Oxygen Mediates Vascular Smooth Muscle Relaxation in Hypoxia

The activation of soluble guanylate cyclase (sGC) by nitric oxide (NO) and other ligands has been extensively investigated for many years. In the present study we considered the effect of molecular oxygen (O2) on sGC both as a direct ligand and its affect on other ligands by measuring cyclic guanosine monophosphate (cGMP) production, as an index of activity, as well as investigating smooth muscle relaxation under hypoxic conditions. Our isolated enzyme studies confirm the function of sGC is impaired under hypoxic conditions and produces cGMP in the presence of O2, importantly in the absence of NO. We also show that while O2 could partially affect the magnitude of sGC stimulation by NO when the latter was present in excess, activation by the NO independent, haem-dependent sGC stimulator 3-(5′-hydroxymethyl-2′-furyl)-1-benzylindazole (YC-1) was unaffected. Our in vitro investigation of smooth muscle relaxation confirmed that O2 alone in the form of a buffer bolus (equilibrated at 95% O2/5% CO2) had the ability to dilate vessels under hypoxic conditions and that this was dependent upon sGC and independent of eNOS. Our studies confirm that O2 can be a direct and important mediator of vasodilation through an increase in cGMP production. In the wider context, these observations are key to understanding the relative roles of O2 versus NO-induced sGC activation.


Introduction
Soluble guanylate cyclase (sGC) is well known as the ''receptor'' for nitric oxide (NO). Binding of this gaseous diatomic molecule to the haem moiety of the enzyme stimulates the conversion of guanosine triphosphate (GTP) to cyclic guanosine monophosphate (cGMP), a nucleotide that is involved in several vital intracellular signalling cascades and physiological processes [1]. Although NO is the preferred ligand for sGC, activating the enzyme several hundred fold over its basal level, other gaseous and synthetic activators, such as carbon monoxide (CO) and the benzylindazole derivative 3-(59-hydroxymethyl-29-furyl)-1-benzylindazole (YC-1) respectively, have been identified. While only a moderate activator, CO would seem to mediate its action via the same mechanism as NO. YC-1 on the other hand employs a totally different strategy, binding to an allosteric site on the enzyme thereby increasing the maximal catalytic rate. Such an action produces a 10-fold increase in basal sGC activity independently of other ligands, while also potentiating the responses to subsequent exposure to such agents [2]. Given the above it is evident that there are more ways than one to activate sGC. Importantly the physiological relevance of actions that modulate the response of sGC to other ligands remains in question.
Perhaps one of the most significant roles for the NO/sGC axis is in the control of vascular tone [3][4][5], a function that is essential to maintaining blood flow and oxygen/nutrient delivery to tissues. As such, in response to NO produced by adjacent endothelial cells, sGC located within vascular smooth muscle is activated and the subsequent production of cGMP mediates vasorelaxation. Since CO can also induce this cGMP-dependent response, it is perhaps a little surprising, given that NO, CO, and molecular oxygen (O 2 ) differ by only one valence electron, that a role for O 2 in this process has not been identified.
Conventional understanding would suggest that O 2 simply does not bind to sGC [6][7][8], at least not in the same way as NO and CO. However changes in O 2 tension are widely recognised to influence vascular tone. For instance, via the regulation of specific potassium channel activity, O 2 plays a major role in the control of pulmonary vascular tone [9,10]. Perhaps of more relevance to an interaction with sGC is our previous demonstration that the relaxation response to endogenous NO stimulation or exogenous NO addition is enhanced the lower the O 2 tension [11], implying an inverse relation between NO effectiveness and O 2 . These factors contribute to the concept of ''hypoxic vasodilation'', an innate physiological response designed to maintain tissue perfusion in the face of falling O 2 concentrations [12][13][14][15][16]. While the exact mechanisms that underlie this response have been the subject of active research and debate for many years, it is now widely accepted that activation of sGC is intrinsically involved.
The introduction of oxygenated red blood cells (RBCs) to hypoxic tissues is now well recognised to immediately induce vasorelaxation. However, while these cells would certainly, and very quickly, release O 2 under such conditions, the latter has been overlooked as a potential mediator of the relaxation response in favour of more conventional activators of s'GC. To this end, oxygenated haemoglobin-derived nitrosothiol (HbSNO) [17], nitrite (NO 2 2 )-derived NO as a consequence of deoxyhaemoglobin nitrite reductase activity [15,20], and RBC-derived adenosine triphosphate (ATP) stimulation of endothelium-dependent release of NO and prostacyclin [21,22] have all been postulated to mediate ''hypoxic vasodilation''.
While all or any of the above could contribute to RBC-induced vasodilation in the acute setting, it would seem that none provides a totally clear mechanism. The data described in this manuscript advocates an alternative and more straight-forward candidate, molecular O 2 . Therefore, the aim of the present study was to investigate whether O 2 could act as a possible direct ligand for sGC and/or a modulator of the actions of other preferred ligands. To accomplish these aims two model systems were used; an isolated sGC enzyme system and an isolated blood vessel system, both of which allowed for tight control of the local O 2 environment. Importantly, these models not only allowed us to probe O 2 -dependent mechanisms but also to investigate how these align with the known features of the putative sGC-dependent relaxants described above.

Ethics Statement
Animal. In accordance with the United Kingdom Animal (Scientific Procedures) Act of 1986, this study did not require a Home Office project license because no regulated procedures were carried out. White, male New Zealand rabbits were humanely killed at a designated establishment by sodium pentobarbitone, which is an appropriate method under Schedule 1 of the Act.
Human. Methods requiring human blood samples were fully approved by the local research ethics committee (South East Wales Research Ethics Committee) and the joint NHS/University Research and Development Office (Cardiff and Vale University Health Board). All healthy volunteers gave written informed consent.

Isolated sGC Study
All experiments were performed in an Invivo 2 Hypoxia Workstation 400 (Ruskin). Normoxic experiments were carried out at 37uC and 20% O 2 /5% CO 2 via a 25% O 2 /5% CO 2 gas cylinder (BOC). Hypoxic experiments were maintained at 37uC and ,0% O 2 /5% CO 2 . Reagents were allowed to equilibrate for 1 hour before all tests were completed. Precise buffer O 2 concentrations for these experiments are also shown in Table 1.
The pure sGC enzyme was reconstituted in buffer (Tris 50 mmol/L, DTT 1 mmol/L, 0.5% BSA, pH 7.4) and diluted 1 in 200 in assay buffer (Tris 50 mmol/L, EGTA 100 mmol/L, MgCl 2 0.3 mmol/L, 0.045% BSA, pH 7.4. GTP, dissolved in equimolar MgCl 2 , was then added to a final concentration of 1 mmol/L to start the reaction. As appropriate, NOC-9 (to give final concentrations of 0.118, 1.118, 11.8 and 118 mmol/L) or YC-1 (100 mmol/L) were added at the same time as the GTP. In experiments utilising superoxide dismutase (SOD) and catalase (CAT), agents were added 60 minutes prior to GTP. Hyperoxic samples were prepared by perfusing the buffer-enzyme mix with 95% O 2 /5% CO 2 . All reactions were incubated for 10 minutes at 37uC, after which boiling inactivation buffer (Tris 50 mmol/L, EDTA 4 mmol/L, pH 7.5) was added in 4 times excess. Samples were then heated to boiling point before storage at 220uC for further analysis.

cGMP ELISA
Sample cGMP content was measured by commercial ELISA (R&D Systems) as directed in the kit manual.
For isometric tension recording, rings were suspended in tissue baths containing 5 ml of Krebs at 37uC and gassed with 95% O 2 / 5% CO 2 . Resting tension was set to 2 g. Signals from the transducer (AD Instruments, Chalgrove, UK) were amplified and visualised on the Powerlab/Chart 4 for Windows software. All rings were allowed to equilibrate for 60 minutes prior to experimentation.
Constriction-relaxation exercises to 1 mmol/L phenylephrine (PE) and 10 mmol/L acetylcholine (ACh) (Sigma Aldrich), respectively, were performed in order to establish both smooth muscle and endothelial integrity.
In order to equilibrate the rings in defined levels of hypoxia, the gas supply was switched to the appropriate O 2 content (ensuring balance of N 2 and 5% CO 2 ) for 10 minutes. Constriction to PE was tested across a broad range of concentrations and the appropriate PE concentration to achieve 80% of maximal attainable constriction used in all studies. For hypoxic studies (0% O 2 ), rings were typically exposed to 3 mmol/L PE in order to achieve a similar sub-maximal constriction as observed in normoxic rings. The O 2 concentration measured in the tissue bath equilibrated under hypoxic conditions was 0.9% (equivalent to 9 mmol/L O 2 ) as we have detailed previously [11].

Krebs Samples
Krebs solution was gassed at 95% O 2 /5% CO 2 in order to achieve 95% O 2 samples. 21% O 2 samples were allowed to equilibrate with the atmosphere. 0% O 2 samples were gassed with 95% N 2 /5% CO 2 . In order to accurately measure O 2 concentration within these equilibrated Krebs buffer samples, electron paramagnetic resonance (EPR) oximetry was undertaken utilising N 15 per-deuterated tempone (2,2,6,6-tetra-methyl-4piperidone; PDT) as the O 2 -sensitive probe. In brief, the spectral line width obtained from PDT (,5 mmol/L) shows a linear relationship with O 2 concentration and sample readings can be measured against known standards. This technique measures O 2 with an accuracy of 60.2% O 2 in our hands as we have described in detail previously [23]. The precise O 2 concentrations within ''95%'', ''21%'' and ''0%'' buffer are shown in Table 1.

Haemoglobin (Hb) Samples
Venous RBCs were lysed and diluted 1 in 3 in PBS (0.9% sodium chloride w/v), of which 250 ml was then loaded on to a Sephadex G25 M column (GE Healthcare). Samples of pure Hb were collected in fractions for use in the myograph studies. The HbO 2 content was verified by blood gas analysis as described (see 'Venous RBC samples'). The volume used in the myograph studies gave a final O 2 concentration in the tissue bath of 30.9 mmol/L which was equal to the O 2 content of the tissue bath following addition of either buffer or RBC samples.

Venous RBC Samples
Blood was drawn from the antecubital vein of healthy volunteers and immediately centrifuged at 12006g for 5 minutes. The plasma layer and buffy coat were removed and replaced with an equivalent volume of gassed (95% O 2 /5% CO 2 ) phosphate buffered saline (PBS). Following re-centrifugation at 12006g for 5 minutes the PBS layer was removed. In order to achieve RBCs of varying O 2 saturation, the cells were firstly diluted ,1:3 with PBS and loaded on to a thin film rotating tonometer, which was purged with O 2 /CO 2 to achieve higher saturations or N 2 /CO 2 for lower saturations as we have described previously [24]. The O 2 saturation of representative RBC samples was measured on an OSM3 Hemoximeter (Radiometer, Copenhagen).

In vitro NO Analysis
Phosphate buffered saline (PBS), pH 7.4, was kept at 37uC in a reaction vessel purged with a flow of nitrogen gas feeding into a Nitric Oxide Analyser (NOA) 280i (Analytix) for on-line ozonebased chemiluminescence (OBC) detection of NO. NOC-9 was reconstituted in 0.1M sodium hydroxide (NaOH) to give a final stock concentration of 24 mmol/L and kept on ice in the dark. For the experimental samples, the NOC-9 was further diluted in 1 ml of PBS (protected from light in a sealed vessel) to give the following concentrations (in mmol/L): 2.4; 1.2; 0.24; 0.12 and 0.024. In order to compare NO release under different conditions, the PBS was either pre-equilibrated to 0% or 95% O 2 by vigorous bubbling. Samples were incubated for 10 minutes at 37uC, after which 200 ml of the gas layer was drawn up using a Hamilton syringe and injected immediately into the purge vessel for OBC analysis. After the signal trace returned to baseline values, 200 ml of the corresponding PBS sample was drawn up and injected into the purge vessel. Data were recorded in real time and presented as area under the curve (Liquid software).

Statistical Analysis
All data were analysed by one-way ANOVA plus Tukey's or Dunnett's post hoc test or Student's t test as appropriate. Pearson's correlations were assessed and coefficients are shown where appropriate. All data are expressed as mean 6 standard error (SEM). Differences were considered significant where p,0.05.

Isolated Pure sGC Activity in Normoxia/Hypoxia
A striking observation of these experiments was the difference in basal sGC activity in normoxia (,100 pmol/ml) compared to hypoxia (,30-40 pmol/ml) (Figure 1a).
In separate experiments, YC-1 (100 mmol/L), a compound known to stimulate sGC without direct interaction with the sGC haem group [2], was added to sGC under hypoxic and normoxic conditions as above. While the addition of YC-1 enhanced the amount of cGMP produced by the enzyme, importantly the increase from baseline was similar in both normoxia and hypoxia (,46.20 pmol/ml & ,44.86 pmol/ml, respectively) ( Figure 1d).
To deduce whether superoxide (O 2 2 ) was having a direct influence on sGC activity in normoxia, SOD and CAT were incubated with the enzyme to observe any change in cGMP levels produced. Compared to control levels, SOD and CAT did not significantly affect the activity of the isolated enzyme ( Figure 1e).

Effects of Buffer Containing Increasing O 2 on Vascular Tone in Hypoxia
Tetrameric Hb has the ability to carry four O 2 molecules [25]. The O 2 content of a RBC suspension is therefore much greater than buffer equilibrated with the same O 2 concentration. In order to attain a comparable O 2 delivery, Krebs buffer was gassed with 95% O 2 /5% CO 2 , either in a sealed glass bottle (high O 2 saturation) or an open container allowed to equilibrate with atmospheric air (21% O 2 ), or with 95% N 2 /5% CO 2 (0% O 2 ). 200 ml of the appropriate Krebs buffer samples were then added to tissue baths containing hypoxic PE-pre-constricted aortic rings as described above. A 3 fold greater (p,0.01) relaxation was observed following addition of highly oxygenated versus minimally oxygenated samples (Figure 2).

Effects of Administering a Constant O 2 Supply to Hypoxic Tissue
In separate experiments, aortic tissue was incubated in hypoxia and pre-constricted with PE as described above. The gas supply was then switched to 95% O 2 /5% CO 2 permanently. An instantaneous and transient relaxation response, akin to that

Influence of Endothelial NO Synthase (eNOS) on O 2mediated Vasorelaxation
Experiments were repeated in the presence of the eNOS inhibitor L-N G -monomethyl arginine (LNMMA, 300 mmol/L for 30 minutes). Figure 4 demonstrates that this agent has no effect on the oxygenated buffer-induced relaxations (n = 4, **p,0.01 cf. 0% O 2 ).

Vascular Effects of Superoxide and Hydrogen Peroxide
It is well acknowledged that reperfusion of O 2 to hypoxic vessels causes the generation of O 2 radical species [26]. Therefore, it was necessary to ascertain whether O 2 2 had a role in the myography experiments conducted here. SOD and CAT linked to polyethylene glycol (PEG) were therefore used to abrogate any effects of intracellular O 2 2 and hydrogen peroxide (H 2 O 2 ) generation, respectively. Neither PEG-SOD, PEG-CAT, nor both together had a significant effect on the O 2 -induced relaxation of vessels in hypoxic conditions ( Figure 5).

Effects of Oxygenated RBCs on Vascular Tone in Hypoxia
RBCs equilibrated at a higher O 2 saturation (98.2260.45% O 2 , n = 13) caused significantly (p,0.001) more relaxation than those of partial saturation (51.4366.16% O 2 , n = 4) or low saturation (20.4065.28% O 2 , n = 13) when administered to a hypoxic tissue bath ( Figure 6). These data confirm a positive relationship between RBC O 2 saturation and the relaxation induced in hypoxic aortic tissue (Pearson's correlation coefficient r = 0.815, p,0.0001).

Effects of RBC O 2 Cycling upon Hypoxic Vasorelaxation
Highly oxygenated RBC (98.2260.45% O 2 , n = 13) were administered to hypoxic aortic rings to induce relaxation. In a parallel sample, highly oxygenated RBC were first deoxygenated to ,20% O 2 (20.6861.66% O 2 ) before reoxygenation (97.7060.39% O 2 , n = 4) and subsequent addition to hypoxic rings. There was no significant difference in the relaxation induced post deoxy-oxy cycling of RBCs compared with controls (8.2862.97% vs. 12.0063.60%).
The RBC O 2 cycling prompted further investigation into the movement of the relaxing moiety in and out of the cell. RBC were deoxygenated to a low saturation (22.3861.69% O 2 ) before resuspension in fresh oxygenated buffer to give a saturation of ,98% (98.0560.65% O 2 , n = 4). As above, there was no significant difference between the relaxations induced by buffer-   replaced RBCs compared with controls (12.9761.89% vs. 12.0063.60%).

Comparison of O 2 -mediated Vasorelaxation by RBCs, Buffer or Isolated Hb
In additional experiments, purified oxygenated Hb was prepared from RBCs and applied to tissue baths containing hypoxic PE-pre-constricted aortic rings as described above. Samples of RBC, buffer or Hb were prepared such that addition to the tissue bath resulted in final O 2 concentrations of ,30 mmol/ L. Figure 7 shows grouped results confirming very similar relaxation responses for RBCs, Hb and oxygenated buffer (n = 4, p.0.05).

Effects of Oxygenated RBCs on Vascular Tone at Different Tissue pO 2
In separate experiments, aortic tissue was perfused for 10 minutes with 1%, 2%, 5%, 21% or 95% O 2 /5%CO 2 . Constriction to PE was normalised to 80% of maximum at each O 2 tension, then fully oxygenated RBCs were administered. This is in contrast to most of the experiment described above in which varying O 2 was introduced to hypoxic tissue or isolated sGC.   Significantly greater relaxation was induced at tissue incubated in 1% (***p,0.001 cf. 95% O 2 ) and 2% O 2 (*p,0.05 cf. 95% O 2 ) (n = 6) ( Figure 8).

Effects of O 2 on the Liberation of NO from NOC9
In order to test whether production of NO from NOC-9 was affected under normoxic or hypoxic conditions, we measured NO released from NOC-9 into both the gaseous and liquid phases of a PBS sample across a range of NOC-9 concentrations. Our data confirms that the quantity of NO detected in both the gas and liquid phases (quantified as area under the curve), are not significantly different under normoxic versus hypoxic conditions (n = 3, p,0.05) (Figure 9).

Discussion
The aim of the present study was to investigate the effects of O 2 on sGC function, and in particular, on responses of the enzyme to known ligands and stimulators/activators. Our data conclusively shows that molecular O 2 has the ability to increase sGC activity in the absence of NO. Moreover, exposure of sGC to hypoxic conditions revealed a novel modulatory effect of O 2 on NOstimulated sGC activity. Our isolated vessel model confirms that the sGC-mediated transient relaxation of tissues in hypoxic conditions can occur simply by injecting a bolus of oxygenated buffer. As such, the data presents an important and hitherto undescribed role for O 2 in the production of cGMP.
As previously described, and when compared to NO, sGC can be activated to a lesser extent by other diatomic species such as CO [28]. However, despite the obvious structural similarity, the ability of O 2 to interact with sGC, and specifically with the haem moiety to form a stable complex, has been overlooked [6][7][8]29,30]. Importantly, the data described herein goes against conventional thinking and supports a role for O 2 in upregulating baseline sGC activity as evidenced by greater cGMP production in our isolated enzyme system under normoxic compared to hypoxic conditions. Moreover, the presence of O 2 is also shown to have a modulatory effect on the subsequent activation of sGC by NO. Thus following addition of NO to hypoxic sGC, similar large increases in cGMP were observed irrespective of NO concentration. Conversely, a concentration dependent response to NO was revealed for normoxic sGC, the largest response to 23.6 mmol/L NO being similar to that produced by 100-fold less NO under hypoxic conditions. As such, the removal of O 2 from sGC both decreases baseline enzyme activity and makes it more sensitive to activation by lower concentrations of the NO. That the NO and haemindependent sGC activator, YC-1, produced a different response, similar increases in cGMP above baseline being observed in both normoxia and hypoxia, would support a possible role for the interaction of O 2 with the sGC in a way that modulates the haem group. sGC has been reported to function optimally at intracellular O 2 concentrations between 20 and 40 mmol/L [31] indicating the ability of sGC to distinguish between ligands. Importantly, we confirmed that the amount of NO released by the NO donor NOC-9 was similar under both hypoxic/normoxic experimental conditions and thus cannot be accountable for the results achieved. While a stable isolatable complex of O 2 and sGC has not been identified, it is possible that intracellular concentrations of O 2 could have an influence on enzyme activity. The majority of studies conducted with sGC are under anaerobic conditions [8,32,33] however Ullrich and colleagues [34] observed a decrease in cGMP produced by platelet sGC following N 2  addition compared to basal conditions. In addition, Soret band analysis by UV spectrophotometry show subtle changes in sGC binding under oxygenated conditions [35].
In order to substantiate our isolated enzyme data, we next developed an in vitro vascular model. Isolated rabbit aortic rings pre-constricted with PE under hypoxic conditions were exposed to single bolus additions of buffer equilibrated with varying O 2 concentrations. In such experiments, transient and concentrationdependent relaxation responses to O 2 were produced. The involvement of sGC and classic cGMP-dependent downstream signalling in smooth muscle to induce relaxation was confirmed by the fact that the O 2 -dependent responses were inhibited in the presence of ODQ, the irreversible haem-site inhibitor of sGC [36,37]. Since eNOS function is close to normal even under hypoxic conditions in this vascular preparation [38] a role for endothelium-dependent NO in the responses described above was also investigated. In that the relaxation induced by O 2 was insensitive to eNOS inhibition by LNMMA, such an involvement was ruled out. Together these observations support the role of O 2 in sGC activation already described above.
The in vitro isolated vascular system used in the present study relates closely to ischaemia reperfusion, where a vessel of low oxygenation quickly becomes reperfused with O 2 , for example following removal of an occlusion [39]. Under such circumstances many studies have shown evidence for the generation of free radical species [26,[40][41][42][43]. Taken with the recently observed inhibitory effect of O 2 2 on sGC [44] it was important to confirm that O 2 2/ H 2 O 2 were not involved in the sGC-driven effects we observe following addition of O 2 to the tissues. To this end experiments were repeated in the presence of the appropriate cell permeable inhibitors (PEG-SOD and PEG-CAT). That the relaxation responses to O 2 were again insensitive to these agents ruled out this possibility.
In order to further translate our study, we repeated our in vitro studies, substituting the O 2 buffer bolus for highly O 2 -saturated human RBCs. When oxygenated, RBCs have the ability to change Hb conformation into the R state. Indeed, in this state Hb has the capacity to offload O 2 and in vivo this is conveyed by the transfer of O 2 from the blood to respiring tissue beds. Importantly, such a mechanism is crucial for RBCs to give up O 2 when encountering a hypoxic environment. Therefore, by altering the saturation of RBCs, we could change the proportions of R and T state Hb and therefore potentially show the preference for the RBCs to give up O 2 in the R state. The highly saturated (98.2260.45% O 2 ) RBCs induced a significantly greater relaxation of hypoxic rabbit aortic rings compared with RBCs of low saturation (20.4065.28% O 2 )/ mainly T state Hb. Consequently, purified Hb was used in further experiments to confirm the relaxant species was released directly from Hb without the need for RBCs per se. Interestingly, RBCs, Hb and buffer boluses of comparable O 2 content induced very similar relaxations of the hypoxic vessel tested in the present study. Additional experimentation demonstrated that the relaxant species is not depleted by repeat deoxygenation/re-oxygenation cycles, confirming the species is readily replenishable and is essentially delivered by simply replacing the Krebs buffer surrounding the hypoxic RBCs with adequately oxygenated buffer.
For some time it has been suggested that RBC-induced vasorelaxation is linked to Hb allostery, although a particular interest has focussed on the now well established vasodilatory capacity of HbSNO. We ourselves followed this line of investigation demonstrating greater RBC-induced relaxations with increasing HbSNO [45], and an altered HbSNO to relaxation response in highly glycosylated RBCs that we linked to an effect on the Hb O 2 saturation curve [11]. Concurrently, we also measured the transient release of O 2 into the tissue bath but at the time took this to be a consequence of Hb transition rather than the relaxant species. Subsequent studies have later confirmed the potential for O 2 in buffer alone to induce relaxation, but considered this control considerably less than the relaxation induced by RBCs [15]. Importantly, the samples and the buffer controls were perfused with the same gas, but crucially were not equalised in terms of O 2 content in the latter study.
Further studies have confirmed the importance of Hb allostery in RBC-mediated hypoxic vasorelaxation [20,46] although they contend partially oxygenated Hb acts as a NO 2 2 reductase that converts readily available NO 2 2 to free NO to exhibit vasodilatation. Whilst both HbSNO and NO 2 2 exhibit properties that suggest they could contribute to RBC-induced relaxation [15,[18][19][20]45], neither completely satisfies the characteristics of a readily replenishable, freely diffusing relaxant species that is present in simply oxygenated buffer as we show here.
A potential criticism of our model system is that we induce an artificial condition in tissue whereby a defined unit of O 2 (either buffer or RBCs) is introduced to an essentially hypoxic environment. To address this, we conducted studies under conditions where either fully oxygenated RBCs or buffer samples were administered to baths containing tissue equilibrated at various O 2 concentrations. Of physiological significance, O 2 -mediated relaxation was inversely related to the extent of tissue oxygenation, in complete concurrence with established HbO 2 delivery and matching the need for dilation across the whole O 2 range. Furthermore, we conducted studies in which O 2 was confirmed as a relaxant of blood vessel segments since the first response to administration of a constant supply of O 2 being re-introduced into a hypoxic setting was transient relaxation.
In previous work we had demonstrated how vessel relaxation response to exogenous addition of NO is enhanced with lower O 2 tension [11]. This had been ascribed to the influence of O 2 on competing mechanisms, including direct effects on constriction. However, taken together with the data we show herein, would indicate a more direct interplay between the effects of NO and O 2 on sGC. Clearly, NO is the preferred ligand as shown by its ability to fully activate isolated sGC in our model system at a concentration as low as 10 mmol/L even in the presence of 210 mmol/L O 2 (''normoxia''). Under conditions where NO is relatively low, O 2 may influence sGC in two ways; the influence of weak sGC activation by O 2 may accrue greater importance, and the effect of NO will be enhanced.
In conclusion, our data has revealed the novel finding that in the presence of O 2 , sGC can stimulate an increased production of cGMP from GTP compared to under hypoxic conditions. The use of YC-1 in our isolated enzyme model clearly enhanced cGMP production however the compound was not affected by the presence of O 2 , perhaps suggesting the effects of O 2 on the enzyme may be occurring at a site independent to that of YC-1. Furthermore, we show that O 2 can modulate the effect of NO (concentration-dependent) upon the enzyme, indicative of an interaction of O 2 with the haem of sGC or an indirect involvement with haem via an allosteric site. Armed with the knowledge that ,6% of sGC within vascular tissue is stimulated by NO under normal conditions [47], our findings may have important consequences in the wider context of vascular relaxation via stimulation of sGC across the physiological and pathological O 2 range.

Author Contributions
Manuscript structure: DL PEJ. Conceived and designed the experiments: JD AGP DL PEJ. Performed the experiments: JD AGP. Analyzed the data: