Hypoxic pulmonary vasodilation: a paradigm shift with a hydrogen sulfide mechanism

Abstract

Hypoxic pulmonary vasoconstriction (HVC), an intrinsic and assumed ubiquitous response of mammalian pulmonary blood vessels, matches regional ventilation to perfusion via an unknown O₂-sensing mechanism. Global pulmonary hypoxia experienced by individuals suffering from chronic obstructive pulmonary disease or numerous hypoventilation syndromes, including sleep apnea, often produces maladaptive pulmonary hypertension, but pulmonary hypertension is not observed in diving mammals, where profound hypoxia is routine. Here we examined the response of cow and sea lion pulmonary arteries (PA) to hypoxia and observed the expected HVC in the former and a unique hypoxic vasodilation in resistance vessels in the latter. We then used this disparate response to examine the O₂-sensing mechanism. In both animals, exogenous H₂S mimicked the vasoactive effects of hypoxia in isolated PA. H₂S-synthesizing enzymes, cystathionine β-synthase, cystathionine γ-lyase, and 3-mercaptopyruvate sulfur transferase, were identified in lung tissue from both animals by one-dimensional Western blot analysis and immunohistochemistry. The relationship between H₂S production/consumption and O₂ was examined in real time by use of amperometric H₂S and O₂ sensors. H₂S was produced by sea lion and cow lung homogenate in the absence of O₂, but it was rapidly consumed when O₂ was present. Furthermore, consumption of exogenous H₂S by cow lung homogenate, PA smooth muscle cells, and heart mitochondria was O₂ dependent and exhibited maximal sensitivity at physiologically relevant Po₂ levels. These studies show that HVC is not an intrinsic property of PA and provide further evidence for O₂-dependent H₂S metabolism in O₂ sensing.

it is well known that hypoxia dilates mammalian systemic vessels and constricts pulmonary vessels (29). Hypoxic vasoconstriction (HVC) is believed to be a unique attribute of the pulmonary resistance vessels of all mammals, and it is an intrinsic feature of pulmonary vascular smooth muscle cells (1, 10). Although HVC helps match regional ventilation to perfusion, it can have deleterious consequences during global pulmonary hypoxia associated with high altitude, chronic obstructive pulmonary disease, sleep apnea, and a variety of other hypoventilation syndromes (20, 22, 31).

Despite intense scrutiny and obvious clinical importance, the mechanism coupling the responses of pulmonary hypoxia to HVC, the so-called “O₂ sensor,” remains to be identified (1, 7, 27, 30, 33). Recent studies from our laboratory suggest that O₂-dependent metabolism of constitutively produced and vasoactive H₂S is the pivotal O₂ transducer (15). Support for this hypothesis has been obtained from studies on respiratory and systemic vessels from a variety of vertebrates and includes the following: 1) hypoxia and H₂S uniquely evoke the same response in isolated blood vessels from all vertebrate classes; 2) H₂S is enzymatically generated by vessels and adjacent tissue; 3) tissue H₂S concentration is inversely related to O₂ concentration; i.e., endogenous and exogenous H₂S is consumed in the presence of O₂, while endogenous H₂S rises in hypoxia; 4) hypoxia and H₂S appear to utilize a common activation process; and 5) hypoxia and H₂S depolarize bovine pulmonary vascular smooth muscle cells.

Diving mammals such as pinnipeds routinely experience profound hypoxia as a result of prolonged underwater activity, or terrestrial sleep apnea. Their arterial Po₂ often falls below 15–20 mmHg, which is well below the ∼25-mmHg level at which humans lose consciousness (11, 18, 21). Surprisingly, pulmonary arterial blood pressure does not increase in hypoxic pinnipeds, even during sleep apnea, when cardiac output remains constant (17). This finding is in marked contrast with the intense pulmonary hypertension that would be expected in terrestrial mammals under similar conditions (4, 13), and it suggests that the response of pinniped pulmonary arteries to hypoxia is different from that of their terrestrial counterparts.

In the present studies, we explored the possibility that pinniped pulmonary arteries were less sensitive to hypoxia by comparing the hypoxic response of isolated conductance pulmonary arteries (CPA) and resistance pulmonary arterioles (RPA) of the California sea lion (Zalophus californianus) with that of the cow (Bos tarus). The surprising observation that hypoxia dilated sea lion pulmonary arteries allowed us to further examine our hypothesis that O₂-dependent metabolism of H₂S is the vascular O₂ sensor.

MATERIALS AND METHODS

Animals

Holstein cow (Bos taurus, Mammalia) lungs were obtained from a nearby abattoir, placed in 4°C Krebs-Henseleit mammalian saline, and transported to the laboratory (South Bend or Milwaukee). The pulmonary arteries (4th–6th generation) were dissected out and stored in buffer at 4°C until use.

Lung tissue was obtained from euthanized (due to nonrepairable injuries or neurological problems associated with domoic acid toxicity) or fatally injured California sea lions (Zalophus californianus) that were delivered to the SeaWorld Adventure Park veterinary facility as part of the marine mammal rescue program of Southern California. The tissue was placed in a Kapak seal-a-meal bag with lactated Ringer solution and shipped overnight on ice to South Bend and Milwaukee. The pulmonary vessels were dissected free on the following morning and prepared for myography. Lungs with vessels that did not exhibit passive stress-relaxation or contract when exposed to 80 mM KCl were discarded. The suitability of this method of tissue procurement was verified in our laboratory by storage of cow lungs under similar conditions; no adverse effects were observed. All procedures were approved by the respective Institutional Animal Care and Use Committees.

Conductance Vessel Myography

Vessels were cut into 3- to 8-mm-long segments, mounted on 280-μm-diameter stainless steel wire hooks, and suspended in 5-ml water-jacketed smooth muscle baths filled with the appropriate buffer at 37°C. They were aerated with 95% air-5% CO₂. One hook was stationary; the other was connected to a force-displacement transducer (model FT03C, Grass Instruments, West Warwick, RI). Tension was measured on a polygraph (model 7E or 7F, Grass). Polygraph sensitivity was set to detect changes as small as 5 mg. Data were archived on a personal computer at 1 Hz using Labtech Notebook software (Laboratory Technologies, Andover, MA) or SoftWire (Measurement Computing, Middleboro, MA). The chart recorders and software were calibrated before each experiment.

Length-tension relationships were derived from KCl-contracted vessels and used to apply a reasonable baseline (resting) tension (∼1,000 mg) for 0.5–1 h before experimentation. In a typical experiment, vessels were contracted twice with 80 mM KCl, and resting tension was reestablished over 30- to 45-min equilibration periods after the rinse and again before experimentation. Vessels were then precontracted with the thromboxane A₂ mimetic U-46619 (10⁻⁶ M) and, when stable, exposed to hypoxia (95% N₂-5% CO₂, <5 mmHg Po₂).

To determine O₂ sensitivity of cow conductance vessels, we exposed cow CPA to progressively lower Po₂ by mixing 95% air-5% CO₂ and 95% N₂-5% CO₂ with a Wösthoff gas-mixing pump. In preliminary experiments, we observed that hypoxic contractions were stronger in precontracted vessels. To determine whether the U-46619 precontraction affected vessel sensitivity to hypoxia, the Po₂-response curves were obtained from otherwise unstimulated vessels and from vessels precontracted with 10⁻⁸ or 10⁻⁶ M U-46619.

To examine the contribution of potential sulfur donors to HVC, we exposed cow vessels to four consecutive periods of 20 of min hypoxia-30 min of normoxia. Potential sulfur donors, cysteine (Cys), reduced glutathione (GSH), oxidized glutathione (GSSG), 3-mercpatopyruvate (3-MP), or Cys + α-ketoglutarate (Cys + α-KG), all at 1 mM, were added after the second hypoxia exposure. Control vessels were exposed to four consecutive hypoxia treatments.

Resistance Vessel Myography

Isolated pulmonary arteries were cannulated with glass cannulas and placed in a water-jacketed plastic chamber. An artery segment was tied in place on the proximal cannula, and the lumen was flushed with Krebs-Henseleit-buffered (KHB) saline. The distal end of the artery was tied onto the distal cannula, and all side branches were tied off. The vessel was bathed in and superfused with KHB saline at 37°C and aerated with room air. A micrometer connected to the proximal cannula was used to remove slack from the artery. A KHB saline-filled syringe proximal to a pressure transducer on the inflow cannula was raised or lowered to set transmural pressure at the corresponding in vivo pressure (10 mmHg). A color video camera mounted on a stereomicroscope above the vessel chamber projected the artery image on a video monitor, and the external arterial diameter (±1.5 μm) was measured on a screen using a video scaler. The vessel diameter was always measured at the same point on the arterial wall using various distinguishing features, such as adhering connective tissue and side branches, located near the site. Diameters were measured immediately after the artery was mounted, after equilibration, and throughout the experimental protocols. Drugs were added to the superfusate, and hypoxia was achieved by gassing the superfusate with 95% N₂-5% CO₂, which consistently lowered the chamber Po₂ to <50 mmHg.

One-Dimensional Western Blotting

Bovine pulmonary artery smooth muscle cells (PASMC) were grown to confluence, placed on ice, and lysed using 1% Triton X-100 lysis buffer. Sea lion pulmonary arteries were cut into small pieces. Cells and arteries were treated with a protease inhibitor cocktail and 100× phenylmethylsulfonyl fluoride and centrifuged at 22,600 g at 4°C for 15 min. The supernatant was collected and loaded onto a gel at uniform protein concentration. A Bis-Tris 4–12% 1.5-mm precast gel was run with a NuPAGE MES SDS running buffer and SeeBlue prestained molecular weight standard at 200 V for ∼45 min, transferred to a 0.22-μm nitrocellulose membrane, and stained with Ponceau S for total protein load. The membrane was blocked for 30 min with BSA in TBS-Tween 20 buffer and probed overnight at 4°C with the appropriate primary antibody for the H₂S-synthesizing enzymes, cystathionine β-synthase (CBS; Santa Cruz Biotechnology, Santa Cruz, CA), cystathionine γ-lyase (CSE; Novus Biologicals, Littleton, CO), or 3-mercaptopyruvate sulfur transferase (3-MST; Sigma-Aldrich, St. Louis, MO; all at 1:2,000 dilution) and then for 1 h with the secondary antibody in 2% dry milk-TBS-Tween 20 solution. Enhanced chemiluminescence agent was applied, and the membrane was developed on CL-X Posure film. The image was viewed on a Kodak IS200 MMt Imaging Station.

Immunohistochemistry

Lung tissue was cut into small pieces and immersion fixed in fresh 4% paraformaldehyde overnight at 4°C. Tissues were then saturated with 30% sucrose in PBS and cryosectioned (Leica) at −20°C. Sections (10 μm thick) were dried on glass slides, permeabilized, and blocked in PBS containing 0.2% Tween 20 and 1% BSA for 1 h. Sections were incubated overnight at 4°C in rabbit polyclonal anti-CBS (Santa Cruz Biotechnology), anti-CSE (Sigma GenoSys), or anti-3-MST (Sigma); all primary antibodies were diluted 1:1,000 in blocking/permeabilization buffer. Endogenous peroxidase was then quenched by 1 h of incubation in 0.3% H₂O₂, and slides were washed five times for 10 min each in PBS with 0.2% Tween. Horseradish peroxidase-conjugated anti-rabbit secondary antibody (Vector Labs; 1:2,000 dilution) was incubated for 2 h at room temperature, and wash steps were repeated. Respective enzyme expression was then developed by 5 min of incubation in NOVARed horseradish peroxidase substrate (Vector Labs) to create a red/brown precipitate. Tissue sections were counterstained in hematoxylin for 1 min and ammonia for 30 s (to lighten and brighten the hematoxylin blue). Images were acquired with a Leica DMLFS microscope (objectives: ×10, NA 0.30 and ×40, NA 0.55), MicroPublisher 3.3 RTV camera, and QCapture software.

Tissue H₂S Production

Cow or sea lion lung tissue (1.0 g) with no large vessels or airways was processed with a homogenizer (Ultraturax model SDT, Ika-Werk) with an N-10 probe on ice in 4 ml of homogenization buffer and then centrifuged until the rotor speed reached 10,000 rpm. Homogenate (1.5 ml) was added to a water-jacketed glass metabolism chamber with ports for the H₂S and O₂ sensors and maintained at 37°C (see below). The homogenate was bubbled with N₂ until O₂ reached zero; then the stopper was lowered until no headspace remained and no bubbles were visible. The homogenate was spiked to 100 μM pyridoxal-5-phosphate and allowed to incubate for 2 min, and Cys was added to a concentration of 1 mM. After 10 min, α-KG was added to a concentration of 1 mM, and the homogenate was allowed to incubate for 20 min. Air (1.5 μl) was injected, and the homogenate was allowed to consume all the O₂ and resume production of H₂S. Tissue was homogenized immediately before each run, and a sample was assayed for protein using the Lowry method.

Effect of O₂ on H₂S Consumption

The following studies were performed only on cow tissues because of the limited availability of tissues from sea lions.

Homogenized cow lung.

Lung tissue (0.5 g) with no large vessels or airways was homogenized as described above on ice in 4 ml of homogenization buffer. In the metabolism chamber (37°C), 200 μl of supernatant were added to 1.3 ml of homogenization buffer and bubbled with air and N₂, which were mixed with a Wösthoff gas-mixing pump to the desired Po₂. The stopper of the chamber was lowered to remove all headspace, with care taken to ensure that no bubbles were trapped. The homogenate was spiked to 2 μM Na₂S and allowed to incubate for ∼4 min. Consumption rate was taken as a linear regression of the initial one-third of the consumption trace.

Cow heart mitochondria.

Twenty grams of bovine heart ventricle were placed in 200 ml of mitochondria isolation buffer (MIB) with 850 μl of nargarse protease solution (Sigma) and minced on ice for 5 min. The minced tissue was processed on ice intermittently with an Ika Ultraturax T18 Basic homogenizer with an S18 N-19G probe until smooth. The homogenate was centrifuged at 1,300 g for 5 min, and the supernatant was vacuum filtered through four layers of cheesecloth and then centrifuged at 14,500 g for 10 min. The pellet was rinsed with MIB to remove the fluffy layer surrounding the mitochondrial pellet, resuspended in MIB, and centrifuged again at 14,500 g for 10 min. The pellet was rinsed and resuspended in a minimum volume of MIB. Mitochondrial respiratory control ratio buffer (1.5 ml) was placed in the diffusion chamber (37°C) and bubbled with air and N₂, which were mixed using a Wösthoff gas-mixing pump until Po₂ stabilized. The stopper was then lowered, with care taken to ensure that no gas bubbles were trapped. The buffer was then spiked to 2 μM Na₂S, and 5 μl of ∼20 mg/ml mitochondrial suspension were added. The mitochondria were allowed to consume the sulfide for ∼4 min. Consumption rate was taken as a linear regression of the initial one-third of the consumption trace.

Bovine PASMC.

Bovine PASMC were cultured in Cascade Biologics Medium 231 with smooth muscle cell growth supplement at 37°C in an atmosphere of 5% CO₂ in T-75 flasks. Cells were used between passages 6 and 8 and harvested with 0.25% trypsin-EDTA at near confluence. Cells were centrifuged at 1,000 rpm for 5 min and resuspended in ∼3.5 ml of Medium 231 with supplement, counted, and kept on ice until use. Cell suspension (1.5 ml) was transferred to the metabolism chamber and equilibrated to 127 μM O₂ (equivalent to systemic arterial Po₂) or 40 μM O₂ (equivalent to systemic venous Po₂). Once the suspension was equilibrated, the stopper was lowered to remove all air, and the cells were spiked to 2 μM Na₂S. H₂S consumption was then measured in the same batch of cells at 7.5, 2, and 0 μM O₂ in random order and determined at 127 or 40 μM O₂ to ensure continued cell viability. There were no statistically significant differences between the initial and final H₂S consumption rates at 127 or 40 μM O₂.

Metabolism chamber.

The metabolism chamber was constructed in-house and consisted of a lost-wax-cast soda lime glass inner member with O₂ and H₂S sensor ports on the side, orthogonal to each other and to the axis of the chamber and surrounded by an acrylic water jacket. A polyvinylidine difluoride (PVDF) stopper was machined to tightly fit into the opening and sealed with a Viton O-ring. A small hole drilled through the stopper permitted venting of the headspace air when the stopper was lowered into the chamber and provided an access port for injection of drugs or air bubbles with a Hamilton syringe. The chamber was placed on a magnetic stirrer and a PVDF-coated stir bar, constructed from the core of a typical laboratory micro stir bar that had been removed from the Teflon shell and inserted into a new shell of PVDF and heat-sealed, was used to minimize backdiffusion of O₂, which can be a problem with Teflon-coated stir bars. Construction of the H₂S sensor with a sensitivity of 14 nM H₂S gas (∼100 nM total sulfide) is described elsewhere (32). O₂ was measured with a dissolved oxygen electrode (model 125/05D, Instech, Plymouth Meeting, PA) and was calibrated according to the manufacturer's directions and verified with N₂ equilibrated buffer and air-N₂ gas mixtures of 20, 10, 5, 1, and 0% air-balance N₂ obtained from a Wösthoff gas-mixing pump. On the basis of the signal-to-noise ratio, the O₂ sensitivity was estimated to be <0.5 μM (∼0.7 mmHg Po₂). Both sensors were connected to a free radical analyzer (Apollo 4000, WPI, Sarasota, FL). Response time for the H₂S and O₂ sensors was 63% in 8 and 4 s, respectively.

Buffers and Chemicals

Mammalian KHB saline consisted of (in mM) 115 NaCl, 2.51 KCl, 2.46 MgSO₄, 1.91 CaCl₂, 5.56 glucose, 1.38 NaH₂PO₄, and 25 NaHCO₃ (pH 7.4). Lung tissue homogenization buffer consisted of (in mM) 120 KCl, 2 NaH₂PO₄, 3 Na₂HPO₄, and 1 MgSO₄ (pH 7.0). Respiratory control ratio buffer consisted of (in mM) 100 KCl, 50 MOPS, 5 K₂HPO₄, and 1 EGTA (pH 7.4). MIB consisted of (in mM) 220 mannitol, 70 sucrose, 5 MOPS, and 2 EGTA (pH 7.0). H₂S was produced by dissolving Na₂S into deoxygenated buffer under 100% N₂ and titration to pH 7.4 with HCl. Unless otherwise stated, all chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

Calculations

Statistical comparisons were performed with the Student's t-test or, when appropriate, with a paired t-test. Significance was assumed when P ≤ 0.05. The effective Po₂ at which H₂S consumption was half-maximal (P₅₀, or EC₅₀ when expressed as O₂ concentration) was estimated from a curve fit by eye.

RESULTS

Hypoxia (<5 mmHg Po₂) consistently contracted cow CPA (>500 μm diameter; Fig. 1, A and C). The response of sea lion CPA to hypoxia was variable: some vessels were relaxed (Fig. 1, B and C), whereas others contracted (not shown). Because CPA do not make a major contribution to pulmonary vascular resistance and may also exhibit hypoxic vasodilation (HVD) in terrestrial mammals (9), we examined the hypoxic responses of RPA (<400 μm diameter), where HVC is presumed pervasive. Here the results were more striking: moderate hypoxia (<50 mmHg Po₂) constricted all cow RPA but dilated all sea lion RPA (Fig. 1D). The dichotomous hypoxic responses of RPA were not due to differences in the mechanical properties of the two vessels, inasmuch as vessels from both species were contracted by 80 mM KCl and the thromboxane A₂ mimetic U-46619 (10⁻⁶ M, not shown).

Fig. 1.

Effects of hypoxia (N₂) and H₂S (3 × 10⁻⁴ M) on U-46619 (10⁻⁶ M)-precontracted conductance (CPA; >500 μm diameter) and resistance (RPA; <400 μm diameter) pulmonary arteries from cow and sea lion. A and B: representative traces showing hypoxic (100% N₂) constriction of cow CPA and hypoxic and H₂S (1 and 3 × 10⁻⁴ M) dilation of sea lion CPA. C: responses of cow and sea lion CPA to hypoxia and H₂S [1 × 10⁻⁴ M (cow) and 3 × 10⁻⁴ M (sea lion)]. Hypoxia and H₂S consistently constricted cow CPA. Response of sea lion CPA was variable, but vessels relaxed by hypoxia, as shown here, were similarly relaxed by H₂S. D: cow RPA were consistently contracted by hypoxia (<50 mmHg Po₂) and H₂S (3 × 10⁻⁴ M); the same treatments uniformly relaxed sea lion RPA. Values are means ± SE; n = 5 cows and 5 CPA, 4 sea lions and 12 CPA in hypoxia; 2 sea lions and 14 CPA in H₂S; and 9 cows and 9 RPA, 3 sea lions and 5 RPA in hypoxia and H₂S.

The mechanical effects of H₂S were identical to those produced by hypoxia in cow and sea lion pulmonary arteries. H₂S (≥100 μM) contracted cow CPA (Fig. 1, A and C). In sea lion CPA, those vessels that were relaxed by hypoxia were also relaxed by H₂S (Fig. 1, B and C) and those vessels that were contracted by hypoxia were also contracted by H₂S (not shown). The response of RPA to H₂S (300 μM) was also identical to the hypoxic response: H₂S constricted all cow RPA and dilated all sea lion RPA (Fig. 1D).

Figure 2 shows the distribution of H₂S-synthesizing enzymes in sea lion RPA, cultured bovine PASMC, and cultured bovine pulmonary artery endothelial cells. CBS immunoreactivity was not evident in sea lion RPA or in bovine PASMC but was clearly present in bovine pulmonary artery endothelial cells. Immunohistochemical localization of CBS, CSE, and 3-MST antibodies in cow and sea lion lung tissue is shown in Fig. 3. In general, all three enzymes were abundantly distributed in the small airways and alveoli of both animals. There was noticeably less immunoreactivity in the adventitia and media of small vessels, whereas all three enzymes were often evident in the endothelium.

Fig. 2.

Western blots of H₂S-synthesizing enzymes, cystathionine β-synthase (CBS), cystathionine γ-lyase (CSE), and 3-mercaptopyruvate sulfur transferase (3-MST), in sea lion RPA and bovine pulmonary arterial smooth muscle cells (PASMC) and endothelial (ENDO) cells. CSE and 3-MST were abundant in RPA and PASMC, but only 3-MST was present in ENDO cells. CBS immunoreactivity was found only in ENDO cells.

Fig. 3.

Immunohistochemical localization of CBS, CSE, and 3-MST in cow (A) and sea lion (B) lung. Low-power (×10 objective) images show tissue distribution of respective enzymes (red/brown) with nuclear counterstain (blue); higher-power (×40 objective) images show an arteriole and surrounding tissue. Image scale: ×10 = 880 μm wide, ×40 = 220 μm wide.

An amperometric (polarographic) H₂S sensor was used to examine H₂S production and O₂-dependent metabolism in real time in homogenized, but otherwise untreated, lung tissue from cow and sea lion (Fig. 4). H₂S production was observed in lung tissue from both animals in the presence of pyridoxal phosphate (a cofactor for CSE and CBS) and Cys and after addition of α-KG, but only under severely hypoxic conditions. Addition of air, theoretically sufficient to increase O₂ concentration to ∼8.5 μM (in the absence of any concomitant O₂ consumption), immediately led to consumption of previously formed H₂S and prevented H₂S from increasing until nearly all the available O₂ was consumed. As the O₂ concentration fell, H₂S production resumed. These studies show that lung tissue from both animals produces H₂S and that O₂-dependent H₂S consumption inversely couples tissue H₂S concentration to ambient O₂ concentration.

Fig. 4.

Real-time measurement showing substrate- and O₂-dependent H₂S production by homogenized cow (A and B) and sea lion (C and D) lung. A and C: representative simultaneous measurements of H₂S (sulfide) and O₂ concentrations after addition of pyridoxyl 5′-phosphate (100 μM; time 0) followed by cysteine (Cys; 1 mM), α-ketoglutarate (α-KG; 1 mM), and air (1.5 μl, ∼8.5 μM O₂) and recovery of H₂S production after injected O₂ was consumed (rec). B and D: rate of H₂S production or consumption after each treatment for cow and sea lion, respectively. Addition of O₂ consumes extant H₂S and prevents further accumulation until O₂ concentration falls. Values are means ± SE.

To further examine the relationship between H₂S consumption and O₂ availability, we measured the rate of exogenous H₂S consumption as a function of Po₂ in bovine lung homogenate, cultured bovine PASMC, and mitochondria extracted from bovine hearts (Fig. 5). H₂S consumption in lung homogenate was maximal until Po₂ fell below 20 mmHg and decreased thereafter. The estimated Po₂ for half-maximal H₂S consumption (P₅₀) by lung homogenate was ∼3.2 mmHg (EC₅₀, 4 μM O₂). H₂S consumption by bovine PASMC was maximal in the range of arterial and venous plasma Po₂, 100 and 32 mmHg, respectively (127 and 40 μM O₂) but fell to half-maximal at 6 mmHg (7.5 μM O₂). The rate of H₂S consumption by bovine PASMC was <10% of maximal at 1.5 mmHg (2 μM O₂). H₂S consumption by bovine heart mitochondria was sustained until Po₂ fell below ∼5 mmHg, with an apparent P₅₀ of ∼0.8 mmHg (EC₅₀ ≤1 μM O₂). These studies show a dose-dependent relationship between O₂ concentration and H₂S consumption in a variety of tissue preparations.

Fig. 5.

Effect of O₂ on consumption of exogenous H₂S by bovine lung homogenate (n = 3), cultured bovine PASMC (n = 6–10 replicates), and bovine heart mitochondria (n = 6 replicates). Samples were spiked with 2 μM Na₂S, and H₂S consumption was continuously measured with an amperometric sensor. Rate of H₂S consumption by all samples decreased when O₂ concentration fell below 10 μM. Approximate O₂ concentration (or Po₂) at half-maximal H₂S consumption (EC₅₀ or P₅₀, respectively) was 4 μM O₂ (∼3.2 mmHg Po₂) for lung homogenate and ≤1 μM O₂ (no more than ∼0.8 mmHg Po₂) for bovine heart mitochondria. H₂S consumption by PASMC was maximal at O₂ concentrations equivalent to pulmonary arterial and venous blood (40 and 127 μM O₂, respectively), but at 7.5 μM O₂ it was reduced to ∼35% of maximum. Values are means ± SE. Curves were fit by eye.

The effects of prestimulation on force development and O₂ sensitivity of HVC were examined in cow CPA. Prestimulation with 10⁻⁶ M U-46619 significantly increased the strength of maximal HVC nearly three- to fivefold (Fig. 6A) but did not significantly increase the O₂ sensitivity (Fig. 6B). The P₅₀ values for 0, 10⁻⁸, and 10⁻⁶ M U-46619 were ∼6, 7, and 10 mmHg, respectively. These studies show that although prestimulation affects the mechanical process of HVC, it has little, if any, effect on the O₂-sensing mechanism.

Fig. 6.

Effect of prestimulation with the thromboxane A₂ mimetic U-46619 on maximal (100% N₂) hypoxic contraction (A) or O₂ sensitivity (B) of cow CPA. Prestimulation significantly increased the magnitude of 2nd, 3rd, and 4th hypoxic contractions 3- to 4-fold (A) but did not affect O₂ sensitivity (B). Values are means ± SE; n = 7 in each group in A and n = 5 for 10⁻⁶ M U-46619, 4 for 10⁻⁸ M U-46619, and 3 for no prestimulation (0 U-46619) in B. Curves were fit by eye.

Potential sulfur donors, Cys, GSH, GSSG, 3-MP, and Cys + α-KG, were added to prestimulated and unstimulated cow CPA to examine their ability to augment HVC. Cys and 3-MP slightly, but significantly, decreased the third HVC in prestimulated CPA (Fig. 7A), whereas HVC was significantly, and substantially, increased by all sulfur donors, except 3-MP and, at the fourth HVC, by GSSG in unstimulated vessels (Fig. 7B).

Fig. 7.

Effects of potential sulfur donors, Cys, reduced glutathione (GSH), oxidized glutathionine (GSSG), 3-mercpatopyruvate (3-MP), or Cys + α-ketoglutarate (α-KG), all at 1 mM, on the magnitude of hypoxic contraction of cow CPA that were prestimulated with 10⁻⁶ M U-46619 (A) or unstimulated (B). Control vessels were exposed to 4 consecutive periods of hypoxia; sulfur donors were added after the 2nd hypoxic exposure. Values are means ± SE of number of vessels shown in parentheses below bars; values for the 3rd (gray bars) and 4th (black bars) hypoxic contractions were normalized relative to values for the second (white bars) hypoxic contraction. *Significantly different from respective control (Con).

DISCUSSION

To our knowledge, our study is the first observation of HVD in a mammalian RPA. This is noteworthy in several aspects. 1) It challenges the paradigm that HVC is a universal attribute of the mammalian pulmonary vasculature (28). 2) HVD appears to be of physiological benefit during routine dives and sleep apnea. 3) It demonstrates the evolutionary plasticity of fundamental vascular responses that enables them to fit the needs of the organism. 4) The disparate responses to hypoxia in cow and sea lion pulmonary arteries provide a unique model system with which to dissect vascular O₂-sensing and signal transduction mechanism(s). This allowed us to provide additional evidence that O₂-dependent H₂S metabolism is involved in vascular O₂ sensing.

Hypoxic Pulmonary Vasodilation

Given the pervasive thought that hypoxia constricts the mammalian pulmonary microcirculation, we initially hypothesized that HVC would have to be blunted in diving mammals such as pinnipeds to minimize an otherwise anticipated pulmonary hypertension. The HVD that we observed, although unanticipated, is clearly of greater homeostatic value, inasmuch as it will actually reduce pulmonary vascular resistance and potentially lower the metabolic load on the right ventricle. HVD may be even more beneficial during sleep apnea. It is well known that cardiac output decreases during a dive. The reduction in pulmonary blood flow could potentially offset HVC and minimize the potential for pulmonary hypertension. However, unlike a dive, cardiac output does not change during the entire apneic period when the animal is sleeping on land (17), even though the arterial Po₂ may be as low as that encountered in a dive (11, 21). Under these circumstances, even a modest HVC would produce pulmonary hypertension. HVD ensures that this does not occur.

H₂S and O₂ Sensing

We previously proposed that O₂-dependent metabolism of H₂S serves as a vascular and chemoreceptor O₂ sensor (14–16). In this model, the constitutive cytosolic production of vasoactive H₂S is offset by mitochondrial H₂S oxidation. The oxidative capacity of tissues is such that tissue H₂S concentration is nil during normoxia, but as O₂ levels fall, the capacity to oxidize H₂S progressively fails and tissue H₂S increases. The existence, as well as sensitivity and efficacy, of this mechanism depends on a number of criteria. 1) The effects of H₂S must mimic the hypoxic response, implying that these two stimuli utilize the same initial activation pathway. 2) Tissues must be capable of H₂S synthesis. 3) The rate of H₂S production or H₂S metabolism must be intimately coupled to O₂ availability, i.e., Po₂. 4) O₂-H₂S coupling must occur at physiologically relevant levels of O₂. We previously demonstrated various aspects of these criteria in a variety of vertebrates from hagfish and lamprey to rats and cows. The uniqueness of the present study is that it examines these criteria in the same organ of two mammals where the vascular responses to hypoxia are exactly the opposite.

The similarity of hypoxic and H₂S responses in cow and sea lion pulmonary vessels, despite the opposite outcome (Fig. 1), is consistent with our previous observations that hypoxia and H₂S produce identical responses in systemic and respiratory conductance vessels from vertebrates of every class, from hagfish to mammals, whether this response is contraction, relaxation, or multiphasic (15). Furthermore, the difference between cow and sea lion vessels is specific for hypoxia and H₂S, inasmuch as other stimuli, such as 80 mM KCl and U-46619, consistently constricted both vessels. Collectively, these observations support criterion 1 and suggest that hypoxia and H₂S regulate vascular diameter via a common pathway that is upstream from the ultimate mechanical outcome, i.e., HVC or HVD.

Figures 2 and 3 show that pulmonary vessels and adjacent lung tissue have appropriate enzymes for H₂S synthesis, and Fig. 4 demonstrates that H₂S can be produced by these tissues under the appropriate conditions. H₂S can be produced directly from Cys desulfuration via the enzymes CSE (also known as CGL; EC 4.4.1.1) and CBS (EC 4.2.1.22) or from Cys transamination to 3-mercaptopyruvate (with α-KG, often serving as the amine acceptor) and subsequent desulfuration by the enzyme 3-MST (EC 2.8.1.2) (8, 19). CSE is the only desulfurating enzyme that has been identified in mammalian vessels (5, 25, 35, 36), although we have identified CBS mRNA in trout arteries and veins (16). CSE is found in smooth muscle and endothelial cells, and it is especially abundant in pulmonary arteries (5, 25, 34–36). 3-MST appears to be the major pathway for H₂S synthesis in the brain (19), but to our knowledge, our study is the first evidence of 3-MST in the vasculature.

The capacity for H₂S synthesis in the airways is particularly noteworthy, inasmuch as it suggests that these tissues are also involved in O₂ sensing. This is consistent with the well-known observation in terrestrial mammals that airway hypoxia is a more potent stimulus of HVC than mixed-venous (systemic) hypoxemia (13). Teleologically, it makes sense that the tissue that is the most immediately exposed to hypoxia (airways) or is the most intimately sensitive to it (systemic tissues) would be a leading contributor to the vasoactive response, be it constriction or dilation. This also explains why H₂S is produced by so many extravascular tissues.

Addition of potential sulfur donors to otherwise unstimulated cow CPA greatly augmented HVC (Fig. 7A), suggesting that vascular smooth muscle can utilize sulfur via variety of biochemical pathways to increase H₂S production. This effect was not noticed when the vessels were precontracted with U-46619. Although U-46619 may augment H₂S production, it seems more probable that U-46619 prestimulation augments HVC independent of H₂S production, and this masks the more subtle effects of the sulfur donors on HVC in otherwise unstimulated vessels. Precontraction is well known to enhance HVC, and in many isolated preparations it is an absolute requirement for HVC (28). It has been suggested that prestimulation enhances mobilization and turnover of Ca²⁺ from intracellular stores, which then increases the availability of Ca²⁺ for contraction (2, 28). The fact that prestimulation increased the magnitude of HVC three- to fourfold in our experiments without affecting O₂ sensitivity supports the suggestion that the effects of U-46619 are also indirect. Collectively, these studies plus the observations, in real time, of tissue H₂S production (Fig. 4) support criterion 2 for H₂S-mediated vasoactivity.

Figure 4 also shows that, in the presence of O₂, any endogenously produced H₂S is rapidly consumed and that H₂S concentrations will remain low until O₂ levels fall. This satisfies criterion 3, an inverse relationship between O₂ and H₂S.

Because it is not possible to measure intracellular H₂S concentrations and there is still some uncertainty about subcellular O₂ distribution, we can only indirectly address criterion 4, i.e., tissue H₂S is inversely coupled to O₂ at physiologically relevant Po₂ levels (or O₂ concentrations). Figures 5 and 6 are, to our knowledge, the first attempt to do so. Unfortunately, there was not enough tissue from sea lions for a comparative analysis; nevertheless, we are confident that the cow samples are representative of a variety of O₂-sensing tissues. The P₅₀ for HVC in cow CPA was 6–11 mmHg for otherwise unstimulated vessels and vessels precontracted with U-46619 (Fig. 5B), even though the force of HVC was almost four times greater in the maximally precontracted vessels (Fig. 5A). Thus it is reasonable to assume that the P₅₀ for the O₂-sensing mechanism is relatively independent of the contractile process and must occur upstream from it. We can then compare this P₅₀ with the Po₂ at which consumption is impaired: ∼3 mmHg for homogenized lung tissue and ∼6 mmHg for cultured PASMC (Fig. 5). The P₅₀ values for H₂S consumption and vascular activation are reasonably similar and may be even closer if one corrects for the increased O₂ diffusion distance that would be expected in an intact CPA. This argument is fortified by the even lower P₅₀ of purified mitochondria (Fig. 5), where diffusion distances are even shorter. Although it is possible that the different O₂ sensitivities of lung homogenate, cultured smooth muscle cells, and mitochondria are due to tissue-specific variations, it is clear that, in all preparations, at low O₂ concentrations the ability to metabolize H₂S progressively fails.

The efficacy of O₂-dependent H₂S metabolism can be put into the context of the effects of brief physiological hypoxia on vascular/tissue Po₂. In rats breathing room air (6), arterial, arteriolar, tissue (near venous capillaries), and venular Po₂ levels are 98, 52, 35, and 27 mmHg, respectively. These levels are typical of Po₂ in a variety of tissues (23), and all are sufficiently high to rapidly and completely consume H₂S in our various preparations (Fig. 5). Moderate hypoxia produced by ventilation with 7% O₂ for 1 min lowered these values to 32, 16, 11.4, and 9.6 mmHg, respectively. Under these conditions, venular and tissue Po₂ would be sufficiently low to begin to impair H₂S consumption by PASMC and homogenized lung. There is considerably more uncertainty in the literature surrounding mitochondrial O₂ concentrations, but because this is the metabolic O₂ sink, it is most likely well below that of the interstitium and cytosol. Recently, using a noninvasive optical technique to estimate mitochondrial Po₂ in the rat heart in vivo, Mik et al. (12) found that, in normoxic rats, Po₂ was 10–20 mmHg in 26% of the mitochondria and 0–10 mmHg in only 10% of the mitochondria. Ventilating the rats with 10% O₂ increased the fraction of mitochondria in the 0- to 10-mmHg Po₂ range to 46%. These results show that even moderate hypoxia produces a cascade of progressive hypoxia in all compartments, ultimately resulting in tissue and/or mitochondrial Po₂ levels that are not sufficient to sustain H₂S consumption. Certainly, the ∼15-mmHg Po₂ observed in arterial and venous blood of elephant seals during prolonged sleep apnea (21) would push intracellular and mitochondrial Po₂ of pulmonary and systemic tissues well into the range of impaired H₂S consumption.

The efficacy of hypoxia-driven H₂S liberation at physiologically relevant Po₂ across species, despite opposing mechanical responses of vessels, demonstrates a certain ubiquity of this mechanism in vascular O₂ sensing. Ward (26) evaluated the most popular putative O₂ sensors in the context of vascular and tissue Po₂ (Fig. 8). Of these, NADPH oxidase appeared to have the greatest dynamic range, progressing almost linearly from 15% activity at 2 mmHg to >80% activity at 120 mmHg. Other sensors, such as heme oxygenase-2 and mitochondrial cytochromes a and a₃, were ∼100% active over the entire range of Po₂ (0–120 mmHg), suggesting that they may not contribute to O₂ sensitivity per se, whereas the factor inhibiting hypoxia-inducible factor-1α, cytochrome P-450 monooxygenase-ω, and prolyl hydroxylase domain proteins were <20% active at ≤20 mmHg Po₂. By comparison, H₂S consumption increases from essentially 0 to 100% between 0 and 20 mmHg, spanning the range of intracellular and mitochondrial Po₂ levels that are expected to occur in normoxia and hypoxia, respectively. It is evident from Fig. 8 that the O₂ sensitivity of H₂S metabolism is most efficient in the expected range of intracellular and mitochondrial P₅₀. Furthermore, it is over this range of P₅₀ levels that HVC in CPA is activated. In fact, the Po₂ for HVC in otherwise unstimulated CPA intersects with the Po₂ for H₂S consumption by PASMC and homogenized lung tissue near their respective P₅₀ values (Fig. 8).

Fig. 8.

O₂ sensitivity of H₂S consumption by tissues and mitochondria (heavy lines, solid symbols; adapted from Fig. 5) to other putative O₂-sensing mechanisms (thin dotted lines, open symbols; adapted from Ref. 26) in the context of physiological mitochondrial, cytosolic, tissue, and alveolar/arterial Po₂ (double-headed arrows at top; adapted from Refs. 12 and 26). O₂ sensitivity of hypoxic vasoconstriction of cow CPA (heavy dotted lines, adapted from Fig. 6) is shown for comparison. H₂S fills a gap in sensitivity in the Po₂ range most likely encountered by mitochondria and cytosol during hypoxia and intersects with Po₂-response curves of pulmonary arteries near their P₅₀. Abbreviations and symbols from the present study: lung, homogenized lung tissue (·); mito, isolated mitochondria (♦); PSAMC (■); unstimulated pulmonary arteries (); pulmonary arteries precontracted with 10⁻⁶ M U-46619 (). Abbreviations from Ref. 26 (all open symbols): HO-2, heme oxygenase-2 (octagons); Cyt aa3, mitochondrial cytochromes a and a₃ (triangles); NOX2, NADPH oxidase (diamonds); FIH-1, factor inhibiting hypoxia-inducible factor-1α (HIF-1α); CYPω, cytochrome P-450 monooxygenase-ω (squares); PHD, prolyl hydroxylase domain proteins involved in O₂-dependent hydroxylation of protein residues on HIF-1α (circles).

Along somewhat similar lines, Vanderkooi et al. (24) compared tissue Po₂ levels with the K_m for 60 O₂-consuming enzymes and, with the exception of cytochrome c oxidase, all were outside the physiological range of O₂ concentration, especially considering that intracellular and mitochondrial O₂ is lower than whole tissue O₂. The assumption that an “apparent” K_m for H₂S consumption is equivalent to the EC₅₀ (∼4 μM O₂; Fig. 5) provides additional support for H₂S metabolism in O₂ sensing. Recent studies showing that free H₂S in tissues is in the low nanomolar range or below (3) are also suggestive of the efficacy of this metabolic pathway in normoxic tissues.

Perspectives

H₂S and O₂ have a long geochemical history of mutual exclusivity: the presence of one in the environment implies the absence of the other. Eukaryotic cells also have a long history with H₂S, and, in fact, a number of the great extinctions appear to be associated with ambient hypoxia and elevated H₂S. It seems reasonable to assume that the animals that survived these dramatic conditions developed methods to cope with abnormal levels of both gases. The present study supports the hypothesis that this reciprocity has been employed as a convenient O₂-sensing/signaling mechanism in anatomically homologous blood vessels with dichotomous hypoxic responses. These studies are also the first to show hypoxic vasodilation in mammalian RPA and argue that HVC is not an invariant attribute of the mammalian pulmonary circulation. Because pinnipeds evolved relatively recently (35–38 million years ago), our findings also suggest that the vertebrate vasculature has considerable plasticity in its ability to utilize a fundamental O₂-sensing mechanism and convert this downstream into completely opposite effector responses suited to the homeostatic requirements of the animal.

GRANTS

This work was supported in part by National Science Foundation Grant IOS 0641436 (K. R. Olson) and National Center for Research Resources IDeA Networks of Biomedical Research Excellence Grant P20 RR-016454 (S. E. Bearden).

DISCLOSURES

No conflicts of interest are declared by the author(s).