Oxygen dependency of hydrogen sulfide-mediated vasoconstriction in cyclostome aortas

Hypoxic vasoconstriction (HVC) was first observed in the mammalian pulmonary vasculature by von Euler and Liljestrand(von Euler and Liljestrand,1946) and it is now generally accepted that in mammals this response is unique to the pulmonary circulation, whereas hypoxic vasodilation(HVD) is the prominent response of systemic vessels(Weir and Archer, 1995). Although HVC and HVD may be modulated by endothelial-derived and/or circulating substances(Félétou et al.,1995; Jacobs and Zeldin,2001; Kerkhof et al., 2001; Liu et al., 2001; Aaronson et al., 2002; Deussen et al., 2006), the basic responses are intrinsic to the vascular smooth muscle cell(Madden et al., 1992). In non-mammalian vertebrates, HVC has been observed in both systemic and respiratory conductance vessels (Olson et al., 2001; Russell et al.,2001; Smith et al.,2001; Russell et al.,2007) and HVC appears to be an intrinsic response of vascular smooth muscle cells in the cyclostome dorsal aorta as well(Olson et al., 2001). HVD has only recently been systematically examined in non-mammalian vertebrates(Russell et al., 2007) and while it is common in systemic vessels it is not necessarily the predominant response.

How vascular smooth muscle cells `sense' hypoxia and transduce this into a mechanical response, either HVC or HVD, is unknown. We recently proposed that the metabolism of H₂S is involved in the O₂-sensing signal transduction process. Our model is based on the balance between constitutive cellular production of vasoactive hydrogen sulfide(H₂S) and its oxidation to inactive products by available O₂ (Olson et al.,2006). Furthermore, this model appears to be applicable to both HVC and HVD and evidence for a H₂S-mediated hypoxic relaxation has even been observed in the trout urinary bladder(Dombkowski et al., 2006).

There is also relatively little information on the mechanism through which H₂S elicits mechanical responses in the vasculature. H₂S-mediated vasodilation has been demonstrated in mammalian systemic vessels and at least part of this response is due to H₂S opening of ATP sensitive potassium (K_ATP) channels on the vascular smooth muscle cell and to release of nitric oxide (NO) from the endothelium(Zhao et al., 2001; Zhao and Wang, 2002; Wang et al., 2004). H₂S-mediated vasoconstriction has been demonstrated in mammalian pulmonary vessels (Olson et al.,2006) and in a variety of both pulmonary and systemic vessels from non-mammalian vertebrates (Dombkowski et al., 2005). Although it is unlikely that H₂S contractions are mediated through either K_ATP channels or endothelial-derived vasoconstrictor substances, the mechanism(s) of H₂S-mediated vasoconstriction is unknown.

Recently, Koenitzer et al. (Koenitzer et al., 2007) examined the effects of H₂S on rat thoracic aortas at high (200 μmol l^–1, ∼150 mmHg) and low (40 μmol l^–1, ∼30 mmHg) partial pressures of O₂ and showed that vascular relaxation was more sensitive to H₂S at low oxygen concentration ([O₂]) and that H₂S-mediated contractions were present at high, but not low[O₂]. They postulated that the decreased sensitivity of the H₂S-mediated vasorelaxation at high [O₂] was due to the combined effect of rapid oxidation (and therefore inactivation) of vasodilatory H₂S plus the generation of a vasoconstrictor oxidation product of H₂S that would compete with the H₂S relaxation. Thus H₂S does not directly produce vasoconstriction. The identity of this oxidation product was not determined.

There are other possible explanations for the results of Koenitzer et al.(Koenitzer et al., 2007) that seem equally or more plausible. First, Koenitzer et al.(Koenitzer et al., 2007) only examined rat aortas and these vessels relax when exposed to either hypoxia or lower (and perhaps more physiological?) concentrations of H₂S and thus a contraction would not normally be expected. Second, our theory of H₂S metabolism in vascular O₂ sensing predicts that as[O₂] falls, endogenous [H₂S] increases. Thus at low P_O2 we would expect greater sensitivity to exogenous H₂S when applied against a background of elevated endogenous H₂S, consistent with the observations of Koenitzer et al. (Koenitzer et al., 2007). Furthermore, we also think that H₂S directly produces vasoconstriction in vessels that exhibit hypoxic vasoconstriction (e.g. hagfish and lamprey aortas) because it seems unlikely to us that cellular concentrations of an oxidation product of H₂S would be increasing when P_O2 is falling.

In the present study we examined the interaction between [O₂]and [H₂S] in the dorsal aorta of the most ancient extant craniate,the hagfish. This vessel was chosen because it has a mono-phasic,[O₂]-dependent HVC that is endothelium independent, and does not involve K_ATP channels, products of lipoxygenase, cyclooxygenase,cytochrome P₄₅₀ enzyme activity, or α-adrenergic, muscarinic,nicotinic, purinergic or serotoninergic receptors(Olson et al., 2001). If our hypotheses that H₂S directly produces vasoconstriction and that endogenous H₂S increases when P_O2falls is correct, we expect to see an increased sensitivity of H₂S-constriction at low [O₂], not an unmasking of H₂S relaxation. We also provide additional evidence for H₂S metabolism in the O₂ sensing mechanism by examining the contribution of its precursor, cysteine, and the effects of inhibitors of H₂S synthesis on HVC. The effects of H₂S, cysteine and enzyme inhibitors on vessel O₂ consumption were measured to determine whether H₂S exposure increased Ṁ_O2 and inhibitors decreased it. For comparison, we examined the O₂sensitivity of H₂S contraction in lamprey aortas. These vessels are identical to hagfish aortas in their response to hypoxia(Olson et al., 2001) and evidence for the role of H₂S in O₂ sensing has been described (Olson et al.,2006).

New Zealand hagfish, Eptatretus cirrhatus Forster (1122±165 g, N=26) were collected in Akaroa Harbour, New Zealand, and held in 14°C seawater aquaria at the University of Canterbury, Christchurch, NZ for at least 1 week prior to use. They were anesthetized in a combination of AQUI-S™ (200 p.p.m.; Lower Hutt, New Zealand), benzocaine (400 p.p.m.)and MS-222 (400 p.p.m.). The ventral and dorsal aortas (VA and DA) and afferent and efferent branchial arteries (ABA and EBA) were removed cleaned of excess fat and blood and placed in 4°C hagfish Hepes saline until use(within 1–2 days). The saline was changed daily prior to use.

Sea lamprey (Petromyzon marinus L.; 130–450 g) were captured by the US Geological Survey, Biological Resources Division, in Michigan during the spring-summer spawning migration and airlifted to Indiana University School of Medicine–South Bend (IUSM-SB). They were housed in 500 l rectangular tanks in aerated, flowing well water (15°C), and exposed to a 12 h:12 h light:dark photoperiod. They were not fed. Lamprey were anesthetized in benzocaine (1:5000, wt:vol), and the vessels were dissected out and placed in Cortland buffered saline at 4°C until use.

Salmon (Oncorhynchus tshawytscha Walbaum) were obtained from a nearby hatchery, anaesthetized with 22 p.p.m. AQUI-S™ in their holding tanks and then killed by pithing the brain and proximal spinal cord. The ventral aorta and afferent branchial arteries were rapidly excised and stored at 4°C in freshwater salmon Ringer's solution; the dorsal aorta is firmly attached to the vertebral column and cannot be removed intact. Storage and preparation of the salmon vessels for respirometry was identical to that described below for hagfish.

Experimental procedures were approved by the University of Canterbury's Animal Ethics Committee and the IUSM-SB IACUC.

Vessels were cut transaxially into 3–4 mm long segments, mounted on 280 μm diameter stainless-steel hooks and suspended in 20 ml,water-jacketed (12°C) smooth muscle chambers and bubbled with room air. Tension was measured with Grass FT03C force-displacement transducers (Grass Instruments, West Warwick, RI, USA) and collected electronically using Biopac model MP35 (Biopac Systems Inc., Goleta, CA, USA), or measured with MLT0210 isometric force transducers (ADInstruments, Castle Hill, Waverley, NSW,Australia) using Powerlab® systems with bridge amplifiers (ADInstruments). Data were archived at 2 Hz on notebook computers. A resting tension of 500±50 mg (Olson et al.,2001) was applied to the vessels for 45–60 min prior to experimentation. Vessels were maximally contracted with the acetylcholine analog carbamylcholine chloride (carbachol, 10 μmol l^–1; Fig. 1) until tension plateaued(30–45 min), then rinsed four times with buffer and resting tension re-established over the ensuing 60 min. They were then contracted a second time with 10 μmol l^–1 carbachol, the rinse repeated, and the vessels were allowed to stabilize and resting tension re-established for the next 1–2 h. The tension produced by the second application of carbachol was used as the reference contraction for subsequent experiments.

Cumulative carbachol dose–response curves were initially obtained for efferent branchial arteries (Fig. 1) and dorsal aortas (not shown). Carbachol at 10 μmol l^–1 produced maximum contraction in both vessels. Therefore,carbachol at this concentration was used twice at the beginning of experiments, first for initial activation and second for a reference contraction. A third carbachol (10 μmol l^–1) was also applied at the end of many experiments to determine if other treatments had non-specific effects on vascular reactivity.

The dissolution of Na₂S in water produces H₂S and HS^– (collectively referred to in this study as H₂S) and increases pH. Because extracellular alkalinity can contract vascular smooth muscle independently of other exogenous stimuli(Smith et al., 2006), the buffering capacity of Hepes samples over the range of 1 μmol l^–1 to 10 mmol l^–1 H₂S was measured in triplicate using an Orion 911600 semi-micro pH electrode (Beverly,MA USA) and a PHM 84 pH meter (Radiometer, Copenhagen, Denmark).

Cumulative H₂S dose–response curves (1 μmol l^–1 to 1 mmol l^–1) were obtained for otherwise unstimulated, normoxic (bubbled with room air; 21% O₂)dorsal aortas and efferent branchial arteries. To determine if H₂S relaxed vessels, a second series of experiments were conducted with the vessels pre-contracted with 150 mmol l^–1 KCl or 0.3 μmol l^–1 carbachol prior to the H₂S doses. In pilot studies, H₂S had no effect on ventral aortas or afferent branchial arteries (N=4) and these vessels were not examined further.

The effect of graded hypoxia on the H₂S dose response of hagfish dorsal aortas was examined by initially aerating groups of vessels with either 100% room air (P_O2=157 mmHg), 6% air/94%N₂ (P_O2=10 mmHg), or 100%N₂ (P_O2<1 mmHg) for 20–30 min prior to and during the H₂S treatments (1 mmHg=0.133 kPa). The air/N₂ mixture was controlled with a Wöstoff type 1M 300/a-F gas mixing pump (H. Wöstoff, Bochum, Germany). The P_O2 was measured in one myograph chamber using a Microelectrodes MI-730 oxygen electrode and meter (Bedford, NH).

The effect of moderate hypoxia on the H₂S dose response of lamprey dorsal aortas was examined using the following protocol. Vessels were contracted twice with 80 mmol l^–1 KCl, washed twice after each contraction, and then vigorously bubbled with 100% N₂ to produce a maximal HVC. After recovery (normoxia) the flow of N₂ was reduced to produce HVC that was 20±6% of the maximal HVC. This is equivalent to a bath P_O2 of 20–30 mmHg(Olson et al., 2001). Cumulative doses of H₂S (10 nmol l^–1–1 mmol l^–1) were applied during this moderate hypoxia.

The involvement of H₂S in hagfish HVC was examined by measuring the response of hypoxia-contracted (100% N₂) aortas to serial additions of the substrate for H₂S synthesis: l-cysteine(0.1, 1 and 10 mmol l^–1), amino-oxyacetate (AOA; 0.1, 1 and 4 mmol l^–1), an inhibitor of cystathionine β-synthase(CBS), d,l-propargylglycine (PPG; 0.1, 1 and 4 mmol l^–1), an inhibitor of cystathionine λ-lyase, or hydroxylamine (0.01, 0.1 and 1 mmol l^–1), a general inhibitor of pyridoxyl 5′-phosphate-dependent enzymes. Following a standard carbachol (10 μmol l^–1) contraction the vessels were thoroughly washed and gassed with 100% N₂ for 20–30 min until the hypoxic contraction stabilized. Cumulative doses of cysteine or inhibitors were applied during the hypoxic contraction followed by a final application of 10 μmol l^–1 carbachol. Vessels were not washed prior to the final carbachol. The effects of hypoxia, cysteine, inhibitors and final carbachol were normalized relative to the reference carbachol contraction.

The relationship between P_O2 and Ṁ_O2 was measured in vessels that were allowed to deplete the oxygen content from air saturation(P_O2 ∼155 mmHg) down to zero. These experiments showed that the vessels could efficiently regulate their Ṁ_O2 between a P_O2 of 15 and 115 mmHg (see Results). Subsequent experiments on the effects of H₂S, cysteine and inhibitors of H₂S production were performed between 40 and 60 mmHg P_O2 where Ṁ_O2 was otherwise independent of P_O2.

A cumulative H₂S concentration–Ṁ_O2response was established for 1 μmol l^–1–1 mmol l^–1 H₂S in both un-contracted and carbachol (100μmol l^–1)-contracted vessels. The effects of the H₂S precursor, l-cysteine (1 and 1 mmol l^–1) and inhibitors of H₂S production, AOA and HA(both 10 μmol l^–1–1 mmol l^–1), on Ṁ_O2 were also determined.

In the initial studies on hagfish vessels it became evident that contracting the vessels with carbachol had no effect on oxygen consumption. This was unexpected as it has been shown that contracting mammalian vessels increases oxygen consumption (Koenitzer et al., 2007). To determine whether this was an actual physiological difference or an experimental artifact, we repeated the studies with ventral aortas and afferent branchial arteries isolated from chinook salmon using the general protocol described above for hagfish vessels.

The composition of hagfish Hepes-buffered saline was (in mmol l^–1): 497.95 NaCl, 8.05 KCl, 5.10 CaCl₂, 9.00 MgCl₂, 3.04 MgSO₄, 3.00 Hepes[N-(2-hydroxyethyl)piperazine-N=–(2-ethane-sulfonic acid)] acid form, 6.99 Hepes sodium salt, 5.55 glucose, pH 7.8. The composition of lamprey Cortland saline was (in mmol l^–1): 124 NaCl, 3 KCl, 2 CaCl₂, 0.57 MgSO₄, 12 NaHCO₃,0.09 NaH₂PO₄, 1.8 Na₂HPO₄, 5.5 glucose, pH 7.8. The composition of salmon Ringer was (in mmol l^–1; 136.89 NaCl; 2.11 KCl; 0.99 MgCl₂; 1.30 CaCl₂; 3.00 Hepes acid form; 6.99 Hepes sodium salt; 0.30 sodium glutamate; 0.40 l-glutamine; 0.02 sodium aspartate; 0.05 dl-carnitine; 10.00 glucose, pH 7.6. AOA was purchased from ACROS Organics (Morris Plains, NJ, USA) all other chemicals were purchased from Sigma Chemical Co. (St Louis, MO, USA).

Concentration response curves were expresses as a percentage of the maximal response. Vessel responses were normalized to the second carbachol contraction produced prior to experimentation. At the end of an experiment the vessel was blotted on paper toweling, weighed and vessel tension was normalized to wet mass, i.e. mg tension g^–1 wet mass. Because the hypoxic responses of individual vessels were reproducible(Olson et al., 2001), each vessel served as its own control and treatment effects were statistically examined by paired t-test or repeated measures tests. Results are presented as mean ± s.e.m. Student's t-test and analysis of variance (ANOVA) were used for comparisons between vessels. Significance was assumed when P⩽0.05.

Significant differences in rates of Ṁ_O2 and responses to drugs were determined using a repeated measures ANOVA. Where significant differences were calculated between means, Tukey's post-hoc tests showed which means were significantly different from each other. Paired Student's t-tests were used to detect differences between carbachol-treated and -untreated vessels in the Ṁ_O2 data(controls and at each concentration of H₂S). Significance was assumed when P⩽0.05. All analyses were performed in Prism 4.00(Graphpad software, San Diego, CA, USA).

Carbachol produced a dose-dependent contraction of efferent branchial arteries (Fig. 1) with an EC₅₀ of 0.275±0.148 μmol l^–1(N=4). Peak force was achieved at a concentration of 10 μmol l^–1 carbachol.

The effects of increasing concentrations of H₂S on pH of hagfish Hepes buffer is shown in Fig. 2. Buffering was very efficient between 1 μmol l^–1 and 1 mmol l^–1 H₂S and increased less than 0.3 pH unit between 1 mmol l^–1 and 3 mmol l^–1 H₂S. However, pH increased nearly 2.5 units between 3 and 10 mmol l^–1 H₂S. H₂S contractions appeared to be independent of pH between 1 μmol l^–1 and 1 mmol l^–1 H₂S, but concentrations above 1 mmol l^–1 alkalinized the medium and this appeared to greatly augment H₂S contractions. H₂S concentrations were limited in subsequent experiments to 1 mmol l^–1 in order to minimize the possibility of pH interference and to avoid complications that might result from changes in ionic strength or composition due to increasing the buffering capacity, or titration.

H₂S produced dose-dependent contractions in otherwise unstimulated dorsal aortas and efferent branchial arteries of hagfish(Fig. 3) but had no affect on ventral aortas or afferent branchial arteries (not shown). H₂S produced essentially identical dose-dependent contractions in KCl (150 mmol l^–1, N=4) and carbachol (0.3 μmol l^–1, N=4) pre-contracted dorsal aortas and in carbachol (0.3 μmol l^–1, N=4) pre-contracted efferent branchial arteries (data not shown for pre-contracted vessels). There was no obvious H₂S-mediated relaxation in either pre-contracted or otherwise unstimulated vessels.

Hagfish dorsal aortas bubbled with 100% N₂ were significantly more sensitive to low H₂S concentrations than aortas bubbled with room air and the H₂S dose–response curve of hypoxic (anoxic)vessels appeared to have two components(Fig. 3). H₂S-mediated contractions of hagfish aortas bubbled with 6% air/94%N₂ (data not shown) were not significantly different from aortas bubbled with room air.

H₂S produced dose-dependent contractions of lamprey dorsal aortas (Fig. 3). Moderate hypoxia increased H₂S sensitivity between 10 and 300 μmol l^–1 H₂S (P value at 30 μmol l^–1 H₂S was 0.053). Hypoxia did not affect the magnitude of the 1 mmol l^–1 H₂S contraction which in hypoxia was 41±5% and in normoxia 43±4% of a 80 mmol l^–1 KCl contraction. H₂S at concentrations between 10 nmol l^–1 to 1 μmol l^–1 did not affect either normoxic or hypoxic vessels (not shown).

In these experiments vessels were contracted with 10 μmol l^–1 carbachol, washed, then continuously contracted with 100%N₂ aeration. During the hypoxic contraction, the vessels were given cumulative additions of cysteine or inhibitors and this was followed, without washing the vessels, by 10 μmol l^–1 carbachol.

The effect of l-cysteine, a substrate for H₂S synthesis, on hypoxic contractions of hagfish dorsal aortas is shown in Fig. 4, top left panel. 100μmol l^–1 cysteine produced a consistent, but statistically insignificant increase in the force of the N₂ contraction. Increasing cysteine to 1 mmol l^–1 significantly(P<0.05) contracted the vessels to approximately double that in the original N₂ contraction. Raising cysteine to 10 mmol l^–1 produced an immediate relaxation back to the pre-cysteine(N₂) level (P<0.05). The carbachol (10 μmol l^–1) contraction at the end of the experiment, in the presence of N₂ and 10 mmol l^–1 cysteine, was not significantly different from the reference carbachol contraction(90±14% of reference, N=7).

The effect of the cystathionine β-synthase (CBS) inhibitor,amino-oxyacetate (AOA), on hypoxic vasoconstriction of the dorsal aorta is shown in Fig. 4 top right panel. Hypoxic contractions were unaffected by 100 μmol l^–1 and 1 mmol l^–1 AOA. 4 mmol l^–1 AOA completely inhibited the hypoxic contraction(P<0.05). The carbachol (10 μmol l^–1)contraction at the end of the experiment, in the presence of 4 mmol l^–1 AOA, was not significantly different from the reference carbachol contraction (111±12% of reference, N=8).

As shown in Fig. 4, lower left panel, the cystathionine λ-lyase (CSE) inhibitor, propargyl glycine (PPG; between 100 μmol l^–1 and 4 mmol l^–1) had no effect on the hypoxic contraction. A carbachol(10 μmol l^–1) contraction at the end of the experiment, in the presence of PPG, was similarly unaffected (104±7% of reference, N=8).

Hydroxylamine (HA), an uncoupler of pyridoxyl 5′-phosphate-dependent enzymes including CBS and CSE, at 10 μmol l^–1significantly increased the force of the N₂ contraction(Fig. 4, lower right panel). Increasing HA to 100 μmol l^–1 and 1 mmol l^–1 produced slight, but statistically insignificant, further increases in tension. The carbachol (10 μmol l^–1)contraction at the end of the experiment, in the presence of HA, was not significantly different from the reference carbachol contraction(103±12% of reference, N=8).

The relationship between P_O2 and oxygen consumption (Ṁ_O2)in uncontracted and carbachol pre-contracted hagfish dorsal aortas is shown in Fig. 5A. Ṁ_O2 was well maintained around 2.4 pmol mg^–1 min^–1between 15 and 115 mmHg P_O2 but doubled between 115 and 155 mmHg and fell to zero as P_O2approached zero. Ṁ_O2 fell to 90%of the regulated rate at a P_O2 of 12 mmHg and the P_O2 at which the regulated Ṁ_O2 fell to half(P₅₀) was 3 mmHg. Pre-treating hagfish aortas with 100 μmol l^–1 carbachol did not significantly affect Ṁ_O2. Ṁ_O2 was also well-regulated in unstimulated salmon vessels between 15 and 115 mmHg P_O2 but the rate of oxygen consumption per unit tissue mass was five times that of hagfish aortas(Fig. 5B). Pre-treating salmon vessels with 100 μmol l^–1 carbachol nearly doubled Ṁ_O2 at all but the lowest P_O2.

The effects of H₂S, cysteine, AOA and HA on Ṁ_O2 are summarized in Fig. 6. 10μmol l^–1 H₂S significantly stimulated Ṁ_O2 whereas 100μmol l^–1 and 1 mmol l^–1 significantly inhibited Ṁ_O2. A 12-fold increase in Ṁ_O2 was produced by 10 mmol l^–1 cysteine; in many experiments 1 mmol l^–1 cysteine often appeared to increase Ṁ_O2 as well,although this was not statistically significant. Ṁ_O2 was inhibited by either 10 mmol l^–1 AOA or 10 mmol l^–1 HA. Other concentrations of AOA (10 μmol l^–1–1 mmol l^–1) and HA (10 μmol l^–1–1 mmol l^–1) did not significantly affect Ṁ_O2. Pre-contraction with 100 μmol l^–1 carbacol did not significantly affect Ṁ_O2 in vessels treated with 10 μmol l^–1, 100 μmol l^–1or 1 mmol l^–1 H₂S(Fig. 6), 10 mmol l^–1 cysteine, 10 mmol l^–1 AOA or 10 mmol l^–1 HA (N=5 for all; data not shown), although Ṁ_O2 of vessels in 10 μmol l^–1 H₂S was significantly greater than Ṁ_O2 of vessels in 100 μmol l^–1 H₂S(Fig. 6).

The present studies support our hypotheses that, (1) H₂S directly produces vasoconstriction, and (2) the metabolism of H₂S is involved in the oxygen sensing and/or signal transduction cascade in hypoxic vasoconstriction of the hagfish aorta.

As shown in Fig. 3, the H₂S sensitivity in both hagfish and lamprey aortas increased when P_O2 was decreased. Hagfish vessels bubbled with 100% N₂ responded to 1 μmol l^–1 H₂S,whereas 100 μmol l^–1 H₂S was required to contract vessels bubbled with room air(Fig. 3). Similarly, the apparent H₂S thresholds for hypoxic(P_O2 ∼20–30 mmHg) and normoxic lamprey vessels were 10 and 30 μmol l^–1 H₂S,respectively (Fig. 3). Furthermore, the H₂S dose–response curves for both animals were left-shifted in low P_O. Thus although the effect of hypoxia on the H₂S response of lamprey vessels was less dramatic than that of hagfish vessels (probably because the hypoxia was less severe),the basic responses were, nevertheless, quite similar. As described below,these results support the hypothesis that H₂S has direct vasoconstrictory activity in specific vessels.

Koenitzer et al. (Koenitzer et al.,2007) observed a bi-phasic effect of H₂S on rat aortas;low H₂S concentrations produced dilation and high concentrations (1 mmol l^–1) produced contraction. They also found that H₂S-mediated dilation of rat aortas became more sensitive to H₂S at low P_O2. These authors(Koenitzer et al., 2007)suggested that the H₂S-mediated vasoconstriction of aortas bubbled with room air was due to an oxidation product of H₂S, not H₂S itself, and that the reason rat aortas became more sensitive to H₂S during hypoxia was because in the absence of this putative oxidation product there was no offsetting constrictory stimulus to compete with the direct H₂S dilation. Our findings argue against these hypotheses. First, H₂S only constricted hagfish and lamprey dorsal aortas and therefore the increased sensitivity observed at low P_O2 could not be due to removal of a competing(in this case dilatory) process. Second, it seems unlikely that production of this hypothetical vasoconstrictory oxidation product of H₂S would increase when the vessels are bubbled with 100% N₂. An alternative,and we think more plausible, explanation for the increased H₂S sensitivity, and one that is consistent with our(Olson et al., 2006)hypothesis of H₂S involvement in HVC (see below), is that when P_O2 falls endogenous H₂S increases. Therefore, less exogenous H₂S is required for vasoconstriction.

Ali et al. (Ali et al.,2006) and Kubo et al. (Kubo et al., 2007) observed the opposite effects of Koenitzer et al.(Koenitzer et al., 2007), i.e. low H₂S concentrations (<200 μmol l^–1)contracted, and elevated H₂S concentrations (200–1600 μmol l^–1) relaxed rat aortic rings. They attributed the low-dose H₂S contraction to H₂S combining with NO and thereby removing the tonic NO-mediated vasodilation(Ali et al., 2006), or directly inhibiting endothelial nitric oxide synthase(Kubo et al., 2007). This also is unlikely to occur in either hagfish or lamprey dorsal aortas because, (1)there is at present no evidence for endothelial NO production by either of these vessels (Olson et al.,2001), (2) exogenous NO produces a modest contraction in the ventral aorta of the hagfish, Myxine glutinosa(Evans and Harrie, 2001), and(3) NO synthesis from l-arginine and O₂ would be expected to be reduced during prolonged hypoxia. Our studies suggest that H₂S may directly constrict specific vessels and that this response is an intrinsic property of the smooth muscle cells. Clearly, however,variations in this response can be achieved through H₂S interactions with other vasoregulatory mechanisms.

Our model of the role of H₂S metabolism in oxygen sensing and/or signal transduction appears to accommodate both hypoxic vasoconstriction (HVC)and hypoxic vasodilation (HVD) in vertebrate smooth muscle (Olson et al.,2007). This model is based on the balance between H₂S production by vascular tissue and its inactivation through oxidation, and it provides a simple and rapid mechanism that couples the concentration of a vasoactive molecule directly to P_O2. The model is supported by observations that the responses of a wide variety of vessels(either constriction, dilation or multi-phasic) to hypoxia and H₂S are identical, H₂S is constitutively produced by blood vessels,cysteine the metabolic precursor of H₂S, augments HVC and inhibitors of H₂S production inhibit HVC and HVD. The present study provides additional support for the involvement of H₂S in HVC in hagfish vessels.

The responses of New Zealand hagfish vessels to H₂S are in many respects similar to those produced by hypoxia. H₂S and hypoxia(Olson et al., 2001), appear to be exclusively vasoconstrictory in hagfish dorsal aortas and efferent branchial arteries because they consistently contracted both un-stimulated and pre-contracted vessels. Conversely, neither H₂S nor hypoxia produced a sustained response in ventral aortas or afferent branchial arteries. H₂S and hypoxic(Olson et al., 2001)contractions of aortas and efferent branchial arteries were also unaffected by pre-contraction with KCl. Furthermore, because KCl pre-contraction presumably depolarizes smooth muscle cells it is likely that the mechanism of H₂S excitation is independent of cell depolarization; evidence for depolarization-independent HVC in these vessels has also been presented previously (Olson et al.,2001). This is in contrast to the H₂S-mediated relaxation of rat aorta (Zhao et al.,2001; Zhao and Wang,2002) and trout efferent branchial arteries(Dombkowski et al., 2004) where elevated KCl partially inhibits the response. Collectively, these findings suggest that H₂S contraction and HVC have a common, or at least similar, excitation pathway in hagfish vessels. This is consistent with other studies (Olson et al., 2001; Dombkowski et al., 2004; Dombkowski et al., 2005; Olson et al., 2006; Russell et al., 2007) that have shown that the vascular response to hypoxia is identical to that of H₂S irrespective of whether this response is contraction,relaxation, multi-phasic, or, as in the case of hagfish ventral aorta and afferent branchial arteries, no response at all. We have also observed identical hypoxic and H₂S responses in trout urinary bladder(Dombkowski et al., 2006) and thus hypoxia and H₂S appear to have a common, or at least similar excitation pathway in vertebrate smooth muscle in general.

Cysteine, which is presumed to be the precursor of H₂S production in animals (Julian et al.,2002) increases the magnitude of HVC at lower concentrations(Fig. 4) suggesting that it increases tissue production of H₂S. This is consistent with a cysteine-enhanced HVC in lamprey dorsal aortas and bovine pulmonary arteries and enhanced HVD observed in rat thoracic aortas(Olson et al., 2006). Further elevation of the cysteine concentration (10 mmol l^–1)inexplicably reduced the HVC in hagfish dorsal aorta. This may be due to a feedback-type inhibitory effect of cysteine on H₂S production, as we (R.D., S. Head, N. Whitfield and K.O., unpublished observation) have also observed elevated cysteine (10 mmol l^–1 or 100 mmol l^–1) inhibition of H₂S production in homogenized bovine heart or trout vessels, respectively, which is consistent with feedback inhibition. Alternatively, cysteine at a concentration of 10 mmol l^–1 may be toxic to smooth muscle cells. However, the fact that carbachol contractions on top of 10 mmol l^–1 cysteine were not significantly different from the reference contractions, suggests that cytotoxicity of cysteine at 10 mmol l^–1 is unlikely, and the reason for this inhibition remains to be identified.

The effects of inhibitors of H₂S production provide additional evidence for H₂S signaling in hypoxic responses. As shown in Fig. 4, amino-oxyacetate (AOA),an inhibitor of cystathionine β-synthase (CBS) completely inhibited HVC in hagfish aortas, whereas the cystathionine λ-lyase (CSE) inhibitor,propargyl glycine (PPG) was ineffective(Fig. 4). This suggests that HVC in the hagfish dorsal aorta is dependent upon H₂S synthesis via CBS. This is in contrast to mammalian systemic vessels where CSE,but not CBS, catalyzes H₂S production(Hosoki et al., 1997; Zhao et al., 2003). We(Olson et al., 2006) have also shown that inhibition of CSE, but not CBS, blocked HVD in rat aortas, whereas inhibition of CBS, but not CSE, blocked HVC in bovine pulmonary arteries. Interestingly, in trout, H₂S produces a tri-phasic relaxation-contraction-relaxation(Dombkowski et al., 2004) and these vessels appear to possess both CBS and CSE (G. Yang, R. Wang and K.O.,unpublished observation). These studies not only support the hypothesis of H₂S as a vascular O₂ sensor but they also provide additional evidence that different enzymes for H₂S production, CBS and CSE, may mediate HVC and HVD, respectively in different vessels.

Contractions produced by carbachol while hagfish dorsal aortas were exposed to AOA or PPG were not significantly different from the reference contraction produced by carbachol in the absence of inhibitors. Thus the inhibitory effect of AOA on HVC could not be due to general inhibition of the contractile apparatus. Separate H₂S and ligand-mediated responses have also been observed in other vessels (Dombkowski et al., 2004; Olson et al.,2006), indicating that the pathway for H₂S activation is not shared with some of the more common ligand-mediated mechanisms.

It is not clear why hydroxylamine potentiated HVC in hagfish aortas,although a non-specific effect seems likely. Hydroxylamine inhibits at least 100 enzymes that use pyridoxyl 5′-phosphate as a co-enzyme(Kery et al., 1999; Tang et al., 2005) including CBS and CSE. Although we expected it to act in a manner similar to AOA and inhibit vasoconstriction, it was the most potent constrictor tested, on a molar basis. Interestingly, despite turning the vasa vasorum brown, probably as a result of an action on heme groups(Canty and Driedzic, 1987; Nichols and Weber, 1989), and contracting the vessels, the vessels remained viable and the response to the final carbachol exposure was not diminished.

As shown in Fig. 5, hagfish aortas display a remarkable ability to maintain O₂ consumption(Ṁ_O2) over a wide range of ambient P_O2 (∼15–115 mmHg). It is not clear why the vessels lose their regulatory ability when P_O2 exceeds ∼115 mmHg, but this may be near the maximum P_O2 these vessels encounter in the wild, i.e. in air-saturated seawater (inspired P_O2 of 156 mmHg), arterial P_O2 was 109 mmHg(Forster et al., 1992). When P_O2 falls below ∼15 mmHg, Ṁ_O2 also falls. The P_O2 over which Ṁ_O2 decreases is quite similar to the P_O2 at which HVC increases(dashed line in Fig. 5).

The P_O2 at which Ṁ_O2 begins to decrease in hagfish dorsal aortas (15 mmHg) is somewhat less than the 20 mmHg critical P_O2(P_O2 at which Ṁ_O2 was reduced by 5%) in isolated rat aortas (Koenitzer et al., 2007). However, the Ṁ_O2 for rat aortas [78 pmol mg^–1 min^–1(Koenitzer et al., 2007)] is 32.5 times greater than the Ṁ_O2 for hagfish aortas (2.4 pmol mg^–1 min^–1). Even assuming a Q₁₀ of 2.4, the 25°C temperature difference between our study and that of Koenitzer et al. (Koenitzer et al., 2007) would only account for a sixfold difference in Ṁ_O2. In fact,these differences would probably be even greater if the O₂solubility coefficients were accounted for; mammalian (human) plasma at 37°C is 1.26 μmol l^–1 mmHg^–1 and seawater (with the same osmolarity of hagfish plasma) at 12°C is 1.72μmol l^–1 mmHg^–1(Boutilier et al., 1984).

Koenitzer et al. (Koenitzer et al.,2007) also showed that Ṁ_O2 more than doubled when rat aortas were contracted with phenylephrine. We did not find any difference in Ṁ_O2 between un-contracted and contracted hagfish aortas, perhaps because Ṁ_O2 was so low to begin with, or, more likely, because once hagfish aortas are contracted they are able to maintain tension with little additional energy expenditure. The latter point may be related to the hypoxia tolerance of hagfish vessels where hypoxic contractions can be sustained for 8 h of continuous aeration with 100%N₂ (Olson et al.,2001). Many non-mammalian vertebrates, especially the more`primitive' ones are considerably more hypoxia tolerant than mammals because of their ability to downregulate cellular metabolism and balance ATP demand with ATP supply (Boutilier,2001). Hypoxia tolerance varies across hagfish species and interestingly, E. cirrhatus does not voluntarily tolerate an ambient P_O2 of less than 45 mmHg (82 μmol l^–1) at 11°C (Forster, 1992). Clearly, the lack of an increase in Ṁ_O2was not due to the technique used as carbachol nearly doubled Ṁ_O2 in salmon.

Despite the elevated metabolic rate of rat aortas, the tension (in mg tension mg^–1 wet mass) produced by KCl contraction of rat aortas, which varies from 240 (Olson et al., 2001) to 720 (Resende et al., 2004) is only 2.5–7.5 times greater than a KCl contraction of hagfish dorsal aorta (94±12, N=4; data from this study). Thus is appears in rat vessels that either more oxygen is consumed for non-contractile-related activities, or that force development is energetically less efficient.

In many organisms, O₂ consumption is affected by H₂S. At low [H₂S], O₂ often increases because of the use of H₂S in mitochondrial ATP synthesis or for H₂S detoxification; at elevated [H₂S], O₂ consumption often decreases because of H₂S inhibition of mitochondrial cytochrome c oxidase (Grieshaber and Völkel, 1998), or perhaps even a general metabolic depression(Blackstone et al., 2005). H₂S also affects Ṁ_O2 in un-contracted hagfish aortas (Fig. 6) in a manner consistent with that described by Grieshaber and Völkel (Grieshaber and Völkel,1998). An increase in Ṁ_O2 is also predicted by our (Olson et al.,2006) model of H₂S oxidation by blood vessels as a mechanism to inactivate H₂S during normoxia. Higher (100 μmol l^–1 and 1 mmol l^–1) [H₂S]inhibits Ṁ_O2(Fig. 3) but not tension development (Figs 2, 3). This likely reflects the inherently low energy cost of force development, consistent with our observation that Ṁ_O2 does not change even during maximal carbachol contraction and it also provides a mechanism for sustaining HVC even when mitochondrial energy production is compromised.

The increase in Ṁ_O2 produced by cysteine and the decrease in Ṁ_O2 produced by AOA (Fig. 6) are also consistent with a positive and negative effect on H₂S production by hagfish dorsal aortas. It is not clear why 10 mmol l^–1cysteine appeared to decrease tension, yet increase Ṁ_O2. This suggests that H₂S oxidation continues, although the mechanism that causes contraction is subject to feedback inhibition. However, other explanations are also plausible, i.e. the experimental conditions were different (anoxia in myograph studies, P_O2∼40-50 mmHg in H₂S studies), this cysteine concentration is near the threshold for both processes, or there are temporal differences in responses. The effects of hydroxylamine on Ṁ_O2 do not correlate with its effects on tension and may also be nonspecific as it inhibits many other enzymes (Zollner,1989).

The authors wish to express their gratitude to R. Bishop, University of Canterbury, Christchurch (NZ) and B. Swink, US Geological Survey, Biological Resources Division, Millersberg, MI, USA for help with animal collection. This research was supported in part by National Science Foundation Grant No. IOS 0641436 (K.R.O.). K.R.O. was a recipient of an Erskine Fellowship from the University of Canterbury.