Bench-to-bedside review: Hydrogen sulfide – the third gaseous transmitter: applications for critical care

Hydrogen sulfide (H2S), a gas with the characteristic odor of rotten eggs, is known for its toxicity and as an environmental hazard, inhibition of mitochondrial respiration resulting from blockade of cytochrome c oxidase being the main toxic mechanism. Recently, however, H2S has been recognized as a signaling molecule of the cardiovascular, inflammatory and nervous systems, and therefore, alongside nitric oxide and carbon monoxide, is referred to as the third endogenous gaseous transmitter. Inhalation of gaseous H2S as well as administration of inhibitors of its endogenous production and compounds that donate H2S have been studied in various models of shock. Based on the concept that multiorgan failure secondary to shock, inflammation and sepsis may represent an adaptive hypometabolic reponse to preserve ATP homoeostasis, particular interest has focused on the induction of a hibernation-like suspended animation with H2S. It must be underscored that currently only a limited number of data are available from clinically relevant large animal models. Moreover, several crucial issues warrant further investigation before the clinical application of this concept. First, the impact of hypothermia for any H2S-related organ protection remains a matter of debate. Second, similar to the friend and foe character of nitric oxide, no definitive conclusions can be made as to whether H2S exerts proinflammatory or anti-inflammatory properties. Finally, in addition to the question of dosing and timing (for example, bolus administration versus continuous intravenous infusion), the preferred route of H2S administration remains to be settled – that is, inhaling gaseous H2S versus intra-venous administration of injectable H2S preparations or H2S donors. To date, therefore, while H2S-induced suspended animation in humans may still be referred to as science fiction, there is ample promising preclinical data that this approach is a fascinating new therapeutic perspective for the management of shock states that merits further investigation.


Introduction
Hydrogen sulfide (H 2 S), a colorless, flammable and watersoluble gas with the characteristic odor of rotten eggs, has been known for decades because of its toxicity and as an environmental hazard [1,2]. Inhibition of mitochondrial respiration -more potent than that of cyanide [3] -resulting from blockade of cytochrome c oxidase is the main mechanism of H 2 S toxicity [4,5]. During recent years, however, H 2 S has been recognized as an important signaling molecule of the cardiovascular system, the inflammatory system and the nervous system. Alongside nitric oxide (NO) and carbon monoxide, therefore, H 2 S is now known as the third endogenous gaseotransmitter [1,6].
cystathionine-γ-lyase, both using pyridoxal 5′-phosphate (vitamin B 6 ) as a cofactor [30][31][32]. This synthesis results in low micromolar H 2 S levels in the extracellular space, which can be rapidly consumed and degraded by various tissues. Similarly to NO and carbon monoxide, H 2 S is a lipophilic compound that easily permeates cell membranes without using specific transporters. Via direct inhibition, NO as well as carbon monoxide are involved in the regulation of cystathionine-β-synthase, but not cystathionine-γ-lyase, which can be activated by lipopolysaccharide (LPS) [1,6].
There are three known pathways of H 2 S degradation: mitochondrial oxidation to thiosulfate, which is further converted to sulfite and sulfate; cytosolic methylation to dimethylsulfide; and sulfhemoglobin formation after binding to hemoglobin [6]. Similar to NO and carbon monoxide, H 2 S can also bind to hemoglobin -which was therefore termed the common sink for the three gaseous transmitters [33]. Consequently, saturation with one of these gases might lead to enhanced plasma concentrations and, subsequently, to biological effects of the other gases [1]. Table 1 summarizes the physicochemistry of H 2 S in mammalian tissues.

Mechanisms of H 2 S
H 2 S exerts its effects in biological systems through a variety of interrelated mechanisms (for a review see [1]). Our current knowledge of the biology of H 2 S predominantly stems from in vitro studies in various cell and isolated organ systems, either using cystathionine-γ-lyase inhibitors such as D,L-propargylglycine (PAG) and β-cyanoalanine, or administration of H 2 S gas or H 2 S donors such as sodium disulfide (Na 2 S) and sodium hydrogen sulfide (NaHS). While high (high micromolar to millimolar) levels are invariably accompanied with cytotoxic effects [34] -which result from free radical generation, glutathione deletion, intracellular iron release and proapoptotic action through both the death receptor and mitochondrial pathways [35] -lower (low micromolar) levels have been shown to exert either cytoprotective (antinecrotic or antiapoptotic) effects [10][11][12][13]36] or proapoptotic proper-ties [37-39], depending on the cell type and on the experimental conditions. Cytochrome c oxidase, a component of the oxidative phosphorylation machinery within the mitochondrium, is one intracellular target of H 2 S [4,5]. Both the toxic effects of H 2 S as well as the induction of a so-called "suspended animation" [40,41] are referred to in this inhibition of mitochondrial respiration [42,43], and thus may represent a possible mechanism for the regulation of cellular oxygen consumption [44].
Activation of potassium-dependent ATP channels is another major mechanism of H 2 S, which in turn causes vasodilation, preconditioning against ischemia/reperfusion injury and myocardial protection [45]. Various findings support this concept [1,6,46]: potassium-dependent ATP channel blockers (sulfonylurea derivates -for example, glibenclamide) attenuated the H 2 S-induced vasodilation both in vivo and in vitro [47,48], and stimulation of potassium-dependent ATP channels was demonstrated in the myocardium, pancreatic β cells, neurons and the carotid sinus [6]. Moreover, glibenclamide reversed the otherwise marked Na 2 S-related increase of the hepatic arterial buffer response capacity that counteracts reduction of portal venous flow, whereas PAG decreased this compensatory mechanism [49].
An endothelium-dependent effect seems to contribute to these vasodilatory properties: in human endothelial cells, H 2 S caused direct inhibition of the angiotensin-converting enzyme [50], and, finally, H 2 S can enhance the vasorelaxation induced by NO [51,52]. The interaction between H 2 S and NO with respect to vascular actions is, however, fairly complex: low H 2 S concentrations may cause vasoconstriction as a result of an attenuated vasorelaxant effect of NO due to scavenging of endothelial NO and formation of an inactive nitrosothiol [52-54]. The local oxygen concentration apparently assumes importance for the vasomotor properties of H 2 S as well [55]: while H 2 S had vasodilator properties at 40 μM oxygen concentration (that is, an oxygen partial . While the influence of H 2 S on extracellular signal-regulated kinase seems to be rather comprehensible [25], studies exploring the effect on downstream pathways result in conflictive statements. Jeong and colleagues reported that H 2 S enhances NO production and inducible NO synthase expression by potentiating IL-1β-induced NF-κB in vascular smooth muscle cells [63], which is consistent with the H 2 S-induced NF-κB activation and subsequent proinflammatory cytokine production in IFNγ-primed monocytes [65]. Nevertheless, any H 2 S effect on NF-κB and its transcription-regulated mediators (for example, inducible NO synthase, cytokines and apoptotic factors) may be cell-type dependent and stimulus dependent. In fact, in addition to the above-mentioned decreased NF-κB activation and inducible NO synthase expression in LPSstimulated macrophages [61], H 2 S administration also attenuated inducible NO synthase expression, NO production, as well as TNFα secretion in microglia exposed to LPS [20].
In the context of these contradictory findings, the doses of the H 2 S donors administered may assume particular importance. Even the physiologically relevant concentrations [36,64] might have to be reconsidered due to overestimation of basal H 2 S levels: murine plasma sulfide levels are reported between 10 and 34 μM [21,22], and are increased up to 20 to 65 μM after endotoxin injection [21] or cecal ligation and puncture [22]. A reduction of plasma sulfide concentration from 50 μM to ~25 μM, finally, was reported in patients with coronary heart disease [1], whereas plasma sulfide levels increased from 44 to 150 μM in patients with sepsis [21]. It should be noted, however, that the distinct techniques used by various groups to determine sulfide levels may account for the marked variability in the baseline values reported.

Metabolic effects of H 2 S: induction of suspended animation
Suspended animation is a hibernation-like metabolic status characterized by a marked yet reversible reduction of energy expenditure, which allows nonhibernating species to sustain environmental stress, such as extreme changes in temperature or oxygen deprivation [41,68].
In landmark work, the Roth's group provided evidence that inhaled H 2 S can induce such a suspended animation [40,41]: in awake mice, breathing 80 ppm H 2 S caused a dose-dependent reduction of both the respiratory rate and the heart rate as well as of oxygen uptake and carbon dioxide production, which was ultimately associated with a drop in body core temperature to levels ~2°C above ambient temperature [40]. All these effects were completely reversible after H 2 S washout, and thereafter animals presented with a totally normal behavior. A follow-up study confirmed these observations, and the authors demonstrated using telemetry and echocardiography that the bradycardia-related fall in cardiac output coincided with an unchanged stroke volume and blood pressure. These physiologic effects of inhaled H 2 S were present regardless of the body core temperature investigated (27°C and 35°C) [69].
It is noteworthy that anesthesia may at least partially blunt the myocardial effect of inhaled H 2 S. In mechanically ventilated mice instrumented with left ventricular pressure volume conductance catheters and assigned to 100 ppm inhaled H 2 S, we found that hypothermia alone (27°C) but not normothermic H 2 S inhalation (38°C) decreased the cardiac output due to a fall in heart rate, whereas both the stroke volume as well as the parameters of systolic and diastolic function remained unaffected (Table 2) [70]. Interestingly, inhaled H 2 S in combination with hypothermia, however, was concomitant with the least stimulation of oxygen flux induced by addition of cytochrome c during state 3 respiration with combined complex I and complex II substrates (Figure 1) [71]. Since stimulation by cytochrome c should not occur in intact mitochondria, this finding suggests better preservation of mitochondrial integrity under these conditions [72].
In good agreement with the concept that a controlled reduction in cellular energetic expenditure would allow maintenance of ATP homoeostasis [41] and thus of improving outcome during shock states due to preserved mitochondrial function [73,74], the group of Roth and colleagues subsequently demonstrated that pretreatment with inhaled H 2 S (150 ppm) for only 20 minutes markedly prolonged survival without any apparent detrimental effects for mice exposed to otherwise lethal hypoxia (5% oxygen) [75] and for rats undergoing lethal hemorrhage (60% of the calculated blood volume over 40 minutes) [8]. It is noteworthy that in the latter study the protective effect was comparable when using either inhaled H 2 S or a single intravenous bolus of Na 2 S [75]: parenteral sulfide administration has a number of practical advantages (ease of administration, no need for inhalation delivery systems, no risk of exposure to personnel, no issues related of the characteristic odor of H 2 S gas) and, in particular, avoids the pulmonary irritant effects of inhaled H 2 S, which can be apparent even at low inspiratory gaseous concentrations [76]. Finally, it is noteworthy that hypothermia is not a prerequisite of H 2 S-related cytoprotection during hemorrhage: the H 2 S donor NaHS improved hemodynamics, attenuated metabolic acidosis, and reduced oxidative and nitrosative stress in rats subjected to controlled hemorrhage at a mean blood pressure of 40 mmHg (Figure 2) [9].
The clinical relevance of murine models may be questioned because, due to their large surface area/mass ratio, rodents can rapidly drop their core temperature [77]. In fact, other authors failed to confirm the metabolic effect of inhaled H 2 S in anesthetized and mechanically ventilated piglets (body weight ~6 kg) or in H 2 S-sedated and spontaneously breathing sheep (body weight ~74 kg) exposed to up to 80 or 60 ppm H 2 S, respectively [78,79]. These findings may be due to the dosing or timing of H 2 S, and are in contrast to recent data from our own group: in anesthetized and mechanically ventilated swine (body weight ~45 kg) that underwent transient thoracic aortic balloon occlusion, infusing the intravenous H 2 S donor Na 2 S over 10 hours reduced the heart rate and cardiac output without affecting the stroke volume, thereby reducing oxygen uptake and carbon dioxide production and, ultimately, core temperature [16]. The metabolic effect of H 2 S coincided with an attenua-

Figure 1
Cytochrome c-stimulated mitochondrial oxygen flux in livers from anesthetized and mechanically ventilated mice. Ratio of mitochondrial oxygen flux in homogenized livers from anesthetized and mechanically ventilated mice after addition in relation to before addition of cytochrome c. Since stimulation by cytochrome c should not occur in intact mitochondria, the smallest value (that is, a ratio close to 1.00) suggests preservation of mitochondrial integrity. Animals were subjected to inhaled hydrogen sulfide (H 2 S) (100 ppm over 5 hours) or vehicle gas during normothermia (38°C) and hypothermia (27°C) [63]. Data presented as mean ± standard deviation, n = 8 in each group. #P <0.05 versus control, 38°C.
tion of the early reperfusion-related hyperlactatemiasuggesting a reduced need for anaerobic ATP generation during the ischemia period -and an improved noradrenaline responsiveness, indicating both improved heart function and vasomotor response to catecholamine stimulation [16].

H 2 S-induced cytoprotection during ischemia-reperfusion
Deliberate hypothermia is a cornerstone of the standard procedures to facilitate neurological recovery after cardiac arrest and to improve postoperative organ function after cardiac and transplant surgery. Consequently, several authors investigated the therapeutic potential of H 2 S-induced suspended animation after ischemia-reperfusion injury -and H 2 S protected the lung [14], the liver [12], the kidney ( Figure 3) [17,80], and, in particular, the heart [10,11,13,15,18,62,[81][82][83]. H 2 S administered prior to reperfusion therefore limited the infarct size and preserved left ventricular function in mice [10] and in swine [11].
While these findings were obtained without induction of hypothermia, preserved mitochondrial function documented by an increased complex I and complex II efficiency assumed major importance for the H 2 S-induced cytoprotection [10]. The important role of preserved mitochondrial integrity was further underscored by the fact that 5-hydroxydeconoate, which is referred to as a mitochondrial potassium-dependent ATP-channel blocker, abolished the anti-apoptotic effects of H 2 S [18]. Clearly, anti-inflammatory and anti-apoptotic effects also contributed to the improved postischemic myocardial function: treatment with H 2 S was associated with reduced myocardial myeloperoxidase activity and an absence of the increase in the IL-1β levels (that is, attenuated tissue inflammation [10,18]), as well as complete inhibition of thrombininduced leukocyte rolling, a parameter for leukocyte-endothelium interaction [10]. Moreover, the ischemia-reperfusioninduced activation of p38 mitogen-activated protein kinase, of c-Jun N-terminal kinase and of NF-κB was also attenuated by H 2 S [18]. Finally, H 2 S exerted anti-apoptotic effects as shown by reduced TUNEL staining [10,11] and by expression of cleaved caspase-9 [18], caspase-3 [10,11], poly-ADPribose-polymerase [11] and the cell death-inducing protooncogene c-fos [13].

Controversial role of H 2 S in animal models of inflammation
Despite the promising data mentioned above, it is still a matter of debate whether H 2 S is a metabolic mediator or a toxic gas [84] -particularly given the rather controversial findings on the immune function reported in various models of systemic inflammation. In fact, H 2 S exerted both marked proinflammatory effects [19,[21][22][23][24][25]27,85] and anti-inflammatory effects [9,10,18,20,[28][29][30]. Studies using inhibitors of endogenous H 2 S production such as PAG demonstrated pronounced proinflammatory effects of H 2 S: PAG attenuated organ injury, blunted the increase of the proinflammatory cytokine and chemokine levels as well as the myeloperoxidase activity in the lung and liver, and abolished leukocyte activation and trafficking in LPS-induced endotoxemia [19,21] or cecal ligation and puncture-induced sepsis [22][23][24][25]86]. In good agreement with these findings, the H 2 S donor NaHS significantly aggravated this systemic inflammation [21][22][23][24][25]86]. Although similar results were found during caerulin-induced pancreatitis [27,87], the role of H 2 S during systemic inflammatory diseases is still a matter of debate. Zanardo and colleagues reported reduced leukocyte infiltration and edema formation using the air pouch and carrageenan-induced hindpaw edema model in rats injected with the H 2 S donors NaHS and Na 2 S [30]. Moreover, in mice with acute lung injury induced by combined burn and smoke Hydrogen sulfide-related hemodynamic effects in rats subjected to hemorrhage and subsequent retransfusion. Time course of the difference in (a) mean blood pressure (ΔMAP) and (b) carotid blood flow (ΔCBF) in rats subjected to 60 minutes of hemorrhage (MAP 40 mmHg) and subsequent retransfusion of shed blood. Ten minutes prior to retransfusion, animals received vehicle (n = 11; open circles) or the hydrogen sulfide donor sodium hydrogen sulfide (bolus 0.2 mg/kg, n = 11; closed circles) [9]. Data presented as mean (standard deviation). #P <0.05 versus controls.
inhalation, a single Na 2 S bolus decreased tissue IL-1β levels, increased IL-10 levels, and attenuated protein oxidation in the lung, which ultimately resulted in markedly prolonged survival [28].
Variable dosing and timing make it difficult to definitely conclude on the proinflammatory and/or anti-inflammatory effects of H 2 S: while the median sulfide lethal dose in rats has been described to be approximately 3 mg/kg intravenously [1], studies in the literature report on doses ranging from 0.05 to 5 mg/kg. In addition, there are only a small number of reports on continuous intravenous infusion rather than bolus administration. Finally, the role of the suspended animation-related hypothermia per se remains a matter of debate. While some studies report that spontanoues hypothermia and/or control of fever may worsen the outcome [88], other authors describe decreased inflammation [89] and improved survival after inducing hypothermia in sepsis [90].
We found in anesthetized and mechanically ventilated mice undergoing sham operation for surgical instrumentation that normothermic H 2 S (100 ppm) inhalation (38°C) over 5 hours and hypothermia (27°C) alone comparably attenuated the inflammatory chemokine release (monocyte chemotactic protein-1, macrophage inflammatory protein-2 and growthrelated oncogen/keratinocyte-derived chemokine) in the lung tissue. While H 2 S did not affect the tissue concentrations of TNFα, combining hypothermia and inhaled H 2 S significantly decreased tissue IL-6 expression (Table 3) [91].

Conclusions
Based on the concept that multiorgan failure secondary to shock, inflammation and sepsis may actually be an adaptive hypometabolic reponse to preserve ATP homoeostasis [92] such as has been demonstrated for the septic heart [93] and thus represent one of the organism's strategies to survive under stress conditions, the interest of inducing a hibernation-like suspended animation with H 2 S is obvious. Investigations have currently progressed most for the treatment of myocardial ischemia [94]. It must be underscored, however, that only a relatively small proportion of the published studies was conducted in clinically relevant large animal models [11,16,95], and, furthermore, that the findings reported are controversial [16,78,79].
Moreover, several crucial issues warrant further investigation before the clinical application of this concept. First, the role of hypothermia for any suspended animation-related organ protection is well established [96], but its impact remains a matter of debate for H 2 S-related organ protection. Clearly, in the rodent studies [10,12,18,28], any cytoprotective effect Hydrogen sulfide attenuation of oxidative DNA damage in the kidney after organ ischemia-reperfusion. Oxidative DNA damage (tail moment in the alkaline version of the comet assay [89]) in kidney tissue biopsies prior to (left panel) and after 2 hours of organ ischemia and 8 hours of reperfusion (right panel) in control swine (n = 7; open box plots) and in animals treated with the hydrogen sulfide donor sodium disulfide (Na 2 S) (n = 8; grey box plots). Renal ischemia was induced by inflating the balloon of an intra-aortic catheter positioned at the renal artery orifices. Na 2 S infusion was infused before kidney ischemia (2 mg/kg/hour over 2 hours) as well as during the first 4 hours of reperfusion (1 mg/kg/hour) [72]. Data presented as median (quartiles, range). #P <0.05 versus before ischemia, §P <0.05 versus control. Lung tissue concentrations of monocyte chemotactic protein-1 (MCP-1), macrophage-inflammatory protein-2 (MIP-2), growth-related oncogen/keratinocyte-derived chemokine (KC), TNFα, and IL-6 after inhaling hydrogen sulfide (H 2 S) (100 ppm over 5 hours) during normothermia (38°C) or hypothermia (27°C) [83]. Data presented as median (range), n = 5 in each group. *P <0.05 versus control, 38°C.
was apparent without a change in core body temperature, but localized metabolic effects cannot be excluded [10]. In addition, the role of any H 2 S-related hypothermia remains controversial in the context of systemic inflammation [88]. Second, similar to the friend and foe character of NO, no definitive conclusions can be made as to whether H 2 S exerts proinflammatory or anti-inflammatory properties [1,6,85].
Finally, in addition to the question of dosing and timing (for example, bolus administration versus continuous intravenous infusion), the preferred route of H 2 S administration remains to be settled: while inhaling gaseous H 2 S probably allows easily titrating target blood concentrations, it is well established that this method can also directly cause airway irritation [76].
While H 2 S-induced suspended animation in humans to date may still be referred to as science fiction, there are ample promising preclinical data that this approach is a fascinating new therapeutic perspective for the management of shock states that merits further investigation.

Competing interests
CS is an officer and stockholder of Ikaria (Seattle, WA, USA), a company involved in the commercial development of hydrogen sulfide. PR received research grants from Ikaria. FW, PA and EC declare that they have no competing interests.