Thromboxane A₂ Is a Mediator of Cyclooxygenase-2-dependent Endothelial Migration and Angiogenesis

Introduction

Recent evidence suggests that COX-2 3 metabolic products contribute to neovascularization and may support vasculature-dependent solid tumor growth and metastasis. Selective COX-2 inhibitors are antiangiogenic (1) , and COX-2-null mice are substantially protected in a genetic model of human familial adenomatous polyposis (2) . Forced COX-2 overexpression enhances the metastatic potential of CaCo-2 colon carcinoma cells through processes that are sensitive to COX-2 inhibitors (3) . Coculture of endothelial cells with tumor cells promotes COX-2-dependent endothelial motility and assembly into capillary-like structures (4) , an effect that is attributed to tumor cell release of angiogenic peptides and nitric oxide. Alternatively, eicosanoids synthesized by endothelial COX-2 may contribute to this effect.

COX-2 expression or function is induced in cultured endothelial cells in response to phorbol esters (5 , 6) , basic FGF (7) , hypoxia (8) , cyclic strain (9) , thrombin, interleukin 1α (10) , or interleukin 1β (11) . Hypoxia (12) or lipopolysaccharide administration (13) induce microvascular endothelial COX-2 expression in situ. Moreover, COX-2 inhibitors have been shown to decrease urinary excretion of prostacyclin, a major product of vascular endothelium in human subjects (14) . These findings motivated our efforts to identify a COX-2 product or products that are capable of functioning as intermediaries of angiogenesis.

Materials and Methods

Eicosanoids and Quantitation.

All of the eicosanoids and eicosanoid agonists were purchased from Cayman Chemical, (Ann Arbor, MI). Eicosanoids were quantified by gas chromatographic negative ion chemical ionization mass spectrometric assays using the precise and accurate stable isotope dilution technique (15) .

Endothelial Migration.

Confluent human renal microvascular endothelial cells were grown to confluency and serum-depleted in medium containing 1% (w/v) bovine albumin for 18 h prior to assay (16) . Triplicate circular “wounds” (600–900 μm in diameter) were generated in confluent endothelial monolayers within a single well, using a rotating silicon-tipped drill bit mounted on a drill press, to avoid scoring subjacent surfaces. Medium was supplemented at the time of wounding with test agents at concentrations indicated in the figures. Residual fractional wound areas were measured using a Bioquant (Nashville, TN) software package calibrated to a Nikon Diaphot microscope. Mean fractional residual areas of three wounds, calculated at each of two or three time points (see Fig. 1b ⇓ ), were used to derive linear regressions, reflecting migration rates (expressed as percentage closure per h ± 95% confidence intervals).

Mouse Corneal Angiogenesis Assay.

Hydron pellets incorporating sucralfate with vehicle alone, basic FGF (a kind gift from Scios, Inc.), or bFGF in combination with other agents indicated in the figure legends were surgically implanted into corneal stromal micropockets, created 1 mm medial to the lateral corneal limbus of C57BL mice (7–10 weeks old), as described previously (17) . On day 5, corneas were photographed at an incipient angle of 35–50° from the polar axis in the meridian containing the pellet, using a Zeiss slit lamp. Images were digitalized and processed by subtractive color filters (Adobe Photoshop Version 4.0), as displayed (see Figs. 3 ⇓ and 4 ⇓ ). Images were analyzed using Bioquant image analysis software to determine the fraction of the two-dimensional total corneal image that was vascularized and the fraction of pixels within that area (regional density) or within the corneal perimeter (total density) that exceeded a threshold matching visible capillaries.

The dose of VU08 used (5 mg/kg) was selected based on anti-inflammatory responses in a carrageenan foot pad assay and on effects of this dose to suppress TXA₂ production by endogenous COX activity in prostate tumor tissue issue (85% inhibition at 3 days daily i.p. injection). Concentrations of SQ29548 were selected based on a dose-response experiment optimizing for inhibition of FGF-induced angiogenesis (data not shown).

Results and Discussion

Fig. 1a ⇓ provides the profile of eicosanoids produced by cultured microvascular endothelial cells under COX-2-induced conditions, following stimulation by PMA. Prostaglandin E₂, TXA₂ (measured indirectly as its thromboxane B₂ metabolite), and prostaglandin F-2α were the dominant PMA-induced products, and induced endothelial production of each was blocked by coincident exposure to a COX-2-selective inhibitor. Notably, basal endothelial capacity to produce these metabolites was maintained in the presence of COX-2 inhibition, consistent with production by constitutive endothelial COX-1.

To assess functional consequences of endothelial COX-2 inhibition, we evaluated endothelial motility. The rates at which endothelial cell migration closed replicate circular wounds in confluent monolayers were determined by quantitating residual wound areas in digital images that were captured at multiple points during a 12-h time course. PMA reproducibly stimulated the rate of endothelial migration over that of untreated cells (Fig. 1b) ⇓ , and the PMA-induced migration was blocked by two different COX-2-selective inhibitors, NS398 and VU08.

The effect of COX-2 inhibition to reduce endothelial production of eicosanoids, coupled with its effect to inhibit endothelial migration, led us to ask whether supplementation with specific eicosanoids would reconstitute PMA-induced migration in the presence of COX-2 inhibition. Shown in Fig. 2a ⇓ , the TXA₂ mimetic, U46619, reconstituted a near full migratory response to PMA under COX-2-inhibited conditions. Because endothelial cells produce TXA₂ (Ref. 18 ; Fig. 1a ⇓ ) and express functional TXA₂ receptors (19) , we evaluated further the effects of U46619 and the thromboxane receptor antagonist, SQ29548.

Shown in Fig. 2b ⇓ , U46619 reconstituted PMA induced migration responses in the presence of COX-2 inhibition, with an ED₅₀ in the range of 0.1–1.0 μm, consistent with its affinity for thromboxane receptors (20) . Importantly, Fig. 2c ⇓ shows the TXA₂ receptor antagonist, SQ29548, blocked PMA-induced endothelial migration, with an IC₅₀ of ∼0.1 μm, also consistent with its reported affinity for thromboxane receptors (20) . U46619 alone did not stimulate migration rates above basal levels in the absence of PMA (data not shown), suggesting TXA₂ participates as a requisite but permissive contributor to induced migration.

To test the hypothesis that TXA₂ may be a mediator of angiogenic responses promoted by COX-2, we evaluated effects of systemic COX-2 inhibition upon corneal angiogenesis in a mouse pellet implantation model. Systemic administration of a selective COX-2 inhibitor imposed a marked inhibitory effect on the angiogenic response to basic FGF in the corneal model, reducing the area vascularized by 52% and the density of vascularity within that area by 80% (Fig. 3a) ⇓ . This provided strong evidence that a COX-2 metabolite participates in angiogenic responses to FGF.

Consistent with a role for TXA₂, local administration of the TXA₂ receptor antagonist, SQ29548, in the corneal pellet inhibited FGF-stimulated angiogenesis, reducing the vascularized area by 40% and the vascular density within that vascularized area by 51% (Fig. 3b) ⇓ . Although vascular flow in the ocular circulation is sensitive to TXA₂ mimetics (21) , the response is vasoconstrictive, and the TXA₂ receptor antagonist should promote vasodilatation and capillary filling rather than the attenuation we observed. Thus, TXA₂ appears to function as an in vivo mediator of FGF-stimulated angiogenesis.

The action of the TXA₂ agonist U46619 to reconstitute endothelial migration under COX-2-inhibited conditions (Fig. 2a) ⇓ suggested the possibility of reconstituting corneal angiogenic responses to bFGF in the setting of systemic COX-2 inhibition. Shown in Fig. 4 ⇓ , locally administered U46619 showed a striking capacity to repair the attenuated angiogenic response seen in the setting of COX-2 inhibition, returning the vascularized area to 80% and the vascular density within that area to 95% of levels achieved with FGF in the absence of COX-2 inhibition. U46619 alone was not angiogenic, and it did not amplify on the bFGF response in animals with intact COX-2 function (vehicle).

These findings provide in vivo validation of a critical role for TXA₂ in neovascularization responses. Extrapolation from the endothelial migration responses in vitro (Fig. 2a) ⇓ suggests that a critical threshold level of TXA₂ is required to support angiogenesis in this system, one that is not met under non-COX-2-induced conditions. Although COX-2 induction may lead to endothelial production of TXA₂, other cellular sources may be relevant in the context of neovascularization in specific tissue circumstances.

Platelets generate TXA₂ from endogenous COX-1-derived substrate prostaglandin H₂ and can convert endothelial-derived prostaglandin H₂ to TXA₂ (22) . We speculate that TXA₂ from either source may support angiogenesis adjacent to microthrombi in tumors and other vascular sites. Indeed, reported effects of thromboxane synthase inhibitors and thromboxane receptor antagonists to inhibit metastatic behavior of tumor cells in mouse models were attributed to interruption of adhesive platelet interactions with tumor cells (23) . Our findings suggest that TXA₂ axis antagonists may, alternatively, act primarily to inhibit endothelial responses to angiogenic peptides that are required for tumor vascularization and metastasis. TXA₂ axis antagonists may retain antiangiogenic activity under circumstances in which COX-2 inhibition is ineffective in eliminating TXA₂ production that is dependent upon COX-1-derived substrate.

Although TXA₂ receptor null mice show no overt developmental vascularization defects or disorders of pregnancy, gestation, or delivery (24) , other critical mediators of neovascularization in mature animals, such as α_vβ₃, are not required for developmental vascularization to proceed (25) . The requirement for thromboxane receptors to mediate COX-2 induced responses provides a focal point for intervention and a rationale for application of thromboxane receptor and synthase antagonists to potentiate therapeutic efficacy of COX-2 inhibitors.