Oxidized low density lipoprotein inhibits macrophage apoptosis by blocking ceramide generation, thereby maintaining protein kinase B activation and Bcl-XL levels.

Macrophages play a central role in the development and progression of atherosclerotic lesions. It is well known that oxidized low density lipoprotein (ox-LDL) promotes the recruitment of monocytes (which differentiate to macrophages) into the intima. We reported recently that ox-LDL blocks apoptosis in bone marrow-derived macrophages deprived of macrophage colony-stimulating factor (M-CSF) by a mechanism involving protein kinase B (PKB) (Hundal, R., Salh, B., Schrader, J., Gómez-Muñoz, A., Duronio, V., and Steinbrecher, U. (2001) J. Lipid Res. 42, 1483-1491). The aims of the present study were 1) to define the apoptotic pathway involved in the pro-survival effect of ox-LDL; 2) to determine which PKB target mediated this effect; and 3) to identify mechanisms responsible for PKB activation by ox-LDL. Apoptosis following M-CSF withdrawal was accompanied by activation of the caspase 9-caspase 3 cascade and cytochrome c release from mitochondria, but the caspase 8 pathway was unaffected. M-CSF withdrawal resulted in a marked and selective reduction in Bcl-XL protein and mRNA levels, and this decrease was prevented by ox-LDL. The ability of ox-LDL to preserve Bcl-XL levels was blocked by NFkappaB antagonists, thereby implicating IkappaB kinase as a key PKB target. M-CSF deprivation resulted in activation of acid sphingomyelinase and an increase in ceramide levels. Desipramine (a sphingomyelinase inhibitor) prevented the increase in ceramide and inhibited apoptosis after M-CSF deprivation. Ox-LDL completely blocked the increase in acid sphingomyelinase activity as well as the increase in ceramide after M-CSF deprivation. Pretreatment of macrophages with C2-ceramide reversed the effect of ox-LDL on PKB and macrophage survival. These results indicate that ox-LDL prevents apoptosis in M-CSF-deprived macrophages at least in part by inhibiting acid sphingomyelinase. This in turn prevents ceramide-induced down-regulation of PKB, the activity of which is required to maintain production of Bcl-XL.

Macrophage-derived foam cells play an important role in the development and progression of atherosclerosis (1). They are histologic hallmarks of early and intermediate atherosclerotic lesions and account for a substantial proportion of the volume of such lesions. More recently, foam cells have been shown to be key determinants of plaque instability, in that macrophagerich lesions are prone to undergo plaque rupture and thrombosis. One obvious factor contributing to the propensity of macrophagerich lesions to rupture is that foam cells have much less mechanical strength than the arterial connective tissue matrix. In addition, there is evidence that macrophages directly contribute to matrix degradation by secreting matrix-degrading enzymes known as matrix metalloproteases (MMPs). 1 Macrophages isolated from atheromas constitutively secrete the metalloproteases, MMP-1 and MMP-3 (2,3). Levels of MMP-1 and MMP-3 are increased in unstable lesions (3)(4)(5), and macrophages are responsible for the increased levels of matrix-degrading enzymes in lesions (6).
In view of this critical role of macrophages as determinants of plaque disruption, a comprehensive understanding of the factors that influence macrophage numbers and/or function in atherosclerotic lesions is of obvious importance. Early and intermediate atheromas are highly cellular and are accompanied by evidence of active cell proliferation (7). Although macrophages are often thought to be terminally differentiated cells and incapable of replication, they are the predominant cell type-expressing proliferating cell nuclear antigen in atherosclerotic lesions, even in lesions containing mostly smooth muscle-derived cells (8,9). In contrast to the cellularity of early lesions, the necrotic core that characterizes advanced lesions is composed in large part of lipid debris from macrophages and smooth muscle cells that have undergone apoptotic or necrotic death (7). Thus, signal transduction pathways leading to growth or death are evidently activated at different stages of lesion evolution. Vascular endothelial cells express Fas as well as its ligand FasL but are normally resistant to Fas-mediated apoptosis (10). In contrast, most other cells, including peripheral blood monocytes, express only Fas (11). Endothelial FasL * This work was supported in part by grants from the Heart and Stroke Foundation of British Columbia and Yukon and from the Canadian Institutes of Health Research. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Recipient of a travel award from the University of the Basque Country, Spain.
¶ Supported by a Fellowship from the Heart and Stroke Foundation of Canada.
ʈ To whom correspondence should be addressed: Division of Gastroenterology, 100 -2647 Willow St., Vancouver, British Columbia V5Z 3P1, Canada. Tel.: 604-875-5244; Fax: 604-875-5447; E-mail: usteinbr@interchange.ubc.ca. expression might be expected to attenuate mononuclear cell infiltrates in the arterial intima, as monocytes would be programmed to undergo apoptosis through the Fas/FasL signal by interacting with endothelial cells as they transmigrate into the intima. Macrophage apoptosis (terminal dUTP nick-end labeling positivity) is seen in atherosclerotic lesions, particularly after interventions that reduce cholesterol levels, and this is accompanied by a dramatic and selective loss of macrophages from lesions (12).
Recent results raise the intriguing possibility that oxidized LDL (ox-LDL) might be a mechanistic link between the above relationship between cholesterol levels and macrophage populations in the arterial intima. Ox-LDL has been shown to promote macrophage recruitment and retention in lesions (13), macrophage proliferation (14 -17), and macrophage survival (18). We showed recently (19) that ox-LDL inhibits apoptosis in cultured bone marrow-derived macrophages (BMDM), whereas native LDL or acetyl LDL had no effect. Ox-LDL was shown to cause activation of phosphatidylinositol 3-kinase (PI3K) and its downstream target protein kinase B (PKB), leading to PKBdependent phosphorylation of three proteins involved in apoptosis, the forkhead transcription factor, IB kinase (IK), and Bad (19).
The first objective of the present study was to define the relative importance of the "death receptor" pathway compared with the mitochondrial pathway in the pro-survival effect of ox-LDL. PKB phosphorylates forkhead transcription factor, and this prevents them from inducing transcription of proapoptotic proteins such as FasL (20 -22). The Fas/FasL death receptor pathway leads to caspase 3 activation and apoptosis via caspase 8. In contrast, PKB-dependent phosphorylation of Bad and IK inhibits the mitochondrial apoptosis pathway, which involves cytochrome c release and caspase 9 activation upstream of caspase 3. In this report, we show that caspase 9 but not caspase 8 is activated in BMDM after M-CSF withdrawal, and this activation is blocked by ox-LDL.
The second objective of this study was to determine whether Bcl-2 family members were involved in the inhibition of apoptosis by ox-LDL. We reported previously that ox-LDL-induced PKB activation in BMDM was accompanied by phosphorylation of Bad (a pro-apoptotic Bcl-2 family protein that is sequestered after phosphorylation) (19). However, in more recent experiments we found that Bad phosphorylation was inconsistent and not necessary for the pro-survival effect of ox-LDL. Therefore, we looked for other Bcl-2 family members that might be regulated by PKB via IK and NFB. We focused on Bcl-X L because it contains NFB-binding sites in its promoter (23) and can prevent the mitochondrial release of cytochrome c as well as the subsequent activation of the caspase 9-caspase 3 apoptotic cascade in the cytoplasm (24). In this study, we show that M-CSF withdrawal in BMDM is associated with a selective decrease in Bcl-X L levels and that this decrease is prevented by ox-LDL in parallel with its effects on PKB activity.
The third objective was to determine whether the pro-survival effect of ox-LDL was mediated through an effect on ceramide levels. Apoptosis induced by a variety of agents is associated with ceramide generation, and one of the mechanisms for this is inhibition of the PKB survival pathway (25,26). The present study indicates that M-CSF withdrawal is accompanied by a major increase in ceramide levels. This is shown to be due to activation of acid sphingomyelinase, and this activation of sphingomyelinase was blocked by ox-LDL. That ceramide generation in this model was both necessary and sufficient for apoptosis was demonstrated by the findings that desipramine, a sphingomyelinase inhibitor, completely blocked apoptosis and that addition of ceramide abolished the anti-apoptotic effect of ox-LDL.
Lipoprotein Isolation and Oxidation-LDL (d ϭ 1.019 -1.063) was isolated by sequential ultracentrifugation of EDTA-anticoagulated fasting plasma obtained from healthy normolipidemic volunteers (27). The concentrations of EDTA in LDL preparations were reduced prior to oxidation by dialysis against Dulbecco's phosphate-buffered saline (PBS) containing 10 M EDTA. Oxidation was performed using 200 g/ml LDL in Dulbecco's PBS containing 5 M CuSO 4 incubated at 37°C for 24 h (28). Oxidation was arrested with 40 M butylated hydroxytoluene and 300 M EDTA, and the modified LDL was then washed and concentrated LDL to about 2 mg/ml using Amicon Centriplus 20 ultrafilters (Millipore, Bedford, MA).
Concentration of LDL protein was determined by a modification of the Lowry method with bovine albumin as the standard (28).
Cell Culture-Bone marrow cells were isolated from the femurs of 6 -8-week-old female CD-1 mice as described (18). Cells were plated for 24 h in RPMI 1640 containing 10% FBS and 10% L-cell conditioned medium as the source of M-CSF. The non-adherent cells were removed and cultured in the above medium until 80% confluence was reached (5-7 days). Cells were then harvested using a Teflon cell lifter. To render cells quiescent prior to experiments, they were washed once with RPMI 1640 and incubated overnight (18 h) in RPMI 1640 with 10% FBS but without M-CSF.
Cell Viability Assay-Macrophages were seeded at 10 4 cells/well in 96-well plates and incubated overnight in RPMI 1640 with 10% FBS but without M-CSF. Ox-LDL and/or other compounds were then added, and cell survival was estimated by measuring the rate of reduction of the tetrazolium dye MTS. 20 l/well of MTS/PMS solution was added to microwells containing 100 l of culture medium. The final concentration of MTS was 333 g/ml MTS and that of PMS was 25 M. After incubation for 4 h at 37°C, the absorbance at 490 nm was recorded using an ELISA plate reader. For ceramide or inhibitor studies, cells were pre-incubated with inhibitors, 25 M C 2 -ceramide, or C 2 -dihydroceramide for 1 h prior to the addition of ox-LDL.
Ceramide and Sphingomyelin Determination-Radioactivity in ceramide was assayed after labeling of BMDM with 5 Ci/ml of [ 3 H]palmitate for 24 h in RPMI 1640 with 10% FBS. The cells were washed twice with PBS and were then incubated with or without 25 g/ml ox-LDL or 10 M desipramine. Cells were harvested by scraping into 1 ml of methanol, which was then mixed with 1 ml of chloroform and 0.9 ml of 2 M KCl, 0.2 M H 3 PO 4 . The aqueous phase was discarded, and the chloroform phase was dried under nitrogen. Ceramides were isolated by TLC by using Silica Gel 60-coated glass plates developed with chloroform/methanol/acetic acid (9:1:1 by volume) for half their length and then with petroleum ether/diethyl ether/acetic acid (60:40:1 by volume). Sphingomyelin was isolated by TLC in chloroform/methanol/acetic acid/ formic acid/water (35:15:6:2:1 by volume). Lipids were visualized by iodine and identified by co-chromatography with authentic standards. Radioactivity was measured by scraping the corresponding bands from TLC plates and liquid scintillation counting.
Diacylglycerol Kinase Assay for Ceramide Mass-Ceramide levels were measured using the diacylglycerol kinase method as described previously (29,30). In brief, total cellular lipids were extracted with chloroform/methanol, resuspended in a micellar solution of 7.5% ␤-octyl glucoside and 25 mM dioleoylphosphatidylglycerol, and then diacylglycerol kinase and [␥-32 P]ATP were added. After incubation for 30 min at room temperature, 50 g of unlabeled ceramide 1-phosphate was added as a carrier, and lipids were extracted and separated on Silica 60 TLC plates with chloroform/acetone/methanol/acetic acid/water (10:4:3:2:1 by volume). Ceramide 1-phosphate spots were scraped from the plates and quantitated by scintillation counting with normalization to total lipid radioactivity. The assay was calibrated with a standard curve of authentic ceramide.
Sphingomyelinase Assay-The activities of neutral and acidic sphingomyelinases were determined exactly as described by Liu and Hannun (31) using [N-methyl- 14 C] sphingomyelin as the substrate.
Western Blotting-Macrophages were plated at 5 ϫ 10 5 cells/well in 6-well plates or 5 ϫ 10 6 cells/well in 100-mm dishes. Cells were incubated overnight in RPMI 1640 with 10% FBS but without M-CSF. Ox-LDL and/or other compounds were then added as described above. Cells were harvested and lysed in ice-cold homogenization buffer (20 mM MOPS, pH 7.2, 1% Triton X-100, 50 mM ␤-glycerol phosphate, 5 mM EGTA, 2 mM EDTA, 1 mM sodium vanadate, 25 M ␤-methyl aspartic acid, 1 mM dithiothreitol, 1 mM phenylmethylsulfonyl fluoride, 10 g/ml aprotinin, and 10 g/ml leupeptin). Lysates were centrifuged at 14,000 rpm for 10 min, and the protein content of supernatants was quantified with the Bradford protein assay. 50 g of protein from each sample was loaded onto a 10 or 15% SDS-PAGE separating gel. Gels were calibrated using prestained SDS-PAGE low molecular weight standards. Proteins were then transferred electrophoretically to nitrocellulose paper, blocked for 1 h in Tris-buffered saline (TBS) containing 4% skim milk, 0.01% NaN 3 , and 0.1% Tween 20, and then incubated overnight with the primary antibody in TBS, 0.1% Tween 20 at room temperature. After three washes with TBS, 0.1% Tween 20, membranes were incubated with horseradish peroxidase-conjugated secondary antibody at 1:5000 dilution for 1 h. Bands were visualized by using enhanced chemiluminescence. Intensity of bands was quantified with NIH Image version 1.62 or with a Fluorochem 8000 imaging system (Canberra Packard Canada, Mississauga Ontario, Canada).
Subcellular Fractionation and Cytochrome c Assay-For isolation of nuclei, cell pellets were lysed in 0.5% Triton buffer and centrifuged at 1000 ϫ g for 5 min. The nuclear pellet was washed twice in PBS and resuspended in buffer containing 6 M urea. Nuclei were then sonicated to shear DNA. For subcellular fractionation cells were washed, harvested with a cell lifter, and suspended for 30 min in ice-cold buffer containing 10 mM Tris, pH 7.5, 300 mM sucrose, 10 M aprotinin, 10 M pepstatin A, 10 M leupeptin, and 1 mM phenylmethylsulfonyl fluoride. Cells were subjected to nitrogen cavitation at 60 pounds/square inch for 20 min and then centrifuged at 1000 ϫ g for 5 min to remove nuclei and intact cells. A second centrifugation at 10,000 ϫ g for 30 min was done to obtain cytosol. The pelleted mitochondria were lysed by sonicating for 20 s in buffer containing 1% Triton X-100, 10 mM Tris, pH 7.5, 150 mM NaCl, 10 M aprotinin, 10 M pepstatin A, 10 M leupeptin, and 1 mM phenylmethylsulfonyl fluoride and recentrifuged at 10,000 ϫ g for 30 min. Serial dilutions of cytosolic and mitochondrial samples were assayed for cytochrome c content using an ELISA kit from MBL International Corp. (Boston, MA) according to the manufacturer's instructions.
Reverse Transcriptase PCR for Bcl-X L -Total RNA from BMDM was reverse-transcribed using Superscript II according to the directions of the manufacturer (Invitrogen). By using either cDNA or mouse genomic DNA as a positive control template, Bcl-x L and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) were amplified by PCR, generating a 410-bp fragment for Bcl-X L (forward primer, 5Ј-CTTAATTCCGTGGT-GGTCGACTTTCTCTCC-3Ј; reverse primer, 5Ј-CGGAATTCCGACCC-CAGTTTACTCCATCCC-3Ј) or an 851-bp fragment for GAPDH (forward primer, 5Ј-CGCGCTGAGTATGTCGTGGAGTCT-3Ј; reverse primer, 5Ј-GTATTATGGGGGTCTGGGATGGAA-3Ј). The amplification conditions are as follows: 1 cycle at 95°C for 2 min and then 35 cycles at 94°C for 30 s, 60°C for 1 min, and 72°C for 1 min. Final extension was performed at 72°C for 5 min. The PCR products were electrophoresed in a 1% agarose gel, stained with ethidium bromide, and photographed with a Fluorochem 8000 imaging system.
Flow Cytometry-Macrophages were seeded at 10 6 cells per well in 6-well culture plates. For assessment of DNA content, macrophages were harvested by scraping, fixed in 70% ethanol for 30 min at room temperature, and then stained with 50 g/ml propidium iodide in 0.1% Triton, 0.1 mM EDTA, 0.05 mg/ml RNase A. For assessment of apoptosis and necrosis by vital staining, macrophages were harvested by scraping, suspended in PBS containing 1 g/ml Hoechst 33342 at 37°C for 7 min, and then 1 g/ml propidium iodide was added for a further 5 min. To quantify phosphatidylserine externalization, macrophages were harvested and incubated with annexin V-FITC (Pharmingen) according to the manufacturer's instructions. DNA content was analyzed by flow cytometry (Coulter Electronics, Hialeah, FL) on the FL-3 channel, and Hoechst 33342 or FITC fluorescence was measured on the FL-1 channel, with gating to exclude debris and cellular aggregates. Ten thousand events were counted for each analysis.
Immunofluorescence Microscopy-Macrophages were grown on sterile glass coverslips and then exposed to the indicated experimental conditions. For evaluation of phosphatidylserine externalization, the coverslips were washed twice with cold PBS and then incubated in 10 mM HEPES, 140 mM NaCl, 2.5 mM CaCl 2 , and 5 l of annexin V-FITC (Pharmingen) for 15 min at room temperature in the dark. For localization of cytochrome c, the coverslips were washed three times with cold PBS and then fixed in 3.7% paraformaldehyde and 0.18% Triton X-100 for 10 min at room temperature. Coverslips were then blocked with 1% bovine serum albumin for 15 min, incubated with a monoclonal anti-cytochrome c antibody (Pharmingen) for 45 min, and then with a goat anti-mouse AlexaFluor secondary antibody (Molecular Probes, Portland, OR) for 30 min. Cells were then examined immediately on a Zeiss Axiophot fluorescence microscope equipped with a digital imaging system.
Statistical Analysis-Results were expressed as means Ϯ S.E. Statistical analysis was done using analysis of variance or Student's t test as appropriate, with level of significance set at p Յ 0.05.

Growth Factor Withdrawal Induces Apoptosis in Bone
Marrow-derived Macrophages-When BMDM are cultured in the absence of growth factors, they typically undergo apoptosis after about 24 h (32). Ox-LDL completely prevents the loss of macrophage viability associated with growth factor withdrawal (19). To verify that the changes in macrophage viability in the present experiments were due to apoptosis, quiescent BMDM were incubated for 0 -24 h in the absence of M-CSF and doubly stained with Hoechst 33342 to identify apoptotic cells and with propidium iodide to identify necrotic cells. Flow cytometric analysis showed that by 24 h most cells stain with Hoechst 33342, and about half of these are also propidium iodide-positive ( Fig. 1). This indicates the primary mode of cell death in these experiments is apoptosis but that most cells subsequently undergo secondary necrosis. Cells incubated with ox-LDL showed no increase in Hoechst 33342 staining. Annexin V immunofluorescence microscopy confirmed that M-CSF withdrawal leads to phosphatidylserine exposure ( Fig. 2A) and that this is prevented by ox-LDL (Fig. 2B).
M-CSF Deprivation Selectively Activates the Caspase 9 (Mitochondrial) Apoptosis Pathway-Activation of the executioner caspase 3 may take place either through activation of caspase 8 in death receptor complexes associated with the plasma membrane or via the mitochondrial pathway involving cytochrome c release and caspase 9 activation (33,34). To determine whether apoptosis after M-CSF withdrawal was due to activation of one or both of these pathways, immunoblot analysis of cell lysates for these caspases was performed. As shown in Fig. 3, following M-CSF withdrawal there was a loss of pro-caspases 3 and 9, and a concomitant increase in the fragments corresponding to the respective active caspases. Ox-LDL prevented cleavage of these pro-caspases. There was no evidence of cleavage of procaspase 8 after M-CSF withdrawal. We have demonstrated previously (19) the functional activation of caspase 3 in this system as evidenced by cleavage of its substrate, poly(ADPribose) polymerase. To verify that caspase 9 activation was responsible for cell death after M-CSF withdrawal, the effects of selective caspase inhibitors on macrophage viability after growth factor withdrawal was measured. Results in Fig. 4 indicate that caspase 9 inhibitors partially blocked apoptosis, whereas a caspase 8 inhibitor was without effect. The same concentration of the caspase 8 inhibitor completely blocked caspase 8 activation in interleukin-3-deprived MC-9 cells (data not shown).
To confirm activation of the mitochondrial apoptosis pathway, cytochrome c was measured by enzyme immunoassay in mitochondria and cytosolic fractions prepared from BMDM. As shown in the upper panel of Fig. 5, M-CSF withdrawal triggered cytochrome c release, and this was prevented by ox-LDL. Immunofluorescence microscopy confirmed these findings, in that there was diffuse cytosolic staining for cytochrome c in a proportion of cells incubated without M-CSF, but there was only staining of perinuclear organelles in cells cultured with M-CSF or ox-LDL (Fig. 5, lower panel).
The Anti-apoptotic Effect of Ox-LDL Requires NFB Activation and Maintenance of Bcl-X L Levels-The PI3K/PKB signaling cascade can affect Bcl-2 family members directly, for example by phosphorylation of Bad, or indirectly through effects on other Bcl-2 proteins mediated by NFB (19). We therefore examined the expression of Bcl-2, Bad, Bax, and Bcl-X L at varying times after M-CSF withdrawal in the presence or absence of ox-LDL. As shown in Fig. 6, M-CSF withdrawal caused a selective decrease in the level of Bcl-X L protein and mRNA. These changes were completely prevented by addition of 25 g/ml ox-LDL. The magnitude of the decrease in Bcl-X L protein seemed greater than the decrease in normalized mRNA, so it is possible that non-transcriptional mechanisms are also implicated in the changes in Bcl-X L levels. The NFB inhibitor caffeic acid phenylethyl ester or a peptide inhibitor of NFB (SN 50) blocked the ability of ox-LDL to maintain Bcl-X L levels, as well as its effect on macrophage survival (Fig. 7). Under the conditions of these experiments, the inhibitors blocked nuclear translocation of NFB but did not prevent phosphorylation of PKB at Ser-473 (data not shown). This implies that induction of Bcl-X L expression by NFB is required for the anti-apoptotic effect of ox-LDL but does not rule out the possibility that effects on other Bcl-2 family members contribute to the inhibition of apoptosis by ox-LDL in these cells.
We reported previously (19) that ox-LDL leads to activation of MEK/MAPK as well as PI3K. To verify that the preservation of Bcl-X L levels by ox-LDL was mediated by activation of the PI3K/PKB pathway, phosphorylation of IB-␣, and Bcl-X L levels were examined following pre-treatment with 5 M LY 294002 or 100 nM wortmannin. As shown in Fig. 8, both these PI3K inhibitors blocked PKB activation, IB-␣ phosphorylation, and maintenance of Bcl-X L levels. MEK inhibitors (10 M PD 98059 or 2 M U0126) had no effect on IB-␣ phosphorylation or Bcl-X L levels.
Ox-LDL Blocks Ceramide Generation after Growth Factor Withdrawal-The lipid second messenger ceramide has been implicated in a number of cellular processes, including growth arrest and apoptosis (35,36). Ceramide may induce apoptosis by the inactivation of pro-survival pathways such as PKB, by directly activating pro-death pathways such as the stress-activated protein kinase pathway or by indirectly promoting apoptosis by triggering dephosphorylation of the pro-apoptotic protein Bad which would then be free to inhibit Bcl-X L (37). To test this, cells were preincubated for 24 h with [ 3 H]palmitate to label sphingomyelin (and other lipids). Cells were then washed to remove free [ 3 H]palmitate so that any changes in radioactivity in complex lipids would be due to degradation and not de novo synthesis. Fig. 9A shows that incubation of BMDM in the absence of M-CSF caused a progressive increase in ceramide. This was temporally correlated with a decrease in cell viability (Fig. 9B). Treatment with 25 g/ml of ox-LDL completely prevented ceramide generation as well as the fall in cell viability. Native LDL and acetylated LDL had no effect on ceramide levels, viability, or PKB phosphorylation (not shown). The increase in ceramide was confirmed with an in vitro assay for ceramide based on the ability of diacylglycerol kinase to phosphorylate ceramide (see "Materials and Methods"). With this assay, the ceramide level in quiescent BMDM at time 0 was 1.8 Ϯ 1.1% of total phosphate incorporated into lipid, and this increased after 24 h without M-CSF to 6.2 Ϯ 2.2%, whereas the ceramide level in cells incubated 24 h without M-CSF but with 25 g/ml ox-LDL was 3.4 Ϯ 1.1% (p ϭ 0.05 versus control).
An important aspect in the characterization of the ceramide pathway in apoptosis centers on the enzymes involved in ceramide generation. Some stimuli, such as tumor necrosis factor-␣, FasL, or ionizing radiation activate sphingomyelinase (38 -40), whereas in daunorubicin-induced apoptosis ceramide is generated by de novo synthesis (41). To determine whether ceramide generation in macrophages resulted from degradation of sphingomyelin, we tested the effect of desipramine, a drug that has been shown to reduce sphingomyelinase activity by accelerating its degradation (42). As shown in Fig. 10, desipramine prevented the generation of ceramide following M-CSF withdrawal and caused a corresponding increase in macrophage viability. The effect of desipramine was comparable in magnitude to that of 25 g/ml ox-LDL and was accompanied by a similar decrease in annexin V staining (not shown). As expected, desipramine prevented the degradation of sphingomyelin after M-CSF withdrawal. In control cells, not all of the decrease in labeled sphingomyelin was recovered as labeled ceramide. This probably reflects further metabolism of ceramide by degradation or phosphorylation. Inhibition of ceramide generation by desipramine fully replicated effects of ox-LDL on PKB activation and Bcl-X L levels (not shown). Taken together, these results represent compelling evidence that ox-LDL inhibits apoptosis of cytokine-deprived BMDM at least in part by preventing the generation of the apoptotic second messenger ceramide.
Growth Factor Withdrawal Activates Acid Sphingomyelinase-Induction of apoptosis by tumor necrosis factor-␣ and FasL has been shown to involve activation of acid sphingomyelinase in many cell types (38,39), whereas ionizing radiation induces apoptosis in endothelial cells by activating neutral sphingomyelinase (40). To determine which type of sphingomyelinase was implicated in apoptosis in M-CSF-deprived macrophages, we assayed acid and neutral sphingomyelinase in BMDM at 0, 24, and 30 h after M-CSF withdrawal. As shown in Fig. 11, more than 98% of sphingomyelin hydrolysis was attributable to acid sphingomyelinase. There was an increase in both acidic and neutral sphingomyelinase activity after M-CSF withdrawal, but only acid sphingomyelinase was inhibited by incubation with ox-LDL. This suggests that ox-LDL prevents ceramide accumulation by selective inhibition of acid sphingomyelinase.
Ceramide Prevents Ox-LDL-mediated PKB Activation-The PI3K/PKB pathway has been shown previously to be involved in BMDM survival mediated by M-CSF (43) as well as by ox-LDL (19). Ceramide has been reported to block PKB activation in HMN1 motor neuron cells (44). To determine whether the effect of ceramide in BMDM also involves PKB, BMDM were pre-treated with 25 M C 2 -ceramide or the same concentration of a biologically inactive ceramide analogue, C 2 -dihydroceramide. PKB activation was examined by phospho-specific immunoblotting for the two sites necessary for full enzymatic activity, Ser-473 and Thr-308 (45). As shown in Fig.  12, C 2 -ceramide but not C 2 -dihydroceramide prevented the activation of PKB by ox-LDL by blocking phosphorylation at both of these sites. Fig. 12 also shows that ox-LDL was unable to prevent apoptosis in the presence of exogenous ceramide. These results confirm the importance of PKB activation in ox-LDLmediated macrophage survival and demonstrate a link be-tween the inhibitory effects of ceramide on PKB activity and subsequent apoptotic cell death.
Changes in Apoptotic Markers Parallel Viability Changes-Flow cytometry results shown in Fig. 13 demonstrate that DNA fragmentation and annexin V binding paralleled the changes in viability associated with addition of ox-LDL and/or other compounds. However, ox-LDL did not completely prevent annexin V binding or DNA fragmentation as judged by flow cytometry even though cell viability with the MTS assay was completely preserved. The reason for this is unclear, as in other experiments ox-LDL completely prevented changes in Hoechst 33324 uptake after M-CSF withdrawal and also fully prevented annexin V binding to BMDM by immunofluorescence microscopy (Fig. 2). DISCUSSION The ability of ox-LDL to promote macrophage survival was first reported by Yui and co-workers (46). Later studies from this group provided evidence that this was due to lysophosphatidylcholine in ox-LDL leading to the autocrine release of granulocyte-monocyte-CSF from macrophages (14,15). Although previous results from our laboratory (16) agree with the observation that ox-LDL induced growth, we found that this was independent of lysophosphatidylcholine and was mediated by activation of a PI3K-dependent pathway.
More recently, we reported that ox-LDL promotes survival of bone marrow-derived macrophages by blocking apoptosis, again through a mechanism involving PI3K and PKB (19). Results in the present study show that ox-LDL also increases Bcl-X L levels, and link activation of PKB by ox-LDL and phosphorylation of its downstream target IB-␣ to the up-regulation of Bcl-X L by activation of NFB. Bcl-X L blocks apoptosis by inhibiting cytochrome c release (as do several Bcl-2 family proteins) and in addition prevents activation of the effector caspase 3 by sequestering Apaf-1 (47, 48).
In the present paper, we also define a new mechanism by which ox-LDL prevents apoptosis in BMDM, inhibition of acid sphingomyelinase. This inhibition prevents ceramide generation that normally follows M-CSF withdrawal and therefore blocks ceramide-dependent inhibition of PKB activity and Bcl-X L . Our working model of the mechanisms involved in ox-LDLmediated macrophage survival is outlined in Fig. 14.
In contrast to these reports of growth induction and/or inhibition of apoptosis by ox-LDL, there are many reports that ox-LDL causes apoptosis in macrophages, endothelial cells, or smooth muscle cells (49 -63). Several factors may account for this apparent contradiction, but we believe the most important is that the pro-survival effect of ox-LDL is seen at concentrations between 10 and 75 g/ml, whereas at higher concentrations cytotoxicity predominates (19,54). Nearly all reports of ox-LDL-induced apoptosis or cytotoxicity involve ox-LDL concentrations in excess of 100 g/ml. Other pertinent differences include cell type (transformed cell lines or primary cultures), differences in ox-LDL preparations, and incubation conditions (duration, medium composition, and serum content).  10. Ceramide is generated through sphingomyelin hydrolysis. For lipid analysis, BMDM were seeded at 10 6 cells/well in 6-well plates and pre-incubated overnight in RPMI 1640 with 10% FBS containing 5 Ci/ml of [ 3 H]palmitate but without M-CSF. Cells were then washed to remove unincorporated [ 3 H]palmitate, and one set of dishes was processed as the time 0 control. The remaining cell dishes were incubated for a further 24 h with no addition, with 25 g/ml ox-LDL as a positive control, or with 10 M desipramine to block sphingomyelinase. Radioactivity in ceramide and sphingomyelin was then determined as described under "Materials and Methods" and is presented as the ratio of dpm in ceramide to dpm in sphingomyelin (closed bars). For viability determination, BMDM were seeded at 10 4 cells/well in 96-well plates and treated as described above. Macrophage viability after 24 h was measured by the bioreduction of MTS (open bars). Results for viability are expressed relative to cells treated without ox-LDL at time 0. Data are pooled from three experiments, and ceramide values are means Ϯ S.E. of triplicate samples, and those for viability are means Ϯ S.E. of quadruplicates. *, p Ͻ 0.01 versus control 0, ox-LDL, or desipramine.
FIG. 11. Acid sphingomyelinase is responsible for ceramide generation in BMDM and is inhibited by ox-LDL. BMDM were seeded at 1 ϫ 10 6 cells/well in 6-well plates and incubated in RPMI 1640 with 10% FBS for 0 -30 h in the presence or absence of 25 g/ml ox-LDL. Cells were lysed by three cycles of freeze-thawing and assayed for acid sphingomyelinase (solid bars) or neutral sphingomyelinase (open bars) using a radiolabeled sphingomyelin substrate as described under "Materials and Methods." Note the abscissa is logarithmic. Each point represents the mean Ϯ S.E. of four independent experiments.
Although there are several shared characteristics between the growth-inducing and the anti-apoptotic effect of ox-LDL, we have not yet ascertained if the mechanism for growth induction is identical to that involved in blocking apoptosis. The threshold degree of oxidative modification of LDL seems to be lower for the anti-apoptotic effect than for growth induction (16,19), and the growth effect of ox-LDL is at least partly due to modified apoB (17), whereas the anti-apoptotic effect of ox-LDL can be fully accounted for by lipid extracts of ox-LDL. 2 However, we have not conclusively established that the two actions are attributable to different component(s) of ox-LDL. We have also not established the mechanism by which ox-LDL inhibits acid sphingomyelinase. However, we speculate that this inhibition may be a result of lysosomal dysfunction induced by ox-LDL. We have shown previously (64) that ox-LDL is poorly degraded and accumulates in lysosomes, due in part to resistance of ox-LDL to cathepsins. However, Hoff and colleagues (65,66) demonstrated that reactive aldehydes in ox-LDL can directly inactivate cathepsins, and it seems likely that such reactive components of ox-LDL could also inhibit lysosomal acid sphingomyelinase. It has been reported that 7-ketocholesterol in ox-LDL results in the inhibition of lysosomal sphingomyelinase (67). This presumably involves a mechanism other than direct covalent binding to the enzyme as 7-ketocholesterol is not particularly reactive. It is also possible that ox-LDL inhibits 2 R. Hundal, unpublished data. FIG. 14. Ox-LDL-induced macrophage survival, a working model. Ox-LDL prevents macrophage apoptosis following M-CSF withdrawal by at least two primary mechanisms: by inhibiting acid sphingomyelinase (thereby preventing ceramide generation) and by directly activating the PI3K/PKB pathway. PKB-mediated phosphorylation of IB-␣ leads to the release and activation of NFB, which then maintains expression of Bcl-X L . By inhibiting the release of cytochrome c from mitochondria, Bcl-X L prevents the activation of the caspase 9-caspase 3 cascade and subsequent apoptosis.
FIG. 12. C 2 -ceramide prevents ox-LDL-mediated PKB activation and survival. Upper panel, BMDM were seeded and pre-incubated for immunoblot analysis as described in the legend to Fig.  8. Macrophages were then incubated for 1 h with medium alone, with 25 M C 2ceramide, or with 25 M C 2 -dihydroceramide. Then, 10 g/ml ox-LDL was added for a further 1-h incubation. Phosphorylation of PKB was examined by immunoblotting with antibodies specific to PKB phospho-Ser-473, PKB phospho-Thr-308, and total PKB. Lower panel, BMDM were seeded at 10 4 cells/well in 96-well plates and treated as described above, except that the final incubation with or without ox-LDL was for 24 h. Viability was measured by the bioreduction of MTS. Results are expressed relative to cells treated without ox-LDL at time 0. Data for viability are means Ϯ S.E. of quadruplicate samples from two replicate experiments. Values for ox-LDL and ox-LDL ϩ dihydroceramide differed from control (p Ͻ 0.01). acid sphingomyelinase through an indirect signaling mechanism; for example, it has been reported that overexpression of PKB prevents the formation of ceramide (68). Hence, the prolonged activation of PI3K/PKB in BMDM by ox-LDL might potentiate the direct effect of ox-LDL on acid sphingomyelinase.
The observation that macrophages become resistant to apoptosis after exposure to ox-LDL might be viewed as an adaptive response that would facilitate the removal of damaged cell or tissue components by these cells. There is good evidence that ox-LDL and apoptotic cells share some structural features; for example, ox-LDL competes for the binding and phagocytosis of oxidatively damaged red blood cells by macrophages (69). Phosphatidylserine expression on the outer leaflet of the plasma membrane may play a key role in macrophage recognition of oxidatively damaged or apoptotic cells (70), and binding of phosphatidylserine liposomes to macrophages is inhibited by ox-LDL (71). A number of macrophage receptors have been shown to interact with phosphatidylserine-rich membranes including scavenger receptors CD36, CD68, and CD14, several integrins, and a novel phosphatidylserine receptor, but it is still not clear which of these mediate(s) the internalization of apoptotic cells (72,73).
Two recent, independent studies of the effects of ox-LDL on apoptosis and ceramide formation reached conclusions very different from those in the present report. In an endothelial cell line, LDL oxidized by UV light was reported to increase sphingomyelinase activity and ceramide levels (59). However, this study found that the induction of apoptosis by UV-oxidized LDL did not involve ceramide. Hence, it is likely that a direct toxic effect of this type of oxidized LDL under the conditions employed masked any potential anti-apoptotic effect of ox-LDL. Indeed, we have also observed that at high concentrations (over 100 g/ml), copper-oxidized LDL promotes death in macrophages (19). Copper-oxidized LDL was also reported to stimulate sphingomyelinase activity and induce apoptosis in human umbilical vein endothelial cells (56). In this latter study, ceramide generation was ascribed to a plasma membrane-associated acid sphingomyelinase and was causally related to apoptosis. The apparently contradictory conclusions between this study and ours may reflect differences in cell type or in the nature or concentration of the oxidized LDL used.