H2S (hydrogen sulfide) is a well known and pungent gas recently discovered to be synthesized enzymatically in mammalian and human tissues. In a relatively short period of time, H2S has attracted substantial interest as an endogenous gaseous mediator and potential target for pharmacological manipulation. Studies in animals and humans have shown H2S to be involved in diverse physiological and pathophysiological processes, such as learning and memory, neurodegeneration, regulation of inflammation and blood pressure, and metabolism. However, research is limited by the lack of specific analytical and pharmacological tools which has led to considerable controversy in the literature. Commonly used inhibitors of endogenous H2S synthesis have been well known for decades to interact with other metabolic pathways or even generate NO (nitric oxide). Similarly, commonly used H2S donors release H2S far too quickly to be physiologically relevant, but may have therapeutic applications. In the present review, we discuss the enzymatic synthesis of H2S and its emerging importance as a mediator in physiology and pathology. We also critically discuss the suitability of proposed ‘biomarkers’ of H2S synthesis and metabolism, and highlight the complexities of the currently used pharmacological H2S ‘donor’ molecules and ‘specific’ H2S synthesis inhibitors in their application to studying the role of H2S in human disease.
- amino-oxyacetate (AOAA)
- BCA (β-cyanoalanine)
- cystathionine-γ-lyase (CSE)
- cystathionine-β-synthase (CBS)
- hydrogen sulfide (H2S)
- propargyglycine (PAG)
H2S (hydrogen sulfide/dihydrogen sulfide [1,2]) is a well known and pungent gas with the distinctive smell of rotten eggs. The toxicology of high concentrations of H2S as an environmental pollutant has been studied extensively . Since it was first discovered to be synthesized in mammalian and human tissues, it has attracted considerable interest in a relatively short period of time as an endogenous gaseous mediator and potential pharmacological and therapeutic tool. Studies in animals and humans have shown H2S to be involved in diverse physiological and pathophysiological processes, such as regulation of blood pressure [4,5], inflammation , neurodegenerative diseases  and metabolic disorders, including obesity and diabetes . These findings have been the subject of several explicitly detailed reviews elsewhere. However, the focus of the present review is to highlight that, as with any emerging field in physiology, pharmacology and medicine, research is limited by the tools available, which have generated significant controversy; specifically, the absence of wholly tissue-selective and enzyme-specific inhibitors to target H2S biosynthesis, specific donors which release H2S in a physiological manner and a current lack of suitable ‘biomarkers’ for H2S synthesis and turnover in vivo.
H2S BIOSYNTHESIS: ARE THERE TISSUE-SPECIFIC H2S-SYNTHESIZING ENZYMES?
H2S is a highly lipophilic molecule able to freely penetrate the membranes of cells of all types by diffusion and without the requirement for specialized membrane transporters . In aqueous solution, H2S is weakly acidic (pKa=6.76 at 37 °C) and dissociates to form two dissociation states: HS− (hydrosulfide anion) (pKa=7.04) and S2− (sulfide anion) (pKa=11.96), according to the following sequential reactions: At pH 7.4, approximately 18.5% of the total sulfide exists as the undissociated acid and 81.5% as HS− . It is currently not known whether the biological effects of H2S are mediated directly by H2S itself or by derived species that will also exist at physiological pH, predominantly HS− but also S2−. It is unlikely S2− will play a significant role, since it will only be present at a high pH. As a varied mixture of these species will always exist under physiological conditions irrespective of the source of H2S, it is prudent to use the term ‘H2S’ to encompass the sum of these species present under physiological conditions [11,12].
In mammalian and human tissues, the bulk of endogenous H2S synthesis appears to be from the PLP (pyridoxal-5′-phosphate)-dependent enzymes CSE (cystathionine-γ-lyase; EC 220.127.116.11) and CBS (cystathionine-β-synthase; EC 18.104.22.168) via the amino acids cysteine, homocysteine and cystathionine (summarized in Table 1). A PLP-independent pathway has been proposed in neuronal tissue and rodent large vessel vascular endothelial cells utilizing the desulfuration of 3-mercaptopyruvate by MPST (3-mercaptopyruvate sulfurtransferase; EC 22.214.171.124) with CAT (cysteine aminotransferase; EC 126.96.36.199) [13–15]. However, at present, the importance of this third pathway is not as well characterized as CSE and CBS, and its role in determining H2S synthesis in human tissues is not known.
It been widely suggested that the expression of CSE, CBS, MPST (and CAT) in rodents and humans showed a marked degree of tissue specificity. However, as more researchers investigate the emerging role of H2S in their particular system, this simple and convenient distinction is no longer as clear as once thought. Up until recently, the current literature consensus was that CBS was the predominant source of H2S in the brain and nervous tissue, highly concentrated in cerebellar Purkinje and hippocampal neurons, and that in the vasculature (e.g. smooth muscle and endothelium) CSE was the major source of H2S. However, it has been known for some time that neuronal tissue clearly contains CSE and vascular tissue contains CSE and CBS, and the previous assumption on distinct tissue distribution is not clear. Furthermore, CBS has been shown to be preferentially expressed in radial glia/astrocytes of adult and developing mouse brain, but is not present in neurons , and at least the rodent macrovasculature has an additional pathway for synthesizing H2S via MPST/CAT (summarized in Table 1).
In humans, the gene for CSE is located on chromosome 1 (1p31.1). CSE is a member of the γ-family of PLP-dependent enzymes and is a 405-amino-acid protein consisting of a tetramer formed by two homodimers. The crystal structure of human CSE in the apo form and in complex with PLP has been determined recently . CSE catalyses the α,γ-carbon elimination of cystathionine to produce cysteine, α-oxobutyrate and ammonia (Table 2). Additional cysteine-dependent β- and homocysteine-dependent γ-reactions have been suggested  to generate H2S (Table 2). Additionally CSE may catalyse the β-elimination of cystine (cysteine disulfide) via the formation of thiocysteine, which then decomposes non-enzymatically to H2S . However, in the presence of physiological concentrations of cysteine (~100 μM), homocysteine (10 μM) or cystathionine (5 μM), although the catalytic-centre activity number for CSE-mediated cystathionine cleavage was 5-fold greater than for cysteine and 12-fold higher than for homocysteine, the CSE-catalysed α,β-elimination of cysteine was the predominant source of H2S accounting for ~70% of the H2S produced, whereas the α,γ-elimination of homocysteine accounted for ~29% of the measured H2S .
At least two CSE mRNA splice variants have been demonstrated [19,20] producing long (CSE-l) and truncated (CSE-s) CSE proteins. Although CSE-s has been suggested to be inactive , the experiments to determine this only measured cysteine accumulation from added cystathionine as an index of enzymatic activity but did not examine H2S production. As cysteine will also serve as a preferential substrate for CSE (Table 2), the lack of accumulation of cysteine from CSE-s suggests it was probably consumed in the generation H2S. However, the precise role of these splice variants in regulating CSE activity in terms of H2S production requires further attention and may offer therapeutic potential for controlling endogenous H2S synthesis.
The human CBS gene is located on chromosome 21 (21q22.3) . CBS is a homotetramer consisting of 551-amino-acid subunits which bind two co-factors (haem and PLP) and two substrates (homocysteine and serine) (Table 2). To generate H2S, human CBS can use either cysteine, forming lanthionine (see below), or cysteine plus homocysteine, forming cystathionine. This latter reaction is predicted to predominate under physiological conditions and account for ~96% of the total H2S generated from CBS, whereas the reactions in the absence of homocysteine represent up to 2.6% of the total H2S . Although the CBS gene encodes several mRNAs , the functional product of these mRNA isoforms in terms of H2S synthesis have yet to be examined. The haem component of CBS is reported to function as a cellular redox sensor , which could increase H2S generation in response to oxidative and/or nitrosative stress-mediated cellular injury (see below).
In contrast with CSE and CBS, very little information is currently available with regards to human MPST and H2S synthesis. MPST is a ~33 kDa monomeric or disulfide-linked dimeric protein containing two rhodanese domains. It is located in the cytoplasm and mitochondria, and the human MPST gene is located on chromosome 22 (22q12.3). At least two splice variants of human MPST are present, but as with CSE and CBS, their regulation and role in H2S synthesis are not understood. Although originally described in 1954 , the formation of H2S from MPST has gained renewed interest and has so far been confined to rodent macrovascular endothelium  and brain homogenates , where H2S is generated through 3-mercaptopyruvate, α-oxoglutarate and cysteine (Table 2). Unravelling the relative contribution of MPST to tissue H2S production will be complex since there are currently no available MPST inhibitors. The inhibitors that have been used in isolated cell or tissue homogenates would not exhibit any degree of specificity to be useful in cell culture or in vivo (Table 2).
A fourth potential pathway for endogenous H2S synthesis has been proposed , via the enzyme cysteine lyase (EC 188.8.131.52) and utilizing cysteine with CH3S (methanethiol) or SO32− (sulfite) as substrates in the presence of PLP (Table 2). However work on this enzyme has exclusively focused on fish, amphibians  and birds [28–30]. It is not certain whether this enzyme is present in or is capable of synthesizing H2S in human (or mammalian) cells and, as a PLP-dependent enzyme, its activity is also likely to be inhibited by commonly used inhibitors of CSE and CBS (see below).
TISSUE AND BLOOD LEVELS OF H2S
Tissue production and levels of ‘H2S’ in blood has been the subject of much controversy (reviewed in ). The predominant method employed to evaluate serum or plasma or tissue levels and synthesis of H2S in humans and in animal models of human disease has used a spectrophotometric approach based around Methylene Blue. In this procedure, H2S/HS− and/or aqueous sulfide in biological samples is ‘fixed’ or ‘trapped’ with zinc to prevent loss of H2S through volatilization and aerial oxidation, resulting in the formation of stable ZnS (zinc sulfide) . H2S is then released from ZnS under strongly acidic conditions and, in the presence of DMPD (N,N-dimethyl-p-phenylenediamine) and Fe3+, results in the formation of the heterocyclic thiazine dye Methylene Blue, which is then either measured by spectrophotometry or HPLC. Under these conditions, plasma or serum H2S levels in healthy human adults have been reported to be in the range of 20 to 60 μM , although considerably higher levels have been reported using this method, for example >100 μM  and in rodent plasma in excess of 300 μM . Additional techniques applied to serum or plasma have similarly reported between 20 and 60 μM ‘H2S’ (or derived species), such as sulfide-selective electrodes [31,34,35], microdistillation and ion chromatography , and GC/ion conductance . However, others have provided evidence to suggest that actual levels of ‘free’ H2S in plasma were beyond or on the limit of detection for the Methylene Blue assay and electrochemical detectors [38,39]. More recently fluorimetric-based methods have been developed employing monobrombimane to trap ‘free’ H2S and the resulting dibimane determined by reverse-phase HPLC to show baseline levels of free ‘H2S’ to be in the region of 0.4–0.9 μM . It is therefore likely that other methods measure the total sum of H2S-derived species such as HS− and S2− and possibly other physiological H2S ‘carrier’ molecules that exist at physiological pH and which release H2S under acidic conditions employed in the analytical processes, rather than ‘free’ H2S itself and care should be taken to describe the results as such .
To date it is not clear what role diet (e.g. cysteine intake) and exercise or past and present medication play in determining tissue and blood levels of H2S (and its derived species under physiological conditions) and ‘H2S’-derived metabolites. The influence of these factors and the lack of standardization between laboratories could account for at least some of this biological variation. At present, the ‘absolute’ level of ‘free’ or ‘bound’ H2S in tissues or in blood is uncertain, but if the field follows a similar progression to that of NO then levels of H2S, the identification and suitability of ‘biomarkers’ and/or specific metabolites are likely to present controversy for some time. Nevertheless, using the above approaches in combination with inhibitors of H2S synthesis (see below) have revealed an emerging role for H2S/HS− in human disease processes and physiology (see below).
H2S METABOLISM: IS THERE A SPECIFIC ‘BIOMARKER’ FOR H2S SYNTHESIS OR ‘TURNOVER’ IN VIVO?
Whatever the precise level of systemic and tissue ‘H2S’, measured levels may represent an underestimate of the true extent of H2S synthesis as pathways for H2S removal exist, although it is not clear how rapid these processes occur or how H2S removal is affected by disease. In addition to the controversy over the absolute levels of ‘H2S’ synthesized in tissues and present in blood, the role of H2S in human physiology and pathology is hampered further by a lack of known specific ‘biomarkers’ or end products of H2S metabolism, which are crucial for identifying physiological and pathophysiological processes regulated by H2S (summarized in Table 3).
Intracellular H2S is apparently rapidly oxidized to S2O32− (thiosulfate) by mitochondria with the subsequent conversion into SO32− and SO42− (sulfate) [41,42]. SO32− and SO42− are also produced upon oxidation of H2S by activated neutrophils, where SO32− induced the respiratory burst leading to further H2S oxidation and loss  by several endogenous oxidant species elevated during disease processes, such as NO [44,45], superoxide , ClO− (hypochlorite) , H2O2 (hydrogen peroxide)  and ONOO− (peroxynitrite) . Furthermore, SO32− readily undergoes hepatic metabolism forming SO42− .
SO42− has also been proposed as an index of endogenous H2S, but while stable and reasonably uncomplicated to assay, it is not only formed from H2S oxidation, but can be derived from the direct oxidation of cysteine by cysteine dioxygenase (EC 184.108.40.206), as well as from the oxidation of SO32− by sulfite oxidase (EC 220.127.116.11) , limiting its usefulness as a ‘biomarker’ of H2S synthesis or metabolism. Similarly, S2O32− measurement in urine and blood has historically been used as an index of environmental H2S exposure and inhalation [50,51], but its use also requires careful scrutiny. Elevated levels of S2O32− in urine is observed in patients with DS (Down's syndrome)  and, although blood levels of S2O32− or ‘H2S’ have not yet been assessed, S2O32− has been used to suggest that there is an overproduction of H2S in DS [52,53]. The location of CBS to chromosome 21 supports this possibility. However, as with SO42−, it is unlikely that urinary S2O32− is reliable as an index of H2S as S2O32− levels in blood without detection in urine have also been observed after industrial H2S poisoning [50,51]. In addition, S2O32− ingestion has been proposed to induce a cardioprotective effect by increasing endogenous H2S synthesis in a murine model of chronic heart failure , suggesting it is inappropriate as a ‘biomarker’ of H2S. Furthermore, urinary S2O3− levels in healthy individuals vary greatly and are highly sensitive to small atmospheric fluctuations in H2S . In animal studies where SO32−, SO42− and S2O32− have been determined simultaneously, H2S gas exposure resulted in higher lung levels of SO32− and SO42− rather than of S2O32− . Similarly, incubation of liver, colonic or muscle tissue or plasma with either H2S or CH3S also leads to significant S2O32− and SO42− generation . As such, the metabolism of H2S in vivo is highly complex and individually the reliability of S2O32−, SO42− or SO32− as specific indices of endogenous H2S synthesis and turnover require further attention.
Additional cellular processes for H2S removal also exist, suggesting the possibility for additional specific ‘biomarkers’ or molecular ‘fingerprints’; however, their use may also not be straightforward. H2S is rapidly methylated to CH3S and CH3SCH3 (dimethyl sulfide) by thiol-S-methyltransferase (EC 18.104.22.168) [41,42], although CBS can also catalyse H2S production by β-replacement of cysteine with CH3S . An additional enzymatic removal process, which is potentially important in colonic tissue and erythrocytes, involves rhodanese (thiosulfate:cyanide sulfurtransferase; EC 22.214.171.124), . Rhodanese catalyses the transfer of HS− to a thiophilic acceptor (such as cyanide) to form SCN− (thiocyanate) and SO42−.
A more promising approach to specific ‘biomarkers’ of H2S synthesis and turnover in health and disease may come from the detailed analysis of enzyme-specific metabolic products. The novel amino acids lanthionine and homolanthionine were recently shown to be formed from the CSE-catalysed condensation reactions of cysteine and homocysteine respectively [18,22]. Lanthionine is also formed during H2S synthesis by CSE and CBS from cysteine alone or cysteine plus serine [17,21]. However, it may also be produced without H2S formation from cysteine via cysteine lyase . CSE-catalysed homolanthionine from homocysteine and homoserine has been demonstrated in rat and human liver , and CSE-mediated H2S production in rat liver is also induced by bacterial endotoxin  and the diabetogenic agent streptozotocin , suggesting lanthionine may represent a viable ‘biomarker’ of H2S synthesis. Lanthionine is metabolized with α-oxo amino acids via glutamine transaminase K/pyruvate kyneurenin aminotransferase to cyclic lanthionine ketimine. This ketimine derivative is present in mammalian brains, including the human cerebral cortex , where it may protect neuronal cells from oxidative stress  and in human urine [62,63]. However, lanthionine ketimine may be biologically active in its own right , since metabolic processes for its removal exist such as ketimine reductases  forming 1,4-thiomorfoline-3,5-dicarboxylic acid, although the extent to which this occurs is not clear. It is therefore likely that the use of lanthionine as an index of CSE/CBS activity and H2S production will be highly complex.
In sharp contrast, very little is known about the metabolic fate of homolanthionine, but while lanthionine forms the six-membered ring lanthionine ketimine, homolanthionine is unlikely to form the corresponding eight-membered ring homologue. Homolanthionine has been shown to be produced in the human liver by CSE  and it is present in urine from patients with homocystinuria [65,66]. Interestingly, homolanthionine production was shown to be increased in a human liver biopsy after 3 months of treatment with pyridoxine HCl, and homolanthionine synthesis in rat liver homogenates was dependent on PLP and cystine . However, rats fed 35S-labelled homolanthionine on an 8% casein diet showed [35S]cystine in hair, suggesting homolanthionine is metabolically processed, at least in rodents . However, the yield of [35S]cystine was low (1.5%) and the accumulation or metabolism of [35S]cystine was not assessed in other tissues. Nevertheless, homolanthionine may offer significant promise for the assessment of overall CSE/CBS activity and H2S synthesis in vivo.
The interaction of H2S with haemoglobin, forming sulfhaemoglobin, is well known  and methaemoglobin has been used as a ‘scavenger’ of H2S in ex vivo and in vitro experiments . Recently, Perna et al.  showed significantly lower erythrocyte levels of sulfhaemoglobin in patients with end-stage renal failure and sulfhaemoglobin levels correlated with lower plasma H2S levels, suggesting sulfhaemoglobin was a marker of endogenous H2S synthesis. However, there are many additional mechanisms for the formation of sulfhaemoglobin in vivo, such as exposure to the atmospheric pollutant SO2 (sulfur dioxide) , xenobiotic metabolism (for example, phenacetin , acetanilide , metoclopramide , phenazopyridine , dapsone , metoclopramide [76,77], N-acetylcysteine  and neomycin , sulfanilamide  and sulfanilamide-containing drugs, such as sumatriptan, albeit after ingestion of large quantities ), drug overdose (cimetidine, paracetamol, ibuprofen and naproxen)  or ingestion of toxic substances (for example paint  or shoe dye ). As such, any clinical study which proposes to use sulfhaemoglobin as a specific index of H2S biosynthesis or turnover would require ruling out confounding exogenous mediators of sulfhaemoglobin formation, such as current and past drug therapies or environmental exposure to SO2.
An additional and recently proposed hypothesis is that H2S and/or species derived from it under physiological conditions further interact with protein thiols via the covalent modification of cysteine (S-sulfhydration) in which an -SH group is transferred to a cysteine-SH residue in a protein, yielding perthiol (-SSH) moieties . Although this hypothesis may explain the signalling mechanisms of intracellular and extracellularly generated H2S, at least in part, it is not clear to what extent perthiol formation accounts for the levels of ‘H2S’ observed in blood and tissues. Detailed discussion on the cellular and molecular signalling pathways induced by H2S and sulfide salt H2S donor compounds are expertly reviewed elsewhere [85,86]. However, since perthiol formation would enhance the chemical reactivity of H2S, these species may be too short-lived to represent viable and H2S-specific ‘biomarkers’ or ‘fingerprints’ of H2S production in vivo. Owing to this high reactivity, they have previously been proposed as endogenous antioxidant molecules . Furthermore, perthiols may be formed in vivo through H2S-independent pathways, such as the general metabolism of thiols (for example glutathione), commensal bacteria  or dietary sources including garlic and other Allium species [89,90].
WHAT IS THE ROLE OF ENDOGENOUS H2S IN HUMAN HEALTH AND DISEASE?
Recent animal models and human clinical studies have shown perturbed synthesis of H2S in a variety of physiological processes and pathologies, and have highlighted the potential for modulating H2S synthesis for therapeutic exploitation. These have invariably relied upon: (i) the measurement of ‘H2S’ in body fluids and the production of H2S by isolated tissues and tissue homogenates; (ii) the use of sulfide salt H2S ‘donors’; and (iii) inhibitors of CSE and CBS activity. The complex physiological problems associated with (ii) and (iii) are discussed in detail further below. With the exception of a few cases to be discussed below, the large majority of studies have only been limited to animal models and the human studies have generally been limited to small patient and volunteer sample sizes. Examples of perturbed synthesis in these few human studies and animal models of disease are summarized in Table 4. For this section, we will focus on studies where both animal models and human clinical studies have been investigated.
H2S and the regulation of vascular tone: is H2S a ‘potent’ vasodilator?
The majority of evidence for the physiological role of H2S has been obtained from studies on vascular tissue and has commonly concluded that H2S is a vasodilatory intermediate and perhaps an (or ‘the’) EDHF (endothelium-derived hyperpolarizing factor) . However, the precise role of endogenously generated H2S is not clear and requires critical appraisal. Generally, the application of high micromolar (typically ≥100 μM) concentrations of either H2S gas solutions or H2S donor compounds, such as Na2S (sodium sulfide) or NaSH (sodium hydrosulfide), to isolated and pre-contracted blood vessel preparations, including rodent aorta [45,91], mesenteric  and hepatic  beds or human internal mammary arteries , induce vessel relaxation in an endothelium-dependent and KATP-channel-dependent manner. In agreement with these findings, infusion of NaSH or Na2S solutions into anaesthetized animals has been observed to induce transient systemic blood pressure reduction [45,58], and conditions associated with hypotension, such as sepsis  and haemorrhagic shock , have been shown to have increased plasma levels and tissue production of H2S, whereas administration of inhibitors of CSE in these models normalized blood pressure. Conversely, decreased blood levels of ‘H2S’ have been observed in hypertensive rats [96–98], and genetic knockout studies comparing CSE+/+, CSE−/+ and CSE−/− mice have shown not only a stepwise reduction in plasma H2S levels with loss of CSE, but also increases in systemic blood pressure . However, it should be noted that this finding has not yet been confirmed . Furthermore, genetic manipulation of CBS also modulates systemic blood pressure and plasma levels of H2S , which again challenges the explicit tissue selectivity of the H2S-synthesizing enzymes CSE and CBS.
Clinical studies are emerging to confirm, albeit indirectly, a vasodilatory role of H2S in humans and have suggested that modulating systemic H2S production may represent a viable approach for the treatment of vascular disease. For example, compared with healthy controls, plasma H2S levels were increased 4-fold in patients with sepsis (up to 200 μM in one patient ), but were substantially reduced in hypertensive children . More recently, plasma ‘H2S’ levels in overnight-fasted men significantly negatively correlated with systemic blood pressure and impaired microvascular function in vivo and were significantly lower in overweight volunteers, and lower still in patients with Type 2 diabetes . Furthermore, plasma H2S levels also negatively correlated with glycaemic control [for example fasting glucose and plasma HbA1c (glycated haemoglobin levels], peripheral and central insulin sensitivity and adiposity [BMI (body mass index), waist and hip circumference and waist/hip ratio]. In one study comparing patients with CHD (coronary heart disease) with angiographically normal subjects, the number of affected coronary vessels (as well as plasma glucose levels) correlated with decreased plasma levels of H2S .
The studies described above have led to the wide and general suggestion that H2S is a potent vasodilator (some examples [10,102–105]), but are the effects of H2S on vascular tissue particularly ‘potent’? The concentrations of H2S gas solution or NaSH and Na2S required to induce tissue relaxation, and added as a bolus, have often been used in excess of 200 μM. These concentrations may be several orders of magnitude higher than the ‘free’ levels of H2S gas in blood and severalfold higher than the reported levels of total H2S (for example the sum of H2S, HS−, S2− and acid-labile sulfur) even in the most optimistic of studies, signifying that H2S is not a particularly potent vasodilator at all. Furthermore, glibenclamide and other KATP channel antagonists invariably prevent H2S-induced tissue relaxation, but only when high concentrations of H2S are used and when used at relatively high concentrations themselves (10–20 μM), strongly suggesting that at physiological levels H2S may exert more subtle effects on the vasculature. Additional mechanisms for the regulation of vascular tone have been proposed such as regulation of NO bioavailability [5,106,107], release of NO from nitrosothiols [44,45], inhibition of Cl−/HCO3− channels [108–110], metabolic inhibition  and activation of PKC (protein kinase C) or cAMP-dependent pathways . However, these studies have also used >300 μM H2S gas or NaSH to elicit these effects, and these additional pathways have not been demonstrated by others [91,92,111] and generally not investigated using human vessels. As such, the precise mechanism by which endogenously synthesized H2S regulates vascular tone is not clear.
H2S and respiratory smooth muscle
The toxicology of inhaled environmental H2S, leading to ocular and respiratory distress, has been well documented over several decades and reviewed in great detail elsewhere . More recent studies have strongly suggested that H2S may also function as an endogenous mediator in the lung. CSE and CBS mRNA and protein have been demonstrated in human pulmonary artery smooth muscle , and CSE is expressed in the airway and vascular smooth muscle in rat peripheral lung tissues [97,112,113]. As with conduit arteries, the effects of H2S donors have been investigated on isolated vessels from the lung, but the findings are similarly vague. In pre-contracted guinea-pig bronchial rings, the addition of up to 10 mM NaSH only produced modest tissue relaxation, whereas, in mouse bronchial tissue, significant relaxation was observed, albeit at NaSH concentrations of ≥500 μM . Under these conditions, tissue relaxation was resistant to antagonists or inhibitors of KATP, soluble guanylate cyclase, COX (cyclo-oxygenase)-1, COX-2 and tachykinin, and the precise mechanism(s) for lung smooth muscle relaxation in the lung has not been identified and it is unlikely, given the relatively low rate of CSE- and CBS-derived H2S synthesis [17,18,22,115,116], that bronchial tissue generates H2S at such high concentrations.
H2S has also been suggested to play a role in lung remodelling. For example, NaSH inhibited collagen accumulation in the wall of the pulmonary artery in hypoxia and aortocaval shunting rat models [117,118], and inhibition of endogenous H2S synthesis with AOAA [amino-oxyacetate; also called O-(carboxymethyl) hydroxylamine hemihydrochloride] and to a lesser extent PAG (propargyglycine) increased human airway smooth muscle proliferation and IL (interleukin)-8 secretion, whereas H2S donors substantially inhibited proliferation and IL-8 secretion by inhibiting ERK (extracellular-signal-regulated kinase) 1/2 and p38-dependent signalling pathways .
Clinical studies examining plasma levels of H2S and lung function are starting to emerge, albeit employing small sample sizes. Using a sulfide electrode, Chen et al.  showed that plasma levels of H2S in healthy controls were approximately 35 μM, but almost doubled in patients with stage I COPD (chronic obstructive pulmonary disease). Plasma H2S was decreased with increasing lung obstruction (stage I, ~72 μM; stage II, ~50 μM; stage III, ~40 μM; stage IV, ~48 μM), positively correlated with lung function [predicted FEV1 (forced expiratory volume in 1 s) and negatively correlated with sputum neutrophil count [119,120]. However, plasma H2S levels were unchanged in patients with AECOPD (acute exacerbation of COPD) , were unaffected by theophylline treatment  and serum levels were increased compared with control subjects . Intraperitoneal NaSH administration was recently shown to decrease tobacco-smoke-induced emphysema, lung injury and oxidative stress in mice , but human studies are similarly less clear cut; smoking either decreased plasma H2S  or had no effect on serum levels of H2S . Analysis of sputum levels of H2S would represent a more reliable index of lung H2S biogenesis or turnover, but, to date, these have not been assessed. As such, the role of endogenous H2S in the lung is also not clear.
Is endogenous H2S a regulator of inflammation?
Much has also been written about the diverse role of endogenous H2S in inflammatory signalling (reviewed in explicit detail elsewhere; for example [6,123]), but the precise role of endogenous H2S is not clear. Recently, we showed significantly higher levels of ‘H2S’ in knee joint SF (synovial fluid) aspirates from patients with RA (rheumatoid arthritis) compared with matched plasma or SF obtained from osteoarthritis patients . In that study, SF H2S negatively correlated with SF total white cell and neutrophil count, and positively correlated with disease activity. Increased SF H2S was not unique to RA, since subsequent studies also found increased levels in the SF from patients with psoriatic, reactive and septic arthritides . Increased plasma levels of H2S were also observed in patients with septic shock , and plasma H2S levels negatively correlated with plasma CRP (C-reactive protein) levels in lung infection , suggesting increased H2S synthesis may represent a generalized response to acute as well as chronic tissue injury and inflammation. However, it is not known whether H2S synthesis was elevated to drive the inflammatory response or was elevated to control or to limit tissue inflammation. Animal model studies employing sulfide salts (for example Na2S or NaSH) at high concentrations have not been able to clarify this issue; H2S either induces (or potentiates) inflammation or it is inhibitory. Conversely, inhibition of CSE with PAG (see below) consistently reduced inflammation (reviewed in detail elsewhere [6,123] and summarized in Table 4).
Is H2S an endogenous cytoprotective mediator?
Endogenous H2S has been proposed as a novel cytoprotective mediator , and there is growing evidence of direct and indirect antioxidant effects of H2S. In cell culture experiments, H2S/HS− generated from NaSH has been shown to ‘scavenge’ detrimental pro-inflammatory oxidants, such as H2O2 , ClO− , superoxide , ONOO−  and NO , inhibit cell death induced by these mediators as well as prevent oxidative modification of intracellular proteins [11,12] and LDL (low-density lipoprotein) . In neuronal cells, NaSH inhibited cell death induced by β-amyloid, mediated at least in part via antioxidant effects  and up-regulating intracellular glutathione synthesis through increasing cysteine uptake and elevating γ-glutamylcysteine synthetase activity . NaSH is also reported to degrade lipid peroxides , inhibit the expression and activity of NADPH oxidase [46,128] and up-regulate thioredoxin-1 expression in vascular endothelial cells . Increased hepatic GSH synthesis and decreased lipid peroxidation are also observed with Na2S treatment in a murine hepatic ischaemia/reperfusion injury model , and hepatocytes isolated from CSE−/− mice showed greater sensitivity to oxidative-stress-mediated injury than wild-type mice . Furthermore, glomeruli isolated from CBS−/+ mice showed increased production of endogenous ROS (reactive oxygen species) compared with glomeruli from wild-type animals . In animal models of smoke/burn injury , and myocardial , renal  and hepatic  ischaemia/reperfusion, H2S salt donors reduced the formation of nitrosatively and oxidatively modified cellular proteins, DNA and lipids, suggesting further an ‘antioxidant’ role for H2S.
However, recent detailed kinetic analysis has shown that the rate constants of the reaction between H2S (Na2S) and detrimental oxidant species are not sufficiently high enough for oxidant ‘scavenging’ alone to mediate the cytoprotective effects of H2S , strongly suggesting that other mechanisms must exist. Evidence for this suggestion is in the literature. For example, H2S from NaSH or Na2S induced ERK and Akt signalling pathways in cardiac tissue [79,99,100], and preserved mitochondrial ultrastructure and respiratory chain function in vivo , possibly via a mitochondrial pathway involving the preservation of Bcl-2 signalling, up-regulation of PKC and opening of mitochondrial KATP channels . H2S may also mediate cytoprotection via effects on other intracellular organelles such as the ER (endoplasmic reticulum), and has recently been shown to attenuate ER-stress-dependent cardiomyoctye cell death in a rat model of hyperhomocysteinaemia . H2S has also been proposed as a regulator of angiogenesis by promoting the proliferation of vascular endothelial cells and inhibiting the proliferation or inducing apoptotic cell death in vascular smooth muscle cells by modulation of ERK, p38, p21cip/WAF and caspase-mediated pathways. Furthermore, administration of Na2S or NaSH in vivo prevents myocardial tissue damage and cell loss in vivo [138–142] via up-regulation of Nrf2-dependent signalling pathways . Although the precise mechanism(s) mediating these phenomena have not been identified, it is possible that the signalling was mediated through selective protein sulfhydration . Although this is a highly attractive proposal, analogous to S-nitrosation of proteins by NO the extent to which this occurs is uncertain  and sulfhydration of intracellular proteins has only been demonstrated in isolated cells after the addition of >100 μM NaSH. However, that study  represents a proof-of-principle demonstration of the potential mechanisms by which H2S could exerts it physiological (and pharmacological) effects and in time may assist in dissecting the diverse signalling processes mediated by this intriguing biological gas (reviewed in detail in ).
PERILS AND PITFALLS: PHARMACOLOGICAL AND GENETIC TOOLS
Specific inhibitors of H2S synthesis?
Much of our current knowledge of the biology of H2S stems from the use of inhibitors of CSE such as D,L-PAG and BCA (β-cyanoalanine), and inhibitors of CBS, such as AOAA. These compounds target the PLP-binding site of these enzymes  and, while undoubtedly useful, they are not entirely specific and will target other PLP-dependent enzymes, especially over the wide concentration ranges commonly used in the laboratory for simple experiments on isolated tissues and cells (typically used between 1 and 10 mM). As such, the studies described in the immediately preceding section and summarized in Table 4 should be viewed with the following observations discussed below in mind. PAG is used as an irreversible and ‘specific’ inhibitor of CSE, but it has also been well used over the past few decades as a ‘specific’ inhibitor of other metabolic processes (summarized in Table 2). For example, PAG also inhibits several transamination reactions in muscle, alters amino acid metabolism and is metabolised to a renal toxin resulting in significant proteinuria, glucosuria and polyuria (Table 2). As such, possible diuretic and other renal effects of this PAG metabolite (or PAG itself) in mediating the effects of PAG (and H2S) in vivo should be considered when examining vascular, endocrine and inflammatory pathways. PAG administration to rats also increased whole-brain levels of cystathionine , suggesting that PAG may have little effect on endogenous H2S synthesis in the brain when administered systemically. Furthermore, urinary levels of proposed H2S ‘biomarkers’ such as SO42− are not reduced in rats after PAG treatment .
BCA (Table 2) has also been used to inhibit CSE in a reversible manner  and is used at high millimolar concentrations . Feeding BCA to rats results in the accumulation of free BCA and γ-glutamyl-β-cyanoalanylglycine in the brain, liver, plasma and muscle , suggesting it is widely bioavailable and metabolically processed. However, this compound is well known for its neurotoxic properties via NMDA (N-methyl-D-aspartate)-dependent and -independent mechanisms [149,150] and for its involvement in dietary-induced lathyrism in animals.
Similarly, AOAA is problematic. In other research fields, this compound is used as a general inhibitor of transaminase reactions, mitochondrial oxidative phosphorylation, amino acid transport and protein synthesis (summarized in Table 2). In neuronal systems, AOAA is known to stimulate GABA (γ-aminobutyric acid) synthesis in rat hippocampus and striatum  as well as induce neurotoxicity [152,153], seizures in mice  and neuronal cell loss in the striatum  and hippocampus  through inhibition of mitochondrial respiration , processes remarkably similar to that of H2S. However, it is possible that these effects could have been due, at least in part, to some inhibition of endogenous H2S synthesis. A further complication to the use of high concentrations of AOAA is that it could also perturb H2S metabolism through inhibition of mercaptopyruvate metabolism and MPST [158–160].
Hydroxylamine has also been used as a ‘specific’ inhibitor of CBS at millimolar concentrations [161–163], but its specificity and appropriateness is highly questionable (summarized in Table 2). For example, hydroxylamine is an endogenous molecule formed from the oxidation of NO2− (nitrites), NO3− (nitrates) and NH3 (ammonia) [164,165]. It has also been well established in the literature over several decades that hydroxylamine releases the gaseous mediator NO and forms NO− (nitroxyl ions) [164,166–170]. As such it is likely that many of the effects observed when using this compound and attributed to inhibition of CBS activity (and decreased levels of H2S) could be due to increased NO and/or HNO (nitroxyl) formation, such as vascular reactivity [165,171,172], promotion of oedema  and inflammation , glutamate signalling, and learning and memory .
Although PAG, AOAA and BCA significantly reduce H2S synthesis in various animal models and the activity of CSE and CBS in isolated cells and tissues in vitro, it has been so far assumed that these compounds are freely cell-permeant. The ‘effective’ concentrations of each of these inhibitors in any system have not been evaluated and to date these studies are lacking and are entirely overlooked. Therefore, in the absence of any evidence to the contrary, disparities in cellular permeability between cell types and species could at least partly explain the inconsistencies and controversies in the literature.
In the absence of anything better being currently available to researchers in a new and rapidly expanding field, the use of PAG, BCA and AOAA (but clearly not hydroxylamine), with adequate controls, should still provide a valuable insight into the physiological, pathological and pharmacological role of H2S in the body. However, they must be used with caution and investigators should resist overinterpreting their findings. A detailed analysis of the literature reveals that additional inhibitors have been used to inhibit CSE, CBS and CAT/MPST, but these have not been examined with respect to H2S synthesis. However, these compounds also either bind to the PLP-binding sites of CSE and CBS or have non-specific effects on other PLP-dependent enzymes (summarized in Table 2).
Genetic models lacking CSE [34,99] or CBS  have been generated and have similarly proved controversial. Wang and co-workers , using CSE-knockout animals, have largely supported the studies conducted using these ‘non-specific’ pharmacological tools in the vasculature (e.g. PAG) in that CSE−/− and CSE−/− mice had significantly lower plasma H2S levels, decreased vascular synthesis of H2S and had markedly higher systemic blood pressure than wild-type animals. However, rather than clarify the controversial role of H2S in the vasculature, these studies may have added to them. For example, Ishii et al.  showed that CSE-knockout animals were not hypertensive, and more recent studies [131,176] showed that CBS−/+ mice also had significantly decreased plasma H2S levels compared with wild-type animals, again questioning the earlier assumption that CSE is specifically a vascular enzyme and responsible for the vascular synthesis of H2S. These genetic CSE/CBS-knockout studies, the highly non-specific effects of inhibitors of CSE and CBS in vivo and their use at high concentrations, the high concentrations of exogenous H2S required to dilate vascular smooth muscle and the high concentrations of KATP channel antagonists required to inhibit the vascular effects of high concentrations of added H2S seriously question the mechanisms by which endogenous H2S regulates vascular tone.
H2S ‘donors’ and some notes of caution
The vast majority of studies which have examined the potential role of H2S in health and disease have invariably utilized commercially available sulfide salts such as Na2S and NaSH. Although these compounds have been useful as they can be conveniently used to prepare standardized solutions of H2S and circumvents the requirement for H2S gas cylinders, they are not particularly relevant tools to examine the physiology of H2S in vitro or in vivo. For example, we have shown that the addition of Na2S or NaSH (or saturated solutions of H2S gas prepared from H2S gas cylinders) to aqueous solutions results in the instantaneous release of a bolus of H2S which dissipates in seconds [116,177]. It is highly unlikely that tissues or cells are ever exposed to H2S generated in such a rapid manner which generates very high local concentrations of H2S (as well as HS− and Na+), since endogenously produced H2S through CSE and CBS is relatively slow and sustained and may well be synthesized in a steady ‘flux’ [17,18,22,115,116]. Typically concentrations of Na2S and NaSH have been employed at concentrations between 50 and 1000 μM (and often much higher than this ) and when administered to animals, blood and tissue levels of added sulfide have only been detailed in very few studies . Authors have generally argued that due to the pKa of NaSH the final concentration of H2S is approximately one-third of that of the final concentration of sulfide salt added and HS− accounts for the remaining two-thirds. However, it is currently not known whether the biological and pharmacological effects of H2S are determined by H2S itself or from HS−, which will always be present at physiological pH no matter what the source of H2S or the donor used. Given that the progression of the field is now towards ‘free’ levels of H2S being several orders of magnitude lower than previously considered (nanomolar rather than micromolar) , use of Na2S and NaSH at these concentrations and added or administered as a bolus is physiologically questionable and assumptions that high concentrations of Na2S/NaSH can effectively model the effects of endogenously produced H2S should not be made. More recent determinations of the levels of ‘free’ H2S in plasma are in the region of 0.4–0.9 μM . As such, compounds which release concentrations of H2S at this level and over a much longer period of time will most likely be of more physiological relevance. However, as a novel therapeutic approach to reperfusion injury and stroke, clinical-grade Na2S is showing considerable clinical promise as a pharmacological agent (Table 4), although the precise mechanisms of action have not been defined.
RECENT ADVANCES: SLOW-RELEASE H2S DONOR MOLECULES
Novel H2S donors have been developed and synthesized to circumvent the problem of using sulfide salts such as Na2S and NaSH as a means to expose animals, tissue and cells to H2S generated in a physiological manner (reviewed in [6,179,180]). The majority of these studies have modified existing pharmacological compounds such as NSAIDs (non-steroidal anti-inflammatory drugs) with ADT-OH [5-(4-hydroxyphenyl)-3H-1,2-dithiole-3-thione], examples of which are illustrated in Table 5. The combination of well-established NSAIDs with ADT-OH derivatives have clearly highlighted the pharmacological potential for H2S, and these novel compounds have shown substantial promise in alleviating and limiting gastrointestinal side effects and toxicity of NSAIDS and in the treatment of inflammatory bowel disease, oedema, endotoxic shock and acute inflammation (Table 5). However, the precise mechanism by which ADT-OH ‘releases’ H2S has not been demonstrated and, since dithiolethiones are themselves biologically active, it is possible that some of the observed biological effects associated with ADT-OH derivatives were due to ADT-OH itself rather than released H2S (reviewed in [6,181]). For example, dithiolethiones have been shown to activate Nrf-2-dependent phase II enzymes, such as γ-glutamylcysteine synthetase, and elevate intracellular glutathione, NADPH:quinone oxidireductase, gluathione reductase and catalase [182–185]. However, the role of H2S released from these molecules in these studies was not investigated and it is possible these effects were due, at least in part, to released H2S.
More recently, H2S donors which do not consist of structurally modified established drug molecules such as GYY4137 have been synthesized and characterized [116,177,186]. GYY4137 is a very slow-releasing H2S donor compound which releases two molecules of H2S per molecule of GYY4137 and has been shown to exert prominent endothelium-dependent vasodilatory activity in vivo via KATP-channel-dependent mechanisms, as well as exert prominent anti-inflammatory activity in vitro and in vivo mediated in part via inhibition of NF-κB (nuclear factor κB)/AP-1 (activator protein-1)-dependent pro-inflammatory signalling (Table 5). In human articular chondrocytes and trabecular bone-derived mesenchymal progenitor cells, GYY4137 protected these cells against oxidative-stress-mediated cytotoxicity, induced Akt phosphorylation and preserved mitochondrial function . GYY4137 offers the additional advantage to the researcher over ADT-OH compounds in that its decomposition products appear inactive , allowing the physiological effects of slow release of H2S to be studied directly in the absence of any possible additive, but therapeutically highly useful, pharmacological effects of NSAID, known drug or ADT-OH. However, it should be noted here that the precise metabolic and pharmacokinetic profiles for any of the ADT-OH-containing H2S donors or GYY4137 have yet to be fully elucidated so it is possible that at least some of the reported effects in vivo could be due to metabolism to biologically active intermediates rather than H2S. As such, detailed control experiments are required to confirm any observations are due to H2S and not the parent compound or decomposed (‘spent’) donor.
Despite the recent advancement in the generation of H2S donor molecules, one clear area that has stagnated is the development of inhibitors that are specific for each of the endogenous H2S-synthesizing enzymes. Although CSE- and CBS-knockout animals and cell lines have been generated in some laboratories, these tools are not widely available, may not be applicable to certain experimental conditions and have generated considerable controversy themselves. Until specific inhibitors are identified, the physiological and pathophysiological role of endogenously generated H2S will remain unclear.
H2S is emerging as a highly significant endogenous gaseous mediator which may play a substantial regulatory role in a variety of physiological systems, suggesting the therapeutic potential for manipulation of H2S in disparate human pathologies (summarized in Figure 1). However, understanding the complex physiology and pharmacology of endogenous H2S and pharmacological H2S has, as with any new and emerging field of research, been limited by the availability of specific research tools which has generated the inevitable controversy. In particular, the field is hampered by a lack of wholly enzyme- and tissue-specific inhibitors, H2S donors which release H2S at physiologically relevant rates and robust ‘biomarkers’ of H2S formation and metabolism. These non-specific tools have generated considerable controversy and often contradictory findings. CSE−/− knockout animals have highlighted a crucial role of H2S in the regulation of blood pressure , but these studies are similarly controversial and recent studies using CBS−/−-knockout animals have further suggested that this enzyme is also a significant contributor to vascular H2S generation [131,176]. Neither of these genetic tools have yet been applied to animal models of human disease.
Molecular biology approaches using RNAi (RNA interference) offers the opportunity to selectively inactivate CSE, CBS and MPST, but this may not be suitable in every system such as primary cell cultures, certain cell lines known to be difficult to transfect, animal models and human clinical studies. The development of selective inhibitors which avoid the non-specific effects highlighted in Table 2 are crucial for the field to advance. Slow-releasing H2S-releasing molecules offer the opportunity to study H2S under physiologically relevant conditions, but as with any new experimental tool its use has so far been limited. Similarly, ADT-OH derivatives of NSAIDs have generated a new and exciting class of pharmacological compounds which have shown considerable therapeutic promise for the treatment of inflammatory and vascular conditions. The availability of these new H2S donors which release H2S in a slow and sustained manner will greatly increase our understanding of this remarkable endogenous gaseous mediator.
Abbreviations: ADT-OH, 5-(4-hydroxyphenyl)-3H-1,2-dithiole-3-thione; AOAA, amino-oxyacetate; BCA, β-cyanoalanine; BMI, body mass index; CAT, cysteine aminotransferase; CBS, cystathionine-β-synthase; CHD, coronary heart disease; COPD, chronic obstructive pulmonary disease; AECOPD, acute exacerbation of COPD; COX, cyclo-oxygenase; CRP, C-reactive protein; CSE, cystathionine-γ-lyase; CSE-s, truncated CSE; DS, Down's syndrome; ER, endoplasmic reticulum; ERK, extracellular-signal-regulated kinase; FEV1, forced expiratory volume in 1 s; GABA, γ-aminobutyric acid; Hb1Ac, glycated haemoglobin; IL, interleukin; LPS, lipopolysaccharide; MPST, 3-mercaptopyruvate sulfurtransferase; NMDA, N-methyl-D-aspartate; NF-κB, nuclear factor κB; NSAID, non-steroidal anti-inflammatory drug; PAG, propargyglycine; PGE2, prostaglandin E2; PKC, protein kinase C; PLP, pyridoxal-5′-phosphate; RA, rheumatoid arthritis; SF, synovial fluid; TNF-α, tumour necrosis factor-α
- © The Authors Journal compilation © 2011 Biochemical Society