LC-MS/MS confirms that COX-1 drives vascular prostacyclin whilst gene expression pattern reveals non-vascular sites of COX-2 expression

Abstract

There are two schools of thought regarding the cyclooxygenase (COX) isoform active in the vasculature. Using urinary prostacyclin markers some groups have proposed that vascular COX-2 drives prostacyclin release. In contrast, we and others have found that COX-1, not COX-2, is responsible for vascular prostacyclin production. Our experiments have relied on immunoassays to detect the prostacyclin breakdown product, 6-keto-PGF1α and antibodies to detect COX-2 protein. Whilst these are standard approaches, used by many laboratories, antibody-based techniques are inherently indirect and have been criticized as limiting the conclusions that can be drawn. To address this question, we measured production of prostanoids, including 6-keto-PGF1α, by isolated vessels and in the circulation in vivo using liquid chromatography tandem mass spectrometry and found values essentially identical to those obtained by immunoassay. In addition, we determined expression from the Cox2 gene using a knockin reporter mouse in which luciferase activity reflects Cox2 gene expression. Using this we confirm the aorta to be essentially devoid of Cox2 driven expression. In contrast, thymus, renal medulla, and regions of the brain and gut expressed substantial levels of luciferase activity, which correlated well with COX-2-dependent prostanoid production. These data are consistent with the conclusion that COX-1 drives vascular prostacyclin release and puts the sparse expression of Cox2 in the vasculature in the context of the rest of the body. In doing so, we have identified the thymus, gut, brain and other tissues as target organs for consideration in developing a new understanding of how COX-2 protects the cardiovascular system.

Conflict of interest statement

Competing Interests: JAM and TDW have received funds from GSK and Astra Zeneca and acted as consultants and expert witnesses for Pfizer in cases where anti-inflammatory drugs, including COX-2 inhibitors have been involved. AN has received research funds and acted as consultant for projects related to supplement and drug analysis from/for Allergan, Unilever, Amarin, Equateq, GSK. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials.

Figures

Figure 1. 6-keto-PGF1α production in isolated mouse aorta; measurement by enzyme immunoassay, radio immunoassay, and liquid chromatography tandem mass spectrometry (LC-MS/MS).

Prostacyclin release by isolated rings of mouse aorta stimulated with Ca2+ ionophore A23187 (50µM), measured as the stable breakdown product 6-keto-PGF1α, was not altered byCox2 gene deletion, but was reduced >10-fold byCox1 gene deletion. The pattern and level of 6-keto-PGF1α accumulation was similar whether measured by (a) enzyme immunoassay, (b) radio immunoassay or (c) LC-MS/MS. Representative LC-MS/MS chromatograms show the presence or absence of 6-keto PGF1α in all sample types (retention time 2.81 min; transition ion m/z 369>163). n=4-7. *, p<0.05 by 1-way ANOVA with Bonferonni’s post-test.

Figure 2. Bradykinin-stimulated prostanoid accumulation in the circulation in vivo in wild-type, Cox1 -/-, and Cox2 -/- mice.

Accumulation of the stable prostacyclin breakdown product, 6-keto-PGF1α in plasma after bradykinin administration (100nmol/kg i.v.) is dependent on COX-1 but not COX-2 when measured by LC-MS/MS (a). Representative LC-MS/MS chromatograms show the presence or absence of 6-keto PGF1α in all sample types (retention time 2.81 min; transition ion m/z369>163). Similar data were obtained for plasma levels of PGE2 (b), 13,14-dihydro-15-keto-PGE2 (c), PGD2 (d), TXB2 (e) and (f) PGF2α. Plasma 6-keto-PGF1α levels in all genotypes compare well with those previously published using enzyme immunoassay measurements. n=6. *, p<0.05 by 1-way ANOVA with Bonferonni’s post-hoc test.

Figure 3. Distribution of luciferin-dependent bioluminescence in cardiovascular tissue from Cox2 fLuc/+ mice.

(a) Quantification of basal expression from the aortic tree, vena cava, chambers of the heart and, for comparison, brain fromCox2fLuc/+ mice and (b) and representative images of bioluminescence. Arteries, veins and chambers of the heart were essentially devoid of expression from theCox2 gene, in comparison with the brain as a reference tissue. The only exception to this was weak, but detectable, expression in the region of the aortic arch. n=3.

Figure 4. Distribution of luciferin-dependent bioluminescence in tissues from Cox2 fLuc/+ mice.

(a) Basal expression from organs of theCox2fLuc/+ mice was visualized by bioluminescent imaging of tissues dissected fromCox2fLuc/+ reporter mice after injection of D-luciferin in vivo (125mg/kg i.p.). (b) Imaging data are expressed as maximum luminescent emission from each tissue. Basal Cox2 gene driven luciferase expression was present in many tissues including the vas deferens, brain, intestine, and thymus but was notably low to absent in the aorta (highlighted with red circles). Sub-division of the (c) brain, (d) intestine, (e) kidney and (f) stomach revealed regional expression patterns within each tissue. n=5.

Figure 5. COX-2-dependent prostanoid production by aorta versus other mouse tissues in Cox1 -/- mice.

(a) PGE2 formation, normalized to tissue mass, was measured by immunoassay in supernatants of Ca2+ ionophore A23187 (50µM)-stimulated tissue segments fromCox1-/- mice. Cox1-/- tissues released a variable amount of PGE2 with low levels in the aorta (highlighted in red), and substantially higher levels in the thymus, intestines, renal medulla, brain and vas deferens. This distribution correlates well with luciferase expression in organs of theCox2fLuc/+ mouse, as described in Figures 3 and 4. n=6.