Hydrogen sulfide (H2S) is rapidly emerging as a biologically significant signaling molecule. Studies published before 2000 report low or undetectable H2S (usually as total sulfide) levels in blood or plasma, whereas recent work has reported sulfide concentrations between 10 and 300 μM, suggesting it acts as a circulating signal. In the first series of experiments, we used a recently developed polarographic sensor to measure the baseline level of endogenous H2S gas and turnover of exogenous H2S gas in real time in blood from numerous animals, including lamprey, trout, mouse, rat, pig, and cow. We found that, contrary to recent reports, H2S gas was essentially undetectable (<100 nM total sulfide) in all animals. Furthermore, exogenous sulfide was rapidly removed from blood, plasma, or 5% bovine serum albumin in vitro and from intact trout in vivo. To determine if blood H2S could transiently increase, we measured oxygen-dependent H2S production by trout hearts in vitro and in vivo. H2S has been shown to mediate ischemic preconditioning (IPC) in mammals. IPC is present in trout and, unlike mammals, the trout myocardium obtains its oxygen from relatively hypoxic systemic venous blood. In vitro, myocardial H2S production was inversely related to Po2, whereas we failed to detect H2S in ventral aortic blood from either normoxic or hypoxic fish in vivo. These results provide an autocrine or paracrine mechanism for myocardial coupling of hypoxia to H2S in IPC, i.e., oxygen sensing, but they fail to provide any evidence that H2S signaling is mediated by the circulation.
- hydrogen sulfide metabolism
- vascular signaling
hydrogen sulfide (H2S) has recently joined carbon monoxide and nitric oxide as the third “gasotransmitter” (37). Evidence has been accumulating for H2S involvement in neurological, cardiovascular, gastrointestinal, genitourinary, and endocrine systems where it may have modulatory and/or cytoprotective effects (2, 17, 20, 22, 35). H2S has also been proposed to be both pro- and anti-inflammatory (19, 36).
The effects of sulfide1 typically occur at concentrations between 10 and 300 μM, which are within the range of plasma sulfide levels reported in at least 20 studies since 2000 (summarized in supporting information). This implies that the above actions are continuously modulated by circulating sulfide. However, studies published before 2000 invariably report very low or undetectable levels of plasma sulfide. To our knowledge, no study in the past seven years has reported sulfide levels of 5 μM or less, with the exception of one anecdotal report by Koenitzer et al. (16). This discrepancy between older and more recent studies does not appear to be due to an improvement in sulfide measurement techniques, since the earlier studies all have detection limits of 1 μM or less, and the newer studies, aside from Koenitzer et al. (16), use techniques developed decades ago without any obvious improvements. In fact, recent reviews (20, 35) question whether circulating sulfide can be as high as recent reports claim.
In preliminary studies, we used a polarographic sensor to measure the concentration of sulfide in trout blood. This sensor specifically measures dissolved H2S gas at submicromolar concentrations and functions in real time at physiological pH, thus allowing continuous monitoring of sulfide in unaltered blood samples and avoiding harsh chemical conditions of other commonly used methods. Because we were unable to detect sulfide in trout blood, we extended this study to include a variety of commonly examined vertebrate models. Finding little evidence for free sulfide in any vertebrate blood, we then measured turnover of exogenous sulfide in blood to determine if blood itself could contribute to these low levels, and we found that blood rapidly consumes sulfide. These results suggested that, contrary to the majority of recent reports, free sulfide does not exist in vertebrate blood at levels >100 nM.
It is possible, however, that plasma sulfide concentration ([sulfide]) could be transiently elevated under certain circumstances, even if sulfide is not a tonic signal . One of the likely locations for an elevated plasma sulfide is in blood bathing the trabeculated trout myocardium. First, there is increasing evidence that sulfide production in tissues is inversely related to Po2 (6, 25). Because trout myocardial cells derive most of their oxygen from systemic venous blood (8) these tissues would have a high probability of contributing sulfide to the circulation. Second, ischemic preconditioning has been demonstrated in trout as well as mammalian hearts (9). Because sulfide has been shown to be important in ischemic preconditioning in the rat heart (27, 31) and shown to exert a protective role in myocardial ischemia (39), it seemed likely that not only was a hypoxia-driven increase in myocardial sulfide production responsible for the preconditioning response but that this response would be even greater in the relatively more hypoxic trout myocardium. Accordingly, a second series of experiments was then conducted to determine if trout hearts produced sulfide, if myocardial sulfide production was inversely related to Po2, and if ambient hypoxia was sufficient to increase circulating [sulfide] in trout in vivo.
MATERIALS AND METHODS
Animals and blood sampling.
All animal protocols were approved by Institutional Animal Care and Use Committee review. All blood was collected in heparinized (∼50 USP/ml) syringes or plastic containers.
Sea lamprey (Petromyzon marinus, 0.13–0.45 kg) were trapped in streams feeding into the Great Lakes and maintained at Indiana University School of Medicine-South Bend in 500-liter tanks with aerated, through-flowing well water (14°C) and a 12:12-h light-dark cycle. They were anesthetized in benzocaine (1:5,000 wt/vol) and opened with a midventral incision, and blood was drawn from the posterior cardinal veins.
Rainbow trout (Oncorhynchus mykiss, 0.4–0.7 kg) were obtained from a commercial hatchery and maintained in circulating 2,000-liter tanks with aerated, through-flowing well water (14°C) and a 12:12-h light-dark cycle and fed a maintenance diet with commercial trout pellets until 72 h before experimentation. Blood was drawn from the hemal arch of lightly restrained fish.
Lobund-Wistar (LW) rats (0.35–0.5 kg), Harlan-Sprague-Dawley (HSD) rats (0.4–0.5 kg), and C57 Black/6 mice (∼25 g) were housed on-site and kept on a 12:12-h light-dark cycle with access to food and water ad libitum. LW rats and mice were killed with CO2, and blood was drawn via cardiac puncture while the heart was still beating. HSD rats were heparinized by intraperitoneal injection of 0.1 ml of 10 mg/ml heparin and then anesthetized with 1 ml of 50 mg/ml pentobarbital. Blood was drawn from the hepatic vein (flow cell experiments) or thoracic cavity (metabolism experiments).
Pig (Sus scrofa) and cow (Bos taurus) mixed venous and arterial blood was collected in heparinized containers within 5–10 min after the animals were killed at a local slaughterhouse. Flow cell and diffusion chamber experiments (below) were conducted on-site.
Blood was used within 1 min after collection for flow cell experiments and within 8 min for diffusion chamber experiments; blood for sulfide consumption studies was kept on ice and used the same day. Heparin did not interfere with H2S measurements or sulfide consumption.
Isolation of red blood cell ghosts.
Red blood cell ghosts were isolated from cow blood as described previously (23). Following isolation, an aliquot of ghosts was washed in Krebs' buffer and another in potassium phosphate buffer, both pH 7.4, and resuspended 1:1 in the same buffer as the wash.
Liberation of sulfide from blood by dithiothreitol.
Warenycia et al. (38) reported that 10 mM dithiothreitol (DTT) liberated a nonacid labile sulfide from brain tissue of H2S-poisoned animals. To determine if endogenous or exogenous sulfide was carried in a nonacid labile form in blood, we measured [sulfide] in the metabolism chamber (see below) with the polarographic sensor in trout blood in the presence or absence of 10 mM DTT, the concentrations used by Warenycia et al. (38). Both unspiked and blood spiked with sulfide to a final concentration of 50 μM were examined.
Sulfide production by steelhead and trout hearts in vitro.
Female steelhead (O. mykiss, Skamania strain, 3–7 kg) were captured by the Indiana Department of Natural Resources (DNR) during the fall migration and kept at the Richard Clay Bodine State Fish Hatchery until the spawning season (January-March). The fish were anesthetized in ethyl m-aminobenzoate methanesulfonate (MS-222), and after the spawn was collected by the DNR, the hearts were removed via a midventral incision and placed in 4°C buffered saline with glucose and transported back to the laboratory.
Rainbow trout (O. mykiss, 0.4–0.7 kg) were maintained as described above. They were stunned by a blow to the head, and the heart was removed via a midventral incision. Ventricles (∼0.5 g) were blotted, weighed, minced and placed in 1:3 wt/vol HEPES in the metabolism chamber (see below). The chamber was sealed, and within a minute most of the dissolved oxygen in the chamber was consumed by myocardial metabolism. The evolution of H2S gas was then measured with the polarographic sensor under these hypoxic conditions. Cysteine, initially as the acid salt and later as the free base, was added every 20 min in increasing amounts producing final concentrations of 0.1, 1, and 10 mM. After 20 min exposure to 10 mM cysteine, 100% O2 gas was injected in the chamber (estimated concentration 100 μmol/l), and sulfide production was monitored. After ∼30 min, when sulfide production had returned, the oxygen injection was repeated. Although the acid form of cysteine substantially lowered the pH of the buffered tissue, there was no difference in sulfide production between acid and free base forms.
Effects of hypoxia on plasma sulfide production in trout in vivo.
Trout were instrumented with an extracorporeal loop that pumped blood from a cannula in the ventral aorta across the polarographic H2S sensor and returned it to the fish via the caudal vein. Techniques for cannulating these vessels and utilization of the extracorporeal loop have been described previously (13, 26). Briefly, trout were anesthetized in benzocaine (1:24,000, wt/vol), and a midventral incision was made exposing the bulbus arteriosus and ventral aorta. The ventral aorta was cannulated with a 5-cm-long piece of silicone tubing (0.51 mm ID), and this was connected to polyethylene tubing (PE-60) and exteriorized. The wound was closed with interrupted sutures. The caudal vein was cannulated with PE-50 via a 1-cm incision on the lateral body wall near the caudal peduncle. The fish was revived and placed in a black tube inserted in a 10-liter chamber with aerated, through-flowing water at 14°C. The cannulas were connected to a peristaltic pump such that blood was withdrawn from the ventral aorta (1.5 ml/min) by the pump and pumped across a thermostated polarographic H2S sensor and returned to the fish via the caudal vein. The fish was made hypoxic by temporarily shutting off the in-flowing water and bubbling the chamber with 100% N2 until the Po2 fell to 40 mmHg. The N2 was then turned off, and the Po2 remained at this level for ∼15 min after which the water was turned back on and the chamber was aerated with room air. In two fish, the dorsal aorta was also cannulated (described below) and connected to a blood pressure transducer to ensure that this level of hypoxia produced the classical hypoxic bradycardia. In several experiments, the water Po2 was further decreased down to 20 mmHg before returning to normoxia. The H2S sensor was calibrated in position by substituting a vial containing a volume of HEPES buffer equivalent to the trout's estimated blood volume and serial additions of sulfide (as Na2S).
Plasma sulfide and sulfide metabolism in trout in vivo.
Trout were instrumented with an extracorporeal loop that drew blood from the dorsal aorta and returned it to the caudal vein. The dorsal aorta was cannulated percutaneously through the roof of the mouth with PE-60 tubing as described previously (26), and the caudal vein was cannulated with PE-50 and connected to the peristaltic pump and polarographic H2S sensor as described above. Sulfide (as Na2S) was injected in the return cannula distal to the pump. The stability of sulfide in the extracorporeal loop was examined by substituting a vial containing HEPES buffer for the trout. The volume of buffer was equivalent to the estimated blood volume of the fish.
Polarographic H2S measurements.
The polarographic H2S sensor was constructed after Doeller et al. (5), with the following modifications. A 32-gauge platinum wire anode was melted at the tip to form a bead, heat-sealed in a borosilicate capillary tube, and the tip ground to the widest point of the bead (∼500 μm) and polished. A 32-gauge platinum wire cathode was coiled around the capillary tube, and the whole assembly was inserted in an outer glass tube 2 mm in diameter. An O-ring held a silicone polycarbonate membrane in place and in conjunction with a threaded sleeve secured it in the flow-through cell or metabolism chamber (see below). The sensor was connected to an Apollo 4000 Free Radical Analyzer (WPI, Sarasota, FL) with 100 mV polarizing voltage.
Because the polarographic H2S sensor measures only H2S gas, it was necessary to measure the pH of each sample and calculate the total sulfide from the Henderson-Hasselbalch equation using a pKa of 6.6 (37°C, 140 mM NaCl; mammalian blood) and 6.9 (15°C, 140 mM NaCl; fish blood) after Hershey et al. (12). It should be noted that the pKa of 6.6 at 37°C results in 14% of total sulfide being present as H2S, one-half of the more commonly cited value of 30% (22, 29), which was derived from the pKa of sulfide at 20°C in water (7.05; see Ref. 21). The detection limit of our polarographic H2S sensor was 14 nM H2S gas, i.e., 100 nM total sulfide at pH 7.4, 37°C, and 115 mM NaCl. Sulfide consumption by the sensor at 37°C in 1.0 ml of Krebs buffer spiked to 10 μM Na2S was 3 pmol/min, less than a 0.1%/min decrease in concentration.
An acrylic water-jacketed flow cell (chamber volume ∼5 μl) was constructed for rapid polarographic measurement of sulfide in blood. Blood was drawn directly from the sampling syringe through the cell at 0.5 ml/min with a syringe withdrawal pump within 1 min of removal from the animal. A second sample from the same animal (a second animal if mouse) was spiked with Na2S (10 μM total [sulfide]) and passed through the flow cell to verify that the system was responsive. Iodometric titration of a 100 μM standard in Krebs buffer, pH 7.4, before and after passage through the entire flow-through system verified that there was not a significant loss of sulfide with this apparatus (all n = 7).
A water-jacketed metabolism chamber was constructed for measurement of sulfide consumption. It consisted of a flat-bottom glass tube ∼1.5 × 3 cm with an acrylic stopper machined to fit tightly, the latter with an injection port and ports for the polarographic H2S sensor and either an oxygen sensor or pH electrode. Whole blood (lamprey, trout, rat, pig, cow), bovine plasma, bovine red blood cell ghosts, or fatty acid-free BSA (0, 0.1, 1.0, and 5% in Krebs buffer) was placed in the metabolism chamber, the stopper was lowered until no air remained (final volume 1 ml), and the baseline was allowed to stabilize. Anoxic Krebs BSA solutions were sparged with 5% CO2-95% N2. Krebs buffer and ghosts were spiked with 10 μl of 1 mM Na2S (10 μM final concentration) and allowed to incubate for at least 10 min. In the case of 5% BSA Krebs, incubation was allowed to continue until the sensor signal had returned to baseline, after which the sample was removed and assayed for sulfide with the diffusion chamber/methylene blue method as described below. This was done to determine whether the sulfide was reversibly binding to the BSA. Whole blood was spiked with 10 μl of 1 mM Na2S sequentially.
Colorimetric assay of acid labile sulfide.
Assay of acid-labile sulfide in plasma and Krebs buffer with 5% BSA was based on the method of Stipanuk and Beck (33), modified by Geng et al. (10). Each diffusion chamber consisted of a 25-ml glass vial with three indentations around its wall that supported a polyethylene center well, cut out of a 5-ml test tube. A 3 cm × 3 cm piece of Whatman No. 1 filter paper was folded into quarters and placed in the center well such that no part of the paper touched the walls of the chamber. The vials had a screw cap with a Teflon seal. Zinc acetate (0.4 ml of 1%) was added to the filter paper to trap H2S as ZnS, and two glass beads were added to the chamber to aid in mixing the sample. The blood was divided into three aliquots immediately after removal from the animal. One aliquot of blood was spiked (spiked blood) with Na2S (final concentration 10 μM, 15 μM for LW rat), and all three aliquots were centrifuged for 3–5 min. Following centrifugation, 1 ml of untreated (control) plasma and 1 ml of plasma from the spiked blood (spiked blood) were added to the first and second chambers, respectively, immediately followed by 0.5 ml 50% wt/vol TCA, and quickly capped. A second aliquot of untreated plasma was then added to the third chamber, spiked to 10 μM (15 μM in the case of the LW rat) Na2S (spiked plasma), and TCA added as above. The time between the addition of Na2S to the blood and addition of TCA to the plasma was usually between 6 and 8 min and not more than 10 min. Chambers were incubated for 60 min on a shaker at 37°C (before incubation, chambers containing cow and pig plasma were at room temperature for ∼1 h during transport from the slaughterhouse to the laboratory). Following incubation, the center well with the filter paper and zinc acetate was removed and placed in a 5-ml glass vial containing 3.5 ml water. Next, 0.4 ml of 20 mM N,N-dimethyl-p-phenylenediamine hydrochloride in 7.2 M HCl was added immediately followed by 0.5 ml of 30 mM FeCl3 in 1.2 M HCl, and the vial was capped and gently inverted. Contents of the vials were transferred to 96-well plates, and the absorbance was read at 669 nm after 10 min. Absorbances were compared with those obtained from a Na2S standard curve run in Cortland buffer (when using fish plasma) or Krebs buffer (when using mammalian plasma) in parallel with the experimental samples. The lower detection limit for this assay was 1 μM.
Ion selective electrode assay of sulfide metabolism.
Sulfide antioxidant buffer (AOB) was made by dissolving 25 g sodium salicylate, 6.5 g ascorbic acid, and 8.5 g NaOH in water and diluting to 100 ml. The Accumet silver/sulfide combination ion selective electrode (Fisher Scientific, Pittsburgh, PA) was immersed in a blank solution of 5 ml of AOB and 10 ml Krebs buffer, and the potential was allowed to stabilize. The electrode was then transferred to a solution of 5 ml AOB and 10 ml Krebs buffer with 5% BSA that had been mixed immediately before transfer. At ∼1-h intervals, the electrode was switched between the same blank and BSA solutions. The electrode was rinsed in a separate blank between transfers. The result was compared with a standard curve made by serial additions of 100 mM Na2S to a 1:2 mixture of AOB and Krebs buffer following the manufacturer's instructions. Continuous recording of the electrode potential was done with Biopac MP30 hardware and BSL Pro 3.7 software (Biopac Systems, Goleta, CA).
The following buffers were used: Krebs-Henseleit (mammalian; in mM): 115 NaCl, 2.5 KCl, 2.46 MgSO4, 2 CaCl2·2H2O, 5.6 glucose, 1.38 NaH2PO4, and 25 NaHCO3, pH 7.4; cortland bicarbonate (fish; in mM): 124 NaCl, 74.6 KCl, 0.57 MgSO4, 2 CaCl2, 5.5 glucose, 12 NaHCO3, 0.09 NaH2PO4, and 1.8 Na2HPO4, pH 7.8; HEPES (trout in vivo; in mM): 145 NaCl, 3 KCl, 0.57 MgSO4, 2 CaCl2, 5 glucose, 3 HEPES, and 7 HEPES sodium salt (HEPES-Na), pH 7.8; and potassium phosphate (100 mM KH2PO4/K2HPO4).
Sodium sulfide nonahydrate crystals were placed in a glass syringe that was then sealed with a rubber stopper and sparged with N2. N2-bubbled distilled water containing 50 μM diethylenetriaminepentacetic acid (to chelate iron) was then injected in the syringe through the stopper to dissolve the Na2S. Stock solutions were made fresh daily. This method was verified by iodometric titration (3). Unless otherwise stated, all chemicals were purchased from Sigma-Aldrich (St. Louis, MO).
Half-times for sulfide consumption were determined from exponential decay equations fit by Table Curve (Jandel, Chicago, IL). Statistical significance was determined by Students' t-test or paired t-test, and correlation among groups was determined by a Spearman rank order correlation test using SigmaStat or SigmaPlot (Systat Software, San Jose, CA). Results are given as means ± SE; significance was assumed when P ≤ 0.05.
Sulfide in blood, polarographic measurements.
Sulfide was not detected in blood or plasma from any animal (Table 1). Total sulfide in blood spiked to 10 μM Na2S did not exceed 2 μM, indicating that while the polarographic H2S sensor could detect sulfide in blood, most sulfide was rapidly removed (Table 1 and Fig. 1, A and B).
Polarographic measurement of sulfide metabolism in whole blood, bovine plasma, albumin, and red blood cell ghosts.
Blood from all animals tested consumed sulfide, as did bovine plasma and BSA. When Na2S was added to whole blood, plasma, or BSA, the concentration of sulfide measured with the polarographic H2S sensor decayed exponentially to baseline (Table 2 and Fig. 1, C–E). Sulfide consumption increased with BSA concentration, but consumption was independent of ambient oxygen (Fig. 1F) and was essentially nil in protein-free solutions (Fig. 1E, inset). The half-time for decay of a 10 μM Na2S spike in buffer following addition of an excess of zinc acetate, which very rapidly removes sulfide from solution, was 23.7 ± 0.6 s at 15°C and 13.0 ± 0.2 s at 37°C (all n = 4). Red blood cell ghosts had no effect on sulfide consumption or sensor response (data not shown).
Colorimetric measurement of acid-labile sulfide in plasma and BSA.
Acid-labile sulfide values in plasma are shown in Table 3. Control plasma was ∼2 μM in lamprey, trout, cow, and pig, not detected in LW rat, and 4.3 μM in HSD rat. Acid-labile sulfide in plasma isolated from spiked whole blood had returned to control levels by the time of assay. Recoveries for the spiked plasma were >60%. Krebs buffer with 5% BSA that had been spiked to 10 μM in the metabolism chamber and assayed after the polarographic H2S sensor reading had returned to baseline contained 2.8 ± 0.3 μM acid-labile sulfide, not significantly different from an unspiked control 3.3 ± 0.6 μM (n = 4).
Sulfide production in BSA and trout plasma measured with ion selective electrode.
Sulfide was rapidly liberated from 5% BSA upon mixing with alkaline AOB and exceeded 1 mM by 12 h (Fig. 2). Sulfide was produced at a similar rate from trout plasma (data not shown). Acidification of the 12-h samples produced the characteristic odor of H2S that was not evident when freshly mixed samples of BSA or trout plasma in AOB were acidified.
Liberation of sulfide from blood by DTT.
Addition of the disulfide reductant DTT (final concentration 10 mM) did not generate sulfide from either unspiked trout blood or from blood spiked with 50 μM sulfide (data not shown). No further studies with DTT were conducted.
Sulfide production by steelhead and trout hearts in vitro.
Sulfide generation from minced trout heart increased when cysteine was added to the reaction mixture (Fig. 3). Injection of oxygen transiently produced net sulfide consumption (Fig. 3A) but did not affect the overall rate of sulfide production once the oxygen was consumed (Fig. 3B). These experiments show that tissue [sulfide] is inversely related to oxygen availability.
Effects of hypoxia on plasma sulfide production in trout in vivo.
Sulfide was not detected in blood drawn directly from the ventral aorta of unanesthetized trout via an extracorporeal pump (Fig. 4A). Reducing ambient Po2 from normoxia (∼145 mmHg) to 40 mmHg was sufficient to produce a profound reflex hypoxic bradycardia but did not increase the [sulfide] in ventral aortic blood. In two fish, ambient Po2 was further reduced from 40 to 20 mmHg, also without an effect on blood [sulfide].
Effects of exogenous sulfide on blood sulfide in trout in vivo.
Sulfide was undetectable in blood drawn via an extracorporeal pump from the dorsal aorta of an unanesthetized trout, and injection of an amount of Na2S sufficient to theoretically raise plasma sulfide level to 30 μM produced only a transient 1.6 μM increase in sulfide (Fig. 5). The same amount of Na2S added to a volume of buffer equivalent to the trout's blood volume produced a sustained 30 μM sulfide reading (Fig. 5, inset).
In the present study, we used a polarographic H2S sensor to directly measure H2S gas in blood and plasma without any chemical modification other than adding an anticoagulant. We found that sulfide was undetectable in plasma and blood and that exogenously applied sulfide was rapidly removed. We also found that trout heart tissue in vitro produced sulfide in the presence of cysteine under hypoxic conditions, whereas sulfide was consumed when the tissue was normoxic In vivo, however, blood exiting the heart does not contain measurable levels of sulfide, even when trout were rendered hypoxic by reducing ambient oxygen. Our studies support the predictions of Szabó (35) and Li and Moore (20) and confirm earlier and generally ignored reports that sulfide does not circulate in the plasma at measurable concentrations. They also provide evidence that the balance between sulfide production and its metabolism by available oxygen may be a key component in oxygen-dependent (“oxygen sensing”) responses in a variety of tissues, including ischemic preconditioning in the myocardium. We could find no evidence, however, that sulfide exists in the circulation even as a transient response to environmental hypoxia and therefore it is doubtful that sulfide serves as a blood-borne signaling molecule.
Many of the studies that have implicated sulfide signaling in a variety of physiological processes, including neural, cardiovascular, gastrointestinal, genitourinary, and immune systems, have utilized “physiological” concentrations of sulfide in the range of 30–300 μM. These values are based on colorimetric assays or ion selective electrode assays performed within the past eight years (see list in supporting materials), each of which depends on harsh chemical treatment (strong acid or base, respectively) before analysis. However, using the polarographic H2S sensor constructed in our laboratory, which responded in real time under physiological conditions with a resolution of 100 nM total sulfide, we were unable to find any evidence that sulfide exists in vertebrate blood or plasma (Table 1 and Fig. 1), or that even when added to blood in vitro or in vivo it would remain as free sulfide for any appreciable time (Table 2 and Fig. 5). In fact, when sulfide was added to whole blood, the peak concentration measured typically did not exceed 20% of that which was added (Fig. 1, A–D), and sulfide was also removed, albeit as a slower rate, by plasma and BSA (Table 2 and Fig. 1, A–E). Because we showed that red blood cell ghosts did not affect sulfide disappearance, the rapid fall in sulfide likely reflects entry in the red blood cell where it may be bound to hemoglobin or metabolized.
Sulfide as a paracrine signal in ischemic preconditioning.
To determine if blood [sulfide] could transiently increase, we selected the tissue with the greatest probability of increasing sulfide production, the hypoxic trout myocardium. Our choice was based on several observations. First, there is recent evidence that ischemic preconditioning, the ability of short bouts of ischemia to protect the heart from a subsequent prolonged ischemic period, is mediated in the rat heart by sulfide (27, 31); second, ischemic preconditioning has been demonstrated in trout hearts (9); third, trout hearts are continuously exposed to some degree of hypoxia (8); and fourth, tissue [sulfide] appears to be inversely related to Po2 (25).
Previously, we proposed that sulfide is involved in oxygen sensing in vascular smooth muscle (25) and in the urinary bladder (6). The ability of blood vessels to “sense” oxygen is important in mediating hypoxic vasodilation of systemic vessels and hypoxic vasoconstriction in pulmonary vessels. This helps couple perfusion to metabolism in the former and perfusion to ventilation in the latter. However, the oxygen sensor mediating this process has been the subject of considerable debate (cf. Refs. 1 and 14). In our model, the concentration of vasoactive sulfide is regulated by the simple balance between constitutive sulfide production in the cytosol and the Po2-dependent oxidation of sulfide to inactive metabolites, i.e., sulfite and/or sulfate, in the mitochondria. This model has been supported by demonstration of sulfide production by vascular tissue, the identical effects of hypoxia and sulfide on a variety of vessels, and the ability of inhibitors of sulfide biosynthesis to inhibit the hypoxic responses (25). This is further supported by recent evidence for sulfur cycling between the cell cytoplasm (where it is reduced to H2S) and oxidation in the mitochondria of eukaryotic cells (30). It is also likely that the presence of mitochondria in lamprey and trout red blood cells accounts for the 5- to 10-fold faster rate of sulfide removal by fish blood compared with mammalian blood (Table 2). (These differences are not due to temperature effects on the polarographic H2S sensor, since the sensor recovery after rapid sequestration of sulfide by addition of zinc acetate was faster at 37°C than at 15°C.) Our observation of sulfide production by the hypoxic trout myocardium and the rapid reversal to net sulfide consumption in the presence of oxygen (Fig. 3), in conjunction with related findings in other tissues (6, 25), suggests that oxygen-dependent sulfide metabolism may be a general property of tissues in which Po2 is coupled to a physiological response.
The trout heart has two distinct layers of myocardium, a relatively thin outer compact layer supplied by coronary arteries and an inner spongy myocardium that derives its oxygen directly from systemic venous blood. Under most circumstances, cardiac function is unaffected by coronary artery ligation, implying that much, if not all, routine cardiac work is performed by the spongy layer (32). Even at rest in a normoxic environment, spongy myocardial cells are exposed to moderate hypoxia, since systemic venous Po2 is <50 mmHg; during exercise, venous Po2 may drop to 15 mmHg (8). Because ambient hypoxia only exacerbates this situation (32), it was predicted that ambient hypoxia sufficient to produce a reflex hypoxic bradycardia would also be a sufficient physiological stimulus to increase sulfide in blood. Because this did not occur (Fig. 4), it seemed unlikely that sulfide could serve, even transiently, as a circulating signal.
Our studies do, however, suggest that sulfide is involved in oxygen “sensing”/signal transduction in ischemic preconditioning. Recent studies have shown that application of sulfide mimics ischemic preconditioning in rat hearts by mitigating the pathological consequences of the final long-term myocardial ischemia (27, 31). It is not known how ischemia initiates the sulfide response. Our study suggests that the decrease in oxygen itself is sufficient to produce a net increase in intracellular sulfide and that H2S gas or HS− then acts locally. Ischemic preconditioning has also been demonstrated in trout hearts (9), indirectly suggesting that sulfide may be involved here as well.
Historical perspective of sulfide in the circulation.
There is a very clear demarcation in blood sulfide levels reported in studies published before or after 2000 (summarized in supplemental information) (Supplemental information for this article can be found on the American Journal of Physiology: Regulatory, Integrative and Comparative Physiology website.). Typically, the earlier studies anecdotally and experimentally support rapid consumption of sulfide by blood in vivo and in vitro and when directly measured show that plasma (or blood) sulfide is 2 μM or less (see supplemental information for details). In pioneering studies, Haggard (11) showed that quickly injecting 10 ml of 77 mM Na2S in a dog was rapidly lethal, whereas injecting five times that dose over 20 min did not harm the animal, leaving Haggard to conclude that sulfide was rapidly metabolized. Prior et al. (28) found in rats an inhaled LC50 (concentration at which 50% of the subjects died) of 335 ppm over 6 h of exposure to H2S, while the LC10 over the same time was 299 ppm. Assuming that H2S readily equilibrates across the alveolar membranes, this would produce sulfide values in the plasma of 157 and 143 μM, respectively (see supplemental information for calculations). Prior et al. (28) suggested that the steepness of the dose-response curve was due to an overload of the sulfide detoxification system, implying that the majority of the lower dose is being continuously detoxified. Given this capacity to metabolize sulfide, it is difficult to imagine that there would be any opportunity for sulfide to accumulate under resting conditions. It is also clear that H2S can leave the blood across the alveolar membrane, since intravenously injected NaHS is partially recoverable as H2S in exhaled air (7). If 10 μM total sulfide in plasma equilibrated with alveolar gas, H2S should be readily detectable in exhaled air at a concentration of 22 ppm. However, reported levels of H2S in human exhaled (24) and end-expiratory air (34) are only ∼50 ppb, >400-fold less than predicted. In fact, 22 ppm is within the range of H2S concentrations in the typical human flatus (18) and slightly higher than the Occupational Safety and Health Administration Permissible Exposure Limit (4).
Methodological consideration of recent sulfide measurements.
With only one exception, all studies after 2000 reported that plasma (or blood) sulfide was between 10 and 300 μM. The exception is an anecdotal report of ≤5 μM in rat blood measured with a polarographic H2S sensor similar to ours (16). There are a number of factors that could contribute to the elevated sulfide values in recent studies. These include modifications to the original methods without proper verification of their accuracy, use of impure NaHS as a standard and nonlinearity of standard curves at low [sulfide]. Artificial generation of sulfide from proteins is also quite likely when using the strong alkaline AOB associated with the ion selective sulfide electrode, as clearly shown in Fig. 2. In fact, Khan et al. (15) also observed that directly mixing blood and AOB results in protein desulfuration. These points are considered in greater detail in the supplemental information.
In summary, our studies with the polarographic H2S sensor indicate that circulating free sulfide levels are considerably lower than previously reported, and it is unlikely H2S or sulfide functions as a signaling molecule in the circulation. However, the inverse coupling of myocardial sulfide and Po2 in trout hearts is evidence of the initial oxygen sensing mechanism in ischemic preconditioning and supports a paracrine/autocrine mode of sulfide signaling.
H2S is one of the most ancient gases on earth and has accompanied living organisms throughout the course of their evolution. Not surprisingly, this association has often been exploited to the advantage of the organism, and our perception of sulfide biology has changed from viewing it as a toxic gas to a biologically relevant “gasotransmitter.” The precedent for this was originally established by studies on nitric oxide (NO), and the potential of H2S as a mediator, like NO, of a variety of physiological and pathological processes has caught the attention of basic scientists and clinicians alike. In the present experiments, we provide additional support for the role of sulfide metabolism as an oxygen “sensor” that may provide the long-sought-after couple in hypoxia-mediated excitation and activation in a variety of cells. However, as we also show, we must temper our exuberance for this novel signaling system while the methodology is refined.
This work was supported, in part, by National Science Foundation Grant IOS 0641436.
We thank Y. Gao for technical assistance; V. Western, K. Stewart, and Dr. E. McKee and Martins Custom Butchering for assistance in obtaining blood samples; and B. Bell, D. Meunick, and the staff at the Richard Clay Bodine State Fish Hatchery, Indiana Department of Natural Resources for help in obtaining steelhead tissues.
↵1 There is considerable ambiguity in the literature regarding the terms “hydrogen sulfide,” “H2S,” “sulfide,” and “total sulfide.” H2S is a gas and when dissolved it is also a weak acid in equilibrium with HS− and S2− (although the latter is negligible at physiological pH). It is not currently known whether H2S, HS−, or both are biologically active. Most analytical methods measure total sulfide, i.e., H2S plus HS−. The polarographic sensor used in the present study only responds to the free gas, but to remain consistent with prior literature we converted our measurements to total sulfide using the Henderson Haselbalch equation and the known, or measured, pH of various solutions. For clarity, we use the term H2S only in specific reference to the gas, and in all other instances sulfide will refer to combinations of the gas and anion, i.e., total sulfide.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society