Sphingosine kinase regulates oxidized low density lipoprotein-mediated calcium oscillations and macrophage survival.

We recently reported that oxidized LDL (oxLDL) induces an oscillatory increase in intracellular calcium ([Ca(2+)](i)) levels in macrophages. Furthermore, we have shown that these [Ca(2+)](i) oscillations mediate oxLDL's ability to inhibit macrophage apoptosis in response to growth factor deprivation. However, the signal transduction pathways by which oxLDL induces [Ca(2+)](i) oscillations have not been elucidated. In this study, we show that these oscillations are mediated in part by intracellular mechanisms, as depleting extracellular Ca(2+) did not completely abolish the effect. Inhibiting sarco-endoplasmic reticulum ATPase (SERCA) completely blocked [Ca(2+)](i) oscillations, suggesting a role for Ca(2+) reuptake by the ER. The addition of oxLDL resulted in an almost immediate activation of sphingosine kinase (SK), which can increase sphingosine-1-phosphate (S1P) levels by phosphorylating sphingosine. Moreover, S1P was shown to be as effective as oxLDL in blocking macrophage apoptosis and producing [Ca(2+)](i) oscillations. This suggests that the mechanism in which oxLDL generates [Ca(2+)](i) oscillations may be 1) activation of SK, 2) SK-mediated increase in S1P levels, 3) S1P-mediated Ca(2+) release from intracellular stores, and 4) SERCA-mediated Ca(2+) reuptake back into the ER.

were provided by Bio-Rad (Hercules, CA). BioMax MR fi lm was from Kodak (Rochester, NY).
Relative electrophoretic mobility (R f ) of modifi ed lipoproteins was assessed using a Ciba-Corning (East Walpole, MA) electrophoresis apparatus and Titan agarose gels (Beaumont, TX) in 50 mmol/l barbital buffer, pH 8.6, according to the manufacturer's instructions. BSA was added to lipoprotein samples to ensure reproducible migration distances. Lipoprotein bands were visualized by staining with Fat Red 7B. All oxLDLs used in this study was extensively modifi ed with an R f value у 3 when compared with nLDL.

Cell culture
L929 cells (kindly provided by Dr. J. W. Schrader, Biomedical Research Centre, BC, Canada) were seeded in TufRol TM roller bottles (BD Falcon, San Jose, CA) at a density of 1.5 × 10 4 cells per cm 2 and cultured in media (DMEM, 10% FBS, 2 mmol/l L -glutamine, 1 mmol/l sodium pyruvate, 50 U/ml penicillin, and 50 g/ml streptomycin) containing 20 mmol/l HEPES at 37°C in a 5% CO 2 atmosphere. After 15 days, the media were harvested and centrifuged at 800 g for 10 min. The supernatant was fi lter sterilized through a 0.22 m fi lter. This L929 cell conditioned media (LCM) contains ‫ف‬ 10,000 U/ml of M-CSF ( 40 ). Bone marrow cells were obtained from the femurs of 6-8 week old female CD1 mice (Charles River Laboratories, Wilmington, MA) as previously described ( 15 ). Cells were cultured in media containing 10% LCM for 18 h at 37°C in a 95% humidity atmosphere containing 5% CO 2 . After 18 h, nonadherent cells were isolated and differentiated into macrophages by culturing them in medium containing 10% LCM until 80% confl uence was reached (5-6 days). Cells were washed to remove nonadherent cells and harvested using a rubber cell scraper (Sarstedt, Montreal, QC, Canada).

Calcium imaging
Bone marrow-derived macrophages (BMDMs) were seeded in 6-well plates at 5.0 × 10 4 cells per cm 2 and grown for 24 h. Cells were then washed with Ca 2+ -free DPBS and incubated in Ca 2+free, HEPES-buffered DPBS containing 2 mol/l Fluo-4-AM for 30 min at room temperature. Fluo-4-AM was dissolved with 20% pluronic acid in DMSO to make a 2 mmol/l stock solution. Cells were washed again with DPBS and incubated in HEPES-buffered medium with inhibitors where indicated for 10 min at room temperature to allow for deesterifi cation of the acetoxymethyl group. Medium was then removed, and fresh media containing test compounds and inhibitors where indicated were added. Fluorescence was measured every 0.6 s for 2 min using an inverted Leica TCS SP2 AOBS laser scanning confocal microscope with a 10× objective. Image analysis was performed using Leica LCS software, and fl uorescence of every cell in each fi eld was measured. On average, 68.2 ± 11.1 cells were separately analyzed per condition in each experiment. Cells exhibiting an increase of fl uorescence at mediate cell survival or induce apoptosis ( 19 ). Within the same cell, Ca 2+ signals can have dual roles in response to the same stimulus, depending on the temporal pattern of the Ca 2+ elevations. For example, distinct temporal patterns of Ca 2+ elevation are associated with positive versus negative selection of developing T-cells in the thymus ( 20,21 ). Weak T-cell receptor activation induces Ca 2+ oscillations, whereas strong T-cell receptor activation induces sustained Ca 2+ elevation. The former activates nuclear factor of activated T-cells optimally and thereby upregulates expression of the prosurvival cytokine IL-2, whereas the latter upregulates the proapoptotic BH3-only protein Bim.
Spingosine-1-phosphate (S1P) plays an important role in many cellular processes, including regulation of Ca 2+ signals (22)(23)(24)(25)(26) and cell survival and proliferation (27)(28)(29)(30)(31)(32)(33)(34)(35). Intracellular levels of S1P are tightly regulated by the equilibrium between its formation, which is catalyzed by sphingosine kinase (SK), and its degradation, which is catalyzed by S1P lyase and S1P phosphatases ( 36 ). S1P produced in response to agonists has the ability to function intracellularly as a second messenger or after secretion in an autocrine/paracrine fashion to activate S1P receptors (formerly known as endothelial differentiation gene or Edg receptors) on the cell surface ( 37 ). Although S1P is thought to mobilize [Ca 2+ ] i via interaction with its surface receptors, increasing evidence suggests an important intracellular role for S1P in mediating Ca 2+ increases within the cell ( 38,39 ). However, the exact mechanism in which S1P mediates Ca 2+ mobilization is still uncertain.

LysoPC in oxLDL is not responsible for the generation of [Ca 2+ ] i oscillations
OxLDL and one of its components, lysoPC, have both been shown to induce an increase in [Ca 2+ ] i in macrophages (41)(42)(43)(44). While 10 mol/l lysoPC was able to elicit [Ca 2+ ] i oscillations in BMDMs to some extent, a considerably lower percentage of cells was positive for [Ca 2+ ] i oscillations compared with cells treated with oxLDL [the 25 g/ ml of oxLDL used contains approximately 6 mol/L lysoPC(45)] ( Fig. 1 ). Furthermore, PC treatment elicited a response similar to that of lysoPC ( Fig. 1 ). Hence, even though PC is converted to lysoPC during the LDL oxidation reaction ( 45 ), it is unlikely that the lysoPC content in oxLDL is responsible for the observed [Ca 2+ ] i oscillations.

2+
] i can be mediated by an infl ux of Ca 2+ from the extracellular environment or from intracellular Ca 2+ stores. To assess the contribution of extracellular Ca 2+ , BMDMs were incubated in medium lacking Ca 2+ . Under these conditions, the percentage of cells showing Ca 2+ oscillations in response to oxLDL was reduced to less than half the level observed in cells incubated in media containing Ca 2+ ( Fig. 2 ). This indicates that while the presence of extracellular Ca 2+ is required for the full effect of oxLDL, release from intracellular stores accounts for much of the observed [Ca 2+ ] i oscillations.

2+
] i oscillations During the course of a Ca 2+ transient, the release of calcium from stores is followed by reuptake via a number of pumps and exchangers that remove Ca 2+ from the cytoplasm. SERCA is one of the pumps that returns Ca 2+ from the cytoplasm to the ER ( 46 ). Thapsigargin, an epoxide derivative that selectively prevents Ca 2+ binding to SERCA least 2 times that of background, followed by a decrease in fl uorescence and another increase in fl uorescence, were scored as positive for calcium oscillations. Each condition was performed in duplicate within the experiment, and data shown are representative of at least three independent experiments.

Cell viability assay
BMDMs were seeded in 96-well plates at 5.0 × 10 4 cells per cm 2 and grown for 24 h. Cells were washed and incubated with medium with or without compounds as indicated for 24 h. MTS/ PMS solution was then added to each well to a fi nal concentration of 333 µg/ml MTS and 25 µmol/l PMS. After incubation for 2 h at 37°C, the absorbance at 490 nm was recorded using a Molecular Devices VersaMax microplate reader. Correlation between macrophage number and formation of formazan product has been previously established ( 11 ). Each condition was performed in triplicate within the experiment, and data are representative of at least three independent experiments.

Apoptosis assay
BMDM were seeded in 6-well plates at 5.0 × 10 4 cells per cm 2 and grown for 24 h. Cells were then washed and incubated with medium containing compounds as indicated for 24 h. Cells were harvested using a rubber cell scraper and fi xed in 70% cold ethanol for 30 min. Cells were then washed with DPBS and stained with 3 mol/l propidium iodide in DPBS containing 0.1% Triton X-100 and 0.73 mol/l RNase A. DNA content was analyzed by fl ow cytometry on the FL-3 channel with appropriate gating used to exclude debris and cellular aggregates. Ten thousand events were counted for analysis. Each condition was performed in triplicate within the experiment and is representative of at least three independent experiments.    ( 51 ). Inhibition of RyR-mediated Ca 2+ release with this drug also did not block oxLDL's prosurvival effect ( Fig. 4B ).

S1P generates [Ca 2+ ] i oscillations and promotes macrophage survival
S1P can act as a second messenger to induce Ca 2+ mobilization within the cell ( 38,39 ). We demonstrated that S1P can generate [Ca 2+ ] i oscillations in BMDMs within the same time frame and as effectively as oxLDL ( Fig. 5 ). Furthermore, the same concentration of S1P promoted BMDM survival ( Fig. 6A ) and blocked apoptosis ( Fig. 6B ) to the same extent as oxLDL. S1P arises from the phosphorylation of sphingosine by SK ( 36 ). It has been previously reported that oxLDL can activate SK in SMCs ( 52,53 ), suggesting that the induction of [Ca 2+ ] i oscillations by oxLDL in BMDM might be mediated by its ability to increase phosphorylation of sphingosine.

SK is activated in response to oxLDL
To determine if oxLDL can activate SK in BMDMs, we used an in vitro kinase assay to measure the ability of cell (47)(48)(49), completely blocked oxLDL-generated [Ca 2+ ] i oscillations ( Fig. 3 ). This suggests that SERCA is responsible for Ca 2+ reuptake to produce [Ca 2+ ] i oscillations.

Inhibition of phospholipase C or ryanodine receptor does not block oxLDL-mediated macrophage survival
Two well-studied mechanisms of Ca 2+ release from intracellular stores involve inositol-1,4,5-triphosphate receptors (IP 3 Rs) and ryanodine receptors (RyRs). Activation of phospholipase C (PLC) results in the conversion of phosphatidylinositol-4,5-bisphosphate to diacylglycerol and IP 3 . The IP 3 stimulates IP 3 R-mediated Ca 2+ release from the ER. U-73122 is a selective inhibitor of PLC in this pathway (IC 50 ‫ف‬ 3 µmol/L) ( 50 ). Inhibition of PLC by U-73122 did not selectively inhibit oxLDL's prosurvival effect ( Fig. 4A ), in that it also decreased survival in control cells incubated without ox-LDL. Dantrolene inhibits Ca 2+ release from RyR channels   signifi cant amounts of S1P ( 55 ), does not elicit a Ca 2+ response similar to S1P, suggesting that endogenous production, perhaps in the plasma membrane, may be required to induce calcium oscillation ( 56 ). Our results do not exclude a role for other components of oxLDL, such as oxysterols, in stimulating intracellular [Ca 2+ ] i oscillations. However, the fact that the effect of oxLDL was mimicked by exogenous S1P and inhibited by an SK inhibitor suggests that S1P plays a major role. In addition, a recent study in U937 macrophages found that increased intracellular Ca 2+ in response to 7 ␤ -hydroxycholesterol was mediated by infl ux of extracellular Ca 2+ and was nonoscillatory ( 57 ). Both of these features are different from our fi ndings with oxLDL-induced Ca 2+ signaling in BMDM.
The exact mechanism in which S1P mediates Ca 2+ mobilization is still uncertain. Ca 2+ release mediated by S1P occurs independently of IP 3 Rs and RyRs ( 58 ). One possible candidate is SCaMPER (for sphingolipid Ca 2+ release mediating protein of the ER) ( 58 ). SCaMPER is a 181 amino acid protein that was fi rst shown to mediate sphingolipidgated Ca 2+ release from intracellular stores in Xenopus laevis oocytes. More recently, antisense knockdown of SCaMPER mRNA was shown to substantially reduce sphingolipid-induced calcium release in human and rat cardiomyocytes ( 59 ). However, SCaMPER shares no similarity to lysates to phosphorylate sphingosine. There was a 1.5-fold increase in SK activity almost immediately after the addition of oxLDL ( Fig. 7 ). This rapid activation of SK lends plausibility to the suggestion it could be a mechanism for mediating the [Ca 2+ ] i oscillations observed in response to oxLDL.

2+ ] i oscillation and macrophage survival
To test if SK activation is required for the ability of ox-LDL to induce [Ca 2+ ] i oscillation, we used a selective inhibitor of SK ( 54 ). Figure 7 shows that this compound effectively blocks SK activation by oxLDL. It also completely blocked oxLDL-generated [Ca 2+ ] i oscillations ( Fig. 8 ) and oxLDL-mediated macrophage survival ( Fig. 9 ). These results strongly suggest that oxLDL-induced Ca 2+ mobilization is mediated by increased generation of S1P via SK activation.

DISCUSSION
A number of groups have reported that oxLDL induces an increase in [Ca 2+ ] i (41)(42)(43)(44), and our studies extend this observation by demonstrating that at least in macrophages, this is an oscillatory increase. These oscillations involved Ca 2+ release from intracellular stores and required SERCA to return cytosolic Ca 2+ to the ER. S1P is known to induce intracellular calcium release in other cells ( 37 ), and we found it to be as effective as oxLDL at inducing [Ca 2+ ] i oscillations in BMDM. The addition of oxLDL resulted in an almost immediate activation of SK, which is the major cellular pathway to production of S1P.
Inhibition of SK activation blocked not only oxLDLgenerated [Ca 2+ ] i oscillations but also oxLDL-mediated macrophage survival. This links Ca 2+ signaling with the prosurvival effects of oxLDL. Delivery of S1P by oxLDL itself is unlikely because S1P is lost during the oxidation process ( 55 ). Furthermore, native LDL, which contains  ulated by p38 MAPK. Together, these results indicated that oxLDL can positively regulate eEF2 kinase activity by both 1 ) generating an oscillatory increase in [Ca 2+ ] i and 2 ) inhibiting its negative regulation by p38 MAPK. The only known substrate of eEF2K is eEF2, a monomeric GTPase that regulates peptide chain elongation. Phosphorylation of eEF2 inhibits its activity, thereby reducing the rate of protein synthesis. In keeping with its ability to activate eEF2K, addition of oxLDL results in a decrease in overall protein synthesis in BMDMs ( 16 ). Paradoxically, this effect of eEF2K activation has been shown to result in increased viability of cells under conditions of metabolic stress (e.g., growth factor withdrawal).
The results described herein extend our previous observations and suggest a model in which oxLDL activates SK, triggering S1P-mediated oscillatory Ca 2+ release from intracellular stores, which in turn leads to activation of eEF2K and energy conservation via inhibition of protein synthesis, culminating in increased macrophage survival.
any other known [Ca 2+ ] i channels and is a small protein with only one transmembrane domain ( 58 ). Thus, it is unlikely to itself be a Ca 2+ channel. Furthermore, a study showed that there is little correlation between its intracellular location and that of known [Ca 2+ ] i stores ( 60 ).
We recently reported that oxLDL-mediated [Ca 2+ ] i oscillations lead to activation of the Ca 2+ /calmodulin-dependent kinase, eEF2 kinase ( 16 ). Both the increase in [Ca 2+ ] i oscillations and the activation of eEF2 kinase were blocked by BAPTA-AM, an intracellular Ca 2+ chelator. Addition of oxLDL also resulted in the phosphorylation of eEF2, the only known substrate of eEF2 kinase. The eEF2 kinase selective inhibitors TS-4 and TX-1918 blocked the ability of oxLDL to promote survival of BMDMs. Withdrawal of M-CSF resulted in the activation of p38 mitogen-activated protein kinase (MAPK), an effect that is blocked with the addition of oxLDL, and eEF2 kinase can be negatively reg-