A Practical Look at the Chemistry and Biology of Hydrogen Sulfide

Abstract

Significance: Hydrogen sulfide (H₂S) is garnering increasing interest as a biologically relevant signaling molecule. The effects of H₂S have now been observed in virtually every organ system and numerous physiological processes. Recent Advances: These studies have not only opened a new field of “gasotransmitter” biology, they have also led to the development of synthetic H₂S “donating” compounds with the potential to be parlayed into a variety of therapeutic applications. Critical Issues: Often lost in the exuberance of this new field is a critical examination or understanding of practical aspects of H₂S chemistry and biology. This is especially notable in the areas of handling and measuring H₂S, evaluating biosynthetic and metabolic pathways, and separating physiological from pharmacological responses. Future Directions: This brief review describes some of the pitfalls in H₂S chemistry and biology that can lead or have already led to misleading or erroneous conclusions. The intent is to allow individuals entering or already in this burgeoning field to critically analyze the literature and to assist them in the design of future experiments. Antioxid. Redox Signal. 17, 32–44.

Introduction

Fifteen years ago the pioneering studies of Kimura's group suggested that hydrogen sulfide (H₂S) was a biologically relevant signaling molecule (1). Since then, the field of H₂S biology has exploded, seemingly without much of the initial skepticism or reservation that accompanied its gaseous predecessor, nitric oxide (NO). Excitement over this molecule, from basic research to translational medicine, has spawned nearly as many reviews as primary research articles (cf. 17, 24, 32, 38, 44, 49, 55, 56, 68, 79). In many instances H₂S seems to fit the criteria for a “gasotransmitter” as adapted from the well-known criteria for neurotransmitters by Wang (74); i.e., gasotransmitters must 1) be a small molecule of gas, 2) be freely membrane permeable, 3) be endogenously and enzymatically generated in a regulated manner, 4) have well-defined specific functions at physiologically relevant concentrations, and 5) act on specific cellular targets. The plethora of observations that exogenous H₂S affects virtually every organ system and tissue appears to bear this out. Not surprisingly, a variety of synthetic H₂S “donating” compounds have been synthesized and are currently being used in a number of clinical trials (55).

Surprisingly few reviews, however, have looked beyond the exuberance and critically examined the field “warts and all.” In one seminal review, appropriately titled “Endogenous production of H₂S in the gastrointestinal tract: still in search of a physiologic function,” Linden et al. (42) reviewed and modified Wang's five criteria for a gaseous signaling molecule; i.e., it must 1) be a gas, 2) be endogenously and enzymatically generated in a regulated manner, 3) with exogenous application, cause a well-defined physiologic effect at physiologically relevant concentrations that mimics the effect of the endogenously produced H₂S on tissue activity, 4) act at specific cellular targets, as demonstrated by competitive antagonism, 5) employ a specific mechanism of inactivation. The authors then went on to show that (with emphasis on the gastrointestinal tract) H₂S fulfills the first criterion, partially fulfills the second criterion (a mechanism for regulating H₂S production has yet to be identified), partially fills the third criterion (the definition of “physiologically relevant concentrations” remains to be determined), partially fills the fourth criterion (competitive antagonism and receptor binding kinetics remain to be determined), and fulfills the fifth criterion since tissue oxidation of H₂S is well known. The authors conclude that while H₂S is a promising candidate as a gasotransmitter, a number of questions remain to be answered.

Innovation.

H₂S biology is an exciting field that is attracting an ever-increasing number of investigators with wide and varied interests. This review emphasizes a number of areas that have either become erroneously entrenched in the literature or are especially problematic in working with this gas.

“Man is so intelligent that he feels impelled to invent theories to account for what happens in the world. Unfortunately, he is not quite intelligent enough, in most cases, to find correct explanations.”—Aldous Huxley

The present review takes another critical look at H₂S chemistry and biology. The intent is to examine areas that have either been overlooked, generated confusion, produced conflicting opinions, or raised interesting and pressing questions. Clearly, not all areas in this expanding field can be covered in detail in this brief review, and I apologize for topics or significant articles that are not covered.

H₂S Chemistry

Hughes et al. (25) reviewed the practical aspects of preparing and working with H₂S under a variety of conditions. The following paragraphs emphasize related aspects of H₂S measurement and chemistry in physiological solutions and buffers (see also Table 1).

Table 1.

Common Methods for Measuring H₂S in Blood and Tissues

Method	Sensitivity	Blood (μM)	Tissue (μM)	Interferences	Anoxia	Real-time	Subcellular
Methylene blue	>1 μM	20–300	40–200	Protein, acid-labile sulfides, absorption spectra non-linear, time sensitive	yes	no	no
Sulfide ISE	>1 μM	20–300	20–300	Strong alkalinity required, time-sensitive liberation of sulfide from protein	no	no	no
GC-FPD, IC	low nM	nd	nd	High pH required, IC requires oxidation to sulfate	yes	no	no
MBB	low nM	0.1–1		Unknown sulfide “sink” in blood, tissue extraction suggested, reaction products stable	yes	no	no
Headspace, CG	low nM	0.01	low nM		yes	no	no
Polarographic (amperometric)	0.2 μM	nd–<1	nd–<1	Electrode sensitive to temperature, pressure, requires frequent calibration	no	yes	no
SSFD	>1 μM			Time delay, sensitivity	no	no	yes

GC-FPD, gas chromatography with flame photometric detector; IC, ion chromatography; ISE, ion-selective electrode; MBB, monobrombimane; SSFD, sulfide-sensitive fluorescent dye; nd, not detectable.

Measuring H₂S

A number of recent reviews have summarized the advantages and disadvantages of extant methods for measuring H₂S in buffers, biological fluids, and tissue samples (25, 53, 72), and the reader is referred to these for further details. This section focuses on the half dozen methods most frequently used for H₂S measurements in biological samples. Emphasis is placed on their limitations and potential for error.

The most commonly used method is the colorimetric generation of methylene blue by the reaction of H₂S with N,N-dimethyl-p-phenylenediamine sulfate. This method was developed for measuring H₂S in aqueous solutions and has more recently been adapted (generally inappropriately) to measuring H₂S in plasma and biological fluids in which it produces spuriously high values (53). In addition, the methylene blue method lacks sensitivity at low H₂S concentrations, and Hughes et al. (25) have recently shown that due to the formation of dimers and trimers of methylene blue, the absorption spectra of aqueous solutions of methylene blue do not obey Beer's law. In fact Beer's law only seems applicable at sulfide concentrations below 1 μM. These values are well below the 20–300 μM commonly reported in biological samples using this method. Because the methylene blue reaction proceeds under acidic conditions, it is not possible to separate free H₂S from acid-labile H₂S derived from iron–sulfur groups in cytochromes and other iron centers; the latter may be several thousandfold in excess of free H₂S (cf. 37). This method requires that the sample is mixed with analyte 20 to 30 minutes before the color is fully developed and the color intensity changes with time. The methylene blue method may be suitable in experiments in which there is no free protein, such as buffer-perfused organs. However, it cannot be used for continuous measurement of H₂S under physiological conditions, in real-time, or for simultaneous measurement of O₂.

Sulfide-specific ion-selective electrodes (ISEs) that measure S²⁻ have also been used on biological samples, but they too are not without error. Formation of S²⁻ requires a strong (pH>11) alkaline solution, generally referred to as the “antioxidant buffer,” to drive the equilibrium between H₂S, HS⁻, and S²⁻ to favor S²⁻. These alkaline conditions appear to promote hydroxyl replacement of cysteine sulfur thereby producing erroneously high sulfide concentrations that continue to increase with time (76). Like the methylene blue method, the ISE cannot provide information in real-time on unadulterated samples nor can it be used for simultaneous measurement of H₂S and O₂. Furthermore, the sensitivity appears to be insufficient for most biological samples.

Ubuka et al. (73) developed a method for measuring H₂S and acid-labile sulfide by initially trapping the evolved gas in an alkaline solution followed by gas chromatography with flame photometric detector and ion chromatography. They were able to measure acid-labile sulfide in tissues as low as 100 nmol/g tissue but did not detect free H₂S. The method is more sensitive than either methylene blue or ISE but cannot provide real-time measurements of H₂S in unadulterated samples.

The well-known method of measuring thiols by derivatization with excess monobromobimane (MBB) and subsequent measurement of the stable sulfide-diamine product with reverse phase high pressure liquid chromatography (HPLC) coupled with fluorescence detection has recently been used to measure plasma H₂S (63, 77). This method is described as being “suitable for sensitive quantitative measurement of free hydrogen sulfide in multiple biological samples such as plasma, tissue and cell culture lysates, or media” (63). Reagents must be made up in deoxygenated solutions and MBB must be protected from light. Maximum yield is obtained at a pH of 9.5 and at 1% oxygen or lower. The limit of detection is 5 nM sulfide, and the reaction product is reported to be stable. Real-time measurements of H₂S under normoxic, physiological conditions are not possible.

Analysis of H₂S evolution into headspace gas and subsequent gas chromatography has been used to measure both tissue and plasma H₂S in the nanomolar range (18, 37). By carefully controlling pH, Levitt et al. (37) were able to separate free sulfide (H₂S gas and HS⁻) from acid-labile sulfide. Although this method can be used to measure H₂S in unadulterated biological samples, it is not capable of measuring H₂S in real-time. In addition, these determinations are usually performed under anoxic or severely hypoxic conditions with an extended period of time required for evolution of gas into the headspace.

The polarographic (amperometric) electrode, originally developed by Jeroschewski and Steuckart (29) and modified by Kraus and Doeller (34), operates similar to the standard Clark O₂ electrode and uses a relatively H₂S-gas–specific membrane and polarizing voltage. This method can be used with otherwise unadulterated samples of tissue homogenates or pieces of tissue, with cultured cells, and even in an extracorporeal loop for continuously monitoring plasma H₂S in unanesthetized animals (76). The polarographic electrode directly measures the concentration of dissolved H₂S gas in real-time and the electrode has a detection limit of 10 to 20 nM H₂S gas or ∼100–200 nM total sulfide (76). Because this method only measures H₂S gas, total sulfide must be calculated from concomitant pH measurements. Polarographic electrodes also consume sulfide, albeit slowly. However, this can become problematic when measurements are made in a small volume in which there is an increased probability of lowering the total sulfide concentration, or in unstirred conditions in which electrode consumption can reduce sulfide concentration in the immediate area surrounding the tip of the electrode. Amperometric electrodes are generally pressure and temperature sensitive and prone to drift, necessitating frequent calibration. They can be used in the presence of oxygen and offer the particular advantage of being able to simultaneously compare oxygen and H₂S concentrations (56).

Wintner et al. (77) compared the amperometric response of the sensor to the MBB method. They observed that while both methods produced essentially identical responses when Na₂S (10 μM final concentration) was added to either HEPES buffer or human plasma, the amperometric sensor only detected a brief, transient increase in total sulfide when Na₂S was added to human whole blood, whereas the sulfide measured with the MBB method was more than three times that of the sensor and only slowly decreased over the following 30 minutes (Fig. 1). They attributed the differences between the two methods to the ability of the MBB method to detect a “reversible sulfide sink” in addition to free H₂S. This sink was not due to H₂S gas nor the common oxidation products of sulfide (sulfite, sulfate, or thiosulfate), and the authors proposed that it is a loosely bound moiety, still in the −2 oxidation state. They also suggested it was a form of a persulfide (RS_nH) on the cysteine residues of one or more proteins. It is unlikely that this sulfide sink is either acid-labile sulfide or sulfane sulfur because neither of these appears to be released by H₂S-doped whole blood (55). It remains to be determined whether or not this unknown sulfide can be released under physiological conditions and, if so, what mechanism is involved. If this is indeed a heretofore unidentified sulfide moiety it may help explain how the effects of exogenously administered H₂S persist long after the gas has disappeared.

FIG. 1.

Comparison of H₂S measured in buffer, human plasma, and human whole blood with the amperometric H₂S sensor and the monobrombimane (MBB) method. Na₂S (final concentration 10 μM) was added to HEPES buffer (50 mM HEPES, 155 mM NaCl, 3 mM glutathione, 1 nM diethylenetriaminepentaacetic acid, pH 8.0), human plasma, or human whole blood. Both methods produce similar results in HEPES buffer and plasma. However, in whole blood there was only a transient spike in sulfide when measured with the amperometric sensor, whereas more than three times as much sulfide was measured with the monobromobimane (MBB) method and sulfide concentration remained elevated for at least 30 minutes. Redrawn from Wintner et al. (77), with permission.

Sulfide-sensitive fluorescent dyes, similar to the well-known ion-sensitive dyes such as the calcium reporter Fura-2 are currently being developed (43, 60). These have the potential to resolve many of the key questions regarding H₂S signaling mechanisms and effector pathways. To date, however, they appear to lack the sensitivity to measure endogenous H₂S at submicromolar levels and in real-time.

Sources of H₂S

The two most common sources of H₂S are the sulfide salts, sodium hydrogen sulfide (NaHS), and sodium sulfide (Na₂S). These compounds are frequently and erroneously called H₂S “donating compounds” and it has been reported that they slowly release H₂S when dissolved (47). However, other studies (13, 39) have shown that when either Na₂S or NaHS is dissolved, H₂S is formed essentially as fast as crystal solvation. The major difference between the study of Muzzaffar et al. (47) and those of DeLeon et al. (13) and Li et al. (39) was that the former measured sulfide with the methylene blue technique, whereas the latter two studies measured H₂S gas in real-time with polarographic electrodes. Calcium sulfide (CaS) has also been used as a H₂S “donor” (41), but this does not appear to convey any advantage over the other sulfide salts and, although not measured, CaS presumably forms H₂S upon solvation at the same rate as NaHS and Na₂S. H₂S gas can also be passed through distilled water or buffers but this is far less convenient and often considerably more hazardous (25).

A number of true H₂S “donating” drugs are also available (5, 45, 55, 59). These reportedly have the advantage of a slower and more sustained release of H₂S than the sulfide salts, and they can be combined with other drugs to achieve multiple therapeutic effects. Some of these drugs also require an enzymatic process to liberate H₂S, which should enhance H₂S delivery to the target cells; i.e., ACS6 release of H₂S is approximately four times greater when incubated with endothelial cells than in buffer (47). Rarely has the release of H₂S from these compounds been measured in real-time. Li et al. (39) reported a slow release of H₂S from GYY4137 [morpholin-4-ium 4 methoxyphenyl(morpholino) phosphinodithioate] in buffer when measured with the polarographic electrode; however, the release of H₂S by GYY4137 in blood was inexplicably measured with the methylene blue method and not with the polarographic electrode. This makes any direct comparison difficult.

pH

Dissolved H₂S is a weak acid in equilibrium: H₂S ↔ HS⁻ + H⁺ ↔ S²⁻ + H⁺. Only the first reaction is relevant for biological samples because the pK_a1 is in the physiological range. The pK_a2 for the second reaction is over 12, with some reports suggesting it is as high as 19 (25). For all practical purposes, S²⁻ is insignificant. Both pKs vary with temperature and salinity; the effect of temperature on pK_a1 can be estimated from the equation: pK_a1=3.122+1132/T, where T is degrees Kelvin (52). This is not insignificant because the pK_a1 changes from 6.98 at 20°C to 6.77 at 37°C. A 1 mM solution of NaHS or Na₂S at pH 7.0 will have 490 μM dissolved H₂S at 20°C and at 37°C the dissolved H₂S will be 372 μM, 25% less. A 1 mM NaHS or Na₂S solution at pH 7.4 will have 277 μM dissolved H₂S at 20°C and at 37°C the dissolved H₂S will be reduced by 30% to 191 μM. Thus, failure to correct for temperature can account for as much as a 30% error in estimating the amount of dissolved H₂S under physiological conditions.

H₂S-forming salts require protons from the solvent, and their solvation can have a substantial effect on the pH as it alkalinizes the solution. This can become especially problematic for solutions of Na₂S in which solvation requires two protons to produce H₂S; NaHS only requires one proton. Dombkowski et al. (15) and DeLeon et al. (13) have shown that the buffering capacity of many physiological buffers begins to fail around 1 mM sulfide. In Krebs–Henseleit buffer, commonly used in mammalian preparations, 1 mM Na₂S increases the pH by nearly one-half unit and 10 mM Na₂S increases it by over 2 pH units. As expected, this is considerably greater than that produced by NaHS for which a 10 mM solution increases the pH by a little over 0.5 pH units. It is not always clear whether studies that employ millimolar concentrations of sulfide salts titrate their solutions prior to use.

Oxygen

It is well known that H₂S is oxidized in solution especially in the presence of metal catalysts (9) and biological tissue (57) (discussed in a later section). Spontaneous H₂S oxidation is also pH dependent (8). As shown in Fig. 2, the rate of spontaneous H₂S oxidation increases nearly tenfold over the range of commonly encountered physiological pH (6.5–8.0). Since it appears that much of this spontaneous oxidation of H₂S is due to metal catalysts, especially ferric iron, that exist as impurities in the water, removing them with a chelator such as diethylenetriaminepentaacetic acid (DTPA) will greatly delay spontaneous oxidation. Hughes et al. (25) found that in the absence of DTPA, HS⁻ absorbance was halved in 3 hours, which argues against the rapid oxidation suggested by Wintner et al. (77); whereas, in the presence of 100 μM DTPA it did not change over this period. Of course ion chelation can also remove essential minerals when done in the presence of tissues. However, volatilization of H₂S, as described in the following section, appears to contribute far more to spontaneous H₂S loss than oxidation in a variety of biological experiments.

FIG. 2.

The effect of pH on the rate of spontaneous H₂S oxidation. At pH<6 H₂S oxidation is nil, it then reaches a maximum at pH 8 before falling to a nadir at pH 9. Over the range of physiological pH, ∼6.5 to 8.0 the rate of H₂S oxidation increases approximately tenfold. Modified from Chen and Morris (8).

Volatilization

Because H₂S is a gas, the concentration of H₂S in solution can also be affected by volatilization, and the rate of H₂S evolving from solution is proportional to the surface area at the air–liquid interface. This is especially problematic in most, if not all physiological experiments because the tissues require continuous oxygenation and carbon dioxide must be eliminated. In tissue culture this is usually accomplished with a large surface/volume ratio and a monolayer of cells. With isolated organs or tissues, additional gassing is required and this often entails bubbling the tissue with some mixture of air, oxygen, and carbon dioxide, or circulating the buffer through an oxygenator. Control studies on the rate of H₂S volatilization as a function of the oxygenation process should be used to correct for H₂S loss. While H₂S volatilization is obvious to anybody working in this field, the impact of volatilization on the concentration of H₂S in biological experiments seems to have been overlooked.

To examine the potential for H₂S volatilization under typical laboratory conditions, we (13) measured the concentration of H₂S gas in buffer in real-time with a polarographic electrode under three experimental conditions commonly used in cardiovascular studies, tissue culture well plates in which gases exchange passively across the well surface, muscle myograph baths that are bubbled via a glass frit, and the Langendorff perfused heart apparatus in which buffer drains down a long glass column that is gassed.

As shown in Fig. 3A, there is an exponential loss of H₂S from 24-well tissue culture plates containing a physiological buffer. The rate constant (k) for H₂S volatilization is 0.13 min⁻¹ and the half time (t_½, the time for H₂S concentration to be halved) is 5 minutes (t_½=0.693/k). Because this process is passive, the rate constant and half time are independent of the initial H₂S concentration. This means that 13% of the H₂S remaining in the well is lost every minute and after 30 minutes only 2% of the initial H₂S will be left in the well. Thus, volatilization becomes problematic when exposing cells to H₂S. Lee et al. (36) exposed THP-1 and U118 cells to 10 μM NaHS in 24-well plates in 1 mL of Dulbecco's modified Eagle medium nutrient mixture F12 containing 5% fetal bovine serum and hydroxylamine (to inhibit endogenous H₂S synthesis). They measured H₂S in the cells by the direct methylene blue method and reported continual H₂S accumulation for 12 hours (Fig. 3B). According to the rate constants for H₂S volatilization from our study, the predicted concentration of H₂S after 12 hours should be around 4×10⁻⁴³ μM. Although the methylene blue method liberates acid-labile as well as free sulfide, the amount of acid-labile sulfide is not expected to accumulate during the duration of the experiment. This raises the question of what is being measured with the methylene blue method. Clearly, Lee et al. (36) show something is increasing, as have numerous other experiments (reviewed by Olson [55]). It is doubtful if this is either endogenous or exogenous H₂S. It remains to be determined if there is a different endogenous sulfide that is being measured with the methylene blue method, or if the molecule(s) being measured is some as yet unidentified storage form of H₂S. It is also possible that this is merely an artifact.

FIG. 3.

Comparison of the rate of 10 μM H₂S lost from cell-free 24-well plates (μmol/L) to the rate of H₂S accumulation by confluent U118 cells in 24-well plates (μmol/g protein). (A) H₂S is lost exponentially with a halftime of ∼5 minutes from 24-well plates when measured continuously with the H₂S polarographic electrode. (B) U118 cells accumulated H₂S (open circles) measured by the direct methylene blue method; transient line without symbols re-drawn from panel A. A, B; cell-free well plates redrawn from DeLeon et al. (13). B; H₂S accumulation by U118 cells redrawn from Lee et al. (36).

H₂S is lost from bubbled muscle myographs (k<0.2 min⁻¹ and t_½ <4 min) three times faster than from unbubbled myographs and also faster than from the 24-well tissue culture plates (Fig. 4A) (13). Presumably this is due to the additional air/liquid surface area from the fine bubbles. The rate of H₂S lost from myographs bubbled with 100% pure oxygen was slightly but significantly slower than when bubbled with 100% nitrogen. This shows that H₂S oxidation is insignificant within the typical 30-minute time frame of many experiments. It also suggests that there may be some, as yet unidentified, chemical or physical interaction between H₂S and N₂ or at the air–bubble interface. Because the response time of isolated blood vessels is typically 1–5 minutes, and H₂S is added in 10- to 20-minute intervals in cumulative dose–response experiments, there is little chance that the vessels will ever experience the desired H₂S concentration. In fact, when H₂S is added to the myograph and monitored with the polarographic electrode, the peak H₂S concentration measured is typically <90% of the predicted concentration. In a dose–response study on mouse aortas bubbled with 95% O₂/5% CO₂, Al-Magableh and Hart (2) only measured 91 μM H₂S with a polarographic H₂S sensor in myographs after cumulative additions of 10⁻⁶, 10⁻⁵, 10⁻⁴, and 10⁻³ M NaHS. They hypothesized that this discrepancy was due to differences between H₂S measured by the sensor and the actual dissolved NaHS. However, it is likely that volatilization was a major factor in their observations.

FIG. 4.

Real-time measurement of 10 and 100 μM H₂S volatilization from a muscle myograph (A) and Langendorff perfused heart apparatus (B). (A) In myographs, volatilization is slowest from unbubbled myographs and fastest when bubbled with 100% nitrogen (N₂). (B) in the Langendorff apparatus H₂S is pumped into a 10-mL beaker (simulating the heart) containing the polarographic H₂S sensor, and it then flows down the side of the apparatus where gas is equilibrated. A bolus of H₂S is injected into outflow from the pump and produces a transient increase in H₂S. By the time the bolus makes the second pass across the sensor (1 minute) the measured H₂S concentration is well below predicted values (dotted lines). Modified from DeLeon et al. (13).

H₂S volatilization from the Langendorff perfused heart apparatus is the fastest (Fig. 4B). When injected into the pump effluent, H₂S concentration transiently increases. However, in the time for one complete circulation (1 minute in this preparation) H₂S concentration had fallen to <20% of the desired concentration, at 2 minutes it was <10%, and it continued to decline exponentially thereafter. Half-times for the transient spike were <6 seconds and the remaining H₂S disappeared with half times of <3 minutes.

Rapid volatilization of H₂S creates several problems in designing and interpreting experiments. First, it is evident that the desired H₂S concentration is never attained because the outgassing process occurs immediately upon sample addition. Second, the rapid fall in H₂S concentrations that results from bubbling renders H₂S dose–response determinations rough estimates at best. Third, where the effects of H₂S are demonstrated for hours (or days) after a single treatment it is not clear if this is an actual response to H₂S. Furthermore, it is also unclear if a metabolite or decomposition product of H₂S produces the observed response or if H₂S initiates a long-lived signal cascade. Nevertheless, it seems likely that many of the dose- dependent effects of H₂S and dose–response curves that have been published in the literature need to be revisited as they most likely overestimate the actual tissue exposure. These studies also help explain, in part, why the concentrations of exogenous H₂S needed to produce biological responses are considerably greater than recent estimates of endogenous H₂S in blood or tissue.

H₂S as an oxidant/reductant

H₂S is often regarded as a strong antioxidant (38, 62). Exogenous H₂S has been shown to reduce hypochlorous acid (48) and is likely to reduce peroxynitrite (75), superoxide (7), and hydrogen peroxide (19), and H₂S has been proposed to effectively prevent oxidant injury in the brain (reviewed by Kimura [33]). While there is little doubt that H₂S is a reductant, how effective it is intracellularly remains to be determined. In order to demonstrate antioxidant activity in biological preparations the concentrations of H₂S typically are (or exceed) 100 μM in vitro (cf. 80). In vivo, an intraperitoneal injection of 5 mg/kg NaHS (20) and assuming distribution throughout the entire body water, will raise intracellular H₂S to ∼140 μM. It remains to be determined if endogenous H₂S is sufficiently concentrated in cells to have a similar physiological effect. Furthermore, Kabil and Banerjee (30) point out that H₂S is a relatively weak reducing agent, especially when compared with other intracellular thiols such as glutathione, and Carballal et al. (6) suggest that the protective effects H₂S cannot be completely accounted for by direct reactions with oxidants.

H₂S has also been suggested to increase oxidative activity in cells by inhibiting oxidative phosphorylation and thus favoring superoxide formation. For instance, Eghbal et al. (16) showed that exogenous H₂S (200–400 μM) dose-dependently stimulated the formation of reactive oxygen species (ROS) in isolated hepatocytes. This concentration is considerably greater than that required to inhibit oxidative phosphorylation, which is usually reported to be ∼20–40 μM (3, 11). It will be necessary to determine if these concentrations are approached under physiological conditions. It is also important to note that the H₂S concentrations used in most experiments are considerably greater than 40 μM and these may either provide excessive reductant or superoxide to the cellular milieu making it difficult to separate H₂S-specific effects from general metabolic depression or pharmacological reducing and/or antioxidant actions.

This also raises an interesting conundrum, if endogenous H₂S production increases to the point of inhibiting oxidative phosphorylation and stimulating ROS production, can these ROS then oxidize the H₂S? Could this be a cellular feedback mechanism for limiting H₂S concentration?

Alternative “forms” of H₂S

In addition to dissolved H₂S and HS⁻, reduced sulfide can be associated with iron-sulfur clusters in an acid-labile form and as persulfides, i.e., sulfane sulfur (RS-S). The term acid-labile is derived from the observation that sulfide can be released from these clusters when pH falls below 5.4 (27). Sulfane sulfur appears to be the only product of the tandem enzyme pathway involving the enzymes cysteine aminotransferase and 3-mercaptopyruvate sulfur transferase (3-MST) (30) and has been proposed to be the major pathway for sulfane sulfur (and potentially H₂S production) by the brain (27, 64). H₂S can be liberated from 3-MST–sulfane sulfur by the ubiquitous reductant, thioredoxin, and by dihydrolipoic acid, both present in cells (46). Both acid-labile and sulfane sulfur have been suggested to serve as a relatively stable form of H₂S for intracellular storage and as a site from which H₂S can be readily mobilized. These possibilities are considered in the following sections.

H₂S biology

H₂S in blood

At present there is no consensus regarding the concentration of H₂S in either blood or tissues, and this must be resolved in order to differentiate between physiological and toxicological effects of exogenous H₂S. The overwhelming majority of H₂S concentrations reported in blood of vertebrates since 2000 are in the range of 20–40 μM, and there are at least a half dozen reports of plasma H₂S in excess of 100 μM. Arguments against plasma H₂S in the micromolar range have been summarized in several recent reviews (53, 55) and include the following. First, high blood H₂S concentrations are generally associated with methods that were not developed for use with biological tissue, most often methylene blue method and sulfide ISEs. A good example of this is the revision of plasma H₂S concentration in mice which the original study (78) reported to be 40 μM H₂S in plasma when measured with the methylene blue method. This was subsequently revised down to 1.6 μM by the same group using the MBB method (63). However, even micromolar concentrations are not typically observed when measuring H₂S with polarographic electrodes or with some of the more modern methods employing gas chromatography and HPLC. Whitfield et al. (76) observed barely to nondetectable levels of H₂S in blood plasma from a variety of animals using the polarographic electrode and Levitt et al. (37) reported 0.07 μM H₂S in mouse blood using HPLC analysis of headspace gas. Novel methods or refinements of existing ones will be needed to measure H₂S in blood or plasma, if it exists at all. Second, there is no odor of H₂S in plasma even though the human nose can smell 1 μM in buffer (Table 2). In fact, 45 μM H₂S in plasma would be equivalent to breathing 100 ppm H₂S (Table 2) and this should cause eye and lung irritation and even begin to paralyze the olfactory system. Pulmonary edema appears at 100 μM H₂S. Clearly, these concentrations are unphysiological. Third, H₂S readily equilibrates across respiratory surfaces and, if it existed in plasma, should be readily exhaled. Rapid appearance of H₂S during administration of Na₂S (IK-1001) has been observed in exhaled air from rats (26) and humans (71). As shown in Fig. 5, H₂S began to appear in exhaled air within 5 seconds after initiation of intravenous H₂S infusion in rats and H₂S concentration oscillated with each breath. When infusion was stopped H₂S concentration began to fall by the second breath and was halved within 3 seconds. Similar results were observed when a bolus of H₂S was injected in humans, although with a longer circulation time and longer onset and recovery periods. Toombs et al. (71) also observed that H₂S in exhaled air from humans prior to Na₂S injection was some 20 ppb greater than room air (6 ppb) indicative of continuous H₂S excretion. If this is the case, and with efficient transpulmonary H₂S exchange, this would imply that plasma H₂S is ∼6 nM, clearly well below the resolution of blood H₂S assays. Fourth, with the assumption that H₂S equilibrates across respiratory membranes, Furne et al. (18) calculated that there is not enough sulfur in the body to sustain H₂S production for even 24 hours. Fifth, H₂S is rapidly removed from the circulation by tissues. Norris et al. (51) have shown that the buffer-perfused rat liver removes nearly 97% H₂S at a perfusate concentration of 150 μM H₂S and 50% clearance at 300 μM H₂S. Sixth, H₂S has been shown to be rapidly consumed by tissues in the presence of oxygen (Fig. 6A) (35, 56, 57). Thus, it is unlikely that there is even enough H₂S escaping from normoxic tissue to contribute to plasma levels. If H₂S does exist in the plasma, especially at levels sufficient to be considered physiologically relevant, which seems doubtful, there has yet to be a method that is sensitive enough to detect it.

Table 2.

Comparison of the Known Effects of H₂S Exposure to the Theoretical Equivalent Plasma Sulfide Concentration Assuming Equilibration Across Alveolar Membranes and No Blood or Tissue Metabolism

Ambient H2S (ppm)	Equivalent total plasma sulfide (μM)a	Effects
0.01–0.3	0.004–0.1	Threshold for detection
1–3	0.4–1	Offensive odor, headaches
10	4.5	8-hour occupational exposure limit in Alberta, Canada
15	6.7	15 minute exposure limit in Alberta, Canada
20–50	9–22	Eye and lung irritation
100	45	Olfactory paralysis
250–500	112–225	Pulmonary edema
500	225	Sudden unconsciousness (“knockdown”), death within 4- to 8-hours
1000	450	Immediate collapse, breathing ceases within several breaths

This table shows that the majority of plasma and tissue H₂S concentrations reported in the literature (20–300 μM) would range from noxious to fatal and all would be malodorous.

^aAll except equivalent total plasma sulfide column modified from Guidotti (22). Equivalent plasma sulfide is the theoretical sulfide concentration in plasma calculated after Whitfield et al. (76) (supplemental information) with the following assumptions: 1) H₂S freely equilibrates across the alveolar membranes (26, 71), 2) Henry's Law constant for H₂S at 37°C and 140 mM NaCl is 0.0649 M·atm⁻¹ (12), 3) 20% of total sulfide exists as H₂S gas (52), and 4) there is no H₂S metabolism in blood or tissues. Table modified from Olson (55), with permission.

FIG. 5.

Appearance of H₂S in exhaled air from rats infused intravenously with 0.6 mg/kg/min Na₂S. Upper trace shows that H₂S appears within 5 seconds after the onset of infusion and thereafter oscillates with each breath. At the end of infusion (lower trace) the concentration of H₂S falls within the first few breaths and is halved within 3 seconds. Redrawn after Insko et al. (26), with permission.

FIG. 6.

Inverse correlation between H₂S and O₂ in biological tissues. (A) Simultaneous measurement of H₂S and O₂ concentrations in homogenized rat lung measured in real-time with H₂S and O₂ polarographic electrodes. H₂S is produced in the presence of precursors of H₂S biosynthesis, cysteine and α-ketoglutarate, and in the absence of O₂. Injection of a small air bubble (arrow) increases O₂ concentration and immediately decreases H₂S concentration. H₂S production resumes after the O₂ is consumed. Modified from Olson and Whitfield (56); (B) Solid lines show O₂-dependent H₂S consumption by pulmonary arterial smooth muscle cells (PASMC), homogenized bovine lung (lung), and purified mitochondria (mito) compared with O₂ dependence of hypoxic pulmonary vasoconstriction of isolated bovine pulmonary arteries (vessel; dashed line). Percent activity refers to the degree of H₂S consumption (100%=all H₂S consumed) or percentage of hypoxic contraction (100%=maximum vessel contraction). Modified from Olson et al. (57). Horizontal arrows above the figure refer to the range of Po₂ in mitochondria (mito), cytosol (cyto), and tissues (tiss) reported in the literature. Half maximal H₂S consumption and vessel contraction occurs at approximately the same Po₂ and these Po₂s are below those encountered in normoxia, i.e., hypoxic conditions.

It has been suggested that H₂S may be transported in the blood either as acid-labile sulfide, sulfane sulfur, or some other reversible “sink” (77), and Levitt et al. (37) reported ∼1 μmol acid-labile sulfide in mouse blood. However, blood pH must fall below 5.4 to release H₂S from an acid-labile pool (27), which is highly unlikely. Furthermore, because blood is an oxidizing environment (50), it is difficult to imagine reducing conditions sufficient to release sulfide from a persulfide. Experiments designed to examine these possibilities have not supported either sulfur store. Acidifying mouse blood below 5.5 with trichloroacetic acid only produced 2.5 μM/kg whole blood measured with gas chromatography of headspace gas (37) and addition of 1 or 10 mM dithiothreitol to trout blood did not generate H₂S either in fresh blood or from blood primed with 10–100 μM H₂S (55). There are other possibilities as well; H₂S may be conveyed in the plasma in some other as yet unidentified form as proposed by Wintner et al. (77), although there is no evidence for this at present. It is also possible that H₂S is produced in blood. Searcy and Lee (61) demonstrated low levels of H₂S production by human red blood cells upon addition of elemental sulfur (S₈), but it is doubtful if this has physiological significance. Both cystathionine β-synthase (CBS) and cystathionine γ-lyase (CSE) have recently been shown to be secreted by the liver and endothelial cells, circulate in an active form in human plasma, and are able to generate H₂S from cysteine or homocysteine plus cysteine (4). These authors (4) suggested that local H₂S production could protect the endothelium from elevated homocysteine. It remains to be determined if these plasma enzymes could also contribute to plasma H₂S.

H₂S in tissue

The vast majority of studies reporting the concentration of H₂S in tissues, or the rate of H₂S production by tissues, have also employed either the methylene blue method or ISEs and have been performed under relatively unphysiological conditions. Arguments can be made against these findings on both experimental and theoretical grounds (18, 37, 55).

First, there is the potential that the acidic conditions required for the methylene blue assay generate additional sulfide from acid-labile stores, as described for blood. In fact, one recent study suggested that acid-labile stores in a variety of mouse issues may be anywhere from 10,000 to 40,000 times the concentration of free H₂S (37). Second, these experiments are routinely conducted under anoxic or very hypoxic conditions. This not only prevents the rapid and effective oxidation that normally occurs under physiological conditions but also increases the reducing potential of the milieu and can potentially liberate sulfane sulfur. Third, these assays are generally conducted with unphysiologically high cysteine concentrations (typically 1 or 10 mM). This excess cysteine could artificially increase the rate of H₂S production by shuttling sulfur metabolism through normally minor metabolic pathways (66). Fourth, other important substrates for H₂S production, such as homocysteine, are not typically provided in the reaction mixture and, under physiological conditions, these may be more important than, or as necessary as, cysteine for H₂S production (Fig. 7). For example, the condensation of cysteine and homocysteine may account for 25%–70% of the H₂S generated under more physiological conditions (10, 30, 65). Failure to have sufficient homocysteine in the medium can also bias interpretation of the enzymes involved in H₂S biosynthesis. H₂S generation from cysteine is primarily catalyzed by CSE, whereas H₂S production from the condensation of cysteine and homocysteine is accomplished by CBS. In a recent study, Kabil et al. (31) further demonstrated the effect of substrate concentration on H₂S production by liver; at saturating cysteine and homocysteine concentrations (20 mM) H₂S production was similar for CBS and CSE, whereas at physiological levels of substrate and enzymes, CSE activity accounted for 97% of the H₂S produced. Furthermore, CBS activity can be regulated by a variety of factors that are not included in the medium such as S-adenosylmethionine, a well-known allosteric regulator of CBS. This can also bias both the rate and pathways of H₂S biosynthesis. It should also be noted that these enzymatic studies are generally performed under anoxic conditions (cf. 31). Thus, while providing valuable information on enzymatic activity and important substrates for H₂S production, they do not provide information on tissue H₂S production under truly physiological conditions. Clearly additional studies that replicate the cellular environment are necessary to both determine the rate of H₂S biosynthesis as well as the enzymes involved. Fifth, compounds frequently employed to inhibit either CSE (propargyl glycine and β-cyanoalanine), CBS (aminooxyacetate), or both (hydroxylamine) and thus used to distinguish the rate and/or pathway of H₂S production are not specific for these enzymes and they are often poorly absorbed by tissues (67). Sixth, although limited in number, a number of studies have suggested that free H₂S in tissues is under 50 nM (18, 37) or undetectable (73). Some of the potential sources for error in H₂S chemistry and blood/tissue measurements are summarized in Fig. 8.

FIG. 7.

Pathways of H₂S biosynthesis (see text for details). DHLA, dihydrolipoic acid; CAT, cysteine aminotransferase; CBS, cystathionine β-synthase; CSE, cystathionine γ-ligase; ST, sulfur transferase; TR, thiosulfate reductase; Trx, thioredoxin; 3-MST, 3-mercaptopyruvate sulfur transferase.

FIG. 8.

Potential sources for error in H₂S chemistry and blood/tissue measurements. (1) Solvation of sulfide salts consumes one, or as shown here, two protons thereby alkalinizing the medium. (2) Analytical procedures that require strong acid (pH<5), strong base (pH>10) or reducing agents, such as dithiothreitol (DTT) can liberate H₂S from a variety of organic and inorganic sulfides (X-S). (3) Equilibrium between H₂S ↔ HS⁻ is sensitive to pH and the pK_a1 for this equilibrium is temperature sensitive. (4) H₂S is readily oxidized in the presence of metal catalysts such as ferric iron. (5) H₂S is rapidly removed from plasma by red blood cells (RBCs) or tissue. (6) H₂S is rapidly lost through volatilization when experiments are aerated or across the lungs during respiration.

Tissue H₂S metabolism

Tissue H₂S oxidation also argues against high blood and tissue H₂S concentrations. Essentially all measurements of H₂S production using the methylene blue, MBB, headspace gas methods, or S²⁻ ISEs have been performed in closed containers under anoxic or very hypoxic conditions. A number of studies have noted that H₂S production is increased or enhanced under hypoxic or anoxic conditions (14, 18) or that oxygen (O₂) consumption is increased by H₂S (21, 35). In fact, when measuring H₂S in headspace gas, Furne et al. (21) not only observed very low concentrations of H₂S in tissues when measured under hypoxic conditions, but they also observed that there was net consumption of H₂S by tissues in the presence of O₂. To my knowledge only one study has simultaneously measured H₂S and O₂ concentrations in tissues in real-time, and this study (57) demonstrated that H₂S is consumed by tissues in the presence of oxygen and H₂S production is only observed under anoxic or severely hypoxic conditions (Fig. 6A). In fact, even exogenous H₂S is quickly and efficiently consumed by tissues at oxygen partial pressures (Po₂) >10 mm Hg (Fig. 6B) (57). By comparison, water or tissue samples in equilibrium with room air typically have a Po₂ >140 mm Hg. Furthermore, tissue hypoxia is expected to increase the reducing conditions in the cell; this could liberate H₂S from persulfides (27), or the acidosis that routinely accompanies hypoxia could release acid-labile sulfur.

H₂S in tissue and H₂S infusion, a quantitative approach

It is instructive to examine concentration in tissues and relative infusion rates as both a self-check of reliability and as an indicator of tissue metabolic capacity. Several constants and/or assumptions are useful: 1 mol of H₂S is equivalent to 32 g/L of sulfur, cells are typically 10%–20% protein and 70% water, plasma volume of most mammals is ∼4% of body weight (bwt), extracellular fluid volume is 20% of bwt, and total body water is 60% of bwt. A few simple calculations allow the reader (or investigator) to determine the feasibility (and credibility) of reported values.

H₂S (or sulfide) is often measured in tissue or cells in culture and expressed as moles or weight of H₂S or sulfide per gram protein. Assuming cells are 15% protein, 70% water (cytosol), and 1 g of water equals 1 mL, then every gram of cell protein is equivalent to 4.7 mL of water (70/15=4.66). Therefore, 1 μmol H₂S/g protein=1 μmol/4.7 g water. Normalizing to 1 L of cytosol, 1 μmol H₂S/g protein=214 μM (6.85 mg/L sulfur) in cytosol. When H₂S is reported in micromoles per milligram of protein, 1 μmol of H₂S/mg protein=214 mM H₂S in cytosol. Tissue H₂S concentrations based on a protein range from 6 nmol/L cytosol in liver (0.03 μmol/kg protein) (37) to tens of millimoles per liter of cytosol (36). Looking at this from a different perspective, if 40 μmol/L H₂S is toxic to cells, this is equivalent to 0.19 μmol/g protein (0.19 nmol/mg protein) or 6.1 μg sulfur/g protein (6.1 ng sulfur/mg protein). Lee et al. (36) measured H₂S in cultured U118 cells exposed to a single treatment of 10 μM H₂S and 12 hours later measured 71 μmol/g protein H₂S (Fig. 3). Based on the preceding calculations, this is equivalent to 15,123 μM H₂S in the cytosol. Space does not permit compilation of other values from the literature and the reader is encouraged to evaluate these with the above calculations in mind.

H₂S is also injected as a single bolus or infused over extended periods and the dose is expressed per kilogram of body weight. If 1 μmol/kg bwt H₂S was injected, the H₂S concentration would be 25 μmol/L if it was retained in the plasma, 5 μmol if retained in the extracellular compartment, and 1.7 μmol/L if distributed throughout body water. These values would increase to 1500, 300, and 102 μmol/L if the H₂S was infused for 1 hour and retained within the respective compartment. Bolus injections are typically administered in as small a volume as practicable to minimize volume disturbances and the concentration of H₂S is often quite high; e.g., Wintner et al. (77) injected 70 mM Na₂S in 10 seconds. These concentrations are undoubtedly toxic until the H₂S is diluted and metabolized.

H₂S as a signaling molecule

The development of mutant mice lacking CSE (CSE^−/−) and subsequent observations that plasma H₂S concentration and aortic and heart H₂S production in these mice was less than that of the wild-type mice and that the CSE^−/− mutant developed hypertension more frequently compared to the wild type led the authors to conclude that “These findings provide direct evidence H₂S that is a physiologic vasodilator and regulator of blood pressure” (78). However, this conclusion seems a bit premature. First, as already described, plasma H₂S concentrations measured in this study were unrealistically high (40, 32, and 18 μmol/L for wild type, CSE^−/+, and CSE^−/−, respectively). Second, Ishii et al. (28) also developed CSE^−/− mutant mice that exhibited hypercystathioninemia and hyperhomocysteinemia but these mice remained normotensive. Third, the only physiologically relevant stimulus that has thus far been demonstrated to affect H₂S production or concentration in tissues is the inverse correlation between H₂S and Po₂ described above and this mechanism hardly seems applicable in the study by Yang et al. (78).

The inverse correlation between Po₂ and H₂S, the similarity of tissue responses to H₂S and hypoxia, and the ability of inhibitors of H₂S synthesis to inhibit hypoxic responses and of sulfur donors to augment it, have led to the hypothesis that H₂S production/metabolism functions as an oxygen sensor in vascular and nonvascular smooth muscle and in chemoreceptor cells (reviewed by Olson and Whitfield [56]). Recent studies of the mammalian carotid body (40, 58, 69, 70) have provided additional support for this mechanism. However, this too has been called into question. Haouzi et al. (23) injected rats with sodium nitrite to produce methemoglobin in blood, which is a well-known scavenger of H₂S. They observed that while intravenous injection of H₂S stimulated ventilation in control rats it did not stimulate ventilation in nitrite-treated rats, thus confirming the H₂S-scavenging effect of methemoglobin They also observed that hypoxic hyperventilation was unaffected by nitrite injection and concluded that H₂S was not involved in H₂S sensing by the carotid body. Key to this conclusion is the ability of methemoglobin to create a sink for H₂S in blood that would prevent buildup of H₂S in the glomus cells of the carotid body. Olson (54) argued against this reasoning based on observations that plasma H₂S is already so low that methemoglobin would not be expected to significantly change the gradient of H₂S between the carotid body and blood, and therefore methemoglobin would not be expected to affect intracellular H₂S signaling. Obviously, this is another conundrum of H₂S biology remaining to be resolved.

It is generally assumed that the metabolic fate of H₂S is mitochondrial oxidation, first to thiosulfate and then sulfate. Thus, once formed, H₂S proceeds inexorably to its fate. However, a recent study by Mikami et al. (46) suggested that this might not be always the case. They found that thiosulfate formed from H₂S oxidation in the mitochondria could be reduced by mitochondrial 3-MST or rhodanase in the presence of physiological levels of dihydrolipoic acid (the reduced form of lipoic acid), thereby again forming H₂S. This is an intriguing possibility because it allows recycling of H₂S without consuming additional cysteine or other sulfur-donating biomolecules. This may have special significance in oxygen sensing in which the proposed increase in H₂S as a result of decreased mitochondrial oxidation (56) may be augmented by H₂S production from preformed thiosulfate, especially since the latter would also be expected to be increased by the hypoxia-driven increased reducing conditions in the cell.

Abbreviations Used

CBS

cystathionine β-synthase

CSE

cystathionine γ-lyase

CSE^−/−

mice lacking cystathionine γ-lyase

DTPA

diethylenetriaminepentaacetic acid

GYY4137

morpholin-4-ium 4 methoxyphenyl(morpholino) phosphinodithioate

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

HPLC

high pressure liquid chromatography

H₂S

hydrogen sulfide

IK-1001

sterile Na₂S solution

ISE

ion-selective electrodes

k

rate constant

MBB

monobromobimane

3-MST

3-mercaptopyruvate sulfur transferase

NO

nitric oxide

Po₂

oxygen partial pressure

ppb

parts per billion

ppm

parts per million

ROS

reactive oxygen species

RSnH

persulfide

RS-S

sulfane sulfur

t_½

half time

Acknowledgments

The author's work has been supported by National Science Foundation Grants IBN 0235223, IOS 0641436, and IOS 1051627.