Lactosylceramide Mediates Tumor Necrosis Factor- a -induced Intercellular Adhesion Molecule-1 (ICAM-1) Expression and the Adhesion of Neutrophil in Human Umbilical Vein Endothelial Cells*

The endothelial expression of adhesion molecules by proinflammatory cytokines such as tumor necrosis factor- a (TNF- a ) has been suggested to contribute to the initiation of atherosclerotic plaque formation. Since lactosylceramide (LacCer) accumulates in large quantities in human atherosclerotic plaque, we have explored its role in TNF- a -induced expression of intercellular adhesion molecule-1 (ICAM-1) in human umbilical vein endothelial cells and their consequent adhesion to polymorphonuclear leukocytes (PMNs). We found that TNF- a increased LacCer synthesis by way of stimulating the activity of UDP-galactose:glucosylceramide b (1 3 4)-galactosyltransferase in a time-dependent fashion. The TNF- a -induced expression of ICAM-1 was abrogated by D -1-phenyl-2-decanoylamino-3-morpholino-1-propanol ( D -PDMP), an inhibitor of UDP-galactose:glucosylceram-ide b (1 3 4)-galactosyltransferase. However, the addition of LacCer reversed the D -PDMP effect on TNF- a -induced ICAM-1 expression in human umbilical vein endothelial cells. Northern hybridization analysis of mRNA levels and enzyme-linked immunosorbent assays revealed that


The endothelial expression of adhesion molecules by proinflammatory cytokines such as tumor necrosis factor-␣ (TNF-␣) has been suggested to contribute to the initiation of atherosclerotic plaque formation. Since lactosylceramide (LacCer) accumulates in large quantities in human atherosclerotic plaque, we have explored its role in TNF-␣-induced expression of intercellular adhesion molecule-1 (ICAM-1) in human umbilical vein endothelial cells and their consequent adhesion to polymorphonuclear leukocytes (PMNs). We found that TNF-␣ increased
One of the critical events in the pathogenesis of atherosclerotic lesion formation is the focal accumulation of lipid-laden foam cells beneath an intact arterial endothelial lining (1,2). Localized attachment of circulating monocytes and lymphocytes to the arterial endothelium appears to precede the formation of early foam cell lesions (1,2). Various studies have demonstrated that the intercellular cell adhesion molecule-1 (ICAM-1), 1 E-selectin (endothelial leukocyte adhesion molecule; ELAM-1), and vascular cell adhesion molecule (VCAM-1) are all inducible on the endothelial surface following stimulation with TNF-␣ (3)(4)(5). ELAM-1 serves as a ligand for neutrophils (5), eosinophils (6), monocytes (7), and subpopulation of circulating lymphocytes (8). VCAM-1, which is induced on endothelial cells by IL-1, TNF-␣, or IL-4 (9) serves as a ligand for very late antigen-4 on lymphocytes, monocytes, and eosinophils (10). ICAM-1 is a specific ligand for lymphocyte function-associated antigen-1 (11) and Mac-1 (CD11/CD18), which is expressed on neutrophils and monocytes (12). In vivo studies have suggested that both ICAM-1 and VCAM-1, which are inducible by TNF-␣, are expressed in human atherosclerotic lesions (13). It is well documented that TNF-␣ induces the expression of cell adhesion molecules by activating an oxidantsensitive signal transduction pathway (14,15). However, to the best of our knowledge, the role of glycosphingolipids (GSLs), particularly lactosylceramide (LacCer), in this phenomenon has not been explored.
LacCer, a ubiquitous GSL, plays a pivotal role in the biosynthesis of complex glycosphingolipids (16). Moreover, the level of this GSL is elevated in several proliferative diseases including human polycystic kidney disease (17), familial hypercholesterolemia (18,19), and atherosclerosis (20,21). In addition, recently, there has been a surge in reports suggesting that sphingolipids may be implicated as second messengers in mediating diverse molecular events, including cell proliferation (22)(23)(24) and programmed cell death (apoptosis) (25), and as adhesive molecules (16). In this paper, we demonstrate that TNF-␣ stimulates the synthesis of LacCer by stimulating the activity of a UDP-galactose:glucosylceramide ␤(134)-galactosyltransferase (GalT-2). In turn, LacCer, via the generation of superoxide, up-regulates the expression of ICAM-1 on the surface of 1  Glycosphingolipids and all other chemicals were purchased from Sigma. The purity of glycosphingolipids was assessed by high pressure liquid chromatography and/or high performance thin layer chromatography to be Ͼ99%. Anti-ICAM-1 and anti-SOD antibodies were purchased from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Anti-VCAM-1 and anti-Eselectin were obtained from Pharmingen (San Diego, CA). Cell Tracker dye green and Cell Tracker dye orange were obtained from Molecular Probes, Inc. (Eugene, OR). Diphenylene iodonium (DPI) was from Calbiochem. SOD virus carrying cDNA for superoxide dismutase and control virus carrying the ␤-galactosidase (no SOD) gene was a generous gift from Dr. Ronald G. Crystal (National Institutes of Health). cDNA for ICAM-1 was a gift from Dr. Mark Stipetic (Emory University School of Medicine).
Cells-HUVECs were purchased from Clonetics. Eahy 926 cells, a hybridoma permanent endothelial cell line, derived from HUVECs and human epithelial cell line A549 (26), was a generous gift from Dr. Roger Harrison (University of Bath, United Kingdom). HUVECs were maintained in endothelial cell growth medium supplemented with bovine brain extract (27). Eahy 926 cells were maintained in RPMI 1640 medium (Life Technologies, Inc.) containing 10% fetal calf serum, penicillin (100 units/ml), and streptomycin (0.1 mg/ml).
Vehicle for Glycosphingolipids-Stock solutions of LacCer and other glycosphingolipids were prepared in Me 2 SO as described (23) and added to culture medium to achieve the desired concentrations of LacCer. The final Me 2 SO concentration exposed to cells was 0.01%. DPI stock solutions were also prepared in Me 2 SO and stored at Ϫ20°C until use. Cells incubated with 0.01% Me 2 SO served as a control. An aqueous solution of NAC was prepared in the culture medium.
Cell Surface ICAM-1, VCAM-1, and E-selectin Expression Assays-The quantitative expression of ICAM-1, VCAM-1, and E-selectin on the surface of the endothelial cell monolayers was determined by a modified ELISA in 96-well plates (28). Following incubation with antagonists and agonists, endothelial cell monolayers were fixed with 3.7% formaldehyde (pH 7.4) containing 0.1 M L-lysine monohydrochloride and 0.01 M sodium m-periodate for 20 min at 4°C and then blocked with PBS containing 1% bovine serum albumin and 0.1 M glycine overnight at 4°C. The fixed monolayer was then incubated with mouse monoclonal anti-human ICAM-1 or anti-human VCAM-1 or anti-human E-selectin for 1 h at 37°C. Next it was incubated with peroxidase-conjugated anti-mouse IgG F(abЈ) 2 for 1 h at 37°C. After washing, a developing substrate (0.2% H 2 O 2 , 0.4 mg/ml O-phenylenediamine) was added for 5 min, and the reaction was stopped with 2 N H 2 SO 4 . The plates were read on a spectrophotometric plate reader at 520 nm (28). ICAM-1 expression was assessed qualitatively in Eahy 926 cells by immunofluorescence staining using fluorescein isothiocyanate (FITC)-conjugated IgG following fixing of cells incubated with or without LacCer with 3.7% formaldehyde as described above.
Measurement of GalT-2 Activity-The activity of GalT-2 in cells incubated with TNF-␣ was measured employing UDP-[ 14 C]galactose as a nucleotide sugar donor and glucosylceramide as an acceptor as described previously (29). Briefly, the GalT-2 assay mixture contained 100 g of enzyme preparation, 20 M of cacodylate buffer (pH 6.8), 1.0 mM Mn 2ϩ /Mg 2ϩ , 0.2 mg/ml Triton X-100-cutscum (1:2, v/v), 30 nmol of GlcCer, and 0.1 mmol of UDP-galactose. Assays without exogenous GlcCer served as blanks and were subtracted from all corresponding data points. The assay was terminated by the addition of 25 mol of EDTA plus 2.5 mol of KCl. Chloroform/methanol (2:1, v/v) and 5 g of human kidney GSL were added, and the products were isolated and separated by Whatman SG-81 paper chromatography by developing in chloroform/methanol/water (60:17:2, v/v/v). Chromatogram areas corresponding in migration with standard LacCer were cut, and radioactivity was measured in a Beckman LS-3800 scintillation spectrometer using the background subtract setting and automatic quench setting.
Measurement of LacCer Synthesis-Cultures of HUVECs were incubated in growth medium containing [ 3 H]galactose (5 Ci/ml) for 24 h. Next, the medium was removed, and the monolayer was washed with sterile PBS to remove nonspecifically bound radioactivity. Fresh serumfree medium (5 ml/plate) was added. One set of dishes was incubated with D-PDMP (20 M) in PBS for 2 h. Next, TNF-␣ (300 units/ml) was added to cells incubated with or without D-PDMP. At various time points, cells were harvested and washed with PBS and centrifuged. The cell pellet was subjected to extraction of lipids with organic solvents as described (20). The total lipid extract was then subjected to silicic acid column chromatography. The mixture of glycosphingolipids was eluted with acetone-methanol (9:1, v/v), and carrier LacCer (10 g) was added and dried under N 2 atmosphere. Glycosphingolipids were separated onto high performance thin layer chromatography plates using chloroform/methanol/water (100:42:6, v/v/v) as the developing solvent. The chromatographic plate was dried in air and stained with iodine vapor. The gel area corresponding to LacCer was scraped, and radioactivity was measured employing "liquiscint" (NEN Life Science Products) as a scintillating fluid.
Northern Blot Analysis of ICAM-1 mRNA Expression in HUVECs-Total cellular RNA was isolated (30) from HUVECs that were preincubated with 15 mM NAC for 30 min, 5 M diphenylene iodonium for 30 min, followed by stimulation with the indicated concentrations of Lac-Cer/TNF-␣ for 2 h. Twenty micrograms of total RNA were separated by electrophoresis on 1% agarose gel (25 mM MOPS, pH 7.8, 1 mM EDTA, 1% (w/v) formaldehyde), transferred to a -probe blotting membrane (Bio-Rad) and hybridized with a 32 P-labeled ICAM-1 cDNA probe as described. The transfer of RNA to membrane and equal amount RNA loading was confirmed by ethidium bromide staining.
Measurement of Superoxide Generation in Intact HUVECs-Lucigenin, an acridylium compound that emits light upon interaction with O 2 . , was used to measure O 2 . production by chemiluminescence (23).
Confluently grown endothelial cells were harvested, and O 2 . generation from the intact endothelial cell suspension was measured using a 96well plate containing dark-adapted lucigenin (250 M) in balanced salt solution, as described by Bhunia et al. (23). The viability of the suspended cells was Ͼ90% as determined by trypan blue exclusion. LacCer or other GSL solutions were added as putative agonists, and the resulting increases in photon emission were measured every 20 s thereafter for 10 min in a scintillation counter (Packard TOP counter). The net increase in O 2 . generation at each time point was calculated by compar-

FIG. 1. Effects of TNF-␣ on the activity of GalT-2 and LacCer synthesis in HUVECs.
A, stimulation of GalT-2 activity by TNF-␣. Cells were incubated with TNF-␣. At various time intervals, cells were harvested in phosphate-buffered saline and lysed, and GalT-2 activity was measured as described under "Materials and Methods." E, control; q, TNF-␣ (100 units/ml); OE, TNF-␣ (300 units/ml). B, effects of TNF-␣ and D-PDMP on the incorporation of [ 3 H]galactose into lactosylceramide in cultured HUVECs. Cells were metabolically labeled with [ 3 H]galactose (5 Ci/ml) for 24 h at 37°C. Next, one set of dishes was incubated with fresh medium with and without D-PDMP (20 M) for 2 h at 37°C. Incubation was continued for the indicated time period following the addition of TNF-␣ (300 units/ml). Next, cells were harvested, total lipids were extracted with organic solvents, and glycosphingolipids were purified as described under "Materials and Methods." The radiolabeled glycosphingolipids were separated by high performance thin layer chromatography using chloroform/methanol/water (100:42:6, v/v/v) as the developing solvent. The chromatography plate was stained with iodine vapor, plate area corresponding to LacCer was scraped, and radioactivity was measured using a scintillation counter. The control specific activity was 805 Ϯ 52 cpm/mg of protein. ison with a standard curve generated using xanthine/xanthine oxidase.
Cell Fractionation and NADH/NADPH Oxidase Assay-Confluently grown HUVECs were incubated with or without 5 M LacCer. After various time intervals, the cells were harvested and homogenized, and the membrane and cytosolic fractions were separated by differential centrifugation, as described previously (23). NADH and NADPH oxidase activities were measured in both the cytosolic and the membrane fractions by lucigenin chemiluminescence as described above. In some experiments, NADPH oxidase activity was measured in the membrane fraction in the presence of 1 mM rotenone (a mitochondrial poison) (31). Protein content was measured by the method of Lowry et al. (32) with bovine serum albumin serving as a standard.
Adenovirus Vector-mediated Overexpression of SOD in Endothelial Cells-A replication-incompetent adenovirus (Ad-SOD) carrying the cDNA for human CuZn-SOD (33) and a control virus (Ad-␤-galactosidase) without the SOD gene but encoding only the Escherichia coli lacZ gene (34) were amplified in 293 cells and purified as described by Crawford et al. (33). Confluently grown HUVECs in 96-well plates (1.3 ϫ 10 4 cells/well) were infected with virus containing CuZn-SOD gene in multiplicities of infection (MOI) ranging from 0 to 75. Other plates of endothelial cells were infected with the control virus (Ad-␤galactosidase) at corresponding dilutions. Twenty-four hours later, the cells were washed with PBS to remove uninfecting virus particles. The endothelial cells were then incubated for an additional 48 h in fresh medium. The cells were then fixed with 3.7% formaldehyde in PBS, and the expression of intracellular SOD was measured by ELISA after permeabilization with 0.2% Triton X-100 in PBS for 20 min. The permeabilized endothelial cells were probed with monoclonal anti-human SOD and then with FITC-conjugated goat anti-mouse IgG. FITC fluorescence was measured on a fluorescence plate reader (CytoFluor 2300) at 480 nm (excitation) and 530 nm (emission). The SOD activities in the cell extracts were measured by the method of McCord and Fridovich (35).
Confocal Microscopic Visualization of Adhesion of Human Neutrophils to HUVECs-For the isolation of polymorphonuclear leukocytes (PMNs), human blood was collected in 10 units/ml heparin from laboratory volunteers by a protocol approved by the Johns Hopkins Joint Committee for Clinical Investigations and centrifuged at 1300 ϫ g for 10 min at 4°C. The white blood cell layer was removed and layered over cold Accu-prep gradient and centrifuged at 600 ϫ g for 30 min for the separation of leukocytes. The red blood cell/PMN layer was then resuspended in red blood cell lysing buffer. After incubation at room temperature, the preparation was centrifuged at 1300 ϫ g for 2 min, and this step was repeated until the PMN pellet was free of red blood cells. The PMN pellet was then washed and was found to contain Ͼ96% PMNs following modified Wright-Giemsa staining. The PMNs were labeled by incubation with 5 M fluorescent Cell Tracker dye (green) for 30 min at 37°C. At the same time, LacCer-stimulated (5 M, 4 h, 37°C)/control (0.01% Me 2 SO) endothelial cell monolayers grown in glass chamber slides were also labeled with 10 M Cell Tracker dye (orange). The labeled PMNs were incubated with confluently grown fluorescencelabeled endothelial cell monolayers in the glass chamber slides for 30 min at 37°C. The nonadherent PMNs were then removed by gentle washing with PBS. A drop of 4% glycerol in PBS was added, and a glass coverslip was mounted on the slide and visualized and photographed using confocal microscopy.
Quantitative Measurement of PMN-Endothelial Cell Adhesion and Blockade by ICAM-1 Antibody-Endothelial cells, grown in 24-well plates were incubated with various concentrations (0 -10 M) of LacCer for 4 h at 37°C and washed with PBS. Next, human PMNs labeled with Cell Tracker dye (green) as described above were incubated with the LacCer-stimulated endothelial cells for 30 min. Nonadherent PMNs were removed by gentle washing (three times) with PBS. The residual adherent PMNs were quantified on a fluorescence plate reader (Millipore Corp.) at 480 nm (excitation)/530 nm (emission) as described by Arai et al. (36). In a parallel experiment, following stimulation with LacCer, endothelial cells were preincubated with ICAM-1 antibody (1:2000 dilution) for 30 min before incubation with labeled PMNs as described above. The number of adherent PMNs was expressed as the number of PMNs/mm 2 of endothelial cell monolayer, based on the mean fluorescent intensity of each PMN, determined from a standard curve. To rule out the possibility of nonspecific adhesion stimulated by HU-VEC reactivity in some PMN adhesion assays, endothelial monolayers in one half of each 24-well plate were fixed first with 2% buffered formalin for 5 min and then washed three times with PBS. The endothelial cells on the other half of the plate were left unfixed but were similarly washed before use in adhesion assay.

TNF␣ Stimulates LacCer Synthesis in a Time-and Concentration-dependent Fashion in HUVECs-
We found that TNF-␣ (both 100 and 300 units/ml) stimulated GalT-2 activity in a time-dependent manner (Fig. 1A). Within 5 min, about 1.5-fold stimulation of GalT-2 activity was observed compared with the control. Thereafter, the activity of GalT-2 continued to rise in cells stimulated with TNF-␣. This phenomenon was accompanied by a time-dependent increase in the synthesis of LacCer as evidenced by an increase in the incorporation of [ 3 H]galactose into LacCer in cells stimulated with TNF-␣ (Fig. 1B). For example, 5 min after the stimulation with TNF-␣, a 1.5-fold increase in LacCer synthesis was observed and continued to increase 60 min after stimulation. Preincubation of cells with D-PDMP decreased the synthesis of LacCer ϳ0.4-fold compared with the control. D-PDMP effect was not altered significantly after stimulation of cells with TNF-␣.  Fig. 2A). Moreover, this inhibition by D-PDMP was reversed upon incubation with exogeneous LacCer but not by exogeneous ceramide or glucosylceramide (Fig. 2B), suggesting the involvement of LacCer in TNF-␣induced ICAM-1 expression.
LacCer Stimulates the Expression of ICAM-1 mRNA Levels in HUVECs-As shown in Fig. 4, LacCer exerted a concentrationdependent increase in the mRNA levels of ICAM-1 in HUVECs. A maximum stimulation of ICAM-1 mRNA level occurred following incubation of cells with 5 M LacCer for 2 h. Moreover, this phenomenon was abrogated by preincubation of cells with N-acetyl-L-cysteine, an antioxidant and a scavenger of free oxygen radicals, as well as DPI, an inhibitor of NADPH oxidase. Interestingly, the magnitude of stimulation of ICAM-1 gene transcription by LacCer was comparable with TNF-␣. . Generation in HUVECs-At various time points following stimulation with or without (5 M) LacCer, NADPH oxidase activity was measured in the membrane preparations (Fig. 6). Three-fold stimulation of NADPH oxidase activity was observed at 5 M LacCer, as compared with nonstimulated cell membrane preparations (Fig. 6A). Preincubation of LacCerstimulated/nonstimulated cell membrane preparations with DPI, a potent NADPH oxidase inhibitor (23), attenuated the LacCer induced increase in NADPH-dependent oxidase activity in the membrane preparations (Fig. 6A). In contrast, LacCer did not alter the activity of NADH oxidase activity in stimulated/nonstimulated membrane preparations (Fig. 6B). Neither NADPH oxidase nor NADH oxidase activity was observed in the LacCer-stimulated/unstimulated cell cytosol (data not shown). On the other hand, allopurinol, a specific inhibitor of xanthine oxidase (23), failed to inhibit LacCer-induced O 2 . generation (data not shown). Since rotenone also failed to inhibit NADPH oxidase activity in the membrane fraction, it appears that plasma membrane-associated NADPH oxidase is the source of LacCer-induced O 2 . generation in endothelial cells. . as well as ICAM-1 expression (Fig. 7, A and B). However, exogeneously added SOD did not inhibit the LacCer-induced O 2 . levels in endothelial cells and ICAM-1 expression (Fig. 7, A and B). These findings suggest that the NADPH oxidase-dependent endogenous generation of O 2 . is necessary to mediate LacCer-induced ICAM-1 expression.

Overexpression of Intracellular SOD Inhibits LacCer-induced ICAM-1 Expression-Since exogeneous SOD did not inhibit LacCer-induced O 2
. levels and ICAM-1 expression, we overexpressed SOD via adenovirus-mediated gene transfer. The expression of SOD in Ad-SOD-infected endothelial cells was 2.5-fold higher at 50 MOI compared with that of Ad-␤galactosidase-infected cells, as measured by immunofluorescence using monoclonal anti-SOD (Fig. 8A). The increase in SOD activity in the Ad-SOD-infected endothelial cell lysate was paralleled with an increase in MOI (Fig. 8B). SOD enzymatic activity in Ad-SOD-infected endothelial cells was also increased, dose-dependently, compared with that of the control (Ad-␤-galactosidase-infected) cells (Fig. 8B). In the Ad-␤-galactosidase-infected control cells, LacCer stimulated ICAM-1 ex- q, LacCer. D, effects of TNF-␣ and LacCer on the expression of VCAM-1 in HUVECs. Following incubation with LacCer (5 M) or TNF-␣ (300 units/ml), cells were fixed, and VCAM-1 expression was measured by ELISA using anti-human VCAM-1 antibody. E, effects of TNF-␣ and LacCer on the expression of E-selectin in HUVECs. The protocol in this experiment was similar to that described for D except that the level of E-selectin expression was measured employing anti-human E-selectin antibody. Each point is the mean Ϯ S.D. of five separate experiments. F, immunofluorescence assay of ICAM-1. Eahy 926 cells (endothelial cells) were confluently grown in glass chamber slides and incubated with or without 5 M LacCer for 8 h. Next, cells were fixed in 3.7% formaldehyde, and ICAM-1 expression was monitored by immunofluorescence using ICAM-1 monoclonal antibody followed by incubation with FITC-conjugated anti-mouse IgG and photographed employing fluorescence microscopy. Data presented here are from one of the two separate experiments with identical results. pression (Fig. 8C). However, in cells infected with Ad-SOD, LacCer-induced ICAM-1 expression was proportionally decreased with an increase in MOI and SOD expression. This finding suggests that the LacCer-induced stimulation of ICAM-1 expression was mediated by the intracellular generation of O 2 . .

LacCer Stimulates the Adhesion of Neutrophils to
HUVECs-Confocal microscopic visualization revealed that in control endothelial cells, incubated with vehicle only (0.01% Me 2 SO), neutrophil adhesion was not observed. In contrast, LacCer (5 M) markedly stimulated the adhesion of green fluorescencestained neutrophils to the orange fluorescence-stained endothelial cell monolayer (Fig. 9). Endothelial cells preincubated with NAC or NADPH oxidase inhibitor, DPI, attenuated the LacCer-induced neutrophil adhesion (Fig. 9). This qualitative observation was further substantiated by the quantitative measurement of adhesion. As shown in Fig. 10A, LacCer markedly stimulated the adhesion of PMN to endothelial cells. This phenomenon was abrogated by anti-human ICAM-1 antibody in a concentration-dependent fashion (Fig. 10B), suggesting the involvement of ICAM-1 in LacCer-induced PMN adhesion to HUVECs. DISCUSSION Significant information has been generated supporting the concept that TNF-␣ plays an important role in the adhesion of monocytes and neutrophils to the endothelial cells (3)(4)(5). This phenomenon precedes the production of superoxides, activation of NF-B, and an increase in the expression of cell adhesion molecules such as ICAM-1 in endothelial cells (14,15). This oxidant-sensitive transcriptional pathway plays an important role in vascular lesion formation (1,2). In the present study, we have extended the mechanism over previous observations by demonstrating that TNF-␣ increases the synthesis of LacCer via activating GalT-2. In turn, LacCer induced ICAM-1 expression and PMN adhesion. This may be an early process in the TNF-␣-mediated atherosclerotic lesion formation. Our current hypothesis describing how LacCer serves as a lipid second messenger in the activation of an oxidant-sensitive signal transduction pathway that leads to the adhesion of neutrophil to endothelial cells is summarized in Fig. 11 LacCer is a ubiquitous glycosphingolipid and plays a pivotal role in the biosynthesis of complex glycosphingolipids (16). Moreover, LacCer may contribute to proliferative diseases such as polycystic kidney disease, atherosclerosis (17), and familial hypercholesterolemia (18,19). The biological function of Lac-Cer was illustrated previously with two examples. First, in aortic smooth muscle cells, LacCer stimulated a redox-regulated signaling pathway via activating p21 ras GTP loading, p44 MAPK , and the expression of c-fos (23,24). Second, LacCer stimulated the expression of Mac-1(CD8/CD11) in human neutrophils. In turn, Mac-1 is recognized by ICAM-1 expressed on the surface of endothelial cells, and that leads to the adhesion of PMN (36).
Our earlier studies also revealed that in aortic smooth mus-

FIG. 5. Effects of LacCer concentration and time of incubation on the generation of superoxide in HUVECs.
Confluently grown endothelial cell monolayers were harvested and suspended in balanced salt solution. Next, the rate of generation of superoxide was measured by lucigenin chemiluminescence as described under "Materials and Methods." cle cells, oxidized low density lipoprotein stimulated GalT-2 activity and LacCer production. LacCer in turn served as a surrogate for oxidized low density lipoprotein to induce cell proliferation (37,38). In the present study with endothelial cells, a parallel observation was made in which TNF-␣ also activated GalT-2 and produced LacCer. Accordingly, we focused our studies on the upstream step in this signaling pathway. We employed D-PDMP as an inhibitor of glycosyltrans- ferases (37)(38)(39)(40) to assess whether the inhibition of LacCer synthesis could abrogate ICAM-1 expression. Indeed, as shown in Figs. 1 and 2, in endothelial cells D-PDMP not only inhibited the activity of GalT-2 and LacCer synthesis but also abrogated ICAM-1 expression. Since D-PDMP, an analog of glucosylceramide (GlcCer), also inhibits the synthesis of GlcCer from ceramide, it was not clear whether GlcCer or LacCer was required for the TNF-␣-induced expression of ICAM-1. To address this issue, the following experiment was performed. First, we preincubated cells with D-PDMP followed by stimulation with TNF-␣, GlcCer, and ceramide. LacCer was added next, and ICAM-1 expression was measured. Only LacCer bypassed the inhibitory effect of D-PDMP on ICAM-1 expression, while GlcCer or Cer did not. These observations suggest that LacCer may serve as a lipid second messenger in TNF-␣-induced ICAM-1 expression.
In agreement with previous studies, we found that TNF-␣ up-regulates the expression of VCAM-1 and E-selectin in HU-VECs. However, LacCer specifically stimulates the expression of ICAM-1 in these cells both at transcriptional and translational levels. At present, we cannot explain the biochemical basis of this observation. We can speculate, however, that TNF-␣-mediated expression of VCAM-1 or E-selectin may require additional signaling molecules other than LacCer.
Previous studies have shown that TNF-␣ induces the expression of adhesion molecules via the generation of ROS (14,15). We found that LacCer-induced ICAM-1 expression is also dependent on the generation of ROS. This tenet is supported by the following observations. First, direct measurement by lucigenin chemiluminescence showed that LacCer induces the generation of superoxide in a time-and concentration-dependent manner. We employed lucigenin chemiluminescence to assay LacCer-induced O 2 . generation, since this method detects both intracellular and extracellular O 2 . accurately even at a very low concentration (23,31). Second, antioxidants NAC and pyrrolidone dithiocarbamate (41) and overexpression of superoxide dismutase gene abrogated ICAM-1 expression. Third, DPI, an inhibitor of the NADPH oxidase-dependent ROS-generating system (23), completely abrogated LacCer-induced ICAM-1 expression both at the transcriptional and translational level. Finally, TNF-␣-induced O 2 . generation in HUVECs was abrogated by preincubation of cells with D-PDMP (data not shown). These findings suggest that TNF-␣-induced LacCer production is necessary for the generation of O 2 . , and that in turn induces ICAM-1 expression. Increased expression of ICAM-1 in atherosclerotic plaque has been previously thought to be due to increased O 2 . levels (15,16,(42)(43)(44). Moreover, endothelial cells produce higher amounts of O 2 . in hypercholesterolemia (45). Since the level of LacCer is also elevated in atherosclerotic plaque intima (20,21) and in patients with familial hypercholesterolemia (18,19), it is tempting to speculate that this GSL may serve as a lipid second messenger that may be required to stimulate O 2 . generation and ICAM-1 expression in atherosclerosis. One of the key transcription factors in ROS-regulated gene expression is NF-B (14, 46). We found that LacCer/TNF-␣ stimulated NF-B expression as measured by the binding activity of NF-B to the consensus oligonucleotide sequence spe- cific for NF-B (data not shown). Previous studies have shown that the cytosolic NF-B migrates to the nucleus and participates in nuclear events that ultimately lead to the expression of ICAM-1 (14). Clearly, further studies are required to explore in detail the LacCer-induced phenomenon described above.
The localized attachment of circulating leukocytes and monocytes to the endothelium of inflamed vessels is an important event in the initiation and progression of atherosclerotic lesion formation (47). As a response to injury, the endothelium secretes cytokines, for example, TNF-␣, IL-8, platelet activation factor, and monocyte chemotactic factor (2). Employing confocal microscopy, we found that LacCer facilitated the adhesion of PMNs to endothelial cells. These qualitative observations were substantiated by quantitative analysis of the number of PMNs adhering to the endothelial cells. Thus, LacCer may be implicated in the initiation of atherosclerotic plaque formation.
In summary, the adhesion of leukocytes to endothelial cells is a complex process, important in the pathogenesis of atherosclerosis and inflammatory phenomenon. Our findings indicate that TNF-␣ and oxidized low density lipoprotein implicated in atherosclerosis can activate GalT-2. TNF-␣-mediated activation of GalT-2 can generate LacCer. LacCer, in turn, generates O 2 . and stimulates ICAM-1 expression in HUVECs. By virtue of stimulating the expression of the ligand Mac-1 (CD11/CD18) for ICAM-1 (36), LacCer facilitates the adhesion of leukocytes to activated endothelial cells that overexpress ICAM-1. This constitutes a novel biochemical mechanism in regard to the potential role of LacCer and ROS in TNF-␣-induced initiation and progression of atherosclerosis.