Hypoxia–Reoxygenation Triggers Coronary Vasospasm in Isolated Bovine Coronary Arteries via Tyrosine Nitration of Prostacyclin Synthase

All chemicals, if not indicated otherwise, were obtained from the same sources as previously described 9. N^ω-mono-methyl-l-arginine (L-NMMA), N^ω-nitro-l-arginine methyl ester (L-NAME), and diethylamine NONoate (DEA-NO) were obtained from Cayman SPI. Polyethylene-glycolated superoxide dismutase (PEG-SOD) was obtained from Sigma Chemical Co. ¹⁴C-PGH₂ was prepared from ¹⁴C-arachidonic acid as previously described 4,7. Rabbit anti–PGI₂ synthase antisera were produced in this laboratory as previously described 12.

As previously described 9, bovine coronary arteries (BCA) of the left ventricle were isolated and resuspended in a tissue bath with 15 ml of Krebs buffer, gassed with 95% O₂/5% CO₂ (37°C; pH 7.4), and attached to a force–displacement TFT6V5 transducer coupled to a polygraph for the measurement of isometric tension. Passive tension was adjusted to ∼1 g over a 30-min equilibrating period, and then coronary arteries were preconstricted by addition of the Tx mimetic U46619 (0.001–0.01 μmol/liter). The vasorelaxation after acetylcholine (0.01–1 μmol/liter) was used to demonstrate the presence of endothelium-dependent relaxation. Throughout the experiment, care was taken to avoid any injury to the endothelium.

Experiments were started by obtaining from each spiral a reference response of vasoconstriction–relaxation with angiotensin II (50 nM) for 30 min. After removal of the agonist from the chambers and return of tone to basal levels, the oxygen tension was then reduced abruptly from 95% O₂ /5% CO₂ to 95% N₂ /5% CO₂

and was maintained for 10-, 30-, or 40-min hypoxia. After this phase of hypoxia, 95% O₂/5% CO₂ was resumed (reoxygenation), and the tension was allowed to stabilize for 40 min before addition of agonists. If required, pharmacological agents were added in the organ chamber 40 min before hypoxia and kept during hypoxia–reoxygenation and the second stimulation with the agonists. Tissues were kept for immunohistochemistry or immunoprecipitation and Western blots. The media from the first and second stimulations with angiotensin II were collected and stored at −20°C for prostanoid analysis, as previously described 9, using ELISA kits according to the manufacturer's instructions.

BCA spirals with or without hypoxia–reoxygenation were incubated with 100 μM ¹⁴C-PGH₂ for 3 min as previously described 4,7,9. The reaction was stopped by acidification with 1 N HCl to pH 3.5. The incubation media were extracted with ethyl acetate (3 vol). After centrifugation, the organic phases were evaporated to dryness under nitrogen. Samples were then resuspended in 60 μl of ethyl acetate and subsequently separated by TLC (ethyl acetate/water/iso-octane/acetic acid, 90:100:50:20). Prostanoids were quantified with a PhosphorImager system (ImageQuant; Molecular Dynamics).

Immunohistochemistry, immunoprecipitation, and Western blots of 3-nitrotyrosine–containing protein in hypoxia–reoxygenation-treated BCA were performed as previously described 9.

In BCA, angiotensin II elicited a biphasic response consisting of a primary vasoconstriction that was selectively blocked by losartan, a selective angiotensin 1 receptor blocker, and a secondary vasorelaxation that was mainly PGI₂ dependent 9.

Abrupt decrease of oxygen tension from 95% O₂/5% CO₂ to 95% N₂ /5% CO₂ caused a slight decrease in tension after a transient rise. Reoxygenation (from 95% N₂ /5% CO₂ to 95% O₂ /5% CO₂) did not alter the tension of unstimulated arteries. Although hypoxia–reoxygenation did not affect the initial constriction of angiotensin II, it selectively blunted the angiotensin II–triggered relaxation phase after 30 min of hypoxia. Along with this suppression of the relaxation phase, a second constriction phase developed in parallel, with a decrease of 6-keto-PGF_1α (Fig. 1 A) that closely resembled the pattern seen previously after peroxynitrite pretreatment 9.

Both the COX (cyclo-oxygenase) inhibitor indomethacin and the TxA₂/PGH₂ receptor blocker SQ29548 restored hypoxia–reoxygenation-impaired relaxation without affecting 6-keto-PGF_1α release (Fig. 1 A). This suggests that a COX-derived product, TxA₂ or PGH₂, caused vasoconstriction via the TxA₂/PGH₂ receptor (Fig. 1 A). A role of TxA₂ was excluded, as CGS13080, a TxA₂ synthase inhibitor, did not affect hypoxia–reoxygenation-triggered vasospasm (Fig. 1 A), and the levels of TxB₂ remained low and unaffected after hypoxia–reoxygenation treatment (97 ± 11 vs. 99 ± 15 pg/30 min). The level of 8-iso-PGF_2α, which also could act on the TxA₂/PGH₂ receptor 14, was low and remained unchanged after hypoxia–reoxygenation (43 ± 12 vs. 47 ± 17 pg/30 min).

PGH₂ is known to cause vasoconstriction via the TxA₂/PGH₂ receptor 13. In arterial vessels, PGI₂ is metabolized primarily by PGI₂ synthase to yield PGI₂. Therefore, we postulated that an inhibition on PGI₂ synthase could be the primary cause for the accumulation of PGH₂, which then caused vasoconstriction by stimulating the TxA₂/PGH₂ receptor. Parallel measurements of prostaglandins in the incubation medium further supported this hypothesis. Hypoxia–reoxygenation lowered 6-keto-PGF_1α formation but raised the level of PGE₂, an enzymatic or nonenzymatic metabolite of PGH₂ (Fig. 1 A). As COX activity was slightly decreased after hypoxia–reoxygenation, an excess formation of PGH₂ after hypoxia–reoxygenation could be excluded.

Conclusive evidence for an inactivation of PGI₂ synthase came from the experiments with ¹⁴C-PGH₂ as a substrate for PGI₂ synthase. A significant inhibition on the conversion of ¹⁴C-PGH₂ into 6-keto-PGF_1α (−89 ± 8%) with a concomitant increase of PGE₂ (135 ± 21%) was observed, confirming the selective inhibition of PGI₂ synthase and the reorientation of PGH₂ metabolism toward PGE₂ after hypoxia–reoxygenation.

The sustained vasoconstriction after hypoxia–reoxygenation could also have been the consequence of a decreased sensitivity of vascular smooth muscle to NO. However, NO generated from DEA-NO in arteries with or without hypoxia–reoxygenation resulted in a similar potency to induce vasorelaxation. Similarly, an increased sensitivity of the TxA₂/PGH₂ receptor was excluded, as pD₂ (the negative logarithm of the molar concentration of agonist that elicits a half- maximal response) of U46619, an agonist for the TxA₂/PGH₂ receptor, was identical before and after exposure to hypoxia–reoxygenation (7.7 ± 0.5 vs. 7.8 ± 0.5). A correspondent dose of PGE₂ (20 ng/ml) applied to the bath solution did not cause vasoconstriction (data not shown), indicating that the higher levels of PGE₂ after hypoxia–reoxygenation were not responsible for the second constriction phase.

According to our working hypothesis, hypoxia–reoxygenation could induce peroxynitrite formation, which inactivates PGI₂ synthase. To test whether peroxynitrite was indeed responsible for the hypoxia–reoxygenation-induced vasospasm, the NO synthase inhibitors, both L-NMMA (10⁻⁴ M) and L-NAME (10⁻⁴ M) were concurrently administered during hypoxia–reoxygenation. L-NMMA, which reduced the angiotensin II–induced relaxation phase by about 25% without affecting 6-keto-PGF_1α formation in normal vessels (data not shown), prevented hypoxia–reoxygenation-induced suppression of angiotensin II–induced vasorelaxation and 6-keto-PGF_1α release (Fig. 1 A); L-NAME was as effective as L-NMMA (data not shown).

Alternatively, concurrent administration of PEG-SOD (500 U/ml) abolished hypoxia–reoxygenation-induced secondary vasoconstriction and restored angiotensin II–stimulated vasorelaxation by blunting the hypoxia–reoxygenation-mediated inhibition on 6-keto-PGF_1α formation (Fig. 1 A).

It was important to show that neither NO nor O₂^.− alone could cause vascular dysfunction and inhibition on PGI₂ release. Therefore, preincubation of BCA with an O₂^.−-generating system (10 mU/ml xanthine oxidase/100 μM hypoxanthine) did not significantly affect either angiotensin II–triggered vasorelaxation or 6-keto-PGF_1α release (Fig. 1 B). Alternatively, the incubation of BCA with DEA-NO (20 μM) as an NO-releasing system also failed to affect angiotensin II–induced relaxation and 6-keto-PGF_1α release (Fig. 1 B). However, concomitant administration to the organ baths of both O₂^.−- and NO-generating agents for 40 min caused a loss of PGI₂-dependent vasorelaxation, an indomethacin- and SQ29548-sensitive vasospasm, and a nitration of PGH₂ synthase (Fig. 1 B).

As previously described, the exposure of BCA to peroxynitrite produced a nitrated protein that colocalized with PGI₂ synthase 9. The same technique was applied to hypoxia–reoxygenation-exposed BCA segments after their mechanical responses had been established. Staining with antinitrotyrosine Ab was weakly visible in control tissue (Fig. 2 A), but clearly enhanced stainings emerged in endothelium and smooth muscle after hypoxia–reoxygenation (Fig. 2 B), where the stainings with Ab against PGI₂ synthase were intensively presented (Fig. 1 C). A computer-generated overlay of the stainings with antinitrotyrosine (green) and anti–PGI₂ synthase (red) resulted in the yellow colocalizing patches in vessels after hypoxia–reoxygenation (Fig. 2 D). The presence of L-NMMA or PEG-SOD abolished the increased stainings with antinitrotyrosine Ab (Fig. 2F and Fig. G) but not those with anti–PGI₂ synthase Ab (data not shown). The specificity of the staining with antinitrotyrosine Ab was deduced from its suppression by 10 mM 3-nitrotyrosine (Fig. 2 E) and the lack of staining when antinitrotyrosine Ab was omitted (Fig. 2 H). 3-chloro- or 3-aminotyrosine or phosphotyrosine were ineffective in blocking the stainings with antinitrotyrosine Ab (data not shown).

Further evidence for hypoxia–reoxygenation-triggered tyrosine nitration of PGI₂ synthase came from the immunoprecipitation with both antibodies against 3-nitrotyrosine and PGI₂ synthase. Immunoprecipitation with anti–PGI₂ synthase Ab yielded similar amounts of protein in both hypoxia–reoxygenation-treated arteries and control tissues (Fig. 3 A). In Western blot analysis, a dense band was detected by antinitrotyrosine Ab in the immunoprecipitates with anti–PGI₂ synthase Ab from hypoxia–reoxygenation-treated vessel homogenates against a weakly visible signal in those from the control tissues (Fig. 3 A). In the same homogenates, 3-nitrotyrosine–containing proteins were precipitated with antinitrotyrosine Ab in parallel. Higher amounts of these proteins were recovered in hypoxia–reoxygenation-treated arteries than in those without hypoxia–reoxygenation. Moreover, when these precipitates were blotted and probed with anti–PGI₂ synthase Ab, a dense staining appeared in arteries with hypoxia–reoxygenation (Fig. 3 B). In agreement with the immunohistochemistry results, both PEG-SOD and L-NMMA effectively abolished the increased stainings with both antibodies in the immunoprecipitates with either antinitrotyrosine or anti–PGI₂ synthase (Fig. 3a and Fig. b).

In summary, our results indicate that hypoxia–reoxygenation elicits the formation of O₂^.−, which neutralize NO to form peroxynitrite. Subsequently, peroxynitrite nitrates and inactivates PGI₂ synthase, leaving PGH₂ unmetabolized, which then causes vasospasm and platelet aggregation via the TxA₂/PGH₂ receptor. This finding might offer a new mechanism for coronary vasospasm during hypoxia–reoxygenation, especially in atherosclerotic arteries where PGI₂ synthase is partially nitrated 15.

We dedicate this paper to Prof. Ullrich on the occasion of his 60th birthday for his generous support. We thank Ms. Nie Hong for technical support and Dr. Leist for help with the confocal images.

This work was supported by the Deutsche Forschungsgemeinschaft Forschergruppe We686/18-1.