Tumor necrosis factor-alpha induces adhesion molecule expression through the sphingosine kinase pathway

Abstract

The signaling pathways that couple tumor necrosis factor-alpha (TNFalpha) receptors to functional, especially inflammatory, responses have remained elusive. We report here that TNFalpha induces endothelial cell activation, as measured by the expression of adhesion protein E-selectin and vascular adhesion molecule-1, through the sphingosine kinase (SKase) signaling pathway. Treatment of human umbilical vein endothelial cells with TNFalpha resulted in a rapid SKase activation and sphingosine 1-phosphate (S1P) generation. S1P, but not ceramide or sphingosine, was a potent dose-dependent stimulator of adhesion protein expression. S1P was able to mimic the effect of TNFalpha on endothelial cells leading to extracellular signal-regulated kinases and NF-kappaB activation, whereas ceramide or sphingosine was not. Furthermore, N, N-dimethylsphingosine, an inhibitor of SKase, profoundly inhibited TNFalpha-induced extracellular signal-regulated kinases and NF-kappaB activation and adhesion protein expression. Thus we demonstrate that the SKase pathway through the generation of S1P is critically involved in mediating TNFalpha-induced endothelial cell activation.

Figure 1

TNFα-induced sphingomyelin turnover in HUVEC. (A) HUVEC were labeled with [³H]serine for 48 h and treated with TNFα (100 units/ml), total cellular lipids were extracted, and [³H]sphingomyelin was then resolved by TLC at the desired time point. (B) The unlabeled cells were treated with TNFα as indicated above, and cells were lysed to measure ceramide levels by using the diacylglycerol kinase assay. The results represent mean values ± SD from three independent experiments. ∗, P < 0.01; †, P < 0.001, vs. the basal levels.

Figure 2

Effects of C₂-ceramide and S1P on adhesion molecules expression. (A) HUVEC were treated with an increasing concentration of C₂-ceramide or S1P for 4 h. The cell-surface expression of E-selectin or VCAM-1 was measured by flow cytometry. The data are expressed as percent of TNFα (100 units/ml)-stimulation in the mean fluorescence intensity (M.F.I.). (B) The endothelial cells were treated with vehicle (Nil), S1P (5 μM), C₂-ceramide (10 μM), DMS (5 μM), sphingomyelinase (SMase, 1 unit/ml), d-erythro-(N-myristoylamine)-1-phenyl-1-propanol (D-MAPP, 5 μM), and/or TNFα (TNF, 100 units/ml), respectively, for 4 h, then the cell-surface E-selectin or VCAM-1 was measured. (C) Flow cytometry profiles showed the effect of S1P on expression of E-selectin (Left) and VCAM-1 (Right). Asterisks indicate a negative control profile with the isotype-matched nonrelevant antibody. (D) After the indicated treatment for 4 h, E-selectin mRNA levels were measured by Northern blot assay with α-³²P-labeled cDNA probes (12). Bar graph (Bottom) depicts relative levels of E-selectin mRNA quantified by the PhosphorImager and normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) mRNA. Values in A, B, and D represent mean ± SD from at least three independent experiments. ∗, P < 0.01, compared with TNFα stimulation.

Figure 3

TNFα-induced SKase activation. (A) HUVEC treated with TNFα (TNF, 100 units/ml) at the desired time points, the cytosolic fractions were extracted to measure SKase activity. (Inset) Kinetic study of SKase. The cell extract was prepared from the cells treated with 100 units/ml TNFα for 5 min. The kinase assay was performed with various concentrations of sphingosine in the absence (•) or presence of 5 μM (○) or 10 μM (□) DMS. (B) After treatment with TNFα as described above, the cells were permeabilized to measure the production of S1P in vivo. (Inset) S1P levels in intact cells measured by labeling with [³H]serine. The data in A and B are mean values ± SD of three individual experiments. ∗, P < 0.01; †, P < 0.001, vs. the basal levels; ‡, P < 0.001, vs. TNFα stimulation.

Figure 4

Effect of SKase activity on ERK and JNK activation. (A) The cells were treated with the indicated agents for 30 min, and ERK activities were then assayed with myelin basic protein (MBP) as substrate after immunoprecipitation with antibodies against p42/p44^ERK. The kinase reaction products were separated on SDS/10% PAGE. In parallel, an aliquot of the same cell lysates was blotted with anti-p42/p44^ERK antibodies to ensure equal ERK expression. Bar graph (Bottom) depicts ERK activity quantified by the PhosphorImager. Values represent mean ± SD in percent of control from three independent experiments. ∗, P < 0.01, compared with control; †, P < 0.02, vs. TNFα (TNF) stimulation. (B) JNK activity was measured with glutathione S-transferase-jun-(1–79) as substrate after the treatment as described above. Bar graph (Bottom) shows JNK activity quantified by the radioactivity incorporated into glutathione S-transferase (GST)-jun-(1–79). Data represent mean of two separate experiments.

Figure 5

Effect of SKase activity on NF-κB activation. NF-κB (A) and Oct-1 (B) binding activity were measured by electrophoretic mobility-shift assay after 30-min treatment with a vehicle (lane 1), TNFα (TNF, 100 units/ml, lane 2), DMS (5 μM) plus TNFα (lane 3), and S1P (5 μM, lane 4), respectively. (C) The specific NF-κB binding complexes were identified by the supershift gel assay with anti-p50 and anti-p65 antibodies and by competition analyses with the addition of a 50-fold molar excess of unlabeled NF-κB oligonucleotides. (D) Bar graph depicts relative binding activity of NF-κB quantified by the PhosphorImager from A. Values represent mean ± SD in percent of control from three independent experiments. ∗, P < 0.01, compared with control; †, P < 0.01, vs. TNFα stimulation.