Accumulation of Lysophosphatidylinositol in RAW 264.7 Macrophage Tumor Cells Stimulated by Lipid A Precursors*

~,03-Diacylglucosamine 1-phosphate (lipid X), a monosaccharide precursor of Escherichia coli lipid A, was used to stimulate RAW 264.7 macrophage tumor cells, and the effects on macrophage phospholipid me- tabolism were examined. The addition of E. coli lipid X to the medium of cells that had been uniformly la- beled with 32Pi resulted in a 4-&fold increase in the level of lysophosphatidylinositol. This effect was max- imal at 5 PM lipid X. Lysophosphatidylinositol levels reached a maximum 45 min after stimulation, followed by a gradual decline to near normal levels within 2 h. The formation of lysophosphatidylinositol was depend- ent upon extracellular calcium and was almost completely inhibited when cycloheximide was added at the time of stimulation. The addition of the disaccharide lipid A precursor IVA, commercial lipopolysaccharide phorboll2-myristate 13-acetate (lo” or calcium ionophore A23187 M) to these cells resulted in a similar increase in lysophosphatidylinositol levels, but phosEhatidic acid was inactive. The stimulation by IVA and phorbol myristate acetate was blocked by cycloheximide, but the @timulation by lipopolysaccharide Ftaetz, unpublished observations). This decomposition would contribute 256 cpm to the lysophosphatidylinositol fraction in the absence of lipid X and 226 cpm in its presence. If this correction is made to the data, then the stimulation of lysophosphatidylinositol formation by lipid X is 5.8-fold. When the rapid extraction procedure described in the text is used, phosphatidylinositol is removed prior to acidification, and its breakdown cannot contribute to the observed lysophosphatidylinositol pool.

~,03-Diacylglucosamine 1-phosphate (lipid X), a monosaccharide precursor of Escherichia coli lipid A, was used to stimulate RAW 264.7 macrophage tumor cells, and the effects on macrophage phospholipid metabolism were examined. The addition of E. coli lipid X to the medium of cells that had been uniformly labeled with 32Pi resulted in a 4-&fold increase in the level of lysophosphatidylinositol. This effect was maximal at 5 PM lipid X. Lysophosphatidylinositol levels reached a maximum 45 min after stimulation, followed by a gradual decline to near normal levels within 2 h. The formation of lysophosphatidylinositol was dependent upon extracellular calcium and was almost completely inhibited when cycloheximide was added at the time of stimulation. The addition of the disaccharide lipid A precursor IVA, commercial lipopolysaccharide (1 pg/ml), phorboll2-myristate 13-acetate (lo" M), or calcium ionophore A23187 M) to these cells resulted in a similar increase in lysophosphatidylinositol levels, but phosEhatidic acid was inactive. The stimulation by IVA and phorbol myristate acetate was blocked by cycloheximide, but the @timulation by lipopolysaccharide was only partially blocked. The stimulation by A23187 was unaffected by cycloheximide. The increase in lysophosphatidylinositol levels might be related to the stimulation of arachidonate release and prostaglandin synthesis that is also observed in cells treated with lipid A precursors. The disaccharide precursor, IVA, was at least 100 times more effective than lipid X at stimulating lysophosphatidylinositol formation and prostaglandin release. The relative ability of lipid X and IVA to stimulate these cells correlated well with their effects on other lipopolysaccharideresponsive systems. Macrophage tumor cells also had the ability to inactivate lipid X by dephosphorylating it.
Lipid A (Fig. l), the hydrophobic portion of lipopolysaccharide, serves to anchor lipopolysaccharide to the outer membrane of Gram-negative bacteria (1,2). Lipid A is also responsible for many of the pathophysiological effects associated with Gram-negative infections, including endotoxin-induced shock (3,4). Some of these events are mediated by macrophages, which are stimulated by lipopolysaccharide to produce and secrete prostaglandins (5-7), interleukin-1 (8), neutral proteinases (9), and tumor necrosis factor (10). The mechanisms by which lipopolysaccharide activates macrophages are not well understood. One reason for this is that the structure of lipid A, now believed to be the active component of lipopolysaccharide (1,2,7), was unknown prior to 1983 (11)(12)(13)(14)(15). Recently, a monosaccharide precursor of lipid A was isolated from a mutant strain of Escherichia coli (11,16) in our laboratory. This substance, N',@-diacylglucosamine 1-phosphate (11) (also termed lipid X),' displays some of the biological activities of lipid A (7,(17)(18)(19); but lipid X is unique in that it is nontoxic to animals (20, 21). In some cases, it may actually protect against endotoxin-induced shock associated with Gram-negative bacteremia @I).* Lipid X (Fig. 1) possesses a formal structural resemblance to phosphatidic acid (22). Both have two long-chain fatty acyl moieties attached to a carbohydrate backbone, and both are phosphomonoesters (22). Because of the similarity of lipid X to classical glycerophospholipids (22,23), we considered the possibility that lipids X and A might exert some of their effects on animal cells by perturbing glycerophospholipid metabolism. Consequently, we investigated the effects of lipid X and other lipid A precursors on the composition of membrane phospholipids in RAW 264.7 macrophage tumor cells, a cell line that responds to lipopolysaccharide by synthesizing various proteins including tumor necrosis factor (24,25). We now report that the level of a minor phospholipid, tentatively identified as lysophosphatidylinositol, rises 4-8-fold after a 45-min exposure of these cells to lipid X. The lysophosphatidylinositol response is dose-dependent, is observed with much lower concentrations of the more biologically active disaccharide precursor IVA ( Fig. 1 and Refs. [26][27][28], and is correlated with prostaglandin formation, possibly suggesting the involvement of a specific phospholipase A, activity. The levels of the major glycerophospholipids, as well as lysophosphatidylcholine and lysophosphatidylethanolamine, are not significantly altered. The accumulation of lysophosphatidylinositol is an early response to lipid A precursors and lipopolysaccharide and may provide new molecular insights into the interaction of lipid A with animal cell membranes.
Isolation of Lipid X and ZV, from Biological Sources-Lipid X and IVA were isolated from E. coli strain MN7 (11) and Salmonella typhimurium strain STi50 (26), respectively. Final purification was achieved by HPLC using an Alltech CIS 10-pm reverse-phase column eluted with an acetonitri1e:water:isopropyl alcohol system reported previously (29).
Cell Lines and Culture Conditions-RAW 264.7 cells were obtained from the American Type Culture Collection. They were typically maintained in F-12 medium (GIBCO), supplemented with 10% fetal bovine serum (GIBCO). Cells were grown as suspension cultures in plastic Petri dishes not intended for tissue culture (GIBCO) since the cells did not adhere to these dishes. For stimulation experiments, the cells were plated onto Falcon or Corning tissue culture dishes to which they tightly adhered. Cells were grown and all stimulation experiments were performed at 37 "C, but F-12 medium without serum was used for the stimulation of cells unless otherwise indicated.
Stimulation of Lysophosphutidylimsitol Lubeling in RAW Macrophage Cells-Cells were plated in 60-mm diameter tissue culture dishes at a density of lo5 cells/dish and labeled with 32Pi (5-10 pCi/ ml) for 60-72 h, at which time the cell density was -1-1.5 X lo6 cells/dish. The labeled medium was removed, and F-12 medium (without serum) containing the desired stimulant was added. After incubation for 45 min, the medium was removed, and the cells were washed with 5 ml of phosphate-buffered saline (PBS) (30) and harvested by scraping in 0.8 ml of PBS. The cells were added to 3.0 ml of CHC13:MeOH (1:2, v/v) containing 300 pg of carrier lipid (extracted from mouse liver) and 25 pg of carrier lysophosphatidylinositol. After 15 min at room temperature, 1.0 ml of CHC4 and 1.0 ml of PBS were added to form two phases. After centrifugation at 600 X g, the upper phase was recovered. Care was taken not to include the interfacial material during removal of the upper phase. The upper phase was washed once with 2 ml of a neutral pre-equilibrated lower phase solution to remove any residual bulk lipids. The washed upper phase was acidified by the addition of 50 pl of concentrated HC1 and extracted twice using 2 ml of an acidifiedpre-equilibrated lower phase solution during each wash. The lower phases from the acidic extraction were combined, dried under nitrogen, and chromatographed on thin-layer plates (Merck, Silica Gel 60) using solvent system A. The radioactive species were localized by autoradiography and identified by migration with authentic standard. Radioactivity was quantitated by scraping the band of interest into a scintillation vial containing 1.0 ml of methanol. Next, 10 ml of Patterson and Green scintillation solution (31) was added for counting. When an analysis of the entire phospholipid pool, including lysophosphatidylinositol, was desired, the cells were extracted directly under acidic conditions (16), without prior extraction under neutral conditions. In the latter case, the isolated lipids were separated with a two-dimensional thin-layer chromatography system consisting of solvent system B in the first dimension and solvent system C in the second dimension.
Stimulation of Prostaglandin Formation-Cells were plated at 3 X IO6 cells/35-mm diameter tissue culture dish and allowed to attach for 24 h, essentially as previously described for mouse peritoneal macrophages (32). Medium was removed, and 1 ml of F-12 (with serum) containing 2 pCi of [3H]arachidonic acid was added. After 5 h, the labeled medium was removed, the cells were washed twice with 3-ml portions of serum-free F-12 medium, and 1 ml of serum-free medium, containing stimulant was added. Medium was recovered after 1 h and added to 3.75 ml of CHC13:MeOH (1:2, v/v) containing 300 pg of mouse liver lipid and 20 pg each of PGE,, PGD,, and arachidonic acid. Next, 1.25 ml of CHCb, 1.25 ml of PBS, and 25 pl of concentrated HC1 were added, and the lower phase was recovered after centrifugation at 600 X g for 10 min. The lower phase was dried under nitrogen and analyzed by thin-layer chromatography using solvent system F (32). Radioactive species were localized by fluorography at -65 "C using EN3HANCE (Du Pont-New England Nuclear), and prostaglandin species were identified by their migration with authentic standards. Radioactivity for each species was quantitated as described above.
Preparation of 32P-Labeled Lysophosphutidylinositol-Cells were labeled for several generations with ',Pi. After harvesting the cells and extraction of the labeled lipids, phosphatidylinositol was isolated by two-dimensional thin-layer chromatography (33). Phosphatidylinositol was extracted from the scrapings with CHCl3:MeOH:H,0 (50:50:5), dried under nitrogen, and digested with phospholipase A, using a modification of the method of Trotter et al. (34). The reaction mixture was spotted directly on a thin-layer plate, which was developed in solvent system B, to separate lysophosphatidylinositol from other products.

Uptake and Metabolism of p'C]Lipid X by RAW 264.7 Cells-['4C]
Lipid X (-lo5 cpm/nmol) was prepared as described previously (11) and was added at 1 pg/ml to 1-ml cultures of 1 x lo6 RAW 264.7 macrophages (32) for the times indicated. The medium was then removed and acidified with HCl to pH 1. The cells were scraped into 1 ml of 0.1 M HCl. Both medium and cells were extracted by the method of Bligh and Dyer (see Ref. 16 and 35). ["CILipid X and its metabolites were separated by thin-layer chromatography using either solvent system G or H. Products were visualized by overnight autoradiography using Kodak XAR-5 film and quantified by liquid scintillation counting (31).

RESULTS
Effect of Lipid X on the Membrane Phospholipid Composition of RAW 264.7 Macrophages-The phospholipid composition of RAW 264.7 macrophages uniformly labeled with 32Pi was determined by two-dimensional thin-layer chromatography, followed by autoradiography (Fig. 2) and scintillation * I Analysis of the phospholipids of stimulated and nonstimulated RAW cells by two-dimensional thin-layer chromatography. RAW 264.7 cells were labeled for 60 h at 37 "C in serum-supplemented F-12 medium containing 32Pi (7 pCi/ml). In the above, cells were treated with 5 p~ lipid X ( B ) (added directly to the medium from a 5 mM lipid X stock in 10 mM Tris-C1, pH 8) or buffer ( A ) for 45 min at 37 "C. Medium was then removed, and the cells were washed once with PBS and harvested by scraping in 0.8 ml of PBS. Lipids were extracted by the acidic Bligh-Dyer method (see Ref. 16 and "Experimental Procedures"). The labeled phospholipid species were separated by two-dimensional thin-layer chromatography using solvent system B in the first dimension and solvent system C in the second dimension. The plate was allowed to dry in a fume hood for 15-20 min between dimensions. The individual lipid species were identified on the basis of their migration with authentic standards.  16 and "Experimental Procedures") and analyzed by two-dimensional thin layer chromatography (Fig. 2). The location of each species was identified by autoradiography, and the area corresponding to each labeled species was scraped into a scintillation vial containing methanol. The samples were counted after the addition of 10 ml of Patterson and Green scintillation fluid If this correction is made to the data, then the stimulation of lysophosphatidylinositol formation by lipid X is 5.8-fold. When the rapid extraction procedure described in the text is used, phosphatidylinositol is removed prior to acidification, and its breakdown cannot contribute to the observed lysophosphatidylinositol pool.  (Table I). In these initial experiments, the total phospholipid fraction was extracted under acidic Bligh-Dyer conditions (see Ref. 16). As shown in Fig. 2 and Table I, macrophages treated for 45 min with 5 p~ lipid X (Fig. 2B) had essentially the same phospholipid composition as untreated control cells (Fig. 2 A ) , with the exception of a slowly migrating component (indicated by the arrowheads in Fig. 2).
This material, which migrated with authentic lysophosphati-dylinositol, increased 3.5-fold in the lipid X-treated cells (Table I), rising from 0.2% of the total phospholipid in the absence of lipid X to 0.7% in its presence. Selective Extraction of the Accumulated Lipid and Its Identification as Lysophosphatidylinositol-The putative lysophosphatidylinositol was recovered in the lower, chloroform phase only if the extractions were performed under acidic conditions. This prompted us to devise a more rapid assay for this material in radiolabeled macrophages, employing its partitioning properties as a function of pH. As detailed under "Experimental Procedures," the radiolabeled phospholipids of the macrophages are initially extracted under neutral conditions. All the major diacylglycerophospholipids partition into the lower phase, whereas most of the putative lysophosphatidylinositol remains in the upper phase. When this material is recovered from the upper phase by acidification and washing with a fresh lower phase, it is relatively pure and can be separated from remaining contaminants by one-dimensional thin-layer chromatography, as illustrated in Fig. 3. Quantitation of the putative lysophosphatidylinositol recovered in this manner from cells labeled either with 32Pi or [3H]inositol is shown in Table I1 and is in good agreement with the results obtained by the direct, acidic extraction (Table I and Fig. 2).
The labeled compound that accumulates in the presence of lipid X migrated with authentic lysophosphatidylinositol in four thin-layer chromatography systems (solvent systems A-D). Furthermore, when authentic lysophosphatidylinositol was subjected to the same extraction protocol (Table III), it demonstrated partitioning properties identical to that of the unknown. The data presented above strongly suggest that this substance is lysophosphatidylinositol and that lysophosphatidylinositol accumulates in these cells during exposure to lipid X. It seems likely that the lysophosphatidylinositol is predominantly the 1-monoacyl isomer, but on the basis of the present data, it is not possible to exclude the idea that it arises in vivo as the 2-monoacyl isomer. It was not feasible to isolate enough material for mass spectrometry or NMR spectroscopy. However, the selective extraction method was very useful for monitoring the accumulation of lysophosphatidylinositol under various conditions provided the 71% yield (Table 111) was taken into account.
-+  (34), was added to approximately lo7 cells in 0.8 ml of phosphate-buffered saline. This was immediately added to 3.0 ml of ch1oroform:methanol (1:2, v/v) and extracted using the neutral-acid extraction procedure exactly as described under "Experimental Procedures." The lower phases of the neutral and acidic extractions were dried and counted by liquid scintillation in Patterson and Green solution (31). Unrecovered counts were those that were associated with the interfacial material during the extraction. The values represent the mean f S.D. of four samples.  Time Course of Lysophosphatidylinositol Accumulation and the Effects of Other Agonists-As shown in Fig. 4, lysophosphatidylinositol begins to accumulate above control levels Cells were labeled with 32Pi as described for Fig. 2 and treated with 5 p~ lipid X for the indicated times. After washing and harvesting the cells, lysophosphatidylinositol was isolated as described for Fig. 3. The labeled band corresponding to lysophosphatidylinositol was scraped into a vial, eluted with 1 ml of methanol, and counted following the addition of 10 ml of Patterson and Green scintillation solution (31). Values represent the averages of two samples. Duplicate samples did not vary greater than 10% of the absolute value of the mean. 25 min. Chemically synthesized lipid X has essentially the same effect on the lysophosphatidylinositol pool (Table IV) as lipid X isolated from E. coli strain MN7 (11) and purified by HPLC (29) as described under "Experimental Procedures." In other biological systems that have been examined, the disaccharide precursor IVA or mature lipid A (Fig. 1) are much more potent agonists than is lipid X (17,18,26,28,35). In some systems, lipid X may even be an antagonist (21). Accordingly, HPLC-purified precursor IVA (29) and commercial lipopolysaccharide were examined, and both were also active in stimulating lysophosphatidylinositol release (Table IV). Interestingly, phorbol myristate acetate and the calcium ionophore A23187 mimicked the lipid A metabolites in stimulating lysophosphatidylinositol accumulation (Table IV).
Effects of Cycloheximide and Calcium on Lysophosphatidylinositol Formation-When cycloheximide was added to the

Accumulation of lysophosphatidylinositol in response to various stimuli and its sensitivity to cycloheximide
Cells were labeled for 60 h with 32Pi (5 pCi/ml) and then stimulated with the indicated compounds for 45 min in the presence or absence of 5 pg/ml cycloheximide. Cells were harvested, and lysophosphatidylinositol was isolated using the neutral-acid extraction procedure described under "Experimental Procedures." All values represent the mean of three dishes. medium together with biologically derived lipid X, synthetic lipid X, or precursor IVA, the accumulation of lysophosphatidylinositol was almost completely inhibited (Table IV). This inhibition was also observed when phorbol myristate acetate was used as the stimulator. On the other hand, the effect of lipopolysaccharide was only partially blocked by the addition of cycloheximide, and the action of the calcium ionophore A23187 was not changed by cycloheximide. The ability of A23187 to stimulate lysophosphatidylinositol accumulation prompted us to examine the requirement of the response for extracellular calcium. As demonstrated in Table V Cells-The stimulation of prostaglandin formation by resident peritoneal macrophages is dependent upon extracellular calcium (36) and can be inhibited by the addition of cycloheximide (37). Considering the similarities between this event and lysophosphatidylinositol formation, it was expected that the formation of lysophosphatidylinositol in these cells might be associated with the formation and release of prostaglandins. Therefore, cells which had been prelabeled with [3H] arachidonic acid were treated with the same agonists that had been used to stimulate lysophosphatidylinositol. All agonists examined were also capable of stimulating prostaglandin release (Table VI). Commercial lipopolysaccharide and precursor IVA were the most effective, causing a 9-10-fold increase in the release of both PGD, and PGE2.3 Lipid X, at the concentration used, was only a weak stimulator of prostaglandin release ( Table VI); but in these experiments, lysophosphatidylinositol release was not quantitated simultaneously since the conditions for measuring prostaglandin release are somewhat different (see "Experimental Procedures"). A more detailed examination using identical cell cultures showed that lysophosphatidylinositol and prostaglandin formation may be somehow related (Fig. 5). When various concentrations of lipid X and precursor IVA were used to stimulate cells which had been doubly labeled with 32Pi and [3H] arachidonate, the stimulation of lysophosphatidylinositol formation correlated to a first approximation with prostaglandin Prostaglandins D, and E, were identified solely by thin-layer chromatography, as described under "Experimental Procedures." We do not know why RAW cells elaborate more PGD, than PGEz, whereas mouse peritoneal macrophages generated mostly PGEZ. Cells were labeled for 60 h with 32Pi (5 pCi/ml). Medium was removed and replaced with Puck's saline containing either CaC1, or CaCL and EGTA. Cells were incubated in the presence or absence of 5 pM lipid X for 45 min and then harvested by scraping. Lysophosphatidylinositol was isolated using the neutral-acid extraction procedure described under "Experimental Procedures.'' All values represent the mean f S.D. of three samples. release, especially at lower concentrations. However, subtle differences were observed in the extent of stimulation in the micromolar range (Fig. 5). As shown in Fig. 5, IVA was several orders of magnitude more active than lipid X in stimulating both processes, whereas phosphatidic acid was inactive at all concentrations examined. Lysophosphatidylinositol production preceded the release of both arachidonic acid and prostaglandin D2 (Fig. 6). Cellular lysophosphatidylinositol levels peaked 30-45 min after stimulation with IVA. Arachidonic acid release leveled off after 45 min, when lysophosphatidylinositol levels began to decrease. Prostaglandin Dz was released throughout the 90-min period.

Arachidonic Acid Turnover during Stimulation by Lipid X-
The loss of [3H]arachidonic acid from the various phospholipid species during stimulation by lipid X was also examined (Table VII). There was a significant loss of label from phosphatidylinositol and, possibly, from phosphatidylethanolamine, whereas phosphatidylcholine showed no change. The loss of label from both phosphatidylinositol and phosphatidylethanolamine was inhibited by the addition of cycloheximide, consistent with the view that arachidonate mobilization might account for the increased levels of prostaglandins recovered in the medium. Lipid X also caused a significant increase in diglyceride formation, as judged by [3H]arachidonate labeling, that was not inhibited by the addition of cycloheximide. However, in a separate experiment in which lysophosphatidylinositol was isolated from RAW cells labeled with [3H]arachidonate, control values and stimulated values were identical (data not shown).  (LPZ, A ) . Prostaglandins were isolated as described under "Experimental Procedures," and lysophosphatidylinositol was quantitated as described for Fig. 3. Values are expressed as the ratio of counts recovered in stimulated versus unstimulated control cultures. For lysophosphatidylinositol (A), the control value is 160 cpm, whereas for prostaglandin D2 (B), the value is 200 cpm. All ratios represent the mean of three cultures, and the error bars represent the standard deviations.
Uptake and Metabolism of Lipid X by RAW 264.7 Macrophages during Stimulation-To investigate the physical interactions of lipid X with RAW 264.7 cells, [14C]lipid X (-lo5 cpm/nmol) was prepared from E. coli strain MN7, as described previously (11). The metabolic labeling of MN7 with [1-I4C] acetate allowed for incorporation of label exclusively into the N-and 0-linked (R)-3-hydroxymyristoyl moieties (Fig. 1).
[14C]Lipid X (1 pg/ml) was incubated with cultures of RAW 264.7 cells, and its time-dependent redistribution between medium and cells was monitored, as described in Fig. 7. As [14C]lipid X was depleted from the culture medium, exactly corresponding amounts became cell-associated (Fig. 7). In all cases, the radioactivity added was quantitatively recovered. Furthermore, [I4C]lipid X which became cell-associated was subsequently converted to a less polar species which migrated near the front in solvent system G (Fig. 8). Since only one new radiolabeled species was produced, it was unlikely that the modification was a deacylation of the glucosamine 1phosphate ring. A second solvent system (solvent system H) was employed to resolve lipid X, (R)-3-hydroxymyristate, and N',@-diacylglucosamine with R, values of 0, 0.5, and 0.2, respectively. When the cell-associated [I4C]lipid X was extracted and the products were resolved in solvent system H, the less polar metabolite migrated as a single product with an N2,03-diacylglucosamine chromatographic standard (data not shown).
Neutrophils and macrophages purportedly have the capacity to alter the toxicity of lipopolysaccharide (38); and, as such, it was relevant to determine if the conversion of lipid X to N2,@-diacylglucosamine altered its prostaglandin stimulatory capacity. At 10 p~, N2,03-diacylglucosamine was inactive (not shown), demonstrating that RAW 264.7 cells are able to efficiently internalize lipid X and convert it to a product inactive in the stimulation of prostaglandin release. Interestingly, a steady-state level of cell-associated lipid X (Fig. 8) was achieved at about the same time that lysophosphatidylinositol accumulation is maximal (Fig. 4).

DISCUSSION
The lipid A moiety of lipopolysaccharide triggers many complex physiological responses in animal systems, including fever, shock, and the activation of certain immune cells (1-4). The response of macrophages is especially important since the prostaglandins and proteins elaborated by macrophages, in the presence of lipid A, may be important mediators of the observed pathology (4-10). Recent evidence has implicated tumor necrosis factor, one of several specific proteins synthesized by lipid A-treated macrophages, as the causative agent of shock (10,251. The increased production of tumor necrosis factor can be attributed to increased mRNA synthesis and more efficient translation of pre-existing mRNA (10,39). However, very little is known about the initial interactions of lipid A with animal cell membranes (40) (receptors, second messengers, etc.) or about the pathways that ultimately lead to the synthesis of specific proteins (25,41).
The RAW 264.7 macrophage tumor cell line (24) offers a relatively simple model system with which to probe lipid Aanimal cell interactions. These cells can synthesize prosta- (5 pCi/ml) and [3H]arachidonic acid as described for Fig. 5. The medium was removed, and the cells were washed twice with F-12 medium containing 0.5% endotoxin-free bovine serum albumin. Next, 2 ml of F-12 medium containing 0.5% endotoxin-free bovine serum albumin and 500 nM IVA were added, and cells were incubated at 37 "C. At various times, the medium and cells were harvested separately, as described for   (32). Medium was removed and replaced with serum-free medium containing no additions (control), 5 p~ biological lipid X, or biological lipid X plus 5 pg/ml cycloheximide. After 60 min at 37 "C, medium was removed, and the cells were washed with phosphate-buffered saline and harvested by scraping. The cells and medium were extracted in order to isolate glycerolipids and prostaglandin species, respectively. Prostaglandins were identified by thin-layer chromatography as described under "Experimental Procedures." The individual glycerolipid species were isolated by a two-step development of a one-dimensional thin-layer plate. Total cellular glycerolipids were spotted onto Silica Gel 60 plates and developed to 14 cm with solvent system B. The plate was dried for 10 min. Next, the plate was redeveloped in solvent system E, separating the rapidly migrating neutral lipid species. The bands of interest were located by fluorography at -65 'C, and the individual species were quantitated by liquid scintillation after scrap- glandins (see above) and tumor necrosis factor (10,25) in response to endotoxin. The feasibility of cloning single RAW cells makes it possible to consider the use of somatic cell genetics to dissect the lipid A response. The RAW system also avoids the complication of cell-cell interactions that may play a role in lipid A-triggered B lymphocyte mitogenesis (17).
In this study, we have examined the effects of lipid A precursors on the metabolism of endogenous phospholipids of RAW cells. No pronounced changes in composition were observed, with the exception of a 5-fold increase in the lysophosphatidylinositol content. To our knowledge, this effect of lipid A and its precursors on animal cells has not been reported previously, although mouse peritoneal macrophages have previously been shown to accumulate lysophosphatidylinositol in the presence of A23187 (42). As is the case with other lipid A-mediated effects (17, 18,26,28,35), disaccharide lipid A precursor IVA (Fig. 5 ) and mature lipid A4 are much more effective than are the monosaccharide precursors.
It is very difficult, at this point, to assess the relationship between the build-up of lysophosphatidylinositol and other phenomena associated with the stimulation of the RAW cells. Several factors suggest that lysophosphatidylinositol accumulation might occur in conjunction with prostaglandin formation. I) All the compounds ( i e . lipid A precursors, phorbol, and A23187) that caused lysophosphatidylinositol accumulation also caused prostaglandin release. 2) To a first approximation, lysophosphatidylinositol accumulation correlated with prostaglandin release when different concentrations of lipid X and IVA were used to stimulate the cells (Fig. 5 ) . Furthermore, an analysis of the time course of lipid IVA action revealed that prostaglandin accumulation does not precede the appearance of lysophosphatidylinositol (Fig. 6). 3) When cells that were labeled with [3H]arachidonic acid were stimulated, there was a significant decrease in 3H associated with phosphatidylinositol (Table VII)

FIG. 8. Conversion of ["Cllipid
X to a more rapidly migrating product by RAW 264.7 macrophages. The protocol for this experiment was the same as described for Fig. 6. A, left lane, autoradiogram of [14C]lipid X chromatographed in solvent system G before the addition of cells; right lane, same analysis of cell-associated radioactive lipid after 1 h in culture. B, total cell-associated radioactivity (O), cell-associated radioactivity remaining as lipid X (A), and radioactivity recovered with the rapidly migrating product (U), identified as 2,3diacylglucosamine, as described in the text. DAG-P, 2,3-diacylglucosamine 1phosphate.

A.
FRONT -ORIGINbuild-up triggered by lipid A precursors demonstrated many of the characteristics observed for arachidonate and prostaglandin release from macrophages, including the requirement for extracellular calcium, the sensitivity to cycloheximide, and the ability of A23187 to by-pass the cycloheximide inhibition (36,37).
Although the pattern of lysophosphatidylinositol accumulation and prostaglandin release triggered by lipid A-like molecules in RAW cells is consistent with the predominant involvement of a phospholipase Az, catalyzing the direct release of arachidonate from phosphatidylinositol, the data are only suggestive. We cannot exclude other sources of arachidonate to account for the observed prostaglandin formation. For instance, stimulation of the cells with lipid X caused a significant decrease in arachidonate from phosphatidylethanolamine and an increase in arachidonate-labeled diglyceride ( Table VII), suggesting that a phospholipase C may be activated and that a role for protein kinase C must be considered (19). Furthermore, there may be direct effects of lipid A precursors on protein kinase C itself (19). Both phospholipase A, and C activities have been shown to exist in mouse macrophages (43,44). 5 The ability of cycloheximide to inhibit the accumulation of lysophosphatidylinositol triggered by lipid A precursors could be interpreted to mean that protein synthesis is required to initiate the phospholipase A2 response. The requirement for extracellular calcium (Table V) and the ability of the calcium ionophore A23187 to by-pass the inhibition by cycloheximide (Table IV) suggest that the stimulation of the cells by the lipid A precursors leads to the synthesis of a protein that facilitates calcium influx. Presumably, the increased intracellular calcium might then activate a phospholipase A2, either directly or indirectly.
The possibility must be considered that lysophosphatidylinositol has some of its own functions within RAW cells. Lysophospholipids can destabilize phospholipid bilayers and may be fusigenic (45), suggesting that they might play a role in exocytosis. Perhaps, lysophosphatidylinositol facilitates 'Murine resident peritoneal macrophages (6, 36) respond less vigorously to lipopolysaccharide alone than do RAW 264.7 cells. However, "normal" macrophages do mobilize considerable amounts of arachidonate when incubated with phorbol myristate acetate (6,36). Low concentrations of lipopolysaccharide prime normal macrophages to release even more arachidonate in the presence of phorbol myristate acetate (6,36). We have not observed such synergism with RAW cells (data not shown), perhaps reflecting differences of normal and tumor cell membranes.  the secretion of various proteins that accumulate in the medium of RAW cells exposed to endotoxin (25), just as lysophosphatidylinositol stimulates the release of insulin from pancreatic islet cells (46). Since we do not know whether the lysophosphatidylinositol that accumulates in vivo is the 1-or 2-monoacyl derivative, it remains a possibility that phosphatidylinositol serves as the donor for the fatty acylation of certain membrane proteins, a process that is enhanced in the presence of lipid A precursors (47).
We do not know whether the uptake and dephosphorylation of lipid X (Figs. 7 and 8) are required for the stimulation of lysophosphatidylinositol and prostaglandin metabolism. A priori, lipid A precursors might be imagined to exert their pharmacological effects by interacting with surface proteins or enzymes, not necessitating bulk uptake or metabolism. In the absence of further information, it seems reasonable to suggest that dephosphorylation is part of a detoxification mechanism since the product of lipid X dephosphorylation is biologically inactive (7). It would be interesting to isolate RAW mutant' lines that are unable to take up lipid X, using colony autoradiography (48), and to examine the effects of such mutations on the lysophosphatidylinositol response. As in other, more complex cellular or animal systems (17, 18,26,28,35), the monosaccharide lipid X was, at best, a much weaker agonist than the disaccharide IVA (Fig. 5). In some settings, especially in whole animals, lipid X is nontoxic and may even provide some protection against the lethal effects of lipopolysaccharide injection (20, 21). On the other hand, IVA, like endotoxin, can cause fatal shock (28,359. We have not yet tested the possibility that nonstimulatory concentrations of lipid X (or related monosaccharide analogs) might antagonize the stimulation observed with nanomolar concentrations of IVA or lipopolysaccharide (Fig. 5). In any event, the fact that the RAW system responded to lipid A precursors in a manner that parallels other more complicated biological systems makes it ideally suited for probing the molecular basis of endotoxin physiology.