Male NMRI and myo^−/− mice were used at 6–7 weeks of age; Myo^−/− mice were obtained from a breeding colony at the Heinrich-Heine-Universität, Düsseldorf. Animals were housed under standard lighting conditions (normal 12 h/12 h cycle) and provided normal rodent chow and water ad libitum. Tissue was harvested after euthanasia, using methods approved under Schedule 1 of the Animals (Scientific Procedures) Act 1986. The study was approved by the local University Review Committee for ethical and safety issues (The investigation also conforms with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health, NIH Publication No. 85-23, revised 1996).

Six to 7-week-old male mice were sacrificed by cervical dislocation. The thoracic aorta from the arch to the diaphragm was removed en bloc and immediately immersed in ice-cold buffer. Krebs-Henseleit buffer was used throughout, with the following composition: NaCl 120.0 mM, KCl 4.7 mM, MgSO₄ 1.2 mM, K₂PO₄ 1.2 mM, Glucose 11.0 mM, CaCl₂ 2.5 mM, NaHCO₃ 25 mM. All chemicals were obtained from Sigma-Aldrich Co (Gillingham, Dorset, UK), apart from sodium nitrite (Martindale Pharmaceuticals, Romford, UK) and carboxy-PTIO (Tocris Bioscience, Bristol, UK). All adherent fat and adventitia was stripped and the endothelium removed by bubbling with air. The aorta was cut into four 2 mm rings and hung in a four-well wire myograph (Danish Myo Technology model 610M, A/S, Denmark) in Krebs-Henseleit buffer. Myography wells were open to the air and bubbled with either carbogen (95% O₂/5% CO₂) or nitrogen (N₂), with and without carbon monoxide, CO (95% N₂/5% CO₂ or 75% N₂/5% CO₂/20% CO, respectively). Buffer was allowed to reach steady state. Using this equipment, the nitrogen-containing gas mixtures produced a buffer oxygen tension of 83 ± 2.5 mmHg (11 kPa, i.e. between average arterial and venous conditions), which was stable over time. The rings were attached to an isometric force transducer and the resting tension was increased stepwise to a maximum of 20 mN, which was determined as optimal in preliminary experiments. A typical trace is shown in Figure 1A, and a concentration response curve in Figure 1B (carbogen gassing, wild-type aorta, 1 µM phenylephrine precontraction). Preliminary experiments revealed that the level of constriction achieved using phenylephrine was variable over different oxygen concentrations, and was inadequate under reduced oxygen tension (the precontractile tone achieved using 95% N₂/5% CO₂ reached ∼40% of that in the presence of 95% O₂/5% CO₂). Therefore, 50 mM KCl was substituted for phenylephrine for experiments performed under mildly hypoxic conditions. Each measure was taken as the average of at least two segments and is expressed as a percentage reversal of preconstriction. For inhibitor experiments, 50 µM oxypurinol and 50 nM raloxifene were incubated with the rings for 5 min before the experimental runs, which were otherwise performed in an identical fashion. All data are shown as the mean ± SEM, and P < 0.05 was taken to indicate statistical significance. Bonferroni post hoc correction was used for data with multiple comparisons.

Thoracic aortas from myo^−/− mice were obtained as before. These rings were then exposed to either adenovirus expressing myoglobin (AdMYO) or control virus (AdCtl) (at a concentration of 10⁻⁸ PFU/mL) diluted in Opti-MEM (Sigma), supplemented with 11 mM glucose, for 120 min under continuous bubbling with carbogen gas. Rings were washed with Opti-MEM to remove adherent virus particles, and then incubated at 37°C in Krebs-Henseleit buffer for 270 min. This incubation time was determined in preliminary experiments to be sufficient for recapitulation of the low levels of myoglobin seen in wild-type aorta. Response to nitrite both in the presence and absence of CO was assessed as before. Nitrite was used at a final concentration of 9 mM (∼EC₉₀) in order to focus solely on the effect of myoglobin (i.e. after saturating other pathways).

Fresh aortic tissue was exposed to 10 mM nitrite in Krebs-Henseleit buffer at 37°C for 15 min, with and without the NO scavenger carboxy-PTIO. A phosphodiesterase inhibitor cocktail was used, as described previously.²² cyclic GMP (cGMP) was measured in aortic tissue after homogenization using a commercial ELISA kit (Amersham Biotech, UK).

Both AdMyo and AdCtl (control virus) were generated using the AdEasy XL Adenoviral vector system (Stratagene). Human wild-type myoglobin cDNA was cloned into the pShuttle-IRES-hrGFP-1 vector (Stratagene), allowing co-expression of green fluorescent protein (GFP) with myoglobin (a stop codon was introduced to prevent fusion of the FLAG tag). The vector was linearized by PmeI digestion and recombined with the pAdEasy-1 (viral backbone vector) in BJ5183 Escherichia coli. The recombined adenoviral constructs were transfected into DH10β E. coli to allow higher plasmid yields. After confirmation of recombination by BstXI and Pac-1 restriction digestion and sequencing, Pac1 linearized recombinant Ad plasmids were then transfected into AD293 cells (Stratagene) using oligofectamine (Invitrogen). AD293 cells were scraped from cell culture vessels and lysed by four freeze-thaw vortex cycles. Lysates containing recombinant adenoviruses used to infect further AD293 cells for amplification. The viruses were purified by caesium chloride gradient ultracentrifugation. The viral titre was estimated by infecting HEK293 cells (not expressing capsid proteins) and counting GFP expressing cells.

Descending thoracic aortas were removed after euthanasia as above. They were stripped of all adherent fat and adventitia and snap frozen in liquid nitrogen. Samples were homogenized using a small blade homogenizer in TriReagent (Sigma) and RNA was extracted. cDNA was transcribed from purified RNA using a standard kit (Applied Biosystems) and 40 cycles of PCR were performed (primer sequences available on request). Tissue samples for western blotting were snap frozen and subsequently homogenized in lysis buffer (25 mM Tris/HCl, pH 7.4, 1 mM EDTA, 1% SDS, 1 mM dithiothreitol, and complete-mini protease inhibitor cocktail tablet (Roche)) and then boiled with Laemlli buffer. Homogenates were separated by 15% sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS–PAGE). Proteins were transferred to PVDF membranes (BioRad), saturated with 5% fat-free milk in 0.1% Tween-20 in phosphate buffered saline (PBS) (blocking solution), and hybridized with α-myoglobin rabbit polyclonal antibody (Santa Cruz, (FL-154) sc-25607) at 1:1000 dilution overnight at 4°C. The membranes were washed four times (10 min each at RT) with 0.1% Tween-20 in PBS and then incubated with goat anti-rabbit IgG (HRP) (Abcam ab2721-1) at 1:4000 dilution for 1 h at RT, washed four times, and finally visualized with the ECL advance immunoblotting detection system (GE Healthcare).

Nitrite caused concentration-dependent relaxation of endothelium-denuded aortic rings, which was completely abolished by preincubation with 20 µM oxygenated haemoglobin (oxyHb) (Figure 1C) implying that nitrite acts through the generation of NO. Moreover, addition of 10 µM 1H-(1,2,4)oxadiazolo(4,3-a)quinoxalin-1-one (ODQ) to the organ bath prevented nitrite-induced relaxation (Figure 1C), suggesting the effects were mediated by stimulation of soluble guanylyl cyclase. This was confirmed in a separate set of experiments performed in the presence of a phosphodiesterase inhibitor cocktail. Under these conditions, the exposure of vascular tissue to nitrite caused accumulation of cyclic GMP (cGMP), which was lower in myo^−/− tissue than in wild type and abolished by preincubation with the NO scavenger, carboxy-PTIO (Figure 1D).

Acute vasorelaxation induced by increasing concentrations of sodium nitrite was assessed during exposure to 95% N₂/5% CO₂ and then to 20% CO/75% N₂/5% CO₂. The level of CO was chosen to inhibit only haem proteins, as previously described,²³ and did not in itself cause vasorelaxation. CO inhibited nitrite-induced vasorelaxation in the murine wild-type aortas (P < 0.001 by two-way ANOVA) (Figure 2A) suggesting that an intrinsic vascular haem moiety significantly contributes to vascular nitrite bioconversion. The effect of 50 μM oxypurinol (an inhibitor of xanthine oxidase) and 50 nM raloxifene (an inhibitor of aldehyde dehydrogenase) is shown for comparison. To assess whether myoglobin was this haem moiety, myo^−/− mice²⁴ were compared with wild-type controls. Overall response to nitrite was markedly reduced in myo^−/− rings (Figure 2B) (P < 0.001 by two-way ANOVA), implying that myoglobin significantly contributes to vascular nitrite bioconversion. In contrast to wild-type aortas, CO did not inhibit vasorelaxation in myo^−/− rings (Figure 2B). To identify the sources of residual nitrite bioconversion, we assessed myo^−/− aortic rings for their vasodilator response to an informative level of nitrite (9 mM; as derived from the concentration response curve) in our experimental system. Preincubation time of vascular rings with inhibitors was 5 min with each agent separately or together; optimal concentrations of inhibitors used were determined in preliminary experiments. Fifty micromolar oxypurinol caused a 32% inhibition of vasorelaxation (from 21.3 ± 0.4 to 14.4 ± 1.4%, P < 0.05 by ANOVA) whereas 50 nM raloxifene produced a 32% inhibition (from 21.3 ± 0.4 to 14.3 ± 0.5%, P < 0.05 by ANOVA). The two agents in combination inhibited nitrite-induced vasodilation by 60% (from 21.3 ± 0.4 to 8.5 ± 1.1%, P < 0.001 by ANOVA) (Figure 2C).

Sodium nitroprusside (SNP) was used to assess the effect of CO on vasorelaxation by a nitrite-independent NO donor. Importantly, myo^−/− aortic rings retained their response to NO (i.e. the extent of vasorelaxation to a fixed concentration of SNP (0.1 µM) was comparable in tissues from wt and myo^−/− mice) and relaxation was increased from 37.7 ± 3.5 to 43.7 ± 4.2% by the addition of CO (P = 0.032 by ANOVA). In wild-type mice, relaxation was increased from 42.3 ± 4.7 to 58.7 ± 6.0% (P = 0.029 by ANOVA) (Figure 2D). The difference in magnitude of effect of CO is consistent with greater scavenging of NO by myoglobin in the wild-type mouse.

To confirm the role of myoglobin as a nitrite bioconvertor and to exclude the possibility that the changes noted in the myo^−/− mice were artefactual adaptations to gene manipulation, we investigated the effect of restoring myoglobin in myo^−/− mouse vessels. Transduction of myoglobin augmented the baseline response to nitrite from 28.6 ± 6.5 to 60.7 ± 7.6% (P = 0.024 by ANOVA) (Figure 3A), compared with control virus-treated rings. The pattern of response to CO was similar. Thus, treatment with control virus did not change the effect of CO (from 28.6 ± 6.5 to 31.6 ± 7.0%) (Figure 3A). Crucially, viral transduction of myoglobin reinstated the inhibition of nitrite relaxation by CO seen in wild-type mice (from 60.7 ± 7.6 to 25.8 ± 3.7%, P = 0.014 by ANOVA) (Figure 3A). The concentration of nitrite (∼EC₉₀ at 9 mM) used in this study was chosen to facilitate accurate determination of the vasorelaxing effect of nitrite and the potentially subtle effect of interventions. As expected, endpoint RT–PCR confirmed that myoglobin is present at mRNA level in wild-type murine aortas and absent in myo^−/− aortas (Figure 3B). Western blotting confirmed the presence of myoglobin in AdMYO-treated and wild-type rings, at a similarly low level, and its absence in AdCtl-treated rings (Figure 3C).