Hypoxic vasodilation by red blood cells: evidence for an s-nitrosothiol-based signal

Abstract

Red blood cells (RBCs) have been ascribed an essential role in matching blood flow to local metabolic demand during hypoxic vasodilation. The vasodilatory function of RBCs evidently relies on the allosteric properties of hemoglobin (Hb), which couple the conformation of Hb to tissue oxygen tension (Po(2)) and thereby provide a basis for the graded vasodilatory activity that is inversely proportional to Hb oxygen saturation. Although a large body of evidence indicates that the Po(2)-coupled allosteric transition from R (oxy)-state to T (deoxy)-state subserves the release from Hb of vasodilatory nitric oxide (NO) bioactivity, it has not yet been determined whether the NO-based signal is a necessary and sufficient source of RBC-mediated vasoactivity and it has been suggested that ATP or nitrite may also contribute. We demonstrate here by bioassay that untreated human RBCs rapidly and substantially relax thoracic aorta from both rabbit and mouse at low Po(2) (approximately 1% O(2)) but not at high Po(2) (approximately 21% O(2)). RBC-mediated vasorelaxation is inhibited by prior depletion of S-nitroso-Hb, potentiated by low-molecular-weight thiols, and dependent on cGMP. Furthermore, these relaxations are largely endothelium-independent and unaffected by NO synthase inhibition or nitrite. Robust relaxations by RBCs are also elicited in the absence of endothelial, neuronal or inducible NO synthase. Finally, contractions that appear following resolution of RBC-mediated relaxations are dependent on NO derived from RBCs as well as the endothelium. Our results suggest that an S-nitrosothiol-based signal originating from RBCs mediates hypoxic vasodilation by RBCs, and that vasorelaxation by RBCs dominates NO-based vasoconstriction under hypoxic conditions.

Figure 1

Hypoxic vasodilation (≈1% O₂) by native human RBCs of mouse aortic segments, precontracted with phenylephrine (PE). A, Lack of effect of endothelial denudation or inhibition of NOS activity with L-NAME. Vasorelaxation was greatly attenuated by the guanylate cyclase inhibitor ODQ (n=9 to 34). *P<0.05. B and C, Representative examples of the elimination of acetylcholine (ACh)-induced vasorelaxation following endothelial removal versus preserved RBC-induced vasorelaxation.

Figure 2

Hypoxic vasodilation by native human RBCs of rabbit aortic segments. A, No significant effect on vasorelaxation resulted from inhibition of NOS activity with L-NAME, L-NMMA, or 1400W. Vasorelaxation was greatly attenuated in both intact and denuded rings by treatment with ODQ (n=12 to 110) *P<0.05. B through E, Representative examples illustrate the lack of effect of NOS inhibition, the lack of effect of NOS inhibition combined with denudation, and the elimination of vasorelaxation by ODQ.

Figure 3

Hypoxic vasodilation by native human RBCs of mouse aortic segments. A, Lack of effect of genetic elimination of eNOS (eNOS^-/-). Elimination of iNOS (iNOS^-/-), or nNOS (nNOS^-/-) was associated with enhanced vasorelaxation. Vasorelaxation was greatly attenuated following treatment with the guanylate cyclase inhibitor ODQ in all cases (n=6 to 29). *P<0.05. B through D, Representative tracings illustrate the lack of effect of eNOS elimination and the enhancement of relaxation by nNOS and iNOS elimination.

Figure 4

Hypoxic vasodilation by native human RBCs of rabbit aortic segments. A, RBC-induced vasorelaxation was potentiated by cysteine and by N-acetylcysteine but was unaffected by nitrite or the prostaglandin synthesis inhibitors indomethacin and aspirin (all agents added 2 minutes before RBCs) (n=5 to 11). *P<0.05. B through D, Representative tracings illustrate the augmentation of relaxation by cysteine and the attenuation of relaxation by SNO-Hb depletion (≈80% depletion). E, Addition of nitrite to the bath (even at supraphysiological concentrations of 1 μmol/L, which are higher than those used in Cosby24) before (NO₂^-⇒RBC) or simultaneously with the addition of RBCs (NO₂^-+RBC), or near the termination of RBC-induced vasorelaxation (RBC⇒NO₂^-) had no discernible effect on the magnitude or duration of RBC-induced hypoxic vasodilation.

Figure 5

PO₂ dependence of vasoconstriction by native human RBCs of rabbit aortic segments. RBC-induced contraction terminated the relaxation at ≈1% O₂ and exhibited latencies at ≈21% O₂ (in the absence of relaxation) that were indistinguishable from the latencies of relaxation at ≈1% O₂. Contractions were greater at ≈21% O₂ than at ≈1% O₂ and greatly attenuated by endothelium removal and by ODQ (n=9 to 36). *P<0.05. Representative examples illustrate the attenuation of vasoconstriction by hypoxia, denudation, and exposure to ODQ.