PI is highly enriched in this fatty acid (14). Arachidonic acid is the main precursor in the synthesis of eicosanoids, including prostaglandins, leukotrienes, thromboxanes, and prostacyclins. These arachidonic acid metabolites play important roles in many cellular processes, such as thrombosis (15), inflammation (16,17), cell growth (18), and apoptosis (19). Eicosanoids are synthesized in response to external stimuli in which the release of arachidonic acid from phospholipids by PLA 2 is the rate-limiting step (20). In particular, the prostaglandins, the synthesis of which depends on cyclooxygenase-1 (COX-1) and COX-2, play key roles in cell growth and cell survival as well as in processes such as carcinogenesis and inflammation (21). In general, COX-1 is expressed constitutively, whereas COX-2 can be induced by various physiological stimuli (22)(23)(24). Furthermore, it has been suggested that an increase in arachidonic acid metabolism as mediated by COX-1 or COX-2 depends on proteins that coordinate the release of arachidonic acid from phospholipids (25).
In the present study, we show that PI-TP ␣ is involved in the regulation of proliferation and of apoptosis sensitivity. We present evidence that this regulation occurs through the COX-2-dependent production and secretion of eicosanoid factor(s). One of these factors most likely acts on the G-protein-coupled cannabinoid 1 receptor, thereby displaying both autocrine and paracrine activity.

Preparation of conditioned medium
Cell cultures (75 or 150 cm 2 dishes) were grown to 80-90% confluency. The medium was replaced by 5 or 10 ml of DMEM/ bicarbonate (Bic) containing 0.1% BSA (DBB). This medium was left on the cells for 24 h. After removal, the medium was centrifuged (5 min at 1,000 rpm) to remove floating cells. The supernatant is the conditioned medium (CM). Under standard conditions, quiescent cells were incubated with CM that was derived from an identical surface of cells (i.e., each 9.5 cm 2 well of a six-well dish was incubated with the amount of CM that was conditioned for 24 h by 9.6 cm 2 of cells).

Cell culture and growth assays
All cells were cultured in DMEM containing 10% newborn calf serum (NCS) and buffered with NaHCO 3 (44 mM) in a 7.5% CO 2 -humidified atmosphere at 37 Њ C. NIH3T3 mouse fibroblast cells overexpressing PI-TP ␣ (SPI ␣ cells) and PI-TP ␤ (SPI ␤ cells) were made as described previously (7,8). In this study, we have used two different SPI ␣ cell lines (SPI2 and SPI8), giving essentially identical results.
To determine growth rate, cells (1 ϫ 10 4 per well) were seeded on 24-well plates. After 24 h, the medium was replaced by 1 ml of DMEM/Bic containing 0.1% NCS. After 24-48 h, the medium was replaced by 250 l of CM or 250 l of DMEM/Bic/ 0.1% NCS supplied with extracts of the medium. The volume of medium that had been conditioned by 2 cm 2 of cells was added to each well of a 24-well (2 cm 2 each) plate. After 8 h, [ 3 H]thymidine (0.5 Ci/well in 50 l of DMEM/Bic/0.1% NCS) was added to the cells. After 16 h (overnight), the medium was removed, the cells were washed four times with phosphate-buffered saline, and 1 ml of methanol per well was added. The methanol was removed after 20 min, and the cells were left to dry on air. A total of 0.5 ml of 0.1 N NaOH was added, and the cells were incubated for 30 min at 37 Њ C and scraped of. The cells were mixed with 4.5 ml of scintillation liquid and counted. Controls were cells incubated with 250 l of DMEM/Bic/0.1% NCS. Maximal stimulation was obtained by the addition of DMEM/Bic containing 10% NCS.

Ultraviolet radiation of cell cultures
For ultraviolet (UV) treatment, cells were grown on a 6-or 12well dish. Before UV treatment, the cells were incubated for 4 h in DBB. All of the compounds that were tested for activity were added to this DBB during the incubation. The medium was removed and UV treatment was performed in a Stratalinker (Stratagene) with the indicated dose (standard, 200 J/m 2 ). Cells were incubated overnight (19 h) with pertussis toxin (300 ng/ml) in DMEM containing 10% NCS followed by incubation of pertussis toxin in CM from SPI ␣ cells for 4 h; suramin and the cannabinoid 1 and 2 receptor antagonists SR141716A (1 M) and SR144538 (1 M) were incubated for 4 h in CM from SPI ␣ cells.
After treatment with UV light, 1.5 ml of DBB was added and cells were incubated at 37 Њ C. At the indicated times, cell death was morphologically scored as the percentage of cells that are in the process of blebbing, which have a condensed nucleus. Also, for visualization of condensed nuclei, the cells were grown on glass cover slips and similarly treated. The cells were fixed in cold methanol ( Ϫ 20 Њ C) and subsequently incubated with 1 g/ml DAPI in methanol for 5 min at room temperature. Cells were washed once with methanol and once with PBS and mounted in Mowiol (Hoechst, Frankfurt, Germany) supplemented with 0.1% paraphenylene diamine. Fluorescent DNA-DAPI complexes were visualized with a Leica inverted microscope.

Flow cytometric analysis of DNA fragmentation
Cells were grown in 9 cm 2 dishes to 80% density. Cells were treated as described above (preincubation with CM for 4 h), UV treatment was performed, and 1.5 ml of DBB was added to the cultures. Cells were kept at 37 Њ C. After 2 h, the medium was collected and cultures were washed with 1 ml of PBS. This PBS was combined with the medium and centrifuged to collect floating cells. Cell cultures were incubated with 1 ml of 8 mM EGTA in PBS for 5 min, resuspended, and centrifuged for 5 min at 800 rpm at 4 Њ C. Both cell pellets were combined and resuspended in 100 l of PBS supplemented with 0.5% polyvinyl alcohol (PVA), vortexed, and kept on ice for 30 min. A total of 900 l of 70% ethanol in PBS (0 Њ C) was layered carefully on top of the PBS-PVA. The sample was mixed and stored at Ϫ 20 Њ C.
Before flow cytometric analysis, nuclei were extracted and stained to detect nuclear fragmentation as a sign of apoptosis [the methodology with slight modifications is described by Darzynkiewicz and Li (26)]. Briefly, the cell suspension was centrifuged for 3 min at 2,200 rpm at 4 Њ C. The supernatant was removed, and the pellet was resuspended in 1 ml of PBS and again centrifuged for 3 min at 2,200 rpm. The pellet was resuspended in 1 ml of a mixture of 50% PBS and 50% extraction buffer (0.2 M by guest, on July 20, 2018 www.jlr.org Downloaded from Na 2 HPO 4 , 4 mM citric acid, pH 7.8). The suspension was kept for 10 min at room temperature. Nuclei were collected by centrifugation (3 min at 2,200 rpm), the pellet was resuspended in 1 ml of PBS containing propidium iodide (50 g/ml) and RNase (50 g/ml), and the suspension was incubated in the dark for at least 30 min. Before flow cytometric analysis, the mixture was filtered through a 70 m cell strainer (Falcon) to remove clustered nuclei and other debris. Flow cytometric analysis of the extracted and stained nuclei (and fragments thereof) was performed on a FACScan flow cytometer equipped with a 100 mW argon laser exciting at 488 nm (Becton Dickinson, San Mateo, CA). The propidium iodide fluorescence of nonaggregated nuclear events was detected in fluorescence detector FL-3 (630 nm long-pass emission detector) in linear mode. Because it is likely that an apoptizing nucleus will fall apart into more than one event in the subdiploid area, the term apoptotic events will be used rather than apoptotic nuclei. This technique renders a relatively low scoring of apoptotic cells compared with visual scoring, as nuclear apoptotic events are still at an early stage at 2 h after UV irradiation.

Determination of COX-1 and COX-2
Cells were grown in 21 cm 2 dishes to 80-90% density. Cells were washed twice with PBS and CM was added. Cells were harvested after 5 h. To harvest the cells, the medium was removed, cells were washed twice with PBS, and the dishes were frozen. After thawing, the cells were incubated with 150 l of buffer containing 0.1% Nonidet P40 in 20 mM Tris (pH 7.2) for 5 min at room temperature, scraped off, and put on ice. The cell lysate was centrifuged for 10 min at 14,000 rpm at 4 Њ C, and the supernatant was used to determine protein content using the Bradford assay (27). Equal amounts of protein of all samples were prepared for gel electrophoresis. A total of 20 g of supernatant protein was subjected to SDS-PAGE on a 12.5% gel and analyzed by Western blotting using antibodies against COX-1 and COX-2 (Cayman). Quantification of bands on film was performed by scanning with a Bio-Rad GS 700 imaging densitometer equipped with an integrating program.

Extraction of [ 14 C]arachidonic acid-labeled metabolites from CM
Cells were grown to 60% confluency. The cells were labeled for 24 h with [ 14 C]arachidonic acid (0.1 Ci per well on a sixwell plate in 1 ml of DMEM/Bic containing 10% NCS). To analyze labeled arachidonic acid metabolites in CM, the label medium was removed after 24 h and replaced by DBB. After 24 h, this medium was collected and centrifuged to remove floating cells. Arachidonic acid metabolites in the CM were extracted and separated as described by Tai, Tai, and Hollander (28). Shortly, 0.03 ml of 12 M formic acid was added per milliliter of CM. The mixture was extracted with two 3 ml portions of ethyl acetate. The combined extracts were evaporated under N 2 . The residue was taken up in acetone and quantitatively spotted on a Silica Gel 60 TLC plate. The TLC plate was developed in the organic phase of ethyl acetate-acetic acid-isooctane-water (11:2:5:10, v/v) as the solvent system. Radioactivity was monitored by scanning the plate with a Berthold Tracemaster 20 Automatic TLC-Linear analyzer.

Mitogenic activity in the CM from SPI␣ and wild-type NIH3T3 cells
Previously, we had shown that SPI␣ cells have a cell cycle duration of 13 h compared with 21 h for wild-type NIH3T3 (wtNIH3T3) cells (7). In view of this highly increased rate of proliferation, CM from SPI␣ cells were tested for the production of a mitogenic factor by measuring the incorporation of [ 3 H]thymidine into DNA of serum-starved (24-48 h) wtNIH3T3 cells. As shown in Fig.  1A, the CM from SPI␣ cells increased DNA synthesis by a factor of 1.6. For comparison, CM from wtNIH3T3 cells increased the DNA synthesis by a factor of 1.1. This indicates that SPI␣ cells produce a mitogenic factor that is secreted into the medium.
Given the enhanced degradation of PI in SPI␣ cells as reflected in the relatively high levels of lysoPI and inositol phosphates (7), it is most likely that arachidonic acid also is released, which subsequently can be converted by COXs and lipoxygenases into metabolites to be secreted into the medium (29). To isolate prostaglandins, leukotrienes, and other metabolites from CM, we used a neutral extraction procedure (28). As shown in Fig. 1B, the lipid extract of CM from SPI␣ cells increased DNA synthesis in quiescent wtNIH3T3 cells by 3.5-fold, whereas the extracts of CM from wtNIH3T3 had only a slight effect. By comparing Fig. 1A, B, it appears that the mitogenic activity in the neutral lipid extracted is enhanced relative to that of CM. When the medium was conditioned in the continuous presence of 10 M indomethacin, a nonselective inhibitor of COX-1 and COX-2, the CM was no longer mitogenic for quiescent wtNIH3T3 cells (Fig. 1C). In the presence of 50 M NS398, a selective inhibitor of COX-2, the ensuing CM displayed a similar lack of mitogenic activity, indicating that the mitogenic factor(s) is dependent on COX-2 activity (data not shown).
The aqueous phase of the lipid extract was not active (data not shown), indicating that all mitogenic activity was present in the organic phase. Similarly, the neutral lipid extract from the postnuclear cell lysates was inactive, indicating that all mitogenic activity formed was released in the medium.

Cell survival upon UV irradiation
Because the rate of proliferation is often correlated with apoptosis sensitivity (30,31), we compared the apoptotic sensitivity of the wtNIH3T3 and SPI␣ cells upon UV irradiation. First, apoptosis was analyzed by quantifying the number of blebbing cells at 2 h after exposing the cells to variable doses of UV radiation ( Fig. 2A). Over a range from 20 to 400 J/m 2 , the SPI␣ cells were almost completely resistant toward apoptosis, compared with wtNIH3T3 cells being up to 40% apoptotic. Apoptosis as a function of time upon a UV dose of 200 J/m 2 confirmed that SPI␣ cells are highly resistant toward UV-induced apoptosis (Fig. 2B). Under the same conditions, the apoptosis sensitivity of cells overexpressing PI-TP␤ (SPI␤) was also tested. Compared with the wtNIH3T3 cells, the SPI␤ cells were found to be even more sensitive toward UVinduced apoptosis. This indicates that SPI␣ cells have acquired a high resistance against apoptosis.
Factors that stimulate cell growth often also demonstrate antiapoptotic activity (32,33). To investigate the relationship between the mitogenic factor and apoptosis sensitivity, we investigated whether CM from SPI␣ cells is able to protect wtNIH3T3 cells against UV-induced apoptosis. Upon incubation for 4 h with CM from SPI␣ cells, the wtNIH3T3 cells were fully protected (Fig. 2C). Similar protection was observed when the wild-type cells were incubated with the neutral lipid extract of CM from SPI␣ cells. The CM from SPI␣ cells is also very effective in protecting the highly sensitive SPI␤ cells (34) against UVinduced apoptosis (Fig. 2D). Given that indomethacin inhibited the mitogenic activity of CM from SPI␣ cells (Fig.  1C), we also determined the effect of indomethacin on the antiapoptotic activity. As shown in Fig. 2D, CM collected in the presence of indomethacin (10 M) was less effective in protecting SPI␤ cells against UV-induced apoptosis. However, it is clear that the inhibitory effect of indomethacin on the mitogenic activity of CM from SPI␣ cells is more pronounced than its effect on the antiapoptotic activity (cf. Figs. 1C and 2D). This suggests that either the factor affects cell survival at a much lower concentration than cell growth or that the SPI␣ cells produce an additional antiapoptotic factor independent of COX-1/ COX-2.
In addition to counting blebbing cells, we quantified the percentage of apoptotic cells by analyzing DNA fragmentation by flow cytometric analysis of nuclei that were labeled with propidium iodide (Fig. 3). Flow cytometric analysis carried out at 2 h after UV irradiation confirmed that the extent of apoptosis increased in the order SPI␣ Ͻ wtNIH3T3 Ͻ SPI␤. When SPI␤ cells were incubated for 4 h with CM from SPI␣ cells before UV treatment, the percentage of apoptotic events was significantly decreased, confirming the antiapoptotic activity of the SPI␣ medium. This protective effect was reduced using SPI␣ medium prepared in the presence of indomethacin.
The different sensitivity of NIH3T3 and SPI␣ cells toward UV-induced apoptosis was confirmed by analysis of nuclear condensation using DAPI staining. Before serum starvation, the nuclear DNA of all wtNIH3T3 and SPI␣ cells was intact (Fig. 4A, D). After sensitization by incubation for 4 h with DBB, the nuclei of some wtNIH3T3 cells showed DNA condensation, whereas the nuclei of SPI␣ cells were completely unaffected (Fig. 4B, E). This shows that in the case of wtNIH3T3 cells, serum deprivation by DBB is already an apoptotic signal. Two hours after UV radiation, the nuclear DNA of the wtNIH3T3 cells showed a significant increase in condensation, whereas the effect on the nuclear DNA of SPI␣ cells was still minimal (Fig.  4C, F).

The mitogenic factor is an arachidonic acid metabolite
Indomethacin and NS398 block the production of the mitogenic activity, indicating that the bioactive factor in CM from SPI␣ cells is an arachidonic acid metabolite (Fig.  1C). To obtain additional evidence, we prepared CM from SPI␣ cells labeled for 24 h with [ 14 C]arachidonic acid, extracted the medium with ethyl acetate, and separated the lipid extract by TLC. As shown in Fig. 5  and PGF 2␣ coeluted with two of the arachidonic metabolite peaks. By using an ELISA kit (Cayman) it was determined that SPI␣ cells produced five times more PGE 2 compared with wtNIH3T3 (0.5 and 0.1 g/ml, respectively). However, it was shown that both prostaglandins, either separate or combined, were unable to stimulate the growth of quiescent wtNIH3T3 cells in the concentration range present in CM from SPI␣ cells. Similar results were obtained by using PGE 1 , PGD 2 , PGA 2 , and PGF 1␣ (data not shown), indicating that the COX-2-dependent mitogenic factor must be an as yet unidentified eicasonoid.

COX-1 and COX-2 expression
Given the high production of a mitogenic factor(s) by SPI␣ cells, we investigated the expression of COX-1 and COX-2 in these cells by Western blotting. In SPI␣ cells, COX-1 and COX-2 are increased 4-and 7.5-fold, respectively, compared with wtNIH3T3 cells (Fig. 6). In addition, incubation with CM from SPI␣ cells for 5 h gives rise to a 3-fold increase of COX-2 in wtNIH3T3 cells and a 2-fold increase in SPI␤ cells, in agreement with the increased resistance of these cells toward UV-induced apoptosis (Fig.  2C, D). Under these conditions, CM from SPI␣ cells has little effect on COX-1.

Involvement of a G-protein-coupled receptor
To identify the mode of action of the antiapoptotic activity in CM from SPI␣ cells, inhibitors of G-protein-coupled receptors (GPCRs) were used. Incubation of SPI␤ cells with CM from SPI␣ cells in the presence of suramin (300 M) or pertussis toxin (300 ng/ml) inhibited the antiapoptotic effect by ‫%05-04ف‬ (Fig. 7). A specific group of eicosanoids, the endocannabinoids, has been shown to inhibit neurodegeneration in rat brain by activation of a GPCR, the cannabinoid 1 (CB1) receptor (35,36). Because PI-TP␣ appears to play an important role in maintaining neural integrity (37), we investigated whether there was a relationship between the paracrine stimulation of SPI␤ cells by CM from SPI␣ cells and the CB1 and CB2 receptors. Incubation with the CB1 receptor antagonist SR141716A (1 M) for 4 h reduced the antiapoptotic activity of CM from SPI␣ cells by 40% (Fig. 7). Under these conditions, the CB2 receptor antagonist SR144538 (1 M) appeared to have no effect. On the other hand, in contrast to pertussis toxin, suramin, and SR141716A, which had no effect on the apoptosis sensitivity of the SPI␤ cells under control conditions (i.e., DBB), the CB2 receptor antagonist SR144538 itself showed a distinct antiapoptotic activity (Fig. 7). Hence, we cannot exclude the possibility that the antiapoptotic factor also acts through the activation of the CB2 receptor. Moreover, addition of pertussis toxin, suramin, or SR141716A to CM from SPI␣ cells prepared in the presence of NS398, a specific COX-2 inhibitor, failed to result in a significant further decrease of antiapoptotic activity (data not shown). These results strongly suggest that at least one of the antiapoptotic factors produced by SPI␣ cells is a COX-2-dependent endocannabinoid, acting through the activation of a CB1like receptor.

DISCUSSION
In this paper, we show that NIH3T3 cells overexpressing PI-TP␣ (SPI␣ cells) are almost completely resistant toward UV-induced apoptosis compared with wtNIH3T3 cells. When CM from SPI␣ cells was added to wtNIH3T3 or SPI␤ cells, which are extremely sensitive to UV radiation, these cells also became apoptosis resistant. In addition, CM from SPI␣ cells as well as the neutral lipid extract derived from CM stimulates the growth of quiescent wtNIH3T3 cells. Antiapoptotic and mitogenic activity in the CM was significantly diminished when the SPI␣ cells were incubated with COX-2 inhibitors during preparation of the CM. In agreement with previous observations on SPI␣ cells being enriched in lysoPI (7), CM from SPI␣ cells is highly enriched in arachidonic acid metabolites (e.g., 0.5 g/ml PGE 2 compared with 0.1 g/ml for wtNIH3T3 cells). This enrichment reflects the constitutive activation of PLA 2 by PI-TP␣, resulting in the degradation of PI into lysoPI and arachidonic acid, which is subsequently converted into eicosanoids, among them PGE 2 and PGF 2␣ . When prepared in the presence of the COX-2 inhibitor NS398, some major arachidonic acid metabolites were absent from CM of SPI␣ cells, suggesting that these compounds constituted a major part of the mitogenic and antiapoptotic activity (Fig. 4).
PI-TP␣ is localized in the cytosol and in the nucleus (3,38). The relevance of the nuclear localization is not clear yet. However, the presence of an active PI metabolism in the nucleus is well documented (39). Furthermore, nuclei of several mammalian cells have been shown to contain an active acylation-deacylation cycle involving PLA 2 activity [reviewed in ref. (40)]. Additionally, because mammalian nuclei also contain a significant level of arachidonoylPI (41), it remains to be established whether the PI-TP␣mediated production of lysoPI and arachidonic acid occurs partially in the nucleus.
In many cell types, PGE 2 is the major COX-2-dependent product known to promote cell growth in an autocrine and paracrine manner (42). The mitogenic effects of PGE 2 are mediated by the activation of E-prostanoid receptors (43). When PGE 2 alone or together with PGF 2␣ was tested on quiescent wtNIH3T3 cells, we failed to observe the growth-promoting effect. However, PGE 2 and PGF 2␣ enhanced to some extent the survival of these cells after UV radiation. On the other hand, incubation of SPI␤ cells with these prostaglandins had no effect on the survival of the SPI␤ cells, which remained extremely sensitive (data not shown). This strongly suggests that SPI␣ cells produce another COX-2-dependent mitogenic eicasanoid, different from PGE 2 or PGF 2␣ , the identity of which remains to be established. Similarly, we assume that these metabolites are also active in protecting cells against induced apoptosis.
It is well established that the induction of COX-2 prevents apoptosis by generating antiapoptotic prostaglandins as well as by removing the proapoptotic substrate Fig. 6. Western blot analysis of cyclooxygenase-1 (COX-1; left ) and COX-2 (right) in wtNIH3T3, SPI␣, and SPI␤ cells and in wtNIH3T3 and SPI␤ cells incubated for 4 h with DBB or CM from SPI␣ cells (SPI␣ CM). Cells were lysed, and the cytosolic fractions (aliquots of 25 g of protein) were analyzed by SDS-PAGE and Western blotting using specific anti-COX-1 and anti-COX-2 antibodies as described in Materials and Methods. To ensure that identical amounts of protein were analyzed, gels and blots were routinely checked by Ponceau S/Coomassie Brilliant Blue staining. arachidonic acid (42,44,45). In the case of the SPI␣ cells, the arachidonic acid released by PI-TP␣ from PI is effectively removed by COX-2, the level of which is highly increased compared with wtNIH3T3 and SPI␤ cells (Fig. 6). Incubation of the SPI␤ cells with CM from SPI␣ cells increased the level of COX-2 but not of COX-1. This could explain why wtNIH3T3 and SPI␤ cells are protected against apoptosis by CM from SPI␣ cells. On the other hand, CM from SPI␣ cells prepared in the presence of the COX-1/COX-2 inhibitor indomethacin still showed considerable antiapoptotic activity (Figs. 2D, 3). This indicates that either the factor affects cell survival at a much lower concentration than cell proliferation or that, in addition to producing COX-dependent factors, SPI␣ cells also produce COX-independent antiapoptotic factors.
Further insight into the nature of the factor was provided by the inhibitory effect of both GPCR inhibitors and the CB1 and CB2 receptor antagonists on the antiapoptotic activity of CM from SPI␣ cells (46,47). The 40-50% reduction in antiapoptotic activity induced by pertussis toxin, suramin, and the specific CB1 receptor antagonist SR141716A strongly suggests that the antiapoptotic factor produced by SPI␣ cells is an endocannabinoid, which acts via activation of the CB1 receptor (Fig. 7). CM from SPI␣ cells prepared in the absence or presence of COX-2 inhibitors (indomethacin or NS398) in combination with the GPCR/CB1 inhibitors showed similar antiapoptotic activity (i.e., 60% of the control). This indicates that the activation of the CB1 receptor is solely attributable to the COX-2-dependent cannabinoid (48,49). On the other hand, to date there is no direct evidence for the presence of a CB1 receptor in mouse fibroblasts. Hence, we cannot exclude the possibility that the antagonist SR141716A acts on a CB1-like receptor.
These findings demonstrate a possible link between the production of endocannabinoids and PI-TP␣ activity. Decreased expression of PI-TP␣ in rat brain is known to cause serious neurodegeneration (37), whereas the endocannabinoid anandamide has been shown to act as an endogenous protective factor of the brain against acute neuronal damage (35). Our results provide evidence that the cell survival (e.g., the prevention of neurodegeneration) by PI-TP␣ is linked to the PLA 2 -dependent release of arachidonic acid, part of which is subsequently converted into bioactive eicosanoids by COX-2. The possible mechanism by which cell survival depends on PI-TP␣ is presented in a model (Fig. 8).
Recently, mice lacking PI-TP␣ were shown to die within 2 weeks after birth as a result of massive physiological defects, including spinocerebellar degeneration, intestinal and hepatic steatosis, and hypoglycemia (13). Because PI-TP␣ Ϫ/Ϫ mice develop to term and are phenotypically normal, it appears that PI-TP␣ is not required for embryonic development. However, immediately after birth, these mutant mice failed to thrive, demonstrating among other defects increased apoptosis throughout the cerebellum. The importance of PI-TP␣ for normal brain function is also evident from the vibrator mouse, which has severe neurological disorders attributable to reduced PI-TP␣ levels in the brain (37). Given that PI-TP transfer activity is particularly Fig. 8. The regulatory role of phosphatidylinositol transfer protein (PI-TP␣) in the production of a bioactive eicosanoid. In the model presented, PI-TP␣ activates a PI-specific phospholipase A 2 (PLA 2 ) leading to an increased release of arachidonic acid, which is subsequently converted into eicosanoids by COX-2. Part of these eicosanoids constitute the survival factor(s), which has antiapoptotic activity by acting through the activation of a GPCR, possibly CB1-like receptor. PLC, phospholipase C; PI-3-K, phosphatidylinositol-3-kinase.
by guest, on July 20, 2018 www.jlr.org Downloaded from high in synaptosome and myelin fractions from rat brain as a result of high levels of PI-TP␣ (50), the neuronal cells of mice may be well protected against apoptosis by PI-TP␣ mediating the synthesis of the antiapoptotic factor. Along this line, it will be interesting to investigate whether the early death of PI-TP␣ Ϫ/Ϫ mice after birth can be prevented by the administration (either in food or intravenously) of the eicosanoid factor produced by SPI␣ cells.