Tumor Necrosis Factor–Related Apoptosis-Inducing Ligand (TRAIL) Sequentially Upregulates Nitric Oxide and Prostanoid Production in Primary Human Endothelial Cells

Recombinant Histidine6-tagged TRAIL was produced in bacteria as described.¹⁴ Primary HUVECs, obtained as described previously,¹⁵ were used between the 3rd and 6th passage in vitro. Cells were grown on gelatin-coated tissue culture plates in M199 endothelial growth medium (BioWhittaker) supplemented with 20% FBS, 10 μg/mL heparin, and 50 μg/mL ECGF (Sigma). In some experiments, subconfluent HUVECs were starved in M199 containing 1% FBS for 18 hours before exposure to cytokines. The optimal concentration for both TRAIL (100 ng/mL) and TNF-α (50 ng/mL) was determined in preliminary experiments in which NO and/or prostanoid production was measured after exposure to serial dilutions (0.1 to 5000 ng/mL) of the cytokines. For apoptosis evaluation, cells were analyzed after propidium iodide staining.¹⁴

For RT-PCR analysis of TRAIL receptors, RNA was purified from HUVECs, using the SV total RNA isolation system (Promega). Synthesis of first strand cDNA and amplification were performed using the Access RT-PCR system (Promega) and specific primer sets, following the manufacturer’s protocol.

For flow cytometric analysis of surface TRAIL receptors, the following antibodies were used: polyclonal goat anti-human TRAIL-R1, TRAIL-R2, TRAIL-R3, and TRAIL-R4 (all from R&D System); PE-conjugated rabbit anti-goat Abs (Immunotech; Marseille, France). Flow cytometric analysis was performed by FACScan (Becton Dickinson).

Subconfluent HUVECs were harvested in lysis buffer containing 1%Triton X-100, Pefablock (1 mmol/L), aprotinin (10 μg/mL), pepstatin (1 μg/mL), leupeptin (10 μg/mL), NaF (10 mmol/L), and Na₃VO₄ (1 mmol/L). Equal amounts of protein (50 μg) for each sample were migrated in acrylamide gels, blotted onto nitrocellulose filters, and probed with the following antibodies: anti-phospho-eNOS (P-Ser1177, Cell Signaling Technology), anti-eNOS/NOS Type III (BD Transduction Laboratories), anti-tubulin antibody (Sigma), anti-poly (ADP-ribose) polymerase (PARP), anti-caspase 3, anti-IkBα, and IkBε antibodies (all from Santa Cruz Biotechnology); and anti-COX-1, -COX-2 (Cayman).

For confocal measurement of intracellular NO production and Ca²⁺ flux, HUVECs were plated on glass coverslips, grown to subconfluence and assays were performed as previously described.R16-127474 ^16,17

NOS activity was assayed by measuring the ability of cell lysates to convert l-³H-arginine (185×10³ Bq, 5 μCi Amersham) into l-³H-citrulline, as previously described.R16-127474 ^16,17

For NO-dependent cGMP measurement, HUVECs were seeded in standard 96-well plates, incubated overnight at standard conditions and subsequently treated, as indicated, for 30 minutes at 37°C in culture medium containing 0.6 mmol/L IBMX. After cell lysis, cGMP levels were measured using an enzyme-immunoassay kit (Amersham) according to the manufacturer’s instruction.

When indicated, pharmacological inhibitors or the vehicle (0.25% DMSO) were added to the cells 45 minutes before TRAIL.

Cell migration was assayed using a modified Boyden chamber assay as described previously.¹⁸

For labeling of actin cytoskeleton, HUVECs were seeded on gelatin-coated coverslips and maintained in serum-free media overnight. Fresh media was administered to the cells 2 hours before stimulation with either TRAIL or VEGF for 20 minutes. Cell membranes were permeabilized with 0.1% Triton X-100/PBS and labeling of filamentous actin was attained by staining with Texas Red-X phalloidin (Molecular Probes).¹⁸ Images were captured using a digital fluorescence microscope.

PGE₂, 6-keto-PGF_1α, (a PGI₂ metabolite), and TXB₂ (a TXA₂ metabolite) were detected in the supernatants of cell cultures using previously validated radioimmunoassays.¹⁹

Data were analyzed using the two-tailed, two-sample t test (Minitab, statistical analysis software, State College). Values of P<0.05 were considered significant.

An expanded Materials and Methods section can be found in the online data supplement available at http://www.circresaha.org.

HUVECs represent a valuable model in which to assess the biological activity of TRAIL, because they express all TRAIL transmembrane receptors (TRAIL-R1, TRAIL-R2, TRAIL-R3, and TRAIL-R4) both at the mRNA and surface protein level (Figures 1A and 1B). TRAIL did not induce apoptosis (constantly <5% over background levels) in HUVECs, even when used at high concentrations (up to 5 μg/mL; data not shown), and it was unable to activate the caspase pathway and/or to induce PARP cleavage (Figure 1C). These findings are similar to those reported by other authors.R5-127474 ^5,6

In order to investigate whether TRAIL was able to modulate the NOS/NO pathway in endothelial cells, quiescent HUVECs were exposed to recombinant TRAIL and analyzed for NOS activation and NO production. Of the different isoforms of NOS, HUVECs constitutively express endothelial NOS (type III NOS or eNOS) (Figure 2A). Of note, whereas vascular endothelial cells are capable of generating large amounts of NO through the inducible NOS (type II or iNOS) pathway after stimulation with inflammatory cytokines,⁹ HUVECs are a notable exception, because they are unable to express iNOS even when exposed to a mixture of inflammatory cytokines, such as TNF-α and IL-1.²⁰ When quiescent HUVECs were exposed to recombinant TRAIL, a significant increase in phosphorylation of eNOS at Ser¹¹⁷⁷ was noticed in Western blots, performed with a mAb directed against the phosphorylation site of eNOS (Figure 2A). The increase of eNOS phosphorylation by TRAIL started to be statistically significant (P<0.05) at 5 minutes, peaked at 15 to 30 minutes, and declined thereafter (Figures 2A and 2B). Similar/overlapping results were obtained in experiments performed in the presence of polymyxin B sulfate, used to exclude the possibility that LPS potentially contaminating the TRAIL preparations could confound the results. Moreover, neither His-control peptide nor LPS (used up to 10 μg/mL) affected eNOS phosphorylation and NO production activity, demonstrating that stimulation of NO was specifically due to TRAIL (data not shown).

In parallel experiments, the NOS activity was assessed in HUVEC lysates (Figure 2C). As expected on the basis of the Western blot data illustrated above, TRAIL treatment induced a dose-dependent increase in NOS activity. It should be noticed that the peak of NOS stimulation by TRAIL (100 to 1000 ng/mL) was similar to that induced by insulin (10 nmol/L) (Figure 2C), and comparable to that reported by other authors,R21-127474 R22-127474 R23-127474 ^21–24 in response of potent NOS agonists, such as vascular endothelial growth factor (VEGF), angiotensin II, and fluid shear stress (FSS). The increase of NOS activity induced by TRAIL started to be statistically significant (P<0.05) over background levels of untreated controls at concentrations as low as 10 ng/mL. This concentration is in the range of those reported in the plasma of several groups of patients.²⁵ It should also be noticed that the local concentrations of TRAIL in the intima microenvironment are expected to be significantly higher than the plasmatic concentrations, taking into account that the major source of TRAIL in the vessel wall is represented by medial smooth cell layer.⁷

We next investigated whether eNOS-expressing cells were able to generate bioactive NO. For this purpose, we first measured the formation of cGMP, a good proxy for NO, because soluble guanylate cyclase is activated by nanomolar concentrations of the gas.¹⁷ Exposure to TRAIL resulted in a significant (P<0.01) increase in cGMP over controls (68±2 and 28±3 fmol/10⁶ cells, respectively; n=4), which was inhibited by the presence of L-NAME (34±2 and 30±2 pmol/mg min⁻¹, respectively; n=4), a competitive inhibitor of NOS. In parallel, NO production was also measured by loading HUVECs with the NO-specific fluorescence dye DAF-2 DA, a cell-permeable compound that is converted to DAF-2 by intracellular esterases. In the presence of NO, DAF-2 forms a triazole derivative that emits light at 515 nm on excitation at 489 nm, in proportion to the amount of NO present. When these cultures were excited with light at 489 nm, TRAIL caused an asynchronous increase in fluorescence at 515 nm, indicative of a rise in intracellular NO in approximately 40% of the cells examined, which was completely abrogated by pretreatment of the cells with L-NAME (Figures 3A and 2B), but not by pretreatment of the cells with the calcium chelator BAPTA (Figure 3B). Moreover, calcium-imaging studies performed after exposure to TRAIL clearly indicated that TRAIL failed to induce intracellular Ca²⁺ flux in HUVECs (Figure 3C).

In order to further investigate the intracellular pathway(s) involved in eNOS activation by TRAIL, HUVECs were pretreated with LY294002, a selective pharmacological inhibitor of the PI 3 kinase/Akt pathway. In agreement with other authors’ studies reporting eNOS Ser¹¹⁷⁷ phosphorylation by Akt,R21-127474 R22-127474 R23-127474 ^21–24 LY294002 completely abrogated the TRAIL-induced NO production (data not shown) and the NOS activity, at the same extent of L-NAME (Figure 3D). Taken together, these data indicate that eNOS is activated by TRAIL through a PI 3 kinase/Akt pathway and is independent of the induction of intracellular Ca²⁺ fluxes.

The apparent time-course discrepancy between NO production investigated by DAF-2 dye staining plus confocal microscopy analysis and eNOS phosphorylation, as evaluated by Western blot, can be readily explained by taking into account that Figures 3A and 3B show cells that are immediately activated by TRAIL, whereas data shown in Figure 2 are derived from the bulk HUVEC population, in which the number of responsive cells progressively increases with time.

NO has been proposed to modulate cell migration and to be essential for podokinesis.⁹ We thus investigated whether the activation of NO induced by TRAIL was sufficient to drive endothelial cell migration by measuring the transfilter migration of cells seeded on a membrane separating the lower and upper part of a 6.5-mm Transwell. TRAIL dose-dependently (P<0.05, starting at 1 to 10 ng/mL) promoted endothelial cell migration (Figure 4A). Accordingly, with the notion that cell migration is tightly associated with formation of stress fibers,²² we found that TRAIL, similarly to VEGF, induced profound cytoskeletal reorganization characterized by the formation of transcytoplasmic stress fibers (Figure 4B).

The TRAIL-induced increase in cell migration and the actin reorganization into stress fibers were mediated by NOS activity, as indicated by the inhibition after preincubation with L-NAME (Figures 4A and 4B).

A complex interplay between the NOS and COX pathways has been demonstrated.R26-127474 ^26,27 Therefore, also taking into consideration the key role of prostanoids in vascular biology,¹³ in the next group of experiments we have analyzed the release of prostanoids in HUVEC supernatants collected after treatment with either TRAIL or TNF-α, used as positive control.²⁸ As expected, at 24 hours, TNF-α induced a several fold increase in all prostanoids examined (PGE₂>PGI₂>TXA₂; Table). Of note, TRAIL also induced a significant (P<0.01) increase in prostanoid production (PGE₂=PGI₂>TXA₂) over basal levels detected in unstimulated cells, although lower that that observed in cultures treated with TNF-α (Table). In time-course experiments, the peak induction of prostanoids by TRAIL was reached after 4 to 6 hours of treatment, showing a plateau thereafter. On the other hand, TNF-α induced maximal prostanoid production after 24 hours (data not shown). In parallel experiments, primary HUVECs were treated with TRAIL or TNF-α for 24 hours, washed, and prostanoid production was measured after incubation of equal numbers of viable cells (10⁶) with 10 μmol/L of exogenous AA. This type of examination ensures a constant substrate concentration. The results obtained under these conditions were similar to those obtained with measurement of spontaneous prostanoid release (Table). It is also noteworthy that TRAIL induced an approximately 2-fold induction of both PGI₂, which represent the key prostanoid controlling vascular tone, and PGE₂, which plays a major role in inflammation and vascular permeability.¹³ On the other hand, TNF-α was significantly more efficient in inducing the production of PGE₂.

In the following experiments, we examined whether TRAIL was also able to modulate the expression of COX-1 and COX-2 isoenzymes, the rate limiting enzymes involved in the production of different classes of prostanoids.¹³ In Western blot analysis, unstimulated HUVECs showed a strong constitutive expression of COX-1 and a less intense, but detectable, expression of COX-2 (Figure 5). As expected,²⁸ TNF-α strongly upregulated COX-2 protein from 6 hours onwards without affecting the constitutive COX-1 expression; on the other hand TRAIL did not affect either COX-1 or COX-2 expression (Figure 5).

In parallel experiments, we investigated the ability of TRAIL and TNF-α to modulate NF-κB transcription factor, taking advantage of the fact that NF-κB activation is normally prevented by the inhibitory family of IkB proteins.²⁹ Consistent with the requirement of NF-κB for the transcriptional upregulation of COX-2 expression,R27-127474 ^27,28 TNF-α induced IkBα degradation, followed by NF-κB–driven increase in protein due to resynthesis of the inhibitor (Figure 6). IkBε degradation occurred over much the same time course as IkBα. On the other hand, TRAIL did not induce any degradation of IkBα and IkBε, nor was there any effect on the synthesis of IkBα or IkBε (Figure 6).

We next investigated the inhibition of prostanoid production induced by treatment with either TRAIL or TNF-α by preincubating HUVECs with serial doses of the selective COX-1 and COX-2 pharmacological inhibitors. The nonselective COX-inhibitor indomethacin suppressed basal, TRAIL-, and TNF-α–mediated PGE₂ production (Figures 7A and 7B). On the other hand, in TRAIL-treated cultures, SC-560 (a COX-1 inhibitor) dose-dependently suppressed the production of PGE₂ to a significantly (P<0.05) higher extent than equimolar doses of NS-398 (a COX-2 inhibitor) (Figure 7A). In sharp contrast, in TNF-α–treated cultures, NS-398 was significantly (P<0.01) more potent than SC-560 in suppressing PGE₂ production (Figure 7B). These data clearly indicate that COX-1 and COX-2 play predominant roles in the production of PGE₂ by TRAIL and TNF-α, respectively. To evaluate the possible interplay of the NOS and COX signaling pathways, the effect of the NOS inhibitor L-NAME was also tested on prostanoid production. L-NAME showed minimal effects on basal PGE₂ production, whereas it induced opposite effects on TRAIL- (Figure 7A) and TNF-α– (Figure 7B) treated cultures. In fact, it showed a partial but significant (P<0.01) reduction in the TRAIL-mediated increase of PGE₂ production, while it strongly enhanced (<0.01) TNF-α–mediated increase of PGE₂ production.